Oligomannose Glycopeptide Conjugates Elicit Antibodies Targeting the Glycan Core Rather than Its Extremities

Up to ∼20% of HIV-infected individuals eventually develop broadly neutralizing antibodies (bnAbs), and many of these antibodies (∼40%) target a region of dense high-mannose glycosylation on gp120 of the HIV envelope protein, known as the “high-mannose patch” (HMP). Thus, there have been numerous attempts to develop glycoconjugate vaccine immunogens that structurally mimic the HMP and might elicit bnAbs targeting this conserved neutralization epitope. Herein, we report on the immunogenicity of glycopeptides, designed by in vitro selection, that bind tightly to anti-HMP antibody 2G12. By analyzing the fine carbohydrate specificity of rabbit antibodies elicited by these immunogens, we found that they differ from some natural human bnAbs, such as 2G12 and PGT128, in that they bind primarily to the core structures within the glycan, rather than to the Manα1 → 2Man termini (2G12) or to the whole glycan (PGT128). Antibody specificity for the glycan core may result from extensive serum mannosidase trimming of the immunogen in the vaccinated animals. This finding has broad implications for vaccine design aiming to target glycan-dependent HIV neutralizing antibodies.

Structure of linkers used in conjugation. a CRM197-glycopeptides used in rabbit immunization. b CRM197 + linkers used in rabbit control group in multi-immunogen study. c BSA-glycopeptide used for coating plates in ELISA. d BSA+linkers used in ELISA assay. e Groups 2-6 post-dose 4 serum binding to BSA+linkers depicted in d.

Fig. S2
Group 6 (sequential) serum selectivity for binding peptides, glycopeptides, and CRM197 +linker. Data are presented with geometric mean and geometric standard deviation. Statistical significance was determined by oneway ANOVA followed by multiple comparisons with Tukey's post-hoc test.

Crystal structure determination for 2G12-glycopeptide complexes
IgG 2G12 was expressed in 293 Freestyle cells and purified by protein A chromatography. IgG was cleaved to Fab with 2% papain for 3 hours before inactivation with 200 mM iodoacetamide. The cleavage mixture was applied to a protein A column and the unbound Fab was further purified on a S200 16/60 column (GE).
Fab was mixed with peptide at a 1:3 (Fab:peptide) molar ratio and the complex was purified by size exclusion chromatography with a Superdex 200 16/60 column. Each Fab-glycopeptide complex was concentrated to 10mg/ml. The 2G12-10V1S peptide complex was crystallized in 24-well sitting drop trays (Hampton Research) in condition 6B of Footprint 1 screen 1 , which corresponded to 1.0 M sodium citrate tribasic dihydrate, 10mM sodium borate, pH 8.5. Crystals were cryoprotected with well solution augmented with glycerol to a final concentration of 30%. 2G12-10F5M peptide complex was crystallized in 24-well sitting drop trays with a well solution of 0.2M LiSO4, 17.5% PEG400, 0.1M Tris, pH 8.5. Crystals were cryoprotected with the well solution plus PEG400 at a final concentration of 27.5%. All crystals were cryocooled by rapid plunging into liquid nitrogen and data were collected at SSRL beamline 12-2 using a Dectris Pilatus 6M detector. Data were processed and scaled with HKL-2000 2 and molecular replacement was carried out with Phaser 3 using 2G12 Fab coordinates from PDB 4RBP 4 as a model. Refinement was carried out with Phenix.refine 5 and final statistics for data collection and refinement are outlined in Table S7. For both complexes, the Fab chain identifiers are L and K (light chains) and H and M (heavy chains). The 10V1S glycopeptide has chain identifiers A and C. Glycans in 10F5M are labeled A and B. In Fab1 (LH), the VL domain is paired with the VH' domain, while in Fab2 (KM), the VL' domain is paired with the VH domain (see Fig. 3).
The 2G12-10V1S complex crystals have one domain-swapped Fab dimer and two glycopeptides in the asymmetric unit, with one peptide bound to each Fab within the dimer (Figs. 3 and S13). Glycopeptide A is better ordered and has lower B values than C (38 Å 2 for A versus 57 Å 2 for C), likely due to crystal contacts from a symmetry-related Fab. Thus, glycopeptide A is used for most of the analysis and discussion. Out of the 40-residue peptide, residues 18-33 and 19-33 are visible in peptides A and C. The peptide forms a hairpin, with a type VIII non-hydrogen bonded reverse turn around residues 23-26 (IPWY). Pro24 adopts a cis conformation. Residue 20 is homopropargylglycine (HPG) to which a Man9 glycan is attached. All 9 mannose moieties are visible in the electron density and adopt a gg-gg (3-4´ and 4´-B both adopt a gauche rotamer) rotameric arrangement. The primary glycan binding site in each Fab binds to the terminal Manα1-2 Man moieties from the D3 arm (mannose B and D3) (Fig. 3a). The Fabglycopeptide interface is extensive with 854Å 2 and 884Å 2 buried on the glycopeptide and Fab, respectively (Tables  S2 and S3). About 60% of the glycopeptide contribution is from peptide. The Fab contacts glycopeptide with CDR's L3, H1', H2', and H3', with some contributions from heavy-chain framework residues (Table S2); the largest contribution comes from CDR H2'. There are 10 hydrogen bonds from the glycopeptide to the Fab, with 6 from the peptide component and 4 from the carbohydrate (Table S4).   S14. Superposition of 10V1S with 2G12+Man8 (6MNF). The 10V1S heavy chain and carbohydrate are colored pink and yellow, while the 2G12/Man8 heavy chain and carbohydrate are both colored white. In both the 10V1S structure and 6MNF, there are 6 hydrogen bonds involving ThrH33, ThrH52a, SerH53 and AspH100d to the D3 or 4' mannose moieties.

Fig. S15
Stability of glycopeptide 10F5M in crystallization buffer. UPLC chromatograms of glycopeptide 10F5M are shown a) prior to and b) after incubation in crystal growth medium. The conditions for growth of 10F5M-2G12 cocrystals were room temperature for 2 days in 0.2M Li2SO4, 17.5% PEG400, 0.1M Tris, pH 8.5. To test stability, glycopeptide 10F5M was dissolved in the same mixture for a week at room temperature, then exchanged back to water by Amicon filtration and analyzed by UPLC/ESI/MS.

Fig. S16
ELISA of all groups binding to 293F SOSIP trimer at 200 ng/well.

Conjugation procedure (6M guanidine PBS): BSA-peptide 10F2, 10F5M, 10F8
Peptides 10F2, 10F5M, 10F8 were each dissolved in 0.5% acetic acid for quantification by BCA assay, and appropriate aliquots of appropriate size for the reaction below were lyophilized and redissolved in 6M guanidine PBS buffer for the conjugation. Generally, capping was performed with slightly substoichiometric BME to avoid driving the reversible loss of glycopeptides from the conjugate.

Conjugation procedure (6M guanidine and 1% acetic acid): BSA-peptide 10F6
HPLC-purified lyophilized 10F6 (2.2 mg) was dissolved in 10% acetic acid and diluted to 0.5% acetic acid solution (0.5 ml) before quantification by BCA assay (1.797 mg/mL result). Solution containing 27 µg of peptide 10F6 (15 µL) was transferred to a 0.5 mL low protein binding Eppendorf tube which was lyophilized. The lyophilized peptide 10F6 was then dissolved in 50 µL 6M guanidine PBS buffer (pH 6.5) and placed in a two-neck flask flushed with nitrogen. TCEP (0.6 µL of 100 mM aqueous solution, 10 equiv) was added and the solution stood overnight at room temperature after which time UPLC-ESI-MS showed complete deprotection of the cysteine. Excess TCEP was removed by Amicon centrifugal filter (3-kDa cutoff, 2 rounds of filtration: 20 min and then 30 min) with 6M guanidine in PBS buffer (pH 6.5). The deprotected peptide was added to the activated BSA-maleimide (0.05 mg, 0.68 nmol, 17.7 maleimide linkers, in 42.4 µL of the same buffer) and the solution stood overnight at room temperature under N2. The peptide-BSA conjugates were purified by Amicon centrifugal filter (30-kDa, 4 rounds, 5 min each round) with 1% AcOH/ H2O to remove salts and unreacted peptide. The conjugate was then capped with mercaptoethanol (0.68 µL of 10 mM solution, 10 equiv, 1h), and purified using 30K Amicon filter with 1% AcOH/ H2O (4 rounds: 5 min each round). BCA quantification assay indicated a yield of 34.4 µg of the BSA-peptide 10F6 conjugate.

Synthesis of cyclohexanol-azide
Cyclohexanol-azide was synthesized by previously reported method 8 . Briefly, imidazole-sulfonyl-azide·HCl (0.52 mmol, 109 mg) was added to a solution of 4-aminocylohexanol (0.434 mmol, 50 mg), CuSO4 (4.35 µmol, 1.1 mg), and K2CO3 (0.478 mmol, 66 mg) in 2 mL MeOH. The mixture was stirred overnight, and then concentrated in vacuo. Water (5 mL) and concentrated HCl (0.25 mL) were then added to the residue and the mixture was extracted with 10 mL of EtOAc for 3 times. The organic layer was washed 3 times with brine (10 mL), dried over MgSO4, and concentrated. The crude was purified by flash column chromatography (1:1 hexane/ EtOAc). Productcontaining fractions were combined and dried under vacuum for 1.5 h, affording 34 mg of pure product, corresponding to 55% yield.

Pilot rabbit study
Three groups of three female New Zealand White rabbits were used to test the dose response to CRM197glycopeptide 10V1S conjugates. Each group was immunized subcutaneously with CRM-g10V1S conjugate containing either 10µg, 50 µg or 100 µg antigen formulated in 50µL Adjuplex adjuvant, 4 times at 4-week intervals and blood was collected 2 weeks after each immunization. A prebleed was collected just before the first immunization.

Multi-immunogen rabbit study
Six groups of six female New Zealand White rabbits were used for the multi-immunogen study. Group 1 rabbits as a control group received subcutaneous immunizations of 50 µg CRM197-maleimide (with BME cap) and 50 µL Adjuplex. Group 2,3,4,5 were immunized with CRM197-glycopeptide 10F5M, CRM197-glycopeptide 10F2, CRM197glycopeptide 10F6, CRM197-glycopeptide 10F8, respectively, each containing 50 µg of respective glycopeptides per dose. Group 6 received sequential immunizations of 10F5M, 10F2, 10F6 and 10F8 glycopeptide conjugates with adjuvant as above. Four immunizations were performed at 4-week intervals and blood was collected 2 weeks after each immunization, with a pre-bleed just before the first immunization. All rabbits received 2 booster injections of 50 µg BG505.SOSIP.664 (T332N) and 50 µL Adjuplex. Throughout the study, 5 animals (roughly one per group) died of unknown causes with no obvious relation to the immunizations.

ELISA analysis
High-protein-binding flat-bottomed Maxisorp ELISA plates (Nunc-Immuno) were coated with 120 ng/mL antigen in coating buffer (50 mM carbonate/bicarbonate buffer, pH 9.6, 100 μL/well) and incubated at 4 °C overnight. The wells were washed twice with PBS-0.05% Tween 20 (PBS-T) and then blocked for 1 h at room temperature with 5% fat-free milk PBS-T (200 μL/well). After washing again twice with PBS-T, the wells were then incubated with either 3-fold or 4-fold serial dilutions of rabbit serum (starting at different concentrations: either 1:10 or 1:100) in 1% fat-free milk in PBS-T for 2 h at room temperature. The wells were washed 3 times before incubating with 100 μL of a horseradish peroxidase (HRP) conjugated sheep anti-rabbit antibody (Novex, part number A16172) at 1: 10,000 dilutions for 1 h at room temperature. After 3 washes, the wells were developed by adding 100 μL of 3,3′,5,5′-tetramethylbenzidine (TMB solution, Abcam Ab171522) for 3 min. The reaction was stopped by adding 100 μL of 1 M sulfuric acid and absorbance was measured at 450 nm wavelength. All measurements were performed in triplicate.

In vitro trimming of CRM197-glycopeptide 10F6 conjugates using rabbit serum
Alkyne-CRM-Ac-g10F6 conjugate (30 µg) was added to five 1.5 mL Eppendorf tubes, each containing 0.3 mL of rabbit serum. The serum mixtures were incubated at 37˚C and the reaction was stopped at different time points (4,17,24,48 h) by adding two mannosidase inhibitors (kifunensine and swainsonine, Santa Cruz Biotechnology, catalog # sc-201364 and sc-201362, respectively) directly to the serum mixture to obtain a 5 µg/mL concentration.
A 0 hr timepoint was generated by adding conjugate to serum already containing inhibitors. The quenched reactions were immediately frozen and stored at -80˚C until the last time point sample was obtained. Next, Dde-biotin-picolyl-azide (Click Chemistry Tools, catalog # 1186-5) (25 µL from 10 mM stock in DMSO) and THPTA (50 µL from 100 mM stock in PBS buffer pH 7.5) were added to the serum mixtures. The mixtures were then vortexed briefly to mix. Next, CuSO4 5H2O (50 µL from 20 mM stock in PBS buffer) was added and mixed briefly. Lastly, sodium ascorbate (50 µL from 300 mM stock in PBS buffer) was added and also vortexed to mix. The click reactions were covered with aluminum foil to protect from light and they were stood for 2h at room temperature. Excess Dde-biotin-picolyl-azide linker was removed by buffer exchange through a 30K Amicon filter (6 rounds of filter, 10 min each round). The clicked CRM-Ac-g10F6 conjugates were incubated with 100 µL of NeutrAvidin resin (ThermoFisher, catalog # 29200) in 0.8 mL centrifuge columns for 1.5 h at room temperature. The NeutrAvidin resin was then washed 4 times with 200 µL water. Then, the washed resin was incubated in elution buffer (2% v/v hydrazine solution) for 1h at room temperature and washed 4 times with 200 µL water. The eluant and all the washes were combined and buffer exchanged by 30K Amicon filter with water. The recovered CRM-Ac-g10F6 conjugates were concentrated and lyophilized.

Identification and quantification of intact and trimmed glycoforms of the tryptic glycopeptides from g10F6 by mass spectrometry (MS) analysis:
To determine the presence of G10F6 trimmed glycoforms resulting from serum mannosidase activity on the G10F6 glycopeptide, a pure G10F6-CRM (standard) sample and the five time-point serum-incubated samples (0, 4, 17, 24, and 48 hr) described above were analyzed the same way, as described below. Prior to mass spectrometry analysis, the samples were reduced, alkylated, trypsin digested, SP-C18 desalted and concentrated. Tryptic digestion would be expected to generate two glycopeptides (formylXLXFIR and XQYVYHAPLLTXVR), each containing two homopropargylglycine residues with triazole cyclohexyl linker and glycan attached (X). The data for all MS acquisition experiments were acquired as follows: the mass spectrometer was operated in positive ion mode and the Orbitrap analyzer was used as the detector for all scan events; all data were acquired in Profile mode. The sample ions were introduced into the mass spectrometer (MS) through an Ion Transfer Tube operated at 300 °C. To minimize in-source fragmentation, the source RF Lens was operated at 5%. Data from the MS tandem experiments were acquired with the following scan event parameters: The MS 1 scan was set at a resolution of 120,000 @ m/z 200, over the full scan range m/z 350-1500, 1 µscan/spectrum, maximum injection time (ion accumulation time) of 50 ms with a target automatic gain control (AGC) of 4 x 10 5 ion population. The following filters were applied to the data dependent acquisition scan events: Monoisotopic Peak Determination was set to Peptide; Charge States included 2-7, the Dynamic Exclusion was set to exclude after 1 time, for a duration of 10 s with a ± 10 ppm window, Excluding Isotopes was set to True; the Intensity Threshold was set to 4 x 10 4 ; the Data Dependent Mode was set to Cylce Time (Top Speed Methodology) with a Master Scan every 3 s. The MS 2 scan event used the Quadrupole for Isolation Mode, with an Isolation Window of m/z 1.6; the activation used was HCD at 45 % for the glycopeptide identification focus experiment or 15% for the glycan composition focus experiment. The MS 2 scan range was set to Auto with m/z range set to High; the first mass was fixed to m/z 100; the AGC target was set to 5.5 x 10 5 ion population and maximum injection time of 150 ms; 2 µscan/spectrum.
The relative quantification focus MS-only experiment MS 1 scan followed the same parameters of the tandem MS 1 scan events in the tandem methods, the only difference being that the MS scan range was set as m/z 350-2000. The glycopeptides were identified by manually assigning peptide backbone and glycan loss peaks in the HCD 45% tandem data; the glycan compositions of such glycopeptides were confirmed in the HCD 15% tandem data. To quantify the relative abundances of the g10F6 glycopeptides' glycoform distributions (the full [Man9] and trimmed glycoform versions) for the standard sample (pure glycopeptide-CRM conjugate) and all time point samples (serum-treated samples) the area under the chromatographic peak corresponding to each glycopeptide precursor ion was calculated by the Thermo Scientific Xcalibur's Qual Browser Software, using the extracted ion chromatogram for each observed charge state. The extracted areas were normalized by charge ( = ∑ ( / )). The percentage for each chromatographically resolved Mann glycoform in the standard sample were calculated and then subtracted from the total signal observed for the corresponding Mann glycoform at each time point. The values were also corrected to take into account a small amount of in-source fragmentation (~4-5%) observed for the standard. This corrected amount was then calculated as a percentage distribution and plotted with Excel. The results are shown in Table S10.

General synthetic methods for Man1-5-Cy-N3 preparation
All synthesis reagents were purchased from Sigma-Aldrich, Acros Organics, Fluka, Alfa Aesar or Strem and used without further purification unless otherwise noted. Toluene, THF, DCM, Ethyl Ether and Pentane were deoxygenated by argon purging and dried by passage through activated alumina columns, then stored under argon gas only briefly before use. DriSolv Acetonitrile, DMSO and Methanol were purchased from EMD. Amines (Et3N, iPr2NEt, Pyridine, 2,6-Lutidine and 2,6-di-t-butylpyridine) were refluxed over CaH2 and freshly distilled before use. Glassware was flame dried or dried in a 150 °C oven. For glycosylations, carbohydrate donors and acceptors were azeotropically dried by the following procedure: the intermediate was dissolved in dry toluene, the solution was cooled to -78 °C, vacuum was applied, and the cooling bath was removed to allow the toluene to evaporate while the mixture warmed to room temperature. The flask was then backfilled with dry nitrogen and this procedure was repeated a total of three times. SiliCycle Siliaflash P60 silica was used for flash column chromatography. Analytical thin layer chromatography (TLC) was performed using SiliCycle glass backed plates (Cat# TLG-R10011B323). TLC plates were analyzed by short wave UV illumination, or by staining with dipping in cerium-ammoniummolybdate (CAM) stain (40 g of ammonium pentamolybdate, 1.6 g of cerium (IV) sulfate, 800 mL of diluted sulfuric acid (1:9, with water, v/v)) and heating on a hot plate. All 1 H and 13 C NMR spectra were obtained on a Varian iNova 400 instrument in CDCl3, internally referenced to TMS, or D2O externally referenced to sodium 3-(trimethylsilyl)propanesulfonate. Chemical shifts are reported in parts per million (ppm), and coupling constants are reported in Hz. Coupling is referred to with the following abbreviations (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, dd = doublet of doublets, app d = apparent doublet, app t = apparent triplet). For NMR spectra in which large numbers of resonances are unresolved, only the clearly-resolved "selected signals" are listed in text-format listing of data. LC/MS analysis was performed on a Waters Acquity UPLC equipped with photodiode array and Waters Micromass ZQ4000 mass detector (Column: Waters ACQUITY UPLC BEH ® C18, 1.7 μm, 130Å, 2.1 x 50 mm. Waters ACQUITY UPLC BEH ® HILIC, 1.7 μm, 130Å, 2.1 x 150 mm.). Optical rotation was measured using a Jasco digital polarimeter. Infrared spectra were obtained using a Nicolet IR200 spectrometer with a diamond ATR. DCM stands for dichloromethane, DDQ stands for 2,3-dichloro-5,6-dicyanobenzoquinone, DTBP stands for 2,6-di-tert-butylpyridine, EA stands for ethyl acetate, TES-H stands for triethylsilane, THF stands for tetrahydrofuran, NIS stands for N-Iodosuccinimide.