Guided by the hypothesis that struc-tural modifications in the N-glycan alter its capacity to serve as ligand for lectins, we prepared a neoglycoprotein with the extended LEC14 N-glycan a
Trang 1core-fucosylated biantennary N-glycan
Glycoengineering for enhanced cell binding and serum clearance
of the neoglycoprotein
Sabine Andre´1, Shuji Kojima2, Ingo Prahl3, Martin Lensch1, Carlo Unverzagt3
and Hans-Joachim Gabius1
1 Institut fu¨r Physiologische Chemie, Tiera¨rztliche Fakulta¨t, Ludwig-Maximilians-Universita¨t Mu¨nchen, Germany
2 Faculty of Pharmaceutical Sciences, Tokyo University of Science, Japan
3 Bioorganische Chemie, Universita¨t Bayreuth, Germany
N-Glycosylation is the most frequent and structurally
most variegated form of post-translational
modifica-tion [1–3] Ironically, it is exactly due to this
unsur-passed molecular complexity that progress to assign
functional significance to distinct glycan epitope lags
behind the advances of work on other types of protein
modifications Taking a step to change this situation
was the driving force for our study At first glance, we consider it reasonable to interpret the enormous struc-tural complexity of the carbohydrate part of glycopro-teins as a wide array of signals; this concept provides direction for research [4,5] As documented already at the level of nascent glycoproteins, their N-glycan struc-ture is relevant for quality control, underscoring the
Keywords
drug targeting; galectin; glycosylation; lectin;
tumor imaging
Correspondence
S Andre´, Institut fu¨r Physiologische
Chemie, Tiera¨rztliche Fakulta¨t,
Ludwig-Maximilians-Universita¨t Mu¨nchen,
Veterina¨rstr 13, 80539 Mu¨nchen, Germany
Fax: +49 80 2180 2508
Tel: +49 89 2180 2290
E-mail: Sabine.Andre@lmu.de
(Received 16 December 2004, revised 23
February 2005, accepted 2 March 2005)
doi:10.1111/j.1742-4658.2005.04637.x
A series of enzymatic substitutions modifies the basic structure of complex-type biantennary N-glycans Among them, a b1,2-linked N-ace-tylglucosamine residue is introduced to the central mannose moiety of the core-fucosylated oligosaccharide by N-acetylglucosaminyltransferase VII This so-called LEC14 epitope can undergo galactosylation at the b1,2-linked N-acetylglucosamine residue Guided by the hypothesis that struc-tural modifications in the N-glycan alter its capacity to serve as ligand for lectins, we prepared a neoglycoprotein with the extended LEC14 N-glycan and tested its properties in three different assays In order to allow compar-ison to previous results on other types of biantennary N-glycans the func-tionalization of the glycans for coupling and assay conditions were deliberately kept constant Compared to the core-fucosylated N-glycan no significant change in affinity was seen when testing three
galactoside-speci-fic proteins However, cell positivity in flow cytofluorimetry was enhanced
in six of eight human tumor lines Analysis of biodistribution in tumor-bearing mice revealed an increase of blood clearance by about 40%, yield-ing a favorable tumor⁄ blood ratio Thus, the extended LEC14 motif affects binding properties to cellular lectins on cell surfaces and organs when com-pared to the core-fucosylated biantennary N-glycan The results argue in favor of the concept of viewing substitutions as molecular switches for lectin-binding affinity Moreover, they have potential relevance for glyco-engineering of reagents in tumor imaging
Abbreviations
CHO, Chinese hamster ovary; GlcNAc, N-acetylglucosamine; GlcNAc-TVII, N-acetylglucosaminyltransferase VII; LacNAc, N-acetyllactosamine; LCA, Lens culinaris agglutinin; PSA, Pisum sativum agglutinin; TFA, trifluoroacetic acid.
Trang 2notion that the glycan’s intimate interplay with specific
lectins can be central to realization of its functional
potential [6–8] Because glycans differing in structure
by specific substitutions might react with cell lectins in
a characteristic manner, a route of translating
struc-tural differences into effects, e.g leading to altered cell
adhesion, growth control or endocytic uptake, is
envi-sioned Fittingly, the complexity of glycan structures is
matched by expression of lectin families [5,9–11] Our
aim, in essence, is to systematically measure ligand
properties of N-glycans with different substitutions In
so doing we combine the emerging technology for
chemoenzymatic tailoring of complex-type N-glycans
on a preparative scale with biochemical⁄ cell biological
methods, starting with the preparation of
neoglycopro-teins with a homogeneous sugar part as suitable test
substances Our initial studies with complex-type
bian-tennary N-glycans bearing either a bisecting
N-acetyl-glucosamine (GlcNAc) or a core-fucose (Fuc) residue
have lent support to the validity of our hypothesis
[12–14] These two substitutions act like switches on
lig-and properties With these data in hlig-and, the description
of the naturally occurring core-fucosylated N-glycan
containing an additional b1,2-linked GlcNAc moiety
attached to the central Man residue (LEC14) ([15,16];
for glycan structures see lower part of Fig 1)
promp-ted us to take the next step in our program by
examin-ing its properties as ligand
The approach to track down the LEC14 N-glycan
variant actually exploited an impact of N-glycan
sub-stitutions on lectin binding Development of resistance
to two agglutinins with dual affinity to the core-fucose
unit and the trimannosyl core region, i.e Pisum
sati-vum and Lens culinaris agglutinins (PSA, LCA) [17],
led to the selection of the LEC14 mutant of Chinese
hamster ovary cells (CHO) [15,16] Besides substantial
branching in poly(N-acetyllactosamine) chains implied
by the occurrence of 3,4-disubstituted GlcNAc in
methylation analysis the presence of a b1,2-linked
Glc-NAc moiety attached to the central Man residue in the
trimannoside core was defined [18] It apparently
per-turbs binding to PSA⁄ LCA, the likely cause for the
enhanced lectin resistance, and contributes to the
com-plex changes in the glycoproteomic profile of the
LEC14 mutant vs wild-type cells [18] The enzymatic
introduction of the b1,2-linked GlcNAc moiety into
the biantennary N-glycan by
N-acetylglucosaminyl-transferase VII (GlcNAc-TVII) depends critically on
the presence of the core-fucose unit so that the LEC14
type of glycan will invariably harbor two core
substitu-tions [19] An immediate question concerns the
possi-bility of further processing
While the bisecting GlcNAc residue in b1,4-linkage
to the central mannose unit has only been described as
an acceptor for chain elongation in glycans of Mgat2-null mice [20,21], the question on branch elongation of the LEC14-specific b1,2-linked GlcNAc could initially not be answered unequivocally The extensive treat-ment of glycopeptides by b-galactosidase and N-acetyl-b-d-glucosaminidase to trim the glycan antennae prior
to structural analysis of the core may well have also impaired this branch elongation [18] This situation was resolved by total synthesis of a complete LEC14 N-glycan The availability of material derived from chemical synthesis not only unambiguously confirmed this particular core structure but also provided suffi-cient quantities for glycosyltransferase assays [22] Galactosyltransferase was found to elongate the unu-sual b1,2-linked GlcNAc residue ([23]; for structure of the glycan with the new branch see Fig 1) In accord-ance with the high level of resistaccord-ance to glycosidase treatment of this GlcNAc residue (an indication for rather poor spatial accessibility compared with the GlcNAc moieties in the a1,3- and a1,6-arms) its reac-tivity towards galactosyltransferase was lower than for terminal GlcNAc residues in the antennae [23] Thus, the demonstration of substrate properties of the LEC14 core for galactosylation suggests, but does not prove the presence of this type of triantennary N-gly-can in the complex profile observed for CHO cell glycans Its occurrence might account for enhanced glycopeptide binding to immobilized Ricinus communis agglutinin I relative to wild-type glycans [18] Conse-quently, we addressed the ensuing question whether the addition of an N-acetyllactosamine (LacNAc) unit
to a core-fucosylated biantennary N-glycan at the cen-tral Man residue in b1,2-linkage will alter the ligand properties especially for endogenous receptors Because chemical conjugation of the synthetic oligosaccharides
to a carrier protein is feasible for a spacered N-glycan [23], we prepared a neoglycoprotein from an extended LEC14 N-glycan after suitable linker design The lig-and properties of the resulting neoglycoprotein (A) were tested in three different systems: (a) tested as lig-and immobilized to a plastic surface with five sugar re-ceptors, among them growth-regulatory galectins [24]; (b) tested as ligand for surface receptors of different types of tumor cells; and (c) injected into circulation with monitoring of the time course of biodistribution The data set on the neoglycoprotein prepared from a core-fucosylated N-glycan devoid of the b1,2-substitu-tion (C) and tested previously under identical condi-tions [13] allowed direct comparison to pinpoint any detectable influence of the new b1,2-branch
Trang 4Synthetic background
We have set ourselves the task of systematically
deter-mining ligand properties of N-glycans, here the
per-galactosylated LEC14 dodecasaccharide (Fig 1,
compound 7), by combining chemoenzymatic synthesis
with biochemical⁄ cell biological techniques In order to
imitate the natural presentation of N-glycans on a
gly-coprotein, conjugation of the synthetic product to the
carrier protein, which is otherwise free of ligand
prop-erties, was necessary Toward this end we started from
the protected nonasaccharide 1 [22] to reach the
per-galactosylated LEC14 epitope [23] as strategically
outlined in Fig 1 The resulting spacered
dodecasac-charide was conjugated to BSA after the spacer’s
amino function was activated by reaction with
thio-phosgene to its isothiocyanate N-glycan attachment to
the carrier protein was visualized by gel electrophoretic
analysis revealing the N-glycan-dependent shift of the
molecular mass (lower part of Fig 1) The protein was
turned totally into a glycan carrier, because no staining
was visible at the position of unsubstituted albumin
To determine the incorporation yield, additional
ana-lytical procedures were performed MS gave evidence
for a spectrum of neoglycoproteins with one to four
attached N-glycan chains, and the colorimetric assay
determined an average of 3.2 N-glycans per carrier
molecule [23] Of note for the intended comparison to
the other so far tested complex-type biantennary
N-glycans without⁄ with substitutions, the linker design
could thus be kept constant Even more important, the
incorporation yield of this reaction was only slightly
lower than for the N-glycan with a bisecting GlcNAc
moiety (B) or the unsubstituted form (D) at 3.6
N-gly-cans per carrier protein and the core-fucosylated
N-glycan at 3.9 oligosaccharide chains (C) [12–14]
This result, ensuring rather similar glycan density, was
the prerequisite to proceed to testing the ligand
prop-erties of the extended LEC14 dodecasaccharide using
neoglycoprotein A in three different assay systems
with: (a) purified sugar receptors; (b) tumor cell surfa-ces in vitro; and (c) organ lectins in vivo
Affinity to purified lectins/antibodies The first assay system was designed to simulate proper-ties of glycoproteins presented on a cell surface by adsorbing the neoglycoprotein to the plastic surface of microtiter plate wells The homogeneity of the structure
of the synthetic N-glycan, rigorously controlled by our analytical procedures (see Experimental procedures) and definitively excluding the presence of contaminating glycoforms, will account for a clear-cut correlation between structure and ligand properties Also, the assay deliberately avoided surface immobilization of the car-bohydrate-binding proteins, which might affect their binding properties Under these conditions, carbohy-drate-dependent and saturable binding of toxic mistletoe lectin, the growth⁄ invasion-regulatory galectin-1 and the natural autoantibody was invariably detected (Fig 2) The calculated Scatchard plots gave straight lines in all cases, evidence for a single class of binding sites Although the different sugar receptors home in on the same basic unit, i.e terminal galactose, their affinity
is clearly disparate (Fig 2A–C) The IgG subfraction bound with the highest affinity, followed by the plant agglutinin with two binding sites per B-subunit in the (AB)2-tetramer and the homodimeric endogenous lectin with its two binding sites at opposing sides of the pro-tein Galectins afford the opportunity to further exam-ine the relationship between the spatial presentation of carbohydrate recognition domains and ligand affinity
We thus tested two further members of the galectin fam-ily, i.e galectins-3 and -5 These two monomeric pro-teins share ligand specificity with galectin-1 but not its homodimeric cross-linking design Due to their mono-meric constitution in solution no affinity enhancement
by ligand clustering through a bivalent module is expec-ted Indeed, these two lectins were inferior in terms of binding affinity to galectin-1, their KD values at
820 ± 71 nm 3) and 734 ± 157 nm (galectin-5; Fig 2D) with about three to fivefold increases in Bmax
Fig 1 Chemical and enzymatic steps to produce the LEC14-type N-glycan with the LacNAc branch in b1,2-linkage at the central Man unit starting from the protected nonasaccharide 1 [34,35] (a) (NH 4 ) 2 Ce(NO 3 ) 6 , CH 3 CN-H 2 O, 80 C (71%); (b) 1 Ethylenediamine, n-BuOH, 80 C;
2 Ac2O, pyridine; 3 MeNH2(41%) in H2O (96% for steps 1–3); (c) 1 Propanedithiol, NEt3, MeOH; 2 N-carbobenzoxy-6-amino hexanoic acid
4, TBTU, HOBt, N-methylpyrrolidone (31% for steps 1–2 after RP-HPLC); (d) PdO-H2O ⁄ H 2 , MeOH-AcOH (95%); (e) UDP-Gal (4 eq.), b1,4-ga-lactosyltransferase, alkaline phosphatase (75%); (f) 1 Thiophosgene, CH 2 Cl 2 -H 2 O, NaHCO 3 ; 2 BSA, H 2 O, NaHCO 3 ; 6 days The last two schemes for N-glycan sequences allow structural comparison between the N-glycan of neoglycoprotein A and the previously studied N-gly-cans in neoglycoproteins B–D (upper panel) Gel documentation is added in the bottom panel for visualization of the gel electrophoretic mobility of the carbohydrate-free carrier protein BSA (lanes a, c, e; 0.15 lgÆlane)1) relative to that of the neoglycoprotein A (lanes b, d, f; 0.2 lgÆper lane, 0.15 lgÆper lane and 0.1 lgÆper lane, respectively) Positions of two marker proteins for molecular mass designation (in kDa) are indicated by arrowheads.
Trang 5values These results confirm a striking effect of receptor
topology at constant spatial features of the ligand They
also enable us to set the ligand properties of the
exten-ded LEC14 dodecasaccharide in relation to the so far
studied N-glycans, especially the N-glycans without any
substitution or with core-fucosylation
The presence of the new glycan branch in the
exten-ded LEC14 neoglycoprotein appeared to enhance
affinities toward human proteins and reduce affinity
toward the plant lectin relative to the properties of the
complex-type biantennary N-glycan lacking a
substitu-tion However, the strict dependence of GlcNAc-TVII
activity on core-fucosylation in its substrate [19], and
therefore the presence of this substitution in the
LEC14 epitope, must not be neglected The ensuing
comparison between the properties of the LEC14
dodecasaccharide epitope and the core-fucosylated
decasaccharide clearly revealed that the b1,2-linked
glycan branch did not significantly affect affinity to the
three tested types of sugar receptor in this assay
sys-tem In contrast, the consideration of the data for the
N-glycan with bisecting GlcNAc (B) with reductions in
affinity underscores the sensitivity of this parameter to
other structural alterations in the N-glycan Because
the nature of the carbohydrate-binding protein matters
notably in this respect, it is mandatory to proceed to
test the pergalactosylated LEC14 N-glycan against a
complex panel of binding partners In order to extend mapping the ligand properties of the LEC14 dodeca-saccharide, we thus moved in our analysis from a test system with purified sugar receptors to cell surfaces with an array of lectins To add potential clinical rele-vance we selected tumor cells of different histogenesis representing common cancer types In the same way as isolated lectins these established cell models also offer the advantage for comparative analysis when ade-quately controlled for constant surface properties
Affinity to tumor cell surfaces The first step in the cytofluorimetric analysis was dedi-cated to documenting the carbohydrate-dependent and saturable binding of the labeled neoglycoprotein to cell surfaces (Fig 3) Biotinylated carrier protein without the N-glycan failed to produce a signal above back-ground, excluding interactions by the protein part or its label Mimicking the situation when a glycoprotein encounters a cell surface, the neoglycoprotein A reacted with cell surfaces in a cell type-specific manner (Fig 4) As highlighted by these results, distinct pre-ference of binding was determined for the B- and T-lymphoblastoid cells among the set of leukemia⁄ lymphoma lines and to the mammary carcinoma cells among the carcinoma lines, when measuring staining
Fig 2 Illustration of Scatchard plot analysis of carbohydrate-dependent interaction between the carrier-immobilized N-glycan (A) and the mistletoe lectin (VAA; A), human galectin-1 (B), the lactoside-binding IgG subfraction (C) and rat galectin-5 (D) in a representative experimen-tal series K D values (mean ± SD) are given for the complete set of analytical data with at least four different experimental series for each sugar receptor The extent of total binding (s) was reduced by that of binding which was not inhibitable by glycoinhibitors (h, 75 m M lactose and 1 mg asialofetuinÆmL)1) to calculate the level of carbohydrate-dependent binding (n) (see inset).
Trang 6intensity Extending the data of the solid-phase assay,
the LEC14 dodecasaccharide is a ligand for cellular
lectins To give potential reason to the presence of the
GlcNAc-TVII, as detected in the LEC14 mutant [19],
the comparison between binding data of this neoglyco-protein (A) and that presenting the core-fucosylated N-glycan (C) without the b1,2-branch is expedient
In general, comparison to the other, so far tested
100 0
Fig 3 Semilogarithmic representation of
the fluorescent surface staining of cells of
the human T-lymphoblastoid line CCRF-CEM
in the absence of incubation with the
biotin-ylated neoglycoprotein (negative control;
shaded) and after incubation with increasing
concentrations of neoglycoprotein in two
steps: up to 2 lgÆmL)1(lines with
0.5 lgÆmL)1, 1 lgÆmL)1and 2 lgÆmL)1from
left to right); (A) and up to 50 lgÆmL)1(lines
with 2 lgÆmL)1, 5 lgÆmL)1, 10 lgÆmL)1,
25 lgÆmL)1and 50 lgÆmL)1from left to
right); (B) Controls with an incubation step
using biotinylated carrier protein
(25 lgÆmL)1); (C) instead of the
neoglyco-protein and an inhibition of staining using
glycoinhibitors (75 m M lactose and 1 mg
asialofetuinÆmL)1); (D) document lack of
label ⁄ carrier protein-dependent binding and
the carbohydrate dependence of binding.
10 0 10 1 10 2 10 3 10 4 01
10 0 10 1 10 2 10 3 10 4 01
10 0 10 1 10 2 10 3 10 4 01
10 0 10 1 10 2 10 3 10 4
10 0 10 1 10 2 10 3 10 4
64.4 %/62.3 88.4 %/56.1
74.4 %/29.4 83.3 %/420.6
10 0 10 1 10 2 10 3 10 4 01
10 0 10 1 10 2 10 3 10 4 01
10 0 10 1 10 2 10 3 10 4
Fig 4 Semilogarithmic representation of the binding of the fluorescent indicator (streptavidin ⁄ R-phycoerythrin conjugate) in the absence of the probe during processing (negative control; shaded) and after the incubation step with the biotinylated neoglycoprotein (25 lgÆmL)1; black line) for the B-lymphoblastoid line Croco II (A), the T-lymphoblastoid line CCRF-CEM (B), the erythroleukemia line K-562 (C), the acute myelo-genous leukemia line KG-1 (D), the mammary carcinoma line DU4475 (E) and the colon adenocarcinoma lines C205 (F), SW480 (G) and SW620 (H) Quantitative data on percentage of positive cells (%) and mean channel fluorescence are given in each panel.
Trang 7carrier-immobilized N-glycans revealed rather
favora-ble ligand properties (Fig 5) A tendency for enhanced
binding relative to the core-fucosylated N-glycan
indi-cated that the new branch is not an inert modification
at this level of testing The decrease of cell positivity
for KG-1 cells and rather constant results for K-562
cells can be judged as internal controls for cell-type
specificity in the direct comparison to the
core-fucosyl-ated N-glycan Thus, this assay system revealed several
cases with an improvement of ligand properties with
cell type-dependent characteristics Because the
N-gly-can profile of glycoproteins not only governs cell
surface binding in vitro but also serum survival in
circulation, a parameter of interest for prolonging
bioavailability of pharmaproteins, we next tested the
influence of this N-glycan in biodistribution analysis
in vivo
Biodistribution in vivo Organ retention and blood content of the iodinated neoglycoprotein (A) were monitored after intravenous injection Six time points from 15 min to 12 h were set
to determine the time course of these parameters Hep-atic uptake was rapid and the major route of blood clearance (Table 1) When comparing blood⁄ organ retention of this neoglycoprotein for the four major sites to the cases of the so far tested N-glycans (B–D), blood clearance was best for the neoglycoprotein bear-ing the LEC14 dodecasaccharide (A) (Fig 6) We have deliberately run these experiments in tumor-bearing mice to look at tumor uptake relative to blood back-ground, a factor with impact on sensitivity of tumor imaging The tumor⁄ blood ratio after 1 h was 0.7 for neoglycoprotein (A) but 0.53 for C bearing a core-fucosylated N-glycan with 3.11 ± 0.17% in blood and 1.65 ± 0.06% in the tumor The detected difference adds to the evidence for modulation of ligand proper-ties by the extended LEC14 motif Placing this glycan
at best position in this respect, the ratio for the unsub-stituted nonasaccharide N-glycan (D) was 0.38 (tumor: 1.16 ± 0.09; blood: 3.07 ± 0.06%) and 0.47 (tumor: 1.47 ± 0.09; blood: 3.11 ± 0.14%) for the decasac-charide with the bisecting GlcNAc moiety (B) Regard-ing the individual organ sites no major alteration of uptake and retention after 1 h was detectable except for the N-glycan with bisecting GlcNAc (Fig 6)
Discussion
The basic complex-type biantennary N-glycan is sub-ject to enzymatic substitutions Structural aspects have been mostly clarified but the functional significance of their presence is still a matter of debate Our hypothe-sis interprets occurrence of substitutions as a means to modulate ligand properties in interactions with endo-genous lectins The versatile potential for fine-tuning a respective information transfer would then clearly out-weigh the required investment in coding for the diver-sity of glycosyltransferases and regulation of their activity In other words, glycosyltransferase activities produce lectin-binding epitopes By virtue of adding substitutions, which may not even directly participate
in binding, they might also affect the affinity of the binding sites To demonstrate that a structural alter-ation in an N-glycan changes its binding parameters requires experimental evidence difficult to collect with natural glycoproteins They generally present more than one type of glycan chain and exhibit microhetero-geneity, confounding efforts to establish a direct struc-ture⁄ activity profile To address this issue, we turned
Fig 5 Comparison of the percentage of positive cells (upper panel)
and mean channel fluorescence (bottom panel) in flow
cytofluori-metric analysis using the LEC14-dodecasaccharide-bearing
neogly-coprotein A and neoglyneogly-coproteins with complex-type biantennary
N-glycan ligand parts substituted by bisecting GlcNAc (B) or
core-fucosylation (C) or without any substitution (D) (see Fig 1 for
struc-tural comparison) Data for neoglycoproteins B, C, D have
previously been published [12–14] and are shown for comparison.
The standard deviations within experimental series are generally
below 7.5%.
Trang 8to the synthesis of neoglycoproteins In contrast to free
N-glycans they harbor a homogeneous sugar part and
favorably maintain a local density akin to natural
gly-coproteins In fact, affinity to galectins is sensitive to
changes in local density of glycan chains and improved
by certain modes of clustering [25–27]
As stated above, it is our aim to delineate
struc-ture⁄ activity profiles for N-glycans Toward this end, we
have so far tested three types of biantennary N-glycans
in different assay systems [12–14] They were confronted
with different situations in vitro and in vivo, i.e the
N-glycan as ligand in a matrix simulating a cell membrane
or in solution⁄ serum confronted with lectin-presenting
cell surfaces Evidently, monitoring tumor cells and bio-distribution has relevance to the glycan’s suitability for drug targeting or imaging [28–30] Our previous results with the neoglycoprotein probes, which were kept rather constant in all relevant features (nature of carrier pro-tein, linker chemistry, yield of glycan incorporation), supported the hypothesis given above [12–14] In this report, we examined the impact of the pergalactosylated LEC14 motif, a b1,2-linked LacNAc disaccharide emer-ging from the central bMan unit of a core-fucosylated N-glycan [7] (Fig 1) The bMan moiety of a LEC14 N-glycan is substituted in positions 2, 3 and 6, a remarkable illustration of the capacity of glycan units to engender branching, in contrast with amino acids and nucleotides As outlined in the introduction, a dis-tinct N-acetylglucosaminyltransferase (GlcNAc-TVII) is responsible for the introduction of the b1,2-GlcNAc into mammalian N-glycans [19] Of note, a b1,2-substi-tution at this site of the core is also found in nonmam-malian N-glycan chains, here with xylose as added sugar unit [31–33] This position is thus apparently predis-posed for enzymatic modification Immunologically, this residue is relevant due to its allergenic activity, an indi-cation for accessibility to interactions with immunoglo-bulin E [33] Likewise, the b1,2-GlcNAc residue at this position of a mammalian-type core-fucosylated N-gly-can is a contact point, as shown by its acceptor capacity
in enzymatic galactosylation [23] Moreover, the selec-tion process to isolate the LEC14 mutant cells exploited the detrimental effect of this core substitution on an interaction with a receptor protein, i.e reduction of binding of the plant agglutinins PSA⁄ LCA [15,16] Our results with purified lectins⁄ antibodies reveal
no major influence of this enzymatically elongated branch on ligand properties, when compared with the core-fucosylated N-glycan lacking this branch
Table 1 Biodistribution of LEC14-dodecasaccharide-bearing neoglycoprotein A in Ehrlich-solid-tumor-bearing mice (% injected doseÆg)1 tissue) Each value indicates the mean ± SD for four to five mice.
Fig 6 Comparison of aspects of the biodistribution patterns of
iodi-nated neoglycoproteins with the following complex-type N-glycan
ligand parts 1 h afer injection: LEC14 dodecasaccharide (A) and
complex-type biantennary N-glycans substituted by bisecting
Glc-NAc (B) or core-fucosylation (C) or without any substitution (D) (see
Fig 1 for structural comparison) Data for neoglycoproteins B, C, D
have previously been published [12–14] and are shown for
compar-ison The range of the standard deviation shown for each result by
bars was between 1.95 and 21.6%.
Trang 9Looking at the growth⁄ invasion-regulatory galectin-1,
it might be that the necessary contact to the
subter-minal GlcNAc residue during binding, a factor
contri-buting to ligand selection [34], is spatially hindered
No affinity enhancement relative to the
core-fucosyl-ated decasaccharide was detectable The affinity of
binding of the monomeric galectins-3 and -5 was
con-siderably reduced, arguing in favor of an influence of
the strong cross-linking activity of galectin-1 as a clue
for the functional divergence noted in a tumor cell
system [35,36] No indication for positive
cooperati-vity of galectin-3 binding was observed This binding
mode was operative with laminin as substratum for
this generally monomeric lectin which can form a
small extent of pentamer in solution [37,38] When
testing cell surfaces with their full array of
carbohy-drate-binding proteins, a clear impact of presence of
the new branch was determined This effect hinged
on the cell type, preferentially leading to increased
binding relative to the core-fucosylated decasaccharide
as ligand In addition to its principal value to
delin-eate evidence for a structure⁄ activity correlation this
result signifies that cell-presented lectins in most of
these tumor lines do not share the core specificity
with the plant agglutinins PSA⁄ LCA which would
have been impaired by introducing the b1,2-branch
Our result underscores differences between plant and
mammalian lectins and recommends using
endo-genous lectins for functional glycoproteomic profiling
of clinical samples [39]
The evidence for a contribution of this b1,2-branch
in the LEC14-type dodecasaccharide to overall ligand
properties was supported by the biodistribution
analy-sis, revealing rapid clearance elicited by
pergalactosyl-ated LEC14 epitope In contrast to galectins the
C-type endocytic receptor of hepatocytes
accommo-dates galactose as central contact point [40] This result
can be relevant for an application Actually, tailoring
of the glycan part of pharmaproteins
(glycoengineer-ing) has become a fertile field of research in order to
manipulate cellular uptake and serum half-life [41–47]
The measured rapid clearance of the respective
neogly-coprotein bearing a b1,2-branch constituted by a
Lac-NAc disaccharide can be advantageous when using an
iodinated glycoprotein for imaging, as it lowers the
background The detection of this property
immedi-ately raises the question of how this parameter will be
altered when the b1,2-branch is shifted away from the
central Man unit to the Man residues in the branch
extensions by GlcNAc-TIV or GlcNAc-TV Indeed,
the consequences of hereby generating the two natural
versions of triantennary N-glycans as part of
neoglyco-proteins have not yet been rigorously determined using
our panel of assays Thus, it is our next challenge to address this issue
Experimental procedures
Synthetic and analytical procedures
NMR spectra were recorded on a Bruker AMX 500 spectro-meter (Karlsruhe, Germany) HPLC separations were per-formed on a Pharmacia LKB gradient system 2249 equipped with a Pharmacia LKB Detector VWM 2141 (Freiburg, Germany) For size exclusion chromatography a Pharmacia
Hi Load Superdex 30 column (600· 16 mm) was used, RP-HPLC was performed on a Macherey-Nagel Nucleogel
RP 100–10 column (Du¨ren, Germany, 300 · 25 mm) Carbohydrate-free BSA and bovine b1,4-galactosyltrans-ferase were purchased from Sigma (Munich, Germany), alka-line phosphatase (calf intestine, molecular biology grade) from Roche Diagnostics (Heidelberg, Germany) UDP-galactose was a generous donation from Roche Diagnostics ESI-TOF mass spectra were recorded with methanol⁄ water
as solvent on a Micromass LCT spectrometer connected to
an Agilent HP 1100 HPLC apparatus MALDI-TOF mass spectra were recorded on a Bruker Reflex III using linear mode and an acceleration voltage of 20 kV For sample preparation in MALDI-TOF-MS a solution of the neogly-coprotein (1 lL, 7 mgÆmL)1) in 0.1% (v⁄ v) trifluoroacetic acid (TFA) was mixed with 1.5 lL of 33% acetonitrile in 0.1% (v⁄ v) TFA and 2.5 lL of a saturated solution of sinapinic acid in 0.1% (v⁄ v) TFA and dried in high vacuum The structures of the synthetic N-glycans were routinely confirmed by the following 2D-NMR-experiments: TOCSY, NOESY, HMQC, HMQC-COSY, HMQC-DEPT, and HMQC-TOCSY Signals of NMR spectra were assigned according to the following convention including designation
of spacer atoms illustrated for compound 7 in Fig 1
Preparation of neoglycoprotein A
For conjugation of the derivatized dodecasaccharide to the carrier protein the amino group was transformed into its isothiocyanate In a 1.5 mL plastic vial 6-aminohexanoyl-N-glycan 7 (0.77 mg, 0.34 lmol) was dissolved in sodium hydrogencarbonate (200 lL, 10 mgÆmL)1) followed by addi-tion of dichloromethane (200 lL) and thiophosgene (5 lL, 19.7 lmol) The biphasic mixture was vigorously stirred After 4 h (TLC: isopropanol⁄ 1 m ammonium acetate, 2 : 1) the mixture was centrifuged and the aqueous phase was sep-arated Subsequently, the organic phase was extracted twice with sodium hydrogencarbonate (100 lL, 10 mgÆmL)1) The combined aqueous phases were extracted twice with dichlo-romethane (500 lL) Carbohydrate-free BSA (2 mg) was dis-solved in the aqueous solution of the isothiocyanate, and the reaction vial was kept for 6 days at ambient temperature
Trang 10The reaction mixture was centrifuged, and the clear
superna-tant was fractionated by gel filtration (Pharmacia Hi Load
Superdex 30 (600· 16 mm); eluent: 0.1 m ammonium
hydro-gencarbonate; flow rate 0.75 mLÆmin)1; detection: 214 and
254 nm) Further quality controls were performed by gel
electrophoretic analysis and colorimetric determination of
the average glycan content as described previously [12–14]:
yield, 2.07 mg; Rf amine 7¼ 0.14 (i-propanol ⁄ 1 m
ammo-nium acetate, 2 : 1); Rf isothiocyanate¼ 0.50 (i-propanol ⁄
1 m ammonium acetate, 2 : 1) MALDI-MS: Mcalcd¼ 68694,
70957, 73220, 75484 (1, 2, 3, 4 N-glycans per BSA molecule);
Mfound¼ 68649, 70971, 73242, 75475 (1, 2, 3, 4 N-glycans
per BSA molecule)
The neoglycoprotein A was then used in solid-phase and
cell-binding assays either free of label or for labeling with
the N-hydroxysuccinimide ester derivative of biotin under
conditions identical to the preparation of the other
N-gly-can-bearing probes tested previously [12–14]
Solid-phase assay
The matrix for the assay was established by adsorption of
neoglycoprotein to the surface of plastic microtiter plate
wells under conditions used previously [12–14] Controls for
standardizing coating density were performed with
biotinyl-ated neoglycoprotein using streptavidin–peroxidase
conju-gate as indicator Ligand properties of the N-glycan were
probed with different types of carbohydrate-binding
pro-teins The galactoside-specific agglutinin from mistletoe
(Viscum album L agglutinin, VAA), human galectin-1,
murine galectin-3 and rat galectin-5 as well as the
immuno-globulin G subfraction with preferential affinity to
b-gal-actosides from human serum were isolated and checked for
purity and quaternary structure by gel electrophoresis and
filtration, electrospray ionization MS, ultracentrifugation
and haemagglutination [48–54] Biotinylation was carried
out under activity-preserving conditions, and label
incor-poration was assessed by binding assays with streptavidin–
peroxidase conjugate or a proteomics protocol [48,55]
Binding studies of the sugar receptors to the
glycan-present-ing matrix were performed by stepwise increases of probe
concentration up to saturation with duplicates at each
con-centration and at least four independent series including
controls to determine extent of carbohydrate-dependent
binding by its inhibition using a mixture of 75 mm lactose
and 1 mg asialofetuinÆmL)1, and the data sets were
algebra-ically transformed to obtain KD values and the number of
bound sugar receptor molecules at saturation (Bmax),
fol-lowing the protocol of our previous reports on
neoglyco-proteins with synthetic N-glycans [12–14]
Cell-binding assay
Using the biotinylated neoglycoprotein as probe, automated
flow cytofluorimetric analysis of carbohydrate-dependent
cell surface binding was performed with the following human tumor lines: Croco II (B-lymphoblastoid cell line), CCRF-CEM (T-lymphoblastoid cell line), K-562 (erythro-leukemia cell line), KG-1 (acute myelogenous (erythro-leukemia cell line), DU4475 (mammary carcinoma cell line) as well as C205, SW480 and SW620 (colon adenocarcinoma cell lines) Except for the Croco II line established in our labor-atory [56] the cells were commercially available (American Type Culture Collection, Rockville, MD, USA) and rou-tinely cultured under the recommended conditions The adherent colon carcinoma cells were detached by exposing them to NaCl⁄ Picontaining 2 mm EDTA Prior to the ana-lysis cells were routinely washed carefully with Dulbeccos’s NaCl⁄ Pisolution containing 0.1% (w⁄ v) carbohydrate-free BSA to remove any inhibitory serum glycoproteins and to saturate nonspecific protein-binding sites For this purpose,
an incubation step with ligand-free carrier protein for
30 min at 4C was added prior to the incubation with the labeled neoglycoprotein at this temperature to minimize uptake by endocytosis Carbohydrate-dependent binding of the neoglycoprotein (25 lgÆmL)1) to the cells (8· 106 cellsÆmL)1) was assessed in a FACScan instrument (Becton-Dickinson, Heidelberg, Germany) with the fluorescent indicator conjugate streptavidin⁄ R-phycoerythrin (1 : 40; Sigma) Controls to assess carbohydrate-independent bind-ing of the carrier via its protein part or label and to docu-ment sugar inhibition were run in each series, as previously described [12–14]
Analysis of in vivo biodistribution
Radiolabeling of the neoglycoprotein was performed by the chloramine-T method [57] Briefly, fresh chloramine-T and sodium metabisulfite solutions were prepared, and the neo-glycoprotein was dissolved in NaCl⁄ Pi (pH 7.2) at a con-centration of 1 mg proteinÆmL)1 A 10 lL portion of
125I-labelled NaI (74 MBqÆmL)1 NaCl⁄ Pi) solution was added to 50 lL of the neoglycoprotein-containing solution, subsequently 10 lL of chloramine-T (3 mgÆmL)1H2O) solu-tion were added, and the mixture was incubated at room temperature for 3 min Thereafter, chloramine-T solution was pipetted to the above mixture in two further portions at intervals of 3 min, and then the reaction was stopped by add-ing 30 lL of freshly prepared sodium metabisulfite solution (5 mgÆmL)1) Label-free neoglycoprotein (50 lg) was added
as a carrier prior to the separation step by Sephadex G-50 (Pharmacia Biotech, Freiburg, Germany) gel permeation chromatography to remove any reagents from the radioiodi-nated product The specific radioactivity of batches of
125
I-labeled neoglycoprotein was in the range between 8 and
10 MBqÆmg)1 protein To monitor biodistribution of the iodinated product tumor-bearing mice were used [58,59] Approximately 5· 106Ehrlich ascites tumor (EAT) cells had been injected subcutaneously into the right rear leg of male ddY mice of the age of 7 weeks for tumor inoculation On