Toward a prostate specific antigen-based prostate cancer diagnostic assay: Preparation of keyhole limpet hemocyanin-conjugated normal and transformed prostate specific antigen fragments

Article abstract of DOI:10.1021/ja8028137

Prostate specific antigen (PSA) molecules secreted by cancerous and normal prostate cells differ in their N-linked glycan composition, while the peptide backbone appears to be conserved. Antibodies selectively recognizing such differentially glycosylated

Full text of DOI:10.1021/ja8028137

Published on Web 09/18/2008
Toward a Prostate Specific Antigen-Based Prostate Cancer
Diagnostic Assay: Preparation of Keyhole Limpet Hemocyanin
-Conjugated Normal and Transformed Prostate Specific
Antigen Fragments
Vadim Y. Dudkin,† Justin S. Miller,† Anna S. Dudkina,† Christophe Antczak,‡
David A. Scheinberg,‡ and Samuel J. Danishefsky*,†,§
Laboratory for Bioorganic Chemistry, and Molecular Pharmacology and Chemistry Program,
Sloan-Kettering Institute for Cancer Research, 1275 York AVenue, New York, New York 10065,
and Department of Chemistry, Columbia UniVersity, HaVemeyer Hall, 3000 Broadway,
New York, New York 10027
Received April 30, 2008; E-mail: s-danishefsky@mskcc.org
Abstract: Prostate specific antigen (PSA) molecules secreted by cancerous and normal prostate cells differ
in their N-linked glycan composition, while the peptide backbone appears to be conserved. Antibodies
selectively recognizing such differentially glycosylated PSA structures could form a basis for a new diagnostic
assay for prostate cancer. Twenty-amino acid PSA fragments carrying di-, tri-, and tetrabranched complex-
type glycans were prepared by total synthesis and conjugated to maleimide-modified keyhole limpet
hemocyanin (KLH) carrier protein through backbone Cys residues. These glycopeptide/KLH conjugates
were then used for antibody generation.
tics.7 PSA, a 28-kDa glycoprotein secreted exclusively by the
prostatic epithelium, consists of 237 amino acids and contains
Introduction
The development of aberrant protein glycosylation patterns
has long been recognized as one of the indicative changes that
accompany malignant transformation of a cell.1 In particular,
the advanced branching of N-linked carbohydrates has been
associated with the onset of metastasis of several types of
cancer,2 including breast,3 colon,4 and lung.5 Our program,
directed toward the fashioning and evaluation of anticancer
vaccines, is based on exploiting differences in oligosaccharide
epitope expression between normal and transformed cells.6
The work described herein seeks to take advantage of tumor-
selective glycosylation patterns at a different level. Modified
cellular glycosylation patterns may provide a potentially valu-
able, yet largely unexplored, opportunity for enhanced cancer
diagnostics. A recent discovery of advanced branching in the
glycoforms of prostate specific antigen (PSA) isolated from the
LnCaP prostate cancer cell line, and not found in normal prostate
cells, is potentially interesting from the perspective of diagnos-
a single N-glycosylation site.8 The high tissue selectivity of PSA
renders it potentially valuable as a marker for emerging prostate
cancer. The development of immunoassays to measure PSA
levels has provided an important breakthrough in prostate cancer
diagnostics.9 Unfortunately, the very common problem of
interpreting borderline PSA levels, necessary to differentiate
between prostate cancer and benign prostatic hyperplasia, has
consistently complicated the efforts of clinicians.10 A number
of methods have been suggested to overcome this problem,
including determinations of free vs total PSA11 (PSA index)
and monitoring the increase of PSA levels for a particular
patient12 (PSA velocity); however, the utility of such potential
measures remains unverified.13
All current state-of-the-art PSA-directed immunoassays em-
ploy antibodies that recognize only its peptide backbone. Such
(7) (a) Armbruster, D. A. Clin. Chem. 1993, 39, 181. (b) Abrahamsson,
P. A.; Lilja, H.; Oesterling, J. E. Urol. Clin. N. Am. 1997, 24, 353. (c)
Egawa, S. Biomed. Pharmacother. 2001, 55, 130.
† Laboratory for Bioorganic Chemistry, Sloan-Kettering Institute for
Cancer Research.
(8) Okada, T.; Sato, Y.; Kobayashi, N.; Sumida, K.; Satomura, S.;
Matsuura, S.; Takasaki, M.; Endo, T. Biochim. Biophys. Acta 2001,
1525, 149.
‡ Molecular Pharmacology and Chemistry Program, Sloan-Kettering
Institute for Cancer Research.
§ Department of Chemistry, Columbia University.
(1) Dennis, J. W.; Granovsky, M.; Warren, C. E. Biochim. Biophys. Acta
1999, 1473, 21.
(9) Ward, A. M.; Catto, J. W. F.; Hamdy, F. C. Ann. Clin. Biochem. 2001,
38, 633.
(10) (a) Semjonow, A.; Hertle, L. Urol.-Ausg. A 1995, 34, 290. (b)
Semjonow, A.; Brandt, B.; Oberpenning, F.; Roth, S.; Hertle, L.
Prostate 1996, 3.
(2) Dennis, J. W.; Laferte, S.; Waghorne, C.; Breitman, M. L.; Kerbel,
R. S. Science 1987, 236, 582.
(3) Seberger, P. J.; Chaney, W. G. Glycobiology 1999, 9, 235.
(4) Fernandes, B.; Sagman, U.; Auger, M.; Demetrio, M.; Dennis, J. W.
Cancer Res. 1991, 51, 718.
(11) Demura, T.; Shinohara, N.; Tanaka, M.; Enami, N.; Chiba, H.; Togashi,
M.; Ohashi, N.; Nonomura, K.; Koyanagi, T. Cancer 1996, 77, 1137.
(12) Thiel, R.; Pearson, J. D.; Epstein, J. I.; Walsh, P. C.; Carter, H. B.
Urology 1997, 49, 716.
(5) Lu, Y.; Pelling, J. C.; Chaney, W. G. Clin. Exp. Metastasis 1994, 12,
47.
(13) Catalona, W. J.; Southwick, P. C.; Slawin, K. M.; Partin, A. W.;
Brawer, M. K.; Flanigan, R. C.; Patel, A.; Richie, J. P.; Walsh, P. C.;
Scardino, P. T.; Lange, P. H.; Gasior, G. H.; Loveland, K. G.; Bray,
K. R. Urology 2000, 56, 255.
(6) (a) Danishefsky, S. J.; Allen, J. R. Angew. Chem., Int. Ed. 2000, 39,
836. (b) Ouerfelli, O.; Warren, J. D.; Wilson, R. M.; Danishefsky,
S. J. Exp. ReV. Vaccines 2005, 4, 677.
9
13598 J. AM. CHEM. SOC. 2008, 130, 13598–13607
10.1021/ja8028137 CCC: $40.75
2008 American Chemical Society

Toward a New Diagnostic Assay for Prostate Cancer
A R T I C L E S
Scheme 1. N-Linked Branching of Normal (1) and Transformed (2 and 3) Desialylated PSA Glycoforms
Scheme 2. Proposed Identification of Antibodies Sensitive to Neoplastic PSA Glycoforms
assays are therefore unable to differentiate between the various
N-linked glycoforms of the protein. Thus, a potentially valuable
criterion for discrimination between normal and transformed
PSA remains invisible under the current methods.
certainly in line with the expectation of a higher degree of
branching of N-linked glycoforms associated with neoplastic
cells.1
The research described below16 was driven by two goals.
First, total synthesis of the PSA associable oligosaccharide
patterns constitutes a formidable challenge to the state of the
art of the field and, as such, would serve as a testing ground to
evaluate operations at the perimeter of current feasibility.
Moreover, the identification of antibodies sensitive to particular
PSA glycoforms, accessed through total synthesis, could form
the basis for the development of a new immunoassay capable
of elucidating not just levels of PSA, but also the metastatic
propensity of the disease. Our approach toward eliciting such
distinguishing antibodies is summarized in Scheme 2. Animal
immunization with glycosylated PSA fragments conjugated to
a carrier protein (such as keyhole limpet hemocyanin (KLH))
and subsequent screening of serum could lead to the identifica-
tion of the desired antibodies specific for single PSA glycoforms.
It seemed that the total chemical synthesis of these types of
complex and highly branched N-linked glycopeptides had never
Analyses of PSA glycosylation patterns suggest that normal
PSA variants contain only biantennary N-linked glycoforms,8,14
in which the mannose units of the pentasaccharide core are
symmetrically glycosylated with N-acetyllactosamine units at
the 2 and 2′ positions, as in glycopeptide 1 (Scheme 1). These
lactosamines may be further modified through sialylation. By
contrast, PSA samples isolated from prostate cancer cells appear
to display tri- and even tetrabranched N-glycans,15 such as in
structures 2 and 3 (Scheme 1). In these glycopeptides, the core
mannose segments are asymmetrically glycosylated with three
or four N-acetyllactosamines at the 2, 2′, 4, and 6′ positions, as
shown. While the exact structures of these transformed PSA
glycans have not always been determined, these findings are
(14) (a) Belanger, A.; Vanhalbeek, H.; Graves, H. C. B.; Grandbois, K.;
Stamey, T. A.; Huang, L. H.; Poppe, I.; Labrie, F. Prostate 1995, 27,
187.
(16) Dudkin, V. Y.; Miller, J. S.; Danishefsky, S. J. J. Am. Chem. Soc.
2004, 126, 736.
(15) Prakash, S.; Robbins, P. W. Glycobiology 2000, 10, 173.
9
J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008 13599

A R T I C L E S
Dudkin et al.
Scheme 3. A Route Toward the Synthesis of Arbitrarily Branched PSA Glycopeptides from a Single Trisaccharide Precursor
previously been accomplished. Yet, de noVo synthesis would
likely be the only way by which to gain access to completely
homogeneous glycosylated PSA fragments in substantial quanti-
ties. To pursue this study, we elected to synthesize three
glycopeptide fragments, each containing the uneicosapeptide
portion of the PSA backbone flanking the glycosylation site
(PSA(27-47)), and presenting either the di-, tri-, or tetra-
branched glycans found in structures 1, 2, and 3, in unsialylated
form. As sialylation adds a considerable degree of heterogeneity
to glycoprotein, such samples are typically subjected to sialydase
digestion prior to analysis.17 Hence, our constructs did not
contain sialic acids at their nonreducing termini, though in
principle, they could be accommodated. As will be shown
herein, we found that antibody binding to our synthetic
glycopeptide constructs did exhibit modest selectivity, though
binding was weaker than with peptide epitopes.
to be highly “step intensive”. We hoped to achieve maximal
efficiency through the simultaneous glycosylation of the pen-
tasaccharide diol (5), triol (6), or tetraol (7) with either two,
three, or four N-acetyllactosamine units (Scheme 3). In this way,
the construction of both the symmetrically (2,2′) and asym-
metrically (2,2′,4 and 2,4,2′,6′) branched glycans would ulti-
mately be accomplished from a common building block. Each
of the required pentasaccharides was to be derived from
trisaccharide 4, which was deliberately designed to allow for
either sequential or simultaneous exposure of the 3 and 6
hydroxyl groups of the terminal mannose.
Two significant changes were built into our strategic plan
for the total syntheses to be pursued under the PSA program.
In our earlier glycopeptide syntheses, based on glycal assembly,
the reducing end was carried to the end of the synthesis as a
glycal linkage.18 In the syntheses described here, the terminal
glycal is converted at a much earlier stage (see compound 4) to
the GlcNAc pattern, thereby avoiding the need to conduct such
technically demanding steps at the end of the synthesis.
Moreover, by the new scheme, the interior mannose (see ring
C, compound 4) would be introduced by direct ꢀ-mannosylation.
In our earlier approaches to complex mannose-based N-linked
glycopeptides, the corresponding interior ꢀ-linked mannose
moiety had been fashioned through epimerization of a ꢀ-glu-
coside precursor (by an oxidation-reduction sequence at C2).18
By contrast, in the PSA synthesis, we would be employing direct
ꢀ-mannosylation according to Crich,19 using Kahne’s sulfoxide
as the glycosyl donor.20
Results and Discussion
The construction of complex homogeneous glycopeptides can
generally be broadly divided into three synthetic tasks: (1) the
preparation of the peptide sequence, (2) the synthesis of the
glycan region, and (3) the merging of the two domains. While
advances in automated peptide synthesis have greatly simplified
the first task, the latter two still present daunting challenges to
the synthetic organic chemist. In order to achieve efficient
glycopeptide conjugation, we envisioned a route consisting of
amination of the free reducing end of the carbohydrate, followed
by coupling of the resulting ꢀ-amino group with a short Asp-
containing peptide. Finally, native chemical ligation could be
employed to further elongate the peptide chain.
The decision was also made to equip the reducing end of 4
with a tert-butylsilyl (TBS)-protected alcohol, hoping to avoid
We sought to design a highly convergent approach that could
allow for access to all three carbohydrate domains from a single
precursor. This would be of significant advantage, since
sequential preparations of each separate domain would be likely
(18) Wang, Z.-G.; Warren, J. D.; Dudkin, V. Y.; Zhang, X.; Iserloh, U.;
Visser, M.; Eckhardt, M.; Seeberger, P. H.; Danishefsky, S. J.
Tetrahedron 2006, 62, 4954.
(19) (a) Crich, D.; Sun, S. X. J. Am. Chem. Soc. 1998, 120, 435. (b) Crich,
D.; Sun, S. X. Tetrahedron 1998, 54, 8321.
(17) Hilz, H.; Noldus, J.; Hammerer, P.; Buck, F.; Luck, M.; Huland, H.
(20) Kahne, D.; Walker, S.; Cheng, Y.; Van Engen, D. J. Am. Chem. Soc.
1989, 111, 6881.
Eur. Urol. 1999, 36, 286.
9
13600 J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008

Toward a New Diagnostic Assay for Prostate Cancer
A R T I C L E S
Scheme 4. Synthesis of the Key Trisaccharide from Chitobiose Glycala
a Reagents and conditions: (a) I(coll)2ClO4, PhSO2NH2; (b) Et3N, H2O/THF; (c) TBSOTf, 2,6-lutidine, CH2Cl2; (d) NaOMe/MeOH; 53-60% for four
steps; (e) i. Tf2O, DTBMP, CH2Cl2, -78 °C, ii. 11, 85%-91% (ꢀ/R ) 8/1); (f) CAN, MeCN/H2O, 74% for two steps; (g) Bu2BOTf, BH3 ·THF, 72%.
Scheme 5. A Proof-of-Concept Aspartylation/Native Chemical Ligation Sequence with the Common Core Trisaccharidea
a Reagents and conditions: (a) aq. NH4HCO3, 6 days; (b) i. HATU, NEt3, DMSO, 59% for two steps, ii. 20% piperidine/DMF, 68%; (c) 17, 18, PBS
buffer pH 7.4, 78%.
the need for the glycal in the late stages of the synthesis. The
stability of anomeric alcohols to the dissolving metal reduction
conditions required for the global deprotection had been
previously demonstrated in our laboratory.21
sponding mannosyl sulfoxide, 11. Using a modest excess (1.5
equiv) of disaccharide 10, mannosyl coupling could be achieved
with good levels of stereoselectivity (8:1, ꢀ:R), to provide
trisaccharide 12 in 85-91% yields. The excess acceptor 10
could be recovered intact and reused. Deprotection of 12
provided the key trisaccharide building block, 4. Under this
sequence, significant amounts of 4 were obtained, with the
largest scale-up effort yielding 25 g of trisaccharide. Finally,
the 4,6-benzylidene functionality of 4 could be reduced by the
action of borane/tetrahydrofuran (THF) complex and dibutyl-
boron triflate to afford 13. This compound, bearing the two
required acceptor sites at C3 and C6 of the interior mannose,
was destined to be a key building block as the synthesis
unfolded.
The synthesis of trisaccharide 422 commenced with the known
disaccharide glycal 8, itself prepared through glycal assembly
techniques developed in our laboratory. The double bond of 8
was subjected to iodosulfonamidation, and both resulting
iodosulfonamide isomers were subsequently converted to the
reducing disaccharide, 9 (Scheme 4). Interestingly, only the
R-anomer of 9 was formed, possibly due to the stabilization of
the axial hydroxide through intramolecular hydrogen bonding
to the C2 sulfonamide. The R-configuration of the anomeric
hydroxyl group could be preserved through silylation with
excess TBSOTf. Deacetylation of the resultant disaccharide,
under Zemple´n conditions, provided the requisite acceptor 10
in 53-60% yield over four steps. In a defining step in the
synthesis, direct ꢀ-mannosylation, under the Crich glycosylation
protocol, was efficiently achieved through exposure to the
presumed mannosyl triflate, generated in situ from the corre-
The mannosylated chitobiose trisaccharide (13) was then fully
deprotected to afford 14, which was used as a model for testing
the viability of the Kochetkov amination/peptide coupling/native
chemical ligation (NCL) sequence (Scheme 5). In the event,
(21) Iserloh, U.; Dudkin, V. Y.; Wang, Z. G.; Danishefsky, S. J.
(22) Dudkin, V. Y.; Miller, J. S.; Danishefsky, S. J. Tetrahedron Lett. 2003,
44, 1791.
Tetrahedron Lett. 2002, 43, 7027.
9
J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008 13601

A R T I C L E S
Dudkin et al.
Scheme 6. Synthesis of Normal PSA Glycopeptide 1a
a Reagents and conditions: (a) 20, (BrC6H4)3NSbCl6, MeCN, 74%; (b) NaOMe/MeOH, 89% for two steps; (c) MeOTf, DTBP, CH2Cl2, 60%; (d) i.
ethylenediamine, n-BuOH/toluene, 90 °C, ii. Ac2O/py, iii. NaOMe/MeOH, 72%; (e) TBAF/AcOH, THF, 76%; (f) i. Na/NH3, -78 °C, ii. Ac2O, iii. NaOMe/
MeOH, 65%; (g) NH4HCO3/H2O; (h) 25, HATU, Hu¨nig’s base, DMSO, 61% from 23; (i) (NH2)2, piperidine, DMF, 62%; (j) 27, MES-Na, pH 7.4, 17%.
Kochetkov amination23 of the trisaccharide proceeded smoothly,
albeit over 6 days at room temperature, to give the ꢀ-glycosyl-
amine (15). Because the starting material containing the free
reducing end hydroxyl group and the product amine differ by
only a single mass unit, the reaction was monitored by
evaluating the product isotope ratios obtained via mass spec-
trometry of the reaction mixture. Removal of the solvent and
the volatile salts was accomplished by repeated lyophilization.
Care was taken to minimize the exposure of the mixture to
aqueous conditions during lyophilization, since the amine is
known to hydrolyze in solution to the anomeric alcohol.
Coupling of the trisaccharide glycosylamine (15) with a model
pentapeptide (CADAS) proceeded uneventfully,24 affording
analytically pure (liquid chromatography/mass spectrometry
(LCMS)) (fluorenylmethoxy)carbonyl (Fmoc)-protected N-
linked trisaccharide in 59% yield (from the free glycan). Fmoc
removal with 20% piperidine in dimethylformamide (DMF)
provided the free trisaccharide pentapeptide (16) in 68% yield.
Native chemical ligation25 performed on an analytical scale with
a 14-mer C-terminal thioester (17) showed substantial comple-
(24) Cohen-Anisfeld, S. T.; Lansbury, P. T. J. Am. Chem. Soc. 1993, 115,
10531.
(25) (a) Dawson, P. E.; Muir, T. W.; Clark-Lewis, I.; Kent, S. B. H. Science
1994, 266, 776. (b) Tolbert, T. J.; Wong, C. H. J. Am. Chem. Soc.
2000, 122, 5421.
(23) Likhosherstov, L. M.; Novikova, O. S.; Derevitskaja, V. A.; Kochetkov,
N. K. Carbohydr. Res. 1986, 146, C1.
9
13602 J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008

Toward a New Diagnostic Assay for Prostate Cancer
A R T I C L E S
Scheme 7. Synthesis of Selectively Benzoylated Mannosyl
with excess R-mannosyl donor, 20, proved to be most efficient,
furnishing the pentasaccharide adduct in 74% yield, with little
or no concomitant orthoester formation. Saponification of the
two benzoyl groups afforded the pentasaccharide diol, 5.
Subsequent two-fold glycosylation with disaccharide donor 21
under MeOTf activation proceeded smoothly. Following con-
version of the phthalimide functionalities to acetates and
liberation of the anomeric hydroxyl group, compound 22,
representing the fully protected nonasaccharide domain of the
target system (1), was in hand. At this stage, we were able to
take advantage of an earlier remarkable finding to the effect
that the integrity of the “free” reducing end hemiacetals in such
systems can be maintained even under Birch debenzylation
conditions. Global deprotection, followed by amine-specific
diacetylation, yielded the free glycan 23 as a mixture of anomers.
The latter was convertable to the free ꢀ-glycosylamine, 24, under
Kochetkov amination conditions. Coupling of 24 with excess
equivalents of hexapeptide 25, followed by removal of the Fmoc
and 4,4-dimethyl-2,6-dioxocyclohex-1-ylidene (ivDde) protect-
ing groups, provided glycoconjugate 26. Finally, compound 26
was subjected to NCL with pentadecapeptide 27, as shown, to
provide construct 1, representing the fully deprotected normal
PSA(27-47) glycopeptide fragment (Scheme 6).
Having successfully synthesized the nontransformed, di-
branched glycopeptide fragment of PSA, we next set out to
synthesize the significantly more challenging tri- and tetra-
branched transformed glycopeptides, 2 and 3. The first task
would be the preparation of R-mannosyl donors 30 and 31,
which would be required for the assembly of the pentasaccha-
rides 6 and 7. Toward this end, the R-mannosyl unit, 28, was
selectively benzylated at the 3 position, as shown. The resultant
intermediate, 29, could be readily converted to donor 30,
possessing benzoylation at the 2 and 4 positions, or to donor
31, in which the 2 and 6 positions were selectively benzoylated
(Scheme 7).
Donorsa
a Reagents and conditions: (a) i. Bu2SnO, toluene, ii. BnBr (89%); (b) i.
NaBH3CN, HCl, THF, 77%, ii. BzCl, DMAP, 98%; (c) i. Bu2BOTf, BH3
THF, 97%, ii. BzCl, DMAP 84%.
tion within 2-3 h, but technical challenges arose which,
unaddressed, would have impeded the isolation of the desired
product. LCMS indicated that the mercaptoethanesulfonate
(MES) thioester derivative of the N-terminal coupling partner
(used in excess) essentially coeluted with the desired material,
as did its carboxylic acid derivative. Simply allowing the
thioester to hydrolyze would be insufficient as a method for
removing the 14-residue byproduct. Fortunately, trapping of the
MES thioester intermediate with cysteamine proved to be an
effective method for altering the compound’s polarity, thus
allowing its effective separation from the desired product, even
on larger scale reactions. By adding a 20-min cysteamine
“quench” following reaction completion, it was possible to
repeat the NCL on a large scale, and to isolate analytically pure
N-linked trisaccharide nonadecapeptide (19) in 78% yield.
Having demonstrated the feasibility of the amination/ligation
sequence with the model trisaccharide system, we sought to
prepare the normal, dibranched PSA glycopeptide, 1. It would
first be necessary to prepare the pentasaccharide diol, 5, through
two-fold glycosylation of trisaccharide 13. This transformation
would require two high-yielding glycosylation events. A number
of protocols were assessed for their ability to provide maximal
yield and minimal orthoester formation. Ultimately, bis-glyco-
sylation of 13 under Sinay radical cation activation conditions26
With these differentially protected monosaccharide units in
hand, we had the key building blocks required for the completion
Scheme 8. Synthesis of Pentasaccharides with Selectively Exposed Branching Pointsa
a Reagents and conditions: (a) 30, (BrC6H4)3NSbCl6, MeCN, 71%; (b) BH3 ·THF, Bu2BOTf, THF, 85%; (c) 20, (BrC6H4)3NSbCl6, MeCN; (d) NaOMe/
MeOH, 72% for two steps; (e) 31, (BrC6H4)3NSbCl6, MeCN, 74%; (f) NaOMe/MeOH, 92%.
9
J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008 13603

A R T I C L E S
Dudkin et al.
Scheme 9. Synthesis of Tribranched Transformed PSA Glycopeptide 2a
a Reagents and conditions: (a) MeOTf, DTBP, CH2Cl2, 41%; (b) i. ethylenediamine, n-BuOH/toluene, 90 °C, ii. Ac2O/py, iii. NaOMe/MeOH, 77%; (c)
TBAF/AcOH, THF, 95%; (d) i. Na/NH3, -78 °C, ii. Ac2O, iii. NaOMe/MeOH, 94%; (e) NH4HCO3/H2O; (f) 25, HATU, Hu¨nig’s base, DMSO, 25% from
37; (g) (NH2)2, piperidine, DMF, 66%; (h) 27, MES-Na, pH 7.4, 38%.
of the syntheses of our target glycopeptides, 2 and 3. The
objective would now be the assembly of pentasaccharide units
6 and 7, presenting appropriately positioned free hydroxyl
groups, which would ultimately serve as the sites of tri- or
tetrabranching. In the event, treatment of trisaccharide 4 with
the R-mannosyl donor 30, followed by regiospecific reductive
cleavage of the 4,6-benzylidene moiety, afforded tetrasaccharide
intermediate 32, possessing two sites of benzoylation and a free
C-6 hydroxyl group, as shown.
Tetrasaccharide 32 served as a common intermediate in the
syntheses of pentasaccharides 6 and 7. The synthesis of 6, a
key intermediate en route to the tribranched glycopeptide (2),
was accomplished through glycosylation of 32 with thioman-
noside 20, again by implementation of the Sinay radical cation
9
13604 J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008

Toward a New Diagnostic Assay for Prostate Cancer
A R T I C L E S
Scheme 10a
a Reagents and conditions: (a) MeOTf, DTBP, CH2Cl2, 19%; (b) i. ethylenediamine, n-BuOH/toluene, 90 °C, ii. Ac2O/py, iii. NaOMe/MeOH, 79%; (c)
TBAF/AcOH, THF, 97%; (d) i. Na/NH3, -78 °C, ii. Ac2O, iii. NaOMe/MeOH, 81%; (e) NH4HCO3/H2O; (f) 25, HATU, Hu¨nig’s base, DMSO, 25% from
42; (g) (NH2)2, piperidine, DMF, 52% for three steps; (h) 27, MES-Na, pH 7.4, 65%.
activation protocol. Saponification of the benzoate protecting
groups afforded the requisite pentasaccharide triol acceptor, 6.
In a similar fashion, 32 was readily advanced to the pentasac-
charide tetraol, 7, an intermediate in the synthesis of the
tetrabranched glycopeptide (3). Mannosylation of 32 with the
2,6-di-O-benzoyl donor, 31, provided pentasaccharide 34, which
readily underwent saponification to afford tetraol 7. The
program, which had been devised to reach pentasaccharides with
well-identified sites for concurrent glycosylation, was imple-
mentable in a relatively straightforward way (Scheme 8).
The synthesis of the tribranched glycopeptide, 2, from
pentasaccharide 6 is summarized in Scheme 9. Three-fold
ꢀ-lactosylation was achieved in 41% yield through exposure
of 6 to excess ꢀ-lactosamine donor 21. The steps required for
9
J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008 13605

A R T I C L E S
Dudkin et al.
Scheme 11. Conjugation of Glycopeptides to KLHa
activated KLH yielded a conjugate bearing 300 residues, as
assessed by carbohydrate high-performance liquid chromatog-
raphy analysis after total hydrolysis (data not shown). The
conjugating efficiency was 20%, which is relatively standard
in the field.
BALB/c mice were immunized with glycopeptide 2 conju-
gated to KLH, and high-titer antisera were elicited for the
antigen. About 1500 hybridoma supernatants were screened for
the presence of antibodies recognizing glycopeptide 2 in a
standard enzyme-linked immunosorbent assay (ELISA). About
10% of the hybridomas were found to be positive and were
tested for recognition of the nontransformed glycopeptide 1.
Our goal was to select antibodies binding preferentially to the
tribranched, transformed glycopeptide 2 compared to the normal,
dibranched glycopeptide 1, thereby allowing us to discard
antibodies that were likely to be directed against the peptidic
moiety of the antigen. When tested in parallel in at least two
independent experiments, a selectivity ranging from 2- to 4-fold
in favor of the tribranched glycopeptide 2 compared to the
dibranched glycopeptide 1 was confirmed for 17 clones, which
were selected for further experiments (Figure 1).
Selected hybridoma supernatants were evaluated for binding
to LnCaP tumor PSA in a sandwich ELISA. These tests were
done on hybridoma supernatant fluids and not purified mono-
clonal antibodies, and antibodies generated against carbohydrates
are often of lower affinity. Therefore, absolute titers of individual
hybridomas cannot be compared. In addition, we had purposely
immunized the mice against the PSA glycopeptide 2, which is
devoid of any sialyl group, in order to prevent the generation
of antibodies directed against this ubiquitous moiety. LnCap
PSA carbohydrates, however, are likely to be sialylated, which
could affect both of the recognized epitopes through masking
or conformational changes. Antibodies recognizing nonsialylated
PSA carbohydrates might therefore not bind to their sialylated
counterparts. To test this hypothesis, we measured the binding
of our hybridoma supernatants to desialylated LnCap PSA. A
positive control monoclonal antibody included in these tests,
H117, which is reactive with the peptide moiety of PSA,27
showed strong binding to both sialylated and desialylated
peptides, as expected. Small increases in the binding to LnCap
PSA after desialylation were observed for four of the six tested
supernatants, sometimes resulting in up to a 2-fold increase in
binding (data not shown). This result suggests that some of the
generated antibodies recognized a desialylated carbohydrate
moiety of native LnCap PSA better than the sialylated forms;
however, only small changes were observed. The small differ-
ence in binding observed after desialylation of LnCap PSA is
consistent with a report in which LnCap PSA has little sialic
acid.28 Desialylation of LnCap PSA would therefore be expected
to affect only moderately the binding of the antibodies that we
have generated against the desialylated glycopeptide 2. Interest-
ingly, when we measured the binding of the 17 selected
hybridoma supernatants to the serum of a patient with prostate
cancer, we observed an increase in binding after desialylation
for 13 of 17 supernatants tested ranging from 2- to 3-fold (data
not shown). While this result could not be repeated due to the
limited amounts of hybridoma supernatant fluids available, we
have identified a trend suggesting that the antibodies raised
a Reagents and Conditions: a) maleimidobutyric acid, PBS EDTA pH
7.0, rt; b) LC-SMCC, PBS EDTA pH 7.0, rt; c) glycopeptide, PBS EDTA
pH 6.8, rt.
the transformation of 35 to 2 are analogous to those described
for the dibranched glycopeptide 1 (Scheme 6). The phthalimide
protecting groups were replaced with N-acetyl functionalities,
and the anomeric hydroxyl group was liberated, as shown.
Global benzyl deprotection proceeded uneventfully to yield
intermediate 37, which smoothly underwent Kochetkov ami-
nation to afford the free ꢀ-glycosylamine, 38. Coupling of 38
with hexapeptide 25 and subsequent NCL proceeded without
incident to yield the tribranched glycopeptide unit 2.
The assembly of the tetrabranched glycopeptide (3) was
accomplished through a route analogous to those described
above for the syntheses of the di- and tribranched systems. Thus,
as summarized in Scheme 10, the pentasaccharide tetraol, 7,
underwent four-fold glycosylation with 21 to yield 40. As would
be expected, this high complexity-building transformation
proceeded in modest yield (19%). Nonetheless, the subsequent
Kochetkov amination/aspartylation/NCL sequence proceeded
smoothly to provide the tridecasaccharide-uneicosapeptide
glycoconjugate 3 in homogeneous form as judged by MS and
NMR analysis.
In order to elicit a sustained immune response against the
tribranched transformed glycopeptide 2, we chose KLH as the
protein carrier. KLH has proven to be effective in numerous
vaccine applications because of its many foreign epitopes, large
molecular mass, and poor solubility. Using the unique thiol on
the glycopeptide as a convenient handle for conjugation to a
maleimide-functionalized KLH, we prepared an extensively
functionalized KLH system. In order to prevent precipitation
due to resilient thiol (from cysteine) cross-linking, KLH was
first reacted with maleimidobutyric acid in order to block the
free thiols (Scheme 11). This preliminary step allowed for a
higher degree of maleimide functionalization, and also provided
stability to the construct. The lysines of KLH were derivatized
using a large excess of succinimidyl 4-[N-maleimidomethyl]-
cyclohexane-1-carboxylate (SMCC). Up to 1000 maleimide
moieties were grafted on KLH using this protocol, as assessed
by 35S-cysteine labeling. Coupling of glycopeptide 2 to this
(26) Zhang, Y. M.; Mallet, J. M.; Sinay, P. Carbohyd. Res. 1992, 236, 73.
(27) Leinonen, J.; Wu, P.; Stenman, U.-H. Clin. Chem. 2002, 48, 2208.
(28) Peracaula, R.; Tabare´s, G.; Royle, L.; Harvey, D. J.; Dwek, R. A.;
Rudd, P. M.; de Llorens, R. Glycobiology 2003, 13, 457.
9
13606 J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008

Toward a New Diagnostic Assay for Prostate Cancer
A R T I C L E S
Figure 1. Binding of selected hybridoma supernatants to the tribranched glycopeptide 2 compared to the dibranched glycopeptide 1. The data presented are
from one representative experiment.
against the nonsialylated glycopeptide 2 can recognize the
desialylated glycan moiety of PSA from a patient serum.
We note also that the determination of the actual glycobiology
and structural chemistry of these glycans is still a work in
progress.29 As firmer answers to these highly complex structural
issues are available, the synthetic chemistry addressed to
simulating nuances of the various PSA related antigens can be
adapted. In this way, one can hope to improve the cor-
respondence of the chemical construct with the cell surface,
thereby achieving more selective and more discerning induced
antibody-natural antigen affinities.
carrier proteins can be achieved through maleimide modification/
Michael addition. The possibility of selective recognition of the
11-mer carbohydrate vs the 9-mer as well as a significant
increase in binding upon sialidase digestion of PSA suggests
the potential of this methodology for providing a path toward
a new, more selective prostate cancer diagnostic assay. While
a trend suggesting binding of the newly generated antibodies
to both LnCap PSA and serum from a prostate cancer patient
after desialylation has been detected, further optimization of
both antigen structure and antibody preparation is required
before a practical use of the assay can be projected.
Conclusion
Acknowledgment. Support for this work was provided by the
NIH (CA23766 and CA55349 to D.A.S. and CA103823 to S.J.D.).
We thank Dr. H. Lilja for the gift of the H117 antibody for PSA
assays, Dr. G. Ragupathi for assistance on KLH conjugate analysis,
and the Monoclonal Antibody Core Facility for help with antibody
development. J.S.M. is the P.I. of DAMD 17-03-1-0016.
In summary, we have shown the feasibility of preparation of
highly complex N-linked glycopeptides through entirely chemi-
cal means. In this case, di-, tri-, and tetrabranched glycans have
been synthesized from a single trisaccharide precursor with most
diversity introduction performed at the end of the oligosaccha-
ride assembly. Amination of such glycans followed by coupling/
NCL sequence allows attachment of long peptide motifs. We
have also demonstrated that significant glycopeptide loading on
Supporting Information Available: Experimental procedures
and compound characterization data. This information is avail-
able free of charge via the Internet at http://pubs.acs.org.
(29) Tabare´s, G.; Radcliffe, C. M.; Barrabe´s, S.; Ram´ırez, M.; Nu´ria, A.;
Hoesel, W.; Dwek, R. A.; Rudd, P. M.; Peracaula, R.; de Llorens, R.
Glycobiology 2006, 13, 132.
JA8028137
9
J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008 13607