PRINCIPLES OF TISSUE ENGINEERING 3RD EDITION - PART 2 (end) doc

With cell therapies based on potentially autologous nonpancreatic cells, targeted by gene expression vectors in vivo, or retrieved surgically, engineered ex vivo, and returned to the ho

I Introduction

II Cell Types for Pancreatic Substitutes

III Construct Technology

IV In Vivo Implantation

V Concluding Remarks

VI Acknowledgments VII References

Bioartifi cial Pancreas

Athanassios Sambanis

Principles of Tissue Engineering, 3 rd Edition

ed by Lanza, Langer, and Vacanti

Copyright © 2007, Elsevier, Inc.

All rights reserved.

I INTRODUCTION

Diabetes is a signifi cant health problem, affecting an

estimated 20.8 million people in the United States alone,

with nearly 1.8 million affl icted with type 1 diabetes [http://

diabetes.niddk.nih.gov/dm/pubs/statistics/index.htm#7]

Type 1 diabetes results from the loss of insulin-producing

cell mass (the β-cells of pancreatic islets) due to

autoim-mune attack Type 2 diabetes has a more complicated

disease etiology and can be the result of not producing

enough insulin and/or the body’s developing a resistance to

insulin Although initially controlled by diet, exercise, and

oral medication, type 2 diabetes often progresses toward

insulin dependence It is estimated that insulin-dependent

diabetics (both types 1 and 2) exceed 4 million people in the

United States Although insulin-dependent diabetes (IDD)

is considered a chronic disease, even the most vigilant

insulin therapy cannot reproduce the precise metabolic

control present in the nondiseased state The poor temporal

match between glucose load and insulin activity leads to a

number of complications, including increased risk of heart

disease, kidney failure, blindness, and amputation due to

peripheral nerve damage Providing more physiological

control would alleviate many of the diabetes-related health

problems, as suggested by fi ndings from the Diabetes

Control and Complications Trial (The Diabetes Control and

Complications Trial Research Group, 1997) and its

continu-ation study (DCCT/EDIC NEJM 353(25):2643–53, 2005)

Cell-based therapies, which provide continuous regulation

of blood glucose through physiologic secretion of insulin, have the potential to revolutionize diabetes care

Several directions are being considered for cell-based therapies of IDD, including implantation of immunopro-tected allogeneic or xenogeneic islets, of continuous cell lines, or of engineered non-β-cells For allogeneic islet trans-plantation, a protocol developed by physicians at the Uni-

versity of Edmonton (Shapiro et al., 2001a, 2001b, 2001c,

Bigam and Shapiro, 2004) has dramatically improved the survivability of grafts The protocol uses human islets from cadaveric donors, which are implanted in the liver of care-fully selected diabetic recipients via portal vein injection The success of the Edmonton protocol is attributed to two modifi cations relative to earlier islet transplantation studies: the use of a higher number of islets and the implementation

of a more benign, steroid-free immunosuppressive regimen However, two barriers prevent the widespread application

of this therapy The fi rst is the limited availability of human tissue, because generally more than one cadaveric donor pancreas is needed for the treatment of a single recipient The second is the need for life-long immunosuppression, which, even with the more benign protocols, results in long-term side effects to the patients

A tissue-engineered pancreatic substitute aims to address these limitations by using alternative cell sources, relaxing the cell availability limitation, and by reducing or eliminating the immunosuppressive regimen necessary for survival of the graft A number of signifi cant challenges are

facing the development of such a substitute, however These

include procuring cells at clinically relevant quantities; the

immune acceptance of the cells, which is exacerbated in

type 1 diabetes by the resident autoimmunity in the patients;

and the fact that diabetes is not an immediately

life-threat-ening disease, so any other therapy will have to be more

effi cacious and/or less invasive than the current standard

treatment of daily blood glucose monitoring and insulin

injections

In general, developing a functional living-tissue

replace-ment requires advances and integration of several types of

technology (Nerem and Sambanis, 1995) These are (1) cell

technology, which addresses the procurement of functional

cells at the levels needed for clinical applications; (2)

con-struct technology, which involves combining the cells with

biomaterials in functional three-dimensional confi

gura-tions Construct manufacturing at the appropriate scale,

and preservation, as needed for off-the-shelf availability,

also fall under this set of technologies; (3) technologies for

in vivo integration, which address the issues of construct

immune acceptance, in vivo safety and effi cacy, and

moni-toring of construct integrity and function postimplantation

The same three types of technology need also be developed

for a pancreatic substitute It should be noted, however, that

the critical technologies differ, depending on the type of

cells used With allogeneic or xenogeneic islets or beta-cells,

the major challenge is the immune acceptance of the

implant In this case, encapsulation of the cells in

permse-lective membranes, which allow passage of

low-molecular-weight nutrients and metabolites but exclude larger

antibodies and cytotoxic cells of the host, may assist the

immune acceptance of the graft With cell therapies based

on potentially autologous nonpancreatic cells, targeted

by gene expression vectors in vivo, or retrieved surgically,

engineered ex vivo, and returned to the host, the major

challenge is engineering insulin secretion in precise response

to physiologic stimuli Lastly, with stem or progenitor cells, the primary hurdle is their reproducible differentiation into cells of the pancreatic β-phenotype Figure 42.1 shows schematically the two general therapeutic approaches based

on allo- or xenogeneic cells (Fig 42.1A) or autologous cells (Fig 42.1B)

This chapter is therefore organized as follows We fi rst describe the types of cells that have been used or are of potential use in engineering a pancreatic substitute We then discuss issues of construct technology, specifi cally encapsulation methods and the relevant biomaterials, man-ufacturing issues, and preservation of the constructs The

challenges of in vivo integration and results from in vivo

experiments with pancreatic substitutes are presented next

We conclude by offering a perspective on the current status and the future challenges in developing an effi cacious, clini-cally applicable bioartifi cial pancreas

II CELL TYPES FOR PANCREATIC SUBSTITUTES

Islets

Despite several efforts, the in vitro expansion of primary

human islets has met with limited success Adult human islets are diffi cult to propagate in culture, and their expan-sion leads to dedifferentiation, generally manifested as loss

of insulin secretory capacity Although there exist reports on

the redifferentiation of expanded islets (Lechner et al., 2005;

Ouziel-Yahalom et al., 2006) and of nonislet pancreatic cells, which are discarded after islet isolation (Todorov et al., 2006), the phenotypic stability and the in vivo effi cacy of

these cells remain unclear Additionally, with expanded and

A Allo- or Xenogeneic Cells B Autologous Cells

Implantation

Cell retrieval

Ex vivo

manipulation

Cell storage

Cell storage

Cell amplification

Cell encapsulation

Implantation

Capsule storage

Ex vivo

manipulation

Cell storage

Cell storage

Cell amplification

Cell encapsulation

Implantation

Capsule storage Capsule storage

Cell

gene therapy

FIG 42.1 Approaches for bioartifi cial pancreas development using allo- or xenogeneic cells (A) and autologous cells (B) In (A), islets are procured from

pancreatic tissue, or cell lines are thawed from cryostorage and expanded in culture; cells are encapsulated for immunoprotection before they are implanted

to achieve a therapeutic effect; encapsulated cells may also be cryopreserved for inventory management and sterility testing In (B), cells are retrieved

surgi-cally from the patient; manipulated ex vivo phenotypisurgi-cally and/or genetisurgi-cally in order to express β-cell characteristics, and in particular physiologically

responsive insulin secretion; the cells are implanted for a therapeutic effect either by themselves or, preferably, after incorporation in a three-dimensional

substitute; some of the cells may be cryopreserved for later use by the same individual In in vivo gene therapy approaches, a transgene for insulin

expres-sion is directly introduced into the host and expressed by cells in nonpancreatic tissues

I I C E L L T Y P E S F O R P A N C R E A T I C S U B S T I T U T E S • 621

redifferentiated islets, it remains unknown whether the

insulin-secreting cells arose from the redifferentiation of

mature endocrine cells or from an indigenous stem or

pro-genitor cell population in the tissue isolate (Todorov et al.,

2006)

Animal, such as porcine, islets are amply available, and

porcine insulin is very similar to human, differing by only

one amino acid residue However, the potential use of

porcine tissue is hampered by the unlikely but possible

transmission of porcine endogenous retroviruses (PERV) to

human hosts as well as by the strong xenograft

immuno-genicity that they elicit Use of closed, PERV-free herds is

reasonably expected to alleviate the fi rst problem With

regard to immunogenicity, a combination of less

immuno-genic islets, islet encapsulation in permselective barriers,

and host immunosuppression may yield long-term survival

of the implant The use of transgenic pigs that do not express

the α-Gal (α[1,3]-galactose) epitope is one possible approach

for reducing the immunogenicity of the islets Studies also

indicate that neonatal pig islets induce a lower T-cell

reac-tivity than adult islets (Bloch et al., 1999), even though the

α-Gal epitope is abundant in neonatal islets as well (Rayat

et al., 1998) Furthermore, it is possible that the primary

antigenic components in islet tissue are the ductal epithelial

and vascular endothelial cells, which express prominently

the α-Gal epitope; on the other hand, β-cells express the

epitope immediately after isolation but not after

mainte-nance in culture (Heald et al., 1999) It should also be noted

that the large-scale isolation of porcine islets under

condi-tions of purity and sterility that will be needed for eventual

regulatory approval pose some major technical hurdles,

which have not been addressed yet

-Cell Lines

Recognizing the substantial diffi culties involved with

the procurement and amplifi cation of pancreatic islets,

several investigators have pursued the development of

con-tinuous cell lines, which can be amplifi ed in culture yet

retain key differentiated properties of normal β-cells One of

the fi rst successful developments in this area was the

gen-eration of the βTC family of insulinomas, derived from

transgenic mice carrying a hybrid insulin-promoted simian

virus 40 tumor antigen gene; these cells retained their

dif-ferentiated features for about 50 passages in culture (Efrat

et al., 1988) The hypersensitive glucose responsiveness of

the initial βTC lines was reportedly corrected in subsequent

lines by ensuring expression of glucokinase and of the

high-K m glucose transporter Glut2, and with no or low expression

of hexokinase and of the low-K m transporter Glut1 (Efrat et

al., 1993; Knaack et al., 1994) A similar approach was used

to develop the mouse MIN-6 cell line that exhibits

glucose-responsive secretion of endogenous insulin (Miyazaki et al.,

1990) Subsequently, Efrat and coworkers developed the

βTC-tet cell line, in which expression of the SV40 T antigen

(Tag) oncoprotein is tightly and reversibly regulated by

tet-racycline Thus, cells proliferate when Tag is expressed, and

shutting off Tag expression halts cell growth (Efrat et al.,

1995; Efrat, 1998) Such reversible transformation is an elegant approach in generating a supply of β-cells via pro-liferation of an inoculum, followed by arrest of the growth

of cells when the desirable population size is reached When retained in capsules, proliferating cells do not grow uncon-trollably, since the dissolved-oxygen concentration in the surrounding milieu can support up to a certain number of viable, metabolically active cells in the capsule volume This number of viable cells is maintained through equilibration

of cell growth and death processes (Papas et al., 1999a,

1999b) Thus, growth arrest is useful primarily in preventing the growth of cells that have escaped from broken capsules

in vivo and in reducing the cellular turnover in the capsules

The latter reduces the number of accumulated dead cells in the implant and thus the antigenic load to the host affected

by proteins from dead and lysed cells that pass through the capsule material

In a different approach, Newgard and coworkers (Clark

et al., 1997) carried out a stepwise introduction of genes

related to β-cell performance into a poorly secreting rat insulinoma (RIN) line In particular, RIN cells were itera-tively engineered to stably express multiple copies of the insulin gene, the glucose transporter Glut2, and the gluco-kinase gene, which are deemed essential for proper expres-sion of β-cell function Although this is an interesting methodology, it is doubtful that all genes necessary for reproducing β-cell function can be identifi ed and stably expressed in a host cell Recently, signifi cant progress was made toward establishing a human pancreatic β-cell line that appears functionally equivalent to normal β-cells

(Narushima et al., 2005) This was accomplished through a

complicated procedure involving retroviral transfection of primary β cells with the SV40 large T antigen and cDNAs of human telomerase reverse transcriptase This resulted in a reversibly immortalized human β-cell clone, which secreted insulin in response to glucose, expressed β-cell transcrip-tional factors, prohormone convertases 1/3 and 2, which process proinsulin to mature insulin, and restored normo-glycemia upon implantation in diabetic immunodefi cient

mice (Narushima et al., 2005).

With regard to β-cell lines capable of proliferation under the appropriate conditions, key issues that remain to be addressed include (1) their long-term phenotypic stability,

in vitro and in vivo; (2) their potential tumorigenicity, if cells

escape from an encapsulation device, especially when these cells are allografts that may evade the hosts’ immune defenses for a longer period of time than acutely rejected xenografts; and (3) their possible recognition by the auto-immune rejection mechanism in type 1 diabetic hosts

Engineered Non– Pancreatic Cells

The use of non–β pancreatic cells from the same patient, engineered for insulin secretion, relaxes both the cell availability and immune acceptance limitations that exist with other types of cells It has been shown that the A-chain/

C-peptide and B-chain/C-peptide cleavage sites on the

pro-insulin gene can be mutated so that the ubiquitous

endo-peptidase furin recognizes and completely processes

proinsulin into mature insulin absent of any intermediates

(Yanagita et al., 1992) Based on this concept, several

non-endocrine cell lines have been successfully transfected to

produce immunoreactive insulin, including hepatocytes,

myoblasts, and fi broblasts (Yanagita et al., 1993) In a

differ-ent approach, Lee and coworkers (2000) expressed a

syn-thetic single-chain insulin analog, which does not require

posttranslational processing, in hepatocytes Although

recombinant insulin expression is relatively straightforward,

a key remaining challenge is achieving the tight regulation

of insulin secretion in response to physiologic stimuli, which

is needed for achieving normoglycemia in higher animals

and, eventually, humans

One approach for achieving regulation of insulin

secre-tion is through regulasecre-tion of biosynthesis at the gene

tran-scription level, as realized in hepatocytes by Thule et al

(Thule et al., 2000; Thule and Liu, 2000) and Lee et al (2000)

Besides the ability to confer transcriptional-level regulation,

hepatocytes are particularly attractive as producers of

recombinant insulin due to their high synthetic and

secre-tory capacity and their expression of glucokinase and Glut2

(Cha et al., 2000; Lannoy et al., 2002) Hepatic delivery by

viral vectors and expression of the glucose-responsive

insulin transgene in diabetic rats controlled the

hypergly-cemic state for extended periods of time (Lee et al., 2000;

Thule and Liu, 2000; Olson et al., 2003) Nevertheless,

trans-criptional regulation is sluggish, involving long time lags

between stimulation of cells with a secretagogue and

induced insulin secretion as well as between removal of the

secretagogue and down-regulation of the secretory response

(Tang and Sambanis, 2003) The latter is physiologically

more important, because it means that the cells continue to

secrete insulin after glucose has been down-regulated, thus

resulting in potentially serious hypoglycemic excursions

Increasing the number of stimulatory glucose elements in a

promoter enhances the cellular metabolic responsiveness

in vitro (Thule et al., 2000) With regard to secretion

down-regulation, Tang and Sambanis (2003) hypothesized that the

slow kinetics of this process following removal of the

tran-scriptional activator are due to the stability of the

preproin-sulin mRNA, which continues to become translated after

transcription has been turned off Using a modifi ed

prepro-insulin cDNA that produced an mRNA with two more copies

of the insulin gene downstream of the stop codon resulted

in preproinsulin mRNA subjected to nonsense mediated

decay and thus destabilized This signifi cantly expedited the

kinetics of secretion down-regulation on turning off

tran-scription (Tang and Sambanis, 2003) Thus, the

combina-tion of optimal transcripcombina-tional regulacombina-tion with transgene

message destabilization promises further improvements in

insulin secretion dynamics from transcriptionally regulated

hepatic cells It should be noted, however, that despite the

time delays inherent in transcriptional regulation, hepatic insulin gene therapy is suffi cient to sustain vascular nitric oxide production and inhibit acute development of diabetes-associated endothelial dysfunction in diabetic

rats (Thule et al., 2006) Hence, many aspects of the

thera-peutic potential of hepatic insulin expression remain to be explored

Another appealing target cell type is endocrine cells, which possess a regulated secretory pathway and the enzymatic machinery needed to process authentic proinsu-lin into insulin Early work in this area involved expression

of recombinant insulin in the anterior pituitary mouse

AtT20 cell line (Moore et al., 1983), which can be sub

-jected to repeated episodes of induced insulin secretion

using nonmetabolic secretagogues (Sambanis et al., 1990)

Cotransfection with genes encoding the glucose transporter Glut-2 and glucokinase resulted in glucose-responsive

insulin secretion (Hughes et al., 1992, 1993) However,

limi-tations of this approach include possible instabilities in the cellular phenotype and the continued secretion of endoge-nous hormones, such as adrenocorticotropic hormone from AtT-20 cells, which are not compatible with prandial metabolism

In this regard, endocrine cells of the intestinal lium, or enteroendocrine cells, are especially promising

epithe-Enteroendocrine cells secrete their incretin products in a tightly controlled manner that closely parallels the secretion

of insulin following oral glucose load in human subjects;

incretin hormones are fully compatible with prandial

metabolism and glucose regulation (Schirra et al., 1996;

Kieffer and Habener, 1999) As with β-cells, enteroendocrine cells are polar, with sensing microvilli on their luminal side and secretory granules docked at the basolateral side, adja-cent to capillaries Released incretin hormones include the glucagon-like peptides (GLP-1 and GLP-2) from intestinal L-cells and glucose-dependent insulinotropic polypeptide (GIP) from K-cells, which potentiate insulin production from the pancreas after a meal (Drucker, 2002) The impor-tance of enteroendocrine cells (and, in particular, L-cells) was fi rst put forward by Creutzfeldt (1974), whose primary interest in these cells was for the prospect of using GLP-1 for the treatment of type 2 diabetes Furthermore, ground-

breaking work by Cheung et al (2000) demonstrated that

insulin produced and secreted by genetically modifi ed intestinal K-cells of transgenic mice prevented the animals from becoming diabetic after injection with streptozotocin (STZ), which specifi cally kills the β-cells of the pancreas

This is an important proof-of-concept study, which showed that enteroendocrine cell–produced insulin can provide regulation of blood glucose levels Subsequent work with a human intestinal L-cell line demonstrated that these cells can be effectively transduced to express recombinant human insulin, which colocalizes in secretory vesicles with endog-enous GLP-1 and thus is secreted with identical kinetics to GLP-1 in response to stimuli (Tang and Sambanis, 2003,

2004) The intestinal tract could be considered an attractive

target for gene therapy because of its large size, making it

the largest endocrine organ in the body (Wang et al., 2004);

however, enteroendocrine cell gene therapy faces serious

diffi culties due to anatomic complexity, with the

entero-endocrine cells being located at the base of invaginations of

the gut mucosa called crypts, the very harsh conditions in

the stomach and intestine, and the rapid turnover of the

intestinal epithelium

Contrary to direct in vivo gene delivery, ex vivo

gene therapy involves retrieving the target cells surgically,

culturing them and possibly expanding them in vitro,

genetically engineering them to express the desired

proper-ties, and then returning them to the host, either as such or

in a three-dimensional tissue substitute It is generally

thought that the ex vivo approach is advantageous, for

it allows for the thorough characterization of the genetically

engineered cells prior to implantation, possibly for the

preservation of some of the cells for later use by the

same individual and, importantly, for localization and

retrievability of the implant However, the challenges

imposed by the ex vivo approach, including the surgical

retrieval, culturing, and in vitro genetic engineering of the

target cells are signifi cant, so such methods are currently

under development

Differentiated Stem or Progenitor Cells

Naturally, throughout life, islets turn over slowly, and

new, small islets are continually generated from ductal

pro-genitors (Finegood et al., 1995; Bonner-Weir and Sharma,

2002) There is also considerable evidence that adult

plu-ripotent stem cells may be a possible source of new islets

(Bonner-Weir et al., 2000; Ramiya et al., 2000; Kojima et al.,

2003) However, efforts to regenerate β-cells in vitro or in

vivo by differentiation of embryonic or adult stem or

pan-creatic progenitor cells have produced mixed results

Insulin-producing, glucose-responsive cells, as well as other

pan-creatic endocrine cells, have been generated from mouse

embryonic stem cells (Lumelsky et al., 2001)

Insulin-secret-ing cells obtained from embryonic stem cells reversed

hyperglycemia when implanted in mice rendered diabetic

by STZ injection (Soria et al., 2000) In another study, mouse

embryonic stem cells transfected to express constitutively

Pax4, a transcription factor essential for β-cell development,

differentiated into insulin-producing cells and normalized

blood glucose when implanted in STZ-diabetic mice

(Blyszczuk et al., 2003) On the other hand, other studies do

not support differentiation of embryonic stem cells into the

β-cell phenotype (Rajagopal et al., 2003) Overall, the mixed

and somewhat inconsistent results point to the

consider-able work that needs to be done before stem or progenitor

cells can be reliably differentiated into β cells at a clinically

relevant scale Harnessing the in vivo regenerative capacity

of the pancreatic endocrine system may present a

promis-ing alternative approach

Engineering of Cells for Enhanced

Survival In Vivo

Because islets and other insulin-secreting cells

experi-ence stressful conditions during in vitro handling and in

vivo postimplantation, several strategies have been

imple-mented to enhance islet or nonislet cell survival in atic substitutes Strategies generally focus on improving the immune acceptance of the graft, enhancing its resistance to cytokines, and reducing its susceptibility to apoptosis Phe-notypic manipulations include extended culturing of neo-natal and pig islets at 37°C, which apparently reduces their immunogenicity, possibly by down-regulating the major histocompatibility class 1 antigens on the islet surface; islet pretreatment with TGF-β1; and enzymatic treatment of pig islets with α-galactosidase to reduce the a-galactosyl epitope

pancre-on islets (Prokop, 2001) However, the permanency of these modifi cations is unknown For instance, a-galactosyl epi-topes reappear on islets 48 hours after treatment with α-galactosidase With proliferative cell lines destined for recombinant insulin expression, selection of clones resis-

tant to cytokines appears feasible (Chen et al., 2000) Gene

chip analysis of resistant cells may then be used to identify the genes responsible for conferring cytokine resistance

Genetic modifi cations for improving survival in vivo

may offer prolonged expression of the desired properties relative to phenotypic manipulations, but they also present the possibility of modifying the islets in additional, undesir-able ways Notable among the various proposed approaches, reviewed in Jun and Yoon (2005), are the expression of the immunomodulating cytokines IL-4 or a combination of IL-

10 and TGF-β, which promoted graft survival by preventing immune attack in mice; and the expression of the antiapop-totic bcl-2 gene using a replication defective herpes simplex virus, which resulted in protection of β-cells from a cytokine mixture of interleukin-1β, TNF-α, and IFN-γ in vitro.

III CONSTRUCT TECHNOLOGY

Construct technology focuses on associating cells with biocompatible materials in functional three-dimensional confi gurations Depending on the type of cells used, the primary function of the construct can be one or more of the following: to immunoprotect the cells postimplantation, to

enable cell function, to localize insulin delivery in vivo, or

to provide retrievability of the implanted cells

Encapsulated Cell Systems

Encapsulation for immunoprotection involves rounding the cells with a permselective barrier, in essence

sur-an ultrafi ltration membrsur-ane, which allows passage of molecular-weight nutrients and metabolites, including insulin, but excludes larger antibodies and cytototoxic cells

low-of the host Figure 42.2 summarizes the common types low-of encapsulation devices, which include spherical microcap-sules, tubular or planar diffusion chambers, thin sheets, and vascular devices

I I I C O N S T R U C T T E C H N O L O G Y • 623

Encapsulation can be pursued via one of two general

approaches With capsules fabricated using water-based

chemistry, cells are fi rst suspended in un-cross-linked

polymer, which is then extruded as droplets into a solution

of the cross-linking agent A typical example here is the very

commonly used alginate encapsulation Alginate is a

complex mixture of polysaccharides obtained from

sea-weeds, which forms a viscous solution in physiologic saline

Islets or other insulin-secreting cells are suspended in

sodium alginate, and droplets are extruded into a solution

of calcium chloride Calcium cross-links alginate,

instanta-neously trapping the cells within the gel The size of the

droplets, hence also of the cross-linked beads, can be

con-trolled by fl owing air parallel to the extrusion needle so that

droplets detach at a smaller size than if they were allowed

to fall by gravity; or by using an electrostatic droplet

genera-tor, in which droplets are detached from the needle by

adjusting the electrostatic potential between the needle and

the calcium chloride bath Capsules generated this way can

have diameters from a few hundred micrometers to more

than one millimeter Alginate by itself is relatively

perme-able; to generate the permselective barrier, beads are treated

with a polycationic solution, such as l-lysine or

poly-l-ornithine The reaction time between alginate and the

polycation determines the molecular weight cutoff of the

resulting membrane Poly-l-lysine is highly infl ammatory

in vivo, however, so beads are coated with a fi nal layer of

alginate to improve their biocompatibility Hence, calcium

alginate/poly-l-lysine/alginate (APA) beads are fi nally

formed Treating the beads with a calcium chelator, such as

citrate, presumably liquefi es the inner core, forming APA

membranes Other materials that have been used for cell

microencapsulation include agarose, photo-cross-linked

poly(ethylene glycol), and (ethyl methacrylate, methyl

methacrylate, and dimethylaminoethyl methacrylate)

copolymers (Mikos et al., 1994; Sefton and Kharlip, 1994)

Advantages of hydrogel microcapsules include a high

surface-to-volume ratio, and thus good transport

proper-ties, as well as ease of handling and implantation Small beads can be implanted in the peritoneal cavity of animals simply by injection, without the need for incision Other common implantation sites include the subcutaneous space and the kidney capsules Disadvantages include the fragility

of the beads, especially if the cross-linking cation becomes chelated by compounds present in the surrounding milieu

or released by lysed cells, and the lack of easy retrievability once the beads have been dispersed in the peritoneal cavity

of a host Earlier problems caused by the variable tion of alginates and the presence of endotoxins have been resolved through the development and commercial avail-ability of ultrapure alginates of well-defi ned molecular

composi-weight and composition (Sambanis, 2000; Stabler et al.,

2001)

Hydrogels impose little diffusional resistance to solutes, and indeed effective diffusivities in calcium alginate and agarose hydrogels are in the range of 50–100% of the corre-sponding diffusivities in water (Tziampazis and Sambanis, 1995; Lundberg and Kuchel, 1997) However, with conven-tional microencapsulation, the volume of the hydrogel con-tributes signifi cantly to the total volume of the implant For example, with a 500-µm microcapsule containing a 300-µm islet, the polymer volume constitutes 78% of the total capsule volume Additionally, conventional microcapsules may not

be appropriate for hepatic portal vein implantation because, besides their higher implant volume relative to the same number of naked islets, they may result in higher portal vein pressure and more incidences of blood coagulation in the liver To address this problem, methods have been devel-oped for islet encapsulation in thin conformal polymeric coats Materials that have been used for conformal coating include photopolymerized poly(ethylene glycol) diacrylate

(Hill et al., 1997; Cruise et al., 1999) and hydroxyethyl

methacrylate-methyl methacrylate (HEMA-MMA) (May

and Sefton, 1999; Sefton et al., 2000).

Encapsulated cell systems can also be fabricated by forming the permselective membrane in a tubular or disc-

pre-Vascular device

Microcapsules

Membrane chambers

Titanium housing

Immunobarrier membranes

Silicone rubber spacer

Immunobarrier membranes

Silicone rubber spacer

shaped confi guration, fi lling the construct with a suspension

of islets or other insulin-secreting cells in an appropriate

extracellular matrix, and then sealing the device This

approach is particularly useful when organic solvents or

other chemicals harsh to the cells are needed for the

fabrica-tion of the membranes Membrane chambers can be of

tubular or planar geometry (Fig 42.2) The cells are

sur-rounded by the semipermeable membrane and can be

implanted intraperitoneally, subcutaneously, or at other

sites Membrane materials used in fabricating these devices

include polyacrylonitrile-polyvinyl chloride (PAN-PVC)

copolymers, polypropylene, polycarbonate, cellulose nitrate,

and polyacrylonitrile-sodium methallylsulfonate (AN69)

(Lanza et al., 1992; Mikos et al., 1994; Prevost et al., 1997;

Delaunay et al., 1998; Sambanis, 2000) Typical values of

device thickness or fi ber diameter are 0.5–1 mm

Advan-tages of membrane chambers are the relative ease of

han-dling, the fl exibility with regard to the matrix in which the

cells are embedded, and retrievability after implantation A

major disadvantage is their inferior transport properties,

since the surface-to-volume ratio is smaller than that of

microcapsules and diffusional distances are longer

Constructs connected to the vasculature via an

arterio-venus shunt consist of a semipermeable tube

sur-rounded by the cell compartment (Fig 42.2) The tube

is connected to the vasculature, and transport of solutes

between the blood and the cell compartment occurs via

the pores in the tube wall A distinct advantage of the

vascular device is the improved transport of nutrients

and metabolites, which occurs by both diffusion and

con-vection However, the major surgery that is needed for

implantation and problems of blood coagulation at the

anastomosis sites have considerably reduced enthusiasm

for these devices

Other Construct Systems

A common approach for improving the oxygenation of

cells in diffusion chambers is to encourage the formation of

neovasculature around the implant This is discussed in the

following “In Vivo Implantation” section Other innovative

approaches that have been proposed include the

electro-chemical generation of oxygen in a device adjacent to a

planar immunobarrier diffusion chamber containing the

insulin-secreting cells (Wu et al., 1999); and the

coencapsu-lation of islets with algae, where the latter produce oxygen

photosynthetically upon illumination (Bloch et al., 2006)

These were in vitro studies, however, and the ability to

translate these approaches to effective in vivo confi

gura-tions remains unknown

In a different design, Cheng et al (2004, 2006)

combined constitutive insulin-secreting cells with a

glucose-responsive material in a disc-shaped construct As indicated

earlier, it is straightforward to genetically engineer

non-β-cells for constitutive insulin secretion; the challenge is

in engineering appropriate cellular responsiveness to

physiologic stimuli In this proposed device, a concanavalin

A (con A)–glycogen material, sandwiched between two ultrafi ltration membranes, acted as a control barrier to insulin release from an adjacent compartment containing the cells Specifi cally, con A–glycogen formed a gel at a low concentration of glucose, which was reversibly con-verted to sol at a high glucose concentration, as glucose displaced glycogen from the gel network Since insulin dif-fusivity is higher through the sol than through the gel, insu-lin released by the cells during low-glucose periods diffused slowly through the gel material; when switched

to high glucose, the insulin accumulated in the cell compartment during the previous cycle was released at a faster rate through the sol-state polymer Overall, this approach converted the constitutive secretion of insulin

by the cells to a glucose-responsive insulin release by the

device (Cheng et al., 2006) Again, these were in vitro studies, and the in vivo effi cacy of this approach remains to be

evaluated

Construct Design and In Vitro Evaluation

Design of three-dimensional encapsulated systems can

be signifi cantly enabled using mathematical models of solute transport through the tissue and of nutrient consumption and metabolite production by the cells Beyond the microvasculature surrounding the construct, transport of solutes occurs by diffusion, unless the construct

is placed in a fl ow environment, in which case convective transport may also occur Due to its low solubility, transport

of oxygen to the cells is the critical issue Models can be used

to evaluate the dimensions and the cell density within the construct so that all cells are suffi ciently nourished and the capsule as a whole is rapidly responsive to changes in the surrounding glucose concentration (Tziampazis and Sambanis, 1995) Experimental and modeling methods for determining transport properties and reaction kinetics have been described previously (Sambanis and Tan, 1999) Furthermore, models can be developed to account for the cellular reorganization that occurs in constructs with time

as a result of cell growth, death, and possibly migration processes Such reorganization is especially signifi cant when proliferating insulin-secreting cell lines are encapsu-

lated in hydrogel matrices (Stabler et al., 2001; Simpson

et al., 2005; Gross et al., 2007).

Pancreatic tissue substitutes should be evaluated in

vitro prior to implantation, in terms of their ability to support

the cells within over prolonged periods of time and to exhibit and maintain their overall secretory properties Long-term cultures can be performed in perfusion bioreactors under

conditions simulating aspects of the in vivo environment

In certain studies, the bioreactors and support perfusion circuits were made compatible with a nuclear magnetic resonance spectrometer This allowed measuring intracel-lular metabolites, such as nucleotide triphosphates, as a function of culture conditions and time, without the need

I I I C O N S T R U C T T E C H N O L O G Y • 625

to extract the encapsulated cells (Papas et al., 1999a, 1999b)

Such studies produce a comprehensive understanding of

the intrinsic tissue function in a well defi ned and controlled

environment prior to introducing the additional complexity

of host–implant interactions in in vivo experiments.

The secretory properties of tissue constructs can be

evaluated with low time resolution in simple static culture

experiments by changing the concentration of glucose in

the medium and measuring the secreted insulin In general,

a square wave of insulin concentration is implemented,

from basal to inducing basal conditions for insulin

secre-tion To evaluate the secretory response with a higher time

resolution, perfusion experiments need to be performed,

in which medium is fl owed around the tissue and secreted

insulin is assayed in the effl uent Again, a square wave

of glucose concentration is generally implemented By

com-paring the secretory dynamics of free and encapsulated

cells, one can ensure that the encapsulation material

does not introduce excessive time lags that might

compro-mise the secretory properties of the construct Indeed,

properly designed hydrogel microcapsules introduce only

minimal secretory time lags (Tziampazis and Sambanis,

1995; Sambanis et al., 2002).

Manufacturing Considerations

Fabrication of pancreatic substitutes of consistent

quality requires the use of cells that are also of consistent

quality Although with clonal, expandable cells this is a

rather straightforward issue, with islets isolated from human

and animal tissues there can be signifi cant variability in the

quantity and quality of the cells in the preparations With

islets from cadaveric human donors, the quality of the

iso-lates is assessed by microscopic observation, viability

stain-ing, and possibly a static insulin secretion test It is generally

recognized, however, that a quantitative, objective

assess-ment of islet quality would help improve the consistency of

the preparations and thus, possibly, the transplantation

outcome

It is conceivable that encapsulated cell systems could

be fabricated at a central location from which they are

dis-tributed to clinical facilities for implantation In this scheme,

preservation of the constructs for long-term storage,

inven-tory management, and, importantly, sterility control would

be essential Cryopreservation appears to be a promising

method for maintaining fabricated constructs for prolonged

time periods Although there have been signifi cant studies

on the cryopreservation of single cells and some tissues, the

problems pertaining to cryopreserving artifi cial tissues are

only beginning to be addressed Cryopreservation of

mac-roencapsulated systems is expected to be particularly

chal-lenging and has not been reported in the literature However,

βTC-cells encapsulated in alginate beads have been

pre-served successfully (Mukherjee et al., 2005; Song et al., 2005)

An especially promising approach involves using high

concentrations of cryoprotective agents so that water is

converted to a glassy, or vitrifi ed, state at low temperatures;

the absence of ice crystals in both the intracellular and extracellular domains appears helpful in maintaining not only cellular viability but also the structure and function of

the surrounding matrix (Mukherjee et al., 2005; Song et al.,

2005)

IV IN VIVO IMPLANTATION

This section highlights results from in vivo experiments

using the different confi gurations outlined earlier Results with encapsulated cell systems are presented fi rst Since

in vivo experiments with non-β-cells engineered for insulin

secretion are at present based mostly on in vivo gene therapy

approaches, these are described next Technologies for the

in vivo monitoring of cells and constructs and the issue of

implant retrievability are then discussed

Encapsulated Cell Systems

In vivo experiments with pancreatic substitutes are

numerous in small animals, limited in large animals, and few in humans Allogeneic and xenogeneic islets in hydrogel microcapsules implanted in diabetic mice and rats have generally restored normoglycemia for prolonged periods of

time In the early study of O’Shea et al (1984), islet allografts

encapsulated in APA membranes were implanted toneally in streptozotocin-induced diabetic rats Of the fi ve animals that received transplants, three remained normo-glycemic for more than 100 days One of these three animals remained normoglycemic 368 days postimplantation In the

intraperi-later study of Lum et al (1992), rat islets encapsulated in APA

membranes and implanted in streptozotocin diabetic mice restored normoglycemia for up to 308 days, with a mean xenograft survival time of 220 days With all recipients, nor-moglycemia was restored within two days postimplanta-tion Control animals receiving single injections of unencapsulated islets remained normoglycemic for less

than two weeks (O’Shea et al., 1984; Lum et al., 1992) More

recently, APA-encapsulated βTC6-F7 insulinomas restored

normoglycemia in diabetic rats for up to 60 days (Mamujee

et al., 1997), and APA-encapsulated βTC-tet insulinomas in

NOD mice for at least eight weeks (Black et al., 2006) In the

latter study, it was also observed that no host cell adherence occurred to microcapsules, and there were no signifi cant immune responses to the implant, with cytokine levels being similar to those of sham-operated controls These results are thus indicative of the potential use of an immu-noisolated continuous β-cell line for the treatment of diabe-

tes With the recently developed human cell line (Narushima

et al., 2005), experiments were performed with

unencapsu-lated cells transplanted into streptozotocin-induced betic severe combined immunodefi ciency mice Control of blood glucose levels started within two weeks postimplanta-tion, and mice remained normoglycemic for longer than 30

dia-weeks (Narushima et al., 2005) Besides rodents, long-term

restoration of normoglycemia with microencapsulated islets

has been demonstrated in large animals, including

sponta-neously diabetic dogs, where normoglycemia was achieved

with canine islet allografts for up to 172 days (Soon-Shiong

et al., 1992), and monkeys, where in one animal porcine islet

xenografts normalized hyperglycemia for more than 150

days (Sun et al., 1992) More recently, one of the companies

working on islet encapsulation technology announced that

primate subjects in ongoing studies have continued to

exhibit improved glycemic regulation over a six-month

period after receiving microencapsulated porcine islet

transplants (MicroIslet Inc Press Release, August 7, 2006)

In vivo results with vascular devices are reportedly

mixed Implantation of devices containing allogeneic islets

as arteriovenous shunts in pancreatectomized dogs resulted

in 20–50% of the dogs becoming normoglycemic up to 10

weeks postimplantation without exogenous insulin

admin-istration When xenogeneic bovine or porcine islets were

used, only 10% of the dogs remained normoglycemic 10

weeks postimplantation All dogs were reported diabetic or

dead after 15 weeks (Sullivan et al., 1991) Recently, a

hollow-fi ber device composed of polyethylene-vinyl alcohol hollow-fi bers

and a poly-amino-urethane-coated, nonwoven

polytetra-fl uoroethylene fabric seeded with porcine islets provided

normalization of the blood glucose levels in totally

pancre-atectomized pigs when connected to the vasculature of the

animals (Ikeda et al., 2006) It should be noted, however, that

the overall interest in vascular devices has faded, due to the

surgical and blood coagulation challenges they pose

Although several hypotheses exist, the precise cause of

the eventual in vivo failure of encapsulated cell systems

remains unclear Encapsulation does not completely prevent

the immune recognition of the implant Although direct

cel-lular recognition is prevented, antigens shed by the cells as

a result of secretion and, more importantly, lysis in the

cap-sules eventually pass through the permselective barrier and

are recognized by the antigen-presenting cells of the host

For example, in one study, antibodies against islets in a

tubular diffusion chamber were detected in plasma two to

six weeks postimplantation, suggesting that islet antigens

crossed the membrane and stimulated antibody formation

in the host (Lanza et al., 1994) In another study,

alginate-encapsulated islets were lysed in vitro by nitric oxide

pro-duced by activated macrophages (Wiegand et al., 1993)

Passage of low-molecular-weight molecules cannot be

prevented by immunoprotective membranes imposing a

molecular weight cutoff on the order of 50 kDa It should be

noted that in one human study involving encapsulated

allo-geneic islets, the patient had to be provided with low levels

of immunosuppression (Soon-Shiong et al., 1994) In a more

recent report, also with encapsulated allografts implanted

peritoneally, type 1 diabetic patients remained

nonimmu-nosuppressed but were unable to withdraw exogenous

insulin (Calafi ore et al., 2006).

Nonspecifi c infl ammation may also occur around the

implant and develop into a fi brous capsule, reducing the

oxygen available to the cells within The fi brotic layer has been found to consist of several layers of fi broblasts and collagen with polymorphonuclear leukocytes, macrophages, and lymphocytes The surface roughness of the membrane may also trigger infl ammatory responses In one study, membranes with smooth outside surfaces exhibited a minimal fi brotic reaction 10 weeks postimplantation, regardless of the type of encapsulated cells, whereas rough surfaces elicited a fi brotic response even one week postim-

plantation (Lanza et al., 1991) Use of high-purity materials

also helps to minimize infl ammatory reactions If a material

is intrinsically infl ammatory, such as poly-l-lysine, it can be coated with a layer of noninfl ammatory material, such as alginate, to minimize the host’s reaction Such coverage may not be suffi ciently permanent, though, resulting in the eventual fi brosis of the implant Indeed, several investi-gators report improved results with plain alginate beads without a poly-l-lysine layer, especially when allogeneic cells are used in the capsules

Provision of nutrients to and removal of metabolites from encapsulated cells can be especially challenging

in vivo Normal pancreatic islets are highly vascularized

and thus well oxygenated There exists evidence that unencapsulated islets injected in the portal system of the liver become revascularized, which enhances their engraft-ment and function On the other hand, encapsulation prevents revascularization, so the implanted tissue is nour-ished by diffusion alone Promotion of vascularization around the immunoprotective membrane increases the

oxygenation of the implanted islets (Prokop et al., 2001)

Interestingly, transformed cells, such as the βTC3 line of mouse insulinomas, are more tolerant of hypoxic conditions than intact islets; such cells may thus function better

than islets in implanted capsules (Papas et al., 1996)

However, with transformed cells, too, enhanced ation increases the density of functional cells that can be effectively maintained within the implant volume Vascular-ization is dependent on the microarchitecture of the mate-rial, which should have pores 0.8–8.0 µm in size, allowing

oxygen-permeation of host endothelial cells (Brauker et al., 1992,

1995) Vascularization is also enhanced by the delivery of angiogenic agents, such as FGF-2 and VEGF, possibly with

controlled-release devices (Sakurai et al., 2003) Although

vascularization can be promoted around a cell-seeded device, improved success has been reported if a cell-free device is fi rst implanted and vascularized and the cells are then introduced One example of this procedure involved placing a cylindrical stainless steel mesh in the subcutane-ous space of rats, with the islets introduced 40 days later

(Pileggi et al., 2006) Replacement of a vascularized implant

is challenging, however, due to the bleeding that occurs A solution to this problem may entail the design of a device that can be emptied and refi lled with a suspension of cells

in an extracellular matrix without disturbing the housing and the associated vascular network

I V I N V I V O I M P L A N T A T I O N • 627

Gene and Cell-Based Therapies

In vivo effi cacy studies with gene therapy and non-

β-cells genetically engineered for insulin secretion are

gener-ally limited to small animals Intraportal injection of

recombinant adenovirus expressing furin-compatible

insulin under the control of a glucose-responsive promoter

containing elements of the rat liver pyruvate kinase gene

restored near-normal glycemia in streptozotocin diabetic

rats for periods of 1–12 weeks (Thule and Liu, 2000) With

hepatic delivery of a recombinant adeno-associated virus

expressing a single-chain insulin analog under the control

of an l-type pyruvate kinase promoter, Lee and coworkers

(2000) controlled blood glucose levels in streptozotocin

dia-betic rats and NOD mice for periods longer than 20 weeks

However, transiently low blood glucose levels observed

three to fi ve hours after glucose loading indicated a

draw-back of the transcriptional regulation of insulin expression,

which may result in hypoglycemic episodes (Lee et al., 2000)

Possible approaches toward ameliorating this problem

include optimizing the number of glucose-regulatory and

insulin-sensing elements in the promoter (Jun and Yoon,

2005) and destabilizing the preproinsulin mRNA; the latter

has been shown to expedite signifi cantly the

down-regula-tion of secredown-regula-tion dynamics from transcripdown-regula-tionally controlled

cells on removal of the secretory stimulus (Tang and

Sam-banis, 2003)

In vivo gene therapy with small animals has also shown

success when the target cells for insulin expression were

intestinal endocrine K- or gastric G-cells Using a transgene

expressing human insulin under the control of the

glucose-dependent insulinotropic peptide (GIP) promoter, Cheung

et al (2000) expressed insulin specifi cally in gut K-cells of

transgenic mice, which protected them from developing

diabetes following STZ-mediated destruction of the native

β-cells Similarly, use of a tissue-specifi c promoter to express

insulin in gastric G-cells of mice resulted in insulin release

into circulation in response to meal-associated stimuli,

sug-gesting that G-cell insulin expression is benefi cial in the

amelioration of diabetes (Lu et al., 2005) Translation of

these approaches to adult animals and, eventually, humans,

requires the development of effective methods of gene

delivery to intestinal endocrine or gastric cells in vivo or the

development of effective ex vivo gene therapy approaches.

In Vivo Monitoring

Monitoring of the number and function of

insulin-secreting cells in vivo would provide valuable information

directly on the implant and possibly offer early indications

of implant failure Additionally, in animal experiments, the

ability to monitor an implant noninvasively reduces the

number of animals that are needed in the experimental

design and helps establish a critical link between

implanta-tion and endpoint physiologic effects, the latter commonly

being blood glucose levels and animal weight

Imaging techniques can provide unique insight into the

structure/function relationship of a construct in vitro and

in vivo There are several imaging modalities that have been

applied to monitor tissue-engineered constructs, including computed tomography (CT), positron emission tomogra-phy (PET), optical techniques, and nuclear magnetic reso-nance (NMR) imaging and spectroscopy Among these, NMR offers the unique advantage of providing information

on both construct integrity and function, without the need

to modify the cells genetically (e.g., through the expression

of green fl uorescent protein, used in optical methods) or the introduction of radioactive labels (e.g., PET agents)

Furthermore, since magnetic fi elds penetrate uniformly throughout the sample, NMR is ideally suited to monitor constructs implanted at deep-seated locations Its disad-vantage is its low sensitivity Whereas optical and radionu-clide techniques can detect tracer quantities, NMR detects metabolites that are available in the millimolar or, in some cases, submillimolar range

The ability to monitor noninvasively in vivo a pancreatic substitute by NMR was reported recently (Stabler et al.,

2005) Agarose disc-shaped constructs containing mouse insulinoma βTC3-cells were implanted in the peritoneal cavity of mice Construct integrity was visualized by MR imaging and the metabolic activity of the cells within by water-suppressed 1H NMR spectroscopy (Fig 42.3) Control experiments established that the total choline (TCho) reso-nance at 3.2 ppm, which is attributed to three choline metabolites, correlated positively and linearly with the number of viable cells within the construct, measured with

an independent assay To obtain the TCho signal in vivo

without interference from the surrounding host tissue, such

as peritoneal fat, the central agarose disc containing the cells had to be surrounded by cell-free agarose buffer zones This ensured that the MR signal arose only from the implanted cells, even as the construct moved due to animal breathing

A second problem that had to be resolved was that glucose diffusing into the construct produced a resonance that inter-fered with the TCho resonance at 3.2 ppm For this, a unique glucose resonance at 3.85 ppm was used to correct for the interference at 3.2 ppm so that a corrected signal, uniquely attributed to TCho, could be obtained The latter correlated positively and linearly with the number of viable cells, mea-sured with an independent assay, on the constructs postex-plantation (Fig 42.3) Hence, with the appropriate implanted construct architecture and signal processing, the number of viable cells in an implant could be followed in the same

animal as a function of time (Stabler et al., 2005).

Labeling of cells with magnetic nanoparticles, which can be detected by magnetic resonance, and genetically engineering cells so that they express a fl uorescent or lumi-nescent marker that can be optically detected are other methods being pursued to track the location and possibly

function of implanted cells in vivo It is expected that

devel-opment of robust monitoring methodologies will be helpful

not only in experimental development studies but also in

eventual clinical applications

Retrievability

The issue of construct retrievability needs to be

consid-ered for all pertinent applications Useful lifetimes of

con-structs are limited, so repeated implantation of cells will be

required It is as yet unclear whether retrieval of constructs

will be necessary at the end of their useful lives Long-term

studies on the safety challenges posed by accumulated

implants in the host need to be carried out to address this

question

V CONCLUDING REMARKS

Tight glycemic regulation in insulin-dependent

diabet-ics signifi cantly improves their overall health and reduces

the long-term complications of the disease A pancreatic

substitute holds signifi cant promise at accomplishing this

in a relatively noninvasive way However, to justify the

improved outcome, a substitute needs to be not only effi

ca-cious in terms of insulin secretion, but also

immunologi-cally acceptable A number of approaches are being pursued

to address this obstacle and additionally develop constructs that can be manufactured at a clinically relevant scale However, as the problems are being thoroughly investigated, their solutions become more challenging Encapsulation in permselective barriers improves the immune acceptance of allo- and xenografts, but it is doubtful that encapsulation will, by itself, ensure the long-term survival and function of the implant in nonimmunosuppressed hosts The develop-ment of specifi c, benign immune suppression protocols that work in concert with encapsulation appears necessary Reducing the immunogenicity of the implanted cells and modifying them so that they better withstand the encapsu-

lation and in vivo environment are appropriate strategies

To ensure that substitutes can be fabricated at the necessary scale, methods to expand pancreatic islets in culture, to produce β-cells from stem cells, or to generate expandable β-cell lines with appropriate phenotypic characteristics need to be pursued In alternative approaches involving gene therapy of non-β-cells, or the ex vivo engineering of

non-β-cells retrieved surgically from the host, the major problem is not that of cell procurement or immune accep-tance but, rather, of ensuring precise regulation of insulin

cells/mL agarose implanted in the peritoneal cavity of a mouse (A) 1

H NMR image obtained with a surface coil The inner disc, containing the cells, is distinguishable from the surrounding cell-free buffer zone, implemented to exclude spectroscopic signal from the surrounding host

tissue (B) Localized, water-suppressed 1

H NMR spectrum from the cells contained in the inner disc Resonances due to total choline (TCho), glucose, and

lactate are clearly visible The time needed to collect the spectrum was 13 min (C) Correlation between the glucose-corrected TCho resonance at 3.2 ppm

and the viable cell number obtained postexplantation using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide] assay Adapted from

Stabler et al (2005).

secretion by glucose or other physiologic stimuli This poses

a different set of problems, which, however, are equally

challenging to those of β-cell procurement and immune

acceptance Methods for the preservation of substitutes and

for the noninvasive monitoring of their integrity and

func-tionality in vivo are integral parts of construct development

and characterization with regard to construct

manufactur-ing and assessment of in vivo effi cacy, respectively.

As in many aspects of life, with challenges come

oppor-tunities It is essential that multiple approaches be pursued

in parallel, because it is currently unclear which ones will

eventually develop into viable therapeutic procedures If more than one method evolves into a clinical application, this would be welcome news, because it may allow fl exibility

in the personalization of therapy For instance, in an adult type 2 insulin-dependent diabetic, use of an encapsulated allograft with low-level immunosuppression might consti-tute an appropriate therapeutic modality In a juvenile type

1 diabetic with aggressive autoimmunity, however, use of autologous genetically engineered non-β-cells, which are not recognized by the resident autoimmunity, may consti-tute the therapeutic method of choice

VI ACKNOWLEDGMENTS

The studies in the author’s and coinvestigators’ laboratories

referenced in this chapter were supported by grants from

the Georgia Tech/Emory Center for the Engineering of Living

Tissues (GTEC), a National Science Foundation Engineering

Research Center; and by grants from the National Institutes

of Health, EmTech Bio, and the Juvenile Diabetes Research Foundation international The author also wishes to thank Indra Neil Mukherjee and Heather Bara for critically review-ing the manuscript, as well as Drs Constantinidis and Thulé for helpful discussions

VII REFERENCES

Bigam, D L., and Shapiro, A J (2004) Pancreatic transplantation:

beta-cell replacement Curr Treat Options Gastroenterol 7, 329–341.

Black, S P., Constantinidis, I., Cui, H., Tucker-Burden, C., Weber, C J.,

and Safl ey, S A (2006) Immune responses to an encapsulated

alloge-neic islet beta-cell line in diabetic NOD mice Biochem Biophys Res

Commun 340, 236–243.

Bloch, K., Assa, S., Lazard, D., Abramov, N., Shalitin, S., Weintrob, N.,

Josefsberg, Z., Rapoport, M., and Vardi, P (1999) Neonatal pig islets

induce a lower T-cell response than adult pig islets in IDDM patients

Transplantation 67, 748–752.

Bloch, K., Papismedov, E., Yavriyants, K., Vorobeychik, M., Beer, S., and

Vardi, P (2006) Photosynthetic oxygen generator for bioartifi cial

pan-creas Tissue Eng 12, 337–344.

Blyszczuk, P., Czyz, J., Kania, G., Wagner, M., Roll, U., St-Onge, L., and

Wobus, A M (2003) Expression of Pax4 in embryonic stem cells

promotes differentiation of nestin-positive progenitor and

insulin-producing cells Proc Natl Acad Sci U.S.A 100, 998–1003.

Bonner-Weir, S., and Sharma, A (2002) Pancreatic stem cells

J Pathol 197, 519–526.

Bonner-Weir, S., Taneja, M., Weir, G C., Tatarkiewicz, K., Song, K H.,

Sharma, A., and O’Neil, J J (2000) In vitro cultivation of human islets

from expanded ductal tissue Proc Natl Acad Sci U.S.A 97,

7999–8004.

Brauker, J., Martinson, L A., Hill, R S., Young, S K., Carr-Brendel, V E.,

and Johnson, R C (1992) Neovascularization of immunoisolation

membranes: the effect of membrane architecture and encapsulated

tissue Transplant Proc 24, 2924.

Brauker, J H., Carr-Brendel, V E., Martinson, L A., Crudele, J.,

Johnston, W D., and Johnson, R C (1995) Neovascularization of

syn-thetic membranes directed by membrane microarchitecture

J Biomed Mater Res 29, 1517–1524.

Calafi ore, R., Basta, G., Luca, G., Lemmi, A., Montanucci, M P.,

Calabrese, G., Racanicchi, L., Mancuso, F., and Brunetti, P (2006)

Microencapsulated pancreatic islet allografts into

nonimmunosup-pressed patients with type 1 diabetes: fi rst two cases Diabetes Care 29,

Chen, G., Hohmeier, H E., Gasa, R., Tran, V V., and Newgard, C B

(2000) Selection of insulinoma cell lines with resistance to

interleukin-1beta- and gamma-interferon-induced cytotoxicity Diabetes 49,

562–570.

Cheng, S Y., Gross, J., and Sambanis, A (2004) Hybrid pancreatic tissue substitute consisting of recombinant insulin-secreting cells and

glucose-responsive material Biotechnol Bioeng 87, 863–873.

Cheng, S Y., Constantinidis, I., and Sambanis, A (2006) Use of responsive material to regulate insulin release from constitutively

glucose-secreting cells Biotechnol Bioeng 93, 1079–1088.

Cheung, A T., Dayanandan, B., Lewis, J T., Korbutt, G S., Rajotte, R V., Bryer-Ash, M., Boylan, M O., Wolfe, M M., and Kieffer, T J (2000)

Glucose-dependent insulin release from genetically engineered K cells

Science 290, 1959–1962.

Clark, S A., Quaade, C., Constandy, H., Hansen, P., Halban, P., Ferber, S., Newgard, C B., and Normington, K (1997) Novel insulinoma cell lines produced by iterative engineering of GLUT2, glucokinase, and

human insulin expression Diabetes 46, 958–967.

Creutzfeldt, W (1974) [Clinical signifi cance of gastrointestinal

hor-mones] Verh Dtsch Ges Inn Med 80, 330–338.

Cruise, G M., Hegre, O D., Lamberti, F V., Hager, S R., Hill, R., Scharp,

D S., and Hubbell, J A (1999) In vitro and in vivo performance

of porcine islets encapsulated in interfacially photopolymerized

poly(ethylene glycol) diacrylate membranes Cell Transplant 8,

293–306.

Delaunay, C., Darquy, S., Honiger, J., Capron, F., Rouault, C., and Reach,

G (1998) Glucose-insulin kinetics of a bioartifi cial pancreas made of

an AN69 hydrogel hollow fi ber containing porcine islets and implanted

in diabetic mice Artif Organs 22, 291–299.

Drucker, D J (2002) Biological actions and therapeutic potential of the

glucagon-like peptides Gastroenterology 122, 531–544.

Efrat, S (1998) Cell-based therapy for insulin-dependent diabetes

mel-litus Eur J Endocrinol 138, 129–133.

Efrat, S., Linde, S., Kofod, H., Spector, D., Delannoy, M., Grant, S.,

Hanahan, D., and Baekkeskov, S (1988) Beta-cell lines derived from

transgenic mice expressing a hybrid insulin gene-oncogene Proc Natl

Acad Sci U.S.A 85, 9037–9041.

Efrat, S., Leiser, M., Surana, M., Tal, M., Fusco-Demane, D., and Fleischer,

N (1993) Murine insulinoma cell line with normal glucose-regulated

insulin secretion Diabetes 42, 901–907.

Efrat, S., Fusco-DeMane, D., Lemberg, H., al Emran, O., and Wang, X

(1995) Conditional transformation of a pancreatic beta-cell line derived

from transgenic mice expressing a tetracycline-regulated oncogene

Proc Natl Acad Sci U.S.A 92, 3576–3580.

Finegood, D T., Scaglia, L., and Bonner-Weir, S (1995) Dynamics of

beta-cell mass in the growing rat pancreas Estimation with a simple

mathematical model Diabetes 44, 249–256.

Gross, J D., Constantinidis, I., and Sambanis, A (2007) Modeling of

encapsulated cell systems J Theor Biol 244, 500–510.

Heald, K A., Carless N., Jay, T R., Boucher, N., and Downing, R (1999)

Expression of the GALalpha(1–3)GAL epitope on pig islets J Mol Med

77, 169–171.

Hill, R S., Cruise, G M., Hager, S R., Lamberti, F V., Yu, X., Garufi s,

C L., Yu, Y., Mundwiler, K E., Cole, J F., Hubbell, J A., Hegre, O D.,

and Scharp, D W (1997) Immunoisolation of adult porcine islets for

the treatment of diabetes mellitus The use of photopolymerizable

polyethylene glycol in the conformal coating of mass-isolated porcine

islets Ann N.Y Acad Sci 831, 332–343.

Hughes, S D., Johnson, J H., Quaade, C., and Newgard, C B (1992)

Engineering of glucose-stimulated insulin secretion and biosynthesis

in non-islet cells Proc Natl Acad Sci U.S.A 89, 688–692.

Hughes, S D., Quaade, C., Johnson, J H., Ferber, S., and Newgard, C B

(1993) Transfection of AtT-20ins cells with GLUT-2 but not GLUT-1

confers glucose-stimulated insulin secretion Relationship to glucose

metabolism J Biol Chem 268, 15205–15212.

Ikeda, H., Kobayashi, N., Tanaka, Y., Nakaji, S., Yong, C., Okitsu, T.,

Oshita, M., Matsumoto, S., Noguchi, H., Narushima, M., Tanaka, K.,

Miki, A., Rivas-Carrillo, J D., Soto-Gutierrez, A., Navarro-Alvarez, N.,

Tanaka, K., Jun, H S., Tanaka, N., and Yoon, J W (2006) A newly

developed bioartifi cial pancreas successfully controls blood glucose

in totally pancreatectomized diabetic pigs Tissue Eng 12, 1799–1809.

Jun, H S., and Yoon, J W (2005) Approaches for the cure of type 1

dia-betes by cellular and gene therapy Curr Gene Ther 5, 249–262.

Kieffer, T J., and Habener, J F (1999) The glucagon-like peptides

Endocr Rev 20, 876–913.

Knaack, D., Fiore, D M., Surana, M., Leiser, M., Laurance, M.,

Fusco-DeMane, D., Hegre, O D., Fleischer, N., and Efrat, S (1994) Clonal

insulinoma cell line that stably maintains correct glucose

responsive-ness Diabetes 43, 1413–1417.

Kojima, H., Fujimiya, M., Matsumura, K., Younan, P., Imaeda, H., Maeda,

M., and Chan, L (2003) NeuroD-betacellulin gene therapy induces islet

neogenesis in the liver and reverses diabetes in mice Nat Med 9,

596–603.

Lannoy, V J., Decaux, J F., Pierreux, C E., Lemaigre, F P., and Rousseau,

G G (2002) Liver glucokinase gene expression is controlled by the

onecut transcription factor hepatocyte nuclear factor-6 Diabetologia

45, 1136–1141.

Lanza, R P., Butler, D H., Borland, K M., Staruk, J E., Faustman, D L.,

Solomon, B A., Muller, T E., Rupp, R G., Maki, T., Monaco, A P., et al

(1991) Xenotransplantation of canine, bovine, and porcine islets in

diabetic rats without immunosuppression Proc Natl Acad Sci U.S.A

88, 11100–11104.

Lanza, R P., Sullivan, S J., and Chick, W L (1992) Perspectives in

dia-betes Islet transplantation with immunoisolation Diabetes 41,

1503–1510.

Lanza, R P., Beyer, A M., and Chick, W L (1994) Xenogenic humoral responses to islets transplanted in biohybrid diffusion chambers

Transplantation 57, 1371–1375.

Lechner, A., Nolan, A L., Blacken, R A., and Habener, J F (2005)

Redif-ferentiation of insulin-secreting cells after in vitro expansion of adult

human pancreatic islet tissue Biochem Biophys Res Commun 327,

581–588.

Lee, H C., Kim, S J., Kim, K S., Shin, H C., and Yoon, J W (2000) Remission in models of type 1 diabetes by gene therapy using a single-

chain insulin analogue Nature 408, 483–488.

Lu, Y.-C., Sternini, C., Rozengurt, E., and Zhokova, E (2005) Release of transgenic human insulin from gastric G cells: a novel approach for the

amelioration of diabetes Endocrinology 146, 2610–2619.

Lum, Z P., Krestow, M., Tai, I T., Vacek, I., and Sun, A M (1992) grafts of rat islets into diabetic mice An evaluation of new smaller

Xeno-capsules Transplantation 53, 1180–1183.

Lumelsky, N., Blondel, O., Laeng, P., Velasco, I., Ravin, R., and McKay, R (2001) Differentiation of embryonic stem cells to insulin-secreting

structures similar to pancreatic islets Science 292, 1389–1394.

Lundberg, P., and Kuchel, P W (1997) Diffusion of solutes in agarose and alginate gels: 1H and 23Na PFGSE and 23Na TQF NMR

studies Magn Reson Med 37, 44–52.

Mamujee, S N., Zhou, D., Wheeler, M B., Vacek, I., and Sun, A M (1997) Evaluation of immunoisolated insulin-secreting beta TC6-F7

cells as a bioartifi cial pancreas Ann Transplant 2, 27–32.

May, M H., and Sefton, M V (1999) Conformal coating of small

parti-cles and cell aggregates at a liquid–liquid interface Ann N.Y Acad Sci

875, 126–134.

Mikos, A., Papadaki, M., Kouvroukoglou, S., Ishaug, S., and Thomson,

R (1994) Mini-review: islet transplantation to create a bioartifi cial

pan-creas Biotechnol Bioeng 43, 673–677.

Miyazaki, J., Araki, K., Yamato, E., Ikegami, H., Asano, T., Shibasaki, Y., Oka, Y., and Yamamura, K (1990) Establishment of a pancreatic beta cell line that retains glucose-inducible insulin secretion: special refer-

ence to expression of glucose transporter isoforms Endocrinology 127,

Nerem, R M., and Sambanis, A (1995) Tissue engineering: from biology

to biological substitutes Tissue Eng 1, 3–13.

V I I R E F E R E N C E S • 631

O’Shea, G M., Goosen, M F., and Sun, A M (1984) Prolonged survival

of transplanted islets of Langerhans encapsulated in a biocompatible

membrane Biochim Biophys Acta 804, 133–136.

Olson, D E., Paveglio, S A., Huey, P U., Porter, M H., and Thule, P M

(2003) Glucose-responsive hepatic insulin gene therapy of

spontane-ously diabetic BB/Wor rats Hum Gene Ther 14, 1401–1413.

Ouziel-Yahalom, L., Zalzman, M., Anker-Kitai, L., Knoller, S., Bar, Y.,

Glandt, M., Herold, K., and Efrat, S (2006) Expansion and

redifferentia-tion of adult human pancreatic islet cells Biochem Biophys Res

Commun 341, 291–298.

Papas, K K., Long, R C Jr., Constantinidis, I., and Sambanis, A (1996)

Effects of oxygen on metabolic and secretory activities of beta TC3 cells

Biochim Biophys Acta 1291, 163–166.

Papas, K K., Long, R C., Jr., Sambanis, A., and Constantinidis, I (1999a)

Development of a bioartifi cial pancreas: I Long-term propagation and

basal and induced secretion from entrapped betaTC3 cell cultures

Bio-technol Bioeng 66, 219–230.

Papas, K K., Long, R C., Jr., Sambanis, A., and Constantinidis, I (1999b)

Development of a bioartifi cial pancreas: II Effects of oxygen on

long-term entrapped betaTC3 cell cultures Biotechnol Bioeng 66,

231–237.

Pileggi, A., Molano, R D., Ricordi, C., Zahr, E., Collins, J., Valdes, R., and

Inverardi, L (2006) Reversal of diabetes by pancreatic islet

transplanta-tion into a subcutaneous, neovascularized device Transplantatransplanta-tion 81,

1318–1324.

Prevost, P., Flori, S., Collier, C., Muscat, E., and Rolland, E (1997)

Appli-cation of AN69 hydrogel to islet encapsulation Evaluation in

strepto-zotocin-induced diabetic rat model Ann N.Y Acad Sci 831, 344–349.

Prokop, A (2001) Bioartifi cial pancreas: materials, devices, function,

and limitations Diabetes Technol Ther 3, 431–449.

Prokop, A., Kozlov, E., Nun Non, S., Dikov, M M., Sephel, G C., Whitsitt,

J S., and Davidson, J M (2001) Towards retrievable vascularized

bio-artifi cial pancreas: induction and long-lasting stability of polymeric

mesh implant vascularized with the help of acidic and basic fi broblast

growth factors and hydrogel coating Diabetes Technol Ther 3,

245–261.

Rajagopal, J., Anderson, W J., Kume, S., Martinez, O I., and Melton,

D A (2003) Insulin staining of ES cell progeny from insulin uptake

Science 299, 363.

Ramiya, V K., Maraist, M., Arfors, K E., Schatz, D A., Peck, A B., and

Cornelius, J G (2000) Reversal of insulin-dependent diabetes using

islets generated in vitro from pancreatic stem cells Nat Med 6,

278–282.

Rayat, G R., Rajotte, R V., Elliott, J F., and Korbutt, G S (1998)

Expres-sion of Gal alpha(1,3)gal on neonatal porcine islet beta-cells and

sus-ceptibility to human antibody/complement lysis Diabetes 47, 1406–

1411.

Sakurai, T., Satake, A., Nagata, N., Gu, Y., Hiura, A., Doo-Hoon, K., Hori,

H., Tabata, Y., Sumi, S., and Inoue, K (2003) The development of new

immunoisolatory devices possessing the ability to induce

neovascular-ization Cell Transplant 12, 527–535.

Sambanis, A (2000) Engineering challenges in the development of an

encapsulated cell system for treatment of type 1 diabetes Diabetes

Technol Therap 2, 81–89.

Sambanis, A., and Tan, S A (1999) Quantitative modeling of limitations

caused by diffusion In “Methods in Molecular Medicine, Vol 18: Tissue

Engineering Methods and Protocols” (J R Morgan and M L Yarmush,

eds.) Humana Press, Totowa, NJ.

Sambanis, A., Stephanopoulos, G., Sinskey, A J., and Lodish, H F

(1990) Use of regulated secretion in protein production from animal

cells: an evaluation with the AtT-20 model cell line Biotechnol Bioeng

35, 771–780.

Sambanis, A., Tang, S.-C., Cheng, S.-Y., Stabler, C L., Long, R C J., and Constantinidis, I (2002) Core technologies in tissue engineering and

their application to the bioartifi cial pancreas In “Tissue Engineering

for Therapeutic Use” (Y Ikada, Y Umakoshi and T Hotta, eds.),

pp 5–18 Elsevier, Boston.

Schirra, J., Katschinski, M., Weidmann, C., Schafer, T., Wank, U., Arnold, R., and Goke, B (1996) Gastric emptying and release of incretin hor-

mones after glucose ingestion in humans J Clin Invest 97, 92–103.

Sefton, M., and Kharlip, L (1994) Insulin release from rat pancreatic

islets microencapsulated in a HEMA-MMA polyacrylate In “Pancreatic

Islet Transplantation Volume III: Immunoisolation of Pancreatic Islets”

(R Lanza and W Chick, eds.) RG Landes, Georgetown, TX.

Sefton, M V., May, M H., Lahooti, S., and Babensee, J E (2000) Making microencapsulation work: conformal coating, immobilization gels and

in vivo performance J Controlled Release 65, 173–186.

Shapiro, A M., Ryan, E A., and Lakey, J R (2001a) Clinical islet

trans-plant — state of the art Transtrans-plant Proc 33, 3502–3503.

Shapiro, A M., Ryan, E A., and Lakey, J R (2001b) Diabetes Islet cell

transplantation Lancet 358 (Suppl), S21.

Shapiro, A M., Ryan, E A., and Lakey, J R (2001c) Pancreatic islet

transplantation in the treatment of diabetes mellitus Best Pract Res

Clin Endocrinol Metab 15, 241–264.

Simpson, N E., Khokhlova, N., Oca-Cossio, J A., McFarlane, S S., Simpson, C P., and Constantinidis, I (2005) Effects of growth regula- tion on conditionally transformed alginate-entrapped insulin secreting

cell lines in vitro Biomaterials 26, 4633–4641.

Song, Y C., Chen, Z Z., Mukherjee, N., Lightfoot, F G., Taylor, M J., Brockbank, K G., and Sambanis, A (2005) Vitrifi cation of tissue-

engineered pancreatic substitute Transplant Proc 37, 253–255.

Soon-Shiong, P., Feldman, E., Nelson, R., Heintz, R., Merideth, N., Sandford, P., Zheng, T., and Komtebedde, J (1992) Long-term reversal

of diabetes in the large animal model by encapsulated islet

transplanta-tion Transplant Proc 24, 2946–2947.

Soon-Shiong, P., Heintz, R E., Merideth, N., Yao, Q X., Yao, Z., Zheng,

T., Murphy, M., Moloney, M K., Schmehl, M., Harris, M., et al (1994)

Insulin independence in a type 1 diabetic patient after encapsulated

islet transplantation Lancet 343, 950–951.

Soria, B., Roche, E., Berna, G., Leon-Quinto, T., Reig, J A., and Martin,

F (2000) Insulin-secreting cells derived from embryonic stem cells

nor-malize glycemia in streptozotocin-induced diabetic mice Diabetes 49,

157–162.

Stabler, C., Wilks, K., Sambanis, A., and Constantinidis, I (2001) The

effects of alginate composition on encapsulated betaTC3 cells

Bio-materials 22, 1301–1310.

Stabler, C L., Long, R C., Jr., Constantinidis, I., and Sambanis, A (2005)

In vivo noninvasive monitoring of a tissue-engineered construct using

1H-NMR spectroscopy Cell Transplant 14, 139–149.

Sullivan, S J., Maki, T., Borland, K M., Mahoney, M D., Solomon, B A., Muller, T E., Monaco, A P., and Chick, W L (1991) Biohybrid artifi cial pancreas: long-term implantation studies in diabetic, pancreatecto-

mized dogs Science 252, 718–721.

Sun, A M., Vacek, I., Sun, Y L., Ma, X., and Zhou, D (1992) In vitro and

in vivo evaluation of microencapsulated porcine islets Asaio J 38,

125–127.

Tang, S.-C., and Sambanis, A (2003a) Preproinsulin mRNA engineering

and its application to the regulation of insulin secretion from human

hepatomas FEBS Lett 537, 193–197.

Tang, S C., and Sambanis, A (2003b) Development of genetically

engi-neered human intestinal cells for regulated insulin secretion using

rAAV-mediated gene transfer Biochem Biophys Res Commun 303,

645–652.

Tang, S C., and Sambanis, A (2004) Differential rAAV2 transduction

effi ciencies and insulin secretion profi les in pure and coculture models

of human enteroendocrine L-cells and enterocytes J Gene Med 6,

1003–1013.

Thule, P M., and Liu, J M (2000) Regulated hepatic insulin gene

therapy of STZ-diabetic rats Gene Ther 7, 1744–1752.

Thule, P M., Liu, J., and Phillips, L S (2000) Glucose-regulated

produc-tion of human insulin in rat hepatocytes Gene Ther 7, 205–214.

Thule, P M., Campbell, A G., Kleinhenz, D J., Olson, D E.,

Boutwell, J J., Sutliff, R L., and Hart, C M (2006) Hepatic insulin gene

therapy prevents deterioration of vascular function and improves

adi-pocytokine profi le in STZ-diabetic rats Am J Physiol Endocrinol

Metab 290, E114–E122.

Todorov, I., Omori, K., Pascual, M., Rawson, J., Nair, I., Valiente, L.,

Vuong, T., Matsuda, T., Orr, C., Ferreri, K., Smith, C V., Kandeel, F., and

Mullen, Y (2006) Generation of human islets through expansion and

differentiation of non-islet pancreatic cells discarded (pancreatic

discard) after islet isolation Pancreas 32, 130–138.

Tziampazis, E., and Sambanis, A (1995) Tissue engineering of a artifi cial pancreas: modeling the cell environment and device func -

bio-tion Biotechnol Prog 11, 115–126.

Wang, S., Liu, J., Li, L., and Wice, B M (2004) Individual subtypes of enteroendocrine cells in the mouse small intestine exhibit unique pat-

terns of inositol 1,4,5-trisphosphate receptor expression J Histochem

Cytochem 52, 53–63.

Wiegand, F., Kroncke, K D., and Kolb-Bachofen, V (1993) phage-generated nitric oxide as cytotoxic factor in destruction of alginate-encapsulated islets Protection by arginine analogs and/or

Macro-coencapsulated erythrocytes Transplantation 56, 1206–1212.

Wu, H., Avgoustiniatos, E S., Swette, L., Bonner-Weir, S., Weir, G C., and

Colton, C K (1999) In situ electrochemical oxygen generation with an

immunoisolation device Ann N.Y Acad Sci 875, 105–125.

Yanagita, M., Nakayama, K., and Takeuchi, T (1992) Processing of mutated proinsulin with tetrabasic cleavage sites to bioactive insulin in

the nonendocrine cell line, COS-7 FEBS Lett 311, 55–59.

Yanagita, M., Hoshino, H., Nakayama, K., and Takeuchi, T (1993) cessing of mutated proinsulin with tetrabasic cleavage sites to mature

Pro-insulin refl ects the expression of furin in nonendocrine cell lines

Endo-crinology 133, 639–644.

V I I R E F E R E N C E S • 633

I Introduction

II Engineering to Generate Insulin-Producing

Cells

III Engineering to Improve Islet Survival

IV Vectors for Engineering Islets and Beta-Cells

V Conclusion

VI References

Engineering Pancreatic Beta-Cells

Hee-Sook Jun and Ji-Won Yoon

Principles of Tissue Engineering, 3 rd Edition

ed by Lanza, Langer, and Vacanti

Copyright © 2007, Elsevier, Inc.

All rights reserved.

I INTRODUCTION

The use of islet transplantation as a treatment for

dia-betes has been hampered by the limited availability of

human islets; therefore, new sources of insulin-producing

cells are needed Expansion of beta-cells by the generation

of reversibly immortalized beta-cells and creation of

insulin-producing cells by exogenous expression of insulin in

non-beta-cells have been investigated as new sources of

beta-cells Recently, embryonic and adult stem cells or

pan-creatic progenitor cells have been engineered to

differenti-ate into insulin-producing cells, demonstrating the possible

use of these cells for betacell replacement Despite signifi

-cant progress, further studies are needed to generate truly

functional insulin-producing cells In addition, the

engi-neering of beta-cells to protect them from immune attack

and to improve viability has been tried Although the

useful-ness of engineered beta-cells has yet to be clinically proven,

studies utilizing different engineering strategies and careful

analysis of the resulting insulin-producing cells may offer

potential methods to cure diabetes

Diabetes mellitus is a metabolic disease characterized

by uncontrolled hyperglycemia, which results in long-term

clinical problems, including retinopathy, neuropathy,

nephropathy, and heart disease Diabetes affects over 150

million people worldwide and is considered an epidemic of

the 21st century Blood glucose homeostasis is controlled by

endocrine beta-cells, located in the islets of Langerhans in

the pancreas When the concentration of blood glucose rises after a meal, insulin is produced and released from beta-cells Insulin then induces glucose uptake by cells in the body and converts glucose to glycogen in the liver When blood glucose concentration becomes low, glycogen is broken down to glucose in the liver and glucose is released into the blood

There are two major forms of diabetes: type 1 diabetes, also known as insulin-dependent diabetes mellitus, and type 2 diabetes, also known as non-insulin-dependent dia-betes mellitus Both types are thought to result from a reduc-tion in the number of insulin-producing beta-cells and defi cits in beta-cell function In type 1 diabetes, beta-cells are destroyed by autoimmune responses, resulting in a lack

of insulin (reviewed in Adorini et al., 2002; Yoon and Jun,

2005) In type 2 diabetes, both inadequate beta-cell function and insulin resistance of peripheral tissues contribute to the development of hyperglycemia, leading to eventual reduc-tion in the number of beta-cells (reviewed in LeRoith, 2002) Intensive exogenous insulin therapy has been used for the treatment of type 1 diabetes, but it does not restore the tight control of blood glucose levels or completely prevent the development of complications In addition, multiple daily injections are cumbersome and sometimes cause poten-tially life-threatening hypoglycemia Islet transplantation has been considered an alternative and safe method for the

treatment of diabetes (reviewed in Hatipoglu et al., 2005)

636 C H A P T E R F O R T Y - T H R E E • E N G I N E E R I N G P A N C R E A T I C B E T A - C E L L S

With the improvement of islet isolation techniques, the

success rate for independence from exogenous insulin is

increasing However, the lack of suffi cient islets to meet the

demands of patients and the side effects of

immunosup-pressive drugs that are required to prevent alloimmune and

autoimmune attack against islet grafts are the major

limita-tions of islet transplantation Therefore, various alternative

sources of insulin-producing cells are being investigated to

provide a suffi cient supply for the treatment of type 1

diabetes

In this chapter, we discuss the use of cell engineering to

produce and expand insulin-producing beta-cells; to create

insulin-producing cells from non-beta-cells, embryonic

stem cells, and adult stem cells; and to improve islet graft

survival Due to the publisher’s restrictions, we are unable

to cite all the references for primary data

II ENGINEERING TO GENERATE

INSULIN-PRODUCING CELLS

Engineering Pancreatic Beta-Cells

The pancreas is composed of endocrine and exocrine

tissues The endocrine pancreas occupies less than 5% of

the pancreatic tissue mass and is composed of cell clusters

called the islets of Langerhans The islets of Langerhans

contain insulin-producing beta-cells (about 80% of cells in

the islets), glucagon-producing alpha-cells,

somatostatin-producing delta-cells, and pancreatic

polypeptide-produc-ing cells The exocrine pancreas occupies more than

95% of the pancreas and is composed of ascinar and

ductal cells, which produce digestive enzymes The beta-cell

mass is dynamic and increases in response to

environmen-tal changes such as pancreatic injury and physiological

changes such as insulin resistance In addition, mature

beta-cells can replicate throughout life, although at a low

level

One approach to produce beta-cells for replacement

therapy is to expand mature beta-cells in vitro However,

because mature beta-cells have limited proliferative

capac-ity in culture, the expression of oncogenes has been tried as

a method to establish beta-cell lines The expression of

simian virus (SV) 40 large T antigen in beta-cells under the

control of the tet-on and tet-off regulatory system in

trans-genic mice resulted in a stable beta-cell line that could be

expanded in vitro These cells produced less insulin in the

transformed state when T antigen was expressed, but insulin

production increased after growth was arrested by cessation

of T antigen expression, and insulin secretion was regulated

as in normal mouse islets When these cells were

trans-planted into streptozotocin-induced diabetic mice, the mice

became normoglycemic, and normoglycemia was

main-tained for a prolonged time, without any treatment to

prevent oncogene expression (Milo-Landesman et al., 2001)

In addition to beta-cell expansion, cell engineering has

been used to improve beta-cell function Rat insulinoma

cells that showed decreased glucose-responsive insulin secretion were transfected with a plasmid encoding a mutated form of GLP-1 that is resistant to the degrading enzyme dipeptidyl-peptidase IV These engineered cells had increased insulin secretion in response to glucose, as com-

pared with untransfected control cells (Islam et al., 2005).

Expansion of human primary pancreatic islet cells has also been tried Primary adult islet cells could be stimulated

to divide when grown on an extracellular matrix in the ence of hepatocyte growth factor/scatter factor, but growth was arrested after 10–15 cell divisions, due to cellular senes-

pres-cence (Beattie et al., 1999) Transformation of adult human

pancreatic islets with a retroviral vector expressing SV40 large T antigen and H-rasVal 12 oncogenes resulted in extended life span, but eventually the cells entered a crisis phase fol-lowed by altered morphology, lack of proliferation, and cell death, suggesting that immortalization of human beta-cells

is more diffi cult than that of rodent beta-cells However, introduction of human telomerase reverse transcriptase (hTERT) resulted in successful immortalization (Halvorsen

et al., 1999), because human cells do not express telomerase

This immortalized cell line, βlox5, initially expressed low levels of insulin, but insulin production subsequently fell to undetectable levels as a result of the loss of expression of key insulin gene transcription factors A combination of the introduction of a beta-cell transcription factor (Pdx-1), treatment with exendin-4 (a glucagon-like peptide-1 [GLP-1] homolog), and cell–cell contact was required to recover beta-cell differentiated function and glucose-responsive

insulin production (de la Tour et al., 2001) However, Pdx-1

expression in this cell line resulted in a signifi cant decrease

in the growth rate of the cells When streptozotocin-induced diabetic animals were transplanted with these Pdx-1-expressing cells, substantial levels of circulating human C-peptide were detected and diabetes was remitted However, 10% of the animals developed tumors, even though the oncogenes and hTERT gene had been fl oxed by loxP sites so that they could be deleted by expression of Cre recombi-nase This suggests that the Cre-expressing adenovirus and/

or Cre-mediated deletion of the oncogenes was ineffi cient

(de la Tour et al., 2001).

The limitations of previously engineered beta-cell lines point to a need for a human beta-cell line that is func-tionally equivalent to primary beta-cells, can be expanded indefi nitely, and can be rendered nontumorigenic In another approach to establish a reversibly immortalized human beta-cell line, human islets were transduced with a combination of retroviral vectors expressing SV40 T antigen, hTERT, and enhanced green fl uorescent protein to immor-talize and mark terminally differentiated pancreatic beta-cells These genes were fl oxed by loxP sites to allow excision

of the immortalizing genes Among 271 clones screened for tumorigenicity, 253 clones were selected for further study, and only one of these (NAKT-15) expressed insulin and the necessary beta-cell transcription factors, such as Isl-1, Pax-

6, Nkx6.1, Pdx-1, prohormone convertases, and secretory

granule proteins Addition of factors that enhance insulin

expression and secretion during culture of the beta-cell line,

such as troglitazone, a peroxisome proliferator-activated

receptor-γ activator, and nicotinamide, helped to maintain

the function of beta-cells, and culture of these cells on

Matrigel matrix facilitated aggregate formation Removal of

the immortalizing genes by Cre recombinase expression

stopped cell proliferation and increased the expression of

beta-cell-specifi c transcription factors, resulting in

rever-sion of the cells These reverted NAKT-15 cells were

func-tionally similar to normal human islets with respect to

insulin secretion in response to glucose and nonglucose

secretagogues, although the insulin content and amount of

secreted insulin were lower than for human islets However,

NAKT-15 cells were able to remit diabetes and clear

exoge-nous glucose when transplanted into diabetic severe

com-bined immunodefi ciency (SCID) mice The insulin content

of these cells was higher in vivo than in vitro, suggesting that

the microenvironment may enhance cellular differentiation

(Narushima et al., 2005).

For clinical application of reversibly immortalized

human beta-cells, safety issues, particularly tumorigenicity,

should be considered Reducing or eliminating

tumorige-nicity may be possible by using multiple selection

proce-dures In the case of NAKT-15 cells, nontumorigenic clones

were fi rst selected by screening for tumor formation in SCID

mice After infection of Cexpressing adenovirus to

re-move the SV40 T antigen and hTERT, SV40T-negative cells

were selected in the presence of a neomycin analog (the

neomycin-resistance gene was positioned to be expressed

after the loxP-fl anked genes were deleted), and

hTERT-neg-ative cells were selected by purifi cation of enhanced green

fl uorescent protein-negative cells Finally,

SV40T/hTERT-negative cells were selected by the addition of ganciclover,

because the cells had been transduced with a suicide gene,

herpes simplex thymidine kinase, which renders them

susceptible to ganciclover These multiple selection

pro-cedures resulted in no tumor development in SCID mice

when reverted NAKT-15 cells were transplanted (Narushima

et al., 2005), although the possibility of tumorigenesis could

not be completely eliminated Nevertheless, there are

advan-tages of reversibly immortalized human beta-cells as

com-pared with primary beta-cells They can be easily expanded

to obtain suffi cient cells for transplantation and genetically

manipulated in vitro prior to transplantation, for example,

to confer resistance to immune attack

Establishment of insulin-producing beta-cell lines by

reversible immortalization of primary islets is a promising

approach for replacing insulin injections, for a beta-cell line

can provide an abundant source of beta-cells for

transplan-tation In addition, beta-cell lines can be genetically

mani-pulated to improve their function and survival However, the

functionality of the cell lines and safety issues remain to be

further studied

Engineering Surrogate Beta-Cells

Non-beta-cells that are genetically engineered to produce insulin may have an advantage over intact islets or engineered beta-cells for transplantation therapy, because non-beta-cells should not be recognized by beta-cell-specifi c autoimmune responses Pancreatic beta-cells have unique characteristics specifi c to the production of insulin, such as specifi c peptidases, glucose-sensing systems, and secretory granules that can release insulin promptly by exocytosis in response to extracellular glucose levels There-fore, the ideal target cell to engineer for insulin production would be non-beta-cells possessing similar characteristics

A variety of cell types, including fi broblasts, hepatocytes, neuroendocrine cells, and muscle cells, have been engi-neered to produce insulin, with varying degrees of success

(reviewed in Xu et al., 2003; Yoon and Jun, 2002).

Neuroendocrine cells have received considerable tion because they have characteristics similar to those of beta-cells and contain components of the regulated secre-tory pathway, including prohormone convertases 2 and 3 and secretory granules A mouse corticotrophic cell line derived from the anterior pituitary, AtT20, expressed active insulin after transfection with the insulin gene under the control of a viral or metallothionein promoter but lacked glucose responsiveness Cotransfection of genes encoding glucose transporter (GLUT)2 and glucokinase conferred glucose-responsive insulin secretion in insulin-expressing AtT20 cells Transgenic expression of insulin in the interme-diate lobe of the pituitary of nonobese diabetic (NOD) mice under the control of the pro-opiomelanocortin promoter resulted in the production of biologically active insulin Transplantation of this insulin-producing pituitary tissue into diabetic NOD mice restored normoglycemia, but insulin secretion was not properly regulated by glucose Engineer-ing primary rat pituitary cells to coexpress GLP-1 receptor and human insulin resulted in GLP-1-induced insulin secre-

atten-tion (Wu et al., 2003).

Intestinal K-cells have been explored as possible gate beta-cells K-cells are endocrine cells located in the gut that secrete the hormone glucose-dependent insulinotropic polypeptide (GIP), which facilitates insulin release after a meal K-cells are also glucose responsive, have exocytotic mechanisms, and contain the necessary enzymes for pro-cessing proinsulin to insulin A murine intestinal cell line containing K-cells transfected with human insulin DNA cloned under the control of the GIP promoter produced bio-logically active insulin in response to glucose, and transgenic mice expressing the human proinsulin gene under the GIP promoter were protected from diabetes after treatment with

surro-streptozotocin (Cheung et al., 2000) These results suggest

that K-cells may have great potential as surrogate beta-cells.The strategy of engineering hepatocytes to produce

insulin has been widely studied (Nett et al., 2003)

Hepato-cytes have advantages for engineering as insulin-producing

638 C H A P T E R F O R T Y - T H R E E • E N G I N E E R I N G P A N C R E A T I C B E T A - C E L L S

cells because they express components of a glucose-sensing

system somewhat similar to that in pancreatic beta-cells,

such as GLUT2 and glucokinase In addition, there are

several hepatocyte-specifi c gene promoters that respond to

changes in glucose concentrations The L-type pyruvate

kinase promoter (LPK) and Spot14 promoter have been

investigated as regulatory elements for glucose-responsive

insulin production in liver Using a chimeric promoter

com-posed of three copies of the stimulatory glucose-responsive

element from the LPK promoter and an inhibitory

respon-sive element from the insulin-like growth factor–binding

protein-1 basal promoter, the expression of a modifi ed

human proinsulin gene was stimulated by glucose and

inhibited by insulin in hepatocytes Engineering of rat

hepa-toma cells to express insulin under the control of the

glucose-6-phosphatase promoter resulted in the stimulation of

insulin production by glucose and self-limitation by insulin

However, insulin expression by the glucose-6-phosphatase

promoter was low because of negative feedback by the

pro-duced insulin It was recently reported that human

hepa-toma cells transduced with a furin-cleavable human

preproinsulin gene under the control of the GLUT2

pro-moter expressed insulin in response to glucose (Burkhardt

et al., 2005).

A drawback for the regulation of insulin production

by glucose-responsive promoters in hepatocytes is slow as

compared with the rapid release by exocytosis from

beta-cells Because a longer period of time is required for

transcriptional regulation to change the plasma levels of

insulin in response to changes in blood glucose,

hypoglyce-mia may occur Therefore, the development of systems that

mimic insulin secretory dynamics is required Strategies

that utilize synthetic promoters composed of multiple

copies of glucose-responsive elements for the induction of

high levels of insulin expression, insulin-sensitive elements

for feedback regulation, and methods to control of the

half-life of insulin mRNA so that it rapidly degrades may make

it possible to mimic insulin production in a

glucose-responsive fashion in non-beta-cells

Another consideration is that most non-beta-cells do

not have the appropriate endoproteases to convert

proinsu-lin to insuproinsu-lin or secretory granules from which insuproinsu-lin can

be rapidly released in response to physiological stimuli One

approach is the mutation of the proinsulin gene so that it

can be cleaved and converted to insulin by the protease

furin, which is expressed in a wide variety of cells Another

approach is the development of a single-chain insulin

analog, which shows insulin activity without the

require-ment for processing Artifi cially regulated insulin secretion

in non-beta-cells has been tried by expressing insulin as a

fusion protein containing an aggregation domain, which

accumulates in the endoplasmic reticulum and is secreted

when a drug that induces disaggregation is administered

Although engineering somatic non-beta-cells to

pro-duce insulin is a very attractive method, no method has yet

succeeded in imitating normal beta-cells regarding the rapid and tight regulation of glucose within a narrow physi-ological range Improvements, including better control of glucose-responsive transcription of transgenic insulin mRNA and artifi cial secretory systems, provide hope for the potential use of insulin-producing non-beta-cells to cure diabetes

Engineering Stem and Progenitor Cells

An exciting advance in the last few years is the ment of cell therapy strategies using stem cells Stem cells are characterized by the ability to proliferate extensively and differentiate into one or more specialized cell types Both embryonic and adult stem cells have been investigated as alternative sources for the generation of insulin-producing pancreatic islets Although spontaneous differentiation of beta-cells from stem cells can be observed, engineering of stem cells for forced expression of key beta-cell or endocrine differentiation factors should be more effi cient for driving beta-cell differentiation

develop-Engineering Embryonic Stem Cells

In principle, embryonic stem (ES) cells have the tial to generate unlimited quantities of insulin-producing cells ES cells can be expanded indefi nitely in the undiffer-entiated state and differentiated into functional beta-cells

poten-However, generation of fully differentiated beta-cells from

ES cells has been diffi cult and controversial Beta-cell ferentiation from ES cells as determined on the basis of immunohistochemical evidence alone has been questioned, because insulin immunoreactivity can also result from insulin absorption from the medium as well as from genuine beta-cell differentiation Therefore, these types of results should be interpreted with caution

dif-Some promising results have been reported for the differentiation of insulin-producing cells from mouse and human ES cells (reviewed in Bonner-Weir and Weir, 2005;

Jun and Yoon, 2005; Montanya, 2004; Stoffel et al., 2004)

Pancreatic endocrine cells, including insulin-producing cells, could be generated from mouse embryonic stem cells

by a fi ve-step protocol, including the enrichment of positive cells from embryoid bodies, and these cells secreted insulin in response to glucose and other insulin secreta-gogues, such as tolbutamide and carbachol However, these cells could not remit hyperglycemica when transplanted into diabetic mice A modifi ed protocol, in which a phos-phoinositide kinase inhibitor was added to the medium to inhibit cell proliferation, resulted in improved insulin content and glucose-dependent insulin release To enrich insulin-producing cells from mouse ES cells, a neomycin-resistance gene regulated by the insulin promoter was transferred to ES cells, which drove differentiation of insulin-secreting cells, and transplantation of these cells restored normoglycemia in streptozotocin-induced diabetic mice

nestin-In another report, mouse ES cells were transduced with a

plasmid containing the Nkx6.1 promoter gene, followed by

a neomycin-resistance gene to select the Nkx6.1-positive

cells, and were differentiated in the presence of exogenous

differentiating factors The selected Nkx6.1-positive cells

coexpressed insulin and Pdx-1, and transplantation of these

cells into streptozotocin-induced diabetic mice resulted in

normoglycemia

Exogenous expression of beta-cell transcription factors

has been used as a strategy to drive the differentiation of

insulin-producing cells from ES cells Overexpression of

Pax4 in mouse ES cells promoted the differentiation

of nestin-positive progenitor and insulin-producing cells,

and these cells secreted insulin in response to glucose

and normalized blood glucose when transplanted into

diabetic mice (Blyszczuk et al., 2003) In the same study,

the expression of Pdx-1 did not have a signifi cant effect

on the differentiation of insulin-producing cells from ES

cells However, another study demonstrated that the

regulated expression of Pdx-1 in a murine ES cell line

by the tet-off system enhanced the expression of insulin

and other beta-cell transcription factors (Miyazaki et al.,

2004)

It was shown that human ES cells can spontaneously

differentiate in vitro into insulin-producing beta-cells,

evi-denced by the secretion of insulin and expression of other

beta-cell markers (Assady et al., 2001) Differentiation of

insulin-expressing cells from human ES cells was promoted

when they were cultured in conditioned medium in the

presence of low glucose and fi broblast growth factor,

fol-lowed by nicotinamide (Segev et al., 2004) A recent report

suggested that human ES cells differentiated into

beta-cell-like clusters when cotransplanted with mouse dorsal

pancreas (Brolen et al., 2005) Although several in vitro

studies suggest the possibility of generating

insulin-expressing cells from human ES cells, differentiation of truly

functional beta-cells from human ES cells has not yet been

reported

Because of their proliferative ability and capacity to

dif-ferentiate in culture, ES cells have received much attention

as a potential source of unlimited quantities of beta-cells for

transplantation therapy for diabetes However, use of ES

cells has ethical concerns, and the mechanisms by which ES

cells differentiate to produce islets and beta-cells are not

well understood Therefore, further studies are needed to

understand the details of the endoderm and beta-cell

differentiation process so that an effective protocol for

differentiating ES cells into insulin-producing cells can be

developed

Engineering Adult Stem and Progenitor Cells

As with ES cells, adult stem cells have the potential to

differentiate into other cell lineages, but they do not bring

the ethical diffi culties associated with ES cells Beta-cell

neogenesis in adults has been reported in animal models of

experimentally induced pancreatic damage, suggesting the

presence of adult stem/progenitor cells These adult stem/progenitor cells could be potential sources for the produc-tion of new insulin-producing cells (reviewed in Jun and Yoon, 2005; Montanya, 2004; Nir and Dor, 2005) Bone marrow, mesenchymal splenocytes, neural stem cells, liver oval stem cells, and pancreatic stem cells have been inves-tigated for their potential to differentiate into insulin-producing cells

A large body of evidence suggests that the adult atic ducts are the main site of beta-cell progenitors Through-out life, the islets of Langerhans turn over slowly, and new small islets are continuously generated by differentiation of

pancre-ductal progenitors (Finegood et al., 1995) It was found that isletlike clusters were generated in vitro from mouse pan-

creatic ducts and ductal tissue–enriched human pancreatic islets In addition, multipotent precursor cells clonally iden-tifi ed from pancreatic islets and ductal populations could differentiate into cells with beta-cell function The expres-sion of the Pdx-1 gene or treatment of ductal cells with Pdx-

1 protein increased the number of insulin-positive cells or induced insulin expression Ectopic expression of neuro-genin 3, a critical factor for the development of the endo-crine pancreas in humans, in pancreatic ductal cells led to their conversion into insulin-expressing cells In addition, treatment of human islets containing both ductal and ascinar cells with a combination of epidermal growth factor and gastrin induced neogenesis of islet beta-cells from the ducts and increased the functional beta-cell mass In addi-tion to ductal cells, exocrine acinar cells and other endo-crine cells can generate beta-cells An alpha-cell line transfected with Pdx-1 expressed insulin when treated with betacellulin It was shown that treatment of rat exocrine pancreatic cells with epidermal growth factor and leukemia inhibitory factor could induce differentiation into insulin-

producing beta-cells (Baeyens et al., 2005) Considerable

evidence suggests that beta-cells in the pancreatic islets can

be dedifferentiated, expanded, and redifferentiated into beta-cells by inducing the epithelial–mesenchymal transi-

tion process (Lechner et al., 2005) Nonendocrine

pancre-atic epithelial cells also have been reported to differentiate

into beta-cells (Hao et al., 2006) These results suggest that

pancreatic stem/progenitor cells are the source of new islets

There is also the possibility of manipulating genitor cells from other organs to transform into the beta-cell phenotype (reviewed in Montanya, 2004; Nir and Dor, 2005) Although there are controversies regarding the differen-tiation of bone marrow–derived stem cells into insulin-producing cells, some successful studies have been reported

stem/pro-In vitro differentiation of mouse bone marrow cells resulted

in the expression of genes related to pancreatic beta-cell development and function These differentiated cells released insulin in response to glucose and reversed hyperglycemia when transplanted into diabetic mice In addition, ectopic expression of key transcription factors of the endocrine

640 C H A P T E R F O R T Y - T H R E E • E N G I N E E R I N G P A N C R E A T I C B E T A - C E L L S

pancreas developmental pathway, such as IPF1, HLXB9, and

FOXA2, in combination with conditioned media in human

bone marrow mesenchymal stem cells differentiated them

into insulin-expressing cells (Moriscot et al., 2005).

Because the liver and intestinal epithelium are derived

from gut endoderm, as is the pancreas, the generation of

islets from both developing and adult liver and intestinal

cells has been tried Rat hepatic oval stem cells could

dif-ferentiate into insulin-producing isletlike cells when

cul-tured in a high-glucose environment Fetal human liver

progenitor cells and mouse hepatocytes could differentiate

into insulin-producing cells when engineered to produce

Pdx-1, and transplantation of these cells reversed

hypergly-cemia in mice It was reported that adult human liver cells

engineered to express Pdx-1 produced insulin and secreted

it in a glucose-regulated manner Transplantation of these

engineered cells under the renal capsule of diabetic mice

resulted in prolonged reduction of hyperglycemia (Sapir

et al., 2005) As well, ectopic islet neogenesis in the liver

could be induced by gene therapy with a combination of

NeuroD, a transcription factor downstream of Pdx-1, and

betacellulin, which reversed diabetes in

streptozotocin-treated diabetic mice Expression of Pdx-1 in a rat

entero-cyte cell line in combination with betacellulin treatment or

coexpression of Isl-1 resulted in the expression of insulin

Treatment of developing as well as adult mouse intestinal

cells with GLP-1 induced insulin production, and

transplan-tation of these cells into streptozotocin-induced diabetic

mice remitted diabetes A recent study showed that neural

progenitor cells could generate glucose-responsive,

insulin-producing cells when exposed in vitro to a series of signals

for pancreatic islet development (Hori et al., 2005) These

results suggest that the controlled differentiation of liver or

intestinal cells into insulin-producing cells may provide an

alternative source of beta-cells

The use of adult stem/progenitor cells for generating

beta-cells for transplantation therapy appears to be

promis-ing, although most of the studies have only been done in

animal models Further studies on the mechanisms for the

differentiation of adult stem/progenitor cells into

insulin-producing beta-cells and the characterization of the newly

generated beta-cells are required before these cells can be

considered for clinical application

III ENGINEERING TO IMPROVE

ISLET SURVIVAL

A hurdle to overcome for islet transplantation therapy

is the rejection and autoimmune attack against the

trans-planted beta-cells Immunosuppressive drugs have been

used successfully, but they have many side effects

There-fore, it is desirable to develop drug-free strategies for the

induction of tolerance to transplanted islet or beta-cells A

variety of approaches to protect islet grafts have been

studied, such as bone marrow transplantation, treatment

with anti-T-cell agents, and inhibition of activation of antigen-presenting cells Another approach is engineering islets or beta-cells to express therapeutic genes to improve islet viability and function, such as genes for cytokines, antiapoptotic molecules, antioxidants, immunoregulatory molecules, and growth factors (reviewed in Giannoukakis and Trucco, 2005; Jun and Yoon, 2005; Van Linthout and Madeddu, 2005) (Table 43.1)

With regard to cytokines, introduction of genes for interleukin (IL)-4 or a combination of IL-10 and transform-ing growth factor-β improved islet graft survival by prevent-ing immune attack in mice In addition, islets expressing the p40 subunit of IL-12 could maintain normoglycemia when transplanted into diabetic NOD recipients by decreasing interferon-γ production and increasing transforming growth

Table 43.1 Engineering islets for beta-cell survival

Cytokine expression Erythropoietin

Interleukin (IL)-1 receptor antagonist

IL-4IL-10IL-12p40Transforming growth factor-β

expression Bcl-xL

Dominant-negative MyD88Fas ligand

Flice-like inhibitory protein

IκB kinase inhibitorTumor necrosis factor receptor- immunoglobulin (Ig)

Glutathione peroxidaseHeme oxygenase-1Manganese superoxide dismutaseImmunoregulatory Adenoviral E3 genes

antigen-4-IgDipeptide boronic acidIndoleamine 2,3-dioxygenase

Vascular endothelial growth factor

factor-β at the transplantation site Islets engineered to

produce IL-1β receptor antagonist were also found to be

more resistant to rejection Adenoviral-mediated gene

trans-fer of erythropoietin, a cytokine that promotes survival, in

islets resulted in protection of islets from apoptosis in

culture and destruction in vivo.

Expression of antiapoptotic molecules such as Bcl-2,

Bcl-xL, and A20, which inhibit nuclear factor-κB activation,

or an IκB kinase inhibitor was shown to protect from

apop-tosis In addition, the expression of soluble human Fas

ligand, dominant negative MyD88, fl ice-like inhibitory

protein, or tumor necrosis factor receptor-immunoglobulin

(Ig) improved allogeneic islet graft survival A recent study

demonstrated that silencing Fas expression with small

inter-fering RNA in mouse insulinoma cells inhibited

Fas-medi-ated beta-cell apoptosis (Burkhardt et al., 2006).

Pancreatic islets are sensitive to oxidative stress because

they produce relatively low amounts of antioxidant enzymes

Thus, expression of antioxidant molecules such as catalase,

glutathione perioxidase, and manganese superoxide

dis-mutase in islets or insulinoma cells could protect against

oxidative stress and cytokine-induced damage In addition,

expression of heme oxygenase in pancreatic islets protected

against IL-1β-induced islet damage It was also found that

delivery of a c-Jun-terminal kinase inhibitory peptide into

isolated islets by the protein transduction system prevented

apoptosis (Noguchi et al., 2005).

Expression of immunoregulatory molecules that affect

T-cell activation and proliferation have been tried

Expres-sion in islets of cytotoxic T-lymphocyte

antigen-4-immuno-globulin, which down-regulates T-cell activation, or CD40-Ig,

which blocks CD40–CD40 ligand interactions, prolonged

allogenic and xenogeneic graft survival Transplantation

of islets overexpressing indoleamine 2,3-dioxygenase

pro-longed survival in NOD/SCID mice after adoptive transfer

of diabetogenic T-cells, probably by inhibiting T-cell

prolif-eration by the depletion of tryptophan at the

transplanta-tion site Similarly, a proteasome inhibitor, dipeptide boronic

acid, was found to prevent islet allograft rejection by

sup-pressing the proliferation of T-cells Expression of

adenovi-ral E3 transgenes in beta-cells was found to prevent islet

destruction by autoimmune attack through the inhibition of

major histocompatibility complex I expression

With regard to growth factors, adenoviral-mediated

transfer of hepatocyte growth factor resulted in an improved

islet transplant outcome in animal models As well,

expres-sion of insulinlike growth factor-1 in human islets prevented

IL-1βinduced betacell dysfunction and apoptosis Insuffi

-cient revascularization of transplanted islets can deprive

them of oxygen and nutrients, contributing to graft failure

Therefore the expression of vascular endothelial growth

factor, a key angiogenic molecule, enhanced islet

revascu-larization and improved the long-term survival of murine

islets after transplantation into the renal capsule of diabetic

mice

Another strategy to protect islets from immune attack

is microencapsulation of islets within synthetic polymers

(Kizilel et al., 2005) Encapsulation of islets within a

semi-permeable membrane, such as nate, blocks the passage of larger cells but allows the passage

alginatepolyllysinealgiof small molecules, thus conferring protection from auto

-i mmune attack However, th-is method has l-im-itat-ions for the long-term survival of islets within the microcapsules because of the lack of biocompatibility, ischemia, and limited protection from cytokine-induced damage To over-come these limitations, a bioartifi cial pancreas has been developed, in which blood fl ows through artifi cial vessels in close proximity to insulin-producing cells

A variety of approaches for engineering islets or cells for improved islet graft survival and escape from immune rejection have been successful in animal models However, the effi cacy of these approaches in human dia-betic patients remains to be determined

beta-IV VECTORS FOR ENGINEERING ISLETS AND BETA-CELLS

The cells of the pancreas divide very slowly; therefore gene transfer vehicles that can transduce quiescent cells have been used for the delivery of transgenes, such as nonviral plasmids and vectors based on lentivirus, adeno-virus, helper-dependent adenovirus, adeno-associated virus (AAV), and herpes simplex virus In addition, protein trans-duction using the cell penetrating peptide from HIV-1 trans-

acting protein (reviewed in Becker-Hapak et al., 2001) has

been successfully used to engineer islets (Table 43.2) However, the choice of vector needs to be carefully made so that the vector itself does not affect islet function or viability

Nonviral methods are considered safe, cost effective, and simple to use and do not induce an immune response, but they generally have a lower gene transfer effi cacy as compared to viral-mediated gene transfer (reviewed in Nishikawa and Huang, 2001) Nonviral methods for transferring genetic material include the direct injection

of DNA, either naked or enclosed in a liposome, poration, and the gene gun method Cationic lipid and polymer-based plasmid delivery has been used to trans-duce islets, and the expression of cytotoxic T-lymphocyte antigen-4 by biolistically transfected islets improved graft survival

electro-Viral vectors (reviewed in Walther and Stein, 2000) have been widely used as a method of gene transfer to engineer islets and beta-cell surrogates Retroviral vectors derived from Moloney murine leukemia virus can carry a gene effi -ciently and integrate it in a stable manner within the host chromosomal DNA, facilitating long-term expression of the gene For immortalization of human islets, retroviral vectors expressing oncogenes or telomerase genes have been used

(Narushima et al., 2005) Although most retroviral vectors

642 C H A P T E R F O R T Y - T H R E E • E N G I N E E R I N G P A N C R E A T I C B E T A - C E L L S

only infect proliferating cells, the lentivirus genus of

retro-viruses, which includes the human immunodefi ciency virus,

has all the advantages of Moloney murine leukemia virus–

derived retroviral vectors and can infect nondividing as well

as dividing cells Lentiviral vectors have been successfully

used to transduce islets with marker proteins (Okitsu et al.,

2003)

The adenoviral vector can harbor up to 30 Kb of foreign

DNA and can transduce nondividing cells with high effi

-ciency In addition, a relatively high titer of virus, about 1012

plaque-forming units/mL, can be produced The transferred

genes are not integrated into the host genome, but remain

as nonreplicating extrachromosomal DNA within the

nucleus Although there is no risk of alteration in cellular

genotype by insertional mutation, the duration of gene

expression may be short, and a strong cellular immune

response to the viral proteins and, in some cases, to the

transgene may be induced Adenoviral vectors have been

widely used to transduce islets for proof-of-concept

experi-ments in vitro and in vivo Although adenoviral vectors are

toxic because of de novo synthesized viral proteins, islet

viability and functional characteristics were not affected

when transduced in vitro However, transduction of islets

with a high dose of recombinant adenovirus (500 MOI) markedly reduced glucose-stimulated insulin secretion, suggesting that an optimal dose is required to result in effi -cient transduction without compromising islet function In general, adenoviral vectors result in transient transgene expression; however, long-term (20-week) expression of the transgene was observed in islets transduced with β-galacto-sidase and transplanted into syngeneic diabetic mice It was reported that double-genetic modifi cation of the adeno-virus fi ber with RGD polylysine motifs signifi cantly reduced toxicity, infl ammation, and immune responses (Contreras

et al., 2003).

A new generation of adenovirus vectors has been oped that are completely devoid of all viral protein–coding sequences and are therefore less immunogenic and less toxic Although these gutless viruses require the presence of

devel-a helper virus for replicdevel-ation, contdevel-amindevel-ation by the helper virus can be avoided by genetically engineering a condi-tional defect in the packaging domain of the helper virus or

fl anking the packing signal with loxP expression sites and encoding Cre recombinase in the supporting cell line In

Table 43.2 Vectors used for islet and beta-cell engineering

Less toxic compared with viral vectors Transient expression

Random integration into host chromosomal DNA

Infects dividing and nondividing cells Limited insertion capacity (8 Kb)

DNA

Infects dividing and nondividing cells Short-term expression

compared with adenoviral vectorLarge insertion capacity

Long-term expressionInfects dividing and nondividing cells

Transduces many cell types

addition, the gutless vectors are known to have a prolonged

expression of the transgene However, there is no report

about islet engineering using these vectors

AAVs are nonpathogenic, replication-defective

parvo-viruses that can infect both dividing and nondividing cells

AAVs generally have low immunogenicity; however, the

generation of neutralizing antibodies may limit

readminis-tration This problem can be overcome by selective capsid

modifi cation of AAV to evade recognition by preexisting

anti-bodies or by direct administration of AAV to the target tissue

The recombinant AAV vector integrates randomly into the

host chromosome or may stay in the episomal state There

is a limitation in the size of the DNA that can be inserted (a

maximum of 4.8 Kb); however, larger inserts can be split over

two vectors and delivered simultaneously, because AAVs

tend to form concatemers, although the effi ciency of

duction is often reduced (Young et al., 2006) Effi cient

trans-duction of islets was achieved using a high dose of AAVs with

an improved recombinant AAV purifi cation method, which

improved infectious titers and yield Transduction of islets

with AAV5 is more effi cient than with AAV2, due to the low

number of receptors for AAV2 on islet cells AAV1 was found

to be the most effi cient serotype in transducing murine islets

(Loiler et al., 2003) However, it was recently demonstrated

that intact human and murine islets could be effi ciently

transduced with a double-stranded AAV2-based vector, and

the transduced murine islets showed normal glucose

respon-siveness and viability (Rehman et al., 2005).

Herpes simplex virus type 1 (HSV-1) has also been used

as a viral vector Based on the persistence of latent herpes

virus after infection, HSV-1 is attractive for its effi cient

infec-tivity in a wide range of target cells and its ability to infect

both dividing and nondividing cells, including islets

Trans-fection of human islets with Bcl-2 protected beta-cells from

cytokine-induced damage However, the transduction may

be unstable, and potential health risks of this vector remain

to be determined

Protein transduction is an emerging technology to

deliver therapeutic proteins into cells as an alternative to

gene therapy This method uses peptides that can penetrate the cell membrane, such as antennapedia peptide, the HSV VP22 protein, and human immunodefi ciency virus TAT protein transduction domain The therapeutic molecule

is linked to the penetrating peptide as a fusion protein, which is then used to transduce the cell (Becker-Hapak

et al., 2001) The protein transduction method is not

im-munogenic and can transduce a variety of cell types, but it has a short half-life Delivery of antiapoptotic proteins such as Bcl-xL or anti-oxidant enzymes such as copper-zinc superoxide dismutase and heme oxygenase by protein transduction in human and rodent islets effi ciently trans-duced the islets and improved their viability without

affecting islet function (Embury et al., 2001; Mendoza et al.,

2005)

V CONCLUSION

Engineering beta-cell lines and non-beta-cells, entiating embryonic and adult stem cells, and transdiffer-entiating non-beta-cells have been studied as methods to provide new beta-cells for cell therapy for diabetes Expan-sion of functional beta-cells by generation of reversibly immortalized human beta-cell lines has been reported, but the techniques have not been clinically proven Generation

differ-of insulin-producing cells from non-beta-cells is an tive method, but it has yet to achieve tight regulation of glucose-responsive insulin secretion Differentiation of insulin-producing cells from ES cells and adult stem/progenitor cells is also a promising alternative to produce beta-cells; however, a better understanding of the mecha-nisms for the differentiation of beta-cells is needed to develop a successful strategy to engineer beta-cells from stem cells Engineering of islets and beta-cells to improve the survival of islet transplants has also been investigated Although much progress has been made, engineered beta-cells need to be carefully analyzed for true beta-cell function and possible tumorigenicity It is hoped that continued research on beta-cell engineering will offer a potential cure for diabetes in the future

attrac-VI REFERENCES

Adorini, L., Gregori, S., and Harrison, L C (2002) Understanding

autoimmune diabetes: insights from mouse models Trends Mol Med

8, 31–38.

Assady, S., Maor, G., Amit, M., Itskovitz-Eldor, J., Skorecki, K L., and

Tzukerman, M (2001) Insulin production by human embryonic stem

cells Diabetes 50, 1691–1697.

Baeyens, L., De Breuck, S., Lardon, J., Mfopou, J K., Rooman, I., and

Bouwens, L (2005) In vitro generation of insulin-producing beta cells

from adult exocrine pancreatic cells Diabetologia 48, 49–57.

Beattie, G M., Itkin-Ansari, P., Cirulli, V., Leibowitz, G., Lopez, A D.,

Bossie, S., Mally, M I., Levine, F., and Hayek, A (1999) Sustained

proliferation of PDX-1+ cells derived from human islets Diabetes 48,

1013–1019.

Becker-Hapak, M., McAllister, S S., and Dowdy, S F (2001)

TAT-mediated protein transduction into mammalian cells Methods 24,

247–256.

Blyszczuk, P., Czyz, J., Kania, G., Wagner, M., Roll, U., St-Onge, L., and Wobus, A M (2003) Expression of Pax4 in embryonic stem cells promotes differentiation of nestin-positive progenitor and insulin-

producing cells Proc Natl Acad Sci U.S.A 100, 998–1003.

Bonner-Weir, S., and Weir, G C (2005) New sources of pancreatic

beta-cells Nat Biotechnol 23, 857–861.

Brolen, G K., Heins, N., Edsbagge, J., and Semb, H (2005) Signals from the embryonic mouse pancreas induce differentiation of human

embryonic stem cells into insulin-producing beta-cell-like cells

Diabe-tes 54, 2867–2874.

644 C H A P T E R F O R T Y - T H R E E • E N G I N E E R I N G P A N C R E A T I C B E T A - C E L L S

Burkhardt, B R., Parker, M J., Zhang, Y C., Song, S., Wasserfall, C H.,

and Atkinson, M A (2005) Glucose transporter-2 (GLUT2) promoter–

mediated transgenic insulin production reduces hyperglycemia in

dia-betic mice FEBS Lett 579, 5759–5764.

Burkhardt, B R., Lyle, R., Qian, K., Arnold, A S., Cheng, H., Atkinson,

M A., and Zhang, Y C (2006) Effi cient delivery of siRNA into

cytokine-stimulated insulinoma cells silences Fas expression and inhibits

Fas-mediated apoptosis FEBS Lett 580, 553–560.

Cheung, A T., Dayanandan, B., Lewis, J T., Korbutt, G S., Rajotte, R V.,

Bryer-Ash, M., Boylan, M O., Wolfe, M M., and Kieffer, T J (2000)

Glucose-dependent insulin release from genetically engineered K cells

Science 290, 1959–1962.

Contreras, J L., Wu, H., Smyth, C A., Eckstein, C P., Young, C J., Seki,

T., Bilbao, G., Curiel, D T., and Eckhoff, D E (2003) Double genetic

modifi cation of adenovirus fi ber with RGD polylysine motifs signifi

-cantly enhances gene transfer to isolated human pancreatic islets

Transplantation 76, 252–261.

de la Tour, D., Halvorsen, T., Demeterco, C., Tyrberg, B., Itkin-Ansari, P.,

Loy, M., Yoo, S J., Hao, E., Bossie, S., and Levine, F (2001) Beta-cell

dif-ferentiation from a human pancreatic cell line in vitro and in vivo Mol

Endocrinol 15, 476–483.

Embury, J., Klein, D., Pileggi, A., Ribeiro, M., Jayaraman, S., Molano,

R D., Fraker, C., Kenyon, N., Ricordi, C., Inverardi, L., et al (2001)

Pro-teins linked to a protein transduction domain effi ciently transduce

pancreatic islets Diabetes 50, 1706–1713.

Finegood, D T., Scaglia, L., and Bonner-Weir, S (1995) Dynamics of

beta-cell mass in the growing rat pancreas Estimation with a simple

mathematical model Diabetes 44, 249–256.

Giannoukakis, N., and Trucco, M (2005) Gene therapy for type 1

dia-betes Am J Ther 12, 512–528.

Halvorsen, T L., Leibowitz, G., and Levine, F (1999) Telomerase activity

is suffi cient to allow transformed cells to escape from crisis Mol Cell

Biol 19, 1864–1870.

Hao, E., Tyrberg, B., Itkin-Ansari, P., Lakey, J R., Geron, I., Monosov,

E Z., Barcova, M., Mercola, M., and Levine, F (2006) Beta-cell

differen-tiation from nonendocrine epithelial cells of the adult human pancreas

Nat Med 12, 310–316.

Hatipoglu, B., Benedetti, E., and Oberholzer, J (2005) Islet

trans-plantation: current status and future directions Curr Diab Rep 5,

311–316.

Hori, Y., Gu, X., Xie, X., and Kim, S K (2005) Differentiation of

insulin-producing cells from human neural progenitor cells PLoS Med 2,

e103.

Islam, M S., Rahman, S A., Mirzaei, Z., and Islam, K B (2005)

Engineered beta-cells secreting dipeptidyl peptidase IV–resistant

glucagon-like peptide-1 show enhanced glucose responsiveness Life

Sci 76, 1239–1248.

Jun, H S., and Yoon, J W (2005) Approaches for the cure of type 1

dia-betes by cellular and gene therapy Curr Gene Ther 5, 249–262.

Kizilel, S., Garfi nkel, M., and Opara, E (2005) The bioartifi cial

pancreas: progress and challenges Diabetes Technol Ther 7, 968–985.

Lechner, A., Nolan, A L., Blacken, R A., and Habener, J F (2005)

Redif-ferentiation of insulin-secreting cells after in vitro expansion of adult

human pancreatic islet tissue Biochem Biophys Res Commun 327,

581–588.

LeRoith, D (2002) Beta-cell dysfunction and insulin resistance in type

2 diabetes: role of metabolic and genetic abnormalities Am J Med

113(Suppl 6A), 3S–11S.

Loiler, S A., Conlon, T J., Song, S., Tang, Q., Warrington, K H., Agarwal,

A., Kapturczak, M., Li, C., Ricordi, C., Atkinson, M A., et al (2003)

Targeting recombinant adeno-associated virus vectors to enhance

gene transfer to pancreatic islets and liver Gene Ther 10, 1551–

1558.

Mendoza, V., Klein, D., Ichii, H., Ribeiro, M M., Ricordi, C., Hankeln, T., Burmester, T., and Pastori, R L (2005) Protection of islets in culture by delivery of oxygen-binding neuroglobin via protein transduction

Transplant Proc 37, 237–240.

Milo-Landesman, D., Surana, M., Berkovich, I., Compagni, A., Christofori, G., Fleischer, N., and Efrat, S (2001) Correction of hyper- glycemia in diabetic mice transplanted with reversibly immortalized

pancreatic beta cells controlled by the tet-on regulatory system Cell

Transplant 10, 645–650.

Miyazaki, S., Yamato, E., and Miyazaki, J (2004) Regulated expression

of pdx-1 promotes in vitro differentiation of insulin-producing cells

from embryonic stem cells Diabetes 53, 1030–1037.

Montanya, E (2004) Islet- and stem-cell-based tissue engineering in

diabetes Curr Opin Biotechnol 15, 435–440.

Moriscot, C., de Fraipont, F., Richard, M J., Marchand, M., Savatier, P., Bosco, D., Favrot, M., and Benhamou, P Y (2005) Human bone marrow mesenchymal stem cells can express insulin and key transcription factors of the endocrine pancreas developmental pathway upon genetic

and/or microenvironmental manipulation in vitro Stem Cells 23,

594–603.

Narushima, M., Kobayashi, N., Okitsu, T., Tanaka, Y., Li, S A., Chen, Y.,

Miki, A., Tanaka, K., Nakaji, S., Takei, K., et al (2005) A human beta-cell line for transplantation therapy to control type 1 diabetes Nat Biotech-

nol 23, 1274–1282.

Nett, P C., Sollinger, H W., and Alam, T (2003) Hepatic insulin gene

therapy in insulin-dependent diabetes mellitus Am J Transplant 3,

1197–1203.

Nir, T., and Dor, Y (2005) How to make pancreatic beta cells —

prospects for cell therapy in diabetes Curr Opin Biotechnol 16, 524–

529.

Nishikawa, M., and Huang, L (2001) Nonviral vectors in the new

mil-lennium: delivery barriers in gene transfer Hum Gene Ther 12,

861–870.

Noguchi, H., Nakai, Y., Matsumoto, S., Kawaguchi, M., Ueda, M., Okitsu,

T., Iwanaga, Y., Yonekawa, Y., Nagata, H., Minami, K., et al (2005)

Cell-permeable peptide of JNK inhibitor prevents islet apoptosis

immedi-ately after isolation and improves islet graft function Am J Transplant

5, 1848–1855.

Okitsu, T., Kobayashi, N., Totsugawa, T., Maruyama, M., Noguchi, H., Watanabe, T., Matsumura, T., Fujiwara, T., and Tanaka, N (2003) Len- tiviral vector–mediated gene delivery into nondividing isolated islet

cells Transplant Proc 35, 483.

Rehman, K K., Wang, Z., Bottino, R., Balamurugan, A N., Trucco, M.,

Li, J., Xiao, X., and Robbins, P D (2005) Effi cient gene delivery to human and rodent islets with double-stranded (ds) AAV-based vectors

Natl Acad Sci U.S.A 102, 7964–7969.

Segev, H., Fishman, B., Ziskind, A., Shulman, M., and Itskovitz-Eldor, J

(2004) Differentiation of human embryonic stem cells into

insulin-producing clusters Stem Cells 22, 265–274.

Stoffel, M., Vallier, L., and Pedersen, R A (2004) Navigating the pathway

from embryonic stem cells to beta cells Semin Cell Dev Biol 15,

327–336.

Van Linthout, S., and Madeddu, P (2005) Ex vivo gene transfer for

improvement of transplanted pancreatic islet viability and function

Curr Pharm Des 11, 2927–2940.

Walther, W., and Stein, U (2000) Viral vectors for gene transfer: a review

of their use in the treatment of human diseases Drugs 60, 249–271.

Wu, L., Nicholson, W., Wu, C Y., Xu, M., McGaha, A., Shiota, M., and

Powers, A C (2003) Engineering physiologically regulated insulin

secretion in non-beta cells by expressing glucagon-like peptide 1

recep-tor Gene Ther 10, 1712–1720.

Xu, R., Li, H., Tse, L Y., Kung, H F., Lu, H., and Lam, K S (2003) Diabetes

gene therapy: potential and challenges Curr Gene Ther 3, 65–82.

Yoon, J W., and Jun, H S (2002) Recent advances in insulin gene

therapy for type 1 diabetes Trends Mol Med 8, 62–68.

Yoon, J W., and Jun, H S (2005) Autoimmune destruction of pancreatic

beta cells Am J Ther 12, 580–591.

Young, L S., Searle, P F., Onion, D., and Mautner, V (2006) Viral gene

therapy strategies: from basic science to clinical application J Pathol

208, 299–318.

I Introdution

II Structure and Morphology of the Thymus

III In Vitro T-Cell Differentiation

IV Thymus Organogenesis

V Summary

VI Acknowledgments VII References

Thymus and Parathyroid

Organogenesis

Craig Scott Nowell, Ellen Richie, Nancy Ruth Manley, and Catherine Clare Blackburn

Principles of Tissue Engineering, 3 rd Edition

ed by Lanza, Langer, and Vacanti

Copyright © 2007, Elsevier, Inc.

All rights reserved.

I INTRODUTION

The thymus is the principal site of T-cell development

and therefore is of central importance within the immune

system: Congenital athymia results in profound

immunode-fi ciency (Flanagan, 1966; Dodson et al., 1969; J Frank et al.,

1999), while perturbed thymic function can lead to

autoim-munity Although highly active in early life, the thymus

undergoes premature involution, such that de novo T-cell

development diminishes signifi cantly with age This has

implications for immune function in the aging population

and in clinical procedures such as bone marrow and solid

organ transplantation, where thymic function is required

for T-cell reconstitution and/or tolerance induction Interest

therefore exists in enhancing immune reconstitution

through regenerative or cell therapies for boosting thymus

activity in vivo or for providing customized in

vitro–gener-ated T-cell repertoires for adoptive transfer The success of

such strategies is likely to depend on a detailed knowledge

of the mechanisms regulating thymus development and

homeostasis Here, we review current understanding of

cel-lular and molecular regulation of thymus organogenesis,

focusing on the epithelial component of the thymic stroma,

which provides many of the specialist functions required

to mediate T-cell differentiation and T-cell repertoire

mesenchymal cells, bone marrow (BM)–derived cells, culature and the uniquely specialized thymic epithelium

vas-(Boyd et al., 1993).

The mature thymus is encapsulated and lobulated and contains three principal histologically defi ned regions: the cortex, the medulla and the subcapsule (Fig 44.1) The capsule and trabeculae consist of a thick layer of connective tissue and are separated from the cortex by a thin layer of

simple epithelium, the subcapsule (Boyd et al., 1993) The

cortex and medulla each contain open networks of lial cells, which are densely packed with thymocytes (Van

epithe-Vliet et al., 1985; Boyd et al., 1993; van Ewijk et al., 1994), and

each of these regions contains several different cally and phenotypically distinct epithelial subtypes (see later) The outer cortex also contains fi broblasts, and the organ as a whole is heavily vascularized BM-derived stromal cells are found in both compartments, macrophages being distributed throughout the organ, while thymic dendritic

cells — which are required for imposition of tolerance on

the emerging T-cell repertoire — are found predominantly

at the corticomedullary junction and in the medulla itself

(Boyd et al., 1993).

Thymus structure is intimately linked to its principal

function, to support T-cell development This encompasses

the linked processes of T-cell differentiation and T-cell

rep-ertoire selection, which together ensure that the peripheral

T-cell repertoire is populated predominantly by T-cells that

have propensity to bind antigen in the context of self-major

histocompatibility antigens (MHC) but that do not bind

self-peptides T-cell development has been extensively

reviewed elsewhere (Zuniga-Pfl ucker and Lenardo, 1996;

Petrie, 2002; Rothenberg and Dionne, 2002) and is not

dis-cussed in detail herein In brief, hematopoietic progenitors

enter the postnatal thymus at the corticomedullary

junc-tion, and subsequent T-cell development is then regulated

such that thymocytes at different stages of development are

found in different intrathymic locations T-cell

differentia-tion from the earliest postcolonizadifferentia-tion stages of thymocyte

maturation [termed triple negative (TN) cells because they

do not express CD3 or the coreceptors CD4 and CD8]

through to the CD4+CD8+ double positive (DP) stage occurs

in the cortex, and the cortex itself can be subdivided into four regions based on the localization of different thymo-cyte populations Thus, zone 1 contains the colonizing population of hematopoietic progenitor cells; these early thymocytes undergo proliferative expansion in zone 2; T-cell lineage commitment is completed in zone 3; and in zone 4, thymocytes differentiate to the DP stage of develop-ment, characterized by expression of both CD4 and CD8

coreceptors (Lind et al., 2001; Porritt et al., 2004) DP cells

are then screened for their propensity to recognize

self-MHC, a process termed positive selection, and those selected

to mature into CD4+ or CD8+ single positive (SP) cells migrate

into the medulla (Kurobe et al., 2006) Central tolerance is

established by deletion of self-reactive thymocytes at the

DP–SP transition, in a process termed negative selection (Baldwin et al., 2005), and is thought to occur principally

at the corticomedullary junction and in the medulla SP

FIG 44.1. Histology of the postnatal thymus The postnatal thymus is surrounded by a capsule consisting of mesenchymal cells and connective tissue

that penetrates into the thymus at regular intervals to form trabeculae Underlying the capsule and trabeculae is the subcapsular epithelium, consisting of

a layer of simple epithelium, which overlies the outer cortex The cortex is populated with cortical thymic epithelial cells (cTEC), macrophages, and developing

thymocytes at the triple negative (TN) and double positive (DP) stages of development Thymocytes enter the thymus at the corticomedullary junction (CMJ)

via the vasculature and migrate through the cortex to the subcapsule as they differentiate The cortex can be divided into four zones based on the

differentia-tion status of thymocytes that reside within it Thus, zone 1 contains the most immature TN1 thymocytes and zone 4 contains thymocytes undergoing the

TN4–DP transition DP thymocytes are then screened for propensity to recognize self-MHC, a process termed positive selection, and those selected to mature

into CD4+ or CD8+ single positive (SP) cells migrate into the medulla, where they undergo the fi nal stages of maturation before being exported to the periphery

Central tolerance is established by deletion of self-reactive thymocytes in a process termed negative selection, and it is thought to occur principally at the

corticomedullary junction and in the medulla: Negative selection is mediated by both thymic dendritic cells and medullary TECs, which supply self-peptides

to medullary dendritic cells (DCs) in a process termed cross-presentation Medullary TECs are also required for the generation of CD4+CD25+ T regulatory

(Treg) cells and natural killer T-cells, both of which actively repress self-reactive T-cells

I I I I N V I T R O T - C E L L D I F F E R E N T I A T I O N • 649

thymocytes proliferate and undergo the fi nal stages of

T-cell maturation in the medulla before exiting into the

peri-pheral immune system The outward migration of

thymo-cytes, from the corticomedullary junction to the outer

cortex, is regulated by chemokines (Plotkin et al., 2003), as

is the migration of positively selected cells from the cortex

into the medulla (Ueno et al., 2004; Kurobe et al., 2006).

Thymic Epithelial Cells

The thymic epithelium (TE) can be usefully classifi ed

into three broad subtypes: subcapsular/subtrabecular,

cor-tical and medullary thymic epithelial cells Within these

subdivisions, ultrastructural and immunohistochemical

analyses have revealed at least six different subsets (van de

Wijngaert et al., 1983), assigned as “clusters of thymic

epi-thelial staining” (CTES) types I, II, III, IIIB, IIIC, and IV

(Brekelmans and van Ewijk, 1990) based on different mAb

staining profi les

Thus, subcapsular/subtrabecular epithelium consists

of type 1 epithelial cells in a simple epithelial layer These

cells are MHC Class II negative (Boyd et al., 1992) and are

reactive to CTES II mAbs (Godfrey et al., 1990) The

outer-most cortical subpopulation comprises type II epithelial

cells, characterized by their pale appearance in

electronmi-crographs (van de Wijngaert et al., 1984) Immediately

adja-cent to the type II epithelia are type III thymic epithelial

cells, which show intermediate electron lucency (van de

Wijngaert et al., 1984), while the innermost cTEC are type IV

cells, which have high electron lucency and

oval/spindle-shaped nuclei (van de Wijngaert et al., 1984) Ultrastructural

analysis has also revealed large complexes of type II and

type III cells and developing thymocytes (van de Wijngaert

et al., 1984), termed thymic nurse cells (TNC) (Wekerle and

Ketelsen, 1980a, 1980b) No mAbs are currently known to

identify individual subpopulations corresponding to the

types II–IV cells just described; however, all cTEC stain with

CTES III mAbs (van de Wijngaert et al., 1984), and types II

and III cells are strongly MHC Class II positive (Boyd et al.,

1993) Medullary TEC (mTEC) predominantly express

determinants reactive to CTES II and IV mAbs, with type

III cells also identifi ed by ultrastructural analysis (van de

Wijngaert et al., 1984) In addition, type V epithelia,

classi-fi ed as undifferentiated cells, exist in small isolated clusters

at the corticomedullary junction (van de Wijngaert et al.,

1984), along with type VI cells, which have been proposed

to be precursors of differentiated epithelial cells (von

Gaudecker et al., 1986) All mTEC express MHC Class I,

while MHC Class II expression is variable (Jenkinson

et al., 1981; Farr and Nakane, 1983; Bofi ll et al., 1985; Surh

et al., 1992).

The different thymic epithelial subpopulations are also

defi ned by differential expression of cytokeratins (K) Two

cortical populations have been identifi ed: a predominant

K5−K14−K8+K18+ subset and a minor subset also consisting

of K5+K14−K8+K18+ cells (Klug et al., 1998) Most mTEC

display a K5+K14+K8−K18− phenotype (Klug et al., 1998) and

also express the antigen reactive to mAb MTS10 (Godfrey

et al., 1990), and a minor K5−K14−K8+K18+ MTS10−

medul-lary subset is also present (Klug et al., 1998).

While precise functions are not ascribed to all of these TEC subpopulations, the clear functional dichotomy between the cortical and medullary compartments is refl ected in functional differences between the cortical and medullary thymic epithelial cell types Notably, cortical thymic epithelial cells (cTEC) are believed to express a ligand

required for positive selection (Anderson et al., 1994), while

a subset of medullary thymic epithelial cells (mTEC) expresses AIRE1 (AIRE1 positively regulates expression of a cohort of tissue or developmentally restricted genes that play an essential role in the induction of central tolerance)

(M S Anderson et al., 2002) This high level of phenotypic

and functional heterogeneity presents a signifi cant lenge for attempts to support full T-cell development,

chal-including proper repertoire selection in vitro, and is also

pertinent to cell replacement or regenerative strategies for

enhancing thymus activity in vivo.

III IN VITRO T-CELL DIFFERENTIATION

The ability to generate T-cells in culture is widely used

as a tool for investigating the regulation of T-cell

differentia-tion (Hare et al., 1999) and is also of interest for clinical and

pharmaceutical purposes Several methodologies exist that

are based on the use of ex vivo thymic tissue Thus, fetal thymic organ culture (FTOC) utilizes ex vivo thymic lobes

usually derived from E15.5–E16.5 mouse embryos or trimester human fetuses to support the differentiation of T-cell progenitors from endogeneous or exogenous sources

second-(Jenkinson and Owen, 1990; Yeoman et al., 1993; Barcena

et al., 1994; Plum et al., 1994; Cumano et al., 1996)

The technique of reaggregate fetal thymic organ culture (RFTOC), in which defi ned TEC subpopulations are ob-tained by cell purifi cation techniques, reaggregated with

fi broblasts and defi ned lymphocyte populations, and then

cultured further in vitro, was developed as an extension

of FTOC and has proved invaluable for assessing the role

of individual stromal components during specifi c stages of

T-cell maturation (Jenkinson et al., 1992; G Anderson et al.,

1994) In addition, this approach has recently been adapted for testing the potency of different fetal and adult TEC

subpopulations (Bennett et al., 2002; Gill et al., 2002; Rossi

et al., 2006).

Recently, it has been demonstrated that a

tantalum-coated carbon matrix can be used to generate an in vitro

thymic organoid when seeded with ex vivo murine thymic

stromal cells (Poznansky et al., 2000) When these structures

were cocultured with human CD34+ haematopoietic genitors, effi cient generation of mature CD4 and CD8 SP T-cells was observed after 14 days The T-cells generated in this system were functional, as demonstrated by their prolifera-tive response to mitogenic stimuli, and demonstrated a diverse TCR repertoire comparable to that of peripheral

pro-blood T-cells (Poznansky et al., 2000) These fi ndings

dem-onstrate that the utilization of three-dimensional matrices

in conjunction with thymic stromal cells can provide an

effi cient and reproducible method of in vitro T-cell

genera-tion However, this approach currently relies on seeding

with ex vivo thymus tissue and therefore is not highly

scal-able in its present form

In vitro T-cell differentiation has also been investigated

using a derivative of the BM stromal cell line OP-9, which

expresses the Notch ligand Delta-like 1 (OP-9 DL-1) Recent

studies demonstrate that OP-9 DL-1 monolayers can support

the generation of CD4+CD8+ DP thymocytes from mouse

fetal liver–, adult bone marrow–, or ES cell–derived

hema-topoietic progenitors (Schmitt and Zuniga-Pfl ucker, 2002;

Schmitt et al., 2004) Small numbers of CD8+ SP T-cells were

also produced, although CD4+ SP T-cells were largely absent

This system was recently shown to support T-cell

develop-ment from human cord blood– and human bone marrow–

derived CD34+ cells (De Smedt et al., 2002; La Motte-Mohs

et al., 2005) However, although it has been proposed that

this system presents a scalable means of supporting in vitro

T-cell differentiation (Lehar and Bevan, 2002), it remains

unclear to what extent the T-cells generated on OP9-DL1

cells undergo positive and negative selection In an

interest-ing alternative to generation of mature T-cells, this system

has recently been used as a means of expanding a CD4−CD8−

DN precursor thymocyte population, which resulted in

improved T-cell reconstitution after adoptive transfer of

these DN cells in a mouse model of hematopoietic stem cell

transplantation (Zakrzewski et al., 2006) In addition, similar

to RFTOC, the OP-9 DL-1 system is a powerful experimental

tool for examining aspects of T-cell differentiation (Porritt et

al., 2004).

T-cell differentiation in vitro has also been achieved

using preparations of cells derived from human skin in

con-junction with a tantalum-coated carbon matrix (Clark et al.,

2005) In this system, cutaneous keratinocytes and fi

bro-blasts were seeded onto the matrix and supplied with human

CD34+ hematopoietic progenitor cells After a period of

three to four weeks, CD3+ T-cells were produced that

exhib-ited functional maturity and were tolerant to self-MHC as

assessed by the mixed lymphocyte reaction Gene

expres-sion analysis demonstrated that the cells used to seed the

matrix expressed transcription factors associated with TEC

function, such as Foxn1, Aire, and Hoxa3, although at much

lower levels than in the normal thymus However, the effi

-ciency of thymopoiesis in this system was low, with a

rela-tively low number of mature T-cells generated despite the

addition of a cocktail of prolymphopoietic cytokines

IV THYMUS ORGANOGENESIS

Cellular Regulation of

Early Thymus Organogenesis

The thymus arises in the pharyngeal region of the

developing embryo, in a common primordium with the

parathyroid gland This common primordium develops from the third pharyngeal pouch (3PP), one of a series of bilateral outpocketings of pharyngeal endoderm, termed

the pharyngeal pouches, which form sequentially in a

rostral-to-caudal manner

In the mouse, outgrowth of the 3PP occurs from

approx-imately E9.0 (Gordon et al., 2004) At this stage, the

epithe-lium of the 3PP consists of a single layer of columnar epithelium surrounded by a condensing population of neural crest cells (NCC) that will eventually form the capsule

(Le Lievre and Le Douarin, 1975; Jiang et al., 2000) Overt

thymus organogenesis is evident from between E10.5 and E11.0, at which stage the epithelium begins to proliferate,

assuming a stratifi ed organization (Itoi et al., 2001)

Follow-ing this, at E12.5, the primordia separate from the pharynx and begin to resolve into discrete thymus and parathyroid organs The thymus primordium subsequently migrates to its fi nal anatomical location at the midline, while the para-thyroid primodium associates with the lateral margins of the thyroid (Manley and Capecchi, 1995, 1998) In the case of the thymus at least, this migration is active and follows the

path of the carotid artery and vagus nerve (Su et al., 2001).

Within the common primordium, the prospective thymus is located in the ventral domain of the third pouch and the prospective parathyroid in the dorsal aspect Pat-terning of these prospective organ domains appears to occur early in organogenesis, for the parathyroid domain

is delineated by the transcription factor Gcm2 as early as E9.5 Possible mechanisms regulating the establishment/

maintenance of these domains are discussed later

The mesenchymal capsule surrounding the thymus mordium is derived from the migratory neural crest, a tran-sient population formed between the neural tube and the surface ectoderm In the mouse, NCC migrate into the pha-ryngeal region from E9 Elegant chick-quail chimera studies provided the fi rst evidence that NCC are the source of mes-enchymal cells in the thymus (Le Lievre and Le Douarin, 1975), and this was recently confi rmed in the mouse by heri-

pri-table genetic labeling in vivo (Jiang et al., 2000).

Colonization of the mouse thymus with hematopoietic progenitor cells occurs around E11.5 (Owen and Ritter, 1969;

Cordier and Haumont, 1980; Jotereau et al., 1987) Because

vascularization has not occurred by this stage, the fi rst nizing cells migrate through the perithymic mesenchyme into the thymic epithelium These cells have been reported

colo-to exhibit comparatively low T-cell progenicolo-tor activity, while

a second colonizing wave, which arrives between E12 and E14, appears to display much higher levels of T-cell potential

upon in vivo transfer (Douagi et al., 2000).

The epithelial cells within the thymic primordium tinue to proliferate strongly after E12.5, at least partly in response to factors supplied by the mesenchymal capsule

con-Concomitantly, TEC differentiation commences, with the

fi rst evidence of differentiation into cortical and medullary

cell types appearing by E12.5 (Bennett et al., 2002)

De-velopment of the two compartments then proceeds in a

lymphocyte-independent manner until E15.5 (Klug et al.,

2002; Jenkinson et al., 2005) The expression of MHC II

and MHC I on the surface of thymic epithelial cells is fi rst

detected at E13.5 and ∼E16, respectively (Jenkinson et al.,

1981; Van Vliet et al., 1984), and is followed by the

appear-ance of CD4+ and CD8+ SP thymocytes at E15.5 and E17.5

(Jenkinson et al., 1981; Van Vliet et al., 1984) Although a

functional thymus is present in neonates, the full

organiza-tion of the stroma is not achieved until two to three weeks

postnatally in the mouse

Origin of Thymic Epithelial Cells

The precise embryonic origins of the thymic

epithe-lium were until recently a matter of long-standing

contro-versy; confl icting hypotheses suggested that the epithelium

had a dual endodermal/ectodermal origin (Cordier and

Heremans, 1975; Cordier and Haumont, 1980) or derived

solely from the pharyngeal endoderm (Le Douarin and

Jotereau, 1975; Manley and Blackburn, 2003; Blackburn and

Manley, 2004; Gordon et al., 2004) However, recent work

from our laboratories has provided defi nitive evidence for a

single endodermal origin in mice (Gordon et al., 2004)

through histological, fate, and potency analysis of the

pha-ryngeal region These data demonstrated that although the

3PP endoderm and third pharyngeal cleft ectoderm make

contact at E10.5, as proposed in the dual-origin hypothesis,

the germ layers subsequently separate, with apoptosis

oc-curring in the contact region Lineage tracing of pharyngeal

surface ectoderm of E10.5 mouse embryos also failed to fi nd

evidence for an ectodermal contribution to the thymic

pri-mordium, and, fi nally, transplantation of pharyngeal

endo-derm isolated from E8.5–E9.0 embryos (i.e., prior to initiation

of overt thymus organogenesis) indicated that the grafted

endoderm was suffi cient for complete thymus

organogene-sis, similar to previous results obtained using chick-quail

chimeras (Le Douarin and Jotereau, 1975) Thus, pharyngeal

endoderm alone is suffi cient for the generation of both

corti-cal and medullary thymic epithelial compartments

Thymic Epithelial Progenitor Cells (TEPC)

The phenotype of TEPC has been of considerable

interest Evidence suggestive of a progenitor/stem cell

activity was initially provided by analysis of a subset of

human thymic epithelial tumours that were found to contain

cells that could generate both cortical and medullary

sub-populations This suggested that the tumourigenic targets

were epithelial progenitor/stem cells (Schluep et al., 1988)

In addition, ontogenic studies suggested that the early

thymus primordium in both mouse and human might be

characterized by coexpression of markers that later

segre-gated to either the cortical or medullary epithelium (Lampert

and Ritter, 1988) However, the fi rst genetic indication of a

TEPC phenotype was provided by a study addressing the

nature of the defect in nude mice, which fail to develop a

functional thymus due to a single base deletion in the

tran-scription factor Foxn1 (Blackburn et al., 1996) Here, sis of allophenic nude-wild-type aggregation chimeras demonstrated that cells homozygous for the nude mutation

analy-were unable to contribute to the major thymic epithelial

subsets, establishing that the nude gene product (Foxn1) is

required cell-autonomously for the development and/or

maintenance of all mature TEC However, a few

nude-derived cells were present in the thymi of adult chimeras, and phenotypic analysis indicated that these cells expressed

determinants reactive to mAbs MTS20 (Godfrey et al., 1990)

and MTS24 but did not express markers associated with mature TEC, including MHC Class II Based on these fi nd-

ings, we suggested that in the absence of Foxn1, TEC lineage

cells undergo maturational arrest and persist as MTS20+24+

progenitors (Blackburn et al., 1996) This hypothesis was

recently confi rmed by an elegant study that demonstrated that functional thymus tissue containing organized cortical and medullary regions is generated on reactivation of a con-ditional null allele of Foxn1 in the postnatal thymus (Bleul

et al., 2006) Since clonal reactivation of Foxn1 was achieved

in this study, it demonstrates unequivocally that in the absence of Foxn1, some persisting TECs have bipotent pro-genitor activity

Further data regarding the phenotype of thymic lial progenitors came from analysis of mice, with a second-ary block in thymus development resulting from a primary

epithe-T-cell differentiation defect The thymi of postnatal CD3e26tg

mice, in which thymocyte development is blocked at the CD44+CD25− TN1 stage (Hollander et al., 1995), contain

principally epithelial cells that coexpress K5 and K8 (Klug

et al., 1998) — which, in the normal postnatal thymus, are

predominantly restricted to the medulla and cortex tively and are coexpressed by only a small population of cells at the corticomedullary junction In this study, Klug and colleagues (1998) demonstrated that transplantation of

respec-CD3e26tg thymi into Ragl −/− mice, which sustain a later block

in T-cell differentiation, resulted in the development of K5−K8+ cells, suggesting that the K5+K8+ cells are progenitors

of cTEC

More recently, two studies have addressed the typic and functional properties of MTS20+24+ cells within the fetal mouse thymus directly Ontogenic analysis has demonstrated that the proportion of MTS20+24+ epithelial cells was highest in the early thymus primordium, decreas-

pheno-ing to less than 1% in the postnatal thymus (Bennett et al.,

2002), consistent with the expression profi le expected of markers of fetal tissue progenitor cells Phenotypic analysis

of the MTS20+24+ and MTS20−24− populations of the E12.5 thymus indicated that all cells in the MTS20+24+ population coexpressed K5 and K8, while none expressed TEC differen-

tiation markers, including MHC II (Bennett et al., 2002)

Importantly, the functional capacity of isolated MTS20+24+cells and MTS20−24− cells was then determined via ectopic transplantation This analysis demonstrated that

I V T H Y M U S O R G A N O G E N E S I S • 651

the MTS20+24+ population was suffi cient for establishment

of a functional thymus containing both cortical and

medullary TEC populations, while, in this assay, the

MTS20−24− population fulfi lled none of these functions

(Bennett et al., 2002; Gill et al., 2002) The studies described

earlier clearly demonstrated that progenitor activity resided

in the MTS20+24+ population However, although these data

strongly suggested the existence of common thymic

epi-thelial progenitor cells, this was not addressed at the clonal

level in either study

With regard to this issue, a recent study indicates that

during initial organogenesis a bipotent progenitor exists

that can form both cortical and medullary TECs and

sug-gests that this activity may persist in the postnatal thymus

(Bleul et al., 2006) The approach used was to perform a

lineage trace using hK14CreERt2 transgenic mice crossed onto

a ROSA26/silent eYFP background In this system,

sponta-neous Cre recombinase activity in thymic epithelial cells

activated the expression of eYFP in a small number of TECs

at around 14 days postpartum Analysis of the thymi by fl

uo-rescence microscopy revealed that the majority of eYFP+

cells were present in clusters and were not evenly

distrib-uted throughout the stroma, suggesting that fl uorescent

cells were derived from a single recombination event in cells

with proliferative capacity The location of the cell clusters

was mixed; a small proportion of eYFP+ clusters were

restricted to either the cortex or the medulla, but the

major-ity (76%) appeared to span the corticomedullary junction

These results are consistent with the presence of a bipotent

progenitor that may give rise to intermediate progenitors

committed to either the cortical or medullary lineage

However, a major caveat for this interpretation is that stem

cell activity per se was not assayed in these experiments and

that the data would also be consistent with proliferation of

differentiated epithelial cells In an elegant extension of

these experiments, the same hK14CreERt2 deletor strain was

used to reactivate a conditional null allele of Foxn1 in

post-natal mice (discussed earlier) Here, clonal activation of

Foxn1 resulted in the generation of small regions of thymus

tissue that contained both cortical and medullary TEC,

pro-viding conclusive evidence for the existence of a common

progenitor in initial organogenesis

Human Thymus Development

Early human thymus development closely parallels that

of the mouse Thus, the thymus forms from the third

pha-ryngeal pouch in a common primordium with the

parathy-roid gland The third pharyngeal pouch is evident from early

in week 6 of human fetal development, and initially it

devel-ops as a tubelike lateral expansion from the pharynx, which

makes contact with the ectoderm of the third pharyngeal

cleft (Weller, 1933; Norris, 1938) A single endodermal origin

has not been demonstrated directly for the human thymic

epithelium However, since the thymus has a single

endo-dermal origin in mice and avians (Le Douarin and Jotereau,

1975), it is reasonable to assume that this is also the case in humans Within the human common thymus/parathyroid primordia, the thymus and parathyroid domains are located ventrally and dorsally and are surrounded by condensing neural crest–derived mesenchyme from the onset of devel-opment The thymus component of this primordium begins

to migrate ventrally from week 7 to mid-week 8, forming a highly lobulated, elongated, cordlike structure The upper part of this structure normally disappears at separation of the two organ rudiments, leaving the parathyroid in the approximate location in which it will remain throughout adulthood (Norris, 1938) The bilateral thymic primordia continue to migrate toward the midline, where they eventu-ally meet and attach at the pericardium — the permanent location of the thymus into adulthood — by mid-week 8 (Norris, 1938) As in the mouse, the human early thymus primordium appears to contain undifferentiated epithelial

cells (Bennett et al., 2002), which express some markers that

are later restricted to either cortical or medullary ments (Lampert and Ritter, 1988; A Farley and CCB, unpub-lished data) Nascent medullary development is evident from week 8, and by week 16 distinct cortical and medullary compartments are present Other cell types penetrate the thymus from week 8, including mesenchymal, vascular and lymphoid cells, and mature lymphocytes begin to leave the thymus to seed the peripheral immune tissues between weeks 14 and 16 (van Dyke, 1941; Lobach and Haynes, 1986)

compart-Cervical Thymus in Mouse and Human

The presence of a cervical thymus in humans has been recorded for some time (van Dyke, 1941; Tovi and Mares, 1978; Ashour, 1995), and recent publications indicate that

an ectopic cervical thymus is also a common occurrence in

at least some mouse strains (Dooley et al., 2006; Terszowski

et al., 2006) In terms of size and cellularity, the cervical

thymus is much smaller than the thoracic thymus However, the morphology of the two structures is very similar, with organized cortical and medullary regions and similar expres-

sion patterns of cytokeratin molecules (Dooley et al., 2006;

Terszowski et al., 2006) Furthermore, the cervical thymus

expresses the transcription factors Foxn1 and Aire and can produce functional T-cells that are tolerant to self-antigens

(Dooley et al., 2006; Terszowski et al., 2006).

The origin of the cervical thymus is at present unclear

A plausible hypothesis is that it may arise from remnants of the thymus domain of the 3PP that become detached from the organ during separation of the thymus and parathyroid domains Several alternative explanations exist, and the identifi cation of cervical thymi in mice will allow the embry-onic origins of these structures to be addressed experimen-tally It appears that in mice the cervical thymus may mature

postnatally (Terszowski et al., 2006), while in humans it is

clearly present in the second trimester of fetal development

Although the presence of Foxn1+ epithelial cells has not

been reported in the cervical regions of developing mouse

embryos, because it is now clear that cells specifi ed to the

thymic epithelial lineage retain their identity in the absence

of Foxn1 expression (Bleul et al., 2006), Foxn1 may not be an

appropriate lineage marker for tracing the origins of this

tissue

Molecular Regulation of Thymus and

Parathyroid Organogenesis

Although the regulation of thymus organogenesis is

incompletely understood, studies of classical and

geneti-cally engineered mouse mutants have begun to reveal a

network of transcription factors and signaling molecules

that act in the pharyngeal endoderm and surrounding

mesenchyme and mesoderm to regulate thymus and

parathyroid organogenesis The principal components of this network are discussed next and are summarized in Fig 44.2

Molecular Control of Early Organogenesis

The earliest events in thymus organogenesis occur prior

to overt organ development and relate to molecular control

of 3PP formation The T-box transcription factor Tbx1, noic acid (RA) signaling, and fi broblast growth factor 8 (Fgf8) signaling have been implicated as important regulators of this process

reti-Tbx1 was recently identifi ed as the gene responsible for cardiovascular and glandular defects in Df1 mice, which

carry a large deletion of chromosome 16 (Lindsay et al.,

1999) Df1 heterozygotes closely phenocopy a human

FIG 44.2. Molecular regulation of early thymus

organogenesis (A) At approximately E8.0–E8.5

the formation of the third pharyngeal pouch (3PP)

is initiated in the pharyngeal endoderm and is

dependent on the expression of Tbx1 and retinoic

acid (RA) and fi broblast growth factor 8 (Fgf8)

signaling (pink) (B) At E9.5 the 3PP has formed

and is surrounded by mesenchymal cells of

meso-dermal and neural crest cell (NCC) origin

Contin-ued development is dependent on the expression

of the transcription factors Hoxa3, Pax1, Pax9,

Eya1, and Six1 (green) (C) Bone morphogenetic

protein (BMP) (blue) and sonic hedgehog (Shh)

(pink) signaling occur at E10.5 in the 3PP

endo-derm in the ventral and dorsal aspects,

respec-tively These factors may be involved in the

specifi cation of the 3PP into thymus- and

para-thyroid-specifi c domains (D) At E11.5 epithelial

cells in the ventral domain of the 3PP express the

transcription factor Foxn1 (light blue) and will

form the thymic epithelium Epithelial cells in the

dorsal domain express the transcription factor

Gcm2 (purple) and will form the parathyroid gland

The differentiation and maintenance of both of

these cell types are dependent on these factors

I V T H Y M U S O R G A N O G E N E S I S • 653

condition known as 22q11.2 deletion syndrome (22q11.2DS,

or DiGeorge Syndrome), in which a deletion in

chromo-some 22 covering an interval of approximately 30 genes

(Scambler, 2000) results in a range of defects including

thymus aplasia or, more frequently, hypoplasia (Paylor et al.,

2001; Taddei, Morishima et al., 2001) Thus the Df1 mouse

represents a useful model for 22q11.2DS and has allowed

the identifi cation of Tbx1 as a critical early regulator of

pha-ryngeal development

During development, Tbx1 is expressed in the

geal endoderm and the core mesenchyme of the

pharyn-geal arches from approximately E7.5 and continues to be

expressed in a variety of structures until E12.5 (Chapman et

al., 1996; Hu et al., 2004; Xu et al., 2004) Tbx1 mutants have

severe defects in the pharyngeal region, including abnormal

patterning of the fi rst pharyngeal arch; hypoplasia of the

second arch; and absence of the third, fourth and sixth

arches and pouches (Jerome and Papaioannou, 2001) As a

result of this, Tbx1−/− mutants lack both thymus and

para-thyroid and display a spectrum of cardiovascular

abnor-malities and craniofacial defects (Jerome and Papaioannou,

2001) The phenotype of Tbx1−/− animals suggests an

impor-tant role for this gene in the segmentation of the pharyngeal

region Supporting evidence for this hypothesis was

pro-vided by an elegant study addressing the temporal

require-ment for Tbx1 in the developrequire-ment of the pharyngeal region

Deletion of Tbx1 at E8.5, during the formation of the 3PP,

resulted in complete absence of thymus and parathyroid,

and complementary fate-mapping experiments

demonstrated that cells that express Tbx1 at E8.5 contribute signifi

-cantly to the thymic primordium (Xu et al., 2005) However,

although deletion of Tbx1 at E9.5/E10.5 (after initial

forma-tion of the 3PP) caused morphological defects in the thymus,

these were not as severe as the aplasia seen after deletion at

E8.5, and fate mapping of cells expressing Tbx1 cells at E9.5/

E10.5 revealed only a small contribution to the thymus (Xu

et al., 2005) Taken together, these data suggest that Tbx1

is required for establishment of the 3PP but that it is

not directly required for subsequent thymus development

Thus, Tbx1 may infl uence later stages of thymus

organo-genesis in a non-cell-autonomous manner Furthermore,

although Tbx1 is haplo-insuffi cient with respect to thymus

development (Lindsay et al., 2001), the basis of this insuffi

-ciency remains to be determined and may result either from

secondary effects resulting from mild defects in pouch

for-mation or from dosage effects related to factor provision by

non-NCC mesenchymal cells

A role for RA in 3PP formation was suggested by

experi-ments in which RA antagonist was administered to

whole-embryo cultures Here, blockade of RA signaling at E8.0

resulted in the absence of the growth factors Fgf8 and Fgf3

in the 3PP endoderm and impaired NCC migration to the

third and fourth pharyngeal arches (Wendling et al., 2000)

Expression of the transcription factor Pax9 (see later) was

also absent in the third pouch, but was expanded in the

second pouch endoderm These data suggest that RA ing is required for the specifi cation of the third pouch endo-derm, which confers subsequent competence to support

signal-NCC migration In vivo evidence for a role for RA signaling

was subsequently provided by the fi nding that fetal mice lacking RA receptors α and β display thymus agenesis and

ectopia (Ghyselinck et al., 1997).

There is considerable evidence that Fgf8 is required during the early stages of thymus and parathyroid develop-ment Fgf8 is expressed in the early gut endoderm and in the endoderm and ectoderm of the pharyngeal pouches and clefts Mice carrying hypomorphic alleles of Fgf8 show defects in thymus development ranging from hypoplasia to

complete aplasia (Abu-Issa et al., 2002; D U Frank et al.,

2002): The initial impairment in thymus and parathyroid organogenesis is likely to occur at an early stage in develop-ment in these mice, because the third and fourth pharyn-geal arches and pouches are usually hypoplastic/aplastic, in addition to other abnormalities in the pharyngeal region

(Abu-Issa et al., 2002; D U Frank et al., 2002).

In terms of the cell types affected by impaired Fgf8 naling, similarities between the phenotype of Fgf8 hypo-morphs and the spectrum of abnormalities found during experimental NCC ablation (Bockman and Kirby, 1984;

sig-Conway et al., 1997) suggest that the glandular defects may

result from defective NCC migration/differentiation or vival In support of this, NCC of Fgf8 hypomorphs show

sur-increased levels of apoptosis (Abu-Issa et al., 2002; D U

Frank et al., 2002) and reduced expression of Fgf10 (Frank

et al., 2002), a factor that may mediate proliferation of the

pharyngeal endoderm (D U Frank et al., 2002) There is

also a mild reduction in the expression of genes associated

with differentiated NCC (Abu-Issa et al., 2002), suggesting

that the maintenance of NCC is perturbed Taken together, these data implicate Fgf8 in maintaining a competent NCC population that can contribute to thymus organogenesis

However, Fgf8 may also act specifi cally on the 3PP derm, since ablation of Fgf8 in the endoderm and ectoderm

endo-or ectoderm alone results in different phenotypes; ablation

in the ectoderm alone causes vascular and craniofacial

defects seen in Fgf8 hypomorphs (Macatee et al., 2003),

whereas when Fgf8 is also deleted in the endoderm, glandular defects are evident, including thymus hypo-

plasia and ectopia (Macatee et al., 2003) Since the NCC

defects were the same in both cases, Fgf8 may directly infl uence development of the endoderm, although it remains possible that the differences observed result from

a dosage effect

Transcription Factors and Regulation of 3PP Development

After the initial onset of 3PP formation, continued development is dependent on several transcription factors, most notably Hoxa3, Pax1, Pax 9, Eya1, Six1, and Six4 All of these transcription factors are expressed in the 3PP endo-derm from approximately E9.5 to E10.5 and, with the excep-

tion of Pax1 and Pax9, are also expressed in associated NCC

and the ectoderm

Absence of functional Hoxa3 or Eya1 results in the

failure to initiate overt thymus and parathyroid

organogen-esis once the 3PP has formed, revealing the essential roles

of these factors (Manley and Capecchi, 1995, 1998; Xu et al.,

2002; Zou et al., 2006) Tbx1 and Fgf8 are both

down-regulated in the 3PP of Eya1−/− mice at E9.5 (Zou et al., 2006),

indicating that Eya1 plays a role in the regulation of each of

these factors Lack of Six1, Six1 and 4, Pax1, or Pax9 causes

much less severe phenotypes The common primordium

begins to develop in Six1−/− mice, and patterning into thymus

and parathyroid domains (see the next section) is initiated

However, subsequent apoptosis of endodermally derived

cells in the common primordium leads to complete

disap-pearance of the organ rudiment by E12.5 (Zou et al., 2006)

A similar phenotype is evident in Six1−/−;Six4−/− embryos,

though the size of the primordium is further diminished in

the double mutants, indicating synergy between these gene

products (Zou et al., 2006) Loss of function mutations in

Pax9 results in failure of the primordia to migrate to the

mediastinum and in severe hypoplasia from E14.5

How-ever, the thymic lobes are vascularized and contain

lym-phocytes (Hetzer-Egger et al., 2002) Pax1−/− mutants show

relatively mild thymus hypoplasia and aberrant TEC

differ-entiation (Wallin et al., 1996; Su and Manley, 2000; Su et al.,

2001) Since Pax1 and 9 are highly homologous, these

phe-notypes may refl ect functional redundancy, as

demon-strated in other tissues (Peters and Balling, 1999)

Expression of Hoxa3, Eya1 and Six1 in multiple germ

layers complicates interpretation of the respective null

phenotypes for these genes However, Hoxa3 appears to

regulate Pax1 and Pax9 either directly or indirectly, since

Pax1 and Pax9 expression is initiated normally in Hoxa3−/−

mutants but fails to be maintained at wild-type levels

beyond E10.5 (Manley and Capecchi, 1995) Furthermore,

Hoxa3+/−;Pax1−/− compound mutants show delayed

separa-tion of the thymus/parathyroid primordium from the

pharynx, resulting in thymic ectopia and a more severe

hypoplasia than that seen in Pax1−/− single mutants (Su

et al., 2001).

It has been suggested that Eya1 and Six1 act downstream

of the Hox/Pax genes, for Eya1−/− embryos show normal

expression of Hoxa3, Pax1, and Pax9 but reduced expression

of Six1 in the endoderm of the third and fourth pouches

and ectoderm of the second, third, and fourth pharyngeal

arches (Xu et al., 2002) However, while it is likely that Six1

acts downstream of Eya1, recent evidence indicates that

Eya1 and Six1 do not act downstream of the Pax genes This

has been shown by analysis of Pax9−/− and Pax1−/−Pax9−/−

mutants, which show normal expression of Eya1 and Six1

in the 3PP, and of Eya1−/−;Six1−/− double mutants, which

lack expression of Pax1 in the E10.5 3PP (Zou et al., 2006)

Interestingly, Pax9 expression is unaffected in Eya1/Six1

double mutants (Zou et al., 2006) However, it remains

possible that Eya1 and Six1 may be regulated by Hoxa3 pendently of Pax1 and 9 function

inde-The data just reviewed indicate the power of pound-mutant analysis for unraveling genetic interactions However, further work is required to elucidate exactly how these and other transcription regulators cooperate in the development of the 3PP endoderm and to determine the precise role of each of these factors at the cellular level

com-Specifi cation of the Thymus and Parathyroid

Prior to the overt formation of the thymus and roid, the 3PP is specifi ed into organ-specifi c domains At E9.5, epithelial cells within the anterior dorsal aspect initi-ate expression of Gcm2, a transcription factor required for

parathy-the development of parathy-the parathyroid (Gordon et al., 2001)

Current evidence suggests that Gcm2 acts downstream of Eya1 and Hoxa3, because Gcm2 is down-regulated in Hoxa3−/−, Eya1−/− and Hoxa3+/−;Pax1−/− compound mutant

mice (Su et al., 2001; Xu et al., 2002; Blackburn and Manley,

2004) The fi rst thymus specifi c transcription factor, Foxn1,

is expressed at functionally relevant levels in the ventral domain of the 3PP from approximately E11.25, although

low levels can be detected by PCR from E10.5 (Gordon et al., 2001; Balciunaite et al., 2002) Foxn1, a forkhead class tran-

scription factor, is required for TEC differentiation and hair development, and is discussed in more detail in the next section Upstream regulation of Foxn1 is currently not understood, and analysis of mutants in potential upstream regulators have not been informative in this regard Hoxa3−/−and Eya1−/− embryos do not express Foxn1, but this is due to the block in primordium formation prior to the onset of Foxn1 expression Although Foxn1 expression is unaltered

in Pax9−/− and Hoxa3+/−;Pax1−/− mutants (Hetzer-Egger et al.,

2002; Su and Manley, 2000), the possibility remains of dancy between Pax1 and Pax9 Although Six1−/− mutants display reduced Foxn1 expression, the 3PP exhibits increased

redun-cell death in the absence of Six1 (Zou et al., 2006), and

there-fore the reduced expression of Foxn1 may refl ect poor vival of Foxn1+ cells rather than a direct interaction between Foxn1 and Six1 Further studies are required to determine whether a regulatory relationship exists between these genes and Foxn1

sur-The expression patterns of Foxn1 and Gcm2 clearly defi ne the thymus and parathyroid domains of the 3PP However, these factors do not appear to be responsible for specifi cation of their respective organs, indicating that specifi cation must be mediated by an upstream factor or factors With respect to Foxn1, this model is supported by the fi nding that epithelial cells in the ventral domain of the 3PP express the thymus specifi c cytokine IL-7, a factor required for thymocyte differentiation throughout develop-

ment, at E11.5 (von Freeden-Jeffry et al., 1995; Zamisch et

al., 2005), making IL-7 one of the earliest currently

identi-fi ed markers of thymus identity In the thymus primordium, IL-7 is regulated independently of Foxn1, because Foxn1−/−

I V T H Y M U S O R G A N O G E N E S I S • 655

embryos show normal IL-7 expression (Zamisch et al., 2005),

indicating that thymus identity is specifi ed in the absence

of Foxn1 Transplantation experiments also support this

model, because E9.0 pharyngeal endoderm, which has not

yet formed the 3PP, gives rise to a functional thymus when

grafted ectopically (Gordon et al., 2004), indicating that at

this developmental stage some cells are already specifi ed to

the thymic epithelial lineage Although Rhox4 has

previ-ously been identifi ed as an early marker of thymus identity

in the 3PP, the human orthologues of this gene are not

expressed in human thymus organaogenesis, making it

unlikely that it plays a critical role in lineage specifi cation

(Morris et al., 2006).

Similarly, our recent studies conclude that Gcm2 is not

required for specifi cation of the parathyroid, for other

para-thyroid-specifi c markers, including CCL21 and CaSR, are

initiated but not maintained in Gcm2−/− mice (Z Liu, S Yu,

and N R M., unpublished) Thus, Gcm2 may play an

analo-gous role in parathyroid development to that of Foxn1 in the

thymus

How, then, are the thymus and parathyroid domains

within the 3PP established? Evidence suggests that

oppos-ing gradients of bone morphogenetic proteins (BMP) and

sonic hedgehog (Shh) may play an important role in this

process During thymus development, BMP4 expression is

fi rst detected at E9.5, when it is expressed by a small number

of mesenchymal cells in the third pharyngeal arch (Patel

et al., 2006) By E10.5, the BMP4 expression domain has

expanded to include the ventral 3PP endoderm and the

adjacent mesenchyme but remains absent from the dorsal

3PP This expression pattern is maintained at E11.5, and by

E12.5 BMP4 is expressed throughout the thymic

primor-dium and the surrounding mesenchymal capsule (Patel

et al., 2006) The expression pattern of BMP4 suggests that

it may be responsible for the initiation of Foxn1 expression,

because BMP4 is restricted to the ventral domain of the

3PP immediately prior to the onset of Foxn1 expression

Furthermore, Noggin expression is restricted to the dorsal

anterior region of the 3PP at E10.5 and E11.5 and there is

some in vitro evidence that BMP4 can directly regulate

Foxn1 expression (Tsai et al., 2003; A Farley and C C B.,

unpublished data)

In vivo evidence of a role for BMPs in thymus

organo-genesis has been provided by a transgenic approach in

which the BMP inhibitor Noggin is driven by the Foxn1

pro-moter, thus impairing BMP signaling in the thymic stroma

(Bleul and Boehm, 2005) These mice have hypoplastic and

cystic thymi that fail to migrate to their normal position

above the heart It is highly likely that this is due to a direct

effect on the thymic stroma, because impaired development

is evident prior to the immigration of lymphocytes, which

are able to differentiate into T-cells Furthermore, mediators

of BMP signaling, such as Msx1 and phosphorylated

Smad proteins, are down-regulated in both the epithelium

and surrounding mesenchyme, suggesting that impaired

communication between these cell types is the mechanism responsible for the phenotype observed Interestingly, Foxn1 expression was not overtly affected in this transgenic model, arguing against a role for BMP in the regulation of this tran-scription factor However, the inhibition of BMP signaling is driven by the Foxn1 promoter, and thus it may occur after the point at which it is required for initiation of Foxn1 expression In addition, since BMP2 and BMP7 are also expressed in the 3PP and common primordium (C C B and

N R M., unpublished data), it is highly likely that dancy operates between different BMP family members It may thus be more pertinent to view BMPs in terms of the signaling they mediate during organogenesis rather than which specifi c member of the BMP family is involved

redun-During development of the 3PP, expression of the secreted glycoprotein Shh is restricted to cells of the pouch opening at E10.5 and E11.5, although its receptor, Patched1,

is expressed by cells in close proximity to this region Scott and Manley, 2005) Analysis of Shh−/− embryos revealed that both the BMP4 and Foxn1 expression domains are expanded in the 3PP, while the corresponding Gcm2+ para-thyroid domain is lost in these mutants (Moore-Scott and Manley, 2005) Thus, the role of Shh in thymus organogen-esis may be to oppose the action of BMP4 to allow the speci-

(Moore-fi cation and development of the parathyroid

Wnt glycoproteins may also be important in regulating Foxn1 expression Wnts are expressed by the thymic stroma and lymphoid cells, although Wnt receptors are expressed

exclusively by TECs (Balciunaite et al., 2002) The earliest

reported expression of Wnt family members during thymus development is at E10.5, immediately prior to strong Foxn1 expression, when the epithelium of the 3PP and adjacent

cells express Wnt4 (Balciunaite et al., 2002) Given this

expression pattern and the fi nding that TEC lines that

over-express Wnt4 display elevated levels of Foxn1 (Balciunaite et

al., 2002), it is possible that Wnt4 cooperates with BMP4 to

regulate Foxn1 Wnt 1, Wnt 4, and Wnt1, 4-null mice all exhibit hypoplastic thymi characterized by reduced T-cell numbers but normal thymocyte developmental progres-sion However, because no histological analysis of the thymi

in these mutants has been presented, it is not possible to evaluate whether the primary effect is on the thymic epithe-

lium or on thymocytes (Mulroy et al., 2002; Staal and Clevers,

2005) As with BMP family members, functional redundancy

is again a possibility, because other Wnt family members are also expressed in the 3PP and surrounding mesenchyme

(Mulroy et al., 2002; Balciunaite et al., 2002; C C B and

N R M., unpublished data)

Foxn1 and the Regulation of TEC Differentiation

As discussed earlier, the development of functionally mature TECs from the 3PP endoderm is cell autonomously

dependent on Foxn1 (Blackburn et al., 1996) Adult nude

mice, which lack functional Foxn1, retain a cystic, phoid thymus consisting predominantly of apparently

alym-immature epithelial cells (Cordier, 1974; Cordier and

Haumont, 1980; Gordon et al., 2001) These and other data

(see earlier) suggest that lack of Foxn1 results in

develop-mental arrest of thymic-epithelial-lineage cells at the

founder/progenitor-cell stage of development Thus,

whereas TEPCs form independent of Foxn1, their

differen-tiation into mature TEC subtypes depends on it In the

thymus, Foxn1 is expressed by all thymic epithelial cells

throughout development and is maintained postnatally

(Nehls et al., 1996) It is also expressed in the hair follicles

and epidermis, where it is required for normal development

of the skin (Flanagan, 1966)

The precise role of Foxn1 has not been completely

elu-cidated in either the thymic or the cutaneous epithelial

lineages However, considerable evidence suggests that it

regulates the balance between epithelial proliferation and

differentiation Keratinocytes derived from Foxn1−/− mice

have reduced proliferative capacity in vitro and prematurely

express markers associated with terminal differentiation

(Brissette et al., 1996) Furthermore, when Foxn1 is

over-expressed, markers associated with earlier stages of

differ-entiation are up-regulated and later markers are absent

(Brissette et al., 1996) More recently it was shown that

primary human keratinocytes can be induced to initiate

ter-minal differentiation by transient expression of Foxn1 but

that completion of the differentiation program is dependent

on the levels of activated Akt, which may in turn be

regu-lated by Foxn1 (Janes et al., 2004) Thus, in the epidermis

Foxn1 may function to ensure that terminal differentiation

proceeds in a temporally regulated manner Whether this

is the case in the thymus is unclear The early thymic

rudiment in nude mice also shows reduced proliferation

(Itoi et al., 2001) However, the epithelial cells resemble

immature progenitors and do not express markers of

terminal differentiation

There is some evidence to suggest that in the thymus,

Foxn1 is required for lymphoepithelial cross-talk

Compara-tive gene expression analysis between nude and wild-type

thymus suggests that PD1 Ligand is a target of Foxn1,

because it is down-regulated in nude embryos (Bleul and

Boehm, 2001) The receptor for PD1 ligand is expressed by

thymocytes and has been implicated in thymocyte survival,

proliferation and positive selection (Nishimura et al., 1996,

2000), lending weight to the hypothesis that Foxn1 may

regulate genes required for lymphoepithelial cross-talk

A separate publication suggested a role for Foxn1 in

cross-talk by analyzing Foxn1∆/∆ mice, which express a splice

variant lacking exon 3 of the N-terminal domain (Su et al.,

2003) Skin and hair development is normal in these mice,

but the differentiation of TECs is suspended at a time

equi-valent to E13.5 Unlike the nude thymus, the thymic stroma

can attract lymphocytes and mediate T-cell differentiation,

although there are postnatal abnormalities in thymocyte

maturation and a severe reduction in thymocyte numbers

The thymus-specifi c defect in Foxn1∆/∆ mice resembles to

some degree the phenotype seen in hCD3ε26 mice, which have secondary blocks in TEC maturation due to suspended

thymocyte development (Klug et al., 1998, 2002) Based on

this, we suggested that the N-terminal domain is required for lymphoepithelial cross-talk and as such has a thymus-specifi c function However, the impairment of TEC differen-tiation in Foxn1∆/∆ mice is more profound than in hCD3ε26 mutants and occurs prior to the stage when TEC develop-ment is dependant on lymphoepithelial cross-talk It is therefore unclear if Foxn1 directly regulates genes involved

in cross-talk or if its primary role is to regulate TEC entiation, which has downstream effects on lymphoepithe-lial communication

differ-Molecular Regulation of TEC Proliferation

Following the formation of the thymic primordium and the commitment of the epithelial cells to the TEC lineage, the thymus undergoes a period of expansion involving both the proliferation of stromal cells and an increase in thymocyte numbers With regard to TECs, the growth factors

Fgf7 and Fgf10 have been shown to be involved In vitro

experiments have demonstrated that both Fgf7 and Fgf10, which are expressed by the perithymic mesenchyme, can

stimulate the proliferation of fetal TECs (Suniara et al., 2000)

Furthermore, mice lacking Fgfr2IIIb, the receptor for Fgf7 and Fgf10, have severely hypoplastic thymi, although they can support T-cell maturation, and Fgf10−/− mutants also

develop hypoplastic thymi (Revest et al., 2001).

Noncanonical NF-kB Signaling Regulates mTEC Development

The development of mTECs depends on activation of the NF-kB signaling pathway, specifi cally the noncanonical pathway that culminates in RelB activation Medullary TEC development is severely compromised in RelB-defi cient

mice (Burkly et al., 1995; Weih et al., 1995) In the

nonca-nonical RelB activation pathway, ligand engagement of receptors in the TNFR family, such as lymphotoxinβ recep-tor (LTβR), activates NF-kB-inducing kinase (NIK), which phosphorylates homodimers of the downstream kinase, Ikkα Activated Ikkα in turn phosphorylates the C-terminal region of NF-kB2 (p100), leading to ubiquitin-dependent degradation and release of the N-terminal polypeptide, p52 The formation of RelB/p52 heterodimers permits shuttling

of RelB from the cytoplasm into the nucleus, where it

func-tions as a transcriptional regulator (Bonizzi et al., 2004) A

range of medullary defects occurs in mice that are defi cient

in various components upstream of RelB in the alternative NF-kB activation pathway Targeted disruption of the LTβR gene results in disorganized medullary regions that contain

reduced numbers of both major mTECs subsets (Boehm et

al., 2003) Mice with a naturally occurring mutation in NIK

(Barcena et al., 1994) and Ikkα knockout mice have severe defects in medullary formation, including impairment of mTECs development and reduced expression of AIRE and

I V T H Y M U S O R G A N O G E N E S I S • 657

tissue-restricted antigens (Kajiura et al., 2004; Kinoshita

et al., 2006; E Richie, unpublished observations) A similar

medullary phenotype is found in TRAF6 knockout

mice (Akiyama et al., 2005) Interestingly, each of these

mutant strains develops autoimmune manifestations,

indicating a breakdown in the establishment of central

tolerance

Role of Medullary Thymic Epithelial Cells in

Establishing Central Tolerance

Medullary TECs have the unique ability to express genes

encoding a wide array of antigens that initially were thought

to be expressed only in peripheral tissues (Derbinski et al.,

2001) AIRE is expressed by mTECs and plays a prominent

role in regulating tissue-restricted antigen expression In

humans, homozygous AIRE mutations result in a severe,

multiorgan autoimmune disease termed autoimmune

poly-glandular syndrome type 1 (Villasenor et al., 2005) AIRE

knockout mice develop a similar autoimmune phenotype

due to defective clonal deletion, resulting in persistence

of self-reactive thymocytes (Anderson et al., 2002; Liston

et al., 2003) Although AIRE is clearly important for

regula-tion of tissue-restricted antigen expression in mTECs, other

as-yet-undefi ned factors are involved since certain

tissue-restricted antigens are expressed even in the absence of

AIRE (Derbinski et al., 2005) Medullary TECs also affect

central tolerance by supplying self-peptides to medullary

dendritic cells (DCs) in a process termed cross-presentation

(Gallegos and Bevan, 2004) Medullary TECs are also required

for the generation of CD4+CD25+ T regulatory (Treg) cells

and natural killer T-cells, both of which actively repress

self-reactive T-cells (Kronenberg and Rudensky, 2005; Kim et al.,

2006) Given that AIRE and tissue-restricted antigens are

highly expressed by mTECs, it is likely that the autoimmune

phenotype in mice defi cient in various components of the

RelB activation pathway is a result of failed mTEC ment Taken together it is now clear that mTECs contribute directly and indirectly to establishing central tolerance and averting the development of autoimmunity

develop-V SUMMARY

Thymus organogenesis is a complex process in which a three-dimensional organ forms from the endoderm of the 3PP The thymic epithelium, a critical regulator of thymo-poiesis, comprises many subtypes of TEC, all of which arise from a common progenitor Recent studies have begun to clarify the lineage relationships between these cell types and to identify the molecular factors that govern thymus organogenesis However, several signifi cant questions remain unresolved:

• What mechanisms regulate formation of the 3PP from the endoderm?

• How does the network of transcription factors and naling molecules that are expressed in the endoderm and surrounding mesenchyme/mesoderm control sub-sequent 3PP development?

sig-• What factor or factors specify thymus and parathyroid lineages within the 3PP?

• What are the factors responsible for the maintenance, proliferation and differentiation of TEPCs?

Additionally, important questions remain regarding static maintenance of the mature thymus, including eluci-dation of whether the postnatal organ is maintained by a stem cell mechanism, and the cellular and molecular mech-anisms that operate to induce thymic involution Resolution

homeo-of these issues will permit the design homeo-of rational strategies for therapeutic reconstitution of the adaptive immune system The development of such strategies should signifi -cantly impact the health of the aging population and other immunocompromised individuals

VI ACKNOWLEDGMENTS

We wish to thank Lucy Morris (Carnegie Institution of

Washington, D.C, and Howard Hughes Medical Institute,

Baltimore), for critical reading of the manuscript, and our

funding bodies, Leukaemia Research (CCB, CSN), the EU (CCB, CSN), and the NIH (NRM, ER), for support

VII REFERENCES

Abu-Issa, R., Smyth, G., et al (2002) Fgf8 is required for pharyngeal

arch and cardiovascular development in the mouse Development

129(19), 4613–4625.

Akiyama, T., Maeda, S., et al (2005) Dependence of self-tolerance on

TRAF6-directed development of thymic stroma Science 308(5719),

248–251.

Anderson, G., Owen, J J., et al (1994) Thymic epithelial cells provide

unique signals for positive selection of CD4+CD8+ thymocytes in vitro

J Exp Med 179(6), 2027–2031.

Anderson, M S., Venanzi, E S., et al (2002) Projection of an

immuno-logical self-shadow within the thymus by the aire protein Science

298(5597), 1395–1401.

Ashour, M (1995) Prevalence of ectopic thymic tissue in myasthenia

gravis and its clinical signifi cance J Thorac Cardiovasc Surg 109(4),

632–635.

Balciunaite, G., Keller, M P., et al (2002) Wnt glycoproteins regulate the expression of FoxN1, the gene defective in nude mice Nat Immunol

3(11), 1102–1108.

Baldwin, T A., Sandau, M M., et al (2005) The timing of TCR alpha

expression critically infl uences T-cell development and selection

J Exp Med 202(1), 111–121.

Barcena, A., Galy, A H., et al (1994) Lymphoid and myeloid

differentia-tion of fetal liver CD34+-lineage cells in human thymic organ culture

J Exp Med 180(1), 123–132.

Bennett, A R., Farley, A., et al (2002) Identifi cation and

characteriza-tion of thymic epithelial progenitor cells Immunity 16(6), 803–814.

Blackburn, C C., and Manley, N R (2004) Developing a new paradigm

for thymus organogenesis Nat Rev Immunol 4(4), 278–289.

Blackburn, C C., Augustine, C L., et al (1996) The nu gene acts

cell-autonomously and is required for differentiation of thymic epithelial

progenitors Proc Natl Acad Sci U.S.A 93(12), 5742–5746.

Bleul, C C., and Boehm, T (2001) Laser capture microdissection-based

expression profi ling identifi es PD1-ligand as a target of the nude locus

gene product Eur J Immunol 31(8), 2497–2503.

Bleul, C C., and Boehm, T (2005) BMP signaling is required for normal

thymus development J Immunol 175(8), 5213–5221.

Bleul, C C., Corbeaux, T., et al (2006) Formation of a functional thymus

initiated by a postnatal epithelial progenitor cell Nature 441(7096),

992–996.

Bockman, D E., and Kirby, M L (1984) Dependence of thymus

devel-opment on derivatives of the neural crest Science 223(4635), 498–500.

Boehm, T., Scheu, S., et al (2003) Thymic medullary epithelial cell

dif-ferentiation, thymocyte emigration, and the control of autoimmunity

require lympho-epithelial cross-talk via LTbetaR J Exp Med 198(5),

757–769.

Bofi ll, M., Janossy, G., et al (1985) Microenvironments in the normal

thymus and the thymus in myasthenia gravis Am J Pathol 119(3),

462–473.

Bonizzi, G., Bebien, M., et al (2004) Activation of IKKalpha target genes

depends on recognition of specifi c kappaB binding sites by RelB:p52

dimers Embo J 23(21), 4202–4210.

Boyd, R L., Wilson, T J., et al (1992) Phenotypic characterization of

chicken thymic stromal elements Dev Immunol 2(1), 51–66.

Boyd, R L., Tucek, C L., et al (1993) The thymic microenvironment

Immunol Today 14(9), 445–459.

Brekelmans, P., and van Ewijk, W (1990) Phenotypic characterization

of murine thymic microenvironments Semin Immunol 2(1), 13–24.

Brissette, J L., Li, J., et al (1996) The product of the mouse nude locus,

Whn, regulates the balance between epithelial cell growth and

differ-entiation Genes Dev 10(17), 2212–2221.

Burkly, L., Hession, C., et al (1995) Expression of relB is required for

the development of thymic medulla and dendritic cells Nature

373(6514), 531–536.

Chapman, D L., Garvey, N., et al (1996) Expression of the T-box family

genes, Tbx1–Tbx5, during early mouse development Dev Dyn 206(4),

379–390.

Clark, R A., Yamanaka, K., et al (2005) Human skin cells support

thymus-independent T cell development J Clin Invest 115(11),

3239–3249.

Conway, S J., Henderson, D J., et al (1997) Pax3 is required for cardiac

neural crest migration in the mouse: evidence from the splotch (Sp2H)

mutant Development 124(2), 505–514.

Cordier, A C (1974) Ultrastructure of the thymus in “Nude” mice

J Ultrastruct Res 47(20), 26–40.

Cordier, A C., and Haumont, S M (1980) Development of thymus,

parathyroids, and ultimo-branchial bodies in NMRI and nude mice

Am J Anat 157(3), 227–263.

Cordier, A C., and Heremans, J F (1975) Nude mouse embryo:

ectodermal nature of the primordial thymic defect Scand J Immunol

4(2), 193–196.

Cumano, A., Dieterlen-Lievre, F., et al (1996) Lymphoid potential,

probed before circulation in mouse, is restricted to caudal

intraembry-onic splanchnopleura Cell 86(6), 907–916.

De Smedt, M., Reynvoet, K., et al (2002) Active form of Notch imposes

T cell fate in human progenitor cells J Immunol 169(6), 3021–3029.

Derbinski, J., Schulte, A., et al (2001) Promiscuous gene expression in medullary thymic epithelial cells mirrors the peripheral self Nat

Immunol 2(11), 1032–1039.

Derbinski, J., Gabler, J., et al (2005) Promiscuous gene expression in

thymic epithelial cells is regulated at multiple levels J Exp Med 202(1),

Douagi, I., Andre, I., et al (2000) Characterization of T cell precursor

activity in the murine fetal thymus: evidence for an input of T cell

pre-cursors between days 12 and 14 of gestation Eur J Immunol 30(8),

2201–2210.

Farr, A G., and Nakane, P K (1983) Cells bearing Ia antigens in the

murine thymus An ultrastructural study Am J Pathol 111(1), 88–97.

Flanagan, S P (1966) “Nude,” a new hairless gene with pleiotropic

effects in the mouse Genet Res 8(3), 295–309.

Frank, D U., Fotheringham, L K., et al (2002) An Fgf8 mouse mutant

phenocopies human 22q11 deletion syndrome Development 129(19),

4591–4603.

Frank, J., Pignata, C., et al (1999) Exposing the human nude phenotype

[letter] Nature 398(6727), 473–474.

Gallegos, A M., and Bevan, M J (2004) Driven to autoimmunity: the

nod mouse Cell 117(2), 149–151.

Ghyselinck, N B., Dupe, V., et al (1997) Role of the retinoic acid

recep-tor beta (RARbeta) during mouse development Int J Dev Biol 41(3),

425–447.

Gill, J., Malin, M., et al (2002) Generation of a complete thymic

micro-environment by MTS24(+) thymic epithelial cells Nat Immunol 3(7), 635–642.

Godfrey, D I., Izon, D J., et al (1990) The phenotypic heterogeneity of

mouse thymic stromal cells Immunology 70(1), 66–74.

Gordon, J., Bennett, A R., et al (2001) Gcm2 and Foxn1 mark early

parathyroid- and thymus-specifi c domains in the developing third

pharyngeal pouch Mech Dev 103(1–2), 141–143.

Gordon, J., Wilson, V A., et al (2004) Functional evidence for a single

endodermal origin for the thymic epithelium Nat Immunol 5(5),

546–553.

Hare, K J., Jenkinson, E J., et al (1999) In vitro models of T cell

development Semin Immunol 11(1), 3–12.

Hetzer-Egger, C., Schorpp, M., et al (2002) Thymopoiesis requires

Pax9 function in thymic epithelial cells Eur J Immunol 32(4), 1175–

1181.

Hollander, G A., Wang, B., et al (1995) Developmental control point in

the induction of thymic cortex regulated by a subpopulation of

prothy-mocytes Nature 373, 350–353.

Hu, T., Yamagishi, H., et al (2004) Tbx1 regulates fi broblast growth

factors in the anterior heart fi eld through a reinforcing autoregulatory

loop involving forkhead transcription factors Development 131(21),

5491–5502.

V I I R E F E R E N C E S • 659

Itoi, M., Kawamoto, H., et al (2001) Two distinct steps of immigration

of haematopoietic progenitors into the early thymus anlage Int

Immunol 13, 1203–1211.

Janes, S M., Ofstad, T A., et al (2004) Transient activation of FOXN1

in keratinocytes induces a transcriptional program that promotes

terminal differentiation: contrasting roles of FOXN1 and Akt J Cell Sci

117(Pt 18), 4157–4168.

Jenkinson, E J., and Owen, J J (1990) T-cell differentiation in thymus

organ cultures Semin Immunol 2(1), 51–58.

Jenkinson, E J., Van Ewijk, W., et al (1981) Major histocompati bility

complex antigen expression on the epithelium of the developing

thymus in normal and nude mice J Exp Med 153(2), 280–292.

Jenkinson, E J., Anderson, G., et al (1992) Studies on T cell maturation

on defi ned thymic stromal cell populations in vitro J Exp Med 176(3),

845–853.

Jenkinson, W E., Rossi, S W., et al (2005) Development of functional

thymic epithelial cells occurs independently of lymphostromal

interac-tions Mech Dev 122(12), 1294–1299.

Jerome, L A., and Papaioannou, V E (2001) DiGeorge syndrome

phe-notype in mice mutant for the T-box gene, Tbx1 Nat Genet 27(3),

286–291.

Jiang, X., Rowitch, D H., et al (2000) Fate of the mammalian cardiac

neural crest Development 127(8), 1607–1616.

Jotereau, F., Heuze, F., et al (1987) Cell kinetics in the fetal mouse

thymus: precursor cell input, proliferation, and emigration J Immunol

138(4), 1026–1030.

Kajiura, F., Sun, S., et al (2004) NF-kappa B-inducing kinase establishes

self-tolerance in a thymic stroma-dependent manner J Immunol

172(4), 2067–2075.

Kim, H J., Hwang, S J., et al (2006) NKT cells play critical roles in the

induction of oral tolerance by inducing regulatory T cells producing

IL-10 and transforming growth factor beta, and by clonally deleting

antigen-specifi c T cells Immunology 118(1), 101–111.

Kinoshita, D., Hirota, F., et al (2006) Essential role of IkappaB kinase

alpha in thymic organogenesis required for the establishment of

self-tolerance J Immunol 176(7), 3995–4002.

Klug, D B., Carter, C., et al (1998) Interdependence of cortical thymic

epithelial cell differentiation and T-lineage commitment Proc Natl

Acad Sci U.S.A 95(20), 11822–11827.

Klug, D B., Carter, C., et al (2002) Cutting edge:

thymocyte-indepen-dent and thymocyte-depenthymocyte-indepen-dent phases of epithelial patterning in the

fetal thymus J Immunol 169(6), 2842–2845.

Kronenberg, M., and Rudensky, A (2005) Regulation of immunity by

self-reactive T cells Nature 435(7042), 598–604.

Kurobe, H., Liu, C., et al (2006) CCR7-dependent cortex-to-medulla

migration of positively selected thymocytes is essential for establishing

central tolerance Immunity 24(2), 165–177.

La Motte-Mohs, R N., Herer, E., et al (2005) Induction of T-cell

devel-opment from human cord blood hematopoietic stem cells by Delta-like

1 in vitro Blood 105(4), 1431–1439.

Lampert, I A., and Ritter, M A (1988) The origin of the diverse

epithe-lial cells of the thymus: is there a common stem cell? In “Thymus

Update” (M D Kendall and M A Ritter, eds.), pp 5–25 Harwood

Aca-demic, London.

Le Douarin, N M., and Jotereau, F V (1975) Tracing of cells of the avian

thymus through embryonic life in interspecifi c chimeras J Exp Med

142(1), 17–40.

Le Lievre, C S., and Le Douarin, N M (1975) Mesenchymal derivatives

of the neural crest: analysis of chimaeric quail and chick embryos

J Embryol Exp Morphol 34(1), 125–154.

Lehar, S M., and Bevan, M J (2002) T cell development in culture

Immunity 17(6), 689–692.

Lind, E F., Prockop, S E., et al (2001) Mapping precursor movement

through the postnatal thymus reveals specifi c microenvironments

sup-porting defi ned stages of early lymphoid development J Exp Med

194(2), 127–134.

Lindsay, E A., Botta, A., et al (1999) Congenital heart disease in mice

defi cient for the DiGeorge syndrome region Nature 401(6751),

379–383.

Lindsay, E A., Vitelli, F., et al (2001) Tbx1 haploinsuffi cieny in the DiGeorge syndrome region causes aortic arch defects in mice Nature

410(6824), 97–101.

Liston, A., Lesage, S., et al (2003) Aire regulates negative selection of

organ-specifi c T cells Nat Immunol 4(4), 350–354.

Lobach, D F., and Haynes, B F (1986) Ontogeny of the human thumus

during fetal development J Clin Immunol 7, 81–97.

Macatee, T L., Hammond, B P., et al (2003) Ablation of specifi c

expres-sion domains reveals discrete functions of ectoderm- and derived FGF8 during cardiovascular and pharyngeal development

endoderm-Development 130(25), 6361–6374.

Manley, N R., and Blackburn, C C (2003) A developmental look at thymus organogenesis: where do the non-hematopoietic cells in the

thymus come from? Curr Opin Immunol 15(2), 225–232.

Manley, N R., and Capecchi, M R (1995) The role of Hoxa-3 in mouse

thymus and thyroid development Development 121(7), 1989–2003.

Manley, N R., and Capecchi, M R (1998) Hox group 3 paralogs late the development and migration of the thymus, thyroid, and para-

regu-thyroid glands Dev Biol 195(1), 1–15.

Moore-Scott, B A., and Manley, N R (2005) Differential expression of Sonic hedgehog along the anterior–posterior axis regulates patterning

of pharyngeal pouch endoderm and pharyngeal endoderm-derived

organs Dev Biol 278(2), 323–335.

Morris, L., Gordon, J., et al (2006) Identifi cation of a tandem

duplicated array in the Rhox locus on mouse chromosome X

Mammalian Genome, in press.

Mulroy, T., McMahon, J A., et al (2002) Wnt-1 and Wnt-4 regulate

thymic cellularity Eur J Immunol 32(4), 967–971.

Nehls, M., Kyewski, B., et al (1996) Two genetically separable steps in

the differentiation of thymic epithelium Science 272(5263), 886–889.

Nishimura, H., Agata, Y., et al (1996) Developmentally regulated

expression of the PD-1 protein on the surface of double-negative (CD4–

CD8–) thymocytes Int Immunol 8(5), 773–780.

Nishimura, H., Honjo, T., et al (2000) Facilitation of beta selection and

modifi cation of positive selection in the thymus of PD-1-defi cient mice

Owen, J J., and Ritter, M A (1969) Tissue interaction in the

develop-ment of thymus lymphocytes J Exp Med 129(2), 431–442.

Patel, S R., Gordon, J., et al (2006) Bmp4 and Noggin expression during

early thymus and parathyroid organogenesis Gene Expr Patterns 6(8),

794–799.