Strong Differential Regulation of Serum and Mucosal IgA Responses as Revealed in CD28-Deficient Mice Using Cholera Toxin Adjuvant

Animals

CD28^−/− (28), CTLA4-Hγ1 Tg mice (20), and C57BL/6 WT mice were bred under pathogen-free conditions using microisolator cages and sterile work benches at the Department of Medical Microbiology and Immunology (University of Göteborg, Sweden). Breeding pairs were kindly provided by T. Mak (Toronto, Ontario, Canada) (CD28^−/−) and P. Lane (University of Birmingham, U.K.) (CTLA4-Hγ1-Tg). When indicated experiments were conducted with mice generated by crossing and intercrossing the CTLA4-Hγ1 Tg onto the CD28^−/− mice. The genotype of the CD28^−/− and CTLA4-Hγ1 positive or negative mice were confirmed by PCR using appropriate primers, by FACS analysis of phenotype using CD28-specific Abs, and by ELISA for detection of the CTLA4-Hγ1 protein, as previously described (19, 28). All experiments were conducted with sex-matched, 8- to 14-wk-old mice. All animal studies have been approved by the Institutional Review Board.

Immunizations

Mice were given three oral immunizations at 10 days apart with keyhole limpet hemocyanin (KLH) (Sigma-Aldrich, St. Louis, MO) at 2.0 mg/dose in the presence or absence of 10 μg/dose of CT (List Biological Laboratories, Campbell, CA) (29). Six to 8 days following the final immunization, mice were sacrificed and the immune responses were analyzed. i.p. immunizations were performed twice at 10 days apart with DNP-KLH (Boehringer Mannheim, Indianapolis, IN) at 200 μg/dose in the absence or presence of 1 μg/dose CT (List Biological Laboratories).

Preparation of lymphoid cells

LP lymphocytes were prepared as described (30). Briefly, the small intestines were removed and the PP were carefully excised and discarded. The intestines were opened up, cut in small pieces, and washed six times in calcium- and magnesium-free HBSS (CMF-HBSS; Life Technologies, Paisley, U.K), supplemented with 25 mM HEPES (Life Technologies), and incubated four times, 15 min each time, with prewarmed (37°C) CMF-HBSS containing 5 mM EDTA (Merck, Darmstadt, Germany) and 10% heat-inactivated horse serum (Life Technologies). After each incubation, the supernatant containing the sloughed epithelial cells and the intraepithelial lymphocytes were discarded. After four incubations, the epithelium was removed and the supernatants were clear. To block any remaining EDTA activity, the tissue pieces were incubated for 15 min with RPMI 1640 (Flow Laboratories, Irvine, U.K.) containing 25 mM HEPES and 10% inactivated horse serum. Next, the tissue pieces were incubated three times for 60 min with collagenase (120 IU/ml; Sigma-Aldrich) dissolved in RPMI 1640 containing 25 mM HEPES and 20% inactivated horse serum (Life Technologies). After each incubation, the suspension of cells were collected, washed, and centrifuged and stored in IMDM containing 10% FCS. Spleen lymphoid cells were prepared by teasing the tissue through a nylon screen. RBCs were lysed by osmotic shock and single-cell suspensions were prepared and washed twice in CMF-HBSS (Life Technologies) containing 10% FCS (Life Technologies).

Ligated loop test

For evaluation of protection against CT-induced diarrhea/fluid loss, the method described by Lange and Holmgren (31) was used. Briefly, the abdomen was opened under light ether anesthesia, and a 6- to 8-cm loop was ligated in the middle part of the small intestine. CT (List Biological Laboratories), 2.5 μg in 0.2 ml of PBS, was injected into the loop, and the abdomen was closed. After 4 h the mice were sacrificed, whereafter the loop with its fluid content was weighed and its length was determined. Unimmunized mice of both strains were equally sensitive to CT injected in ligated small intestinal loops. Values for fluid accumulation in the ligated loops, reflecting the degree of immune protection, were expressed as the weight per length ratio in milligrams per centimeter ± SD of 5–7 mice per group.

ELISPOT assay

LP lymphocytes were analyzed for Ag-specific or total IgA Ab production at the single-cell level (spot-forming cell; SFC) using the ELISPOT assay (32). For the Ag-specific analyses, petri dishes were coated with KLH (Sigma-Aldrich) at 100 μg/ml or 3 nmol/ml ganglioside GM1 (Sigma-Aldrich) followed by 3 μg/ml of CT (List Biological Laboratories). Lymphocytes in complete IMDM were added at at 1 × 10⁶ cells/ml, and 400 μl/well, in duplicates to coated petri dishes after blocking with 0.2% BSA/PBS. Cells were allowed to incubate for 2.5–4 h at 37°C. Petri dish SFC were visualized using a two-step procedure involving a first incubation at 4°C overnight with goat anti-mouse IgA (Cappel; Organon Teknika, West Chester, PA) followed by a second HRP-labeled rabbit anti-goat Ig (DAKO, Glostrup, Denmark) both at 1/200 dilution for 2 h at room temperature. Total Ig SFC in splenic cells were detected by sequential inubations of HRP-labeled rabbit anti-mouse Ig (DAKO) followed by HRP-labeled anti-rabbit Ig (DAKO). The SFC reaction was visualized by adding the HRP substrate: paraphenylendiamine at 0.5 mg/ml and 0.01% H₂O₂ in 1% agar in PBS.

Intestinal lavage samples

Intestinal secretions for Ab determinations were collected as described in detail (33). Briefly, the small intestines were taken out, rinsed in PBS, and carefully injected with 1.8 ml of a protease inhibitor solution consisting of 0.1 mg/ml soybean trypsin inhibitor (Sigma-Aldrich), 50 mM EDTA (Sigma-Aldrich), and 1 mM PMSF (Boehringer Mannheim) in PBS. After incubation for 10 min at room temperature, the intestinal content was then transferred to a test tube, vigorously vortexed, sonicated, centrifuged, and transferred to a microfuge tube with PMSF at a final concentration of 1 mM and 5% FCS. The supernatants were stored at −70°C until analyzed.

Serum and gut lavage ELISA

Total or specific Ig concentrations were determined by ELISA (33, 34). Polystyrene microtiter plates (Nunc, Roskilde, Denmark) were coated overnight at 4°C with unlabeled rabbit anti-mouse IgG1, IgG2a, IgG3, IgM (Southern Biotechnology Associates, Birmingham, AL), or purified monoclonal IgA (BD PharMingen, San Diego, CA) Abs at 5 μg/ml. For Ag-specific Ab, plates were coated with KLH (Sigma-Aldrich) at 100 μg/ml or GM1 ganglioside (Sigma-Aldrich) at 0.5 nmol/ml followed by CT (List Biological Laboratories) at 0.5 μg/ml. Serum was diluted at 1/100 or 1/200 and lavage at 1/10 in PBS before added in 3- or 2-fold dilutions in corresponding subwells and incubated overnight at 4°C. The following day, alkaline phosphatase-conjugated isotype-specific rabbit anti-mouse Abs (Southern Biotechnology Associates) at 1/300 dilution were added to the wells. Nitrophenyl-phosphatase substrates tablets (Sigma-Aldrich) at 1 mg/ml in ethanolamine buffer (pH 9.6) were used to visualize the alkaline phosphatase-labeled Abs. The enzymatic reactions were read at 405 nm using a Titertek Multiscan MS spectrophotometer (Labsystems, Stockholm, Sweden). Ab titers were defined as the interpolated OD reading giving rise to an absorbance of 0.4 above background, which consistently gave OD readings on the linear part of the curve. Titers were given as log₁₀ titers ± SD.

Immunohistochemistry

Frozen sections (5 μm) of gut mucosa from naive or immunized mice were prepared on microslides using a cryostat (model 1720; Leitz, Weltzar, Germany). The sections were stored at −70°C and transferred directly to 50% acetone in H₂O for 30 s, followed by fixation in 100% acetone for 5 min after which they were washed in PBS two times for 5 min. After washing, the sections were incubated with horse serum 20% in PBS for 15 min in a humidified box at room temperature to avoid unspecific binding to the tissue section. Detection of GC reactions and IgA containing cells in PP or mesenteric lymph nodes (MLN) was achieved by single or double labeling with FITC-conjugated peanut (Arachis hypogaea) hemagglutinin (PNA; Sigma-Aldrich) and/or biotin-labeled rat anti-mouse IgA mAb (BD PharMingen) followed by Texas Red-conjugated streptavidin (Sera-Lab, Sussex, U.K.). The slides were mounted in Fluorescent mounting media (DAKO), evaluated, and photographed using DAS Mikroskop, LEICA DMLD (Leica, Weltzar, Germany).

Mutational analysis

LP lymphocytes from WT and CD28^−/− mice were prepared as above and used for purification of IgA⁺ plasma cells by MACS enrichment using biotinylated goat anti-IgA Abs (Southern Biotechnology Associates) and streptavidin-coated beads. The highly enriched gut IgA⁺ plasma cells were >95% pure, as assessed by FACS analysis and contained no other surface Ig⁺ cells (i.e., <1%). Single-cell suspensions of C57/BL6 splenocytes were prepared as above and B220 (RA3-6B2) positive cells were enriched by MACS to 98% purity as determined by FACS. DNA from the IgA⁺ and B220⁺ populations was prepared using TRIzol (Life Technologies) and used for subsequent PCR. Total mutational load in IgA⁺ or B220⁺ cells from either WT or CD28^−/− mice was compared using the strategy developed by Jolly et al. (35), analyzing intronic sequences from the J558 V_H family. The splenic B220⁺ cells served as a negative control for the experiment. Amplifications were performed using a polymerase mix containing a proofreading activity (Expand High Fidelity System; Roche, Basel, Switzerland). Amplified material was blunt end cloned into pGEM-3Z cut with SmaI, sequenced, and analyzed using the computer software VectorNTI (Informax, Bethesda, MD).

Statistical analysis

We used Student’s t test for independent samples for analysis of significance.

Unaltered total gut IgA production despite lack of GC in the PP

Our previous studies in CTLA4-Hγ1 Tg mice showed a paradox in the regulation of gut mucosal IgA immunity (21). We observed that the block in CD80/86 costimulation resulted in severely impaired specific mucosal and serum IgA responses following oral immunizations, despite seemingly normal IgA-inductive sites in PP and normal levels of total gut and serum IgA (21). In this report, we have extended our studies to CD28^−/− mice. In contrast to CTLA4-Hγ1 Tg mice, naive CD28^−/− mice exhibited no GC in PP and greatly reduced levels of serum IgA (Fig. 1⇓). Despite this, CD28^−/− mice had almost normal levels of total gut IgA in lavage and IgA⁺ plasma cells in the LP (Fig. 1⇓). Serum IgM was unaltered, whereas IgG1 and IgG2a Abs were substantially reduced compared with WT mice (Fig. 1⇓). Thus, IgA production in the gut LP can be independent of CD28 signaling, whereas serum IgA appears to be dependent on this pathway, suggesting a differential regulation of mucosal and systemic IgA responses. Furthermore, GC in the PP were not required for IgA production in the gut LP, indicating that the mucosal IgA B cell progeny can be derived from sites other than PP.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Intact gut mucosal IgA production while serum IgA levels are low in CD28^−/− mice. The total IgA concentration in serum (A) and in gut lavage (B) was determined in CD28^−/− (□) or WT (▪) mice. A, For comparison total serum IgG1, IgG2a, and IgM were determined in both strains. The bars represent mean values and are given as log₁₀ titers ± SD of three experiments with five mice in each group. C, Normal distribution and frequency of IgA⁺ cells (green) in a frozen section from gut small intestinal LP of CD28^−/− mice. The frequency of IgA⁺ cells/visual field with 4–5 villie was 28 ± 4 in CD28^−/− and 32 ± 5 in WT mice. Frozen sections of gut tissue from WT (D) or CD28^−/− (E) mice were double-labeled with PNA-FITC (green) and anti-IgA-Texas Red Abs (red), showing the complete absence of GC and IgA⁺ cells in PP of CD28^−/− mice. Frozen sections are representative of four separate experiments given similar results with 3–4 mice in each group. ∗, Significance, p < 0.05.

Strong gut mucosal IgA responses to oral immunization in CD28-deficient mice

Next, we analyzed the IgA response to an oral immunization with KLH and CT adjuvant in CD28^−/− mice and compared the result with that of CTLA4-Hγ1 Tg and WT mice. Unexpectedly, and in contrast to the poor IgA response in CTLA4-Hγ1 Tg mice, CD28^−/− mice demonstrated near normal gut mucosal IgA SFC responses to both KLH and CT (Fig. 2⇓). Particularly, the response to the strong mucosal immunogen CT differed significantly between CD28^−/− and CTLA4-Hγ1 Tg mice, with >95% reduced, or even undetectable, anti-CT IgA SFC activity in the latter strain, while the specific gut IgA LP response in CD28^−/− mice was not significantly different from that of WT mice (Fig. 2⇓). Therefore, the ability to respond to KLH or CT after oral immunization appeared to be independent of CD28 signaling, but blocked by CTLA4-Hγ1 protein, as seen in the Tg animals. Moreover, we found no correlation between a specific IgA response in the gut LP and the presence or absence of GC and IgA⁺ B cells in the PP, since the CTLA4-Hγ1 Tg mice had GC in PP, but could not respond (21), whereas immunized CD28^−/− mice had no GC in PP or MLN (data not shown), but responded normally to an oral immunization (Fig. 2⇓E). Of note, also other peripheral lymph nodes and the spleen were lacking GC formations following both oral and systemic immunizations, which agrees well with earlier reports using CD28^−/− mice (12).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Specific gut IgA responses following oral immunizations can be CD28-independent KLH- (A and B) or CT- (C, D, and F) specific Ab-producing cells (SFC) in small intestinal LP following oral immunizations of CD28^−/− or CTLA4-Hγ1 Tg mice (□) or WT mice (▪) with KLH admixed with CT adjuvant. The bars represent mean SFC values per 10⁷ cells ± SEM of three experiments with six mice per group. E, Frozen section of PP from orally immunized CD28^−/− mice, showing the complete absence of GC (green) and IgA⁺ cells (red). For comparison with WT controls, see Fig. 1⇑. ∗, Significance, p < 0.05.

To prove that the specific gut mucosal IgA response was dependent on a CTLA4-Hγ1-blockable costimulatory pathway and independent of CD28 signaling, we extended our analysis to include CD28^−/− mice that did or did not express the CTLA4-Hγ1 Tg. By cross and intercross breeding we obtained CTLA4-Hγ1 Tg positive or negative CD28^−/− mice that were given oral immunizations with CT. We found that the presence of the CTLA4-Hγ1 Tg greatly impaired the gut IgA anti-CT SFC response also in CD28^−/− mice (Fig. 2⇑F). Thus, this result corroborated the data in WT mice, and clearly demonstrated the presence of an alternative, CD28-independent, costimulatory pathway in the gut mucosal immune system.

Evidence of a differential regulation of local mucosal and systemic IgA immunity

Whereas the local IgA response, especially that against CT, was clearly independent of the CD28-signaling pathway, we investigated whether this was also the case for systemic responses following oral or i.p immunizations with CT. However, contrary to the gut IgA responses, both CD28^−/− and CTLA4-Hγ1 Tg mice exhibited poor or no specific SFC responses in the spleen following either oral or i.p. immunizations, as illustrated in Fig. 3⇓. Likewise, serum anti-CT Ab responses, including IgA, were also dramatically impaired after oral (Fig. 4⇓) or i.p. immunizations (data not shown). Thus, CD28 signaling was clearly required for systemic immunity, including serum IgA, whereas mucosal IgA responses appeared to be independent of signaling through the B7-CD28 pathway.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Impaired systemic responses in CD28^−/− and CTLA4-Hγ1-Tg mice after oral as well as systemic immunizations. Mice were immunized orally (A and B) or i.p (C and D) with CT. Specific systemic responses in the spleen were recorded following a booster immunization. CT-specific responses are given as mean splenic total anti-CT SFC activity in CTLA4-Hγ1-Tg or CD28^−/− mice (□) or WT (▪) mice, as indicated. The SFC activity is expressed as mean SFC values per 10⁷ cells ± SD from three pairs of mice per group. This is one representative experiment of four. ∗, Significance, p < 0.05.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Strongly impaired serum Ab responses to oral immunizations with CT in CD28^−/− mice. Anti-CT serum isotype-specific titers are given as mean log₁₀ titers ± SD for six mice per group and three identical experiments. ∗, Significance, p < 0.05.

Gut LP IgA⁺ cells from CD28^−/− mice are modified by somatic hypermutations

Given the ability to raise near normal specific gut IgA responses in CD28^−/− mice, despite the complete absence of GC in the PP or MLN, we analyzed whether gut IgA LP plasma cells were modified by somatic hypermutations following oral immunizations. After extensive cell sorting using MACS technology to obtain highly pure IgA⁺ plasma cells (36) from the gut LP of immunized WT and CD28^−/− mice, we used the strategy developed by Jolly et al. (35) to analyze the presence or absence of hypermutation. This method uses the fact that mutations are found extending into the J-C intron. Thus, by using a degenerate V gene primer, recognizing all V genes of the J558 V gene family, together with a primer hybridizing in the intronic enhancer region, a global analysis, irrespective of specific V gene usage, can be performed. By this approach we found that IgA⁺ plasma cells in the LP of the CD28^−/− mice carried modifications by somatic hypermutation (Fig. 5⇓). Although the frequency of mutation (1/300 bp) was reduced compared with the WT mice (1/80 bp), the mutational load of the CD28^−/− plasma cells were clearly above the background PCR error rate, as shown by the simultaneous analysis of naive B cells (1/2500). Thus, despite the absence of GC, gut IgA⁺ plasma cells in CD28^−/− mice were modified by somatic hypermutation, a characteristic of B cells undergoing maturational processes normally associated with clonal expansion and differentiation in GC in the GALT in WT mice.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

IgA⁺ plasma cells from gut LP of CD28^−/− mice display modifications by somatic hypermutation in the absence of PP GC. Analysis of the levels of somatic hypermutation in IgA⁺ LP plasma cells from either WT or CD28^−/− mice. DNA were prepared from MACS-enriched cells and used for amplification of Ig intronic sequences using a proofreading polymerase. Purified splenic B cells served as a negative control for the PCR. Each diagram represents the analysis of a 240-bp fragment from 30 individual clones and shows the number of mutations per sequence. The average frequency of mutation is indicated below each diagram.

Gut IgA responses in the CD28^−/− mice are functionally normal

Somatic hypermutations in CD28^−/− gut IgA⁺ cells were clearly found, but whether these mutations also represented differentiation and maturation of specific IgA B cell clones following oral immunization was unclear. Therefore, we orally immunized CD28^−/− and WT mice with CT and subsequently subjected them to an intestinal ligated loop test to assess whether their gut IgA responses were functional and could neutralize toxin in an in vivo challenge model (31). As illustrated in Fig. 6⇓, we found strong and comparable antitoxic protection in both CD28^−/− and WT mice, indicating that gut IgA B cell clones in CD28^−/− mice could both expand and mature to production of functionally relevant mucosal IgA Abs. Immunized mice from both strains resisted an intestinal challenge with CT and showed >75% protection compared with unimmunized control mice (Fig. 6⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Intact antitoxic protection in gut mucosa following oral immunization with CT in CD28^−/− mice. We determined functional antitoxic protective immunity in gut-ligated loops of perorally CT-immunized CD28^−/− or WT mice. Three oral immunizations with 10 μg/dose of CT were given and 7 days following the final dose, mice were analyzed for antitoxic protection against CT-induced diarrhea by the ligated loop test. Fluid accumulation in response to challenge with CT, 2.5 μg/loop, was determined and compared with that observed in CT-challenged loops of unimmunized control mice. Values are given in milligrams per centimeter ±SD of 5–7 mice in each group with immunized (□) and unimmunized control (▪) mice. The data are representative of two identical experiments giving similar results. ∗, Significance, p < 0.05.

The adjuvant effect of CT can be independent of CD28 signaling

In previous studies we have demonstrated that CT dramatically increases expression of both CD80 and CD86 on B cells (24) and others have reported similar findings in macrophages and dendritic cells (23, 37). Accordingly, it has been proposed that CT exerts much of its adjuvant function through enhancing costimulation, and especially signaling through the B7-CD28 pathway (24, 38, 39). However, in the present study, we unexpectedly found that CT greatly augmented mucosal IgA anti-KLH responses in CD28^−/− mice and that the IgA responses appeared to be functionally normal. Extending the analysis, we determined the enhancing effect of admixed CT on specific serum Ab responses following i.p immunizations with DNP-KLH. We found consistently increased DNP-specific serum titers when CT adjuvant was used (Fig. 7⇓). As expected, responses to DNP-KLH of all isotypes were much lower in CD28^−/− compared with WT mice, but the enhancing effect of CT was comparable (Fig. 7⇓A). Calculated as a relative change, CD28^−/− and WT mice had comparable increases in serum specific titers, as a function of CT adjuvant (Fig. 7⇓B). On average the specific serum titers were 70–90% higher in CT adjuvant-treated as opposed to nontreated mice. Thus, CT exerted both mucosal and systemic adjuvant function in the complete absence of CD28. The difference in responsiveness to mucosal as opposed to systemic immunizations with protein Ag rather reflected the degree of dependence on CD28, with mucosal IgA being relatively independent, whereas systemic/serum IgA responses are highly dependent on signaling through the CD28 pathway.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

CT exerts adjuvant effects on humoral immune responses in CD28^−/− mice. To assess the relative enhancing effect of CT adjuvant on humoral immune responses we immunized (A) CD28^−/− (□) and WT (▪) mice twice i.p with DNP-KLH in the absence or presence of CT striped (▨) and determined the DNP-specific serum Ab response by ELISA. The results are given as mean log₁₀ titers ± SD of three identical experiments. The enhancing effect of CT adjuvant on the specific Ab titers in CD28^−/− mice was calculated as relative increase and compared with that observed in WT mice. The gray area of the bars represent the mean enhancement in percentage of CT-adjuvant of three separate experiments (B).