Arid5a Mediates an IL-17–Dependent Pathway That Drives Autoimmunity but Not Antifungal Host Defense Free

This article is featured in Top Reads, p.

Interleukin 17 (IL-17A) is a member of a subclass of cytokines that are structurally and functionally distinct from other inflammatory cytokines. Although produced mainly by T lymphocytes (“type 17” cells, including γδ- and αβ-T cells, ILC3, and NKT cells), IL-17 predominantly activates downstream signals in nonhematopoietic cells to protect against mucosal pathogens, notably the commensal fungus Candida albicans (1, 2). Conversely, IL-17 and type 17 cells have been implicated in the pathogenesis of many autoimmune and inflammatory diseases (3), highlighted in the success of anticytokine drugs targeting IL-17 or the IL-17 receptor. Although unquestionably these drugs have improved patients’ disease activity and quality of life, their use is associated with adverse side effects including worsening bowel inflammation in patients with inflammatory bowel disease and the frequent development of opportunistic fungal infections, especially oral and esophageal candidiasis (4–7).

IL-17 signals through a heterodimeric receptor composed of IL-17RA and IL-17RC, although some alternative configurations of the receptor have recently been described (8–10). Binding of IL-17 to the IL-17R complex promotes recruitment of the adaptor Act1 (11–13). In turn, Act1 binds to multiple TNFR-associated factor (TRAF) factors that promote alternative downstream signaling cascades (14). TRAF6 activates several transcription factors, including NF-κB, IκBξ, and C/EBP family members. Additionally, Act1 initiates recruitment of TRAF2 and TRAF5, which initiates a signaling cascade resulting in a prolonged half-life of many IL-17–dependent mRNA transcripts. The effects of IL-17 signaling on mRNA fate is orchestrated by a complex array of RNA-binding proteins (RBPs) (9). RBPs shape the output of inflammatory effector genes by stabilizing or degrading transcripts that encode inflammatory mediators. Some RBPs, including HuR (Elavl1) and IMP2 (Igf2bp2), act as positive regulators of mRNA expression, resulting in target mRNA stabilization or enhanced translation (15–17). Alternatively, RBPs such as Regnase-1, splicing factor 2 (SF2), and Roquins negatively regulate mRNA expression, facilitating degradation of unstable transcripts (18–20). These pathways are interlinked because many transcription factors operative in the IL-17 pathway are encoded by RNAs that are subject to posttranscriptional control, for example, IκBξ (Nfkbiz) and C/EBPs (Cebpb, Cebpd).

AT-rich interaction domain 5A (Arid5a) is another RBP implicated in IL-17 signaling and Th17-driven diseases. Arid5a enhances target mRNA transcript stability and translation, in part by offsetting the destabilizing activity of the endoribonuclease Regnase-1 (15, 21–23). Arid5a controls expression of genes downstream of the Th17/IL-17R axis through control of Il6 and Nfkbiz, as well as Th17 cell responses via Il6, STAT3, TBX21 (Tbet), and OX40 transcripts (21, 23, 24). To date, the precise mechanisms by which IL-17 regulates Arid5a function in autoimmunity are not fully understood. Even less is known about Arid5a function in the context of host defense, although Arid5a mRNA is upregulated in an IL-17–dependent manner in oropharyngeal candidiasis (OPC) in the murine oral mucosa (15, 25). In this study, we created a line of Arid5a^−/− mice to assess the role of Arid5a in autoimmunity versus fungal infections. These data reveal a surprising distinction used between signaling pathways critical in host defense compared with those required in autoimmunity.

The Arid5a knockout allele was fortuitously generated during an attempt to create a conditional deletion with CRISPR-Cas9 (26, 27). Briefly, the targeting strategy was to introduce a LoxP site in intron 2 and a Myc-tag at the C-terminal with another LoxP site just after the stop codon. In the process of genotyping potential founders, we identified a mouse carrying a deletion between the two SpyCas9 target sites, with a deletion of 3359 bp on chromosome 1 between position 36,316,840 and 36,320,199. Fertilized embryos (C57BL/6J, The Jackson Laboratory) produced by natural mating were microinjected in the cytoplasm with a mixture of 0.33 µM EnGen Cas9 protein (New England Biolabs, M0646T), Arid5a-guide2 and Arid5a-guide5 (21.23 ng/µl [∼0.66 µM]), and two single stranded oligonucleotides, that is, Arid5a-Myc-HDR and Arid5a-guide5-HDR (0.5 µM, “Ultramers” from Integrated DNA Technologies). Injected zygotes were cultured overnight, and two-cell embryos were transferred to pseudopregnant CD1 mice. The Single guide RNA (sgRNA) templates were generated by PCR (26) and used for synthesis of sgRNAs using an NEB HiScribe T7 quick high yield RNA synthesis kit (New England Biolabs, E2050S). The sgRNAs were purified using a MEGAclear kit (Thermo Fisher Scientific). The target sequences of the sgRNAs are 5′-ATAGGCTCTGGCCTACAGTTTGG-3′ for Arid5a-guide2 at chr1:36377037-36377060 and 5′-GTAAAAGCCAAATGCGCCCCAGG-3′ for Arid5a-guide5 at chr1:36373679-36373702 (coordinates from GRCm38/mm10 assembly).

The sequences of the single-stranded oligonucleotides are 5′-tcgaactcagagagttctgcctgcttctcctgagtgctaggattaaaggtgtgtgccaccactgcctgggAATTCATAACTTCGTATAATGTATGCTATACGAAGTTATgcgcatttggcttttaccatacggttgagggactctgaccctgccttccaggaactgagttggataa-3′ for Arid5a-guide5-HDR and 5′-TGGCACATGCCACCCGTCACAACCTATGCGGCACCTCACTTCTTCCACCTCAACACCAAACTGGAGCAGAAACTCATCTCAGAAGAGGATCTGTAAGAATTCATAACTTCGTATAATGTATGCTATACGAAGTTATAGAGCCTATCCTGCTATGCTGTGGAGGATTTGATGGGCAGCTGCCGCCATTATCTCAGGCC-3′ for Arid5a-Myc-HDR.

The potential founders were identified by PCR, and the sequence of the knockout allele was determined by sequencing. Arid5a^−/− mice or littermates were genotyped with the following primers: F31, 5′-ACCTTCTGGCTACAACAGGC-3′; R31, 5′-ACCTACACACTTGCTCCTGC-3′, and F51, 5′-ATATGGGTGCTAGGCAAGGC-3′.

The WT allele is detected with primer F31 and R31 and produces a product of 564 bp. The knockout allele is detected with primer F51 and R31 and produces a product of 595 bp.

OPC was induced in 7- to 9-wk-old mice of both sexes by sublingual inoculation with C. albicans strain CAF2-1–saturated cotton balls for 75 min under anesthesia, as described previously (28, 29). Tongue homogenates were prepared on gentleMACS homogenizers (Miltenyi Biotec) with C-tubes, and CFU were determined by serial dilution plating on yeast extract–peptone–dextrose (YDP) with ampicillin. Limit of detection is ∼30 CFU/g. For systemic candidiasis, mice were injected i.v. with 2–7 × 10⁵ C. albicans strain SC5412 in PBS. At 3 d postinfection, kidney tissues were homogenized in C-tubes, and CFU were determined as above.

Mice (males 6–18 wk of age) were immunized s.c. in four sites on the back with 100 μg of myelin oligodendrocyte glycoprotein (MOG) peptide 35–55 (Bio-Synthesis) emulsified with CFA with Mycobacterium tuberculosis strain H37Ra (DIFCO laboratories) and administered 100 ng of pertussis toxin (List Biological Laboratories) i.p. on day 0 and day 2, as described (30). Mice were scored daily as follows: 1) flaccid tail, 2) impaired righting reflex and hindlimb weakness, 3) partial hindlimb paralysis, 4) complete hindlimb paralysis, 5) hindlimb paralysis with partial forelimb paralysis, 6) moribund.

Tongues were harvested and digested with collagenase IV (0.7 mg/ml) in HBSS for 30–45 min at 37°C. Cells were separated by Percoll gradient centrifugation. Abs were from the following sources: anti-CD45 and anti-CD11b (BioLegend), anti-CD4 and anti-F4/80 (Invitrogen), anti-TCRβ, anti-CD8, and anti-CD19 (eBioscience), and anti-Ly6G and anti-Ly6C (BD Biosciences). Dead cells were excluded using Ghost Dye (eBioscience). Data were acquired with an LSRFortessa and analyzed using FlowJo software (Tree Star).

Collected single-cell suspensions from lymph nodes (LNs) were filtered and dead cells were excluded using Ghost Dye (eBioscience). LNs were cultured in complete medium (RPMI 1640 medium containing 10% FCS, supplemented with l-glutamine and antibiotics) with 50 ng/ml PMA and 500 ng/ml ionomycin (Sigma-Aldrich) in the presence of GolgiPlug (BD Biosciences) for 4 h, followed by staining with Ghost Dye, CD4, and IL-17 (BioLegend).

Naive splenocytes and thymocytes were isolated from 6-wk-old Arid5a^−/− and Arid5a^+/+ male mice. Flow cytometry was performed on single-cell suspensions.

Western blotting was performed as described (15). Abs against Arid5a were from Abcam (ab81149). Blots were imaged with a FluorChem E imager (ProteinSimple, Santa Clara CA).

Data were analyzed on GraphPad Prism. Each symbol represents one mouse unless indicated. Significance was defined as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

All of the experiments were conducted following National Institutes of Health guidelines under protocols approved by the University of Pittsburgh Institutional Animal Care and Use Committee.

To elucidate the role of Arid5a in vivo, CRISPR-Cas9 was used to create Arid5a-deficient mice with a deletion in exons 3–7 (Fig. 1A; see Materials and Methods for details). Founders were validated by PCR (Fig. 1B) and bred to homozygosity. Mice were born at expected Mendelian ratios and exhibited normal fertility and baseline body weights (data not shown). Arid5a is expressed at variable levels across tissues, with high expression in thymus, medium levels in spleen, and very low levels in kidney (Fig. 1C). Confirming that mice lacked Arid5a protein expression, tissues from Arid5a^−/− mice displayed undetectable levels of Arid5a compared with Arid5a^+/+ mice.

Previous studies using an independently generated Arid5a knockout mouse demonstrated a role for Arid5a in regulating several immune processes (21–24, 31–33). Therefore, we examined the baseline status of immune compartments in Arid5a^+/+ and Arid5a^−/− littermates in spleen (Fig. 2A) and thymus (Fig. 2B). As shown, no observable differences between Arid5a^+/+ and Arid5a^−/− mice were detected by flow cytometry in thymic or splenic hematopoietic cells, including TCRβ⁺ cells, B cells, neutrophils, monocytes, or macrophages. Therefore, the absence of Arid5a does not appear to impact normal development of immune cells.

Several reports have described a role for Arid5a in driving pathogenesis in mouse models of autoimmunity, with emphasis on control of Th17 cell differentiation (21, 23). Accordingly, we subjected Arid5a^−/− mice and Arid5a^+/+ littermate controls to a standard model of experimental autoimmune encephalomyelitis (EAE), a strongly IL-17–dependent model of multiple sclerosis (17, 34–37). Mice were injected with a MOG peptide (MOG_35–55) emulsified with CFA. Mice were assessed daily for signs of ascending paralysis on a standard scoring system. Arid5a^+/+ control mice developed a typical onset and clinical course of EAE, peaking at 14–16 d postimmunization. As expected, mice lacking the IL-17 receptor adaptor, Act1, and Arid5a^−/− were resistant to disease (Fig. 3A). Concomitant with reduced disease scores, there was marked reduction in EAE incidence in Arid5a^−/− mice (Fig. 3B). Th17 cells are pathogenic in EAE (3, 30, 38, 39). To determine whether Arid5a affects Th17 cells during EAE, mice were immunized with MOG, and LNs and spinal cords were harvested on day 12. Cells isolated from LNs were stimulated with PMA and ionomycin and stained for CD4 and intracellular IL-17A. There were reduced numbers of CD4⁺IL-17⁺ cells in Arid5a^−/− compared with Arid5a^+/+ mice (Fig. 3C). Arid5a is known to regulate Il6 in many cell types (15, 21), and indeed there was a trend to decreased expression in the CNS (Fig. 3D). However, surprisingly, Stat3 mRNA expression was unchanged (Fig. 3D) (22). These findings confirm prior work in an independent system that Arid5a is required for development of EAE (21), and they also functionally verify that this line of Arid5a-knockout mice has the expected phenotype with respect to IL-17–driven autoimmunity.

IL-17R signaling is established to drive oral immunity to the commensal fungus C. albicans (2, 28, 40). Moreover, we previously observed that Arid5a mRNA is upregulated in an IL-17–dependent manner in the oral mucosa of mice during OPC (15). Because Arid5a drives cellular responses to IL-17 through posttranscriptional control of mRNA in vitro (15, 41), we predicted that Arid5a^−/− mice would likely be more susceptible to OPC than WT mice, which clear infection rapidly and do not exhibit overt signs of disease. To test this hypothesis, Arid5a^−/− mice were subjected to OPC by sublingual exposure with a cotton ball soaked with 2 × 10⁷ cells/ml C. albicans strain CAF2-1 (29). As a control, Act1^−/− mice (which lack the ability to respond to IL-17 signaling) (42, 43) showed high fungal loads at 5 d postinfection (Fig. 4A) and progressively lost weight throughout the time course of infection (Fig. 4B). In contrast, Arid5a^+/+ and Arid5a^−/− mice cleared the infection by day 5 and returned to normal body weights (Fig. 4A, 4B).

Disseminated candidiasis is a serious nosocomial infection associated with high rates of morbidity and mortality in humans (2). In mouse models, IL-17R signaling mediates immunity to candidiasis (44, 45). Because Arid5a controls genes that drive Th17 differentiation as well as genes downstream of the IL-17 receptor (14, 41), we postulated that Arid5a deficiency would render mice susceptible to disseminated candidiasis. Therefore, we subjected Arid5a^−/− mice to disseminated candidiasis by i.v. inoculation of 2 × 10⁵ C. albicans yeast cells (strain SC5314). Arid5a^−/− mice had similar fungal loads as Arid5a^+/+ mice (Fig. 4C). Moreover, whereas Act1^−/− mice succumbed to disease by day 7, most Arid5a^+/+ and Arid5a^−/− mice survived beyond day 10 and did not show statistically different survival patterns, indicating that Arid5a is not required for immunity to systemic C. albicans infections (Fig. 4D). Collectively, these data indicate that Arid5a is important for autoimmunity but does not play an important role in the response to C. albicans.

Arid5a was originally described for its ability to interact with AT-rich DNA elements in DNA, modulating cellular proliferation, differentiation, and development (46, 47). Arid5a belongs to the ARID family that consists of 15 proteins, which can be divided into seven subgroups based on the degree of similarity within the ARID domain. The ARID domain is a helix-turn-helix motif essential for binding to DNA elements (46, 47) and can also interact with cis-acting elements located in the 3′ untranslated region of certain inflammatory RNAs, such as Il6, Tbx21, Stat3, and Ox40 (15, 21–24, 48). Binding of Arid5a to these transcripts promotes their longevity and/or facilitates their translation, although the underlying mechanisms are not well understood. Additionally, Arid5a offsets the action of a potent endoribonuclease, Regnase-1, which degrades target transcripts and thereby serves as a negative regulator of inflammation (21). Arid5a acts in hematopoietic cells (macrophages, T cells) and nonhematopoietic cells (fibroblasts, epithelial cells) (14, 41). Thus, Arid5a is a newly recognized RBP that is an important posttranscriptional modulator in multiple immune pathways.

The Arid5a^−/− mice described in the present study are not identical to the original published Arid5a^−/− mouse strain (21). The latter lacks exons 1–3, and a major part of the DNA binding domain was replaced by a floxed Neo-cassette. In the Arid5a^−/− mice described in the present study, LoxP sites were introduced into intron 2 and immediately following the stop codon, creating a 3359-bp deletion. Despite these differences, both Arid5a^−/− lines are resistant to EAE (21), lending confidence that Arid5a plays an important role in driving Th17/IL-17–driven autoimmune inflammation.

IL-17 mediates effects through upregulation of target RNA accumulation, which can occur transcriptionally but also posttranscriptionally (9). Even so, the data presented in the present study indicate that Arid5a^−/− mice are, unexpectedly, resistant to oral and systemic candidiasis even though IL-17R signals are essential for immunity to this condition (44, 49, 50) and Arid5a functions downstream of IL-17 to upregulate many IL-17 target genes (15, 25). Arid5a has been shown to act in T cells to stabilize genes such as Il6, Stat3, Nfkbiz, Ox40, and Tbx21 (encoding Tbet) that drive Th17 differentiation and IL-17 production (22, 41). One explanation to reconcile these observations may be that the early responses to C. albicans in mice are innate in nature, as this fungus is not a commensal microbe in mice (51). Rather than conventional Ag-specific Th17 cells, IL-17 in these naive settings comes primarily from a combination of γδ T cells and innate-acting CD4⁺ αβ T cells; additionally, adaptive T cell memory responses can be established in mice that appear to reflect human T cells responses (52–58). In line with this, Il6^−/− and Ccr6^−/− mice are resistant to OPC, and Th1 cells are dispensable for protection against oral C. albicans (28, 52, 59). Thus, the genes regulated by Arid5a might not be important for protection against C. albicans oral infections.

Anti–IL-17 therapy is not associated with risks for systemic candidiasis, likely because the major mechanisms of antifungal control at cutaneous/mucosal surfaces is through neutrophil recruitment and β-defensin production whereas regulation of circulating neutrophils are less IL-17–dependent (4, 50, 60, 61). In contrast, during systemic candidiasis IFN-γ, IL-12, as well as IL-17 signaling contribute to effective immunity against C. albicans (62–64). Previous studies showed that Arid5a regulates expression of IFN-γ in Th1 cells via control of the master Th1 transcription factor Tbet (24). Given this, we were surprised to find that Arid5a^−/− mice showed the same susceptibility to systemic candidiasis as their wild-type littermate counterparts. It is possible that other mRNA stabilizing factors either in T cells or in IL-17–responsive cells (likely renal tubular epithelial cells [45], although NK cells have been reported [65]) can compensate for the loss of Arid5a during C. albicans infection, but more work is needed to dissect this phenomenon.

The use of biologics targeting IL-17 or the IL-17 receptor was a major advance in the treatment of psoriasis, psoriatic arthritis, and ankylosing spondylitis (10, 66, 67). Unexpectedly, IL-17 blockade failed in Crohn’s disease, which has been attributed to a vital tissue-protective role of IL-17 in the intestine (5, 68–70). Intriguingly, IL-17 has been linked to controlling social behavior in mice (71), and blockade of IL-17RA was linked to a small risk of suicide in humans (72), although this is controversial (73). Thus, an ideal therapeutic approach to blocking IL-17 would inhibit its pathogenic effects while sparing host-defensive pathways. Based on these and other data, Arid5a may be an attractive candidate for selective blockade in IL-17–driven diseases, although achieving this is not trivial. Interestingly, recent advances in approaches to understand and target RBPs suggests potential ways to disrupt Arid5a (or other RBPs) interactions with specific target transcripts (74). For example, an elegant study in the IL-17 system showed that aptamers that recognize the binding element of Act1 can interfere with its mRNA stabilizing capacity, and are effective at reducing inflammation in vivo (75). Indeed, the development and potential uses of RNA therapeutics is becoming increasingly appreciated (76, 77). Arid5a was reported to be a target of the antipsychotic drug chlorpromazine (22), although this remains to be confirmed. Targeting individual pathway components such as Arid5a potentially allows for a more targeted approach to treat IL-17–mediated inflammatory/autoimmune diseases while preserving the protective role of IL-17 in host defense.

We thank B.M. Coleman, N. Amatya, and R. Bechara for valuable input. The graphical abstract was created with BioRender.com.

S.L.G. has consulted for Aclaris Therapeutics and Eli Lilly on aspects of IL-17 signaling. The other authors have no financial conflicts of interest.