OUP user menu

Acid-detoxified Inaba lipopolysaccharide (pmLPS) is a superior cholera conjugate vaccine immunogen than hydrazine-detoxified lipopolysaccharide and induces vibriocidal and protective antibodies

Cyrille Grandjean, Terri K. Wade, David Ropartz, Logan Ernst, William F. Wade
DOI: http://dx.doi.org/10.1111/2049-632X.12022 136-158 First published online: 1 March 2013


Worldwide, in endemic areas of cholera, the group most burdened with cholera is children. This is especially vexing as young children (2–5 years of age) do not respond as well, or for as long as adults do, to the current killed oral cholera vaccines (OCV). Conjugate vaccines based on the hapten-carrier paradigm have been developed for several bacterial pathogens that cause widespread and severe diseases in young children. We and others have studied different formulations of Vibrio cholerae (Vc) O1 lipopolysaccharide (LPS, a T-independent antigen) conjugates. Detoxified LPS is a central component of a LPS-based conjugate vaccine. pmLPS, which is detoxified by acid treatment, is a superior immunogen compared with hydrazine-detoxified LPS (DetAcLPS) that has altered lipid A acyl chains. The other feature of pmLPS is the ability to link carrier proteins to a core region of sugar. pmLPS readily induced vibriocidal antibodies following one intraperitoneal dose in a MPL-type adjuvant One dose of the pmLPS conjugate was suggestive of being protective; a booster resulted in protective antibodies for infant mice challenged with virulent cholera.

Key words
  • Vibrio cholerae
  • cholera
  • conjugate vaccine
  • lipopolysaccharide


Recent events in Haiti, Sierra Leone, Ghana, and other West African countries highlight the menace of epidemic and endemic cholera. There are two OCV that are WHO-approved: Dukoral and Shanchol. Dukoral and Shanchol are killed, whole-cell (kW-C) formulations representing both biotypes of Vc and the two main O1 serotypes, Inaba and Ogawa. Some feel that the utility of OCV is settled science with the current vaccine representing the best chance of providing immunity against cholera (Desai & Clemens, 2012). This conclusion is likely true for those individuals living in endemic areas with the exception of young children between 2 and 5 years of age (Deen et al., 2008; Harris et al., 2012). This cohort does respond to the OCV in a manner similar to older individuals, but they consistently have their immunity wane faster than older individuals (Sinclair et al., 2011; Wade, 2011; Harris et al., 2012). The anti-LPS antibody response is well known for those infected by Vc or vaccinated with OCV (Ahmed et al., 1970; Glass et al., 1985). The vibriocidal antibody titers of patient sera are considered a good, but surrogate, marker of protection (Glass et al., 1985; Khan et al., 1987). There are only limited reports, but children do produce antibody-secreting cells (ASC) that secrete anti-LPS antibodies, but their ACS responses trend lower than older individuals (Leung et al., 2011). If there is an intrinsic defect in the young in their response to respond to LPS, a type II T-independent antigen, then adjustments in the presentation of the protective LPS epitopes may be required.

Vibrio cholerae lipopolysaccharide (LPS) is a first-line immunogen for cholera vaccines (Wade, 2011). The O-specific polysaccharide (O-SP) of Vc's O1 serotypes is composed of 12–18 substituted perosamines [α-(1→2) linked 4-amino-4,6-dideoxy-d-mannoses] whose amino group is acylated with 3-deoxy-l-glycero-tetronic acid residues (Chatterjee & Chaudhuri, 2003). The terminal O-SP sugar defines both the Inaba and Ogawa O1 serotypes and several protective epitopes (Hisatsune et al., 1993, 1996). Native LPS induces inflammation, an unwanted occurrence at the inoculation site for parenteral Vc vaccines. What is not so appreciated is how the removal or alteration of the lipid A moieties modifies Vc's LPS immunogenicity especially for young children (Gupta et al., 1992; Grandjean et al., 2009).

Conjugate vaccines are superior to nonconjugate vaccines for immunizing young children against disease due to encapsulated bacteria. These successful vaccines have changed the circulating serotypes of several pathogens (Picazo et al., 2012). The use of parenteral conjugate vaccines in pediatric medicine is more prevalent than mucosally delivered conjugate vaccines that are still experimental, but the success of parenteral conjugate vaccines has presaged the introduction of orally delivered conjugate vaccines that are being developed in animal systems (Buchanan et al., 2001; Ren et al., 2011). The translation of conjugate vaccine technology to issues of immunizing young children against cholera is dependent on finding a highly optimal vaccine configuration before the challenges of oral immunization are addressed. It is not clear whether the effectiveness of parenteral conjugate vaccination will translate to oral priming and induction of immunity. To address the issue of how to optimize the anti-LPS response for young children, the hapten-carrier paradigm was applied by several groups that include Dr. Kabir's pioneering work (Grandjean et al., 2009; Meeks et al., 2004; Chernyak et al., 2002; Gupta et al., 1998 Gupta et al., 1992; Kabir, 1987). Several reports from Robbins' group showed that hydrazine-detoxified LPS (deacylated LPS or DeAcLPS) coupled to cholera toxin (CT) was immunogenic in mice and humans (Gupta et al., 1992, 1998). The human trials suggested that two doses of the Inaba-CT conjugates induced IgG-based vibriocidal antibodies. The use of synthetic Inaba and Ogawa conjugates that obviated the need to purify or detoxify Vc LPS has been reported (Chernyak et al., 2002; Meeks et al., 2004). The main difference between synthetic O-SP protein conjugates and those derived from detoxified LPS was the lack of protection induced by Inaba synthetic conjugates (Meeks et al., 2004), but the facile induction of protective antibody by Ogawa-based synthetic conjugates (Chernyak et al., 2002).

The methodology to produce different cholera conjugate vaccines has explored the role of the type of attachment to the carrier, the carrier used, and the method used to detoxify the LPS (Kabir, 1987; Gupta et al., 1992; Meeks et al., 2004; Cabrera et al., 2006; Grandjean et al., 2009; Paulovicová et al., 2010). A recent work by one of us showed that acid-detoxified Inaba LPS (pmLPS) conjugated to different carrier proteins induced vibriocidal antibodies that reacted with both Vc O1 serotypes. No switching was measurable of the anti-LPS antibody from IgM to IgG – an event that is typical in response to conjugate vaccine. We extend these observations and clearly showed that pmLPS is the premier immunogenic form of detoxified Inaba LPS. Adjuvant and intraperitoneal inoculation of the conjugates was superior to subcutaneous delivery. The potential for Inaba LPS conjugates to induce vibriocidal and thus protective antibodies was apparent after one dose of the pmLPS conjugate.

Materials and methods

Bacterial strains and vectors

The bacterial strains used in this study were as follows: N16961, Vibrio cholerae, El Tor, Inaba (ATCC® 39315TM), Pet15b vector that carries an N-terminal His-Tag® sequence, Novagen 69257 that was used to clone El Tor CBP-A and TcpF (Muse et al., 2012). His-tagged ET TcpA clone (pET15b/tcpA1) in Origami was provided by R. Taylor, Geisel School of Medicine at Dartmouth (Muse et al., 2012). Origami2 (DE3) cells (Novagen) were transformed with plasmid DNA for the individual Vc protein colonization factors. The CBP-A construct was also expressed in Origami to facilitate folding and enhance yield of the recombinant protein. His-tagged immunogens were purified with a Ni-column following manufacturer's instructions (GE Healthcare). Glycerol stocks were maintained at −80 °C. Inaba LPS was purified from V. cholerae strain 569B using the hot phenol–chloroform method as described (Chernyak et al., 2002). Extracted LPS was treated with DNase, Rnase, and proteinase K to reduce contaminating molecular species.

Generation of Inaba protein conjugates

Hydrazine detoxification of Vc LPS: DetAcLPS

Detoxification and derivatization of LPS were performed according to Gupta et al., 1992, with slight modifications. Briefly, freshly freeze-dried LPS (42 mg) was treated with hydrazine monohydrate (4.2 mL) at 37 °C for 2 h. After cooling, the crude reaction mixture was mixed with cold acetone until precipitation and centrifuged (7000 g for 30 min) at 4 °C. The precipitate was washed with acetone (twice) and dissolved in 0.15 M NaCl (40 mL) and centrifuged (7000 g for 5 h) at 4 °C. The supernatant was dialyzed exhaustively against deionized water, passed through a 0.22-μm-pore-size filter, and then freeze-dried. The resulting product is referred to as DetAcLPS.

Acid detoxification of Vc LPS: pmLPS

Lipopolysaccharide (100 mg) in 10 mL of 1% (v/v) aqueous acetic acid was heated at 100 °C for 90 min (Gupta et al., 1992; Grandjean et al., 2009). Precipitated lipid A was removed by low-speed centrifugation (800 g for 15 min) at 4 °C. The supernatant was purified by RP-HPLC. The collected fractions were diluted in water and freeze-dried to give pmLPS in 20–30% yield, depending upon the preparations. RP-HPLC separations were performed on a Nucleosil C18 300-5 C18 (Macherey-Nagel, France) (300 Å, 4.6 × 250 mm) column at a flow rate of 1 mL min−1 (analysis), or on a Nucleosil (Macherey-Nagel, France) (300 Å, 5 μm, 10 × 250 mm) column at a flow rate of 3 mL min−1 (semi-preparative), with ELSD and UV (215 nm) detection: gradient: 0% B for 5 min, 0–30% B over 25 min, or 0–25% B over 25 min, respectively; solvent system A: H2O; solvent system B: CH3CN.

Conjugate preparation using the cyanogen bromide/adipic acid dihydrazide (CNBr/ADH) activation procedure

Preparation of native LPS-bovine serum albumen: wt LPS-BSA

Purified Inaba LPS (5 mg/250 μL H2O) was brought to pH 10.5 with 1 N NaOH, and an equal weight of CNBr (5 M solution in CH3CN) was added. The reaction pH was maintained between 10 and 11 with 1 M NaOH for 3 min. An equal volume of 0.5 M ADH in 0.5 M NaHCO3 was then added, and the pH was adjusted to 8.5. The reaction mixture was stirred at room temperature for 1 h and at 4 °C overnight. This mixture was then dialyzed against water and freeze-dried to provide LPS-ADH. To generate the wt LPS-Bovine serum albumen (BSA) conjugate, LPS-ADH (1.2 mg) was added to BSA (1.7 mg) in solution in 0.15 M NaCl (200 μL). The pH was adjusted to 5.5 (indicator paper) using 0.1 N HCl. EDC (2 mg, actual concentration 0.05 M) was then added, and the pH was maintained at 5.5–6 for 1 h. The crude reaction mixture was then extensively dialyzed (cutoff 30 kDa) against 25 mM PBS, pH 7.3, to give wt LPS-BSA conjugate. Sugar concentration was determined by high-performance anion-exchange chromatography following TFA hydrolysis of the conjugate using LPS as a standard (Talaga et al., 2002).

The native LPS-chitin-binding protein A (CBP-A) conjugate (wt LPS-CBP-A), the hydrazine-detoxified LPS-BSA (DetAcLPS-BSA), the hydrazine-detoxified LPS-CBP-A (DetAcLPS-CBP-A), and the hydrazine-detoxified LPS-toxin co-regulated pilus A (TcpA) conjugate (DetAcLPS-TcpA) were prepared analogously using equal amount of either wt LPS or DetAcLPS and protein at 1.2 mg, 5.4 mg, 1.2 mg, and 4 mg, respectively.

Conjugate preparation using the thiol/maleimide coupling chemistry

Preparation of acid-detoxified LPS-BSA and acid-detoxified LPS-CBP-A conjugates: pmLPS-BSA, pmLPS-(ms)-CBP-A, and pmLPS-(sm)-CBP-A

pmLPS-BSA: To a solution of pmLPS (15 mg) in 0.1 M potassium phosphate buffer, pH 7.3 (1.5 mL), was added 3-(maleimido)propionic acid N-hydroxysuccinimide ester, in three portions every 45 min [3 × 3.43 mg dissolved in CH3CN-H2O 1 : 1 (200 μL)]. Following an additional period of 45 min, pH of the crude reaction mixture was adjusted to 5.5 by adding AcOH, dialyzed against water containing 0.1% (v/v) AcOH, and freeze-dried to give an inseparable mixture (by RP-HPLC) of derivatized and nonderivatized pmLPS further designated as pmLPS-mal.

In parallel, to BSA (Sigma, A0281) (30.44 mg, 20 mg mL−1 in PBS 0.1 M pH 7.5) was added S-acetylthioacetic acid N-hydroxysuccinimidyl ester (NHS-SATA) (3 × 6.82 mg of a 60 mg mL−1 solution in CH3CN, 3 × 50 equivalents), in three portions every 45 min. Following an additional reaction period of 45 min, the crude reaction mixture was dialyzed (cutoff 30 kDa) against 0.1 M potassium phosphate buffer, pH 6.6, at 4 °C to eliminate excess reagent.

SATA-activated BSA (3.5 mg) in 0.1 M potassium phosphate buffer solution, pH 6.6, was next added to pmLPS-mal at about a 1 : 10 molar ratio (based on a 30% conversion of pmLPS, estimated from our previous investigations) (Grandjean et al., 2009). The reaction mixture was buffered at a 0.5 M concentration by addition of 1 M potassium phosphate buffer, pH 6.6. Then, NH2OH.HCl (10 μL of a 2 M solution in 1 M potassium phosphate buffer, pH 6.6), was added to the mixture, and the coupling was carried out for 2 h at room temperature. pmLPS-BSA conjugate was finally obtained following dialysis (cutoff 30 kDa) against PBS 10 mM, pH 7.4, to remove excess reagents.

pmLPS-(ms)-CBP-A was prepared similarly from SATA-activated CBP (obtained from 1.2 mg of CBP-A).

pmLPS-(sm)-CBP-A: To a solution of pmLPS (10 mg) dissolved in PBS pH 7.3, 20 mM (1 mL) NHS-SATA was added (3 × 1.27 mg, 3 × 25 equivalent, dissolved in 50 μL of CH3CN) in three portions every 45 min. The pH of the reaction mixture was controlled (indicator paper) and maintained at pH 7–7.5 by addition of 0.5 M aqueous NaOH. Following an additional reaction period of 40 min, the crude reaction mixture was dialyzed (cutoff 3.5 kDa) against water at 4 °C and then freeze-dried to give pmLPS-SATA (mixture of derivatized and nonderivatized pmLPS).

In parallel, to a solution of CBP (6 mg) in PBS pH 7.3, 20 mM (1 mL) 3-(maleimido)propionic acid N-hydroxysuccinimide ester was added (3 × 0.87 mg, 3 × 30 equivalent, dissolved in 20 μL of CH3CN) in three portions every 45 min. Following an additional reaction period of 40 min, the crude reaction mixture was dialyzed (cutoff of 10 kDa) against 0.1 M potassium phosphate buffer (pH 6.0) at 4 °C.

Following dialysis, maleimide-activated CBP-A (1.2 mg) in 0.1 M potassium phosphate buffer solution, pH 6 (200 μL), was reacted with pmLPS-SATA (1.3 mg) in 1 M potassium phosphate buffer (pH 6.5) (200 μL). Then NH2OH.HCl (15 μL of a 1 M solution in 1 M potassium phosphate buffer, pH 6.5) was added to the mixture, and the coupling was conducted for 2 h at room temperature. pmLPS-(sm)-CBP-A was purified by dialysis (cutoff 20 kDa) against PBS 20 mM, pH 7.3.

Determination of the sugar concentrations

Sugar concentration in every conjugate formulation was determined by high-performance anion-exchange chromatography following TFA hydrolysis of the conjugate using the corresponding polysaccharide hapten (LPS, DetAcLPS, or pmLPS) as a standard (Talaga et al., 2002). 50 μL of sample was mixed with 12.5 μL of a 10 N TFA (final concentration 2N) in a glass tube sealed with a screwcap. After 2 h at 100 C, solutions were freeze-dried and then redissolved in 50 μL of water, and the solutions were transferred to autosampler vials. In separate tubes, standard mixtures of known amount of disaccharide 3 (from 0 to 50 μg) were treated similarly and were used for chemical identification and quantification. Chromatography was performed for the samples using a Dionex ICS 3000 system and a pulsed amperometric detector. 10 μL of the extract was injected through a 4 × 50 mm Propac PA1 precolumn (Dionex), before separation of anionic compounds on a 4 × 250 mm Propac PA1 column (Dionex) at 35 °C, at a flow rate of 1 mL min−1. Peak analysis was performed using chromeleon software, version 7.0. Solvent A was H2O; solvent B was a 100 mM NaOH solution; and solvent C was a 100 mM NaOH containing 0.5 M sodium acetate solution. Anionic compounds were eluted with a multi-step gradient as follows: 0–7 min, 20 mM NaOH; 7–15 min, 20–100 mM NaOH; 15–60 min, 100–0 mM NaOH, and simultaneous gradient of NaAc from 0 to 500 mM. After every run, the column was re-equilibrated in 20 mM NaOH for 10 min. Analyses were carried out in triplicates.


Adult female BALB/c mice were obtained from the National Cancer Institute (Bethesda, MD) for generation of antisera. Untimed, pregnant CD-1 mice were purchased from Charles River (Raleigh, NC) and they were the source of the 3- to 5-day-old pups for the infant mouse challenge assay (Meeks et al., 2004; Muse et al., 2012). Animals were housed at the Geisel School of Medicine at Dartmouth's Animal Resource Center at Dartmouth-Hitchcock Medical Center, which is accredited by the American Association for Accreditation of Laboratory Animal Care.

Unless otherwise noted, adult mice were immunized twice intraperitoneally (i.p.) on days 0 and 21 with some conjugates emulsified in the MPL®+TDM Adjuvant system (formally known as RIBI®) from Sigma (Table 1). Retro-orbital puncture on days 0 (prebleed, PB) and 21 (primary bleed) was used to obtain sera. After 42 days, mice were anesthetized and ensanguined (secondary bleed). The protection assay used pooled group sera, but individual sera were analyzed for endpoint titers (ELISA and vibriocidal).

View this table:
Table 1

Characterization of the conjugates

ConjugateLinkerYield (%)Composition (mgmL−1) Protein CHO
  • CHO, carbohydrate.

  • The yield is calculated on the basis of the weight of the protein in the conjugate compared with the starting protein.


The LPS- and carrier-specific serum antibody endpoint titers were determined as described (Chernyak et al., 2002; Meeks et al., 2004; Wade et al., 2006). Endpoint titers are the reciprocal of the antibody dilution of the last positive well following the subtraction of twice the average optical density (Dynatech Laboratories MRX microplate reader) of the negative control (PB) wells. A more detailed description of the ELISA and vibriocidal assay has been given (Muse et al., 2012). The ELISA and vibriocidal endpoint titers for the positive and control sera used in this study are provided in the insets. To control for nonspecific immunity of mice in the various experimental groups, PB serum samples were used as the negative controls for ELISA and vibriocidal assays. Prebleed endpoint titers in the ELISA analyses were assessed at the baseline dilution of either 100 or 200 (reciprocal of the starting serum dilution). Those experimental sera that still contained OD units after subtracting the background were considered positive, but the utility of the response in protection or vibriocidal assays was dubious. The positive control was pooled BALB/c mouse sera from mice immunized four times (i.p.) 21 days apart with 9.0 μg (first two doses) or 5.0 μg (last two doses) of purified Inaba LPS. The ELISA endpoint titers for the positive control were 25600 (IgM) and 2560 (IgG), while the vibriocidal endpoint titer was 320.

Vibriocidal microtiter assay

Fournier group's microtiter method was used to determine vibriocidal endpoint (Boutonnier et al., 2003; Meeks et al., 2004; Muse et al., 2012). Vibriocidal activity of mouse antisera is reported as the reciprocal of the antibody dilution of the well showing no growth. A titer > 40 was considered vibriocidal. The negative control included Vc N16961 bacteria and complement only; the positive control sera were described above and generated from purified Inaba LPS. The PB sera showed the same results as the bacteria and complement-only wells. This reports no effect on metabolism (viability) and thus positive ODs. The endpoint titers were converted to log10 for statistical analysis and graphical presentation.

Infant mouse challenge assay

Three- to five-day-old CD-1 neonates were used in the cholera challenge model to assess the protective quality of anti-Vc colonization factor antibodies (Chernyak et al., 2002; Taylor et al., 2004). For some experiments, cultures of Vc N16961 were grown for 17 h in 2 mL of Luria–Bertani (LB) broth, pH 6.5, at 30 °C on a rotating wheel. Twenty-five microliters of bacteria representing 2–12 LD50 was combined with 25 μL of pooled PB serum, pooled anti-Inaba LPS hyperimmune serum, or pooled primary or secondary antiserum diluted 1 : 10 with normal heat-inactivated mouse serum (Ab Serotec) immediately before intragastric administration to the infant mice. Challenged mice were kept at 30 °C and monitored every 4 h, beginning 24 h after challenge for clinical signs of cholera: dehydration, loss of vitality, and diarrhea (Taylor et al., 2004).

In other studies to hyperinduce TcpA and TcpF expression, we used AKI broth to grow Vc for the challenge assays. Vc was grown in AKI broth for 3 h without shaking at 30 °C and then transferred into culture tubes and incubated at 30 °C on a rotating wheel for 17 h (Iwanaga et al., 1986). The growth of Vc at 30 °C in LB and AKI is both compatible with Vc colonization, but different numbers of bacteria in the challenge inoculum yield different virulence (W. F. Wade personal observation). The number of LB-grown bacteria (≈107) when compared with the same number of AKI-grown Vc are much less virulent with 50% (LB-grown) of the mice dying during the test periods compared with 100% (AKI-grown). If mice are normalized for the same percent death (50%) over the observation period, then 1–3 × 106 AKI-grown Vc is equivalent to 107 LB-grown Vc (WF Wade personal observation).

Statistical analysis

Antibody titers were scored as endpoint titers for individual serum sample for all groups. Sera that did not generate a signal in the ELISA at the starting dilutions were considered negative, but designated as 1 for the endpoint titer, so log10 transformation could be performed, and means and standard deviations calculated were analyzed by analysis of variance (anova) and Tukey's post-test comparison, both of which are available on GraphPad Prism software (Version 5, GraphPad Software, San Diego, CA). The survival curves were also generated using GraphPad Prism software, which uses the Kaplan–Meier method to generate curves. To test the null hypothesis that survival curves did not differ, the Mantel–Haenzel test was used. P values > 0.05 are not considered significant.


Generation and characterization of the conjugates

Detoxified Inaba LPS structures

Vibrio cholerae O1 LPS-based antigens are one of the principle targets of the adaptive immune response to cholera, whether for infected individuals or those given OCV (Glass et al., 1985; Khan et al., 1987). Conjugation of detoxified Vc LPS to a protein carrier might overcome the lack of long-term protection of young children that is described by the absence of long-term memory B cells following immunization with the kW-C cholera vaccine preparations currently in use (Leung et al., 2011; Patel et al., 2012). In this study, we compared experimental cholera vaccine conjugates based on using either purified native (wt) or two types of detoxified Inaba Vc LPS (Fig. 1). We wanted to investigate wt LPS as a hapten to determine whether conjugation might mitigate TLR4 binding, but retain enough agonist activity to enhance anti-LPS responses, which are typically limited to detoxified LPS (Gupta et al., 1992).

Figure 1

Differences in detoxification of Inaba LPS: pmLPS vs. DetAcLPS. (a) The top structure represents detoxified Inaba LPS (pmLPS) derived using a less stringent treatment method of purified LPS, which removes the lipid A moiety but preserves most of the core sugars except the fructose and the downstream Kdo, hydrolyzed and rearranged, respectively. The amino group on the glucosamine residue that was the point for carrier attachment as shown in Fig. 1b offers a unique point of derivatization with linkers and was used for attachment of the conjugates' carrier proteins. (b) Hydrazine treatment of purified LPS results in a detoxified Inaba LPS (DetAcLPS) that includes the O-SP, the core sugars, and part of the lipid A moiety.

One method to detoxify LPS used mild acid hydrolysis (Gupta et al., 1992) and results in a LPS structure termed pmLPS (Fig. 1a). pmLPS results in the release of a fructofuranose, as well as the rearrangement of a 3-deoxy-manno-2-octulosonic acid (Kdo) sugar residue in the core region. The other method is based on hydrazine treatment (Gupta et al., 1992) that removes some of the acyl chains from the Lipid A component of LPS (Fig. 1b). The O-SP structure of pmLPS is not appreciably altered, and pmLPS was easily purified by RP-HPLC (Grandjean et al., 2009) (Fig. 2b). Resonances of the most abundant protons, that is, those of the N-(3-deoxy-l- glycero-tetronyl)-d-perosamines, were readily observed on the 1H NMR spectrum of the pmLPS and consistent with the data previously reported (Fig. 3) (Hisatsune et al., 1993; Zhang & Kovác, 1997).

Figure 2

Analysis of DetAcLPS and pmLPS. (a) Gel filtration chromatogram of DetAcLPS. Conditions: Superdex 75 column, at a flow rate of 0.6 mL min−1 using H2O as eluent, and UV detection at 215 and 254 nm. (b) Analytical RP-HPLC profile of pure pmLPS, ELSD (main trace), and 215-nm (minor trace) dual detection (for elution conditions, see ‘Material and methods’). DetAcLPS is a detoxified LPS based on hydrazine treatment; pmLPS is a detoxified LPS based on acid treatment. pmLPS stands for polysaccharide moiety of LPS and comprises the O-specific polysaccharide and most of the core residues (see Fig. 1a for detailed structure).

Figure 3

1H NMR (500 MHz, D2O) spectrum of pmLPS following RP-HPLC purification. Only the chemical shifts related to the repeat unit of the O-SP, which is overexpressed compared with the core residues, have been indicated in the spectra.

pmLPS was characterized by both negative and positive MALDI-TOF mass spectrometry (Fig. 4a and b, respectively). The former spectrum (Fig. 4a) contains a series of [M–H]pseudomolecular ions ranging from m/z 4597 to 7077, with a difference of 248 mass unit between every two neighboring signals, which corresponds to the mass of one perosamine of the Vc O-antigen homopolymer. However, each peak is observed at +18 Da compared with our precedent results (Grandjean et al., 2009), suggesting a possible hydration of the present form of the pmLPS (perhaps of the ketoacid function of the downstream residue) (Fig. 4a). The mass spectrum reveals the pmLPS preparation also contains a second series of peaks (peaks observed 192 Da below [M–H]pseudomolecular ions) that have been tentatively attributed to species lacking one heptose of the core. The positive MALDI-TOF spectrum was more complex, although revealing the same polymeric distributions (Fig. 4b). Apart from the [M+H]+pseudomolecular ions ranging from m/z 3113 to 6081, the spectrum also contains potassium adduct ions (mainly ranging from m/z 4389 to 7108). Among the remaining series of peaks, species lacking one tetronyl side chain ( e.g. series of [M+K-102]+ or [M+H-102-18]+ ions) were also observed as previously reported (Grandjean et al., 2009).

Figure 4

Linear negative (a) and positive MALDI-TOF (b) mass spectra of pmLPS, determined on an Autoflex III MALDI-TOF/TOF spectrometer (BrukerDaltonics) using a 2000–10000 m/z mass range. 75 pmol of each sample, dissolved at 2 mg mL−1 in H2O, was mixed with 2,5-dihydroxybenzoic acid at 100 mg mL−1 in H2O/CH3CN/N,N′-dimethylaniline (50 : 50 : 0.2) as matrix and was deposited on MALDI target plate. (For clarity, the m/z corresponding to minor series of peaks has been omitted in the positive spectrum).

Protein carriers for Vc LPS epitopes

Three different proteins were investigated as carriers. BSA and two soluble Vc adhesion/colonization factors that are protective antigens: chitin-binding protein A (CBP-A) and toxin co-regulated pilus A (TcpA). Cholera toxin is not generally considered a protective antigen and therefore was not used. Experimentally, CBP-A and TcpA antibodies are associated with strong protective responses in mice (Kirn & Taylor, 2005; Kirn et al., 2009). DetAcLPS and wt LPS were thus reacted with adipic acid dihydrazide following cyanogen bromide activation and further coupled to the acid residues of BSA and CBP-A, as well as to those of TcpA for the former antigen to give a series of five conjugates (Table 1). This method of conjugation is robust and well adapted to amphiphile derivatives such as DetAcLPS and wt LPS. However, as any nucleophilic group of the polysaccharide can be activated, it can lead to a population of conjugates that might have an effect on antigenic determinants.

Advantage of pmLPS and details of coupling chemistry

For pmLPS, one of us recently established that a unique, free amino group of a glucosamine residue of the core (Fig. 1b) could be used as single-point attachment to afford fully controlled neoglycoconjugates. (Grandjean et al., 2009). This strategy was adopted herein using the thiol/maleimide coupling chemistry because of its advantages. Staudinger ligation is accompanied with substantial antigenic determinant alterations (Grandjean et al., 2005); both thiol/bromoacyl ligation (Grandjean et al., 2009) and the squaric acid chemistry (C. Grandjean unpublished results and Xu et al., 2011) need larger excess of the polysaccharide antigens to reach coupling ratios equivalent to that observed when using thio- and maleimide-functionalized partners. However, we previously observed a substantial antilinker Ab response, a known pitfall of the thiol/maleimide chemistry, although generally observed for aromatic or alicyclic cross-linkers (Peeters et al., 1989; Buskas et al., 2004). To restrain the accessibility of the spacer to the immune system and thus minimize interfering responses, we switched from a (maleimido)hexanoyl to a shorter (maleimido)propanoyl linker whose use was found appropriate in humans (Verez-Bencomo et al., 2004). The amino group of pmLPS was derivatized using ester-activated derivatives of either 3-(2,5-dioxo-2,5-dihydro-1 H-pyrrol-1-yl)propanoic acid or S-acetylthioacetic acid to give pmLPS-mal and pmLPS-SATA, respectively. Differences in hydrophobicity between native and modified pmLPS were, however, not sufficient to obtain their full separation by RP-HPLC as previously performed. The mixtures were therefore reacted as a whole, assuming a 30% conversion on the basis of our previous results (Grandjean et al., 2009), with either BSA or CBP-A, each functionalized with the pending S-acetylthioacyl- or maleimide-reactive group to give pmLPS-BSA, pmLPS-(ms)-CBP-A, and pmLPS-(sm)-CBP-A (Table 1 and Fig. 5). The last two conjugates differ depending on whether the maleimide or the S-acetylthioacetyl groups were introduced on the pmLPS moiety, respectively.

Figure 5

Analytical gel filtration chromatograms of (a) CBP-A and (b) pmLPS-(ms)-CBP-A. Conditions: Superdex 75 column, at a flow rate of 0.6 mL min−1 using H2O as eluent, and UV detection at 215 nm.

Conjugate parameters that effect immunogenicity

Anti-Inaba LPS IgM endpoint titers

We examined the immunogenicity of three carriers and three forms of Inaba LPS that comprised the experimental LPS-protein conjugates we designed. Two of the three forms of LPS (haptenic epitopes) represent well-known detoxified derivatives of O1 Inaba LPS. The third, native LPS, was not detoxified before the conjugation with the hope of attenuating its toxicity, but retaining some of its ability to produce robust vibriocidal and protective antibody in all the mice inoculated (Wade & Wade, 2008). To simplify the graphical data presentation of the 15 experimental groups, we collected the negative (PB sera) and positive controls to present them as insets (Fig. 6a–c).

Figure 6

Serum endpoint titers (ELISA and vibriocidal) of mice immunized with various experimental Inaba-O-SP-carrier conjugates (Table 2). The data for Fig. 6a–c were analyzed by anova, and a Tukey's postcomparison test was run for all possible row comparisons. The P value for the null hypothesis that the means were not significantly different was P < 0.0001 for data in panels A–C. There are 451 comparisons for each panel, which are available upon request. Select comparisons of biological significance are highlighted in the results. Inset: Sera collected from mice prior to immunization were pooled and used as the negative control for the ELISA and vibriocidal assays, and pooled sera from mice immunized four times with wt Inaba LPS served as the positive control. Individual sera for mice in each experimental group were assayed by ELISA for LPS IgM and IgG titers and for vibriocidal activity against N16961. The starting dilutions for the ELISA were either 1 : 100 or 1 : 200 for primary sera and 1 : 200 for secondary sera. (a) Vc Inaba LPS-reactive IgM. Conjugates with TcpA-based carriers with adjuvant (group 1) and without adjuvant (group 2) showed no IgM titer primary response in group 1 compared with low-level primary response for group 2 ( P < 0.001), which when boosted both the secondary IgM titers rose modestly and were not significantly different from each other ( P > 0.05). When BSA was used as the carrier (groups 3 through 8), primary anti-Inaba LPS IgM responses did not significantly differ from each other ( P > 0.05). Group 3 and group 4 were notable as they were significantly different in their booster response due to pmLPS. The primary mean Inaba-specific IgM titers for groups 9, 10, and 12 were statistically equal. Omitting RIBI adjuvant for the experimental groups containing wt LPS-BSA (group 11 vs. group 9) resulted in lower IgM primary titers ( P < 0.001), and when comparing both groups inoculated with adjuvant, group 11 again had a statistically lower titer than group 12 (the wtLPS-CBP-A group, P < 0.001). After a booster, group 11 secondary sera continued to have a lower mean IgM endpoint titers (group 11 vs. groups 9 and 10, P < 0.001, group 11 vs. group 12, P < 0.01). Groups 13–15 compared various forms of LPS (pmLPS or detAc) and different linkers (sm or ms) with CBP-A as the carrier. Group 13 (pmLPS-CBP-A with a sm linker) proved superior in the primary and secondary IgM response compared with both groups 15 (pmLPS-CBP-A with a ms linker) and 14 (detAcLPS-CBP-A). (b) Vc Inaba LPS-reactive IgG. The anti-Inaba LPS IgG endpoint titers for groups 1 and 2 using TcpA as a carrier was only observed after mice received two inoculations of the conjugate DetAc LPS-TcpA in RIBI adjuvant. There were no measurable titers in the primary sera from this group, nor in primary or secondary sera from the same conjugate inoculated without RIBI. In groups 3–8 using BSA as the carrier, no primary IgG response was evident, except for group 8 that had wt LPS added to the pmLPS-BSA conjugate in RIBI. After a second inoculation, all groups responded with elevated IgG titers. Groups 9–12 looked at whether the inclusion of adjuvant would make a difference in IgG responses when using wt LPS conjugated to either BSA or CBP-A as a carrier. There was no significant IgG response in any of the primary sera for all groups, whether adjuvant was included or not. The secondary response was increased by the inclusion of adjuvant for both carriers. Groups 13–15 tested different linkers and the form of LPS and did not result in differences in the primary or secondary anti-Inaba LPS IgG titers. (c) Vibriocidal responses to Vc conjugates. No vibriocidal activity was noted for primary or secondary sera from both groups inoculated with DetAcLPS using TcpA as a carrier, whether RIBI was included (group 1) or omitted (group 2). Sera from other experimental groups of DetAcLPS conjugated to BSA as the carrier, either alone (group 5) or with added VcCF (group 3), also did not have measurable vibriocidal titers in either primary or secondary sera, unless wt LPS was added to the inoculum (group 6) where both primary and secondary vibriocidal titers were observed. Groups 4, 7, and 8 all had pmLPS conjugated to the BSA carrier, and all had vibriocidal activity in both primary and secondary sera, although the inclusion of the VcCF decreased the vibriocidal titer in the primary sera (group 4), while the addition of wt LPS increased both primary and secondary titers (group 8 compared with group 7 and 4, P < 0.0001). When comparing the vibriocidal titers of primary and secondary sera from the groups using wt LPS conjugated to BSA or CBP-A with or without adjuvant (groups 9–12), variable individual response within each group was noted. Groups 13–15 looked at the difference in the form of detoxified LPS and the linker attachment to the carrier. There was no vibriocidal antibodies produced from either primary or secondary sera of group 14 using detAcLPS conjugated to CBP-A. Vibriocidal activity in primary sera was seen ( P < 0.001) when pmLPS was conjugated to CBP-A by use of a sm linker (group 13), but not with the ms linker (group 15), and secondary sera from both groups showed an increase in titers with group 15 significantly lower than group 13 ( P < 0.001). (d) Pearson's product-moment analysis represents the correlation analysis between mean serum anti-LPS endpoint titers and the mean serum vibriocidal titers. The primary and the secondary IgM titers (panels 1 and 2) significantly correlated with the vibriocidal titers (primary, P = 0.0482, and secondary, P = 0.0347). This is in contrast to the primary ( P = 0.0565) and secondary ( P = 0.2451) IgG responses (panels 3 and 4) correlated with the vibriocidal effect of antisera. The rank orders for the comparisons are as follows: Panel 1, Primary IgM: 8>13>7>6>4>12>9=10>11=1=2=3=14=15; Panel 2, Secondary IgM: 13>8>7>4=6>12>>10>11>9>15>1=2=3=5=14=15; Panel 3, Primary IgG: 8>13>7>6>4>12>9=10>11>1=2=3=5=15; Panel 4, Secondary IgG: 13>8>7>6>4>12>10>11>9>15>1=2=3=5=14.

The individual mouse anti-LPS serum endpoint ELISA titers and the vibriocidal titers for primary and secondary sera from mice of the 15 treatment groups are presented in Fig. 6a–c. Comparing the anti-Inaba LPS response to TcpA-based carrier conjugates showed that omission of adjuvant allowed a low, but positive primary IgM titer not seen in the TcpA conjugate inocula that included RIBI (compare groups 1 and 2; P < 0.001, Fig. 6a). A booster immunization increased the secondary endpoint titers in both groups 1 and 2 that did not differ statistically ( P > 0.05). If BSA was used as a carrier, individual mice from groups 3–8 responded with IgM antibody in the primary serum sample. Graphically, some of the means appeared separated by distinct distances, but statistically, the mean IgM endpoint titers did not differ ( P > 0.05) among groups 3–8. A booster immunization increased the titers in secondary sera wherein group 3's mean response was significantly lower than group 4's ( P < 0.05), but group 3's response did not differ from groups 5–8 (all P values comparisons > 0.05). The magnitude of the anti-LPS IgM response was not increased with the addition of wt LPS as an additional adjuvant (group 5 vs. group 6; group 7 vs. group 8, Fig. 6a).

We tested the ability of native LPS (nondetoxified) to serve as the haptenic portion of the Vc LPS conjugates. Groups 9, 10, and 12 statistically had the same mean primary IgM response, but the response of group 11, wt LPS-BSA given without RIBI, was statistically lower than group 9 ( P < 0.001). Comparing the groups (11 and 12) without adjuvant that received wt LPS-based conjugates showed that CBP-A induced higher primary anti-LPS IgM serum endpoint titers ( P < 0.001). The secondary anti-Inaba IgM LPS response of groups 9, 10, and 12 did not differ after a booster, but compared with these groups, group 11 was still lower in its mean IgM endpoint titer (group 11 vs. group 9, P < 0.001; group 11 vs. group 10, P < 0.001; group 11 vs. group 12, P < 0.01). These data show that wt Inaba LPS can retain part of its immunogenicity, but with less than optimal IgM response engendered unless adjuvant or CBP-A is used as part of the inoculum, suggesting that wt LPS conjugates are not equivalent to wt LPS in their ability to consistently and robustly increase anti-IgM serum titers (Wade & Wade, 2008; Muse et al., 2012).

Groups 13–15 were configured to compare the form of LPS and the type of linkage of CBP-A to pmLPS on the immunogenicity of the conjugates (Fig. 6a). While the mean titers of anti-Inaba LPS IgM made in response to carriers with sm linker (group 13, carrier linker derived from maleimide functional groups) appeared higher than those of groups 15 and 14 (carrier linker derived from thioacetyl functional groups), the differences were not significant for either primary or secondary IgM responses ( P > 0.05).

Anti-Inaba LPS IgG endpoint titers

One of the goals of making Vc LPS conjugates is to obtain T-dependent-type responses to Inaba LPS B-cell epitopes. The switch to T dependence is linked to orchestrated germinal center events that should enhance isotype switching (IgM > IgG, IgA) with its attending affinity maturation of the anti-Inaba LPS. The anti-Inaba LPS IgG endpoint titers for conjugates with TcpA as a carrier were only evident if RIBI was used as an adjuvant (Fig. 6b, groups 1 and 2). Of the BSA-based carrier conjugates (groups 3–8), four induced primary anti-Inaba LPS IgG endpoint titers and only group 8 had a notable group response. Of note, these are low-level responses at the initial serum dilution. All the anti-Inaba LPS IgG endpoint titers (in the 1000s) were readily measured in secondary sera (Fig. 6b, filled symbols); however, the group means were not statistically significant compared with groups 3–6. Prism calculated the comparison of groups 7 and 8 as P < 0.05.

The trend for anti-Inaba LPS IgG endpoint titers was similar for groups 9–12 that featured wt Inaba LPS as part of the conjugate. Individual mice from groups 9–12 responded poorly ( P > 0.05) and variably with their initial IgG response to Inaba LPS presented in wt LPS-based conjugates. The inclusion of adjuvant (groups 9 and 10) did not significantly alter the mean primary serum endpoint titers compared with groups 11 and 12. The secondary response was affected by the inclusion of adjuvant but only for group 9 compared with groups 11 and 12. Mice responded with increased titers in their secondary sera as a group if adjuvant was included in the inocula.

The different linkers or the form of LPS did not result in primary or secondary anti-Inaba LPS IgG endpoint titers that were different (groups 13–15, all comparisons P < 0.05), but the responses showed less variance compared with BSA-conjugate-induced and wt LPS-conjugate-induced titers, but were of lower magnitude in the secondary sera compared with the BSA conjugates.

Vibriocidal responses to Vc conjugates

The vibriocidal response to Vc is mainly measured as an anti-LPS IgM response. In vaccine studies, it is considered correlative with protection. A recent report by Patel and co-workers suggested that anti-LPS IgG memory B cells were the best correlate with a decrease in risk of infection from household contacts of patients with cholera (Patel et al., 2012). TcpA as a carrier for DetAcLPS, the hapten group, induced low levels of IgM but not a vibriocidal response measurable in primary or secondary antisera (Fig. 6c, groups 1 and 2). Conjugates based on BSA and DetAcLPS (groups 3 and 5) showed similar results – some IgM and, in these groups, some IgG, but no vibriocidal responses. If wt LPS (group 6) was added to the adjuvant that was part of the DetAcLPS-BSA inocula, the vibriocidal titers were measurable in both primary and secondary sera. The mean vibriocidal endpoint titers of the primary responses of pmLPS-BSA conjugates, group 4 vs. group 7, were significant ( P < 0.001), while that of group 4 vs. group 8 were not P > 0.05 ). Group 8, which received a low dose of 0.5 μg wt LPS with the pmLPS-BSA conjugate, had a more consistent vibriocidal response than group 4 or group 7 ( P < 0.001). A consistent result was that the pmLPS form of LPS was able to induce vibriocidal antibodies, while DetAcLPS forms of detoxified LPS were not.

The vibriocidal response of groups 9–12, which featured wt LPS as the haptenic group of the conjugates, was variable, and in general, the mean primary and secondary vibriocidal endpoint titers were lower than that seen in response to pmLPS-BSA conjugates. Comparing the variability of the individual vibriocidal response of groups 9 and 11 to group 8 suggests that the inclusion of wt LPS as the hapten of a conjugate modified the capacity of the BSA conjugates to induce vibriocidal antibody; however, the use of CBP-A without adjuvant induced similar ( P > 0.05) primary responses to group 7, but the secondary response was significantly different ( P < 0.001). Group 12 induced the best primary and secondary response of groups 9–12, which is interesting as the anti-LPS IgM and IgG responses were comparable among those groups.

Finally, the difference in the form of detoxified LPS and the linker to attach the carrier was evaluated by groups 13–15, wherein sm-based pmLPS (group 13) induced the most robust primary and secondary vibriocidal responses (Fig. 6c, comparing group 13 with group 15, primary sera, P < 0.001, secondary sera, P < 0.001). DetAcLPS-CBP-A conjugate did not induce vibriocidal antibodies in the primary or secondary sera, while the use of ms linkers to attach CBP-A to pmLPS induced some measurable, but statistically lower (group 13 vs. group 15, P < 0.001) vibriocidal antibody titers in the secondary sera. Of note, group 13 induced vibriocidal antibody after one dose.

Correlation of ELISA endpoint titers (IgM and IgG) with vibriocidal endpoint titers

Because there were so many possible group comparisons for ELISA and vibriocidal endpoint titers, some of which were significant and many of which were not, we need a more general means to understand the relationship between the serum antibody titers and vibriocidal antibody titers. Pearson's product-moment correlation analysis is well suited for this type of comparison and is shown in Fig. 6d, panels 1 through 4. The primary and the secondary IgM titers (panels 1 and 2) were significantly correlated with the vibriocidal titers (primary, P = 0.0482, and secondary, P = 0.0347), while neither the primary ( P = 0.0565) nor the secondary (P = 0.2451) IgG responses (panels 3 and 4) correlated with the vibriocidal effect of antisera. Of note, there were multiple groups with notable vibriocidal titers in secondary sera compared with primary sera, which is what is expected after a booster. The primary vibriocidal titers of note that did not generate vibriocidal antibodies were groups with DetAcLPS conjugate (groups 1, 2, 3, 5, and 14). The role of wt LPS and pmLPS at either selecting the right B cells or initial priming conditions seems important to the consequence, that is, vibriocidal capacity. The conclusion from this analysis is that IgM titers correlate best with vibriocidal responses, which is consistent with previous literature that concludes IgM is more evolved per weight basis for a complement-mediated vibriocidal response (Ahmed et al., 1970). The general rank order of the vibriocidal capacity of the group's vibriocidal titers was as follows: 8>13>7>6>4>12>9=10>11>1=2=3=14=15. The upper third of the rank was composed of the same groups with groups 8 and 13 taking preeminence depending on the antibody isotype. The middle rank was generally sera from mice immunized with wt LPS conjugates, while the third grouping was composed of primarily DetAcLPS-protein conjugates.

Serum antibody response to experimental Vc conjugate vaccine carriers

TcpA serum endpoint titers in response to carrier- and noncarrier-bound TcpA

We envisioned Vc LPS conjugates composed of TcpA and Vc Inaba-O-SP epitopes as providing two sources of protective antigen (Osek et al., 1994). To determine whether TcpA retained its immunogenicity when conjugated to Vc-detoxified LPS, we added TcpA (along with TcpF and CBP-A) to the inoculum containing DetAcLPS- or pmLPS-based TcpA conjugates (Fig. 7a, groups 3 and 4). The addition of the three VcCF would also allow us to determine whether there was an additive protective response to that engendered by the LPS component of the conjugate. The primary immune response (IgM) to TcpA was not detectable (Fig. 7a, panel A), while the secondary IgM response (Fig. 7a, panel A) was evident in some individual mice. The primary TcpA antibody response of the IgG isotype was not significantly different for groups 1, 3, and 4 ( P > 0.05). The exception to the moderate anti-TcpA response was group 2's endpoint titers in the primary sera (Fig. 7a, panel B, P < 0.001) compared with means of groups 1, 3, and 4. The secondary anti-TcpA IgG responses were measurable for all groups, but again, group 2 mice did not respond as well to TcpA [group 2 (the lowest mean among groups) vs. groups 1, 3, or 4 ( P < 0.001)]. The response to TcpA as a carrier immunogen was lower than the response to noncarrier TcpA contained in the additional VcCF immunogens (groups 3 and 4). This may be because groups 3 and 4 received a higher total dose of TcpA.

Figure 7

Serum endpoint titers for protein carriers used in experimental conjugates. (a) Anti-TcpA serum endpoint titers (panel 1, IgM, panel 2, IgG). Mice were immunized with TcpA in the form of a conjugate or as added Ag to a conjugate with various carriers (Table 2). In the groups receiving nonconjugate TcpA, the dose was 75% for both inoculations (groups 3 and 4 vs. groups 1 and 2). This was due to the realities of conjugate chemistry and the amount of TcpA we had on hand. Anti-BSA serum endpoint titers (panel 1, IgM primary and secondary responses, panel 2, IgG primary and secondary responses). The doses of BSA for immunization of the groups are reported in Table 2. The range was 39–18.6 μg for the primary inoculation and 19.5–9.3 μg for the secondary. Group 4 with the lowest primary anti-BSA IgM and IgG titer received the lowest dose of BSA and displayed equivalent IgM and IgG secondary titers to groups that received some 2-fold more BSA. (c) Anti-CBP-A serum endpoint titers (panel 1, IgM primary and secondary responses, panel 2, IgG primary and secondary responses). Some groups were immunized with CBP-A in the form of a conjugate or as added Ags to a BSA-based conjugate (Table 2). Again due to the limitations of conjugate chemistry, the protein carrier amounts varied. The largest amount of BSA inoculated was 46.2 μg (groups 10 and 12), and the lowest was 8 μg (group 15) for the primary inoculation and half as much for the booster. Note that for most comparisons, the difference in CBP-A of the conjugate or as a standalone immunogen did not greatly influence the serum endpoint titers. Note that groups 3 and 4 with an inocula of almost four times as much inoculated CBP-A have similar IgM titers and even lower primary IgG titers than group 15 that was inoculated with less CBP-A, although in the form of a conjugate. These data indicate that the epitopes in CBP-A are equally accessible such that the polyclonal antisera to different doses and different forms of CBP-A do not greatly influence the serum titers.

BSA serum endpoint titers in response to BSA-based conjugates with or without additional Vc immunogens

The primary IgM response to BSA of the mice in the groups that received BSA-based conjugates was greater than the PB sera (data not shown) but not remarkable (Fig. 7b, panel A, open symbols). The weakest primary IgM response was that of group 4 that received the pmLPS-BSA conjugate with RIBI. The primary serum endpoint means were not in general statistically significant (data not shown). The secondary IgM response to BSA epitopes (Fig. 7b, panel A, closed symbols) was increased compared with the primary titers. The difference in means between groups 5 and 11, which were the highest and lowest means, respectively, was not significantly different (Fig. 7b, panel A, P > 0.05).

The IgG primary serum endpoint titers to BSA were positive in all groups (Fig. 7b, panel B, open symbols). There was a 10-fold range of response with group 4 pmLPS conjugate immunized mice being statistically lower than groups 3 and 5 ( P < 0.001). While some of the differences were significant, the reason for the differences did not track with an obvious biological function. The range of BSA carrier of the various conjugates was 18.6–39.0 μg (Table 2). Group 4 had the lowest anti-BSA antibody response and received 18.6 μg and 9.3 μg of BSA in the conjugate for the primary and booster. Group 11 received 22.7 μg and had a lower primary and secondary anti-BSA response than group 9 that also received 27 μg of conjugated BSA. A range was observed in the primary IgG anti-BSA responses, but the range was not as great in the secondary sera (Fig. 7b, panel B, closed symbols) where the titers centered around 1:10 million, suggesting BSA was highly immunogenic in the context of a Vc LPS conjugate regardless of the form of LPS or the additional components of the inocula. The lowest secondary response to BSA was that of group 11 that was significantly different ( P < 0.001) compared with group 9 that received the same conjugate but additional adjuvant in the form of RIBI. Of note, this difference did not translate to large differences in the anti-LPS response or the vibriocidal response. The secondary responses of BSA-specific IgG suggest that initial differences in BSA dose per immunization group can be overcome with a booster.

View this table:
Table 2

Different immunogen configurations and adjuvants

GroupLPS form carrierCarbohydrate inoculated (μg)Protein inoculatedAdjuvantLinker
1DetAcLPS-TcpA5.4; 2.720 μg; 10 μgRIBIADH
2DetAcLPS-TcpA5.4; 2.720 μg; 10 μgNoneADH
3DetAcLPS-BSA (VcCF)20; 1039 μg;19.5 μg
15 μg; 7.5 μg
4pmLPS-BSA (VcCF)20; 1018.6 μg;9.3 μg
15 μg; 7.5 μg
5DetAcLPS-BSA20; 1039 μg; 19.5 μgRIBIADH
6DetAcLPS-BSA20; 1039 μg; 19.5 μgRIBI; 0.5 μg wt LPSADH
7pmLPS-BSA20; 1018.6 μg; 9.3 μgRIBIMaleimide/thiol
8pmLPS-BSA20; 1018.6 μg; 9.3 μgRIBI; 0.5 μg wt LPSMaleimide/thiol
9wt LPS-BSA8.3; 4.1522.7 μg; 11.35 μgRIBIADH
10wt LPS-CBP-A20; 1046.2 μg; 23.1 μgRIBIADH
11wt LPS-BSA8.3; 4.1522.7 μg; 11.35 μgnoneADH
12wt LPS-CBP-A20; 1046.2 μg; 23.1 μgnoneADH
13pmLPS-(sm)-CBP-A20; 1012.8 μg; 6.4 μgRIBIThiol/maleimide
14DetAcLPS-CBP-A20; 1018.8 μg; 9.4 μgRIBIADH
15pmLPS-(ms)-CBP-A20; 108 μg; 4 μgRIBIMaleimide/thiol

CBP-A serum endpoint titers in response to carrier- and noncarrier-bound CBP-A

The primary anti-CBP-A IgM response to CBP-A as part of a conjugate (groups 10 and 12–15) was able to induce mean endpoint titers at or near 1000 (Fig. 7c, panel A, open symbols). The exceptions were groups 4 and 13 that received pmLPS-(sm)-CBP-A. Groups 4 and 13 did not show quantitative anti-CBP-A response in the primary response, but had titers that were comparable with many in the group with some comparisons being significant, for example, group 13 vs. group 15, P < 0.01 (Fig. 7c, panel A, open symbols). A booster immunization normalized the anti-CBP-A IgM response at a mean of ≈1:1000 (Fig. 7c, panel A, closed symbols). The delivery of CBP-A not as a conjugate, but as a part of the three VcCF inocula (groups 3 and 4) induced similar responses to CBP-A presented in the form of a conjugate (Fig. 7c, panel A).

The distribution (ranking) of mean anti-CBP-A IgG endpoint titers in primary sera was essentially that of the IgM mean responses (Fig. 7c, panel B, open symbols). The lowest titers (statistically significant, P < 0.001) were represented by groups that received nonconjugate-bound CBP-A (groups 3 and 4) and group 13 that used the sm linker. Depending on the comparison, the P values ranged from < 0.01 to < 0.001. There was no significant difference in mean IgG endpoint titer between groups 3, 4, and 13 for primary sera. The explanation for the differences in anti-CBP-A response groups immunized with CBP-A associated with haptenic groups compared with the response mice in the groups that received the three VcCF may be related to the dose of CBP-A, which was 25% higher in the conjugate groups (Table 2). The form of LPS associated with the CBP-A conjugates did not have a consistent effect on the primary mean endpoint titers. The secondary serum endpoint titer means of the various groups that had anti-CBP-A IgG endpoint titers ranged from 1 : 1 000 000 to 1 : 10 000 000, and the differences among the means were less than those of the primary anti-CBP-A IgG response (Fig. 7c, panel B, closed symbols). Some groups were significantly higher (group 10 vs. group 12, P < 0.01), but many means were statistically the same (groups 13–15, P > 0.05). Yet, groups 13–15 had very different anti-LPS responses and vibriocidal responses that seemed to track with the form of LPS, pmLPS, and the linker rather than the carrier.

Conjugates induce antibodies that variably protect infant mice from virulent Vc

TcpA-based conjugates and inocula

We undertook these studies based on the hypothesis that a conjugate subunit cholera vaccine composed of LPS, and a protein VcCF would be superior to a conjugate with Vc LPS epitopes and a standard carrier. We tested the sera of groups that had received TcpA as part of the inoculation whether in the form of the carrier or the inclusion of TcpA in the set of three VcCF (TcpA, TcpF, and CBP-A) using AKI-grown Vc. The protective capacity of anti-TcpA antibodies was only evident for secondary sera (compare Fig. 8a and b). The inclusion of the three VcCF did not enhance the protection in the primary or secondary sera from mice immunized with DetAcLPS- or pmLPS-based conjugates (Fig. 8a and b, compare open and closed squares).

Figure 8

Protective efficacy of sera from mice immunized with TcpA-based conjugates or nonconjugate-based TcpA. TcpA is a component of TCP, a type IV pilus required for colonization and thus the pathogenesis of Vc. The expression of the tcpA operon is maximized by growing Vc in AKI culture. Pooled sera from individual mice in each group were used for all protection assays. Primary sera protection results are shown in ‘a’, while ‘b’ shows the secondary sera's protection. Primary sera were not protective in this study likely due to the lower virulence of the inocula. However, secondary sera, especially from groups immunized with pmLPS conjugates, were highly protective. (secondary pmLPS-BSA/RIBI vs. PB, P = 0.0064; secondary pmLPS-BSA/RIBI + three VcCF vs. PB, P = 0.0188; secondary DetAcLPS-TcpA/no adjuvant vs. PB, P = 0.1322).

After a booster of the various immunogens (Table 2, groups 1–4), secondary sera provided protection for pmLPS-based immunogens (Fig. 8b, open and filled inverted triangles, group 4, P = 0.0188, and group 7 P = 0.0064). The group immunized with DetAcLPS-TcpA conjugate showed partial protection (open squares), but the survival curve was not significant compared with the PB curve ( P = 0.1322). Group 4, inoculated with DetAcLPS-BSA and the three VcCF, showed no protection compared with PB (open circles). The lack of adjuvant in the DetAcLPS-TcpA-immunized mice increased the number of mice that did not survive, indicating some operative effect of adjuvant and DetAcLPS-TcpA (compare open squares vs. closed squares). We assessed the sera generated against TcpA conjugates in a challenge model that is more suited to assess the anti-LPS antibody response rather than the anti-TcpA protective response (Muse et al., 2012). Neither group 1 nor group 2 that were immunized with DetAcLPS-TcpA conjugates induced protective anti-LPS antibody (data not shown) as is consistent with the lack of a vibriocidal response (Fig. 6c).

BSA conjugates based on pmLPS

TcpA is not a particularly strong immunogen for mice (Muse et al., 2012). While it does induce protective antibodies at higher doses, we wondered whether a more stable protein that has previously been shown to provide for good anti-O-SP antibody responses to Ogawa or Inaba synthetic hexasaccharides would increase the ‘helper’ function of the carrier and thus increase the anti-LPS antibody response (Chernyak et al., 2002; Meeks et al., 2004). As pmLPS proved to be a superior form of detoxified LPS for conjugate-induced antibody responses, we assessed antisera made to conjugate with pmLPS as the hapten (Grandjean et al., 2009). Holding the carrier and form of LPS constant, we asked whether the addition of low levels of wt Inaba LPS in the inocula or the inclusion of the three VcCF would enhance the protective response (Fig. 9a and b). In this experiment that analyzed primary sera, the virulence of the Vc inocula results in 50% mortality, and as such, the survival of neonates passively immunized with known protective anti-Inaba LPS antibody ( P = 0.1721) or with sera from mice immunized with pmLPS-BSA/RIBI ( P = 0.1721), pmLPS-BSA/RIBI with wt LPS ( P = 0.2139), or pmLPS-BSA/and the three VcCF ( P = 0.2366) was not statistically different from the PB protection (Fig. 9a). A booster immunization resulted in the experimental groups and the positive control being significantly different from the survival of the PB-treated mice (Fig. 9b, the group with the lowest percent survival, 8, P < 0.001 compared with PB).

Figure 9

Protection data for BSA-based conjugates. These studies did not investigate the role of anti-TcpA in protective immunity and thus used LB-grown Vc (Muse et al., 2012). (a) Primary sera and (b) secondary sera. Sera were pooled from group members and used at a 1 : 10 dilution with about 10 LD50. Primary sera of mice immunized with pmLPS-BSA conjugates with either adjuvants (RIBI and wt LPS) were protective at about 80%. The inclusion of additional colonization factors reduced the protective effect. The delta for protection between the positive control sera and the negative control was only 30%, and thus, none of the differences in percent protection were significant. This is in contrast to the secondary sera, which was uniformly protective compared with PB sera. Sera from group 4 pmLPS-BSA and the 3 colonization factors protected 100% of the mice, which was not significantly different from the other groups that had protective sera. Groups 4 and 7 had 100%, group 8 had 80% (+ wt LPS).

wt Inaba LPS-based conjugates

We reported that priming mice with low doses of wt Inaba LPS before their immunization with Inaba hexasaccharide-BSA conjugates increased the isolation of protective mAbs from immunized mice (Wade & Wade, 2008). The hypothesis we wished to explore is whether conjugates designed with wt LPS might still retain some mitogen aspects of native LPS and thus coupled with a protein carrier may be partially both a T-independent and T-dependent immunogen. We would expect this to be a very robust humoral antibody response with higher levels of attending vibriocidal and protective antibodies and higher levels of IgG.

There was no statistically significant protection provided by primary sera from groups 9–12 (Fig. 10a). As in the other primary challenge (Fig. 9a), the positive control protection also failed to reach significance as it was included in the same experiment. A booster immunization induced significant protective efficacy for groups that received the conjugate in RIBI [Fig. 9b, groups 9 ( P = 0.0040) and 10 ( P = 0.0177) and for the CBP-A conjugate without RIBI (group 12, P = 0.0040)]. The partial protection afforded by the wt LPS-BSA sera (group 11) that did not receive the inocula in adjuvant was not significant compared with PB ( P = 0.4403).

Figure 10

Protection data for conjugates with wt LPS as hapten conjugated to either BSA or CBP-A. These studies do not depend on induction of TcpA and also used LB-grown Vc (Muse et al., 2012). (a) Primary sera and (b) Secondary sera. Sera were pooled from group members and used at a 1 : 10 dilution and mixed with between approximately 6–12LD50. As in the other protection data presented, the virulence of the inocula for the challenge studies that use primary sera was not as high as it could be. Thus, the protective trends that showed adjuvant and conjugates were superior to conjugate alone did not have a large enough delta to be different from the protection provided by the PB. The other trend, while not significant, was the higher percent protection of CBP-A conjugates over BSA conjugates. (b) The protective trend was, however, supported by statistical analysis of the protective capacity of secondary sera. The only nonprotective conjugate was the wt LPS-BSA conjugate without adjuvant.

CBP-A-based conjugates with different forms of detoxified LPS and linkages for carrier attachment

The VcCF, CBP-A, has been shown to induce antibodies in rabbits that are protective for infant mice challenged with classical Vc (Kirn & Taylor, 2005). We were not able to show consistent protective effects of anti-CBP-A (El Tor) antisera (Muse et al., 2012). However, because CBP-A is such a potent immunogen, inducing primary and secondary titers in the millions and tens of millions, respectively, we examined its ability to enhance the anti-Inaba LPS response induced by pmLPS that emerged as a very viable cholera vaccine immunogen. Previous work by one of us showed that neo-epitopes can be introduced by linking the carrier to the pmLPS structures (Grandjean et al., 2009); therefore, it was important to also test the role of the linker in conjugate immunogenicity as immunodominant linker responses are not desirable. The three CBP-A-based conjugates (groups 13–15) allowed us to test the effect of different linkers as well as the form of LPS (DetAcLPS vs. pmLPS) on the generation of antibodies to conjugates with the same carrier (Fig. 11a and b).

Figure 11

pmLPS is a superior immunogen for providing protective antibody compared with CBP-A-detoxified LPS conjugates. A direct comparison of the effectiveness of DetAc-LPS vs. pmLPS was assessed using CBP-A as the carrier. The influence of the linker used to attach the carrier to pmLPS was also examined. Vc were grown in LB pH 6.5 (Muse et al., 2012). (a) Primary sera and (b) Secondary sera. Pooled group sera were used at a 1 : 10 dilution for both primary and secondary sera that were combined with 12 LD50 or 6 LD50 for the primary and secondary sera challenge, respectively. The trend continued with the secondary sera, but the DetAcLPS-CBP-A performed less robustly than the pmLPS-induced sera. PB for primary was only 50%, the DetAcLPS-CBP-A group was 0% survival.

The protective capacity of primary sera of the different groups (groups 13–15) indicates that none of the sera provide statistically significant protection (Fig. 11a). The reason for the increased apparent virulence of mice gavaged with Vc and DetAcLPS-CBP-A antisera is not known, but has been noted before (Muse et al., 2012). As in the other experiments where the primary sera were not protective, a booster demonstrates that the pmLPS-based CBP-A conjugate secondary sera could protect (Fig. 11b). Graphically, the lowest protection (78%) afforded was by sera from mice immunized with DetAc-CBP-A/RIBI [protective compared with PB ( P = 0.0040)]. Group 14 did not have a measurable vibriocidal endpoint titer suggesting that the protection in secondary sera may be due to anti-CBP-A antibodies or a combination of anti-CBP-A antibodies and nonmeasurable vibriocidal antibody that nonetheless bound LPS in vivo. The impact of the carrier CBP-A was noticeable as DetAcLPS-CBP-A secondary sera were partially protective, while DetAcLPS-BSA-induced secondary sera were not (data not shown).


The relevance of the form of detoxified LPS and route of conjugate inoculation

We examined the forms of detoxified LPS, the type of carrier, and the use of adjuvants to determine how they modulated induction of protective anti-VcLPS antibody responses to Vc LPS conjugates. Our group has shown that i.p. immunization can effectively target B cells with low doses of LPS; therefore, we used i.p. immunization rather than s.c. immunization reported earlier for Vc LPS conjugates (Gupta et al., 1992; Grandjean et al., 2009). The differences in the details of the metrics we used are highlighted in the results, but in sum, adjuvant, pmLPS, and i.p. inoculation with a CBP-A carrier would be considered the optimal parameters for vibriocidal and protective responses to Vc LPS conjugates. On average, the vibriocidal responses after an equal number of immunizations were about a log higher than those reported by Gupta or Grandjean (Gupta et al., 1992; Grandjean et al., 2009). The higher titer antisera were protective against virulent Vc challenge, a component that was not part of the data set reported by either Gupta or Grandjean.

Are differences in protein and carbohydrate amounts between conjugate confounding?

The generation of conjugates for these studies was carried out over a period of time (16 months), and changes were made based on accruing results. A difficulty in making conjugates is that they cannot all be made to the exact specification with respect to concentration of carbohydrate and protein by an academic laboratory with limited resources. The range of carbohydrate in the primary inoculations was 5.4 μg to 20 μg, the latter of which was the mode (Table 2). The lowest anti-LPS serum endpoint titers were associated with DetAcLPS-TcpA conjugates (5.4 μg and 2.7 μg of carbohydrate for primary and secondary inoculations, respectively), but the increase to 20 μg of carbohydrate in DetAcLPS-BSA did not rescue the anti-LPS serum response. This inference is supported by our earlier studies of Ogawa hexasaccharides that showed the small relative differences in the amount of carbohydrate are not a major controlling factor in the humoral response to Vc LPS epitopes (Saksena et al., 2005).

Protein concentration for the first dose ranged from 8 μg to 46 μg with the booster dose being halved. The mean concentration for protein in conjugates was 26.0 μg. The results from groups 13–15 suggest the dose of CBP-A was not an issue in vibriocidal responses comparing pmLPS-(sm)-CBP-A and DetAcLPS-CBP-A, where DetAcLPS-CBP-A-immunized mice had a 47% higher dose of protein. The use of even higher doses (2.5- to 5.8-fold more) of CBP-A for the wt LPS conjugates did not increase the serum endpoint responses to CBP-A ELISA titers, suggesting the dose of protein carriers we used was not limiting.

Role of i.p. inoculation

We did not directly compare s.c. vs. i.p. inoculations because there was not enough reagents for the additional experiments. However, inbred BALB/c mice have consistent responses to the Vc LPS. This and the fact that pmLPS is better than DetAcLPS given i.p. (this study) and also poorly immunogenic if given via the s.c. route suggests that the route of inoculation is important. Intraperitoneal inocula can track to the spleen, thereby interacting with multiple B-cell subsets in the peritoneum and the spleen (Wade, 2011). We conjecture that in our earlier studies using purified LPS, marginal zone B cells were part of the response to bacterial LPS (Oliver et al., 1997). The role of murine B1 B cells in an anti-Vc LPS antibody response is not known. How the three subsets of B cells are primed by pmLPS conjugates needs to be determined. Marginal zone B cells can bind LPS, and they also can present MHC class II–associated peptides to interact with T cells.

A report by Bystricky's group concluded an i.p. booster of Vc LPS conjugates was more effective than an s.c. booster (Paulovicová et al., 2010). Paulovicova's study reported that O-SP-BSA (our pmLPS-BSA) was more effective than DeOAc-LPS-BSA (our DetAcLPS) at inducing a T-helper response. Bystricky's laboratory uses conjugates that are a glucan matrix to which the detoxified LPS and carrier are attached. How this change influences serum endpoint titers with conjugates that also differ in the haptenic groups is not known. It is clear that a glucan matrix is not required for a robust anti-LPS response to pmLPS conjugates.

Role of adjuvant in Vc LPS conjugates

Adjuvants are used to enhance antigen uptake and presentation of epitopes. We assessed the role of adjuvant in anti-LPS conjugate responses. In the case of T-dependent antigens, adjuvants increase isotype switching and somatic mutations, thereby increasing the affinity of the antibody. Adjuvant is required for isotype switching and for enhanced vibriocidal and protective responses to pmLPS conjugates. Adjuvants are sometimes argued to complicate the time and costs of large-scale vaccine production (time and cost). Based on our study, adjuvants are beneficial for the VC-based conjugates for isotype switching and should not be ruled out if the switching is associated with enhanced B-cell memory as both involve follicular interactions with antigen-specific T cells.

Do the carriers for LPS conjugates need to be Vc-protective antigens?

We used two VcCF as carriers. One, TcpA, is a proven protective antigen and the other CBP-A (El Tor) in our hands is variable in its protective responses. A recent publication described the protective potential of a chimeric TcpF-CT vaccine. The induction of antibodies to protective B-cell epitopes was compromised (Price & Holmes, 2012). This report and our recent publication (Muse et al., 2012) suggest that while it is intellectually attractive to combine multiple protective epitopes into one immunogen, it will likely be problematic and not necessary. If not VcCF carrier to deliver pmLPS, what carriers would suffice? Carriers approved by the FDA or other world health organizations and that have been shown to be safe and effective include tetanus toxoid, the mutant diphtheria toxin protein, CRM197, gonococcal outer membrane proteins, and CT (Costantino et al., 2011). It is clear that empirical testing of carriers will be required, and this should be carried out in humans so as not to encounter confounding issues of murine repertoire differences or B-cell activation differences. Multi-component conjugate vaccines have been noted to alter the responses to other immunogens resulting in lower serum response, termed carrier-induced epitopic suppression (Borrow et al., 2011). This may argue for CBP-A as a carrier rather than already approved carriers.

Protective B-cell epitopes on LPS

Serotype-specific B-cell epitopes have been identified for both Inaba and Ogawa LPS and an epitope(s) common to both serotypes (Wang et al., 1998; Villeneuve et al., 2000; Ahmed et al., 2008; Wade & Wade, 2008). Some Vc LPS-protective mAbs (S-20-4, IgG1; MD4, IgA) bind the terminal sugar of Ogawa LPS (Villeneuve et al., 2000). Another mAb (I-24-2) bound both Inaba and Ogawa LPS via an epitope located in the area of the convergence of the core and O-SP domains (Wang et al., 1998). One more common epitope found on Inaba and Ogawa LPS is the tetronic acid substitution on the individual perosamines (Ahmed et al., 2008).

After the generation of pmLPS conjugates, a panel of Vc-specific, protective mAbs should be used to characterize the change in binding per weight for wt LPS and pmLPS that are conjugated to a carrier. The resulting human anti-LPS sera made after phase I trials can be assessed for the percent and distribution of LPS B-cell epitope reactivity compared with cholera patients with naturally induced anti-LPS antibodies. This analytical adjunct will backstop the development of conjugates for human testing based on conjugates that retain and induce the best protective LPS B-cell responses.

Why do linkers matter?

Choice of the coupling chemistry for conjugates and the nature of the resulting spacer arm can have a profound impact on the conjugate structure(s) that affects the immunogenicity of the LPS epitopes. In this study, DetAcLPS was derivatized as reported by Gupta et al. (1992), using the CNBr/ADH activation procedure that resulted in the most effective Inaba LPS conjugates. This procedure has been applied to the preparation of long capsular polysaccharide (Jones, 2005) as well as other LPS conjugates (Polotsky et al., 1994; Gupta et al., 1995). However, considering that virtually any accessible nucleophile site (OH, NH2…) can be activated and that the length of Inaba LPS is shorter, the probability needs to be entertained that the protective antigenic determinant can be modified during the conjugate preparation and thus reduce immunogenicity.

An alternate coupling strategy based on the selective activation of the DetAcLPS amine groups gave disappointing results according to Gupta (Gupta et al., 1992). For this reason, the difficulty to drive the derivatization to completion, and our incapability to remove any unbound DetAcLPS from the conjugate mixture, we did not pursue modification in conjugate based on DetAcLPS but rather with pmLPS. One of us showed in a slightly different system that there was a confounding feature of adding new structures to the Inaba LPS, that is, the linker could serve as an epitope that diverts antibody response to nonprotective immunogen epitopes (Grandjean et al., 2009). In this study, we report the result based on a 10 bond length linker (resulting from the addition of thiol onto 3-maleimidopropionoyl tails) that is amenable for use in humans, but a structure that is less accessible and therefore likely less immunogenic (Verez-Bencomo et al., 2004). Installation of the spacer arm relies on the selective coupling between a maleimide and a thiol group. We report that linkers differing in their orientation (i.e. whether the maleimide or the thiol is introduced on the carrier), given the caveat about carrier dose (60% higher for sm-based conjugates), can influence the level of vibriocidal antibody. Further analyses beyond the scope of this study will be necessary to explain the mechanism that causes this linker effect. Perhaps underscoring the dynamics of the individual immunogens are Boons and co-workers data from a cancer vaccine model wherein ‘ms’ were superior to ‘sm’ conjugates. They reported that unreacted or hydrolyzed maleimide groups remaining on the carrier are the source of competitive epitopes (Buskas et al., 2004). However, such antimaleimide responses are not systematic. They were not observed by one of us in the context of conjugate vaccine for shigellosis (Phalipon et al., 2009).

Xu and co-workers used squaric acid chemistry to link BSA to either Inaba- or Ogawa-O-SP (pmLPS) (Xu et al., 2011). They did not test the conjugates for immunogenicity, rather showed the conjugates were bound by sera from patients with cholera at various stages of recovery. The next-generation pmLPS constructs will have to be tested in humans sooner, rather than later, to ensure immunogenicity of protective LPS immune response. Understanding the structural role of linkers needs to be appreciated for future cholera conjugate vaccine development. These roles range from the prospective of adding new epitopes that are immunodominant, modifying existing protective epitopes, and leaving reactive groups in the semi-purified immunogen that preclude efficient priming due to antigenic competition with Vc-protective antigens.

Why cholera vaccine conjugates?

A recent review by Harris and co-workers noted 11 references that supported their conclusion that, ‘…vaccines (two oral kW-C doses) provide 60–85% protective efficacy for 2–3 years, although protection among young children is of shorter duration’ (Harris et al., 2012). A successful cholera vaccine is now more often discussed as one that induces better and more durable immunity in young children (Wade, 2011). Robbins et al. (1995) conceptualization that IgG antibodies can contribute to protective immunity against cholera was perhaps ill-fated given the extensive push to prove OCV were the only viable approach for cholera vaccines. Yet early cholera vaccine field trials showed parenteral vaccination could protect individuals against cholera with IgG being the main isotype of antibody responding in the main to Vc LPS epitopes (Benenson et al., 1968a,b; Mosley et al., 1970). A meta-analysis of the vaccine literature substantiated that there was indeed better protection available from vaccine protocols that largely induced systemic IgG (Graves et al., 2011). The most recent evidence of the potential role of anti-LPS IgG came from the Harvard/Bangladesh cholera group that reported the predictive element of protection for household associates of cholera cases were those with IgG B-cell memory specific for Vc LPS (Patel et al., 2012). The role of IgG in protection against cholera may be more important than that previously agreed upon, and if an IgG response can be leveraged to enhance oral vaccination to provide durable cholera immunity, especially to children, it is worth pursuing.

Vc LPS conjugate vaccines – then and now

One reason protection may wane for young children immunized with kW-C cholera vaccines is the lack of robust recruitment of B-cell memory to either LPS or other unknown Vc antigens (Patel et al., 2012). To address this problem, several groups constructed Vc LPS-based conjugate vaccines (Kabir,1987; Gupta et al., 1992; Gupta et al., 1998; Chernyak et al., 2002; Meeks et al., 2004; Grandjean et al., 2009). An early conclusion based on this literature, detoxified LPS, pmLPS and DetAcLPS, were poorly immunogenic by themselves for BALB/c mice (Gupta et al., 1992; Grandjean et al., 2009). Two doses of 2.5 μg of DeA-LPS-CTI or II conjugates given subcutaneously (s.c.) to BALB/c mice generally induced low-level Inaba-reactive vibriocidal antibody in primary and secondary sera. The exception was a titer of 5000 to DeA-LPS-CTII that was derivatized with ADH following random CNBr activation, and thus, the potential for multiple attachment sites on the saccharide and protein was possible (Gupta et al., 1992). ELISA of serum Inaba LPS-specific antibody did not show IgG in primary or secondary sera, and only low levels of Inaba-specific IgM was measured (geometric means for CTI and CTII ranged from 40–70, average 45). In contrast, whole-cell Vc induced higher IgM titers (106 and 1114, respectively, for the primary and secondary sera). The conjugate booster increased serum anti-Inaba LPS IgG, but no primary IgG was measurable. The use of alum, a weak adjuvant, did not change the reported immunogenicity of the conjugates (Gupta et al., 1992). The lack of a robust serum IgG in response to DeAc-LPS-CT conjugates compared with that we report for the pmLPS-based conjugates is a major contrast between our results and those of Gupta et al. (1992).

pmLPS conjugated to either tetanus toxoid or BSA were injected s.c. (10–20 μg of carbohydrate conjugate immunogens) into BALB/c mice (Grandjean et al., 2009). Low vibriocidal titers (≈1 : 40) but no class IgG switching in the anti-LPS antibody was observed. The pmLPS used in our study was provided in a similar dose to that of Grandjean and co-workers, but importantly we used RIBI and inoculated the experimental vaccine preparations i.p. These two changes had a large impact on ELISA and vibriocidal endpoint titers as well as class switching.

The route of inoculation (s.c. vs. i.p), the adjuvant choice (alum vs. RIBI), and the type of detoxified LPS (DetAcLPS vs. pmLPS) used to make the conjugates are a notable variable between our current report and the 1992 report by Gupta. Collectively, the increase in carbohydrate, the switch to pmLPS, and the i.p. inoculation of the conjugate are associated with better IgG responses and higher vibriocidal responses. Next-generation animal studies could examine more conjugate parameters in detail, but it may be wiser and more efficient to first determine whether the Vc pmLPS conjugates are immunogenic in humans. Subcutaneous dosing should be compared with i.n., which is a mucosal route that has the potential to induce marginal zone B cells, which we feel are critical to anti-LPS responses (WF Wade personal observation). If anti-LPS IgG B cells play a role in protecting against cholera, the alternate routes of inoculation could supplement the OCV. If OCV and a pmLPS conjugate are provided simultaneously, additive or synergistic establishment of plasma cells and memory B cells could rescue the B-cell memory generation that is absent in young children. A major advantage of a conjugate cholera vaccine may be the entailment of a cholera vaccine protocol into established childhood vaccine programs. These programs already include conjugate vaccines and are supported by government and nongovernmental organization, even though they require multiple doses. If it is not possible to produce a one-dose cholera vaccine for young children, it might be possible to use pmLPS conjugates to establish cholera immunity and memory in young children based on an accepted vaccine schedule for childhood diseases for which conjugates are the accepted immunization vehicle.


DetAcLPS with its partial acylation is an inferior immunogen compared with pmLPS that is completely deacylated. Several carriers can serve with pmLPS, but an advantage might be with CBP-A that was able to induce protective and vibriocidal antibodies without adjuvant. This will be especially useful if CBP-A can be shown to be a protective antigen in humans. Adjuvant was a factor in antibody responses especially the switch to IgG. The type of linker used to attach the carrier moiety to Inaba-O-SP-core structures was important with sm linkers outperforming ms. There is a limited but varied potential set of linkers, and it will be important to know whether the human immune system has a preference. The maturation of Vc cholera conjugates is at a point where phase 1 trials can be considered for safety, as well as the efficacy of the resulting antisera. Demonstrating serum protection and enhanced long-term B-cell memory in humans should be pursued with the current generation of pmLPS-CBP-A-based conjugates.


View Abstract