Báo cáo khoa học: Production of a recombinant mouse monoclonal antibody in transgenic silkworm cocoons pptx

On the other hand, numerous production systems for mAbs have been developed using non-mammalian hosts, Keywords IgG; monoclonal antibody; N-glycosylation; recombinant protein; silk gland

in transgenic silkworm cocoons

Masashi Iizuka1, Shingo Ogawa2, Atsushi Takeuchi1, Shinichi Nakakita3, Yuhki Kubo4,

Yoshitaka Miyawaki4, Jun Hirabayashi3and Masahiro Tomita1

1 Neosilk Co., Ltd, Higashihiroshima, Hiroshima, Japan

2 Research Institute, Koken Co., Ltd, Kita-ku, Tokyo, Japan

3 Life Science Research Center, Kagawa University, Kita-gun, Kagawa, Japan

4 Masuda Chemical Industries Co., Ltd, Takamatsu, Japan

Introduction

mAbs comprise the fastest growing class of therapeutic

proteins; thus, there is an increasing need for their

cost-effective production Current standard procedures for

the production of recombinant mAbs rely on

mamma-lian cell lines as hosts [1] because their use meets current

regulatory requirements However, enormous invest-ment is required for the construction of the bioreactors used to culture the cells and to run the reactors On the other hand, numerous production systems for mAbs have been developed using non-mammalian hosts,

Keywords

IgG; monoclonal antibody; N-glycosylation;

recombinant protein; silk gland

Correspondence

M Tomita, Neosilk Co., Ltd, 3-13-26

Kagamiyama, Higashihiroshima, Hiroshima

739-0046, Japan

Fax: +81 82 431 0654

Tel: +81 82 431 0652

E-mail: mtomita@neosilk.co.jp

(Received 15 June 2009, revised 3 August

2009, accepted 5 August 2009)

doi:10.1111/j.1742-4658.2009.07262.x

In the present study, we describe the production of transgenic silkworms expressing a recombinant mouse mAb in their cocoons Two transgenic lines, L- and H-, were generated that carried cDNAs encoding the L- and H-chains of a mouse IgG mAb, respectively, under the control of the enhancer-linked sericin-1 promoter Cocoon protein analysis indicated that the IgG L- or H-chain was secreted into the cocoons of each line We also produced a transgenic line designated L⁄ H, which carried both cDNAs, by crossing the L- and H-lines This line efficiently produced the recombinant mAb as a fully assembled H2L2 tetramer in its cocoons, with negligible L- or H-chain monomer and H-chain dimer production Thus, the H2L2 tetramer was synthesized in, and secreted from, the middle silk gland cells Crossing of the L⁄ H-line with a transgenic line expressing a baculovirus-derived trans-activator produced a 2.4-fold increase in mAb expression The recombinant mAb was extracted from the cocoons with a buffer containing 3 m urea and purified by protein G affinity column chromato-graphy The antigen-binding affinity of the purified recombinant mAb was identical to that of the native mAb produced by a hybridoma Analysis of the structure of the N-glycans attached to the recombinant mAb revealed that the mAb contained high mannose-, hybrid- and complex-type N-gly-cans By contrast, insect-specific paucimannose-type glycans were not detected Fucose residues a-1,3- and a-1,6-linked to the core N-acetylglu-cosamine residue, both of which are found in insect N-glycans, were not observed in the N-glycans of the mAb

Abbreviations

AAL, Aleuria aurantia lectin; CBB, Coomassie brilliant blue; DsRed, red fluorescent protein; GnT, N-acetylglucosaminyltransferase; HRP, horseradish peroxidase; MGFP, monster green fluorescent protein; MSG, middle silk gland; PA-N-glycans, pyridylaminated-N-glycans; PNGaseF, peptide-N-glycosidase F; PSG, posterior silk gland

including plants [2–4], filamentous fungi [5], chickens [6]

and insect cells [7–9] A single IgG molecule is a

tetra-mer consisting of two H- and two L-chains The

recom-binant mAbs produced by the above non-mammalian

production systems are intact H2L2tetramers with

nor-mal antigen-binding ability N-glycans are attached to

Asn297 of the H-chain constant region in IgG mAbs

Because differences in the structures of these N-glycans

can cause allergic reactions [10] or lead to rapid

clear-ance of the mAbs from the human body [11–13], it is

important to humanize them when the mAbs produced

by non-mammalian hosts are to be used for therapeutic

applications Several attempts have been made to

pro-duce recombinant mAbs with humanized N-glycans

using plants as hosts For example, immunogenic

b-1,2-xylose and a-1,3-fucose residues have been removed

from the glycans by inhibiting b-1,2-xylosyltransferase

and a-1,3-fucosyltransferase, respectively, using RNA

interference or knockout technology [4,14,15]

The silkworm Bombyx mori synthesizes large

amounts of silk proteins in its silk glands and spins

them into silk fibers to build a cocoon This ability to

synthesize silk proteins in large quantities may be

use-ful for the production of recombinant proteins By

increasing the number of reared silkworms, the

proce-dure for protein production can be scaled up with

ease Therefore, the silkworm might be suited as a host

for the mass production of recombinant mAbs

com-pared to mammalian cultured cells and

non-mamma-lian organisms The silk fibers are composed of the

proteins fibroin and sericin, which constitute

approxi-mately 75% and 25%, respectively, of the fiber weight

Fibroin, which constitutes the silk fiber core, is

synthe-sized in the posterior silk gland (PSG) [16] Sericin,

which comprises a group of hydrophilic glue proteins

that surround the fibroin core, is synthesized in the

middle silk gland (MSG) Two sericin genes are known

(ser1 and ser2); however, most sericin proteins are

encoded by ser1 [17–21] One method for generating

germline transgenic silkworms involves the use of

pig-gyBac transposon-derived vectors [22,23] By taking

advantage of PSG- and MSG-specific promoters, we

developed two recombinant expression systems using

transgenic silkworms On the one hand, the

recombi-nant proteins were expressed as fusion proteins with

fibroin in the PSG under control of the fibroin

promoter [23–25] The silk fibers produced by these

silkworms exhibited the properties of both the silk and

the recombinant proteins because the recombinant

pro-teins were embedded in the fibroin fibers On the other

hand, the ser1 promoter was used to express

nant proteins in the MSG In this case, the

recombi-nant proteins were secreted into the hydrophilic sericin

layers without being fused to the silk proteins; thus, they were extractable from the cocoons with mild neu-tral aqueous solutions such as NaCl⁄ Pi or NaCl⁄ Tris [26,27] We previously reported an increase in the expression of recombinant proteins in the MSG using the baculovirus-derived enhancer hr3, the trans-activa-tor IE1 [27] and the 5¢-UTR of baculovirus polyhedrin mRNA [28] Recombinant mRNAs were efficiently transcribed from their transgenes in MSG cells by using both the above enhancer and trans-activator; the amounts of the mRNAs observed reached 30–40% of the endogenous ser1 mRNA level On the other hand, the 5¢-UTR enhanced recombinant protein expression

in the MSG cells at the level of translation, leading to

a 1.5-fold increase in recombinant protein synthesis

In the present study, we generated germline trans-genic silkworms that synthesize both the L- and H-chains of a mouse IgG mAb in their MSG cells and secrete the mAb as an H2L2 tetramer into the sericin layer of their silk fibers Expression of the mAb was increased by introducing the gene encoding the baculo-virus-derived trans-activator The recombinant mAb was extracted and purified from the silk fibers, and the antigen-binding properties of the purified mAb were compared with those of a natural mAb from a hybrid-oma that had been used as the source of the intro-duced IgG genes, demonstrating that the binding properties of the recombinant mAb were identical to those of the hybridoma-derived natural mAb The structures of the N-glycans attached to the recombi-nant mAb were also determined Paucimannose-type N-glycans were not detected, whereas high mannose-, hybrid- and complex-type N-glycans were detected No core fucosylations were found in the N-glycans of the recombinant mAb Major N-glycans in insect cells have paucimannose structures with core fucosylations and high mannose structures, although some variations

in the glycan structure are observed depending on the synthesized glycoproteins Further analysis of the N-glycans from silkworm tissues revealed that the above-described N-glycan structures in the recombinant mAb are a result of the tissue specificity of silk glands

Results

Generation of transgenic silkworms carrying cDNAs encoding a mouse IgG mAb

We constructed two vectors, pIgGL⁄ M1.1MG and

pIg-GH⁄ M1.1R, for the generation of transgenic silkworms expressing a mouse IgG mAb (Fig 1) The former vector contained the cDNA for monster green fluorescent protein (MGFP) as a marker under the control of an

eye- and nervous tissue-specific promoter, 3xP3, plus the

cDNA for the IgG L-chain under the control of the ser1

promoter The latter vector contained the cDNAs for red

fluorescent protein (DsRed) and the IgG H-chain under

the control of the 3xP3 and ser1 promoters, respectively

(Fig 1) pIgGL⁄ M1.1MG and pIgGH ⁄ M1.1R were

injected into 3154 and 2854 eggs, respectively, and the

hatched G0 larvae were allowed to develop to moths G1

embryos from the G0 moths were screened for MGFP or

DsRed fluorescence to obtain transgenic silkworms

Genomic Southern blot analysis of the transgenic

silkworms demonstrated the existence of 13 and 17

independent transgenic lines, respectively, for

pIg-GL⁄ M1.1MG- and pIgGH ⁄ M1.1R in relation to the

chromosomal insertion positions and copy numbers of the

transgenes Transgenic lines with a single-copy transgene

were selected, and the cocoon proteins of the lines were

analyzed by SDS⁄ PAGE The lines with the highest levels

of IgG L- and H-chain expression were used in the

subse-quent experiments as the L- and H-lines, respectively

To generate transgenic silkworms bearing both the

L- and H-chain cDNAs, an L-line worm was crossed

with an H-line worm, and the silkworms in the

subse-quent generation that expressed both MGFP and

DsRed in their eyes were selected The silkworms

car-rying both the L- and H-chain cDNAs were referred

to as L⁄ H-line silkworms

Analysis of recombinant mouse IgG in cocoons

To analyze secreted proteins in the sericin layer of the

silk fibers, all proteins in the layer were dissolved in a

buffer containing 8 m urea, electrophoresed under reducing conditions, and analyzed by western blotting using polyclonal anti-mouse IgG serum Recombinant mouse IgG L- and H-chain was detected in the L- and H-lines, respectively (Fig 2A, lanes 8 and 9) The H-chain was also identified in the H-line proteins by Coomassie brilliant blue (CBB) staining (Fig 2A, lane 4), whereas the L-chain was not, as a result of the presence of endogenous silk proteins with a similar molecular weight (Fig 2A, lane 3) Both L- and H-chains were detected in the cocoon proteins from the L⁄ H-line by CBB staining and western blotting (Fig 2A, lanes 5 and 10) The amount of L- or H-chain in the L⁄ H-line appeared to be higher than that in the L- or H-line The intensity of the H-chain

on the CBB-stained gels was quantified by densitome-try The mean ± SEM amount of H-chain present in 0.1 mg of cocoons from the L⁄ H- and H-lines was

319 ± 1 ng (n = 3) and 139 ± 9 ng (n = 3), respec-tively

To investigate the assembly of the recombinant L- and H-chains, cocoon proteins were analyzed by electrophoresis under nonreducing conditions No L- or H-chain was detected among the cocoon proteins from the L- and H-lines by CBB staining, respectively (Fig 2B, lanes 3 and 4) Western blotting revealed that the L-chain in the L-line cocoons existed as a mono-mer (Fig 2B, lane 8) By contrast, the H-chain was detected as a dimer in the H-line cocoon proteins (Fig 2B, lane 9) Intense bands with an apparent molecular weight that exceeded that of the H-chain dimer were also visible on the blot Although the

ie1

polyA

ie1

polyA

piggyBac

right arm

piggyBac

left arm

IgG H-chain hr3

ser1

DsRed SV40

piggyBac

right arm piggyBac left arm

BmNPVpol 5′-UTR

pIgGH/M1.1R

IgG L-chain hr3

polyA polyA fibL

piggyBac

right arm

piggyBac

left arm BmNPVpol 5′-UTR

pIgGL/M1.1MG

pIE1

Fig 1 Structures of the transformation vectors Three transformation vectors (pIgGL ⁄ M1.1MG, pIgGH ⁄ M1.1R and pIE1) were constructed, each of which contained expression units for selection markers and the recombinant proteins between the right and left arms of piggyBac.

In the selection marker units, the gene encoding DsRed (DsRed) or MGFP (MGFP) was placed between the 3xP3 promoter (P3xP3) and SV40 polyA signal sequence (SV40 polyA) The recombinant protein units were designed to express IgG L- and H-chains in pIgGL ⁄ M1.1MG and pIgGH ⁄ M1.1R, respectively; thus, the L-chain (IgG L-chain) or H-chain (IgG H-chain) cDNA was placed between the BmNPV hr3 enhan-cer, (hr3)-ser1 promoter (Pser1) and fibroin L-chain polyA signal sequence (fibL polyA) The recombinant protein unit in pIE1 was composed

of the ser1 promoter (P ser1 ), IE1 gene (ie1) and ie1 polyA signal sequence (ie1 polyA).

detailed structures of these high molecular weight

products were unclear, the products most likely were

random aggregates of H-chain connected by

inter-chain disulfide bonds CBB staining of the cocoon

pro-teins from the L⁄ H-line revealed a band co-migrating

with the standard IgG H2L2 tetramer (Fig 2B, lanes 5

and 6) In addition to the H2L2 tetramer, small

amounts of H2L and H2were detected among the

pro-teins from the L⁄ H-line by western blotting, which

were also detectable in the standard mouse IgG

(Fig 2B, lanes 10 and 11) Bands with the higher

molecular weight than H2L2, which were assumed to

be derived from random aggregates of the chains, were also detected on the blot as the single expression of the H-chain These aggregates appear in much smaller amounts than the H2L2because the H2L2was detected

as a major product on the CBB-stained gel No L-chain monomer was present among the cocoon proteins (Fig 2B, lane 10) These results suggest that a fully assembled mouse IgG mAb with an H2L2-subunit structure was synthesized in the MSG cells and secreted into the sericin layer of the silk fibers in silk-worms carrying both the IgG L- and H-chain cDNAs

Quantification of L- and H-chain mRNAs in MSGs

As described above, the amounts of L- and H-chain in the cocoons were increased by the co-expression of both chains compared to the expression of either chain We therefore investigated whether these increases arose from an increase in the corresponding mRNAs in the cells Total RNA was extracted from the MSGs of fifth-instar larvae of the L-, H- and L⁄ H-lines, and the L- and H-chain mRNA levels were measured by quanti-tative RT-PCR The amount of sericin-1 mRNA was also determined to allow for normalization of the expression of the L- and H-chains As shown in Table 1, the amount of L- or H-chain mRNA in the

L⁄ H-line was lower than that of the corresponding chain in the L- or H-line, most likely as a result of the co-expression of the two genes from the same promoter These results suggest that the increases in IgG L- and H-chain in the L⁄ H-line cocoons were not caused by the transcriptional regulation of mRNA expression, but

by the regulation of protein synthesis and secretion

Enhanced transgene expression using trans-activator IE1

We previously demonstrated that the baculovirus-derived trans-activator IE1 stimulates the

transcrip-CBB Western

Reducing

1 2 3 4 5 6 7 8 9 1 0 11

H

L

20

30

40

80

50

kD a

H

L

M W L H L/H St W L H L / H S t

Nonreducing

H 2 L 2

H 2 L 2

H 2 L

H 2

L

W L H S t

20

30

40

80

50

120

kD a

220

1 2 3 4 5 6 7 8 9 1 0 1 1

CBB Western

A

B

Fig 2 Analysis of the cocoon proteins in the L-, H- and L ⁄ H-lines.

The proteins in the cocoons of wild-type (W), L- (L), H- (H) or L ⁄ H-line

(L ⁄ H) silkworms were extracted with (A) 8 M urea containing 2%

(v ⁄ v) b-mercaptoethanol and 50 m M Tris–HCl, pH 8.0 (i.e reducing

conditions) or (B) 8 M urea containing 50 mM Tris–HCl, pH 8.0 (i.e.

nonreducing conditions) Aliquots of the extracts were subjected to

SDS ⁄ PAGE Some of the gels were stained with CBB, whereas

others were subjected to western blotting using the rabbit

anti-(mouse IgG) as a primary antibody (western) ‘H2L2’, ‘H2L’, ‘H2’, ‘H’

and ‘L’ to the right of the gel indicate the H 2 L 2 tetramer, H 2 L trimer,

H 2 dimer, H monomer and L monomer, respectively The numbers to

the left of the gel are the molecular masses (kDa) as determined by

the migration of the markers (M) St, commercially available standard

mouse IgG.

Table 1 Copy numbers of mRNAs of the L-chain, H-chain and sericin-1 in MSG cells.

Line L-chain a H-chain a Sericin-1 a

Percentage of L-chain to sercin-1

Percentage of H-chain to sercin-1

L 2.4 ± 0.4 b 0.0 58.3 ± 1.8 4.1 ± 0.6 0.0

L ⁄ H 1.4 ± 0.2 1.6 ± 0.1 47.0 ± 2.5 3.4 ± 0.3 2.9 ± 0.1 a

Copy numbers of mRNAs of L-chain and H-chains of the recombi-nant IgG, and sericin-1 per 10 ng of total RNA The indicated values are 10)5of the actual copy numbers b Data are the mean ± SEM

of the results obtained from three MSGs.

tional activity of the ser1 promoter in the presence of

the baculovirus-derived enhancer hr3 in MSG cells

[27] This mechanism was used to express recombinant

proteins in transgenic silkworms [26,27] However, the

simultaneous introduction of hr3 and ie1 using a single

transformation vector induced the leaky expression of

ie1 in tissues other than the MSG because of the

self-activation of ie1 expression through an interaction

between IE1 and hr3, resulting in high silkworm

mor-tality In the present study, ie1-bearing silkworms were

generated using a transformation vector lacking hr3

but containing ser1 promoter-linked ie1 and crossed

with the L⁄ H-line to obtain silkworms carrying the

genes encoding L-chain, H-chain and IE1 The

resul-tant silkworms, which were designated the L⁄ H ⁄

IE1-line, showed no lethality or abnormalities (data not

shown), and were therefore used in the subsequent

experiments aiming to investigate the increases in

L- and H-chain in the cocoons

The proteins contained in the cocoons of the L⁄

H-and L⁄ H ⁄ IE1-lines were separated by SDS ⁄ PAGE

under reducing conditions and stained with CBB (data

not shown) L- and H-chains in the L⁄ H ⁄ IE1-line

cocoons were more highly expressed than those in the

L⁄ H-line cocoons The intensities of the H-chain bands

on the gels were quantified by densitometry The

mean ± SEM amount of H-chain per 0.1 mg of

cocoon in the L⁄ H- and L⁄ H ⁄ IE1-lines was

319 ± 1 ng (n = 3) and 754 ± 36 ng (n = 3),

respec-tively Thus, the expression of IE1 induced an

approxi-mate 2.4-fold increase in the expression of the IgG

mAb in the silkworms The mAb content in the

cocoons of the L⁄ H ⁄ IE1-line was estimated to be 1.1%

Extraction and purification of recombinant mouse

mAb from cocoons

Recombinant mAb was extracted from L⁄ H ⁄ IE1-line

cocoons at 4C with NaCl ⁄ Pi or a buffered solution

containing urea at a variety of concentrations (in the

range 2–8 m), and the resultant extracts were analyzed

by SDS⁄ PAGE (Fig 3A, lanes 4–10) All the proteins

in the sericin layers were solubilized using 8 m urea and

2% b-mercaptoethanol with heating and then subjected

to SDS⁄ PAGE (Fig 3A, lane 3) The ratios of the

amount of mAb extracted with NaCl⁄ Pi or the

urea-containing solutions to the total amount of mAb in the

sericin layers were calculated by quantifying the band

intensities of the CBB-stained H-chains When the

extracted proteins with NaCl⁄ Pi were analyzed, faint

bands of the H- and L-chains were detected The ratio

of the extracted H-chains to all H-chains in the sericin

layers was estimated to be 8% (Fig 3A, lane 10) In

the case of the extraction with a buffered solution con-taining urea at a concentration of 4 m or less, the amount of the mAb increased as the urea concentration increased; the extraction efficiencies were 18%, 25% and 40% at 2, 3 and 4 m urea, respectively (Fig 3A, lanes 4–6) In the previous studies, recombinant enhanced green fluorescent protein and human serum albumin were efficiently extracted with NaCl⁄ Pi or NaCl⁄ Tris saline from cocoons of transgenic silkworms [26,27] In the present study, however, the addition of urea in the saline was required for the efficient tion of the mAb This difference in the protein extrac-tion may be a result of the difference in the structure of recombinant proteins or their affinities for sericin Endogenous sericin variants were hardly solubilized by urea at concentrations of less than 3 m (Fig 3A, lanes

4, 5 and 10) More than 80% of the mAb was recover-able using a solution containing more than 5 m urea (Fig 3A, lanes 7–9) Under these conditions, however,

a large proportion of the sericin variants were solubi-lized Thus, in our subsequent purification experiment,

L⁄ H ⁄ IE1-line cocoons were treated with a buffered solution containing 3 m urea to reduce the level of contamination in the extract by sericin variants

An L⁄ H ⁄ IE1-line cocoon extract prepared with 3 m urea and 50 mm Tris-HCl (pH 7.4) was dialyzed against 20 mm phosphate buffer (pH 7.0) and sub-jected to protein G affinity column chromatography

As shown in Fig 3B, this process was sufficient to pur-ify the mAb to apparent homogeneity (Fig 3B, lane 5) SDS⁄ PAGE under nonreducing conditions revealed that the purified mAb was fully assembled

H2L2 (Fig 3B, lane 6) As shown in Table 2, we were able to obtain 1.2 mg of purified mAb from 500 mg of

L⁄ H ⁄ IE1-line cocoons

Antigen-binding properties of recombinant mAb The IgG L- and H-chain cDNAs used in the present study were cloned from a mouse hybridoma that pro-duces an IgG mAb against human IgG Therefore, binding of the recombinant mAb to human IgG was analyzed by ELISA to compare the antigen-binding properties of the recombinant mAb with those of the hybridoma-derived one As depicted in Fig 4, both mAbs produced almost identical binding curves to the antigen with similar EC50 values The mouse mAb to human surfactant protein D that was used as a negative control did not bind human IgG at all We also surveyed the binding of the recombinant and hybridoma-derived mAbs to human IgM as a negative control; no binding was detected in either case (data not shown)

Structures of N-glycans attached to recombinant

mAb

The N-glycan profiles of the recombinant mAb

pro-duced by the silkworms were determined The purified

mAb with a protein G column was used for this

deter-mination Protein G, as well as protein A, binds the

CH2 and CH3 domain interface region distal to the

glycosylation site in the CH2 domain of IgG [29], and

the affinity of the IgG-binding proteins for IgG is

unchanged by deglycosylation of IgG [30] Thus, it is

unlikely that any specific mAb glycoform is

preferen-tially selected by purification using protein G

When the pyridylaminated-N-glycans (PA-N-gly-cans) prepared from the mAb were separated by anion exchange chromatography, the glycans were detected only in a flow-through fraction (data not shown), suggesting the absence of negatively-charged saccharides such as sialic acids Subsequently, the PA-N-glycans in the fraction were separated by size fractionation and RP-HPLC Six major PA-N-glycan fractions were obtained, and their structures were analyzed by MALDI-TOF-MS The results obtained are summarized in Table 3 The six PA-N-glycan fractions were identified as GlcNAcMan3GlcNAc2-PA (GNb), Man2Man3GlcNAc2-PA (M5), GlcNAc2Man3 GlcNAc2-PA (GN2), Man3Man3GlcNAc2-PA (M6), Man4Man3GlcNAc2-PA (M7) and Man5Man3 Glc-NAc2-PA (M8) Most of the major N-glycans were high mannose-types such as M5 (51.1%) and M6 (11.9%) On the other hand, significant amounts

of hybrid-type (GNb) and complex-type N-glycans (GN2) having one and two GlcNAc residues at their nonreducing termini, respectively, were detected

at ratios of 18.1% and 11.7%, respectively

Pauci-20

30

40

80

50

60

120

100

kDa

220

Extraction

sM

sA

sP

H

L

Purification

20

30

40

80

50

60

120

100

kDa

220

H

L

H 2 L 2

Urea conc.( M )

2 3 4 5 6 8

Heating

W L/H/IE1

M W L/H/IE1 Extraction Purification

Heating

M

Fig 3 Extraction and purification of recombinant mAb from cocoons (A) Extraction of recombinant mAb from L ⁄ H ⁄ IE1-line cocoons The proteins in the sericin layer of the silk fibers from wild-type (lane 2) or L ⁄ H ⁄ IE1-line (lane 3) silkworms were extracted by maintaining the cocoons at 80 C for 5 min in 8 M urea, 2% (v ⁄ v) b-mercaptoethanol and 50 m M Tris–HCl (pH 8.0), at 10 mg dry weightÆmL)1 The proteins from the L ⁄ H ⁄ IE1-line cocoons were also extracted with 50 m M Tris–HCl (pH 7.4) containing 2, 3, 4, 5, 6 or 8 M urea (lanes 4–9) or NaCl ⁄ Pi (lane 10) at 4 C for 24 h The extracted proteins were separated by SDS ⁄ PAGE and stained with CBB (B) Purification of the recombinant mAb from the L ⁄ H ⁄ IE1-line cocoons Recombinant mAb extracted with 50 m M Tris–HCl (pH 7.4) containing 3 M urea was purified using a protein G column The extract (lane 4) and purified mAb (lane 5) were electrophoresed under reducing conditions The purified mAb was also subjected to SDS ⁄ PAGE under nonreducing conditions (lane 6) The electrophoresed proteins were stained with CBB The numbers to the left of the gel indicate the molecular masses (kDa) as determined by the migration of the markers (M) ‘sM’, ‘sA’, and ‘sP’ to the right of the gel represent sericin M, A and P, respectively ‘H 2 L 2 ’, ‘H’ and ‘L’ represent the H 2 L 2 tetramer, H-chain monomer and L-chain monomer

of the IgG, respectively.

Table 2 Purification of the recombinant mAb from 500 mg of

cocoons.

Purification step

Amount

of mAb (mg)

Recovery (%)

mannose-type N-glycans such as Man3GlcNAc2-PA (M3), which are typically found in insects, were not observed [31] It is also noteworthy that fucose resi-dues linked to the core GlcNAc residue were not detected

To confirm the absence of fucose residues in the N-glycans attached to the mAb, SDS⁄ PAGE with lectin blotting using Aleuria aurantia lectin (AAL) or concanavalin A was performed on the purified recom-binant mAb and standard human IgG from human serum treated with or without peptide-N-glycosidase F (PNGaseF) The AAL lectin used in this analysis rec-ognizes fucose residues a-1,3- and a-1,6-linked to the GlcNAc residue [32] Concanavalin A lectin recogniz-ing mannose was also used as a control CBB stainrecogniz-ing showed that PNGaseF-treated H-chains in both the recombinant mAb and standard human IgG were slightly lower in molecular mass than the correspond-ing untreated chains (Fig 5, lanes 1–4) This indicates that PNGaseF actually removed N-glycans from the H-chains Concanavalin A reacted with both the recombinant and standard H-chains (Fig 5, lanes 5 and 7); however, this reaction was not observed after PNGaseF treatment (Fig 5, lanes 6 and 8), confirming the presence of mannose residues in the N-glycans of both H-chains On the other hand, AAL did not react

Table 3 N-glycan structures of the recombinant mouse mAb produced by silkworms The proposed structure is illustrated using symbols: closed square, open circle and closed circle indicate N-acetylglucosamine, mannose and aminopyridine, respectively ODS, octadecyl silane Abbreviation

of N-glycan

structure

Proposed structure b

Theoretical m ⁄ z (mass + H + ) a

Observed m ⁄ z (mass + H + ) b

% Peak area obtained from HPLC (ODS) c

a

Theoretical m ⁄ z of PA-N-linked glycans was calculated as the monoisotopic mass of (mass + H +

). bObserved m ⁄ z (mass + H)) were obtained from reflector mode MALDI-TOF mass spectra of the labeled N-glycans c % Peak area calculated from the result of RP-HPLC.

Concentration (ng·mL –1 )

Recombinant EC50: 73.9 ng·mL –1

Hybridoma EC50: 99.4 ng·mL –1

Negative control

1.25

1.00

0.75

0.50

0.25

0.00

10 –1 10 0 10 1 10 2 103 10 4

Fig 4 Antigen binding of recombinant mouse mAb The binding of

recombinant mouse mAb to human IgG was analyzed by ELISA.

Recombinant mAb (closed triangles), hybridoma-derived natural

mAb (closed squares) and negative control (anti-human surfactant

protein D mouse IgG 1 ; closed circles) at various concentrations

(3333.33, 1111.11, 370.37, 123.46, 41.15, 13.72, 4.57, 1.52, 0.51

and 0.17 ngÆmL)1) were reacted against human IgG The EC50

val-ues for the binding of the mAbs to the antigen were determined

from binding curves.

with the recombinant H-chain, whereas the standard

H-chain was clearly stained with this lectin (Fig 5, lanes

9 and 11) On the basis of these results, together with

our structural data, we conclude that the N-glycans

attached to the recombinant mouse mAb contained no

detectable a-1,3-linked or a-1,6-linked fucose residues

N-Glycan structures of endogenous proteins in

cocoons and larval tissues

As described above, the recombinant mAb contained

high mannose, hybrid and complex N-glycans On the

other hand, major N-glycans synthesized in insect cell

lines have paucimannose structures with a-1,3- and⁄ or

a-1,6-fucose residues and high mannose structures,

although some variations in the N-glycan structure are

observed depending on the synthesized glycoproteins

[33] To investigate the reason for this difference in

N-glycan structure, we analyzed the N-glycans

con-tained in the cocoons and two larval tissues (MSGs

and fat bodies) of wild-type pnd-w1 silkworms The

results obtained are shown in Table 4

The major N-glycans in the cocoons were M5

(48.5%) and GN2 (36.2%), with small amounts of M3

(1.2%) Fucosylated glycans were not detected in the

cocoons, as in the case of the recombinant mAb Thus,

the N-glycans attached to the endogenous cocoon

proteins were similar to those attached to the mAb

Similar N-glycan structures were noted in the MSG

N-glycans, suggesting that the structural features of

the N-glycans in the cellular glycoproteins of the MSG

cells are comparable to those in the secreted cocoon

glycoproteins The N-glycan structures from the fat

bodies were different from the MSG The major fat

body glycans had a fucosylated paucimannose struc-ture (Man2[Fuc1]GlcNAc2-PA [FM2; 37.4%]) with high mannose structures having more than six man-noses Because FM2 was identified only by MS, it was not possible to determine whether the fucose residues were a-1,3- or a-1,6-linked to the GlcNAc residues The M5 observed in the cocoons and MSGs as a major high mannose-type glycan was not present in the fat bodies We also analyzed the N-glycan struc-tures in the tissues of another silkworm strain, Kinshu, and found that they were essentially the same as those

in the pnd-w1 silkworms (data not shown) From these results, we conclude that the structural features of the N-glycans in the recombinant mAb are attributed to the tissue specificity of the silk glands

Discussion

In the present study, we generated three transgenic lines, L-, H- and L⁄ H-, that synthesized mouse IgG L-chain and H-chain, or both L- and H-chains, respec-tively The L-line silkworms secreted L-chain as a monomer into their cocoons, whereas the H-line silk-worms secreted H-chain as a dimer and higher mole-cular aggregates In the case of the L⁄ H-line, the co-expressed L- and H-chains formed H2L2 tetramers that were secreted as a major product into the cocoons L-chain monomers and H-chain dimers were hardly detected in the L⁄ H-line cocoons The amount of H-chain in the L⁄ H-line cocoons was approximately 2.3-fold higher than that in the H-cocoons Quantita-tive analysis of the H-chain mRNA in the MSG cells revealed that the increase in H-chain in the L⁄ H-line cocoons was not the result of a rise in the mRNA level Thus, H2L2 tetramers were preferentially synthesized and secreted through post-transcriptional regulation

In vertebrate antibody-producing cells, H-chain dimers synthesized in the absence of L-chain expres-sion are not secreted, but are retained within the cells This inhibition of secretion is caused by the stable association of an endoplasmic reticulum-resident stress protein, BiP, with the H-chain dimer [34–36] This mechanism could be present in the MSG cells of silk-worms However, the regulation of IgG secretion may

be insufficient in MSG cells because the L-chain monomers or H-chain dimers were secreted from the cells in the case of the single expression of each chain Similar observations were reported in Drosophila cells transfected with the genes encoding humanized IgG [37] When the H-chain gene was expressed in these cells, H-chain was efficiently secreted as a dimer into the culture medium Furthermore, a Drosophila BiP homolog, hsc72, transiently interacts with the H-chain

KDa

50

25

35

CBB

H

L

AAL

Con A

St

R

St

Fig 5 Analysis of N-glycans in the recombinant mAb by lectin

blot-ting Recombinant mAb was subjected to lectin blotting with AAL

and concanavalin A The purified recombinant mAb (R) and standard

human IgG (St) treated with (+) or without ( )) PNGaseF were

elec-trophoresed on polyacrylamide gradient gels One gel was stained

with CBB (lanes 1–4), whereas the others were subjected to

conca-navalin A (lanes 5–8) or AAL blotting (lanes 9–12) ‘H’ and ‘L’ to the

right of the gel represent the H- and L-chains of IgG, respectively.

The numbers to the left of the gel correspond to the molecular

masses (kDa) as determined by the migration of the markers.

during its secretion [37] Unlike vertebrate BIP,

Drosophila hsc72 dissociates from the H-chain

inde-pendently of the L-chain association, allowing the

secretion of the H-chain as the dimer The silkworm

genome contains a homolog of BiP (NCBI accession

number AB016836), and this gene is expressed in

MSG and PSG cells [38] It is also suggested that

endoplasmic reticulum-resident chaperone proteins

such as BiP are involved in the synthesis and

secre-tion of fibroin in PSG cells [16] Therefore, it is

rea-sonable to assume that the silkworm BiP homolog

and⁄ or other chaperone proteins involved in the

secretion of silk proteins might also function in that

of recombinant IgG Although the function of these

factors in IgG-secretion is not sufficient in MSG cells,

as observed in the Drosophila cells, the factors might

have selectively enhanced the secretion of the IgG

H2L2 tetramers from the cells into the cocoons

Accordingly, we were able to collect the mAb as fully assembled H2L2 tetramers from the cocoons

Previous studies have shown that the major N-gly-cans in insects have pauci and high mannose-structures [9,33,39] Paucimannose-type N-glycans such

as M3 are characteristic of insects and are not found in mammals On the other hand, in the present study, paucimannose-type N-glycans were detected at very low levels in the cocoons and MSGs, whereas N-glycans of this type were present at high levels in the fat bodies Paucimannose-type N-glycans arise from GNb by the removal of a GlcNAc residue by the Golgi membrane-associated enzyme b-N-acetylglucosaminidase [40] In the MSG cells of silkworms, b-N-acetylglucosaminidase activity might be absent or very low, resulting in the nondetection of paucimannose structures in the N-gly-cans of the cocoons On the other hand, it is likely that N-acetylglucosaminyltransferase (GnT)-I and II activity

Table 4 Structures of N-glycans from cocoons, MSGs and fat bodies The proposed structure is illustrated using symbols: closed square, open circle, open diamond, closed triangle and closed circle indicate N-acetylglucosamine, mannose, galactose, fucose and aminopyridine, respectively ODS, octadecyl silane.

Abbreviation of N-glycan structure Proposed structure

% Peak area obtained from HPLC (ODS) a

a

% Peak area calculated from the result of RP-HPLC.

is present in the MSG cells of silkworms as in other

insect cells or tissues [41–43] Therefore, it is reasonable

to find that significant amounts of N-glycans having

GlcNAc residues at their nonreducing termini were

detected among the MSG-synthesized glycoproteins

We also detected large amounts of M5, which is a

pos-sible substrate for GnT-I [44] The accumulation of M5

suggests relatively low GnT-I activity in the MSG cells

No b-1,4-galactose-containing N-glycans were detected

among the MSG-synthesized glycoproteins, implying

little or no b-1,4-galactosyltransferase activity in the

cells This is consistent with previous observations in

other insect cells and tissues [45–47]

One surprising finding obtained from our N-glycan

analysis was the absence of fucose residues among the

MSG-synthesized glycoproteins Previously, the

N-gly-cans of insects such as silkworms were reported to

contain considerable amounts of fucose residues

a-1,3-and⁄ or a-1,6-linked to the core GlcNAc residue

[33,48] For example, the ratios of N-glycans with

a-1,3-fucose and a-1,6-fucose, and both a-1,3- and

a-1,6-fucoses, to the total amount of N-glycans among

the membrane glycoproteins in Sf-21 cells were found

to be 1.8%, 15.1% and 8.8%, respectively [33] The

absence of fucose residues is not attributable to the

silkworm strain used in the present study, but to the

tissue specificity of the silk glands The Drosophila

gen-ome contains genes for a-1,3-fucosyltransferase

(FucTA) and a-1,6- fucosyltransferase (FucT6) [49,50]

These Drosophila enzymes preferred N-glycans with

nonreducing terminal GlcNAc residues as substrates

[49,50] Gene homologs of the Drosophila

fucosyltransferases were also identified from the

silk-worm genome (FucTA homolog, NCBI accession

number CK537398; FucT6 homolog, NCBI accession

number BB987128) Our preliminary analysis

demon-strated that the fucosyltransferase mRNAs were

expressed in the MSG cells (data not shown),

suggest-ing that the absence of the core fucosylation is not a

result of the absence of the fucosyltransferase

expres-sion The shortage of GDP-fucose in the cells might

lead to the prevention of fucosylation

The present study highlights several advantages of

using transgenic silkworms as hosts for the production

of recombinant mAbs Cocoons of transgenic

silk-worms contained fully formed H2L2 tetramers with

appropriate antigen-binding ability The expression of

the mAb was increased up to 1.1% by the introduction

of ie1 into the mAb-expressing silkworms The mAb

was easily extracted and purified from the cocoons

Thus, the present study demonstrated the feasibility of

using transgenic silkworms for the mass production of

recombinant mAbs Silkworms have been used for the

manufacture of silk in the sericultural industry There-fore, the industrial production of recombinant mAbs could be achieved by taking advantage of technologies employed in the sericultural industry, although quality control of the product must be taken into consider-ation The observed structures of the N-glycans further highlight the potential use of transgenic silkworms for mAb production Although the presence of oligoman-nose N-glycans such as M5 is not always favorable for the therapeutic use of the mAb, the absence of the core fucosylation is beneficial Fucose residues a-1,3-linked

to GlcNAc show high antigenicity when administrated

to humans [10] Therefore, the presence of a-1,3-fucose

in the recombinant glycoproteins produced by insect cells has been recognized as an important issue for their use in therapeutic applications In our system, this issue may be solved because no a-1,3-fucose resi-dues were detected among the N-glycans in the recom-binant proteins The absence of a-1,6-fucose comprises yet another reason supporting the use of this system in the production of mAbs The absence of a-1,6-fucose enhances the activity of antibody-dependent cellular cytotoxicity of IgG [51,52] Thus, the present system might be particularly beneficial for the production of therapeutic mAbs, whose main mechanism of action is antibody-dependent cellular cytotoxicity activity

Experimental procedures

Experimental animals

B mori strain pnd-w1 was obtained from the National Institute of Agrobiological Science (Tsukuba, Japan) The larvae were reared at 25C on an artificial diet (Silk Mate

PM, Nosan Corp, Kanagawa, Japan)

Construction of the vectors used to generate the transgenic silkworms

A mouse hybridoma that produces a mouse IgG1 mAb to human IgG was kindly provided by Dr S Usuda (Institute

of Immunology, Tokyo, Japan) Total RNA prepared from the hybridoma using an RNeasy kit (Qiagen, Valencia, CA, USA) was reverse-transcribed to produce cDNA fragments Two cDNA fragments encoding the IgG L- and H-chain variable regions were obtained from the hybridoma cDNAs using a SMART RACE cDNA Amplification Kit (Clon-tech, Palo Alto, CA, USA) in accordance with the manu-facturer’s instructions, and the obtained fragments were sequenced The sequences were used to design PCR primers, and the cDNAs encoding the full-length L- and H-chain ORFs were cloned by PCR from the hybridoma cDNAs The 5¢-UTR sequence of BmNPV polyhedrin [53]