GPR43 regulates marginal zone B‐cell responses to foreign and endogenous antigens

Abstract Marginal zone (MZ) B cells are innate‐like B cells that produce polyreactive antibodies with an affinity for microbial molecular patterns and carbohydrate ligands. MZ B cells have been shown to be important in mediating immunity to various bacteria including Streptococcus pneumoniae and are also implicated in inflammatory syndromes including lupus erythematosus. The intestinal microbiota is responsible for producing short‐chain fatty acids, which can regulate immune cell function by several mechanisms including ligation of the G‐protein‐coupled receptor (GPR)43. Herein, we show that MZ B cells express Gpr43 messenger RNA and that the absence of this receptor impacts on MZ B‐cell surface marker expression and antibody production. In T‐cell‐independent responses to the hapten 4‐hydroxy‐3‐nitrophenylacetic acid (NP), mice deficient in GPR43 displayed higher serum titers of NP‐specific antibodies. Moreover, in response to a pneumococcal polysaccharide vaccine, GPR43‐deficient mice developed robust serum antibody responses and had markedly increased numbers of splenic antibody‐secreting cells, compared with control mice. Finally, serum immunoglobulin M autoantibodies to double‐stranded DNA and phosphatidylcholine were increased in resting 10–15‐week‐old mice lacking GPR43. Taken together, mice lacking GPR43 have heightened antibody responses to T‐cell‐independent antigens, which may be a result of impaired regulation of MZ B cells.


INTRODUCTION
B cells can be broadly subdivided into three types: B-1, marginal zone (MZ) B and follicular (Fo) B cells. 1 Innate B-1 cells develop from the fetal liver and reside in large frequencies in the pleural and peritoneal cavities where they are self-renewing. These cells express B-cell antigen receptors that are enriched for polyreactivity, which afford early protection to several pathogens invading the body such as Streptococcus pneumoniae and Salmonella. Fo B cells express highly diverse monoreactive B-cell receptors, and are a major cell type in secondary lymphoid organs. 1 Fo B cells undergo expansion after stimulation through the B-cell receptor and following interactions with cognate T cells in germinal centers, undergo rounds of somatic hypermutation leading to isotype-switched, high-affinity antibodies against pathogens including bacteria and viruses. Finally, MZ B cells reside in the splenic MZ, adjacent to the red pulp, where they are exposed to blood-borne infections and particulate matter. 2,3 Like B-1 cells, MZ B cells express polyreactive B-cell receptors and are primarily thought to mediate short-term immunity to pathogens including Streptococcus pneumoniae, Haemophilus influenzae and Neisseria meningitidis. 2 Although MZ B cells are typically thought to afford early protection to invading pathogens, MZ B cells can also shuttle antigens into follicles to promote Fo B-cell responses. [3][4][5] Furthermore, MZ B cells can produce antibodies that recognize some forms of DNA and phosphoproteins, which can be expressed on the surface of apoptotic cells. The production of antibodies to endogenous particulate matter is proposed to have a beneficial role in homeostasis; however, such antibodies are also postulated to promote pathogenesis in certain inflammatory diseases including lupus and various cardiac disorders. 6 Thus, finding regulators of MZ B cells may help to define strategies that dampen or prevent autoimmune pathology, while retaining the homeostatic function of natural antibodies produced by these cells.
Short-chain fatty acids (SCFAs), typically acetate, butyrate and propionate, are produced by saccharolytic bacteria such as Firmicutes and Bacteroidetes. 7 SCFAs can be generated through a number of sources including fiber-rich diets and are produced at an approximately 3:1:1 acetate:butyrate:propionate molar ratio in the proximal colon. 7 Most SCFAs are rapidly absorbed into intestinal epithelial cells after secretion into the gut lumen, although some make it into the bloodstream, with high levels of variability reported within the population. 8 Preclinical studies where SCFAs or high-fiber diets are fed to mice have demonstrated that SCFAs may alleviate autoimmune and/or inflammatory pathologies of the bowels, pancreas, lungs and others organs. [9][10][11] Observational studies of populations with different diets have indicated that people with a high-fiber diet have increased levels of SCFA, 12,13 leading to the commencement of several clinical trials to test the efficacy of high-fiber diets in autoimmune and inflammatory settings such as diabetes, cirrhosis, intestinal inflammatory disorders and cardiovascular disease.
Many effects of SCFAs are related directly to their impact on intestinal epithelium, where they modulate cellular metabolism, gene expression and promote barrier integrity. 14 However, there is an increased appreciation that SCFAs regulate lymphocyte functions and immunity. Several studies have shown that SCFAs may promote Bcell responses to intestinal commensals and ingested antigens. 15,16 In mice fed high-fiber diets, an increase in immunoglobulin A (IgA)-producing B cells and decrease in IgE secretion were observed. Furthermore, mice fed SCFAs were shown to develop increased frequencies and numbers of IgA-secreting plasma cells in the intestinal lamina propria, owing to a boost in oxidative and glycolytic metabolism in B cells. 15,16 However, conflicting reports on the impact of SCFAs on antibody production exist. In two independent studies, butyrate and propionate have been shown to lower Aicda (gene for activation-induced deaminase) expression and reduce plasma cell differentiation. 17,18 SCFAs were shown to reduce the level of somatic hypermutation and these effects were thought to rely on the induction of microRNAs. 18 Differences in reports could be a result of a range of factors, including large differences in microbial flora between animal facilities worldwide, which would affect the levels of SCFAs in the gut lumen and in circulation.
Another complicating factor in understanding the role of SCFAs in health relates to their mechanism(s) of action. SCFAs may passively diffuse into cells or be actively transported via the proton-coupled monocarboxylate transporter 1 and the sodium-coupled monocarboxylate transporter 1. 14 Another mode by which they mediate their function is by activating Gprotein-coupled receptors (GPRs) including GPR41, GPR43 and GPR109a. Activation of SCFA receptors has been shown to have a variety of effects typical of GPRs, including the inhibition of cyclic adenosine monophosphate production and induction of calcium influx into the cytoplasm. 19 Mechanistically, activation of GPR43 has been shown to promote the differentiation of IgA + plasma cells in the gut, because they are reduced in frequency in GPR43-deficient mice. 16 Moreover, SCFA receptors have been shown to promote the functions and persistence of suppressive regulatory T and effector CD8 T cells. [20][21][22][23] In genome-wide expression analyses performed as part of the ImmGen consortium, 24   To investigate the expression of SCFA and long-chain fatty acid receptors, we purified MZ B (B220 + CD93 À CD21 + CD23 À ) and Fo B (B220 + CD93 À CD21 À CD23 + ) cells from the spleens of Gpr43 +/+ and Gpr43 À/À mice. 20 Gpr40, Gpr41, Gpr43, Gpr109a and Gpr120 messenger RNA was quantified by real-time PCR. We found that Gpr43 was expressed at higher levels in MZ B cells compared with Fo B cells and expression of Gpr43 messenger RNA was completely absent in MZ B cells from Gpr43 À/À mice (Figure 1a), confirming the genotype of these mice. Next, we analyzed the spleens of Gpr43 +/+ and Gpr43 À/À mice for the presence and phenotype of MZ B cells by flow cytometry and immunohistochemistry. The numbers of transitional (T) 1, T2, Fo B, MZ B and B-1 cells were similar between Gpr43 +/+ and Gpr43 À/À mice (Figure 1b). B-1-cell frequencies also appeared normal in the peritoneal cavity of Gpr43 +/+ and Gpr43 À/À mice (Supplementary figure 1). We noted that the expression of several cell surface receptors expressed by MZ B cells including CD1d, CD9, CD21, CD24, CD38 and IgM was significantly lower in Gpr43 À/À compared with control mice (Figure 1c). Analysis of the spleen by immunofluorescence with antibodies to CD1d, B220 and MOMA showed that the or MZ B cells from Gpr43 +/+ and Gpr43 À/À mice was subjected to quantitative reverse transcription-PCR analysis to measure the expression of Gpr40, Gpr41, Gpr43, Gpr120 and Gpr109a. All data are shown as mean values AE s.e.m. of two independent experiments (n = 4 mice per group), except for Gpr109a which was quantified in a separate experiment with n = 4 Gpr43 +/+ and n = 6 Gpr43 À/À . (b) Splenocytes of Gpr43 +/+ and Gpr43 À/À mice were analyzed by flow cytometry. B220 + CD93 + immature or B220 + CD93 À mature B-cell population are gated. B-cell precursors within the immature population were further divided into T1 (IgM + CD23 À ) or T2 (IgM + CD23 + ) cells. The mature B-cell population was further stained for CD21 and CD23 to divide cells into Fo B (CD21 À CD23 + ) or MZ B cells (CD21 + CD23 À ). B-1 cells were identified as CD19 + B220 À/low cells. Bar graphs show the frequencies or absolute cell numbers of T1, T2, Fo B, MZ B and B-1 cells. (c) Splenocytes from Gpr43 +/+ and Gpr43 À/À mice were stained for MZ B cell markers and mean fluorescence intensity (MFI) levels of the indicated surface molecules are plotted. Data are shown as mean values AE s.e.m. of two independent experiments (n = 5 or 6 mice per group). (d) Immunofluorescence staining of spleen sections from 10-to 12-week-old Gpr43 +/+ and Gpr43 À/À mice with B220 (blue) for B cells, CD1d (red) for MZ B cells and MOMA-1 (green) for marginal metallophilic macrophages. The MZ B-cell layer is marked by white arrows. Slides are representative of four mice per group.
structure of the MZs in the spleen appeared normal in Gpr43 À/À mice (Figure 1d). We also analyzed the serum of 10-15-week-old mice at resting for total IgM, IgG, IgG3 and IgG1 and for the presence of autoantibodies to dsDNA and PC, because these can indicate differences in the function of MZ B cells. While basal serum Ig levels were similar between Gpr43 +/+ and Gpr43 À/À mice (Figure 1e), higher levels of dsDNA-and PC-specific IgM were present in the serum of Gpr43 À/À mice compared with control mice (Figure 1f). Thus, the absence of GPR43 does not affect the development of MZ B cells, but may affect their phenotype and function.
Gene expression and responsiveness to lipopolysaccharide appear normal in MZ B cells from GPR43-deficient mice RNA-sequencing on purified splenic MZ B cells from Gpr43 À/À and Gpr43 +/+ mice was conducted to determine whether gene expression was altered in MZ B cells lacking

GPR43. Gene transcription profiles of MZ B cells from
Gpr43 +/+ and Gpr43 À/À mice were largely overlapping because only Gpr43 and Ceacam2 were significantly reduced in the absence of GPR43, after adjustment for multiple comparisons (Figure 2a). We next tested the responsiveness of purified Gpr43 À/À MZ B cells to lipopolysaccharide in vitro. MZ B cells from Gpr43 +/+ and Gpr43 À/À mice differentiated comparably into plasmablasts (B220 + CD138 + ) and plasma cells (B220 À CD138 + ) following culture over 6 days and secreted similar quantities of IgM (Figure 2b-d), suggesting that responsiveness to lipopolysaccharide was normal.
Enhanced antibody production in Gpr43 À/À mice in a T-cell-independent immunization model  dependent [NP-Chicken Gamma Globulin (CGG)] immunization models. We used high (NP30) and low (NP7) valency NP to detect the quality of the antibody response to NP. At day 6 after immunization with NP-Ficoll, serum from Gpr43 À/À mice contained significantly higher IgM specific to both NP (7) and NP (30), compared with serum from Gpr43 +/+ mice (Figure 3a). Quantification of IgM-secreting NP-specific splenic B cells by ELISpot analysis showed no significant difference between Gpr43 +/+ and Gpr43 À/À mice, although levels were slightly higher in GPR43-deficient mice (Figure 3b). Analysis of T-celldependent B-cell responses following immunization with NP-CGG in alum demonstrated no clear difference in the quantity of NP (7)-specific IgG1 present in the serum of Gpr43 +/+ and Gpr43 À/À mice (Figure 3c), nor in the quantity of NP (7)-specific IgG1-secreting cells in the spleen ( Figure 3d). Thus, GPR43 appeared to restrain MZ B-cell antibody responses to NP-Ficoll immunization, but had little effect on Fo B-cell responses to NP-CGG.
Enhanced antibody production in Gpr43 À/À mice to pneumococcal antigens We next evaluated the response of Gpr43 À/À mice to pneumococcal polysaccharide antigens (PNEUMOVAX). Mice were injected intraperitoneally with PNEUMOVAX and 6 days later serum and spleens were harvested and analyzed for the presence of antibodies and antibodysecreting cells. PNEUMOVAX-specific IgM was present in the serum of Gpr43 À/À mice in much greater quantities than in Gpr43 +/+ mice (Figure 4a). Serum antibodies to type 1 (161) and type 3 (169) carbohydrates were also significantly increased in mice lacking GPR43 (Figure 4a  . Gpr43 À/À mice have heightened responses to T-cell-independent antigens while the response to T-cell-dependent antigens is unaffected. (a) Gpr43 +/+ and Gpr43 À/À mice were immunized intravenously with NP-Ficoll. Six days after immunization, mice were analyzed for serum immunoglobulin M (IgM) to NP (7) or NP(30) by ELISA. (b) NP(7)-or NP(30)-specific IgM antibody-secreting cells (ASCs) from spleens were measured by ELISpot. Representative wells from splenocytes of Gpr43 +/+ and Gpr43 À/À mice are shown. (c) Gpr43 +/+ and Gpr43 À/À mice were injected with NP-CGG in alum. Seven days later, mice received a boost of NP-CGG diluted in phosphate-buffered saline. Fifteen days after the first injection, blood was collected and NP(7)-specific serum IgG1 levels were measured by ELISA. (d) NP(7)-specific IgG1 ASCs from spleens were measured by ELISpot. Representative wells from splenocytes of Gpr43 +/+ and Gpr43 À/À mice are shown. Three independent experiments were performed with NP-Ficoll and two independent experiments were performed with NP-CGG; n = 3 or 6 NP-immunized mice per group per experiment, n = 1 or 2 na€ ıve mice per group per experiment as controls. For ELISA, representative data from one experiment are shown and two-way ANOVA and Sid ak's test were used for comparison. For ELISpot analysis, bar graphs show the pooled results over all experiments and groups were compared using the Mann-Whitney U-test. *P < 0.05, ***P < 0.001. NP, 4-hydroxy-3-nitrophenylacetic acid.
Taken together, these results suggest that GPR43 restrains MZ B-cell responses to T-cell-independent antigens and reduces antibody responses to endogenous dsDNA and PC.

Discussion
SCFAs have been shown to have a range of immune modulatory functions, in particular in the gut and associated lymphoid organs. Because SCFAs can effectuate change via various mechanisms, it can be difficult to pinpoint how immune function is modulated. The results herein suggest that SCFAs regulate the function and maturation of MZ B cells in the spleen by activating GPR43. This finding may have important implications in the setting of lupus erythematosus and in various cardiovascular disorders. In both disease settings, antibodies that recognize phospholipids have been suggested to enhance the formation of clots and promote the development of antibody complexes, which can have deleterious effects on the health of several organs including the heart and kidneys. In a recent study of autoimmunity in lupus-prone mice, addition of SCFAs into drinking water significantly reduced oligonucleotidespecific antibody titers. 18 Although the authors suggested that the impact of SCFAs was independent of GPR43, our results support the possibility that SCFAs regulate antibody production to dsDNA, in part via GPR43.
Trials of high-fiber diets have been started in various clinical settings with the aim of reducing inflammatory symptoms. However, the impact of high-fiber diets or SCFA supplementation on inflammatory disorders in mice has often yielded conflicting results. 25 This may be because of the broad mechanisms of action of SCFAs. Agonists and antagonists of SCFA receptors have been characterized over recent years, which may provide a more targeted therapeutic option going forward. 26 For instance, mice treated with a GPR43 agonist were shown to be more resistant to dextran sulfate sodium-induced colitis. 23 Furthermore, GPR43 agonists have been shown to boost insulin secretion from pancreatic beta cells 27 and at the same time, inhibit insulin signaling into adipocytes, thereby preventing lipid accumulation in adipose tissue. 28 Whether GPR43 (or other SCFA receptors) is expressed by human MZ B cells, and whether they play a regulatory role, is yet to be explored.
Although our study implies that GPR43 may directly regulate MZ B-cell antibody responses, it is possible that B-1 cells, which also produce antibodies to carbohydrate antigens and to circulating dsDNA and PC, 29   . Gpr43 À/À mice have heightened responses to immunization with a pneumococcal polysaccharide vaccine. Gpr43 +/+ and Gpr43 À/À mice were immunized intraperitoneally with PNEUMOVAX pneumococcal polysaccharide vaccine and killed after 6 days. The serum immunoglobulin M (IgM) antibody response to (a) PNEUMOVAX, type 1 polysaccharide antigen 161-X or type 3 antigen 169-X was measured by ELISA. (b) The number of PNEUMOVAX-specific IgM antibody-secreting cells from the spleen was measured by ELISpot. Representative wells from splenocytes of Gpr43 +/+ and Gpr43 À/À mice are shown. For ELISA, two-way ANOVA and Sid ak's test were used and in ELISpot, the Mann-Whitney U-test was used to compare groups. Results represent mean values AE s.e.m. of two independent experiments (n = 5 PNEUMOVAX-immunized and 1 or 2 na€ ıve control mice per group per experiment). **P < 0.01, ***P < 0.001, ****P < 0.0001. altered in function in mice lacking GPR43. Furthermore, GPR43 may regulate MZ B-cell functions at various levels. For instance, whether GPR43 regulates B-cell receptor signaling into MZ B cells remains unclear. Reduced expression of receptors including CD1d and CD21 on MZ B cells also allude to the possibility that other MZ B-cell functions are altered in the absence of GPR43, including antigen presentation to natural killer cells 30 and potentially, antigen capture and transport into lymph node follicles. 4,5 The impact of GPR43 on these functions should be investigated. In all, this study suggests that GPR43 restricts antibody responses to Tcell-independent antigens and dampens antibody responses to the autoantigens dsDNA and PC.

Mice
C57BL/6N wild-type mice (Gpr43 +/+ ) were purchased from Janvier Labs (Rennes, France). GPR43-deficient mice (Gpr43 À/À) were sourced from Astra Zeneca. Mice were housed and bred at Karolinska Institutet animal facility. All animal procedures were approved by the Stockholms Jordbruksverket in Sweden and performed according to approved guidelines.

Quantitative reverse transcription-PCR
Cell pellets were resuspended in TRIzol reagent (Life Technologies, Carlsbad, CA, USA) and RNA was isolated following the manufacture's protocol. RNA was reverse transcribed into complementary DNA utilizing the Super Script IV First Strand Complementary DNA Synthesis kit (Thermo Fisher Scientific, Waltham, MA, USA). Quantification of Ffar1 (Gpr40), Ffar2 (Gpr43), Ffar3 (Gpr41), Ffar4 (Gpr120), Gpr109a or Hprt was performed using SYBR Green dye-based real-time PCR and samples were run on a CFX384 machine (Bio-Rad, Hercules, CA, USA). Data were analyzed using the 2DDCt method where Hprt was used as a housekeeping gene.
Immunization with NP-Ficoll, PNEUMOVAX or NP-CGG Age and sex-matched Gpr43 +/+ and Gpr43 À/À mice were injected intravenously with 50 µg of 2,4,6-trinitrophenyl-Ficoll (TNP-AECM-FICOLL; Biosearch Technologies, Novato, CA, USA), intraperitoneally with NP-CGG in alum or intraperitoneally with a 1:10 dilution of PNEUMOVAX (MSD, Merck Sharp and Dohme Corp, Kenilworth, NJ, USA), which corresponds to 0.5 µg of each polysaccharide antigen. The antigens were diluted in sterile PBS and mice were injected with 100 µL. For T-cell-independent immunizations, mice were killed 6 days later and spleens and blood were taken for fluorescence-activated cell sorting analysis, ELISpot and ELISA assays. For T-cell-dependent immunization with NP-CGG, mice were boosted at day 7 and killed at day 15.

Enzyme-linked immunosorbent assay
To measure antigen-specific antibodies, ELISA plates (Nunc) were coated with 5 µg mL À1 of NP (7)  Corp) or 2.5 µg mL À1 PC-BSA (Athera Biotechnologies, Stockholm, Sweden). Plates were washed with washing buffer (PBS + 0.05% Tween-20) and subsequently blocked with PBS containing 5% dry milk at room temperature (RT) for 1 h. Serum or supernatant was diluted in PBS containing 1% dry milk, was added to the plate in threefold serial dilutions and plates were incubated at RT for 1.5 h. After washing six times with washing buffer, 100 µL of diluted secondary horseradish peroxidase-or biotin-coupled goat antimouse IgG, IgG3, IgG1 or IgM antibody (Southern Biotechnology, Birmingham, AL, USA) was added per well and plates were incubated at RT for 1.5 h. For reactions using a biotinylated detection antibody, wells were washed six times and 100 µL of diluted streptavidin-horseradish peroxidase was added to plates and incubated at RT for 1 h. The assay was developed with 3 0 ,5,5 0 -Tetramethylbenzidine substrate (KPL, Gaithersburg, MD, USA) and the reaction stopped by adding 1 M H 2 SO 4 . Absorbance was read at 450 nm using an Asys Expert 96 ELISA reader (Biochrom, Cambridge, UK).

ELISpot
An ELISpot assay was used to detect NP-or PNEUMOVAXspecific IgM-or IgG-producing cells. MultiScreen Filter plates were pretreated with 70% EtOH, washed in sterile PBS and coated with 5 µg mL À1 [NP(7)-BSA or NP(30)-BSA] or 1:100 PNEUMOVAX diluted in PBS. Plates were incubated overnight at 4°C. The following day, plates were washed in sterile PBS and blocked in complete Roswell Park Memorial Institute medium for 2 h at 37°C. Single-cell suspensions were prepared and added to the plates in triplicate in twofold dilutions, starting at 5 9 10 5 cells/well in 100 µL total volume. Plates were wrapped in plastic foil and incubated for 16 h at 37°C in 5% CO 2 . The next day, cells were removed by washing the plates with PBS + 0.05% Tween-20, followed by incubation with biotinylated goat antimouse IgM or IgG (Mabtech, Nacka Strand, Sweden) antibody diluted in PBS for 2 h at RT. Plates were washed in PBS and ALPconjugated streptavidin (Mabtech) diluted in PBS was added followed by incubation for 45 min at RT. After washing plates with water, they were developed with 100 µL filtered 5bromo-4-chloro-3-indolyl phosphate/NBT-plus substrate (Mabtech). The reaction was stopped by washing plates extensively with tap water. Plates were leftover night to dry and spots were counted using an ELISPOT reader and analyzed using the BioSpot suite (Cellular Technology Limited, Shaker Heights, OH, USA).

In vitro stimulation
About 2 9 10 5 fluorescence-activated cell sorting-sorted MZ B cells or 5 9 10 5 fluorescence-activated cell sorting-sorted FO B cells were cultured in a 48-well plate (TPP, Trasadingen, Switzerland) in 600 µL Roswell Park Memorial Institute complete medium in the presence of absence of 10 µg mL À1 Escherichia coli lipopolysaccharide (0222B4; Sigma, Merck Sharp and Dohme Corp). Cells were kept at 37°C in a 5% CO 2 humidified incubator. On days 3 and 6 of stimulation, cells were harvested for fluorescence-activated cell sorting analysis and supernatant was taken to measure the IgM titer by ELISA.

RNA-sequencing and analysis
Sequencing libraries were constructed using the TruSeq Library Prep kit version 2 (Illumina, San Diego, CA, USA) from messenger RNA from purified MZ B cells from Gpr43 À/À (n = 3) and Gpr43 +/+ (n = 3) mice. Indexed sequence libraries were pooled for multiplexing. Sequencing was performed on an Illumina NextSeq 550 platform as 75-bp long single-reads. Running FastQC (https://www.bioinformatics.babraham.ac.uk/ projects/fastqc/) on the raw fastq files revealed the presence of adapters, poly-A tails and poly-G strings indicative of falsely confident calls on no signal rounds. Trim Galore's (http:// www.bioinformatics.babraham.ac.uk/projects/trim_galore/) autodetect adapter option was used to trim adapters, while -a and -a G{10} options were used to trim off poly-As and poly-Gs. Trimmed files were run through FastQC and multiQC to verify quality. Alignment program STAR was utilized to map reads onto the GRCm38 genome assembly and RSEM was utilized to calculate expression values. The expression values were read into R, where low-count genes (less than 1 CPM for 2 or more samples) were eliminated and intersample CPM values were normalized using the method of trimmed mean of M-values. Differential expression was determined using R's limma package. After adjustment for multiple comparisons, Gpr43 (adjusted P = 0.023) and Ceacam2 (adjusted P = 0.036) were the only two genes differentially expressed between the MZ B from Gpr43 À/À and Gpr43 +/+ mice.

Statistical analysis
When comparing two groups, the Mann-Whitney U-test was used. For ELISA results, two-way ANOVA and Sid ak's test were used to determine statistical differences at different titrations of serum.