Genetic dissection of a major haplotype associated with arthritis reveal FcγR2b and FcγR3 to act additively

Abstract

A haplotype with tightly linked Fc gamma receptor (FcγR) genes is known as a major locus controlling immune responses and autoimmune diseases, including arthritis. Here, we split a congenic fragment derived from the NOD mouse (Cia9) to study its effect on immune response and arthritis in mice. We found that arthritis susceptibility was indeed controlled by the FcγR gene cluster and a recombination between the FcγR2b and FcγR3 loci gave us the opportunity to separately study their impact. We identified the NOD‐derived FcγR2b and FcγR3 alleles as disease‐promoting for arthritis development without impact on antibody secretion. We further found that macrophage‐mediated phagocytosis was directly correlated to FcγR3 expression in the congenic mice. In conclusion, we positioned FcγR2b and FcγR3 alleles as disease regulatory and showed that their genetic polymorphisms independently and additively control innate immune cell activation and arthritis.

The closely linked polymorphic Fcgr2b and Fcgr3 NOD alleles promote the development of arthritis and enhance the inflammatory response capacity.

Introduction

The etiology of chronic autoimmune diseases, such as rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE), are polygenic diseases with numerous loci, each with small effects. However, a tightly linked region including the low‐affinity FcγR genes has shown clear importance [1, 2].

Linkage analyses in mouse models have revealed a major regulatory effect from a locus on chromosome 1 containing low‐affinity FcγR genes (Cia9), associated with both antibody production and disease severity [3, 4]. However, identification of the underlying gene variants has so far been unsuccessful. The problem with this region on chromosome 1 is its high density of polymorphic genes in both mice and humans, of which many genes can have potential importance for the regulation of chronic inflammation [5, 6]. Despite the strong genetic association to autoimmune diseases in this region, no specific FcγR encoded polymorphisms have been identified as disease causative. Instead, the autoimmune disease regulatory function of FcγRs has been mostly studied using different KO mouse models, which has led to some confusion. One problem has been linked genes from embryonic stem (ES) cells. Studies with FcγR2b KO mice show that the surrounding genes rather than FcγR2b regulate immune response [6, 7, 8]. Even when syngenic ES cells are used, effects from genetic manipulation can be seen [9], which makes it difficult to mimic the naturally selected polymorphisms that could regulate disease. Analyzing the individual effect from this composite set of highly polymorphic genes can therefore better be assessed in a more biological setting through natural polymorphisms by genetic mapping of phenotypic associations to relevant inflammatory disorders [5, 10, 11, 12].

Previous studies pinpointed the Cia9 locus on chromosome 1 as being the major locus besides the MHC region to be associated with collagen‐induced arthritis (CIA) [3], the classical mouse model for RA. The association of Cia9 with arthritis was confirmed using a genome‐wide mouse heterogeneous stock analysis, with the contribution of eight inbred mouse strains [13, 14]. CIA severity and the levels of anti‐CII antibodies were significantly increased in a B10.Q mouse with a NOD‐derived Cia9 congenic fragment, which included the FcγR locus [3].

Here, we aimed to better understand the genetic control of chronic inflammation in the Cia9 region by positioning the causative gene(s). We have established four informative overlapping sub‐congenic Cia9 strains (Cia9b, Cia9c, Cia9i, and Cia9k), identifying the FcγR gene cluster and allowing to dissect the function of FcγR2b and FcγR3 independently of each other. We found that FcγR2b and FcγR3 genes additively control inflammatory responses and arthritis.

Results

Congenic mapping of Cia9 identified the arthritis regulatory region to a <1Mb fragment

We have previously shown that a chromosome 1 congenic locus (Cia9, 10 Mb in length) from NOD mice introgressed onto the B10.Q background mediates increased susceptibility to CIA [3, 14]. To identify the underlying loci, we further refined the locus and established four overlapping Cia9 sub‐congenic mouse strains (Cia9b, Cia9c, Cia9i, and Cia9k).

The Cia9b fragment covers the region above the FcγR gene cluster, whereas Cia9c covers the region below the FcγR gene cluster and contains several genes from the signaling lymphocyte activation molecule (SLAM) family, in which polymorphisms have been shown important in maintaining tolerance in lupus [6, 10]. The Cia9i fragment contains the highly polymorphic NOD‐derived FcγR gene cluster, consisting of FcγR2b, FcγR4, and FcγR3. The smaller Cia9k fragment harbors NOD FcγR2b and FcγR4 alleles (Fig. 1).

Figure 1.

Overview of the Cia9 sub‐congenic fragments compared to Fcγ2b KO mice. The positional information of the FcγR2b KO mice was adapted from [7]. The locations, indicated in mega base pairs (Mb), are based on the mouse genome assembly GRCm38/mm10. The different bars represent different fragments. The borders of the congenic fragments are defined by the respective markers, and the region outside applies to the B10.Q background. Cia9 (163.5‐173), Cia9b (163.5‐170.4), Cia9c (171.3‐173.0), Cia9i (169.3‐171.5), Cia9k (169.3‐171.0). Within the Cia9 fragment, the FcγR and SLAM/CD2 gene clusters are highlighted. The arthritis regulatory region (≤1Mb), with corresponding protein‐coding genes, is indicated in red. The dashed blue line shows the recombination between Cia9i and Cia9k, identified by the corresponding SNPs (rs49184774 and rs50943911, sequenced using primer pair 1 (F: TGATTGTTGCCAGGGCTAGG, R: AATGAACCTCCTCTGCAGGC) and primer pair 2 (F: CTGCTGGGTGAAACAAAGGC, R: AGATGGCGGTACTAGGGTGT), respectively). The genes in bold are located in the Cia9k fragment, whereas the rest was contained within the Cia9i fragment.

We then tested which of the recombinant congenic fragments conferred the arthritis susceptibility seen in the original Cia9 fragment by using the type II collagen (CII) specific T and B cell‐dependent CIA model, as well as the T and B cell‐independent collagen antibody‐induced arthritis (CAIA) model [15, 16]. No differences in arthritis development were observed between Cia9b or Cia9c congenic mice and WT mice (Supporting Information Fig. S1), restricting the disease‐regulating interval to less than 1 Mb of the Cia9 locus, i.e., the Cia9i congenic containing the FcγR cluster.

After screening a high number of meiosis, we obtained a recombination within the FcγR gene cluster, excluding the NOD FcγR3 allele from the fragment (Cia9k), (Fig. 1). To investigate the role of the different FcγRs in arthritis development, CIA was induced in WT, Cia9i, and Cia9k mice, and in FcγR2b KO and FcγR3 KO mice for experimental control (Fig. 2A and B). In agreement with previous studies, FcγR2b KO mice developed severe arthritis, whereas FcγR3 KO mice were completely resistant [17, 18, 19, 20]. Due to disease severity, FcγR2b KO mice had to be sacrificed before the second injection with CII. Compared to WT mice, Cia9i congenic mice developed more severe arthritis with earlier disease onset. The arthritis severity of Cia9k mice was milder than Cia9i mice, but more severe compared to WT mice. Despite these differences, similar serum levels of anti‐CII antibodies were analyzed in the congenic mice at day 21 and 57 after immunization (Fig. 2C–E). The serum levels of anti‐CII IgG2b were elevated in FcγR2b KO mice, but which was related to arthritis severity rather than a direct effect on B cell response.

Figure 2.

CIA susceptibility of Cia9i, Cia9k mice, and WT mice. Mice were immunized with CII on day 0 and day 35 and were monitored macroscopically for signs of arthritis. Mean arthritis score (A) and incidence (B) of WT, Cia9i, Cia9k, FcγR2b KO, and FcγR3 KO mice. FcγR2b KO mice had to be sacrificed on day 35 due to disease severity. At day 21 and day 57 after CIA induction, serum samples were collected to assess total Ig (C), IgG1 (D), and IgG2b (E) levels of anti‐CII antibodies, correlated to arthritis development. The values between brackets (A) indicate the number of mice that developed arthritis out of the total number of animals in the experiments. Data show mean + SEM (A) or mean (B–E) and represent pooled data of two individual experiments for WT, Cia9i, and Cia9k mice. Two‐way ANOVA with Tukey's multiple comparison (A), Fisher's exact test among congenics and Chi‐Square test for comparison with FcγR2b KO (B) and Mann‐Whiney U test (C–E) were used and differences were considered statistically significant when p < 0.05 for a 95% confidence interval. Different symbols indicate statistical significance between Cia9i and WT mice (^* p < 0.05, ^** p < 0.01, ^*** p < 0.001), Cia9k and WT mice (^# p < 0.05, ^## p < 0.01), Cia9i and Cia9k mice (^$ p < 0.05), and between FcγR2b KO mice and Cia9i/Cia9k/WT mice (^§ p < 0.05, ^§§ p < 0.01, ^§§§ p < 0.001, ^§§§§ p < 0.0001). The flat lines (A,B) indicate multiple timepoints, and the connected line (A) represent the Area under the curve at day 40–56.

In summary, mice carrying the NOD‐derived FcγR gene cluster (Cia9i) or a part of the FcγR gene cluster (Cia9k) were more susceptible to CIA arthritis disease development compared to WT mice, whereas no differences in antibody levels were observed. Therefore, these results show that FcγR2b and FcγR3 act in concert to determine the magnitude of inflammatory effector cell responses.

Conserved FcγR haplotypes

Aiming to study variations in a multiple genome comparison across the FcγR genes, public data from the Welcome Trust mouse genome project was assessed that consists of 30 common laboratory strains, including the reference genome (C57BL/6J), and 7 wild mouse strains [21, 22]. A total of 4020 SNPs was found in the FcγR region (170.9–171.07) that differed between the 37 strains. Mus Spretus and Mus Castaneus were the strains that differed the most, as expected, since they are distant from Mus Musculus strains. In fact, 1920 out of the 4020 SNPs were unique to either Mus Spretus or Mus Castaneus or were only shared by the two. Looking closer at the remaining 2100 SNPs in the FcγR region, 1239 SNPs (of which 9 are non‐synonymous coding) separated the 35 strains into two distinct haplotype regions covering FcγR2b and FcγR4 (Supporting Information Table S1). Twelve mouse strains, including C57BL/10J and the reference genome, shared haplotype I derived from the M. musculus molossinus (MOLF/EiJ). The M. musculus musculus derived haplotype, haplotype II, was shared by 22 mouse strains including NOD and four wild mouse strains (Supporting Information Table S1). In the six laboratory strains carrying alleles from haplotype I upstream of Fc γ R3, a recombination has occurred with a change to haplotype II: the BALBc, CBA, two DBA strains, and two B57 strains, suggesting that the haplotype polymorphism was selected in the wild mouse population. This suggests that the haplotypes, now splitted in our congenic strains, have been conserved by strong natural selection.

Polymorphisms of the FcγR cluster regulate the inflammatory arthritis effector phase

To examine the inflammatory and not autoimmune phase of arthritis development, we used the CAIA model. Mice with the NOD‐derived FcγR gene cluster (Cia9i and Cia9k) were more susceptible to CAIA, as compared to WT mice (Fig. 3A and B). Disease development before LPS injection, causing antibody‐induced joint inflammation, was elevated in Cia9i and Cia9k congenic mice compared to WT mice (Fig. 3A). This effect was enhanced after LPS stimulation (day 7), increasing inflammatory cell infiltration.

Figure 3.

Polymorphisms in the Cia9i and Cia9k fragment associated with development of antibody‐mediated effector phase of arthritis. Mice were injected i.v. with 4 mg of anti‐CII mAbs cocktail (M2139+CIIC1+CIIC2+UL1) on day 0 and boosted with LPS i.p. on day 7. Mean arthritis score (A) and incidence (B) of WT, Cia9i, and Cia9k mice. The values between brackets (A) indicate the number of mice that developed arthritis out of the total number of animals in the experiments. Data show mean+SEM and represent pooled data of two independent experiments. Two‐way ANOVA with Tukey's multiple comparison (A) and Fisher‘s exact test (B) were used and differences were considered statistically significant when p<0.05 for a 95% confidence interval. The flat lines indicate multiple timepoints. Differences between WT and Cia9i mice: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Differences between Cia9i and Cia9k mice: ^$$ p < 0.01, ^$$$$ p < 0.0001. Differences between WT and Cia9k mice ^# p < 0.05, ^## p < 0.01, ^#### p < 0.0001.

Interestingly, Cia9i mice developed arthritis with higher frequency and severity as compared with the Cia9k mice. To study a possible FcγR3 independent effect of the congenic fragment, we first investigated ROS‐induced phagocytosis of Daudi cells, showing a difference, as compared to WT, by the Cia9i but not the Cia9k fragment (Supporting Information Fig. S2A–D). To investigate whether the Cia9i fragment contained other arthritis regulatory genes outside the FcγR cluster, we used the FcγR independent mannan‐induced psoriasis (MIP) model [23]. No differences in disease development were observed between Cia9i congenic and WT mice (Supporting Information Fig. S2E). Thus, we conclude that there is no other major effect than the FcγR polymorphism in the congenic fragment that could explain the enhanced arthritis seen in both Cia9i and Cia9k.

No observed effect on B cell function

To investigate why Cia9i and Cia9k congenic mice were different in arthritis susceptibility, we analyzed the immune cell populations in spleens from naïve mice but found no differences (Supporting Information Fig. S3). A major candidate for the arthritis susceptibility in Cia9 mice is the FcγR2b gene, which is expressed on myeloid cells and B cells, with B cells lacking expression of FcγR3. It has previously been shown that activated B cells from NOD mice have lower FcγR2b expression than C57BL/6.J mice[24]. This is also true for our congenic mice, both gene and protein expression were lower in LPS‐activated CD3⁻CD19⁺ B cells in vitro as compared with WT B cells, whereas no differences for naïve B cells were observed (Supporting Information Fig. S4A–C). To determine the ability to produce antibodies in vitro, we stimulated total spleen cells and MACS sorted CD3⁻CD19⁺ B cells from CIA induced mice in vitro with OVA, LPS, CII or medium for 5 days, but no differences in anti‐CII antibody production between WT and Cia9i mice were observed (Supporting Information Fig. S4D).

Despite lower expression of FcγR2b on activated B cells from Cia9i and Cia9k mice, no differences in in vitro antibody production by CIA primed B cells were observed, nor could any difference in IgG antibody response after CII immunization be observed (Fig. 2C–E). We conclude that B cells may not be a major mediator of the FcγR gene cluster regulated control of arthritis development.

Increased FcγR3 expression on Cia9i macrophages regulates effector function

To determine the FcγR2b expression on macrophages, we isolated thioglycolate‐elicited peritoneal macrophages (TpMFs) and analyzed gene and protein expression (Figs. 4A–F). The FcγR2b gene expression was drastically reduced in TpMFs of Cia9i and Cia9k congenic mice compared to WT mice. Both Cia9i and Cia9k mice had lower FcγR2b protein expression on un‐stimulated and in vitro LPS stimulated TpMFs. We also observed elevated gene expression of Fc γ R3 on macrophages from Cia9i mice compared to WT. Moreover, elevated levels of FcγR3 protein were found on Cia9i TpMFs compared to WT and Cia9k TpMFs, whereas Cia9k mice showed lower FcγR3 protein expression on in vitro LPS stimulated TpMFs compared to WT TpMFs. No differences in FcγR4 gene or protein expression levels were observed between NOD and B10.Q‐derived congenics. These gene and protein expression differences of the individual FcγRs between genotypes were also found on naïve macrophages (Supporting Information Fig. S5).

Figure 4.

FcγR3 expression levels on macrophages regulation of antibody‐dependent cellular phagocytosis (ADCP). (A–F) Gene and protein expression of FcγR2b (A,B), FcγR3 (C,D), and FcγR4 (E,F) on thioglycollate‐elicited peritoneal macrophages (TpMFs) (CD11b⁺F4/80⁺). (A,E) WT n = 7, Cia9i n = 9, Cia9k n = 9. (A) Gene expression of FcγR2b on FcγR2b KO mice (n = 6) was absent. (C) FcγR3 gene expression using primer/probe sets spanning exon boundary 2–3 (FcγR3) and 1–2 (FcγR3‐1) on un‐stimulated (blank; WT n = 6, Cia9i n = 8, Cia9k n = 9) or in vitro LPS stimulated (grey; n = 6) TpMFs. Horizontal line represents gene expression of FcγR3 KO mice (n = 4). (B,D,F) Representative histogram overlay and normalized protein expression of FcγR2b (B), FcγR3 (D), and FcγR4 (F) on un‐stimulated (left) and in vitro LPS stimulated (right) TpMFs. (B, D) FcγR2b (B) and FcγR3 (D) protein expression were normalized using the MFI from the respective KO mice. WT n = 6, Cia9i n = 8, Cia9k n = 8. (F) Un‐stimulated: WT n = 10, Cia9i n = 9, Cia9k n = 9. LPS: WT n = 11, Cia9i n = 13, Cia9k n = 13. (G) ADCP, Representative flow cytometry contour plots show phagocytosis of Daudi cells (CD11b⁺CFSE⁺) among total Daudi cells (CFSE⁺) with (+RTX, top) and without (‐RTX, bottom) rituximab by macrophages of FcγR3 KO, WT, Cia9i, and Cia9k mice. (H) Phagocytosis of Daudi cells. FcγR3 KO (3 KO) n = 4, WT n = 6, Cia9i n = 6, Cia9i n = 6. The data show mean ± SEM and represent a pool of two (A‐E,H) and three (F) individual experiments. (A,C,E) Mann–Whitney U test was used and differences were considered statistically significant when p < 0.05 for a 95% confidence interval. *p < 0.05, **p < 0.01, ***p < 0.001.

To determine how the observed FcγR protein expression affects FcγR mediated function on macrophages, we used a phagocytosis model. It is known that binding of activating FcγRs (FcγR3 and FcγR4), with pathogen‐bound IgG, directly mediates clearance of the pathogen by degranulation of cytotoxic cells and phagocytosis, whereas FcγR2b inhibits the function of activating FcγRs [25]. With the Cia9k congenic mouse, excluding NOD FcγR3, we investigated antibody‐dependent cellular phagocytosis (ADCP) of rituximab labeled Daudi cells by macrophages from WT, Cia9i and Cia9k mice, using FcγR3 KO mice as control (Fig. 4G). Here we show that Cia9i macrophages, with increased FcγR3 expression, induced more phagocytosis compared to Cia9k and WT macrophages (Fig. 4H). As expected, phagocytosis by FcγR3 KO macrophages was reduced compared to that of the congenic macrophages.

We next compared the efficiency of FcγR mediated phagocytosis in vivo through depletion of regulatory T (Treg) cells with the anti‐CD25 antibody PC61, which is known to be dependent on FcγR3 but not FcγR2b [26]. We found that PC61 reduced the frequency of CD4⁺Foxp3⁺ Treg cells in peripheral blood in WT, Cia9i, Cia9k, and FcγR2b KO mice, but not in FcγR3 KO mice (Supporting Information Fig. S6A–C). Interestingly, the data correlates with our FcγR3 expression data for Cia9i macrophages. Cia9k mice with lower expression of FcγR, showed less efficient Treg cell depletion even when compared to WT mice. This effect was less pronounced in spleen cells 6 days after PC61 Ab. Nevertheless, both WT mice and Cia9i mice showed a reduction in CD4⁺Foxp3⁺ Treg cell levels compared to naïve mice. Moreover, Cia9i mice had fewer CD4⁺Foxp3⁺ Tregs in the spleen 6 days after PC61 Ab compared to WT and Cia9k mice (Supporting Information Fig. S6D–E).

Taken together, Cia9i and Cia9k macrophages had decreased expression of FcγR2b compared to that of WT mice, whereas Cia9i macrophages showed higher expression of FcγR3, which led to increased in vitro and in vivo phagocytosis.

NK cell function is altered by polymorphisms in FcγR3

Since our data indicated a role for FcγR3 in the enhanced arthritis susceptibility of Cia9i mice, we next studied NK cell function, solely expressing FcγR3. We found increased FcγR3 gene expression on CIA primed NK cells of Cia9i mice (Fig. 5A). Moreover, FcγR3 protein expression was upregulated in naïve and IL2 activated NK cells from Cia9i mice, whereas Cia9k mice showed lower FcγR3 protein expression compared to WT mice (Fig. 5B and C). IL‐2 activated NK cells were used for FcγR3 mediated antibody‐dependent cell cytotoxicity (ADCC) assays, with Cia9i NK cells showing more specific lysis at different effector/target ratios (E/T) compared to Cia9k and WT mice (Fig. 5D and E). The strains had similar NK cell frequencies and secreted similar amounts of IFNγ upon PMA/ionomycin activation of NK cells (Supporting Information Fig. S7).

Figure 5.

FcγR3 on NK cells regulating cytotoxicity. (A) Gene expression of FcγR3 on NK cells isolated from spleens of CIA primed WT (n = 19) and Cia9i congenic mice (n = 18) on day 28 after CIA induction. (B) Representative histogram overlay of FcγR3 protein expression on CD3⁻NKp46⁺ NK cells of naïve (left) and IL2 stimulated (right) spleen cells from Cia9k, WT, Cia9i mice, and FcγR3 KO mice. (C) FcγR3 protein expression shown as MFI on CD3⁻NKp46⁺ NK cells of naïve or IL2 stimulated spleen cells, normalized using the MFI from FcγR3 KO mice. WT n = 8, Cia9i n = 9, Cia9k n = 8. D) ADCC of RMA cells by IL2 activated NK effector cells. Percentage of killed RMA cells at effector/target ratio (E/T) 4/1 in the absence (top) and presence (bottom) of anti‐Thy1.2 antibody. E) Specific lysis of RMA target cells by IL2 activated NK cells at E/T ratio 4/1, 11/1, 33/1 and 100/1. FcγR3 KO mice were used as assay control. FcγR3 KO n = 10, WT n = 10, Cia9i n = 10, Cia9k n = 7. The data shown are mean ± SEM and represent a pool of 2 (C,E) and 3 (A) independent experiments. E/T 100/1 in (E) represents 1 individual experiment. Mann–Whitney U test was used and differences were considered statistically significant when p < 0.05 for a 95% confidence interval. *p < 0.05, **p < 0.01, ***p < 0.001, ^**** p < 0.0001.

These data show that activated NK cells from Cia9i mice have upregulated FcγR3 resulting in higher NK cell functionality compared to WT and Cia9k mice, arguing for a role for the FcγR3 polymorphism.

Discussion

The low‐affinity FcγR cluster is located in a conserved haplotype with a strong influence on autoimmune diseases. Here, we have identified the underlying polymorphisms in this haplotype by splitting the effect of the closely linked FcγR2b and FcγR3 genes in congenic mouse strains, Cia9i and Cia9k. This strategy identified both FcγR2b and FcγR3 as regulators of experimental arthritis, regulating independently of each other but contributing to arthritis development additively. Moreover, both genes from the conserved haplotype of mus musculus musculus promoted a pro‐inflammatory effect as compared to the corresponding haplotype from mus musculus molossinus in the B10 mouse.

The FcγRs play an essential role in inflammation and immune response and their functions are quite complex in different pathophysiologic settings. Although nomenclature differs between mice and humans, their function and binding specificities are remarkably similar [27]. The FcγR genes are highly polymorphic and associated with autoimmune diseases. In humans, it has been difficult to identify a disease regulatory polymorphism of FcγR2a and FcγR2b, orthologue of mouse FcγR3 and FcγR2b, due to a high degree of linkage disequilibrium. Of particular interest is that in the mouse the three low to intermediate FcγRs (FcγR2b, FcγR3, and FcγR4) are also strongly linked and inherited in a well‐conserved haplotype. In fact, different subspecies of wild mice have different haplotypes, and inbred mouse strains have inherited different wild mouse‐derived haplotypes [13, 22, 28]. The haplotype polymorphisms could be older than the mouse species as has been suggested for the adjacent SLAM locus [11, 29]. The haplotype from the Mus musculus musculus, common on the Eurasia continent, is today carried by NOD, MRL and NZB strains, which are often more susceptible to various autoimmune diseases. In contrast, the C57.Black strains carry a haplotype from the Mus musculus molossinus, which naturally occurs on the Japanese islands [28, 30]. The occurrence of different haplotypes in inbred strains can help to understand the biological role of FcγR, in particular since the locus in the human population is also polymorphic. To better understand the biologic impact of this genetic information, it is necessary to isolate and study the effect of the conserved haplotype as well as to split the haplotype in order to investigate the effect of single genes.

Using Cia9 congenic mice, we initially found that CIA severity and the levels of anti‐CII IgG1 antibodies were significantly increased in Cia9 congenic mice compared to littermate control mice, which mapped to the NOD FcγR locus [14]. However, since the Cia9 locus consisted of more than 150 genes, the impact of NOD‐derived genes other than FcγR2b could not be excluded. Aside from the FcγR gene cluster, Cia9 also contained the SLAM/CD2 gene cluster, which is important in maintaining tolerance in autoimmune diseases and has been linked to lupus [6, 10]. No effect on arthritis development in congenic mice devoid of the FcγR gene cluster was observed, ruling out the role of the big SLAM/CD2 gene cluster in disease development, and the disease regulatory gene(s) were isolated to the FcγR region. A split recombination within the region showed that the FcγR2b and FcγR3 genes jointly and additively cause the effect on arthritis. However, we cannot exclude an influence of additional genes within these small fragments. Luan et al. [24] described several effects contrasting to the present data but these variabilities are most likely dependent on the older data being based on a very large (25 cM) congenic fragment containing many other immune regulatory genes. Another limitation of the study is that we have not explored the full potential of the FcgR polymorphism on inflammatory responses as we have only investigated selected disease models, activation inducers, and cell types.

With the recombination between the FcγR2b and the FcγR3 loci, we were able to study the single effect of FcγR2b and the combined effect of FcγR2b and FcγR3 on inflammatory responses. We showed that the FcγR2b and FcγR3 alleles operated in concert with an additive effect and primarily controlled the inflammatory effector phase of arthritis, but not the priming autoimmune phase.

To further investigate the role of polymorphic FcγR2b and FcγR3, FcγR‐dependent functions on B cells, NK cells, and macrophages were studied. Impaired FcγR2b expression in B cells in mice has been shown to influence antibody production in an antigen‐independent manner [31]. Despite lower expression of FcγR2b on Cia9i and Cia9k in vitro activated B cells, no significant differences in anti‐CII antibody secretion were observed. In contrast, the increased expression of FcγR3 derived from the NOD allele are likely to play a role in arthritis and showed a more pronounced phagocytosis in vitro and in vivo.

Nevertheless, despite the dramatic reduction of the arthritis prone congenic fragment there are still some additional genes in the FcγR gene cluster flanking region that might impact downstream functions. This has been indicated by the altered NK cell killing function in the Cia9k mice with an isolated NOD derived FcγR2b allele. Whereas the specific lysis by Cia9i and WT NK cells was linked to FcγR3 expression, that of Cia9k NK cells was not. Since FcγR3 expression in Cia9k NK cells was slightly reduced compared to WT NK cells, we expected lower or equal NK cell‐mediated lysis. Interestingly though, specific lysis by Cia9k NK cells was lower than that of Cia9i NK cells, but increased compared to that of WT NK cells. This implies possible involvement of other linked genes within the congenic fragment. The only gene within the Cia9k fragment that has been associated with NK cell‐medicated cytotoxicity is the activating transcription factor 6 (Atf6) [32]. It is possible that without NOD.Q FcγR3, NOD.Q Atf6 still controls cytotoxicity. Nonetheless, with our congenic mice, we were able to study the independent and additive effect of FcγR2b and FcγR3 on inflammation. Our congenic mice could provide a more physiological setting to study FcγR function.

In summary, we found that it is indeed the FcγR gene cluster of the Cia9 region that controls chronic inflammation, and that FcγR3 polymorphism on macrophage regulate effector functions. To conclude, it is the additive effect of genetic polymorphisms in FcγR2b and FcγR3 that regulate inflammation, most likely due to natural haplotype selection.

Materials and methods

Mice

Mice were bred and kept at the Karolinska Institute in Stockholm, Sweden (a specific pathogenic free unit with intraventilated cages). We used the 10‐Mb Cia9 congenic fragment [14], to generate the sub‐congenic fragments derived from NOD on to the B10.Q background (Fig. 1). FcγR2b KO [33] and FcγR3 KO mice [34], generated by gene targeting in 129‐derived ES cells and backcrossed for more than ten generations to C57BL/6.J, were obtained from Jackson Laboratory. They were further backcrossed into B10.Q background for more than ten generations in the MIR animal house and were used as experimental controls for various assays. Genotyping was performed using markers shown in Supporting Information Table S2. Haplotype variation analysis in the genomic region (170.9–171.07Mbp) that harbors FcγR2b, FcγR4, and FcγR3 is based on 4020 SNPs found in the Welcome Trust mouse genome project database (Welcome Sanger Institute, UK), comparing sequencing‐data from 37 different mouse genomes, including the reference genome C57BL/6J [21].

All experimental animal procedures were approved by the local ethics committees and were performed using B10.Q WT littermate control mice. All experiments were performed in a blinded manner with age‐ and sex‐matched groups randomly distributed in cages. Unless stated otherwise, 10–12 weeks old male mice were used for in vivo experiments. Animal model experiments, including serologic measurement, were performed following earlier described protocols for CIA [14], CAIA [16], and MIP [23].

Cells and antibodies

The RMA T leukemia cell line, used for NK cell‐mediated killing, was provided by Dr. M. Johansson (Karolinska Institute, Stockholm, Sweden) and the Daudi human B cell lymphoma cell line was provided by Dr. N. Nagy (MTC, Karolinska Institute, Stockholm, Sweden). The following antibodies were purchased from BD Biosciences (San Jose, CA) or Biolegend (San Diego, CA) and were used for analysis on a LSR‐II flow cytometer (BD Biosciences): anti‐CD45 (30‐F11), ‐CD3 (145‐2C11), ‐TCRβ (H57‐597), ‐CD4 (H129.19), ‐CD8 (53‐6.7), ‐CD19 (6D5), ‐CD45R/B220 (RA3‐6B2), ‐NK1.1 (PK136), ‐NKp46 (29A1.4), ‐CD25 (PC61.5), ‐CD11b (M1/70), F4/80 (BM8), ‐CD11c (HL3), ‐GR‐1 (RB6‐8C5), and ‐IFN (R46A2). Antibodies to iNOS (eBR2a) and Foxp3 (FJK‐16S) were from eBioscience (San Diego, CA). The FITC‐conjugated anti‐mouse FcγR antibodies: anti‐FcγR2b (AT130‐2), ‐FcγR3 (AT154‐2) and ‐FcγR4 (AT137), were generated in Southampton as previously described [35]), and were used at 10, 20, and 10 μg/ml, respectively. Viability of the cells was determined with LIVE/DEAD fixable near‐IR or violet dead cell stain kit (Invitrogen, Carlsbad, CA). Cancer cells were labeled with CellTrace CFSE or CellTrace Violet cell proliferation kit (Invitrogen, Carlsbad, CA). Data analysis was performed using the FlowJo software (TreeStar).

Quantitative real‐time PCR

Total RNA was extracted using Trizol and the PureLink RNA Mini Kit (Ambion, Thermo Fisher Scientific, Waltham, MA, Life Technologies, Inc., Foster City, CA). RNA was reverse transcribed to cDNA using the High Capacity cDNA Reverse Transcription Kit (ABI Applied Biosystems, Foster City, CA). Quantitative real‐time PCR (qRT‐PCR) was performed on a Bio‐Rad CFX96 System (Hercules, CA, USA) using TaqMan™ according to the manufacturer's protocol. TaqMan® Gene Expression Assays (Thermo Fisher Scientific) for FcγR2b (Mm00438875_m1 FAM), FcγR3 (Mm00438882‐m1 FAM, primers/probe spanning exon2‐3), FcγR3‐1 (Mm01290524‐m1 FAM, primers/probe spanning exon1‐2), FcγR4 (Mm00519988_m1 FAM), and the housekeeping genes Actin‐β (Mm00607939_s1 VIC) and GAPDH (Mm99999915_g1 VIC) were used. The relative expression of each FcγR gene was determined after normalization to both housekeeping genes and samples from naïve WT mice using the ΔΔCt method.

B cell analysis

Spleens were harvested and processed into single‐cell suspensions and B cells enriched by positive selection through CD19 microbeads according to the manufacturer's protocol (MACS, Miltenyi Biotec, Bergisch Gladback, Germany). The purified B cells (CD3⁻CD19⁺) were determined to be >90% purity by flow cytometric analysis and used for culture and qRT‐PCR.

Purified B cells and whole splenocytes were cultured in DMEM (Gibco, Thermo Fisher) supplemented with 50 U/ml penicillin, 50 mg/ml streptomycin, 10% HI fetal bovine serum, 50 μM β‐mercaptoethanol, and 10 mM HEPES buffer (complete DMEM) in the presence of 10 μM LPS, 10 μM CII, 10 μM OVA, or medium alone. Cells were stimulated for 5 days in 5% CO₂ at 37°C and anti‐CII Ab production was detected by ELISA using HRP conjugated anti‐kappa mAb as described above.

FcγR2b protein expression was determined on CD3⁻CD19⁺ or CD3⁻CD45R⁺ cells, on total spleen cells and purified B cells, from naïve Cia9i, Cia9k, and WT mice, using flow cytometry analysis of FITC conjugated anti‐FcγR2b (AT130‐2, 10 μg/ml). All flow cytometyric analysis followed published guidelines [36]. Cells were stimulated with LPS for 20 h to activate B cells or left un‐stimulated. FcγR2b KO mice were used as control. FcγR2b gene expression was determined on purified B cells and LPS‐stimulated B cells using qRT‐PCR as well.

Macrophage analysis

Macrophages were collected by peritoneal lavage or differentiated from BM cells. Naïve mice were injected i.p. with 1 ml 3–4% Brewer's thioglycollate (Difco, BD) and peritoneal lavage were taken 4–5 days after. For qRT‐PCR, cells were allowed to adhere to the surface of culture plates for 1–2 h in DMEM supplemented with 50 U/ml penicillin, 50 μg/ml streptomycin, and 10% HI FCS. Non‐adherent cells were washed away and the adherent cells were treated with Trizol for RNA extraction. The adherent cells consisted of more than 90% F4/80⁺CD11b⁺ cells, confirmed by flow cytometry analysis.

For BM‐derived macrophages (BMM), femurs were flushed and cells were cultured at 1.25×10⁵ cells/ml in complete DMEM containing M‐CSF for 7 days in 5% CO² at 37°C. All cells were F4/80⁺CD11b⁺. Thioglycollate‐elicited macrophages (TpMFs) and BMMs were used for flow cytometry analysis of FcγR proteins and for antibody‐dependent cellular phagocytosis (ADCP). For ADCP, cells were cultured in DMEM, 10% FCS at 37°C, 5% CO₂, and allowed to adhere ON.

Peritoneal cells were cultured in complete DMEM with or without 1 μg/ml LPS for 20 hours at 37°C, 5% CO². Macrophages were stained with FITC‐conjugated anti‐mouse FcγR2b (AT130‐2), FcγR3 (AT154‐2), or FcγR4 (AT137) antibodies on F4/80⁺CD11b⁺ pMQs and analyzed on a flow cytometer. Expression was measured as the MFI for each FcγR, using FcγR2b KO and FcγR3 KO mice as control for FcγR2b and FcγR3 expression, respectively. Oxidative burst assays were performed as earlier described [9].

Antibody‐dependent cellular phagocytosis

Antibody‐dependent cellular phagocytosis (ADCP) was determined by flow cytometry. Macrophages were seeded at 5×10⁴ cells/well into 96‐well plates ON. Target Daudi cells were labeled with 5 μM CellTrace™ CFSE for 5 min at 37°C and quenched with FCS for 5 minutes at room temperature. Cells were washed twice with pre‐warmed culture medium and resuspended in culture medium. Labeled Daudi cells were added to the macrophages at a 5/1 E/T ratio with or without rituximab (RTX) (provided by Inger Gjertsson, Rheumatology Unit, Sahlgrenska Hospital, Göteborg, Sweden) at 1 μg/ml for 4 h in 5% CO₂ at 37°C. After 4 h, cells were stained with F4/80 and CD11b and analyzed. ADCP was defined as the percentage of macrophages that had phagocytized. Phagocytosis was calculated as the percentage macrophages (CFSE⁺ CD11b⁺) among total target cells (CFSE⁺) per sample with and without RTX, and was normalized using the no RTX sample as negative control: (RTX sample – no RTX control) / (100% – no RTX control) × 100%.

NK cell culture

NK cells for qRT‐PCR were isolated from spleens of CIA induced Cia9i and WT mice. Non‐NK cells were labeled with a cocktail of biotin‐conjugated antibodies and anti‐biotin microbeads from the MACS NK cell isolation kit II, leaving unlabeled NK cells.

To activate NK cells, splenocytes from naïve Cia9i, Cia9k, WT, and FcγR3 KO mice were cultured for 4–7 days in complete α‐MEM (containing 50 U/ml penicillin, 50 mg/ml streptomycin, 50 μM β‐ME, 10 mM HEPES, 1 mM sodium pyruvate, 2 mM l‐glutamine, and 10% HI FCS) supplemented with human rIL‐2 (1000 U/ml; PeproTech) in 7% CO₂ at 37°C [37]. These cells were used as effector cells in antibody‐dependent cell‐mediated cytotoxicity (ADCC) assays.

FcγR3 protein expression was assessed on CD3⁻NKp46⁺ naïve spleen cells and IL2 stimulated splenocytes, using flow cytometry analysis of FITC conjugated anti‐FcγR3 (AT154‐2, 20 μg/ml). Expression was measured as the MFI, using FcγR3 KO mice as control.

Antibody‐dependent cell‐mediated cytotoxicity by NK cells

RMA cells were labeled with 5 μM CellTrace™ CFSE or CellTrace™ Violet (CTV) as described above. Labeled RMA cells were used as target cells at 5×10³ cells per well in 96‐well round‐bottom plates and pre‐incubated with 5 μg/ml anti‐Thy1.2 (clone 30‐H12, BD) for 10 min at 37°C, and washed with complete α‐MEM. NK effector cells were added to the wells containing RMA cells at effector/target (E/T) ratios 4/1, 11/1, 33/1, and 100/1 and incubated at 37°C for 4 h [37]. To determine the background cytotoxicity, culture medium instead of anti‐Thy1.2 was added as negative control. As positive control, target cells were heated for 30 minutes at 45°C [38]. ADCC was determined by flow cytometry. Cells were labeled with a fixable viability dye and NK cell (CD3⁻NKp46⁺) markers. Specific lysis was calculated with the number of tumor target cells killed per sample: (experimental sample – negative control) / (positive control – negative control) × 100%.

In vivo regulatory T cell depletion

CD25⁺ Treg cells were depleted in vivo using PC61.5 Ab [26, 39]. One day before Treg cell depletion, Cia9i, Cia9k, WT, FcγR2b KO and FcγR3 KO mice were bled by tail bleeding to establish their baseline CD4⁺Foxp3⁺ T cell population. At day 0, mice were injected i.p. with 250 μg anti‐CD25 (PC61.5) mAb. Blood was collected at day 1 and day 3 after PC61 Ab to determine the frequency of CD4⁺Foxp3⁺ T cells. At day 3, mice were given a second injection of 250 μg PC61. Peripheral blood and spleens were collected at day 6 after PC61 Ab and analyzed by flow cytometry for TCRβ⁺/ CD4⁺Foxp3⁺ cells.

Statistical analysis

GraphPad Prism software (San Diego, CA, USA) was used for statistical analysis. Arthritis severity and incidence between the groups of animals were analyzed using Two‐way ANOVA with Tukey's multiple comparison and the Fisher's exact test (and Chi‐Square test when comparing to FcγR2b KO mice) respectively. For all in vitro experiments, the Mann‐Whitney U test was used when comparing data from two groups. Significance was considered when P<0.05 for a 95% confidence interval.

Conflict of Interest

The authors declare no commercial or financial conflict of interest.

Peer review

The peer review history for this article is available at https://publons.com/publon/10.1002/eji.202048605.

Abbreviations

CAIA

collagen antibody‐induced arthritis

CIA

collagen‐induced arthritis

ES

embryonic stem

RA

rheumatoid arthritis

SLAM

signaling lymphocyte activation molecule

SLE

systemic lupus erythematosus

Supporting information

Supporting information