A comprehensive synthetic library of poly-N-acetyl glucosamines enabled vaccine against lethal challenges of … – Nature.com

Synthesis of the PNAG oligosaccharide library

To date, only fully acetylated and fully deacetylated PNAG oligosaccharides have been investigated as immunogens for vaccine studies13,14,15. The availability of a library of PNAG oligosaccharides with systematically varied numbers and locations of free amines can greatly aid in the identification of the maximally protective epitope structures. We aimed to synthesize a comprehensive library of 32 pentasaccharides designated PNAG0PNAG31 (Fig.1) fully covering the free amine space of PNAG. The reducing ends of the target pentasaccharides bear a linker terminated with a disulfide group, which can be reduced for chemoselective conjugation to a carrier protein.

The five-digit number in the bracket for each compound codes for free amine (0) or N-acetamide (1) at residues ABCDE from the non-reducing end to the reducing end of the pentasaccharide, respectively. The five-digit number was then viewed as a binary number and converted to the decimal system as the compound number. For example, 01010 in binary number is equivalent to 10 in the decimal system. Thus, the PNAG pentasaccharide bearing N-acetylation at units B and D only is named as PNAG10.

While several PNAG structures have been synthesized before16,17,18,19,20, a general method for the expeditious construction of a comprehensive PNAG pentasaccharide library is lacking. To accelerate the library synthesis, rather than starting from monosaccharide building blocks for each targeted pentasaccharide, we envisioned the overall efficiency can be significantly enhanced with a divergent strategy. In our synthetic design, the amine groups of strategically protected pentasaccharides are differentiated by orthogonal protective groups for selective deprotection and acetylation. After screening multiple synthetic intermediates, we developed two key linchpin pentasaccharide intermediates (1 and 2), which bear four protective groups, i.e., tert-butyloxycarbonyl (Boc), allyloxycarbonyl (Alloc), 2,2,2-trichloroethoxycarbonyl (Troc), and fluorenylmethoxycarbonyl (Fmoc), on glucosamine units A, B, C, and D. The reducing end glucosamine unit E is N-acetylated (for compound 1) or N-trifluoroacetylated (for compound 2) (Fig.2A).

A Structures of two key linchpin pentasaccharide intermediates (1 and 2). B Synthesis of the reducing end glucosamine building block 8. C Syntheses of compound 13; D Syntheses of compound 1. Ac acetyl, Alloc allyloxycarbonyl, Bz benzoyl, TBDPS tert-butyldiphenylsilyl, Boc tert-butyloxycarbonyl, DIPEA diisopropylethylamine, Fmoc fluorenylmethoxycarbonyl, HATU hexafluorophosphate azabenzotriazole tetramethyl uronium, and Troc 2,2,2-trichloroethoxycarbonyl.

Based on the above design, our synthesis commenced from thioglycoside 3, which glycosylated 3-azido-1-propanol 4 to provide compound 5 in 82% yield (Fig.2B). Upon removal of the Alloc group from 5 and N-acetylation, the resulting compound 6 was subjected to azide reduction, amidation of the free amine with carboxylic acid 7, and protective group adjustments leading to compound 8 in 45% yield for the four steps.

Oligosaccharide assembly started from the CD disaccharide 9 containing N-Troc and N-Alloc groups (Fig.2C). Thioglycoside donor 10 was preactivated with the p-TolSCl/AgOTf promoter system21 at 78C. Upon complete activation, the thioglycosyl acceptor 11 was added to the reaction mixture leading to disaccharide 9 in 83% yield (Fig.2C). In order to extend the glycan chain, the glycosylation of acceptor 8 with disaccharide 9 was performed. When the reaction was first carried out under the pre-mix condition, i.e., 9 and 8 were mixed together followed by the addition of promoter (p-TolSCl/AgOTf or NIS/TfOH22,23), little desired trisaccharide 12 was obtained, which was likely due to the activation of the thioester moiety by the thiophilic promoter. Next, the reaction was explored under the pre-activation condition by activating 9 with the promoter p-TolSCl/AgOTf first, followed by the subsequent addition of acceptor 8. This change of the reaction protocol successfully produced trisaccharide 12 in 71% yield. Replacement of Alloc with Fmoc and removal of TBDPS group from 12 resulted in the trisaccharide 13. To extend 13 to a pentasaccharide, the Troc moiety of disaccharide 9 was replaced with Boc (disaccharide 14, Fig.2D). Pre-activation-based glycosylation of 14 and 13 produced pentasaccharide 1, which contains four different N-protective groups on units A, B, C, and D. Analogously, pentasaccharide 2 was synthesized with four different N-protective groups on units A, B, C, and D, and the N-TFA group on unit E (Supplementary Fig.1).

With the two key pentasaccharides in hand, we explored orthogonal deprotection of pentasaccharides 1 and 2. As an example, the Boc and Alloc groups of compound 2 could be removed by 90% aqueous TFA and Pd(PPh3)4/PhSiH3, while deprotections of Troc and Fmoc were accomplished using Zn/AcOH and 20% piperidine in N,N-dimethylformamide (DMF), respectively, without affecting any other amine protective groups (Fig.3A). These results suggest that the four amine protective groups could be independently removed specifically.

A Orthogonal deprotection of pentasaccharide 2. B Divergent syntheses of 16 PNAG pentasaccharides from the strategically protected pentasaccharide 1. C Divergent syntheses of 16 PNAG pentasaccharides from the strategically protected pentasaccharide 2. Ac acetyl, Alloc allyloxycarbonyl, Bz benzoyl, TBDPS tert-butyldiphenylsilyl, Boc tert-butyloxycarbonyl, DIPEA diisopropylethylamine, DMF dimethylformamide, Fmoc fluorenylmethoxycarbonyl, HATU hexafluorophosphate azabenzotriazole tetramethyl uronium, Troc 2,2,2-trichloroethoxycarbonyl, and TFA trifluoroacetic acid.

With the orthogonal deprotection conditions established, divergent modifications of the key pentasaccharide intermediates were carried out. Treatment of pentasaccharide 1 with 90% TFA cleaved both Boc and TBDPS groups (Fig.3B). Upon acetylation of the newly liberated hydroxyl and amine moieties, the Alloc, Troc, and Fmoc groups were subsequently removed followed by full O- and S-deacylation with 20% hydrazine hydrate in MeOH, affording PNAG17 pentasaccharide in 48% overall yield bearing the N-acetylglucosamine (GlcNAc)-glucosamine (GlcN)-GlcN-GlcN-GlcNAc (10001) sequence. Alternatively, following TFA treatment of 1, the Fmoc group was cleaved, which was then acetylated with subsequent removal of Troc, Alloc, and Bz moieties to produce pentasaccharide PNAG19 with the GlcNAc-GlcN-GlcN-GlcNAc-GlcNAc sequence (10011) in 51% overall yield. Similar divergent modification processes on the two key pentasaccharides 1 and 2 produced the full library of 32 PNAG pentasaccharides with all possible combinations of free amines in each glucosamine unit of the pentasaccharides (Fig.3B, C).

As carbohydrate antigens in general are T cell independent B cell antigens24 and small oligosaccharides alone are not immunogenic25, these types of antigens need to be conjugated to an immunogenic carrier in order to induce anti-carbohydrate IgG antibody responses. The mutant bacteriophage Q (mQ)26 is a powerful carrier and likely highly useful for carbohydrate based conjugate vaccines27,28,29. As PNAG oligosaccharides can potentially contain multiple free amine moieties, we resorted to sulfhydryl chemistry for PNAG/mQ conjugation. The mQ A38K/A40C/D102C was expressed in E. coli, purified, and incubated with the bifunctional linker succinimidyl 3-(bromoacetamido)propionate (SBAP) 19 to react with free amines on the mQ surface (Fig.4A). Upon removal of the excess linker, the SBAP functionalized mQ was added to the PNAG pentasaccharide followed by quenching the unreacted bromoacetamide moieties on mQ with cysteine to avoid any potential side reactions of residual bromoacetamide on mQ upon storage or following vaccination. MALDI-TOF mass spectrometry (MS) analysis of the mQPNAG conjugate showed an average loading of 250 copies of pentasaccharide per particle (Supplementary Fig.2A). It is known that the antigen loading density on Q can significantly impact the levels of antibodies induced against the target antigen29,30. When the loading level of antigen was low (<50 copies per particle), despite the same total amount of antigen administered, the antibody responses induced were low29,30. Increasing the local density of the antigen on the particle (over 100 antigens per particle) can significantly improve the antibody responses, which is presumably due to the more effective crosslinking of the B cell receptors on B cells31. The loading density of PNAG on the mQPNAG conjugate was higher than the threshold antigen level needed for powerful B cell activation.

Syntheses of A mQPNAG, B TTHcPNAG, and C BSAPNAG conjugates. SBAP succinimidyl 3-(bromoacetamido)propionate, TCEP tris(2-carboxyethyl)phosphine, TThc tetanus toxoid heavy chain.

With the mQPNAG conjugates in hand, their abilities to induce anti-PNAG antibodies were evaluated. The conjugate of TT with the PNAG pentasaccharide bearing five free amines (5GlcNH2TT) has undergone a phase 1 human clinical trial32. To compare with our mQPNAG conjugate, we covalently linked PNAG0 (5GlcNH2) with the TT heavy chain (TTHc) using SBAP and achieved an average loading of 4.7 PNAG0 per protein molecule (Fig.4B and Supplementary Fig.2B). The recombinant TTHc is a suitable surrogate of TT33. As the molecular weight of mQ particle (2540kDa for the protein shell) is about 49 times that of the TTHc (MW~52kDa), the overall densities of PNAG0 on mQPNAG0 and TTHcPNAG0 were similar.

Head-to-head comparative immunogenicity studies of the mQPNAG0 and the TTHcPNAG0 conjugates were carried out. Groups of female C57Bl6 mice (n=5 per group) were immunized with freshly prepared mQPNAG0 (8nmol corresponding to 8g of PNAG0 per injection) or the TTHcPNAG0 conjugate (8nmol PNAG0 per injection) on days 0, 14, and 28. Monophosphoryl lipid A (MPLA, 20g) was added to each vaccination as the adjuvant. A control group of mice received a mixture of mQ with PNAG0 at equivalent total amounts of mQ, PNAG0, and MPLA following the same immunization protocol. On day 35, sera were collected from all mice.

To analyze the levels of antibodies generated, enzyme linked immunosorbent assay (ELISA) analyses were performed. To avoid the interference of anti-mQ antibodies in the sera, the 32 PNAG pentasaccharides were conjugated with BSA individually (Fig.4C and Supplementary Fig.3) and used as the ELISA coating antigens. As shown in Fig.5A, mQPNAG0 induced high anti-PNAG IgG titers (EC50 IgG titers GMT 75,613, measured against BSAPNAG0) while the IgM titers were negligible (GMT<1000). Furthermore, high levels of anti-PNAG0 IgG responses were observed more than 1 year after the initial immunization (Fig.5B). The IgG levels could be boosted back to near peak levels after nearly 2 years indicating that the mQ conjugate induced PNAG0 specific memory B cells through immunization. The GMT of 75,613 achieved in mice receiving the mQPNAG0 was significantly (P<0.0001, Dunnetts multiple comparisons test) higher than the anti-PNAG0 IgG titers achieved in mice immunized with the corresponding TTHcPNAG0 conjugate (GMT 4765), highlighting the superior immunogenicity of the mQ carrier for conjugate vaccines. Mice immunized with the admixture of mQ and PNAG0 did not produce any detectable levels of anti-PNAG0 IgG (GMT<1000), accentuating the critical need to covalently conjugate mQ with PNAG0.

A C57Bl6 mouse (n=5 per group) antibody responses at day 35 after immunization. The EC50 value (the fold of serum dilution that gives half-maximal binding) of the IgG titers to the immunizing oligosaccharide was plotted with each symbol representing one animal and the horizontal line is the geometric mean value of the titers within the group. The ELISA titers were determined using the BSAPNAG conjugate containing the same PNAG structure as the immunizing QPNAG construct. One-way ANOVA allowed for rejection of the null hypothesis that all groups have the same mean IgG titers (P<0.0001). Statistical significance was performed by Dunnetts multiple comparisons post-hoc test. ****P<0.0001; B Anti-PNAG0 IgG antibody responses of mice (n=5) immunized with mQPNAG0 monitored over time with mean titers plotted. Data are presented as mean valuesstandard deviation of the titer numbers from five mice. The arrows indicate days of vaccination (days 0, 14, 28, 360, and 655). The antibody responses could be boosted more than 650 days after prime vaccination. Source data are provided as a Source Data file.

As C57Bl6 mice are inbred, to enhance the rigor of our study, we immunized outbred CD1 mice with the mQPNAG0 conjugate following the same immunization protocol. mQPNAG0 was able to elicit comparably high titers of anti-PNAG0 IgG antibodies on day 35 after the primary series of immunization in CD1 mice (Supplementary Fig.4).

The precise PNAG sequences synthesized by pathogens such as S. aureus are not known. Furthermore, the most abundant PNAG structure on cell surfaces that would encompass a highly (8095%) acetylated polysaccharide is not a protective epitope13. To guide vaccine design, we envisioned anti-PNAG mAb F598 could provide valuable information regarding optimal acetylation patterns in a PNAG pentasaccharide. Isolated from a patient who recovered from an S. aureus infection, mAb F598 can protect mice against S. aureus infections34. The 32 PNAG pentasaccharideBSA conjugates were immobilized onto a glycan microarray35. Following incubation of mAb F598 with the microarray and washing, the amount of antibody remaining bound was quantified with a fluorescent secondary antibody. Interestingly, although mAb F598 was initially identified due to binding to deacetylated PNAG with only ~15% N-acetylation34, it had little binding to glycan PNAG0 or any glycans containing only one Ac moiety. Highly acetylated PNAG such as PNAG30 and PNAG31 with four or more consecutive GlcNAcs were among the strongest binders (Fig.6). Both the location and the number of NHAc are important for F598 binding, supporting the idea of an amine/acetylation code. For example, despite having the same total number of NHAcs (4 in the molecules), PNAG23 (10111) is a weak binder with an apparent affinity <5% of that with PNAG30 (11110). Out of the PNAGs with two or three GlcNAc residues, PNAG10 and PNAG26 were the strongest binders, respectively.

A Relative fluorescence unit (RFU) of F598 mAb binding with the library of 32 PNAG pentasaccharides. The glycans are grouped according to the number of NHAc units in the molecule. Each PNAG sequence is printed five times on the glycan microarray. The error bars represent the standard deviations of five individual spots. Data are presented as mean valuesstandard deviation. F598 generally prefers highly acetylated PNAG sequences. Both the location and the number of NHAc units are important determinants of F598 binding. B Quantification of the preference of F598 for acetylation at each site of the PNAG pentasaccharide. The mean values are calculated from the values of the binding intensities of all 32 PNAG sequences to F598. Each PNAG sequence is printed five times on the glycan microarray. Data are presented as mean valuesstandard deviation. Source data are provided as a Source Data file.

To better interpret the binding data, we quantified the GlcNAc binding preference of F598 by computing the preference index (P) for each unit of the pentasaccharide as

$${P}_{i}=frac{mathop{sum}limits_{j}{R}_{j}times {A}_{i}}{mathop{sum}limits_{j}{R}_{j}}$$

(1)

where (i) (AE) is the site of monosaccharide from the non-reducing end to the reducing end, (j) (031) is the serial number of glycan, (R) is the intensity of the binding signal (RLU), and (A) is the code for amine vs acetylation (A=1 for free amine and A=1 for NHAc). P value indicates the conditional probability difference between finding an NHAc or free amine for binding, which ranges from 1 to 1 with 1 and 1 indicating a complete preference for free amine or NHAc, respectively, at the specific site. As shown in Fig.6B, unit B position showed the highest P value of 0.91, suggesting on average that there is a 95.5% chance to find an NHAc moiety rather than a free amine on saccharide B for ligand binding with F598. The P values for sites A, C, and E were between 0.31 and 0.54 indictive of a moderate global preference for N-acetylation. There were almost no preferences for NHAc or free amine for site 5 as the P value at this site was close to 0.

The importance of an NHAc at unit B identified from microarray binding can be rationalized by the crystal structure of F598 complexed with fully acetylated PNAG oligosaccharides (PDB 6be4)36. The binding pocket of F598 could accommodate PNAG with five GlcNAc residues. The NHAc groups on saccharides B and D in the binding pocket were deeply inserted into the groove clamped by the heavy and the light chain of the mAb, forming multiple hydrogen bonds, while the NHAcs on units A, C, and E only had weak to moderate interactions with the antibody. The carbonyl oxygen of the NHAc on saccharide B forms a hydrogen bonding with light chain A32 backbone amide while bridging with light chain R52 residue via a water molecule. The carbonyl oxygen of NHAc on saccharide D also formed hydrogen bonds with light chain A97 backbone amide and the hydroxyl of heavy chain Y50. Those interactions supported the relatively high dependence of NHAc on sites B and D for the binding of F598.

Based on the microarray results and the report that antibodies raised against the fully acetylated PNAG antigen were poorly protective13,14,15, we selected PNAG10 and PNAG26 as new PNAG oligosaccharide antigens for further evaluations. PNAG10 has the strongest binding to F598 among all PNAG structures with two or fewer NHAcs, and PNAG26 is the best binder among all structures with three or fewer NHAcs. Both PNAG10 and PNAG26 have NHAcs on glycan sites B and D. PNAG0 was utilized as a positive control since the corresponding TTPNAG0 construct (5GlcNH2TT) has entered clinical trials [ClinicalTrials.gov Identifier: NCT02853617].

C57/Bl6 mice were immunized with the mQ conjugates of PNAG10 or PNAG26 following the aforementioned immunization protocol (8nmol PNAG, three injections on days 0, 14, and 28 with MPLA adjuvant). ELISA analysis of the immune sera showed significantly enhanced IgG antibody titers against the immunizing antigen (PNAG10 or PNAG26) with GMTs of 191,141 and 227,064 ELISA units, respectively, as compared to pre-immune sera (Fig.5A). Similarly, mQPNAG10 or PNAG26 conjugates induced high levels of anti-PNAG10 and anti-PNAG26 IgG antibodies, respectively, in CD1 mice (Supplementary Fig.4).

To demonstrate the immunogenicity of the mQ conjugates in an additional mammalian species, New Zealand white rabbits were immunized with mQ conjugates of PNAG0, PNAG10, and PNAG26 (8nmol PNAG per injection) following a similar prime-boost protocol as that used in the mouse study. ELISA analysis of the post-immune sera showed that all three constructs induced strong anti-PNAG IgG responses with EC50 titers over 100,000 ELISA units (Fig.7A), while those for the pre-immune sera were below 1000 ELISA units. No side effects due to vaccinations were observed in either rabbits or mice.

A IgG antibody titers to the immunizing PNAG oligosaccharide in rabbit (n=2 per group) sera on day 35 after prime vaccination. B IgG antibody titers in pooled rabbit sera from mQ-conjugate or 5GlcNH2TT conjugate immunized animals (n=2 per group) as well as titer of natural human IgG in pooled human serum against native PNAG polysaccharide purified from Acinetobacter baumannii. The numbers above symbols are the average titer numbers. Titers and 95% confidence intervals (CI) were determined by linear regression using log10 values of the average of replicate serum dilutions to determine the X intercept and 95% CI when Y=0.5 (OD405nm of ELISA plate reading). C Stacked bar graphs depicting the IgG signals at the serum dilution of 1:50,000 for each rabbit (n=2) immunized with mQPNAG0, mQPNAG10, and mQPNAG26 as well as pre-immune sera, respectively, on the array. The complete microarray results are provided in the Source Data file; D Normalized binding of the comprehensive library of PNAG pentasaccharides by IgG antibodies from post-immune sera of rabbits immunized with mQPNAG0, mQPNAG10, and mQPNAG26, respectively, as well as pre-immune sera. PNAG sequences are grouped together according to the total number of acetamides in the molecules. The color scale bar is shown on the right with 100% indicating the strongest binding to a PNAG component and 0% indicating the weakest binder. For each antigen, the two rows represent sera from two rabbits per group immunized with the specific construct. Source data are provided as a Source Data file.

We analyzed next the recognition of native PNAG using PNAG polysaccharide isolated from Acinetobacter baumannii37,38 as the coating antigen for ELISA. As shown in Fig.7B, control sera from rabbits immunized only with the Q carrier did not bind with PNAG. In contrast, sera from rabbits immunized with mQPNAG0, mQPNAG10, and mQPNAG26 exhibited strong binding with mQPNAG26 antiserum having the highest titer (1,584,983 ELISA units) to the native microbial polysaccharide. As a comparison, sera from the conjugate of 5GlcNH2TT13,15 immunized rabbit only gave a titer of 501 ELISA units (Fig.7B). Normal human sera containing natural antibodies to PNAG had an average ELISA titer of 631 ELISA units. These results further highlight the potential of mQPNAG conjugates as vaccines.

Analysis of the microarray binding by post-immune sera revealed selective PNAG epitope recognition by the post-immune sera (Fig.7D). Rabbits immunized with mQPNAG0 produced serum IgG antibodies exhibiting the strongest binding with the immunizing PNAG0 antigenic structure. Other good binders include PNAG1 and PNAG8, both having a single GlcNAc in the structure. Interestingly, for PNAG4 with the sequence of GlcN-GlcN-GlcNAc-GlcN-GlcN, although it also only contains one GlcNAc, it had much lower binding with the sera (about 30% that to PNAG1). This suggests that three or more consecutive GlcNs are important for binding by anti-PNAG0 sera.

mQPNAG10 immunized rabbits produced serum antibodies that preferentially bind to PNAG8 (01000) and PNAG10 (01010), which differ only by the GlcNAc in residue D indicating the non-reducing end GlcN-GlcNAc-GlcN may be the main epitope. Serum antibodies from mQPNAG26 (11010) immunized rabbits preferentially bound to PNAG25 (11001), PNAG26 (11010), PNAG8 (01000), and PNAG16 (10000) suggesting GlcNAc-GlcNAc-GlcN and GlcNAc-GlcN-GlcN may be part of the epitopes being recognized.

For an effective vaccine, it is important to establish that the post-immune sera bind not only the isolated antigen but also the antigen expressed on pathogen cells. We reacted S. aureus ATCC29213 cells with rabbit immune sera and the bound antibodies were detected by a fluorescently labeled anti-rabbit IgG secondary antibody. As shown in Supplementary Fig.5A, fluorescence microscopy images showed stronger binding to bacterial cells by IgG antibodies in mQPNAG10 and mQPNAG26 immune sera compared to sera from mQPNAG0 immunized rabbits or pre-immune sera. To validate pathogen recognition observed in fluorescence images, whole cell ELISA was performed. S. aureus cells were coated on ELISA plates, incubated with rabbit immune sera, and detected by secondary antibodies. The post-immune sera exhibited significantly higher titers in binding with the cells compared to pre-immune sera (Supplementary Fig.6).

For antibody-mediated complement deposition39, we added various immune sera to wells coated with purified PNAG isolated from Acinetobacter baumannii37,38 along with IgG/IgM depleted 2.5% human complement (Fig.8A). After incubation, the immobilized complement component C1q was detected by anti-C1q antibodies. As shown in Fig.8A, sera from mQPNAG10 and mQPNAG26 deposited significantly more C1q than those from mQPNAG0 immunized rabbits, which in turn had more potent C1q binding than antibodies in sera from rabbits immunized with the 5GlcNH2TT conjugate13,15.

A Complement deposition tests were performed as described39 using pooled sera from rabbits (n=2 per group) immunized with mQPNAG conjugates, the 5GlcNH2TT conjugate, or from a sample of pooled normal human sera. Titers and 95% confidence intervals (CI) were determined by linear regression using log10 values of the average of replicate serum dilutions to determine the X intercept and 95% CI when Y=0.5 (OD405nm of ELISA plate reading). P values indicate the significance of the deviation of the slope of the titration curve from zero to identify sera with activity at P<0.05. mQPNAG10 and mQPNAG26 conjugates were more potent than the mQPNAG0 and 5GlcNH2TT conjugate in inducing C1q deposition onto purified PNAG. Normal human serum had no significant C1q depositing activity in spite of having a binding titer to PNAG (see Fig.7B) consistent with prior reports that naturally acquired human antibody to PNAG is not functional due to the inability to activate the complement pathway12,34. Titers were determined by simple linear regression. B Pooled sera from rabbits (n=2 per group) immunized with mQPNAG conjugate led to significantly higher levels of opsonic killing activities against S. aureus cells. Three aliquots were prepared from each pooled serum and the individual values of the three aliquots were presented. Source data are provided as a Source Data file.

The abilities of the post-immune sera to kill bacteria in vitro were evaluated next via the opsonophagocytic killing (OPK) assay. S. aureus cells were treated with pooled rabbit immune sera, followed by the addition of complement/phagocytic cells and quantification of the number of bacterial cells surviving the opsonic killing. As shown in Fig.8B, while the pre-immune sera were completely ineffective, all three constructs induced antibodies with potent in vitro killing activity. mQPNAG26 (EC50: 2534) and mQPNAG10 (EC50: 3045) showed higher EC50 OPK titers as compared to mQPNAG0 (EC50: 1345). Omitting either immune sera, complement or phagocytic cells resulted in complete loss of killing activity (Supplementary Fig.7), indicating the need for all three components for protective immunity.

The efficacy of the various vaccine constructs in protecting against bacterial infection was tested in two mouse bacteremia challenge models. According to the CDC, bloodstream infections by S. aureus are serious threats with nearly 20,000 deaths per year in the USA4. For the in vitro study, we first compared the mQPNAG0 vs TTHcPNAG0 construct. In the active protection model, mice were immunized three times with mQPNAG0 or TTHcPNAG0 at equivalent doses (8nmol PNAG0) (n=20 for each group) (Fig.9). Another group of control mice received a mock injection of saline. Two weeks following the last vaccination, each mouse was challenged via the tail vein with 10*LD50 of the S. aureus strain ATCC29213. Mice that had received saline all died within 2 days of bacterial challenge. On the other hand, 95% of the mice receiving mQPNAG0 were protected against death from this pathogen. The survival rate of the mQ vaccine group was significantly better than TTHcPNAG0 vaccinated group (p=0.0154, logrank test) (Fig.9A). Bacteria were detected in the kidneys of 35% (7 out of 20) of the mice immunized with TTPNAG0, while mQPNAG0 vaccination reduced the recovered levels of S. aureus from the kidneys with bacteria only observed in 5% of the mice (1 out of 20) (Fig.9B). Contingency table analysis of the proportion of the 20 immunized mice in each group with or without detectable S. aureus by Fishers exact test showed significantly (p=0.0436) fewer infected kidneys in the mQPNAG0 immunized group, with a relative risk of 0.68 (95% CI=0.450.93). Thus, disease burden evaluated by the levels of S. aureus in mouse kidneys was significantly better in mQPNAG0 vaccinated group compared to those receiving the TTHcPNAG0 vaccine. These results further support the superior performance of the mQ carrier.

Immunization with mQPNAG0 effectively A protected against S. aureus infection, and B reduced bacterial count in mouse kidney. mQPNAG0 was significantly better than TTHcPNAG0 in protecting mice and reducing disease burden (n=20 for each group). Logrank tests were performed for statistical analysis. P values were presented in the graph. ****P<0.0001. Source data are provided as a Source Data file.

As the mQPNAG0 immunogen gave almost complete protection in the active protection model in mice, we next established a passive protection model to differentiate the various mQPNAG constructs, using rabbit sera transferred to mice. The passive model can be a more stringent test by using more dilute sera for protection. Rabbit sera were diluted 800-fold and administered intraperitoneally to mice, which were then challenged with 10*LD50 (200 million cells) of S. aureus ATCC29213 via the tail vein (Fig.10). While all control mice receiving the pre-immune sera died within 3 days of this challenge, all post-immune sera from PNAG0, PNAG10, or PNAG26 immunized rabbits bestowed significant protection.

Transfer of antisera from mQPNAG immunized rabbits to mice A provided significant protection to mice (n=10 per group) against the lethal challenges by S. aureus ATCC29213. Statistical analysis was performed with the logrank test. ****P<0.0001; and B significantly reduced bacterial count in mouse kidneys. The combination of sera from mQPNAG0 and mQPNAG26 immunized rabbits provided complete protection to mice. Statistical analysis for survival was performed using the logrank test. Analysis of S. aureus cfu/gm was by KruskalWallis non-parametric ANOVA (P<0.0001 for overall effect of serum given). P values for pairwise comparisons are shown on graph by Dunns multiple comparisons test. Transfer of antisera from mQPNAG immunized rabbits to mice C provided significant protection to mice against the lethal challenges by MRSA strain 1058 (n=10 per group); statistical analysis for survival was performed using the logrank test; and D reduced bacterial count in mouse kidneys. Sera from mQPNAG26 immunized rabbits provided the highest protection to mice. The horizontal line represents the median value of each group. Statistical analysis for survival was performed using the logrank test. Analysis of MRSA cfu/gm was by KruskalWallis non-parametric ANOVA (P=0.0967). P values for pairwise comparisons are shown on graph by Dunns multiple comparisons test. Source data are provided as a Source Data file.

Mice receiving sera from mQPNAG26 and mQPNAG10 immunized rabbits showed higher survival rates than those receiving PNAG0 sera (Fig.10A) (60% and 50%, respectively, vs 30%) and lower pathogen load compared to mQPNAG0 sera supporting the in vitro opsonic killing data (Fig.10B). We next tested the combination of two sera. Interestingly, administering the mixed PNAG26 and PNAG0 sera (1:1 ratio with each individual serum diluted 1600 times, which is regarded equivalent in concentration to 1:800 dilution of a single serum) provided 100% protection to mice against the 10*LD50 challenges with S. aureus (Fig.10A). The kidneys of mice receiving the combination of PNAG26 and PNAG0 sera had no detectable bacteria (Fig.10B). The higher protective efficacy observed with the combined sera was presumably because the PNAG polysaccharide can be heterogenous in the amine/acetylation patterns. While some of the sequences such as the fully deacetylated PNAG0 may be rare within the native PNAG polysaccharide, antibodies generated by mQPNAG0 can complement those by mQPNAG26. Thus, the combination of two mQPNAG constructs can broaden bacterial recognition and enhance protection against bacterial challenges.

The emergence of MRSA is a pressing public health concern40. Effective vaccines can provide a complementary tool to combat S. aureus infections and reduce the reliance on antibiotics. The post-immune rabbit sera were tested against multiple MRSA strains including six clinical strains first via immunofluorescent staining (Supplementary Fig.5B and Supplementary Table1). All three mQPNAG sera recognized the seven strains tested highlighting the breadth of immune recognition. A control strain lacking PNAG expression with icaA gene knock out (954) showed negligible binding by the immune sera, indicating the recognition is PNAG dependent. Next, rabbit sera were diluted 800 times and administered to mice, which were then challenged with 10*LD50 (200 million cells) of the MRSA strain 1058 via the tail vein (Fig.10C). Sera from mQPNAG26 immunized rabbits protected 90% of the mice from MRSA-induced death, which was significantly higher than the 40% protection by mQPNAG0 sera. Correspondingly, mice receiving mQPNAG26 rabbit sera had the lowest overall bacterial load in the kidneys of challenged mice (Fig.10D).

As PNAG is expressed in many types of bacteria, we explored the effects of immunization on gut microbiome. To analyze the composition of the gut microbiome, mice were fully immunized with mQPNAG26, and feces were collected on day 0 prior to immunization and day 35 following the initial prime immunization. The microbial species present in the droppings were analyzed via the 16S rRNA sequencing. Despite the significant amounts of anti-PNAG IgG produced in mouse sera, there were no significant changes in the microbial community present in the mouse gut (Supplementary Fig.8). Similar results were reported in a sponsored trial of the 5GlcNH2TT vaccine and shown in the study of anti-PNAG therapy in the setting of graft-versus-host disease41, or in human subjects in phase 1 clinical trials of both the 5GlcNH2TT vaccine or anti-PNAG mAb42,43. These observations corroborate that immunity to PNAG does not significantly alter the gut microbiome in immunized animals highlighting the potential safety of the vaccine.

In summary, numerous pathogens produce PNAG, rendering it a highly attractive target for vaccine development with the conjugate of fully deacetylated PNAG pentasaccharide with TT carrier currently undergoing human clinical trials as an anti-microbial vaccine. In order to enhance the potential protective efficacy, several aspects of the PNAG-based vaccine can be improved. As carbohydrates are typically T cell independent B cell antigens, an immunogenic carrier is critical. We have demonstrated that mQ is a powerful carrier. The mQPNAG conjugate was found to be superior in inducing higher levels of anti-PNAG IgG antibodies as compared to the corresponding PNAG conjugate with the TTHc carrier.

Besides the carrier, another important factor in vaccine design is the identification of the protective epitope(s) of the PNAG antigen, which was hampered by the lack of diverse structurally well-defined PNAG compounds. To gain a deeper understanding of the epitope specificity, a comprehensive library of PNAG pentasaccharides covering all possible combinations of free amine and NHAc has been synthesized. The synthesis is highlighted by a divergent design through the judicious choice of four amine protective groups, which can be orthogonally removed without affecting each other. The library of 32 PNAG pentasaccharides was obtained from just two strategically protected pentasaccharide intermediates, thus significantly enhancing the overall synthetic efficiency.

The availability of the comprehensive library provided an exciting opportunity to probe the epitope specificity through a glycan microarray. Screening of an anti-PNAG mAb F598 on the microarray suggests that the NHAc at unit B plays a critical role in F598 binding. NHAc at unit D could further enhance the binding. This knowledge led to the addition of two PNAG sequences (PNAG10 and PNAG26) beyond the fully deacetylated PNAG0 for vaccine studies.

The mQ conjugates with PNAG10 and PNAG26 were found to elicit IgG antibodies capable of inducing high levels of complement deposition and opsonic killing of bacteria compared to the mQPNAG0 conjugate. Vaccination with mQPNAG conjugate provided effective protection to mice against lethal challenges by S. aureus in both active and passive immunity models. Mice were also effectively protected from MRSA-induced death by the immune sera with significantly reduced bacterial load in the kidneys. The vaccines are biocompatible with no adverse side effects and do not significantly disturb the gut microbiome of the immunized mice. PNAG-based vaccine design guided by the well-defined synthetic library of PNAG is a powerful strategy to develop the next generation of vaccines and more effectively fight against pathogen infections including those by drug resistant strains.

Read more:

A comprehensive synthetic library of poly-N-acetyl glucosamines enabled vaccine against lethal challenges of ... - Nature.com