A Gain-Of-Function Mutation in the Plcg2 Gene Protects Mice from Helicobacter felis-Induced Gastric MALT Lymphoma

Gastric mucosa-associated lymphoid tissue (MALT) lymphomas develop from a chronic Helicobacter infection. Phospholipase C gamma 2 (PLCG2) is important for B-cell survival and proliferation. We used BALB/c mice with a gain-of-function mutation in the Plcg2 gene (Ali5) to analyze its role in the development of gastric MALT lymphoma. Heterozygous BALB/c Plcg2Ali5/+ and wildtype (WT) mice were infected with Helicobacter felis (H. felis) and observed up to 16 months for development of gastric MALT lymphomas. In contrast to our initial hypothesis, Plcg2Ali5/+ mice developed MALT lymphomas less frequently than their WT littermates after long-term infection of 16 months. Infected Plcg2Ali5/+ mice showed downregulation of proinflammatory cytokines and decreased H. felis-specific IgG1 and IgG2a antibody responses. These results suggested a blunted immune response of Plcg2Ali5/+ mice towards H. felis infection. Intriguingly, Plcg2Ali5/+ mice harboured higher numbers of CD73 expressing regulatory T cells (Tregs), possibly responsible for impaired immune response towards Helicobacter infection. We suggest that Plcg2Ali5/+ mice may be protected from developing gastric MALT lymphomas as a result of elevated Treg numbers, reduced response to H. felis and decrease of proinflammatory cytokines.


Introduction
Mucosa-associated lymphoid tissue (MALT) lymphomas are extranodal marginal zone B-cell lymphomas. There is a strong association between Helicobacter pylori (H. pylori) infection and LASvendi GmbH, Soest, Germany) was continuously available. We used heterozygous BALB/c Plcg2 Ali5/+ mice for infection experiments, because homozygous Plcg2 Ali5/Ali5 show strong inflammatory reactions on paws, eyes and internal organs and could not been used for longterm infection studies. During housing, infected animals were monitored 2-3 times a week for health status (e.g. piloerection, reduction of weight) and got points for it. Decision of euthanasia by cervical dislocation (after CO 2 narcotization) was done, if animals got 7 points. The criteria list for decision of euthanasia is shown in S1 Table. Animals which died or have been killed due to the health status during the study (within 6 months after infection) were not included in the analysis. One mouse in group 1, 2 mice in group 3 and 5 mice in group 3 died unexpectedly within 6 months after infection and were not euthanized.

Infection of mice
Three groups of female BALB/c Plcg2 Ali5/+ and WT mice were orally infected with H. felis. Bacteria were harvested in sterile PBS and mice were gavaged with 2 x 10 8 bacteria in 200 μl suspensions on days 1, 3 and 5. Mice were narcotized with CO 2 before cervical dislocation and analyzed at 6, 12 and 16 months after infection (group 1: n = 11; group 2: n = 16; group 3: n = 17 per genotype). A fourth group was infected once with 2 x 10 8 bacteria / mouse every 3rd month for 12 months (n = 11 per genotype). Uninfected animals served as controls. To confirm that mice were infected with comparable bacterial load, we performed quantitative PCR of stomachs from mice which were infected for 16 months. H. felis specific primers against 16S rDNA were used (see SI Material and Methods).

Preparation of bacterial lysate
Helicobacter felis strain CS1 (ATCC49179) was purchased (American Type Culture Collection). H. felis was harvested in PBS, centrifuged at 3000 x g for 10 min and resuspended in 25 μl 1 x TES-Buffer (Sigma-Aldrich) plus 1 μl Ready-Lyse Lysozyme Solution (Epicentre Biotechnologies) and incubated for 60 min at room temperature. After a second centrifugation step at 4°C for 10 min and 20000 x g the supernatants were collected. Protein concentration was determined by BCA Protein Assay Kit (Thermo Fisher Scientific) according to the manufacturer´s instructions.
Collection and preparation of mouse blood and tissue samples RNA from whole blood was collected with RNAprotect Animal Blood Tubes and Kit (Qiagen) according to the manufacturer´s instructions. MNCs were prepared from spleen by disruption tissue with a pair of tweezers in sterile PBS, which were then processed for Ficoll-Hypaque density gradient centrifugation (FICOLL-Paque Plus, Amersham Biosciences). Residual red-blood cells were lysed by red-blood cell lysis buffer [155 mM NH 4 CL, 10 mM KHCO 3 , 0.1 mM EDTA, pH 7.5] for 5 min at room temperature. MLN cells were disrupted as described for spleen cells.
Immunohistochemical stainings were reviewed in a blinded manner by a reference pathologist (M.S.). Histology was examined as i) healthy gastric mucosa, ii) gastric mucosa with lymphoid aggregates indicating chronic gastritis, iii) gastric MALT lymphoma with infiltrates of centrocyte-like cells and replacement of gastric glands (including lymphoepithelial lesions (LEL) and lymphoepithelial destruction (LED)), according to Isaacson and Wright. [20] Immunocytological CD3 and Foxp3 stainings were performed on formalin-fixed paraffin embedded (FFPE) spleens of uninfected and infected mice according to the avidin-biotin complex (ABC) method as previously described [21]. Following primary antibodies were used: anti-CD3 1:500 (clone SP7; Thermo Fisher Scientific) and anti-Foxp3 1:100 (clone FJK-16s; Affymetrix/eBioscience, Inc.). The total number of Foxp3 + Tregs, in 4-6 T-cell areas of the white pulp from each spleen, was determined microscopically in all cases by a single investigator to warrant homogeneity of data collection (Leica DMRB; Leica Microsystems GmbH). By using ImageJ v1.47 software (freeware, National Institutes of Health), the corresponding area of CD3 + T-cells was calculated and Foxp3 + Tregs were determined per 1000 μm 2 .

RNA extraction and reverse transcription
Total RNA extraction from peripheral blood was prepared by using RNeasy Protect Animal Blood Kit (Qiagen). 1 μg of RNA was reverse transcribed to cDNA using Omniscript Reverse Transcription Kit (Qiagen). RT 2 First Strand Kit (SABioscience) was used to transcribe 1 μg RNA from the blood of mice of group one for the performance of the RT 2 Profiler PCR Array. All procedures were carried out according to the manufacturer´s guidelines.

PCR Array of proinflammatory cytokines
To assess proinflammatory cytokines in mouse peripheral blood we used the RT 2 Profiler PCR Array (SABioscience) according to the manufacturer´s instruction. The following amplification protocol was performed on the Mastercycler 1 ep realplex 2 S (Eppendorf AG): 1 x 95°C for 10 min followed by 40 x (95°C for 15 s and 60°C for 1 min). Web-based software of SABioscience/ Qiagen GmbH was used (http://pcrdataanalysis.sabiosciences.com/pcr/arrayanalysis.php).
For determination of H. felis-specific IgG1 and IgG2a antibody levels, ELISA plates were coated with soluble H. felis extract [10 μg/ml in PBS]. Serum samples were applied in a dilution of 1:100 (IgG1) and 1:10 (IgG2a). Secondary antibodies were used as previously described. Concentration of H. felis-specific IgG1 and IgG2a was estimated in relation to a pooled standard serum of infected mice in a titrated 1:4 dilution which was arbitrarily set at 10 labor units/ ml [LU/ml].

Statistical analysis
Differences in MALT lymphoma incidence was calculated with Fisher´s Exact Test. PCR array data were evaluated by the web-based analysis tool of SABioscience/Qiagen GmbH and calculated with t-test or two-way analysis of variance (ANOVA) (http://www.R-project.org). Statistical analysis of two-tailed t-test or Mann-Whitney U-test was performed with GraphPad Prism software version 5.01 (GraphPad Software, Inc.). Statistical significance was assigned at p 0.05.

H. felis-infected Plcg2 Ali5/+ mice are protected from the development of MALT lymphomas
Host gene polymorphism influences the incidence of gastric MALT lymphomas [8][9][10][11]. We investigated whether a gain-of-function mutation in the Plcg2 gene may have an effect on MALT lymphoma development after infection with H. felis. Three groups of Plcg2 Ali5/+ and WT mice were infected and followed for 6, 12 and 16 months, respectively. In group one, no difference could be shown for both genotypes after 6 months of infection (Table 1, group 1): 6 out of 10 (60%) Plcg2 Ali5/+ mice and 6 out of 11 (54%) WT mice (Fisher´s exact test, p = 1) developed chronic gastritis with lymphoid aggregates, but no gastric MALT lymphomas in either group. 12 months after infection (group 2), gastric MALT lymphomas were observed in 5 out of 15 (33%) mutant Plcg2 Ali5/+ mice, as compared to 10 out of 15 (67%) (Fisher´s exact test p = 0.1431) in WT mice. At 16 months of follow-up (group 3), H. felis-infected Plcg2 Ali5/+ mice showed significantly less gastric MALT lymphomas compared to their infected WT littermates (7 out of 14 (50%) vs. 14 out of 15 (93%)) (Fisher´s exact test, p = 0.0142). Noninfected mice developed neither gastritis nor MALT lymphomas (S2 Table). Fig 1 shows the histology of a representative mouse with chronic gastritis, LEL (lymphoepithelial lesions) as well as LED (lymphoepithelial destruction) as characteristics of MALT lymphomas. Subsequently, we investigated the bacterial load in both H. felis infected mice strains of group 3 by quantitative Real-Time PCR and could show that Plcg2 Ali5/+ mice had the same bacterial load as compared to WT mice (S1 Fig). We conclude that the lower MALT lymphoma development in Plcg2 Ali5/+ mice was not the cause of a lower colonization rate.
We also asked whether repeated H. felis infection resulted in a higher prevalence of MALT lymphomas. To investigate this question, we re-infected a fourth group of mice in a 3 months interval for up to one year (Table 2). At the end of the experiment, we found that repeated infection with H. felis resulted in increased incidence of MALT lymphomas in both genotypes (83% Plcg2 Ali5/+ mice vs. 100% WT mice) (Fisher´s exact test p = 1).

Proinflammatory genes are downregulated in H. felis-infected Plcg2 Ali5/+ mice
To investigate whether the lower susceptibility of Ali5 mice for MALT lymphoma development after H. felis infection is reflected in a differential gene expression profile, we performed an array testing for mRNA expression of 91 different proinflammatory genes. Peripheral whole blood samples of infected Plcg2 Ali5/+ mice and WT littermates (followed-up for 6 months) as well as of uninfected mice were available for analysis. As expected for an autoimmune-prone mouse strain, non-infected Plcg2 Ali5/+ mice revealed increases in the expression level of several proinflammatory cytokines and cell surface molecules as compared to uninfected WT mice (Table 3). Especially TNFSF13 (Tumor necrosis factor ligand superfamily member 13) and S100A8 (S100 calcium binding protein A8, calgranulin A) are described to be overexpressed in autoimmune diseases such as systemic lupus erythematodes (SLE) [18,23]. However, this picture changed after Helicobacter infection. Surprisingly, 6 months after infection, several proinflammatory genes were downregulated in peripheral blood cells of H. felis-infected Plcg2 Ali5/+ mice as compared to infected WT mice, including two important inflammatory cytokines, IFNγ and IL-1a. Within the Ali5 genotype we also detected several genes which were downregulated in infected Plcg2 Ali5/+ mice as compared to their uninfected Plcg2 Ali5/+ littermates, such as the CXC chemokine subfamily members CXCL1, CXCL9 and CXCL10.  The two-way ANOVA (analysis of variance) method was used to distinguish between gene profiles, which were dependent on the genotype or the infection status (S3 Table). According to the genotype status and independent of H. felis infection (Plcg2 Ali5/+ versus WT mice), CCL3 (chemokine (C-C Motif) Ligand 3), CD40 and TNFRSF13c were downregulated, whereas CXCR1 (chemokine (C-X-C motif) receptor 1), S100A8 and TLR2 (toll-like receptor 2) were upregulated. Independent of the genotype, infection with H. felis, resulted in downregulation of several genes, such as CXCL5 and TNFRSF13c (infected versus uninfected mice).
To validate the array results, we performed quantitative PCR analysis for TNFRSF13c and S100A8, which were the most significant regulated genes. By this, we confirmed the data shown in S3 Table that these genes were significantly regulated, dependent on genotype status (S4 Table). Protection from MALT lymphoma in Plcg2 Ali5/+ mice is not a result of a distinct B-cell defect To investigate, whether defective Plcg2 Ali5/+ B-cells were responsible for impaired MALT lymphoma development and downregulation of proinflammatory genes, we performed several experiments to analyze different aspects of B-cell function. To test for proliferation, purified spleen CD45R/B220 + B-cells from H. felis-infected and uninfected mice were used for in vitro stimulation with different mitogens. S-phase B-cells (BrdU + cells) were used to analyze proliferating cells. We could not detect any differences between the percentages of S-phase B-cells with different stimulation agents in infected and uninfected cells (Fig 2A and 2B). Additionally, we observed comparable B-cell proliferation in Plcg2 Ali5/+ and WT mice after BCR stimulation. Next, we investigated in vitro whether B-cells of uninfected Plcg2 Ali5/+ mice show a physiological reaction to LPS and IL-4. It is known that activation of B-cells with LPS and IL-4 stimulation results in class switching from IgM to IgG1 [24,25]. Therefore we analyzed membrane IgG1 expression on B-cells after stimulation. Expression of membrane IgG1 on B-cells was inducible in both Plcg2 Ali5/+ and littermate B-cells after LPS and IL-4 mediated class switching (Fig 2C and 2D).

Plcg2 Ali5/+ mice show an impaired H. felis-specific antibody response
After having examined the inducibility of an Ig class switch in Plcg2 Ali5/+ and WT B-cells in vitro by analyzing membrane bound immunoglobulins, we sought to determine in vivo titers of serum antibodies. 12 weeks after H. felis infection, sera from both genotypes were obtained and total and antigen-specific immunoglobulin levels were analyzed by ELISA (Fig 3). After infection, total serum IgG1 and IgG2a levels were elevated in both infected genotypes as compared to their uninfected littermates (Fig 3A and 3B). Concerning H. felis-specific IgG1 and IgG2a, no antibodies could be detected in uninfected mice of both genotypes, demonstrating that crossreactivity to any other Helicobacter species potentially present in the microflora played no role in our investigation. Specific antibodies appeared after H. felis infection in both genotypes, but their titers were significantly lower in infected Plcg2 Ali5/+ animals. H. felis-infected Plcg2 Ali5/+ mice showed 3.5-fold lower IgG1 (p = 0.0307) and 10-fold lower IgG2a levels (p = 0.0016) in comparison to their infected WT littermates (Fig 3C and 3D).

Protection from MALT lymphoma correlates with higher frequency of regulatory T-cells
Depletion of Tregs in Helicobacter-infected mice results in increased gastric inflammation [26]. Higher Treg numbers could explain an impaired immune response towards H. felis. Therefore we measured numbers of Treg in spleen tissue. By using immunohistochemical staining of CD3 and Foxp3 in T-cells of uninfected and infected Plcg2 Ali5/+ and WT mice, we determined the number of Foxp3 + CD3 + Tregs per 1000 μm 2 of area composed of CD3 + T-cells (S2 Fig). Our results revealed significantly more Tregs in spleen tissue of uninfected and infected Plcg2 Ali5/+ mice as compared to WT mice (p = 0.0109 and p = 0,0451) (Fig 4A and 4B).

Discussion
We have found that H. felis-infected Plcg2 Ali5/+ mice showed significantly less gastric MALT lymphomas compared to their WT littermates after 16 months of infection. These results were somehow unexpected, because we had assumed that Plcg2 Ali5/+ mice might show a stronger gastric inflammatory response and thus an increased induction of MALT lymphomas after H. felis infection. This seemed likely, as Plcg2 activates NF-κB [16], a key molecule of MALT lymphoma development [28]. Furthermore, we found that repeated infection with H. felis resulted in increased incidence of MALT lymphomas in both genotypes and after 12 months of followup differences were no longer detectable. We suppose that mice infected regularly eliminate, at least in part, H. felis over a longer period of time and the inflammatory response can diminish. However, repeated infection with H. felis may result in a prolonged inflammatory stimulus that yields earlier and more MALT lymphoma development. This result emphasizes that development of gastric MALT lymphomas is strongly dependent on chronic Helicobacter infection and that repeated antigen-stimulation can overcome the influence of a protective host-related gene polymorphism.
We verified that uninfected BALB/c Plcg2 Ali5/+ mice show an increased proinflammatory cytokine profile in the peripheral blood. These results were in accordance with previously published data for heterozygous Plcg2 Ali5/+ mice with C3H background [18]. Especially TNFSF13 (Tumor necrosis factor ligand superfamily member 13) and S100A8 (S100 calcium binding protein A8, calgranulin A) are described to be overexpressed in autoimmune diseases such as systemic lupus erythematodes (SLE) [18,23]. However, in our model, this picture changed after Helicobacter felis infection. In comparison to WT littermates, H. felis-infected Plcg2 Ali5/+ mice showed less induction of cytokines, such as CCL3, CCL4, CCL5, CCL20 and CXCL1. We did not find a direct link between PLCG2 and cytokine production. Thus, these cytokines are known to be produced by gastric epithelial cells as a result of a Helicobacter infection [29][30][31][32][33][34][35][36]. In addition, H. pylori infection induces infiltration of IFNγ producing CD4 + TH1 cells in the gastric mucosa, which is responsible for the increased mucosal inflammation and destruction [37,38]. Also the proinflammatory cytokine IL-1a, which is known as an acute phase cytokine, plays an important role in early inflammation and is mainly produced by macrophages [39]. H. felis-infected Plcg2 Ali5/+ mice also showed less production of these two important inflammatory cytokines.
Less induction of cytokines in H. felis-infected Plcg2 Ali5/+ mice may be a hint for an impaired recognition of these mutated mice toward H. felis. In fact, we found that H. felis antigen-specific IgG1 and IgG2a titers in sera of Plcg2 Ali5/+ mice were significantly decreased.
The inadequate immune response to H. felis infection was not due to a general B-cell defect in BALB/c Plcg2 Ali5/+ mice. Therefore, our findings argue against a simple B-cell defect in Plcg2 Ali5/+ mice as being decisive for reduced MALT lymphoma development.
Rad et al. have previously shown that depletion of Tregs in Helicobacter infected C57BL/6 mice by an anti-CD25 mouse antibody results in severe gastritis and increased antigen-specific antibody titers [26]. Therefore we speculated that, vice versa, elevated Treg numbers in Plcg2 Ali5/+ mice could lead to decreased inflammatory response to Helicobacter infection and, as we have shown, decreased H. felis-specific antibody titers. In keeping, we show that Plcg2 Ali5/+ mice harbour significantly higher numbers of CD4 + /CD25 + /Foxp3 + Tregs in spleen tissue as compared to WT littermates. Kaparakis et al. have also shown increased H. felis-specific antibody titers after Treg depletion [40]. Tregs are able to inhibit B-cell antibody responses dependent or independent of T-cell stimulation [41]. Suppression of TH1/TH2 cells might be a possible effect that is responsible for reduced antigen-specific antibody titers in Plcg2 Ali5/+ mice.
There are conflicting data about the incidence of Tregs in inflammatory-autoimmune diseases. Some studies showed lower numbers of Tregs in autoimmune individuals with SLE [42,43] and other studies found even more Tregs in autoimmune patients with rheumatoid arthritis, SLE and Sjögren´s syndrome [44][45][46][47][48][49]. Yan et al. have shown that the immunosuppressive Treg function in SLE patients is impaired by IFNα-producing antigen-presenting cells (APCs). However, in vitro experiments show that Tregs of SLE patients are not per se dysfunctional. Tregs exhibited normal activity, if cultured with normal donor APCs [44]. These publications support our findings of both, elevated Treg numbers and increased proinflammatory cytokines in uninfected Plcg2 Ali5/+ mice. We suggest for our mouse model that the H. felis infection reconstitute the immunosuppressive function of Tregs. However, the exact regulatory mechanism is undetermined.
One of the most important factors for suppression of T-cell-mediated immune responses is the enzyme ecto-5`-nucleotidase (CD73). CD73 is expressed particularly by Tregs, but also by tumor cells, myeloid-derived and endothelial cells. This enzyme is known to generate antiinflammatory adenosine that protects tissue damage [50][51][52][53][54]. The importance of CD73 expression was demonstrated in a study by Alam and colleagues [27]. They showed that H. felis infected CD73-deficient mice have increased levels of proinflammatory cytokines and severe gastritis [27]. These data are in line with our findings, that Plcg2 Ali5/+ mice had not only higher numbers of CD4 + /CD25 + /Foxp3 + Tregs in spleen tissue as compared to WT mice, but also higher numbers of CD73 + Tregs. Our results support the hypothesis of Kaparakis et al. that Tregs have a potential role in inhibition of lymphoma development after a lengthy infection time with Helicobacter.
Although Helicobacter spp. induce specific immune responses, the immune system of the host often fails to clear the infection and the bacterium can persist lifelong. Interaction of bacterial virulence factors and specific immune answer of the host influence the outcome of the infection. Several studies showed that different host gene polymorphisms in genes such as IL-1, CTLA4 and GST T1 are associated with increased inflammatory responses and consecutively with higher incidence of MALT lymphomas [8][9][10][11]]. Our mouse model shows for the first time that a gene polymorphism could also be responsible for protection from MALT lymphoma development.
In conclusion, we found hints that elevated Treg numbers in Plcg2 Ali5/+ mice play a decisive role in the outcome of Helicobacter infection. Although Plcγ2 is normally involved in activation of immune cells, we could show that Plcg2 Ali5/+ mice were protected from developing gastric MALT lymphomas after long-term infection with H. felis, most likely as a result of increased numbers of immunosuppressive CD73 + Tregs.
Supporting Information S1 ARRIVE Checklist. The ARRIVE Guidelines Checklist. (DOCX) S1 Fig. Plcg2 Ali5/+ and WT mice of group 3 had the same H. felis colonization rate. The bacterial load in gastric tissue of Plcg2 Ali5/+ (n = 10) and WT mice (n = 12) was measured by quantitative Real-Time PCR, 16 months after H. felis infection. Data represent mean ± SEM and were calculated using Student´s t-test. ns = not significant. (TIF) S2 Fig. CD3 and Foxp3 immunohistochemistry of a murine spleen. A representative (A-B) CD3 and (C-D) Foxp3 immunohistochemical staining of a spleen of an uninfected mouse. The same five T-cell areas (in CD3 and Foxp3 staining) of the white pulp were determined microscopically and total Foxp3 + Tregs were determined with a magnification of x40. By using Ima-geJ v1.47 software total CD3 area was calculated (thin yellow line) and Foxp3 + Tregs were determined in relation to 1000 μm 2 (CD3 + T-cell area). (A and C) Scale bars show 200 μm with an original magnification of x2,5 or (B and D) 20 μm with an magnification of x40. (TIF) S1 Materials and Methods. (DOCX) S1