IL-7 Enables Antibody Responses to Bacterial Polysaccharides by Promoting B Cell Receptor Diversity

Compared with adults, young children exhibit increased susceptibility to a variety of bacterial pathogens that express capsular polysaccharides. Salmonella enterica serovar Typhi (S. Typhi), the causative agent of typhoid fever in humans, is one such pathogen. Global estimates indicate that 21.6 million cases of typhoid fever occur each year, resulting in 226,000 deaths (1). Vaccination of high-risk populations, such as children in disease-endemic areas, is considered the most promising strategy for prevention (2, 3). Two types of licensed vaccines are currently available: a live attenuated vaccine that is not safe for use in young children and a subunit vaccine (Typhim Vi) composed of purified Vi polysaccharide (ViPS), a well-known virulence factor expressed by S. Typhi. A single dose of ViPS vaccine induces a robust Ab response in adults (3). However, like other purified bacterial polysaccharide vaccines, such as the 23-valent pneumococcal polysaccharide (PPS) vaccine Pneumovax 23, ViPS does not induce an optimal response in young children.

Bacterial polysaccharides conjugated to protein carriers are referred to as conjugate vaccines. These vaccines can induce polysaccharide-specific Ab responses in young children. Conjugate vaccines induce T cell–dependent B cell responses and require as many as four doses administered 2–6 mo apart in infants and young children for generating optimal levels of protective Ab (4, 5). In contrast, a single dose of an unconjugated polysaccharide vaccine such as Typhim Vi or Pneumovax 23 elicits a rapid T cell–independent Ab response in older children and adults. Although one ViPS conjugate vaccine has been reported to be effective in a clinical trial in Vietnam (6), another experimental ViPS conjugate vaccine as well as a PPS conjugate vaccine 13 did not yield any increase in Ab titers in a multinational clinical trial, despite the periodic boosting that is commonly employed with conjugate vaccines (7). The public health threat posed by S. Typhi, the emergence of other serotypes of pneumococcal strains not covered by existing vaccines, and the vulnerability of young children to infections due to these pathogens make understanding the mechanisms involved in Ab responses to bacterial polysaccharides in young children of great importance.

The mature B cell population in mice is composed of four subsets: follicular (FO; also referred to as B2), marginal zone (MZ), B1a, and B1b cells (8, 9). These subsets are developmentally, phenotypically, and functionally distinct. Two cytokines, Flt3 ligand (also known as Flk2 ligand) and IL-7, contribute to murine B lymphopoeisis early and later in life, respectively (10). IL-7 is essential for production of human B cells from adult bone marrow, and the dependency of B lymphopoiesis on IL-7 becomes progressively more profound during ontogeny (11, 12). Thus, it appears that the degree of IL-7 dependency in B cell development varies between early and later life in both mice and humans.

B lymphopoeisis early in life is largely IL-7 independent, and young mice do not generate efficient responses to bacterial polysaccharides. In contrast, adults generate B cells predominantly through IL-7–dependent B cell lymphopoeisis, and adult mice can generate efficient protective Ab responses to bacterial polysaccharide Ags. This suggests that the B cells generated early in life in the absence of IL-7 are qualitatively different from those that develop in IL-7–sufficient adults (13). In mice, the B1b cell subset has been shown to generate the bulk of Ab responses to several bacterial polysaccharides, namely, PPS serotype 3, α1-3 dextran, and ViPS (14–16). We have previously found that despite having B1b cells as well as B1a and MZ B cells (13), young wildtype mice or adult mice deficient in either IL-7 or IL-7Rα are severely impaired in mounting Ab responses to polysaccharide Ags (13).

B cells from young mice exhibit a biased V_H gene usage favoring the expression of D_H-proximal V_H gene families such as V_H5 (also referred to as V_H7183) (17, 18). In striking contrast, B cells from adult mice have extensive usage of several D_H-distal V_H gene families such as V_H1 (also referred to as V_HJ558) (17, 18). Interestingly, similar to young wildtype mice, a biased pattern of V_H gene usage favoring D_H-proximal V_H gene families is evident in mice deficient in IL-7 or IL-7Rα (19–22). This indicates that IL-7–mediated signaling is required for the generation of complete diversification of the BCR repertoire (19–22). Collectively, these observations led us to hypothesize that IL-7–mediated BCR diversity generated in the preimmune repertoire is one of the critical factors required for B cell responses to a variety of Ags, including bacterial polysaccharides.

The Institutional Animal Care and Use Committee approved these studies. Mice were housed in microisolator cages with free access to food and water and were maintained in a specific pathogen-free facility. C57BL/6J and BALB/cJ were purchased from The Jackson Laboratory (Bar Harbor, ME). Mice deficient in IL-7 (IL-7^−/−) (23) on the C57BL/6 background were obtained from Dr. Crystal Mackall’s colony at the National Cancer Institute, National Institutes of Health. IL-7 transgenic (Tg) mice (IL-7 Tg) on the C57BL6 background (24) were originally generated by Dr. R. Ceredig (INSERM, Strasbourg, France) and obtained from Dr. A. Feeney (The Scripps Research Institute, La Jolla, CA). V_HJ558 Tg mice were previously generated from a V_HJ558.3-DSP2.2-J_H1 VDJ rearrangement amplified from genomic DNA derived from a BALB/c hybridoma and were bred onto the C57BL/6 background (15). We have crossed the V_HJ558 Tg mice with IL-7^−/− mice to generate V_HJ558 Tg × IL7^−/− mice, and their genotypes were confirmed by PCR. Three-week-old mice were considered young and 8-wk-old or older mice were considered adult.

Two-and-a-half micrograms of ViPS (Typhim Vi; Sanofi Pasteur, Swiftwater, PA, or purified ViPS; Lot 5, U.S. Food and Drug Administration) dissolved in 100 μl of Dulbecco’s PBS (DPBS; Mediatech, Herndon, VA) was used to immunize mice i.p. Previously, immunization with ViPS in the range of 0.25–50 μg in 100 μl of PBS was shown to induce comparable anti-ViPS Ab response (25, 26). For dextran immunizations, 50 μg of α1-6 dextran (B512; MP Biomedicals, Solon, OH) or 50 μg of α1-3 dextran (B1355; a gift from Dr. A. Jeanes) dissolved in 100 μl of DPBS was used to immunize mice i.p. For whole bacterial immunization, mice were injected i.p. with 3 × 10⁸ heat-killed Escherichia coli strain W3110 pDC5, which expresses ViPS (27), or 10⁸ heat-killed, paraformaldehyde-fixed Enterobacter cloacae strain MK7, which expresses α1-3 dextran. Blood samples were obtained 0, 7, 14, 21, and 28 d following immunization.

ViPS-specific IgM and IgG were measured by incubating 96-well microtiter plates (Nunc MultiSorp 467340; Thermo Fisher Scientific Nunc A/S, Roskilde, Denmark) with 2 μg/ml ViPS in DPBS overnight at room temperature. All plates were washed and blocked with 2% BSA in PBS (pH 7.2) for 2 h at room temperature. Blood samples from immunized mice were diluted to 1:25 for IgG detection and 1:50 for IgM detection in blocking buffer, samples were centrifuged (800 × g for 10 min), and cell-free supernatant was used. Bound IgM or IgG was measured using HRP-conjugated goat anti-mouse IgM or IgG (Bethyl Laboratories, Montgomery, TX). Because ViPS-specific mouse IgM and IgG reference standards are not available, and the ViPS-specific Abs in mice are likely to be of oligoclonal nature with various affinities, the Ag-specific Ab levels in the current study were interpreted as nanogram or microgram per milliliter “equivalents” using normal mouse serum IgM or IgG standards (Bethyl Laboratories) as described previously (13, 28–30). Measurement of dextran-specific IgM responses involved coating 96-well microtiter plates (Costar 9017; Corning, Corning, NY) with 1 or 5 μg/ml B512 dextran (α1-6 dextran) or dextran B1355 (α1-3 dextran) and incubating them overnight at 4°C. Plates were subsequently washed and blocked with 2% BSA in PBS (pH 7.2) for 2 h at room temperature. Blood collected from dextran or E. cloacae strain MK7–immunized mice was diluted to 1:250 or 1:500, centrifuged as above, and applied to ELISA plates for 2 h at room temperature. Bound Ab was measured as above.

Serum bactericidal assay (SBA) was performed as previously described (31). In brief, log-phase cultures (OD₆₀₀ of 0.5 at 37°C) of S. Typhi strain Ty2 were prepared in Luria–Bertani (LB) broth with 10 mM NaCl. Bacterial cells were washed in DPBS, and the bacterial cell density was adjusted to 1–3.5 × 10⁴ CFU/ml in DPBS. Serum samples were heat-inactivated by incubating at 56°C for 30 min prior to use in the assay. Ten microliters of S. Typhi cells in DPBS (100–350 CFU) were added to each well of a round-bottom polypropylene 96-well plate containing 50 μl of heat-inactivated serum in serial dilutions, 12.5 μl of baby rabbit complement (Pel-Freeze, Rogers, AR), and 27.5 μl of DPBS. Triplicate samples of each dilution were incubated for 120 min at 37°C with gentle rocking, and 10 μl of this mixture was plated on LB agar plates for counting. Serum bactericidal Ab titers were defined as the reciprocal of the highest dilution that produced >50% killing in relation to control wells containing complement but no mouse serum.

To test the relative protection conferred by ViPS immunization, mice were infected with a chimeric strain of S. Typhimurium (strain RC60) that expresses S. Typhi genes necessary for ViPS synthesis, export, and regulation as in S. typhi (32). Strain RC60 was grown to an OD⁶⁰⁰ of ∼1.0 in LB broth containing 10 mM NaCl. Bacteria were washed twice in DPBS, and 2 × 10⁴ CFU in 100 μl of DPBS was injected i.p. or i.v. The bacterial burden in the blood, liver, and spleen was measured 3 d later as described previously (30, 33) because mice on BALB/c or C57BL6 background succumb to Salmonella infection by 5 d postinfection at this infectious dose owing to the presence of a susceptible Nramp1 allele (33–35). Tissues were processed using a Minilys tissue homogenizer (Bertin Technologies, Montigny-le-Bretonneux, France), and bacterial burden in these tissue homogenates or in blood was measured by counting CFUs on LB agar plates.

ViPS was labeled by reaction of the polysaccharide uronic acid carboxylates with Alexa Fluor 647 (AF647) hydrazide via hydrazone formation. Ablation of polysaccharide carboxylates did not decrease immunogenicity (26), and this method was successfully used for the synthesis of a polysaccharide conjugate vaccine (36). Lyophilized ViPS (4 mg/ml) was dissolved in the reaction buffer (25 mM MES hydrate and 150 mM NaCl (Sigma-Aldrich, St. Louis, MO) (pH 6.05). ViPS was hydrated by heating to 50°C for 30 min and then incubated at room temperature for 6 h followed by low-speed vortexing. The ViPS solution was then purified for removal of low m.w. contaminants and proteins by centrifugal filtration in a Microcon YM-100 centrifugal filter (Millipore, Billerica, MA). The final volume of the ViPS was adjusted to achieve a concentration of 10 mg/ml. AF647 hydrazide (Life Technologies, Frederick, MD) was prepared in DMSO for a concentration of 10 mg/ml. N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride and 1-hydroxybenzotriazole hydrate reactants were dissolved at a concentration of 10 mg/ml in the reaction buffer immediately before use. To the purified polysaccharide solution, reactants were added (with gentle mixing between each addition) in the order of 1-hydroxybenzotriazole hydrate, Alexa Fluor hydrazide, and N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride to achieve a molar ratio of 2000:1 reactant/polysaccharide for each reactant. This gives a theoretical labeling of ∼1 fluorophore for every 10 saccharide units. Reaction proceeded for 24 h at room temperature on a rotating mixer. Product was purified via two rounds of dialysis (24 h at 4°C), employing a 10-kDa dialysis cassette (Thermo Fisher Scientific, Waltham, MA) in a volume of PBS >4000 times the reaction volume, for a total dilution ratio of 1.6 E7:1. The product was analyzed by ultraviolet spectroscopy at 650 nm, and the concentration of AF647 calculated (AF647 extinction coefficient 239,000 [M*cm]⁻¹). Employing both the concentration of AF647 and the known volume, the polysaccharide labeling was calculated to be ∼1 fluorophore every 12–14 saccharide units.

To determine the frequency of B1a, B1b, and B2 cells, peritoneal cavity cells were harvested from individual mice, and the cell concentration was adjusted to 2.5 × 10⁷ cells/ml in staining medium (MEM Eagle with Earle’s salts and without l-glutamine or phenol red [Corning Cellgro, Manassas, VA] with 3% newborn calf serum [HyClone Laboratories, Logan, UT] and 1 mM EDTA). To determine the plasma cell frequency, spleen cells and bone marrow cells from femur/tibia were harvested. After blocking Fc receptors with 2.4G2 Ab (1 μg per 10⁶ cells), an aliquot of 25 μl of peritoneal cavity cells was incubated in a microtiter plate with appropriately diluted Ab. To determine the frequency of FO B and MZ B cells, 25 μl of spleen cells was incubated with 2.4G2 Ab for 15 min and stained with appropriate Abs. The Abs anti–Mac1-AF700 (clone: M1/70), anti-CD5 Pacific Blue (clone: 53-7.3), IL-7Rα–PE (clone: A7R34), anti-CD19–PE-Cy7 (clone: eBio1D3), and TACI-PE (clone: eBio8F10-3) were purchased from eBioscience (San Diego, CA); anti-B220-PercP-Cy5.5 (clone: RA3-6B2) was purchased from Caltag (Burlingame, CA); and anti–IgM^a-PE (clone: DS-1) was purchased from BD Pharmingen (San Jose, CA). Anti-mouse CD138-BV605 (clone: 281-2) was purchased from BioLegend (San Diego, CA). The α1-6 dextran–FITC (FD2000S) was purchased from Sigma-Aldrich. The α1-3 dextran and the α1-3 dextran anti-Id Ab, named EB3-7, which recognizes α1-3 dextran–specific BCR, was conjugated to Alexa 488 according to the manufacturer’s instructions (Molecular Probes; Invitrogen). After staining, cells were washed twice with staining medium, and the preparations were analyzed on an LSR II flow cytometer (Becton Dickinson, Mountain View, CA) using the FACSDiva software (Becton Dickinson). Data were analyzed using the FlowJo software program (Tree Star, San Carlos, CA).

Adult C57BL/6J mice were immunized with 2.5 μg of Typhim Vi vaccine i.p.; 4 d postimmunization, cells from the peritoneal cavity and spleens were isolated and stained with a mixture of fluorescent Abs as well as ViPS-AF647; and 50,000 or 100,000 ViPS⁺ B cells were directly sorted into RNAprotect Cell Reagent (Qiagen) on a BD FACSAria sorter to >97% purity. RNA from FACS-sorted ViPS⁺ B cells was extracted using RNeasy Micro Kit according to the manufacturer’s instructions (Qiagen). Approximately 30 ng of RNA from each of the samples was amplified with unique molecular indices using iRepertoire’s mouse BCR H chain primers and arm-PCR method as described (37). Briefly, arm-PCR is a nested, multiplex PCR process, which is conducted with a set of sequence-specific primers covering almost all 489 of mouse V_H genes (forward primers) and constant C_H primers covering all of the isotypes (reverse primers). Included in the reverse primers is a six-nucleotide molecular identification and partial adaptor A sequence for paired-end sequencing on the Illumina platform, whereas partial adaptor B is associated with the V gene–specific primers. After RT-PCR, PCR products were rescued using a right side 0.8× selection with SPRIselect beads (Beckman Coulter), including an 80% ethanol washing step. Qiagen TopTaq Master Mix was added directly to each bead pellet, and the resuspended pellet was subjected to a two-step PCR, allowing for completion of the partial adaptor sequences. The ∼400 bp–long PCR products were run on a 2% agarose gel and purified using a QIAquick Gel Extraction Kit (Qiagen). Libraries were pooled and sequenced on the Illumina MiSeq using a v2 500 cycle kit and 20% of the flow cell. The output of IgH sequence covers CDR2, CDR3, and the beginning of the C region.

Data presented throughout depict pooled data from at least two independent experiments unless otherwise noted. Statistics were performed using the Prism 5 software program (GraphPad Software, La Jolla, CA).

S. Typhi is a human-restricted pathogen, and experimental models to test ViPS vaccine efficacy are currently limited (38). S. Typhimurium causes a typhoid-like systemic disease in mice. Therefore, S. Typhimurium infection in mice is widely used as an experimental model to understand certain aspects of human typhoid (39). Investigating the role of ViPS-specific Ab responses using S. Typhimurium is limited by the fact that, unlike S. Typhi, S. Typhimurium does not express ViPS and possesses a longer LPS than S. Typhi. To overcome these limitations, a chimeric strain of S. Typhimurium (RC60) was previously engineered (32). This strain expresses all genes that are necessary for ViPS synthesis, export, and regulation as in S. Typhi, and additionally has a deletion of FepE, which controls the length of LPS. Strain RC60 was shown to exhibit the cell surface characteristics of S. Typhi (32).

Ab-mediated protection against S. Typhimurium can be measured in BALB/cJ mice (40). To test the protection afforded by ViPS immunization, young and adult BALB/cJ mice were immunized with unconjugated ViPS, which elicits a T cell–independent response, and 4 wk later were challenged i.p. with strain RC60. Unlike immunized young BALB/cJ mice, immunized adult BALB/cJ mice generated robust ViPS-specific Ab responses (Fig. 1A) and significantly higher serum bactericidal titers against S. Typhi (Fig. 1B). ViPS-immunized adult BALB/cJ mice but not young mice exhibited an ∼10-fold lower bacterial burden in the blood, liver, and spleen compared with unimmunized adult mice (Fig. 1C). However, the differences between the bacterial burden in immunized young mice and immunized adult mice were not statistically significant (Fig. 1C).

We hypothesized that the inability of young mice to respond efficiently to bacterial polysaccharide is due to a limitation in IL-7–driven B lymphopoiesis early in life. Indeed, ViPS-specific IgM and IgG responses in young wildtype mice or adult IL-7^−/− mice were significantly lower than in adult wildtype mice (Fig. 2A). As early as 1 wk postimmunization, both young wildtype (C57BL/6J) and adult IL-7^−/− mice exhibited significantly lower S. Typhi serum bactericidal titers compared with ViPS-immunized adult wildtype (C57BL/6J) mice (Fig. 2B). Moreover, compared with immunized adult IL-7^−/− mice, immunized adult wildtype mice had significantly reduced bacterial burden in blood, liver, and spleen upon i.v. challenge (Fig. 2C). As observed with the RC60 infection in BALB/cJ mice (Fig. 1C), the differences in the bacterial burden between immunized young C57BL/6J mice and immunized adult C57BL/6J mice were not statistically significant (Fig. 2C).

To examine whether enforced expression of IL-7 promotes ViPS responses in young mice, we immunized 3-wk-old IL-7 Tg mice, which constitutively express IL-7 (24). Compared with young IL-7 Tg⁻ mice (i.e., 3-wk-old wildtype), young IL-7 Tg⁺ littermates generated significantly increased ViPS-specific IgM and IgG responses (Fig. 2D). Furthermore, young IL-7 Tg mice generated significantly higher S. Typhi serum bactericidal titers as compared with young wildtype mice, demonstrating the functionality of Abs generated by young IL-7 Tg mice (Fig. 2E). Collectively, these data indicate that IL-7 is required for the generation of ViPS-specific Ab and that enforced expression of IL-7 permits the generation of more efficient polysaccharide-specific Ab responses in young mice.

Immunization with whole bacteria can activate immunostimulatory pathways, such as TLRs, and also provide diverse antigenic targets, both of which may quantitatively and qualitatively improve Ab responses. As such, to examine whether bacteria-associated ViPS induces an Ab response in IL-7^−/− mice, we immunized adult wildtype and IL-7^−/− mice with heat-killed E. coli–expressing ViPS. ViPS-specific IgM and IgG responses in adult IL-7^−/− mice were impaired compared with adult wildtype mice (Fig. 3A). Additionally, immunized adult IL-7^−/− mice had strikingly lower S. Typhi serum bactericidal titers than adult wildtype mice even at 21 d postimmunization (Fig. 3B). These results indicate that antipolysaccharide responses are not efficiently generated in IL-7^−/− mice even in the context of whole bacteria.

To assess the relative protection conferred by immunization with E. coli–expressing ViPS, we challenged immunized adult wildtype and adult IL7^−/− mice with S. Typhimurium strain RC60. Immunized adult wildtype mice showed an order of magnitude lower bacterial burden in all tissues relative to unimmunized adult wildtype or immunized IL-7^−/− mice (Fig. 3C). Moreover, levels of ViPS-specific IgG inversely correlated with bacterial burden in all three tissues tested (Fig. 3D).

In mice, anti-ViPS Ab responses are primarily generated by B1b cells (16). Because young wildtype mice and adult IL-7^−/− mice have B1b cells (13), the reduced anti-ViPS response in adult IL-7^−/− mice and young wildtype mice is likely attributed to qualitative characteristics of B cells that develop in the absence of IL-7. For example, unlike adult wildtype mice, young wildtype mice (17, 18) as well as adult mice deficient in IL-7Rα do not express a high frequency of V_H genes such as V_H1, which are located distal to D_H genes in the IgH locus (19, 20, 22). In fact, we found that adult wildtype and young IL-7 Tg mice use distal V_H genes more frequently than young wildtype and adult IL-7^−/− mice (Supplemental Fig. 1). To test whether the ability of adult wildtype mice to respond to ViPS is associated with a BCR repertoire using distal V_H gene segments, we FACS-purified ViPS-binding B cells from adult C57BL/6J mice immunized with Typhim Vi, the unconjugated ViPS vaccine (Fig. 4A), and these B cells were subjected to IgH deep sequencing analysis as described (37). As predicted, the majority of ViPS⁺ splenic or PerC B cells of adult wildtype mice expressed IgH sequences containing distal V_H gene segments, the most frequent of which were V_H1-53 and V_H1-55 (Fig. 4B). The most frequent CDR3 peptide sequence (deduced from the nucleotide sequence) associated with V_H1-53 was identical for ViPS⁺ splenic and ViPS⁺ PerC B cells (Table I). However, the second most frequent CDR3 peptide sequence associated with V_H1-55 had different J_H segment usage, suggesting that distinct B cell clones specific for ViPS are present in these two anatomical sites. In fact, certain distal V_H gene segments such as V_H1-20 and V_H1-6 were selectively represented in the ViPS⁺ B cells from spleen and PerC, respectively. Although we have not demonstrated that ViPS-specific Abs with distal V_H segments are responsible for the protective immunity, these findings supported our hypothesis that distal V_H usage is associated with ViPS responsiveness in adult mice.

To further test whether IL-7 is required for Ab responses to other bacterial polysaccharides by promoting the usage of distal V_H gene segments, we investigated Ab responses to α-glucans, commonly referred to as dextrans. Dextrans are high m.w. homopolymers comprised of d-glucose molecules with various types of linkages such as α1-6 and α1-3 and have been widely employed to study B cell responses to polysaccharides (15, 41–45). B512 dextran from Leuconostoc mesenteroides contains primarily α1-6 linkages between the d-glucose molecules (46) and induces a robust Ab response in adult C57BL6 mice but not young C57BL6 mice (44) (Fig. 5A). Although certain IgH allotypes are associated with α1-6 dextran recognition, an unbiased analysis of V_H sequences of more than 30 IgM hybridomas generated from B512 dextran–immunized C57BL/6J mice revealed predominant usage of distal V_H genes (i.e., the V_H1 family) (43). Consistent with a role for IL-7 in distal V_H gene expression, adult IL-7^−/− mice are impaired in α1-6 dextran–specific IgM responses (Fig. 5A). Because B1b cells have been shown to recognize a number of polysaccharides (14–16), we tested whether the B1b cell subset also recognizes α1-6 dextran. Indeed, we found that upon immunization, the majority of α1-6 dextran–specific B cells of adult wildtype mice exhibit the phenotype of B1b cells (Fig. 5B).

Unlike dextran B512, B1355 dextran is a homopolymer of d-glucose predominantly with α1-3 linkages (15, 41). The α1-3–linked glucan epitopes are present on a variety of clinically relevant organisms, including the opportunistic bacterial pathogen E. cloacae. Responses to α1-3 dextran in mice are associated with the IgM^a allotype, and therefore immunization with B1355 dextran induces a robust IgM response in BALB/c mice but not in C57BL/6 mice (15). Because the IL-7^−/− mice are on C57BL/6 background, we were unable to directly determine the role of IL-7 in Ab responses to B1355 dextran. Nevertheless, we found that young (3-wk-old) BALB/cJ mice respond poorly to B1355 dextran compared with adult BALB/cJ mice (Fig. 5C). The significantly lower response to α1-3 dextran in young mice strengthens the conclusion that young mice are impaired in responding to a variety of bacterial polysaccharides. Immunization of adult BALB/c mice with dextran B1355 results in a rapid expansion of α1-3 dextran–specific B cells, and a significant percentage of these B cells belong to the B1b cell subset (Fig. 5D), as observed for α1-6 dextran in C57BL/6 mice (Fig. 5B).

Our hypothesis suggests that the limited usage of distal V_H gene segments accounts for the paucity of polysaccharide-specific B cell precursors in young mice. If this is the case, it is likely that mice with enforced expression of a rearranged BCR for a given polysaccharide Ag should respond efficiently to that polysaccharide regardless of age. To test this possibility, we used a BCR IgM H chain (V_HJ558) Tg mouse system (15). The BCR of this Tg mouse is composed of a distal V_H segment and recognizes α1-3 dextran. Naive V_HJ558 Tg mice contain a high frequency of α1-3 dextran–specific B cells, which expand rapidly upon immunization with either heat-killed E. cloacae expressing the α1-3 dextran epitope or with purified B1355 dextran (Fig. 6A). Immunization of 3-wk-old V_HJ558 Tg mice with heat-killed E. cloacae resulted in levels of α1-3 dextran–specific IgM comparable to adult V_HJ558 Tg mice (Fig. 6B). In fact, efficient α1-3 dextran–specific IgM responses were also induced in 1- or 2-wk-old V_HJ558 Tg mice (Fig. 6C), although these responses were approximately half of the response seen in adult V_HJ558 Tg mice (Fig. 6C). As expected, V_HJ558 Tg⁻ littermates did not generate detectable α1-3 dextran–specific IgM. These findings demonstrate that young and even infant mice can respond to polysaccharide Ags, provided they express suitable V_H genes in their preimmune repertoire.

To test whether IL-7 is dispensable for antipolysaccharide responses when appropriate V_H genes are expressed within the preimmune repertoire, we crossed the V_HJ558 Tg mice onto an IL-7–deficient background. Previously, it was shown that IL-7 deletion in mice resulted in a more severe deficiency in FO B cells compared with B1a, B1b, or MZ B cells (13, 47), and we reproduced this finding in the V_HJ558 Tg mice (Fig. 7A). Although the absolute numbers of B1b and MZ B cells were significantly lower in these mice (Fig. 7B), α1-3 dextran–specific (i.e., EB3-7⁺) B1b and MZ B cells were relatively abundant (∼10,000–50,000 cells per mouse) in preimmune V_HJ558 Tg × IL-7^−/− mice (Fig. 7B). These data show that IL-7 deletion in the V_H J558 Tg background results in alterations in mature B cell subset numbers similar to those seen in non-BCR Tg IL-7^−/− mice (13, 47).

TLR ligand contamination has previously been shown to influence bacterial polysaccharide responses (48). To minimize the impact of polyclonal activators such as TLR ligands, we subjected our B1355 preparation to phenol extraction as described previously (48). Unlike crude B1355 preparations, phenol-extracted B1355 did not induce IL-6 secretion by peritoneal exudate cells (Supplemental Fig. 2). Despite a general decrease in the numbers of all mature B cells (Fig. 7B), immunization of V_HJ558 Tg × IL-7^−/− mice with purified B1355 dextran resulted in a rapid α1-3 dextran–specific IgM response within 7 d postimmunization, and Ab levels were comparable to those seen in V_HJ558 Tg × IL-7^+/+ mice (Fig. 7C).

There has been considerable debate regarding the importance of IL-7 for B cell lymphopoiesis in humans. The argument that IL-7 is not critical for human B cell lymphopoiesis is based in part on the fact that humans with IL-7Rα mutations possess peripheral B cells (49). Because of their extreme susceptibility to a variety of infections, IL-7Rα–deficient individuals are typically diagnosed at an early age. All of these individuals undergo hematopoietic stem cell transplantation very early in life (49). This makes it impossible to evaluate the effects on B cell responses due to IL-7Rα loss in humans (49). Nevertheless, like IL-7Rα–deficient humans, mice lacking IL-7 or IL-7Rα possess appreciable numbers of peripheral B cells (13, 47). Our work indicates that those mouse B cells are not capable of responding to polysaccharide Ags, such as ViPS (Figs. 2, 3), because of a restricted BCR repertoire. IL-7Rα is transiently expressed at the pro– and pre–B cell stage of mouse and human B cell development (50). Mature B cells, including B1b cells, do not express IL-7Rα (Supplemental Fig. 3), indicating that IL-7–signaling is not required for mature B cell function. Indeed, we found that α1-3 dextran–specific B cell subsets in V_HJ558 Tg mice expanded upon immunization in the absence of IL-7 (Supplemental Fig. 4). Furthermore, we found that IL-7^−/− mice have no deficiency in mature plasma cells (data not shown) or serum IgM and IgG (13). In fact, the availability of an Ag-specific BCR enabled efficient Ab production even in the absence of IL-7 (Fig. 7). It was previously reported that B1a cells (which are known to express BCRs with D_H–J_H proximal V_H gene segments) of IL-7^−/− mice generate appreciable levels of Ab to a glycolipid Ag (FtL) of Francisella tularensis (51). Thus, the impaired Ab response to ViPS in IL-7^−/− mice is not due to a defect in the differentiation of B cells into Ab-secreting plasma cells but appears to be due to a restriction in BCR diversity associated with proximal V_H usage.

In addition to IL-7, several other factors contribute to the regulation of humoral immunity to polysaccharide Ags early in life. The developing B cells of human infants compared with B cells in older children express significantly reduced levels of terminal deoxynucleotidyl transferase, an enzyme required for BCR junctional diversification (52, 53). Additionally, the impaired responses to polysaccharide Ags in the very young have been attributed to delayed MZ B cell development. Low expression of complement receptor CR1/2 on MZ B cells in the spleens of children under 2 y of age has been suggested as a reason for their impaired responses to polysaccharide Ags (54, 55). A role for accessory cells in the neonatal response to polysaccharides has also been implicated (56). Furthermore, TACI and its ligands BAFF and APRIL have been shown to play an important role in antipolysaccharide Ab responses (57). Indeed, it has been reported that decreased expression of TACI on B cells is associated with diminished humoral responses in neonates (58, 59). These factors could explain our observation of diminished responses in 1- to 2-wk-old V_HJ558 Tg⁺ mice after immunization compared with 3-wk-old or adult V_HJ558 Tg mice (Fig. 6B versus 6C).

In the Streptococcus pneumoniae infection model, immunization with Pneumovax 23 confers statistically significant protection in adult wildtype and young IL-7 Tg mice but not in young wildtype mice (13). The differences in complement-dependent bactericidal Ab data (Figs. 1B, 2B) from ViPS-immunized young and adult mice are consistent with the levels of the ViPS-specific Ab response (Figs. 1A, 2A) and our hypothesis. However, in the S. Typhimurium infection model, we did not find a statistically significant difference in bacterial burden between immunized young and adult mice (Figs. 1C, 2C). Reasons that could account for this apparent discordance between the SBA and bacterial burden in infected mice include the use of different pathogens and different modes of antibacterial activity mediated by Abs in these two experimental settings. For example, we have used S. Typhi in the SBA and S. Typhimurium in the in vivo challenge model. It has been suggested that a correlate of ViPS-mediated protection against S. Typhi is complement-dependent bactericidal activity of the induced Abs (31). Interestingly, mouse complement is not effective in activating the classical complement pathway on certain nontyphoidal Salmonella, including S. Typhimurium (60). It is important to note that in the in vitro SBA, we used baby rabbit serum as a source of complement. Therefore, the bacterial burden data in the in vivo challenge model are not expected to reflect the protection conferred by complement-dependent, Ab-mediated bactericidal activity. The relative control of bacterial burden observed upon ViPS immunization of adult wildtype mice could be due to other defense mechanisms such as opsonophagocytosis, as previously suggested (60, 61). Thus, use of S. Typhimurium strain RC60 as a “surrogate pathogen” to determine in vivo protective immunity to S. Typhi has limited utility.

It has previously been shown that the majority of adult human B cells that bind to PPS serotypes 14 and 4 use distal V_H genes V_H3-74 and V_H3-48, respectively (62), whereas those that bind to Haemophilus influenzae type b polysaccharide and PPS serotypes 6b and 23f use the V_H3-23 located in the center of IgH locus (63–65). These data are consistent with our findings on the distal V_H gene usage associated with ViPS-binding B cells in mice and indicates that the use of a given V_H gene is dependent on the type of polysaccharide Ag. Recent studies indicated that decreased expression of IL-7Rα in B cell precursors at the perinatal stage in humans is responsible for the restricted BCR repertoire of neonates (53). Nevertheless, despite having a restricted BCR repertoire, infants can generate antipolysaccharide responses when immunized with polysaccharide–protein conjugate vaccines. Generating the antipolysaccharide Ab response in infants, however, requires multiple immunizations over a period of several months. Presumably, this vaccination strategy incrementally increases the numbers of otherwise rare polysaccharide-specific B cells with each booster immunization. Although conjugate vaccines for a variety of bacterial pathogens have been proven to be successful in the United States and Europe, surprisingly, in a multinational clinical trial in India, Pakistan, and the Philippines, ViPS and PPS conjugate vaccines failed to induce significant Ab responses in infants despite multiple booster immunizations, and the reasons are not known (7). Therefore, exploration of alternate strategies is needed to develop new approaches to induce antipolysaccharide responses in infants in disease-endemic countries. TLR ligands have been shown to promote polysaccharide-specific Ab responses in young mice (66, 67). However, this requires the administration of TLR ligands either a day prior to or 2 d after the polysaccharide immunization. This timing requirement suggests that TLR ligands can enable polysaccharide-specific Ab responses in young mice by expanding rare polysaccharide-specific B cells in preimmune or postimmune mice. In fact, when TLR ligands were used as adjuvants, an increase in B cell division and polysaccharide-specific plasma cell numbers occurred (67). Previously, it was shown that TLR2 signaling enhances Ab responses to H. influenzae type B polysaccharide (68). Therefore, using TLR ligands as adjuvants could enhance ViPS responses in the young by expanding Ag-specific B cell precursors. Our work suggests that strategies to enhance the breadth of preimmune BCR diversification could not only help overcome the impaired responses to certain polysaccharide vaccines in infants and young children but also reduce the number of booster immunizations required to achieve protective immunity.

We thank Dr. Andreas Bäumler for providing S. Typhi strain Ty2, S. Typhimurium strain RC60, E. coli strain W3110 (pDC5), and for discussion; Dr. Tim Manser and Jennifer F. Wilson for editing this manuscript; and Dr. R. Paul Wilson for discussion. We also thank Drs. Shousun Szu and John Cipollo for providing ViPS and Alexa Fort (iRepertoire) for providing the control mouse B cell V_H sequence data.

The authors have no financial conflicts of interest.