Semisynthetic Glycoconjugate Vaccine Candidates against Escherichia coli O25B Induce Functional IgG Antibodies in Mice.

Extraintestinal pathogenic Escherichia coli (ExPEC) is a major health concern due to emerging antibiotic resistance. Along with O1A, O2, and O6A, E. coli O25B is a major serotype within the ExPEC group, which expresses a unique O-antigen. Clinical studies with a glycoconjugate vaccine of the above-mentioned O-types revealed O25B as the least immunogenic component, inducing relatively weak IgG titers. To evaluate the immunological properties of semisynthetic glycoconjugate vaccine candidates against E. coli O25B, we here report the chemical synthesis of an initial set of five O25B glycan antigens differing in length, from one to three repeat units, and frameshifts of the repeat unit. The oligosaccharide antigens were conjugated to the carrier protein CRM197. The resulting semisynthetic glycoconjugates induced functional IgG antibodies in mice with opsonophagocytic activity against E. coli O25B. Three of the oligosaccharide-CRM197 conjugates elicited functional IgGs in the same order of magnitude as a conventional CRM197 glycoconjugate prepared with native O25B O-antigen and therefore represent promising vaccine candidates for further investigation. Binding studies with two monoclonal antibodies (mAbs) revealed nanomolar anti-O25B IgG responses with nanomolar K D values and with varying binding epitopes. The immunogenicity and mAb binding data now allow for the rational design of additional synthetic antigens for future preclinical studies, with expected further improvements in the functional antibody responses. Moreover, acetylation of a rhamnose residue was shown to be likely dispensable for immunogenicity, as a deacylated antigen was able to elicit strong functional IgG responses. Our findings strongly support the feasibility of a semisynthetic glycoconjugate vaccine against E. coli O25B.

Introduction

Escherichia coli is a member of the normal intestinal microbiota of mammals including humans. However, pathogenic strains exist that can cause potentially deadly infections even in healthy individuals.[1,2] Enteric/diarrheal disease, urinary tract infections (UTIs), sepsis, and meningitis are common clinical manifestations resulting from pathogenic E. coli infections.[3] Despite aggressive infection control measures, the rapid increase of hospital-acquired infections is a major concern.[4,5] The global prevalence of antibiotic resistance in pathogenic E. coli is particularly high, with 65% of isolates exhibiting resistance to three or more drug classes, resulting in treatment-unresponsive infections.[6,7] Hence, this bacterium has been identified as a critical priority pathogen by the WHO. The E. coli pathotypes responsible for extraintestinal infections are referred to as extraintestinal pathogenic E. coli (ExPEC), a group that includes serotype O25B.[8]

The E. coli O25B sequence type 131 (ST131) causes predominantly community-acquired infections, accounts for more than 10% of all ExPEC infections, and is the major cause of E. coli multidrug-resistant infections in the United States.[9−12] Serotype O25B differs from the related O25A serotype in its O-antigen biosynthetic gene cluster and polysaccharide structure.[13−15] The O- and K-antigens are well-described virulence factors that contribute to E. coli survival by interfering with complement-mediated killing.[16−20] These surface antigens of E. coli are highly exposed to the immune system, rendering them attractive targets as vaccine candidates.[21,22] For example, it was observed that levels of antibodies to surface antigens increased in patients with bacteremia,[23] a large subset of which were O-antigen-specific.[24,25] Antibodies against O- and K-antigens were shown to promote phagocytosis and confer protection in preclinical challenge models.[26] Although more than 180 different O-antigens are known for E. coli, which comprises serogroups O1 through O181, the majority of ExPEC infections are caused by serogroups O1A, O2, O6A, and O25B.[27−29] Within the O25 serogroup, up to 75% of isolates were shown to belong to the O25B serotype in a recent screening of clinical isolates from Switzerland.[30]

Glycoconjugate vaccines have proven to be an effective measure against invasive bacterial infections in the past ∼40 years. Covalent conjugation of bacterial polysaccharides to a carrier protein as an immunogen results in a T-cell-dependent response able to induce long-lasting protective immunity. Various glycoconjugate vaccines against Haemophilus influenzae type B, Meningococci, Pneumococci, and recently Salmonella Typhi have been licensed[31−33] and are based on native polysaccharides harvested from bacterial cultures.[34] Synthetic oligosaccharide–protein conjugates (semisynthetic glycoconjugates) have emerged as powerful alternatives[35−37] due to their well-defined chemical antigen structures and lower levels of residual impurities.[38,39] The first clinically approved synthetic glycoconjugate vaccine Quimi-Hib against H. influenzae type B developed in Cuba exhibited an excellent safety profile and 99.7% protective efficacy in children and is comprised of synthetic polyribosylribitol phosphate with an average length of eight repeating units.[40,41] There has been great progress more recently on semisynthetic glycoconjugate vaccines containing size-defined and homogeneous synthetic O-antigen-derived glycan antigens. A Shigella flexneri 2a vaccine (SF2a-TT15 or GlycoShig3) was demonstrated to be safe, well tolerated, and highly immunogenic in a phase I clinical trial[42,43] (ClinicalTrials.gov ID: NCT02797236) and is currently under investigation in a phase II study (ClinicalTrials.gov ID: NCT04602975). Several other fully synthetic glycoconjugate vaccines are in different phases of clinical trials against various drug-resistant bacteria.[44−46]

There is currently no vaccine against ExPEC. Multivalent glycoconjugate vaccines containing various E. coliO-polysaccharides were shown to be safe in humans, but antibody induction was variable and low, particularly for the O25 component.[24,47] A tetravalent glycoconjugate vaccine (ExPEC-4V) aimed against UTIs composed of isolated O-antigen from serotypes O1A, O2, O6A, and O25B conjugated to exotoxin A was proven to be safe but elicited only weak functional antibody response in clinical trials.[28,48−51] More recently, multivalent glycoconjugate vaccines (ExpPEC-9V and ExPEC-10V) are being investigated in clinical trials (ClinicalTrials.gov ID: NCT04899336 and NCT03819049). Despite the recent progress in preparing glycoconjugate vaccines against ExPEC, conventional industrial-scale production would still be challenging due to technical problems associated with harvesting polysaccharides from bacterial cultures.[47] To address this issue and the reported poor immunogenicity of isolated O25B, we here report the chemical synthesis of an initial set of five well-defined O25B oligosaccharide antigens that were shown to be immunogenic in mice when conjugated to the CRM197 carrier protein. Three of the semisynthetic vaccine candidates induced functional IgG antibodies in orders of magnitude comparable to a conventional glycoconjugate vaccine prepared with native isolated O25B O-antigen conjugated to CRM197via single-end conjugation. As expected,[52,53] the magnitude and functional activity of the IgG responses varied based on the number of repeating units and the frameshift. The in vivo immunogenicity results generated in this study allow for the rational design of optimized synthetic oligosaccharide antigens to be evaluated in future. Overall, our results demonstrate the feasibility and potential of a semisynthetic glycoconjugate vaccine approach against O25B and ExPEC infections.

Results and Discussion

The reported structure of the O25B O-antigen pentasaccharide repeating unit (RU) 1 consists of two glucose (Glc), two rhamnose (Rha, out of which one is acetylated at the 2nd position) and one N-acetylglucosamine (GlcNAc) residues (Figure ).[14,30] The immunogenicity of oligosaccharide antigens can potentially be influenced by the length (number of RUs) and by the frameshift of the RU as well.[52,54] To evaluate the impact of the antigen structure on immunogenicity, we first aimed to synthesize five possible oligosaccharide antigens, the natural pentasaccharide RU (1RU), a decasaccharide comprising two natural RUs (2RU), a pentadecasaccharide of three natural RUs (3RU), as well as two frameshifts of the pentasaccharide RU termed FS1 and FS2 (Figure ). Installment of a reducing end aminoethyl linker (C2-linker) will allow for efficient conjugation to the carrier protein.

Figure 1

Reported structure of the natural repeat unit (RU) of the E. coli O25B O-antigen and the designed synthetic oligosaccharide targets 1–5. Glc, glucose; GlcNAc, N-acetylglucosamine; Rha, rhamnose.

The syntheses of the above-listed oligosaccharides are particularly challenging, mainly due to the presence of a central d-glucose that is substituted by four other sugars (Figure ). Such highly branched oligosaccharides can substantially increase the hurdles of chemical synthesis.[55] Complexity of this core d-glucose is further increased by the requirement for orthogonal protecting groups, which could facilitate the α-linkage to the less nucleophilic 2-OAc-l-rhamnose (l-Rha-2-OAc). The C4 hydroxyl group of d-glucose is less nucleophilic, so ideally the first glycosylation should occur at this position. On the other hand, α-stereoselectivity of d-glucose is reported to be favorable by using a 4,6-benzylidene protection to carry out glycosylation with an l-Rha-2-OAc building block (BB). β-linkages are also required, which could be achieved with participating groups (typically esters) at the C2 position. However, the cleavage of conventional esters at the C2 of glucose in the presence of the acetate at the C2 position of l-rhamnose would be problematic. With such inherent complexity, identifying suitable orthogonal protecting groups for the C2–C6 hydroxyl groups of the core d-glucose was challenging. Consequently, we explored various BBs with several combinations of protecting groups to evaluate their coupling compatibility and deprotection strategies. The synthetic strategy suitable for one antigen may not be applicable for the synthesis of other antigens. This means the synthetic strategy must consider various possibilities to assemble the parent RU and then apply the same to higher numbers of RUs.

Synthetic Strategy

Our synthetic approach toward antigens 1–3 was based on the synthesis of key intermediates tetrasaccharide 8, pentasaccharide 9, and hexasaccharide 11 (Figure ). Similarly, for antigens 4 and 5, the key intermediate was trisaccharide 13. Given that tetrasaccharide 8, pentasaccharide 9, and hexasaccharide 11 needed the incorporation of tetrasubstituted glucose, the key was to make an orthogonally substituted disaccharide 6 that already incorporated the acetate required in the final antigens at the C2 position of l-rhamnose. The critical disaccharide 6 could be obtained from the glycosylation of an orthogonally protected glucose 17 containing a nonparticipating benzyl group at C2, a 4,6-O-napthylidene, and a 3-O-chloroacetyl (ClAc) group. The l-rhamnose contained the anomeric paramethoxyphenyl (MP), which was orthogonal to naphthyl (NAP) with acetate at C2, thereby allowing for selective cleavage to perform the respective glycosylations and deprotection sequences to obtain key intermediates 8, 9, and 11 (Figure a). A 2-O-azidomethyl benzoyl (AZMB) group was selected for the C2 position of the terminal glucose BB 14 as it can serve as a participating protecting group that can be cleaved under very mild conditions in the presence of other ester groups.[56] Coupling of intermediate 8 with 10 will yield antigen 1 (1), combining intermediates 8 and 11 will provide antigen 2 (2), whereas coupling of intermediates 8, 10, and 11 will give access to antigen 3 (3).

Figure 2

(a) Synthetic strategy for antigens 1–3. (b) Synthetic approach to assemble antigens 4 and 5. AZMB, 2-(azidomethyl)benzoyl; NAP, 2-napthylmethyl; MP, 4-methoxyphenyl; AcCl, chloroacetyl; Troc, 2,2,2-trichloroethoxycarbonyl; TBS, tert-butyldimethylsilyl.

Antigen 4 (4) and antigen 5 (5) could be obtained from trisaccharide 13, which, in turn, would be obtained from disaccharide 12. The sequential glycosylations of trisaccharide 13 with BBs 14 and 15 will provide antigen 4 (4). Similarly, antigen 5 (5) could be obtained by sequential glycosylations of C2-linker, donors 14 and 16(57) with trisaccharide 13 (Figure b).

Synthesis of Pentasaccharide (9)

In our strategy, disaccharide 6 is a crucial building block to convergently diversify our approach to access the targeted antigens. Our initial synthetic studies toward α-glycosylation between thio l-rhamnoside acceptor and 4,6-O-acetal-mediated corresponding d-glucose imidate or phosphate donors varying promoters and solvents led to predominantly aglycon transfer and resulted in the concomitant decomposition of the product. Although α-stereoselective glycosylation[58] between d-glucoside and l-rhamnoside is not unknown,[59−61] however, under the given set of protecting groups on donor and the presence of 2-OAc on acceptor proved extremely difficult. Attempts toward the glycosylation between donor 17 and various 2-O-orthogonally protected groups such as p-methoxybenzyl (PMB) and p-bromobenzyl (PBB) of l-rhamnoside were also unsuccessful. Efforts to improve the yield and selectivity did not turn out to be better by changing different promoters, temperatures, and solvents (Table S1 in the Supporting Information). Hence, the orthogonal glycosylation between donor 17 and acceptor 18(62) in the presence of tert-butyldimethylsilyl trifluorosulfonate (TBSOTf) gave disaccharide 7 (α/β = 6:1) as an inseparable mixture in 53% (Scheme ). Regioselective reductive ring opening of naphthylidene acetal of 6 using triethylsilane and triflic acid (TfOH) gave 4-OH acceptor 7 in 58% as only α-anomer after separation of the isomers. Donor 19(63) was glycosylated to 4-OH acceptor 7 using N-iodosuccinimide (NIS) and trimethylsilyl trifluorosulfonate (TMSOTf) surprisingly to give trisaccharide 20 as an inseparable mixture of anomers (α/β = 1:4) in 77%. Removal of chloroacetyl in 20 followed by coupling with l-rhamnosyl donor 22(64) gave tetrasaccharide 23 in a 74% yield. The anomers formed in the trisaccharide formation were able to separate at this stage, and 23 was isolated as a pure compound. Orthogonal cleavage of the naphthyl group in the presence of p-methoxyphenyl (MP) was achieved using 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) in MeOH/CH2Cl2 to obtain tetrasaccharide acceptor 24 in an 80% yield. Coupling of glucosyl donor 14 with acceptor 24 in the presence of TMSOTf provided pentasaccharide 9 in an 81% yield. MP cleavage using ceric ammonium nitrate (CAN) followed by the conversion of the corresponding hemiacetal to the imidate gave 25 in an 84% yield over two steps.

Scheme 1

Synthesis of Pentasaccharide 9

DABCO, 1,4-diazabicyclo[2.2.2]octane.

Alternative Synthesis of Pentasaccharide 9

Given that only a modest selectivity of α/β = 6:1 and lower yields were observed for the synthesis of disaccharide 6 and the surprising outcome of only α/β = 1:4 for a β-selective glycosylation to obtain trisaccharide 20 as an inseparable mixture (Scheme ), we explored different donors. One of the best reactions turned out to be the coupling between donor 26(65) and acceptor 18 using NIS/TMSOTf to obtain 27 in 71% with lower selectivity (α/β = 2:1) than obtained for disaccharide 6 as an inseparable mixture (Scheme ). The anomers were separated after the benzylidene cleavage to afford 4,6-diol 28 in 53%. The regioselective glycosylation of 6-OH in 28 was achieved using stoichiometric amounts of donor 14 to give a C6 glycosylated product 29 with a 70% yield. Interestingly, the coupling of trisaccharide 29 with donor 19 gave a clean reaction resulting only in β-isomer 30, thereby overcoming the limitation of selectivity and separation observed in Scheme . Cleavage of the NAP group followed by glycosylation with donor 22 gave pentasaccharide intermediate 9. The alternative strategy proved to be efficient with overall better yields and selectivities.

Scheme 2

Alternative Synthesis of Pentasaccharide 9

PTSA, p-tolylsulfonic acid.

Synthesis of Tetrasaccharide 8

The regioselective ring opening of disaccharide 6 proceeded smoothly using borane tetrahydrofuran (BH3·THF) and TMSOTf at room temperature, and the resulting 6-OH was then used in the next step for glycosylation with the glucose imidate 14 to provide trisaccharide 31 in good yield. Deprotection of the chloroacetyl group was performed using DABCO in the presence of pyridine and ethanol at an elevated temperature to obtain the trisaccharide acceptor 32 (Scheme ). Glycosylation with imidate 22 was purely α-selective and high yielding to provide tetrasaccharide 8. Finally, the OMP group was cleaved using CAN and the resulting lactol was converted to imidate 33.

Scheme 3

Synthesis of Tetrasaccharide 8

Synthesis of Antigen 1 (1)

The above-obtained tetrasaccharide imidate 33 was reacted with acceptor 10 at −30 °C to provide the fully protected pentasaccharide 34 in good yield (Scheme ). The deprotection sequence was performed by initially cleaving the AZMB group, followed by protecting group manipulation on nitrogen in GlcNAc to afford compound 35. Finally, the global deprotection of benzyl utilizing the slightly modified hydrogenation conditions provided antigen 1 (1).

Scheme 4

Synthesis of Antigen 1 (1)

Synthesis of Antigen 2 (2)

Glycosylation of acceptor 10 with pentasaccharide donor 25 employing TMSOTf as a promotor at −30 °C proceeded as expected; the α-linked isomer was formed in good yields, giving access to linker-equipped hexasaccharide 36 (Scheme ). Removal of the tert-butylsilyl (TBS) group at the nonreducing end using HF·pyridine in THF gave the hexasaccharide acceptor 37, which served as a common intermediate for the synthesis of antigens 2 (2) and 3 (3).

Scheme 5

Synthesis of Antigen 2 (2)

Reaction of 37 with tetrasaccharide donor 33 in the presence of TMSOTf at −30 °C gave the fully protected decasaccharide 38 as a single isomer. A subsequent four-step deprotection sequence including AZMB removal under Staudinger conditions leaving the acetates intact, reductive Troc-removal with zinc followed by acetylation of the free amine and overall hydrogenolysis of benzyl ethers gave the desired antigen 2 (2) in a 11% yield over four steps.

Synthesis of Antigen 3 (3)

Assembly of pentadecasaccharide 41 was achieved using a 6+5+4 approach (Scheme ). Glycosylation of hexasaccharide acceptor 37 with donor 25 at −30 °C promoted by TMSOTf gave undecasaccharide 39 in good yield. Following selective TBS removal with HF·pyridine gave 40 in high yield. The acceptor 40 was then smoothly coupled with 33, yielding fully protected pentadecasaccharide 41 in a 78% yield. The protecting groups could be removed in a four-step procedure with a 21% overall yield to obtain antigen 3 (3).

Scheme 6

Synthesis of Antigen 3 (3)

Synthesis of Antigen 4 (4)

The fully protected antigen 4 (4) could be accessed by a direct glycosylation of the C2-linker with donor 25; late-stage linker coupling, however, resulted in an inseparable α/β mixture. Hence, a different synthetic 4+1 approach was applied (Scheme ). Key intermediate 47 was assembled by consecutive glycosylations on the central glucose BB 17. Regioselective reductive ring opening of naphthylidene acetal of 17 using triethylsilane and TfOH gave 4-OH acceptor 17a in an 80% yield. The free hydroxy group was subsequently glycosylated using donor 42(66) giving disaccharide 43 in decent yields as an α/β mixture, which could be separated after the removal of the chloroacetyl group. Next, the glucose unit was glycosylated stereoselectively at the C3 position with imidate 22, followed by NAP cleavage at the C6 position giving 46. The last glycosylation on the central glucose with BB 14 gave an excellent yield of tetrasaccharide 47. Compound 47 served as a thioglycoside donor in the glycosylation with linker-equipped rhamnose 15 using NIS/TMSOTf at 0 °C. As expected, pentasaccharide was only obtained in moderate yields as an α/β mixture (7:2). However, the desired α-isomer 48 could be isolated after AZMB removal using tributyl phosphine. Further three-step modifications of 48 gave semiprotected intermediate, azide-functionalized 49 in decent yields. The final hydrogenolytic cleavage of the benzyl ethers gave antigen 4 (4) in a 73% yield.

Scheme 7

Synthesis of Antigen 4 (4)

Synthesis of Antigen 5 (5)

The synthesis of pentasaccharide FS2 antigen 5 (5) was achieved from the common intermediate 45. The α-selective glycosylation with C2-linker was obtained in good yield but with only 3:1 selectivity in favor of desired product 50 after TBS deprotection (Scheme ). Glycosylation with acetyl rhamnose building block 16 proceeded smoothly with very good yields. The C6-NAP on the glucose was deprotected using standard conditions, and the C6 glycosylation with 14 afforded the fully protected FS2 (52). Deprotection of the AZMB group was carried out first, followed by protecting group manipulation on the amino group. Finally, the chloride on the linker was converted to azide using sodium azide in DMF; the global deprotection of benzyl groups was carried out using standard palladium conditions. The crude FS2 antigen 5 (5) was found to be stable, but over a period of time, the C-2-O-acetate group on the rhamnose[60,67] started to migrate,[68] which was confirmed by NMR indicating a mixture of compounds. So finally, under alkaline conditions, the acetate group was deprotected and the FS2 (-OAc) (53) was obtained.

Scheme 8

Synthesis of Antigen 5 (5)

Preparation and Characterization of Oligosaccharide–CRM197 Conjugates

We next sought to evaluate the potential of oligosaccharides 1–4 and 53 (the nonacetylated version of 5) as vaccine candidates against infections with E. coli O25B. It is established that conjugation of the otherwise poorly immunogenic oligo- or polysaccharides to carrier proteins helps to induce T-cell-dependent glycan-specific immune responses, including immunoglobulin class switch.[69,70] As a carrier protein, we selected CRM197, a detoxified mutant of diphtheria toxin that is used in various marketed conjugate vaccines produced with isolated polysaccharides (such as the pneumococcal vaccine Prevnar). To prepare the conjugates, the primary amine group of antigens 1–5 (compounds 1–4 and 53) was reacted with di-N-succinimidyl adipate in DMSO in the presence of triethylamine (Et3N). The resulting monoester was precipitated, washed, and reacted with primary amines of lysine residues of the CRM197 protein (Figure S1a in the Supporting Information).

The resulting conjugates 1-CRM197, 2-CRM197, 3-CRM197, 4-CRM197, and 53-CRM197 had endotoxin contents lower than 0.01 endotoxin units per microgram protein (EU/μg) and were thus considered safe for in vivo immunization experiments (see Experimental section). Successful conjugation was verified by denaturing SDS-PAGE, revealing shifts of the conjugates toward higher masses compared to the unconjugated carrier protein and no indications for unreacted CRM197 (Figure S1b in the Supporting Information). This finding was confirmed by HPLC-SEC analysis in which elution times of the conjugates were consistently lower than that of CRM197 (Figure S1c in the Supporting Information). To estimate the antigen to carrier protein molar ratios (antigen loading), mass increases of the conjugates compared to CRM197 were determined by MALDI-TOF MS (Figure S1d in the Supporting Information). This analysis yielded average antigen loadings of 10.3 (1-CRM197), 12.6 (2-CRM197), 13.8 (3-CRM197), 10 (4-CRM197), and 12.6 (53-CRM197). These antigen loadings were used to calculate the glycan concentrations of the conjugates.

Semisynthetic Conjugates Are Immunogenic

Pilot studies with native O25B O-antigen CRM197 conjugates showed that female CD-1 mice generated more robust and homogeneous O25B IgG responses than female C57BL/6, BALB/c, or C3H mice (data not shown). The optimized vaccination schedule for unadjuvanted serotype O25B conjugates, determined empirically from previous mouse immunogenicity studies, is shown in Figure a. Mice received three doses of the vaccine candidates at weeks 0, 5, and 13. Blood was collected for serum antibody analyses 2 weeks after the second immunization (week 7; postdose 2 [PD2]) and 1 week after the third immunization (week 14; postdose 3 [PD3]). To evaluate the immunogenicity of the semisynthetic conjugates, 1-CRM197 through 4-CRM197 and 53-CRM197, 25 mice per group were subcutaneously vaccinated each with an amount of conjugate corresponding to 2 μg of glycan antigen. This dose level was selected based on the results from previous vaccination studies (data not shown). In addition, we included one group with a sun-type (single-end conjugation) CRM197 conjugate prepared with the isolated native O25B O-antigen including the core oligosaccharide with an average polysaccharide chain length of 14 kDa, corresponding to 15 repeat units. Two negative control groups, one immunized with unconjugated native O25B O-antigen and one sham-immunized with phosphate-buffered saline (PBS), were also included. We did not use an adjuvant such as AlPO4, which is commonly used for conjugate vaccines, as it may have reduced the ability to discriminate the antibody responses among different conjugate vaccines, as observed in previous studies (data not shown).

Figure 3

O25B-specific IgG responses to semisynthetic and native CRM197-conjugates. (a) Immunization schedule. PD2, postdose 2; PD3, postdose 3. (b) and (c) IgG levels, as measured by the Luminex IgG assay, at week 7 (PD2) (b) and week 14 (PD3) (c) time points. Percentages of mice with IgG levels 5-fold greater than the baseline lower limit of quantitation (LLOQ) are indicated. 1–3RU, 1–3 repeat units; FS1, frameshift 1; FS2, frameshift 2. Statistically significant differences to the unconjugated free O25B free native polysaccharide comparator are marked with asterisks (one-way ANOVA on log-transformed data; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant). IgG level GMTs (μg/mL) and 95% confidence interval (CI) values are shown in Table S2 in the Supporting Information. The horizontal lines are the geometric mean titers (GMTs), and the whiskers indicate the 95% confidence intervals.

Serum levels of isolated O25B antigen-specific IgG elicited by the five semisynthetic CRM197 conjugates after two (PD2) and three (PD3) vaccination doses were determined by a Luminex IgG assay (Figure b,c). IgG levels were compared to those induced by the CRM197 conjugate prepared with native O25B O-antigen. Three doses of the native O25B CRM197 conjugate were required to generate robust IgG titers, with a geometric mean titer (GMT) of 22.0 μg/mL IgG and an 84% responder rate. The semisynthetic conjugates were all substantially more immunogenic than the unconjugated free O25B O-antigen, with responses to the three repeat unit (3RU) trimers (3-CRM197) trending higher than the 2RU (2-CRM197) or 1RU constructs including the FS1 and FS2 frameshifts (1-CRM197, 4-CRM197, and 53-CRM197, respectively) at either time point. The best-performing semisynthetic conjugate 3RU (3-CRM197) was not significantly less immunogenic than the native O25B conjugate, although by PD3 it yielded a broader range of IgG titers among individual mice and fewer responders (68 vs 84%). Among the semisynthetic conjugates at the PD3 time point, the 3RU variant (3-CRM197) elicited the highest levels of O25B O-antigen-specific IgG (4.8 μg/mL). The single repeat unit frameshift FS2 (53-CRM197) elicited comparable levels of IgG as the 2RU conjugate 2-CRM197 (2.5 vs 2.3 μg IgG/mL), with responses to the FS1 (4-CRM197) and 1RU (1-CRM197) variants trending lower (1.4 and 0.3 μg IgG/mL).

Knowing that the semisynthetic conjugates were immunogenic, we next assessed whether the generic spacer moiety (composed of ethyl and adipoyl groups) connecting the synthetic oligosaccharides to the CRM197 protein was immunogenic. It was shown previously that strong spacer-specific antibody responses can suppress antibody responses to weakly immunogenic oligosaccharide antigens, which may explain the relatively poor immunogenicity of the 1RU (1) and FS1 (4) antigens.[71,72] To this end, we prepared conjugates with the bovine serum albumin (BSA) protein and five different monosaccharides with C2, C5, or polyethylene glycol linkers (see Figure S2 in the Supporting Information). The resulting BSA conjugates prepared with the C2-linked monosaccharides contain a generic spacer moiety that is identical to that of the semisynthetic CRM197 conjugates used for vaccination experiments. An in-house generated pool of sera from vaccinated BALB/c mice that showed a strong reactivity against the spacer moiety (alkyl–adipoyl) was used as a positive control. Using these BSA conjugates, we performed an ELISA analysis of pooled antisera from the PD3 time point. IgG signals were either below the level of detection or very weak and only seen at high antiserum concentrations, indicating that the generic spacer moiety of the semisynthetic CRM197 conjugates used in this study was not significantly immunogenic.

Antibodies Elicited by Semisynthetic Conjugates Promote Opsonophagocytosis

Functional immunogenicity of sera from vaccinated mice was assessed initially at the intermediate PD2 time point using a multidrug-resistant unencapsulated O25B:K- strain via an opsonophagocytosis assay (OPA). Individual OPA responses are shown in Figure a. As both IgG and IgM are capable of activating complement, OPA activity reflects the composite antigen-specific functional activities contributed by both antibody isotypes. This likely explains why the functional responses in the O25B:K- strain OPA at PD2 for the single repeat unit constructs [1RU (1-CRM197), FS1 (4-CRM197), and FS2 (53-CRM197)] exceed what might be expected based on their correspondingly weak IgG titers compared with the other conjugate variants (Figure ). Indeed, functional responses (but not the corresponding IgG binding titers) to the three distinct single repeat unit conjugates were significantly higher than the activity of the unconjugated O25B free O-antigen control group, which, in turn, was only marginally above the baseline level observed for unvaccinated control animals (titer of 125 vs 50 for unvaccinated controls). The highest OPA responder rates were observed with the synthetic 2RU conjugate 2-CRM197 (72%) and native conjugate (80%), although GMT titers of these two groups were not significantly higher than any of the other CRM197 conjugate variants.

Figure 4

Results of opsonophagocytosis assays (OPAs) that measure functional antibody responses to CRM197 conjugates with unencapsulated (O25B:K-) strain PFEEC0068 and encapsulated (O25B:K5) strain PFEEC0066. (a) Postdose 2 (PD2) OPA titers generated with the O25B:K- strain. (b) PD2 OPA titers generated with the O25B:K5 strain. (c) Postdose 3 (PD3) OPA titers generated with the O25B:K5 strain. Percentages of mice with OPA titers 4-fold greater than the prevaccination baseline GMT titer of all pooled groups are indicated (value of 50). LOD, limit of detection; 1–3RU, 1–3× repeat units; FS1, frameshift 1; FS2, frameshift 2. Statistical significance relative to the unconjugated free polysaccharide comparator was assessed by one-way ANOVA on log-transformed data: *p < 0.05: **p < 0.01; ***p = 0.001; ****p < 0.0001; ns, not significant. OPA titer GMTs and 95% CI values are shown in Table S3 in the Supporting Information. The horizontal lines indicate geometric mean titers (GMTs), and whiskers show the 95% confidence intervals.

We observed in previous experiments that the unencapsulated O25B:K- E. coli OPA strain is highly sensitive to killing by antigen-specific antibodies. This was also reflected in this study by the nearly 5-fold log10 range in PD2 bactericidal titers observed across the groups vaccinated with the various O25B CRM197 conjugates (Figure a). Thus, we developed a second alternative OPA assay with a strain expressing the K5 (heparosan) capsular polysaccharide, which is highly virulent in mouse lethal challenge models[29] and less susceptible to OPA killing by O-antigen-specific antibodies in vitro. The reduced sensitivity of this strain may be due, in part, to the partial masking of the O25B O-antigen by the K5 capsule, as demonstrated in flow cytometry experiments with bacteria treated to selectively remove the K5 capsule via 60 °C heat treatment or by heparinase digestion.[29] Sera from all groups of mice at PD2 and PD3 time points were evaluated in OPAs with this O25B:K5 strain, and results are shown in Figure b,c. Use of this strain has the practical advantage of potentially providing evidence of functional differences observed among conjugate variants by using a phenotypically distinct clinically relevant isolate as readout.

As anticipated, the OPA geometric mean titers observed with the O25B:K5 strain compared with the O25B:K- strain at the equivalent PD2 time point were substantially (about 50- to 100-fold) lower; overall rates of vaccine responders for all of the CRM197 conjugates were also reduced. However, by the PD3 time point, response rates for the conjugates ranged from 38 to 88%, with significant differences between group GMTs apparent relative to the unconjugated free native O25B O-antigen control. The most immunogenic conjugates were the semisynthetic 2RU (2-CRM197) and 3RU (3-CRM197) constructs and the native O25B O-antigen CRM197 conjugate, with responder rates of 71, 68, and 88%, respectively. The next most immunogenic conjugate was the single repeat unit FS2 (53-CRM197), followed by FS1 (4-CRM197). Of the three single repeat unit constructs evaluated at the PD3 time point, the FS2 (53-CRM197) was the most immunogenic with respect to levels of IgG and OPA activities, although differences were only significant for the FS2 (53-CRM197) vs 1RU (1-CRM197) comparison of IgG titers (p = 0.0011, one-way ANOVA on log-transformed IgG data from FS2 (53-CRM197) vs FS1 (4-CRM197) and 1RU (1-CRM197) groups; Table S2 and Table S3 in the Supporting Information).

Functional Antibodies Recognize Different O25B Epitopes

Synthetic oligosaccharide antigens are useful not only as vaccine antigens but also to gain information on the binding epitope of antiglycan antibodies.[73,74] To elucidate epitope recognition patterns, we subjected three O25B-specific monoclonal antibodies (mAbs), designated as 3E9-11,[75] 71-8, and 80-11,[29] to binding studies using surface plasmon resonance (SPR). mAb 3E9-11 was elicited in mice by whole bacteria, the other two mAbs by an O25B long-chain lattice conjugate.[29] The three mAbs exert comparable functional activities, as measured by OPA activity (Table ). To this end, the mAbs were passed over SPR surfaces functionalized with the various CRM197 conjugates to obtain binding kinetics (Table ). mAb 3E9-11 recognized the 2RU (2), 3RU (3), and the natural antigen (15RU) but did not bind to the 1RU (1), FS1 (4), and FS2 (53) antigens, indicating that this mAb recognizes an epitope that is larger than a pentasaccharide. KD values increased along with the glycan chain length, largely due to slower dissociation constants (koff). mAb 71-8 showed a different binding pattern, recognizing all antigens except for 1RU (1). Its binding epitope is therefore likely smaller than a pentasaccharide. In contrast to mAb 3E9-11, lower KD values for the larger antigens were not observed, as koff values for the 2RU (2), 3RU (3), and 15RU antigens were similar. KD values for FS1 (4) and FS2 (53) were lower than those for 2RU (2) and 3RU (3), mainly due to increased association rates (kon). mAb 80-11 recognized all tested antigens with comparable KD values, suggesting that mAb 80-11 has the smallest binding epitope of the three tested mAbs.

Table 1

SPR-Derived Binding Kinetics of Three Anti-O25B mAbs, 3E9-11, 71-8, and 80-11a

immobilized antigen	KD [nM]	kon [M–1 s–1 × 105]	koff [s–1 × 10–3]
	mAb 3E9-11 (OPA EC50: 31 ng/mL)
1RU-CRM197 (1-CRM197)	ND	ND	ND
2RU-CRM197 (2-CRM197)	752.1 ± 22.7	1.38 ± 0.68	140 ± 72
3RU-CRM197 (3-CRM197)	295.4 ± 79	1.78 ± 0.37	44 ± 2.4
15RU-CRM197	3.6 ± 0.85	23.44 ± 8.1	8.7 ± 3.7
FS1-CRM197 (4-CRM197)	ND	ND	ND
FS2-CRM197 (53-CRM197)	ND	ND	ND
	mAb 71-8 (OPA EC50: 22 ng/mL)
1RU-CRM197 (1-CRM197)	ND	ND	ND
2RU-CRM197 (2-CRM197)	387.4 ± 153.4	0.66 ± 0.39	31.3 ± 20.8
3RU-CRM197 (3-CRM197)	385.3 ± 39.2	0.98 ± 0.68	36.5 ± 21.2
15RU-CRM197	41.8 ± 19.5	5.42 ± 0.76	36.2 ± 31.1
FS1-CRM197 (4-CRM197)	67.1 ± 8.2	10.2 ± 2.7	67.9 ± 18.7
FS2-CRM197 (53-CRM197)	166.5 ± 53.3	5.31 ± 3	73.4 ± 26.1
	mAb 80-11 (OPA EC50: 17 ng/mL)
1RU-CRM197 (1-CRM197)	94.2 ± 38	0.1 ± 0.03	0.7 ± 0.1
2RU-CRM197 (2-CRM197)	41 ± 0.7	0.3 ± 0	1.4 ± 0
3RU-CRM197 (3-CRM197)	29.1 ± 4.2	0.4 ± 0.1	1.2 ± 0
15RU-CRM197	35.3 ± 2.6	0.4 ± 0	1.3 ± 0
FS1-CRM197 (4-CRM197)	30.5 ± 7.7	1.3 ± 0.5	3.5 ± 0.4
FS2-CRM197 (53-CRM197)	20.7 ± 0.2	1.2 ± 0.1	2.4 ± 0.1

Rate constants for the antibodies are the averages of 2–4 independent experiments with standard deviations. Functional (EC50) potencies in OPA with strain PFEEC0068 are indicated. KD, equilibrium dissociation constants; kon, association rate constants; koff, dissociation rate constants; ND, no detectable binding. Representative sensorgrams are shown in Figure S3 in the Supporting Information.

Conclusions

A vaccine against ExPEC, which includes the important member E. coli O25B, is urgently needed. Previous experimental vaccination approaches included glycoconjugates with natural O-antigens attached to a carrier protein[30,76] or outer membrane vesicles.[77] While these approaches can elicit protective immunity, chemically defined antigens may help to improve immunogenicity and safety. Moreover, the isolated O25B polysaccharide was shown to be only weakly immunogenic, suggesting that the isolated O-antigen may not be suitable for a human vaccine.[50] In the present study, we aimed to overcome these limitations by evaluating vaccine candidates based on highly pure and defined glycan antigens obtained by chemical synthesis. An initial set of five oligosaccharide antigens from E. coli O25B were successfully synthesized, three pentasaccharides [1RU (1), FS1 (4), and FS2 (53)], one decasaccharide (2RU; 2), and one pentadecasaccharide (3RU; 3). Using these oligosaccharides, we successfully generated five semisynthetic vaccine candidates via conjugation to the CRM197 carrier protein. Results from the immunogenicity studies in CD-1 mice demonstrated that three of the five semisynthetic O25B O-antigen CRM197 conjugates elicited levels of antigen-specific IgG or functional OPA activity that were not significantly different from the comparator native O25B O-antigen CRM197 conjugate. Of note, the native polysaccharide, in contrast to the synthetic antigens, contained the core oligosaccharide. LPS cores are known to be able to elicit an immune response.[78−81] However, in the present study, the core oligosaccharide likely did not contribute to functional immunogenicity as antibody binding epitopes may be poorly exposed on the surface of clinical strains expressing O-antigens, as confirmed by flow cytometry experiments with a core-specific mAb (Figure S4 in the Supporting Information). This is in agreement with a previous study showing that antibodies against the core oligosaccharide cannot penetrate the O-antigen and capsular polysaccharide layers of Gram-negative bacteria.[80,82,83] There was a trend of increased immunogenicity from the larger oligosaccharides, specifically as measured in OPA. The most immunogenic of the semisynthetic conjugates were the 2RU (2-CRM197), 3RU (3-CRM197), and single repeat unit FS2 (53-CRM197) conjugates. In comparison, the 1RU (1-CRM197) and FS1 (4-CRM197) variants trended toward lower IgG responses and generated lower OPA activity at PD2 and PD3 time points with both serotype O25B OPA reporter strains, without or with a capsule. For the semisynthetic constructs, higher repeat numbers of 2RU (2-CRM197) and 3RU (3-CRM197) were associated with higher IgG titers at PD2 and PD3, which translated into improved functional OPA titers by PD3 compared with single repeat unit constructs 1RU (1-CRM197), FS1 (4-CRM197), and FS2 (53-CRM197). Similar observations were made previously for a Salmonella O-antigen for which two or three RUs resulted in better protection from challenge compared to one RU.[84] The immunogenicity data and the binding studies of the three monoclonal antibodies indicate that an antigen of FS1 or FS2 with two or three repeat units has great potential to yield an even superior IgG response. This warrants further immunization experiments with additional semisynthetic vaccine candidates. While glycan-specific IgGs were induced by all vaccine candidates, we did not detect significant amounts of linker-specific antibodies. Thus, the conjugation strategy proved suitable for the O25B glycan antigens. Kinetic binding studies with three functional mAbs revealed that the immune system is able to mount nanomolar KD antibody responses to O25B (Table ), in contrast to the prevailing notion that antiglycan antibodies are typically weak binders.[85,86] However, it has been demonstrated previously that glycoconjugate vaccines are able to elicit nanomolar-affine antibodies.[74,87,88] The epitopes recognized by the mAbs, despite comparable functional activity, differed in that 3E9-11 required two repeating units to bind (suggesting an epitope larger than a pentasaccharide), whereas 71-8 also bound to the two pentasaccharides FS1 (4) and FS2 (53) and 80-11 bound to all tested antigens (suggesting epitopes smaller than a pentasaccharide). Of note, 71-8 showed lower KD values for FS1 antigen (4) than for all other synthetic antigens. Thus, both the size (number of repeating units) and frameshift can influence the binding strength of anti-O25B antibodies. The immunological importance of the frameshift is also reflected in the superior immunogenicity of the FS1 (4) and FS2 (53) antigens compared to that of 1RU (1) (Figure ). Pure antigens with defined frameshift and nonreducing end terminal monosaccharide can be procured by chemical synthesis but not through the isolation of heterogeneous polysaccharides. An open question that remains is the impact of acetyl groups on the immunogenicity of the synthetic antigens. Previous studies have found that O-acetylation had an impact on the immunogenicity of isolated meningococcal and Salmonella typhi vaccines, whereas in contrast for others such as the semisynthetic Shigella vaccines, no effect of acetylation was observed.[89−93] We found that both acetylated (1 through 4) and the nonacetylated antigen 53 elicited IgG responses with functional activity against E. coli, suggesting that for this particular antigen, acetyl groups may have a limited effect on immunogenicity. Overall, the binding studies with the mAbs suggest that an O25B glycoconjugate vaccine elicits IgGs characterized by heterogeneous epitope recognition patterns. Moreover, antibodies elicited in mice by the 2RU (2) and 3RU (3) antigens overall bound more efficiently to native O-antigen and showed higher in vitro activity against E. coli than antibodies raised with 1RU (1) (Figures and 4). This supports that the larger antigens provide more diverse IgG responses due to the availability of additional epitopes. Therefore, reverse engineering approaches to identify glycan epitopes for vaccine development that rely on a single mAb[52] may miss important protective epitopes. In summary, our study demonstrates the potential of semisynthetic glycoconjugate vaccines against E. coli O25B, which is an important, clinically relevant member of the ExPEC group.

Experimental Section

O-Antigen Purification and Conjugation to CRM197

The fermentation broth was treated with acetic acid to a final concentration of 1–2% (final pH of 4.1). The extraction of O-antigen and delipidation were achieved by heating the acid-treated broth to 100 °C for 2 h. After acid hydrolysis, the batch was cooled to an ambient temperature and 14% NH4OH was added to a final pH of 6.1. The neutralized broth was centrifuged, and the centrate was collected. To the centrate was added 5 M CaCl2 stock solution in sodium phosphate to the final CaCl2 concentration of 0.1–0.2 M, and the resulting slurry was incubated for 30 min at room temperature. The solids were removed by centrifugation, and the centrate was concentrated 10- to 12-fold using a 10 kDa MWCO Sartocon Hydrostart membrane from Sartorius, followed by two diafiltrations against 20 mM citrate pH 6.0 and water, respectively, with 10 diavolumes for each diafiltration. The retentate, which contained the O-antigen, was then purified using a carbon filter (3 M CUNO R32SP). The carbon filtrate was diluted 1:1 (v/v) with 4.0 M ammonium sulfate. The final ammonium sulfate concentration was 2 M. The ammonium sulfate-treated carbon filtrate was further purified using a Sartobind Phenyl membrane (Sartorius) at a loading capacity of 55 mg of O-antigen per mL of membrane volume with 2 M ammonium sulfate as the running buffer. The O-antigen was collected in the flow through. The HIC filtrate was concentrated to the desired concentration level and then buffer-exchanged against water (20 diavolumes) using a 2 kDa MWCO Sartocon Hydrostart membrane (Sartorius). For single-end directional coupling, the O-antigen went through activation/reduction via a linker, dithiopropionic dihydrazide, to insert an amine thiol group at the KDO carbonyl present at the polysaccharide reducing end. The activated O-antigen polysaccharide was then conjugated to Bromo-CRM197.

Preparation of CRM197 Glycoconjugates and Quantification of Protein and Endotoxin

The oligosaccharide antigens (1, 2 mg; 2, 3.4 mg; 3, 4.8 mg; 4, 1.9 mg; 53, 3 mg) were dissolved in DMSO, and 35 equiv of Et3N was added. In a separate tube, 20 equiv of di-N-succinimidyl adipate was dissolved in DMSO and then added to the antigen solution. The reaction mixture was stirred for 2 h at room temperature. The resulting monoester was precipitated with EtOAc, centrifuged, and washed with EtOAc. The remaining EtOAc was removed by lyophilization to obtain a white solid. The solid monoester was dissolved in 100 mM sodium phosphate buffer, pH 7, and about 30 equiv was added to the CRM197 protein (Pfizer) dissolved in the same buffer. The reaction mixture was stirred at room temperature for 21 h. Then, the reaction mixture was washed twice with 100 mM sodium phosphate buffer, pH 7, and three times with phosphate-buffered saline (PBS), pH 7.4, using Amicon Ultra-4 centrifugal filters with 10 kDa MWCO (Merck Millipore, cat. no. UFC801096). Finally, the conjugates were sterile-filtered using Ultrafree-CL 0.22 μm centrifugal filters (Merck Millipore, cat. no. UFC40GV0S). The protein concentrations were determined using the QuantiPro BCA Assay Kit (Sigma-Aldrich, cat. no. QPBCA-1KT) following the manufacturer’s protocol. Endotoxin concentrations were determined using the LAL Chromogenic Endotoxin Quantitation Kit (Fisher Scientific, cat. no. 12117850) according to the manufacturer’s recommendations. This yielded the following values for endotoxin units per microgram protein (EU/μg): 1-CRM197, 0.006 EU/μg; 2-CRM197, 0.005 EU/μg; 3-CRM197, 0.003 EU/μg; 4-CRM197, 0.007 EU/μg; and 53-CRM197, 0.008 EU/μg. The CRM197 glycoconjugate of natural O-antigen (single-end conjugation) was prepared as previously described.[29]

SDS-PAGE

Protein samples were diluted in Laemmli loading buffer (VWR Life Science, cat. no. M337–25ML) and heated at 95 °C for 5 min. An amount of 2.5 μg protein per well was loaded per lane of a 10% Mini-PROTEAN TGX precast protein gel (Bio-Rad, cat. no. 4561033). ROTI Mark TRICOLOR XTRA (cat. no. 2244.1) was used as a protein size marker (5 μL per lane). The gel was run for approx. 45 min at 200 V in tris–glycine buffer with 0.1% (w/v) sodium dodecyl sulfate (SDS), stained with GelCode Blue Safe Protein Stain (Thermo Fisher Scientific, cat. no. 1860957), and destained with deionized water.

Size Exclusion Chromatography (SEC)

CRM197 conjugates were analyzed by SEC to quantify mass differences to the unconjugated CRM197 protein. The samples were diluted in 50 mM tris, 20 mM NaCl, pH 7.2, and run on a Thermo Scientific Dionex UltiMate 3000 RS HPLC system fitted with a Tosoh TSK G2000 column (SWxl, 7.8 mm × 30 cm, 5 μm) and a Tosoh TSK gel Guard column (SWxl 6.0 mm × 4 cm, 7 μm). The flow rate was 1 mL per min.

MALDI-TOF MS

Mass spectra were acquired with an Autoflex Speed MALDI-TOF system (Bruker Daltonics). Samples were spotted using the dried droplet technique with sinapic acid as a matrix on MTP 384 ground steel target plates (Bruker Daltonics). The mass spectrometer was operated in a linear positive mode. Mass spectra were acquired over an m/z range from 30,000 to 210,000, and data were analyzed with FlexAnalysis software provided with the instrument.

Mouse Immunization Studies

Female CD-1 mice (6–8 weeks old upon arrival) obtained from Charles River Lab were dosed at 7–9 weeks old. Vaccination was administered by the subcutaneous route with mice given either 2 μg of conjugate by glycan weight or PBS as a control at weeks 0, 5, and 13. Animals were bled (submandibular) at weeks 0 and 7 and exsanguinated at week 14. Endotoxin level requirements stipulated that all antigens were below 0.1 EU per dose per animal (below 0.05 EU per μg). All semisynthetic and native O25B conjugates met this criterion. Eight groups of 25 female CD-1 mice were vaccinated with semisynthetic conjugates, a native polysaccharide CRM197 conjugate, native O25B O-antigen, or PBS control. Based on the power analysis of OPA and IgG titers at postdose 3 from a previous O25B immunogenicity study (unpublished), it was determined that 25 mice per group would allow for the detection of up to a 7-fold significant difference between groups (two-tailed test, α = 0.05, power > 0.8). Animal studies were conducted according to the Pfizer local and global Institutional Animal Care and Use Committee (IACUC) guidelines at an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC).

O25B O-Antigen-Specific Direct Binding Luminex IgG Assay

Native O25B O-antigen was produced in-house by early bioprocess development (Pfizer). The native O-antigen differs from the synthetic oligosaccharides by the presence of terminal inner and outer R1 core oligosaccharides that remain attached to the O-antigen following cleavage from LPS by acetic acid during the bioprocessing of E. coli strains. These core oligosaccharides are not expected to contribute to antigen functional immunogenicity, as antibody binding epitopes are not exposed on the surface of clinical strains expressing O-antigens (Figure S4 in the Supporting Information). The native polysaccharide was covalently conjugated to poly-l-lysine with CDAP, and the derived poly-l-lysine conjugate was covalently coupled to Luminex magnetic carboxy bead microspheres with EDC/NHS. Beads were incubated with serially diluted individual mouse sera or control mAbs with shaking at 4 °C for 18 h. After washing, bound serotype-specific IgG was detected with a PE-conjugated goat antimouse total IgG mouse secondary antibody (90 min rt incubation). Microplates were read on a FlexMap 3D instrument (Bio-Rad). A serotype-specific mouse IgG mAb (Eco 80-11, generated in-house) was used as an internal standard to quantify IgG levels. A standard curve plot for the mAb titration yielded a linear slope profile across a 103-range of serum dilutions (log luminescence vs log serum dilution). A lower limit of quantitation (LLOQ) of 0.153 μg per mL was calculated from standard curve bias across multiple experiments.

O25B O-Antigen-Specific Opsonophagocytic Assay (OPA)

Clinical E. coli invasive blood isolates were obtained from the Pfizer-sponsored Antimicrobial Testing Leadership and Surveillance (ATLAS) collection, which is maintained by the International Health Management Associates (IHMA) clinical lab. Strains were genotypically characterized by whole genome sequencing (WGS) using the Illumina Miseq platform, including in silico serotyping for the prediction of O-antigen and K-capsule types, and multilocus sequence type (MLST). Expression of the O25B O-antigen on the bacterial surface was confirmed by flow cytometry using rabbit antisera to the RAC/DMSO long-chain lattice conjugate as the primary detection antibody. Growth of strains in LB media resulted in high levels of K-capsule expression by encapsulated strains that masked the detection of underlying O-antigens. Presence of group II K-capsule was diagnosed by the increased detection of O25B O-antigen following 60 °C heat treatment. The presence of K5 heparosan capsule was inferred from an increase in O-antigen detection following heparinase treatment. Two serotype O25B ST131 strains were selected for OPA development: PFEEC0068, an unencapsulated isolate resistant to carbapenem and fluoroquinolone antibiotics; and PFEEC0066, an encapsulated O25B:K5 isolate resistant to penicillin and fluoroquinolone antibiotics that is hypervirulent in mouse lethal challenge models. Prefrozen bacterial stocks were prepared by growing bacteria in DMEM media to an OD600 of between 0.5 and 1.0, and glycerol was added to a final concentration of 20% prior to freezing. For the unencapsulated O25B:K- strain, pretitered thawed bacteria were diluted to 1 ×105 CFU per mL in OPA buffer (Hanks Balanced Salt Solution (Life Technologies), 0.1% gelatin, 1 mM MgCl2, 2.5mM CaCl2) and 20 μL (103 CFU) of the bacterial suspension was opsonized with 10 μL of serially diluted sera for 30 min at RT in a 384-well tissue culture microplate. Subsequently, 10 μL of 2.5% complement (Baby Rabbit Serum, Pel-Freez) and 10 μL of HL60 cells (at 200:1 ratio) were added to each well, and the mixture was shaken at 2000 rpm for 45–60 min at 37 °C in a 5% CO2 incubator. For the encapsulated O25B:K5 strain, bacteria were directly combined with complement and HL60 cells without the preopsonization step and shaken at 2000 rpm for 60 min at 37 °C under 5% CO2. In this case, 4% baby rabbit complement and HL60 cells at a 100:1 ratio were used. After the incubation, 10 μL of each 50 μL reaction was transferred into the corresponding wells of a prewetted 384-well Millipore MultiScreen HTS HV filter plate containing 50 μL of water. After vacuum-filtering the liquid, 50 μL of 50% DMEM was applied and filtered and plate-incubated overnight at 37 °C in a sealed zip-lock bag. The next day, the colonies were enumerated after staining with Coomassie dye using an ImmunoSpot analyzer and ImmunoCapture software. To establish the specificity of OPA activity, immune sera were preincubated with 20 μg per mL of the homologous serotype purified O-antigen prior to the opsonization step. The OPA assay includes control reactions without HL60 cells or complement to demonstrate the dependence of any observed killing on these components. Individual serum OPA titers were calculated using variable slope curve fitting (Microsoft Excel). Combined data were plotted using GraphPad Prism to generate GMTs and associated p values for significance (one-way ANOVA with log-transformed data).

ELISA for Spacer-Specific IgG

Conjugates with the BSA protein and various monosaccharides with different linker moieties (α-glucose-C2-NH2, β-glucose-C2-NH2, β-N-acetylglucosamine (GlcNAc)-C2-NH2, β-glucose-C5-NH2, and β-GlcNAc-ethylene glycol-NH2 [structures, see Figure S1]) were prepared. Briefly, the monosaccharides were activated using a 5-fold molar excess of bis(4-nitrophenyl) adipate in DMSO in the presence of triethylamine for 3 h at room temperature. The reaction mixture was frozen using liquid nitrogen and lyophilized for 20 h. The solid crude product was washed three times with chloroform. The washed reaction product (45 equiv) was then reacted with BSA in 100 mM sodium phosphate buffer, pH 7. The reaction mixture was stirred at room temperature for about 20 h. Then, the reaction mixture was washed twice with 100 mM sodium phosphate buffer, pH 7, and three times with phosphate-buffered saline (PBS), pH 7.4, using Amicon Ultra-4 centrifugal filters with 10 kDa MWCO (Merck Millipore, cat. no. UFC801096). Finally, the conjugates were sterile-filtered using Ultrafree-CL 0.22 μm filters (Merck Millipore, cat. no. UFC40GV0S). The protein concentrations were determined using the QuantiPro BCA Assay Kit (Sigma-Aldrich, cat. no. QPBCA-1KT) according to the manufacturer’s protocol. Monosaccharide-to-protein ratios (loading factors) were determined using MALDI-TOF MS analysis, as described above. The resulting BSA conjugates were coated on ELISA plates at 2 μg glycan per well in PBS at 4 °C overnight. The plates were then blocked with 5% (w/v) skim milk powder in PBS for 1 h at room temperature. Pooled antisera (n = 25 mice) of the postdose 3 time points were diluted in 5% (w/v) skim milk powder in PBS and incubated on the ELISA plates for 1 h at room temperature. As a positive control for spacer-reactive antibodies, we used an in-house generated pool of antisera generated with an oligosaccharide–CRM197 conjugate in BALB/c mice with a strong reactivity to the spacer moiety comprising a C5 linker (not shown). The secondary antibody was goat antimouse IgG-HRP (Dianova, cat. no. 115-035-164) used in a 1:10,000 dilution and was incubated for 30 min at room temperature. The plates were then incubated with a 1-Step Ultra TMB-ELISA Substrate Solution (Thermo Scientific, cat. no. 34028) for approx. 15 min at room temperature. Then, the reaction was stopped with 2 M H2SO4 and absorbance at 450 nm was determined in a 96-well plate reader. Data were visualized with the GraphPad Prism software, version 8.4.3 (GraphPad Software).

Kinetic Measurements of Monoclonal Antibodies

Kinetic measurements were performed on a Biacore T200 instrument at 30 °C with HEPES-buffered saline buffer containing 3 mM EDTA and 0.05% (v/v) P20 (HBS-EP) running buffer. E. coliO-antigen O25B CRM197 conjugate ligands were immobilized on a series S sensor chip CM5 at about 50 resonance units per flow channel. A reference surface functionalized with ethanolamine was prepared as a negative control. Multicycle kinetics was performed with an antibody analyte dilution series starting at 1000, 200, or 50 nM along with no analyte as reference. Analyte dilutions were passed over ligand for 2 min at 50 μL per min flow rate followed by a 10 min dissociation step. For successful regeneration between cycles, an ionic cocktail was used.[94] Biacore T200 evaluation software version 3.1 generated double reference subtracted sensorgrams before applying a 1:1 fit for all tested O25B ligand–antibody combinations.