Differential Effect of JAK1/2 and TYK2 Inhibitors on Type III Interferon Induced Antiviral Defence in Human Intestinal Epithelial Cells

Immune-mediated inflammatory diseases (IMIDs) include autoimmune diseases and other conditions characterized by dysregulated immune responses. IMIDs are common and increasing in incidence worldwide. Cytokines are important mediators of pathology in these diseases. For example, cytokines including interleukin (IL)-6, IL-17 and tumour necrosis factor alpha (TNFα) contribute to synovial inflammation and joint destruction in rheumatoid arthritis. Furthermore, IL-17, IL-23, and TNFα are involved in causing skin inflammation in plaque psoriasis [1]. In type 1 diabetes, production of type I interferon (IFN) in the pancreatic islets during the prediabetic period has been proposed to cause excessive expression of MHC class I which thereby facilitates CD8⁺ T-cell-mediated killing of pancreatic beta cells [2, 3].

Many of the cytokines implicated in the pathogenesis of IMIDs act via receptor-mediated activation of Janus kinase (JAK) and signal transducer and activator of transcription (STAT) signalling pathways (recently reviewed in [4]). There are four known JAK isoforms in humans (JAK1-3 and tyrosine kinase 2 [TYK2]). Cytokines and growth factors bind to membrane‐bound cellular receptors, resulting in the activation of associated JAKs (via phosphorylation). Activated JAKs then phosphorylate the intracellular part of the receptor to allow docking and phosphorylation and thereby activation of STATs, which translocate to the nucleus to initiate gene transcription. Type I and II cytokines, including IFNs, IL-2, IL-4, IL-5, IL-10, IL-13, IL-12, IL-21, IL-22, and IL-23, as well as endocrine- and colony-stimulating factors signal via different combinations of JAK-STATs. They have critical functions in regulating antiviral and antitumoral responses, and various systemic functions, including haematopoiesis, lipid metabolism, and granulopoiesis [4].

Much work has been devoted to developing selective inhibitors of the JAK-STAT signalling pathway as therapeutic options for IMIDs, as well as other types of disease including type I interferonopathies (a subset of autoinflammatory disorders characterized by excess IFNα/β signalling) [5, 6]. JAK inhibitors (JAKis) have demonstrated striking efficacy in the treatment of several diseases (e.g., rheumatoid arthritis, atopic dermatitis, and alopecia areata) and are currently being tested for others (e.g., psoriasis, Crohn’s disease and type 1 diabetes) [1, 7, 8]. The first JAKis to be developed target two or three different JAKs and are thus broad acting. Newer generations of JAKis have shown greater specificity for selected JAKs, although partial selectivity for other JAKs has often been observed [5]. Several adverse events (AEs) were reported in trials testing the early JAKis and safety concerns have also remained when testing more recently developed agents with higher specificity [1, 5, 7, 9‒11]. Commonly reported AEs include respiratory tract infections, nasopharyngitis, diarrhoea and headache. More severe AEs such as dyslipidaemia, certain viral and bacterial infections, anaemia, and malignancies have also been documented, occasionally resulting in failure to gain regulatory approval. The side effects may stem from a lack of selectivity, with higher drug doses associated with worse AEs, but the mechanism of action may also be implicated [5, 10]. Due to this, in part, JAKis selectively targeting TYK2 have been developed and tested in clinical trials for diseases such as plaque psoriasis and psoriatic arthritis [1, 12]. Reports from trials testing one of these TYK2 inhibitors, deucravacitinib (SOTYKTU^™), in moderate-to-severe psoriasis showed that the drug was well tolerated and induced a mid-high rate clinical response without causing the more severe toxicities commonly reported for pan-JAKis (e.g., dyslipidaemia) [13].

Increased risk of infections (including opportunistic infections) such as nasopharyngitis, upper respiratory tract infection, tuberculosis, gastritis, urinary tract infections and herpes zoster (shingles) has been associated with the use of JAKis [10]. This could result from the inhibitors blocking antiviral immunity [14] and leads to the hypothesis that more specific JAKis, which preserve the host antiviral defence, could lower the risk of infections. A recent meta-analysis suggested that the infection risk is similar among approved JAKi drugs [9], but how these compare to more recently developed inhibitors of TYK2 (e.g., deucravacitinib, approved September 2022 [12]) remains to be fully investigated.

IFNs play a central role in antiviral immune defence by inducing the expression of IFN-induced genes (ISGs) encoding proteins with antiviral activity (e.g., ISG15, Mx1, OAS1, and PKR) and by modulating the antiviral immune response [15, 16]. Three groups of IFNs have been described: type I (IFNα, β and related molecules), II (IFNγ), and III (IFNλ1-4, also denoted IL28A, IL28B, IL29, and IFNλ4, respectively). Type I and III IFNs play important roles in the early immune response to viral pathogens (reviewed in [16, 17]). Type II IFN (IFNγ) is instrumental in activating both innate and adaptive cell-mediated immunity. Each IFN family signals via a unique receptor, which is followed by the activation of JAK/STATs. Type I (IFNαβ) and III (IFNλ) IFNs signal via their own unique receptors and activation of JAK1/TYK2, while type II (IFNγ) receptor signalling is mediated via JAK1/JAK2 [16]. Due to these similarities and differences, JAK1 and JAK1/2 inhibitors block the effect of all three IFN families, while TYK2 specific inhibitors are expected to suppress type I and III but not type II IFN signalling. However, recent studies have questioned whether TYK2 is essential for type III IFN (IFNλ) signalling [16, 18, 19].

Many viruses that are common in humans infect via mucosal sites including the intestinal and respiratory tracts. These include, for example, entero-, noro-, rota-, and influenza viruses. Several studies have documented the paramount role of type III IFNs in protecting the intestinal barrier, while type I IFNs are important for preventing systemic viral spread [16, 17, 20‒23]. However, despite mounting evidence for the important role of type III IFNs in protecting the gut from viral infection, virtually nothing is known about whether treatment with JAKis interferes with the antiviral defence mechanisms of the human intestinal epithelium.

The aim of the present study was to investigate whether clinically approved JAKis affect type III IFN-induced antiviral responses in epithelial cells lining the gastrointestinal tract. As type III IFNs (and type I IFNs) signal via JAK1/TYK2, we focused on JAKis that inhibit these signalling molecules and not JAK3. Furthermore, we examined whether there was any difference in the effect on the antiviral response if JAK1/2 or TYK2 was inhibited. To this end, we studied the effect of the JAK1/2 inhibitor baricitinib and the TYK2 selective inhibitor deucravacitinib on human intestinal epithelial cells (IECs; Caco-2 and HT-29). MxA, a key mediator of cellular antiviral responses, served as an indicator for the IFN-induced cellular antiviral response. Drug concentrations commonly used for in vitro studies as well as those that are clinically relevant were employed.

Caco-2 cells originally obtained from Dr. Aspenström Fagerlund, the Swedish National Food Agency, Stockholm, Sweden, were cultured in DMEM (Gibco, Life Technologies Fisher Scientific, Göteborg, Sweden) supplemented with 20% heat-inactivated fetal bovine serum (FBS, Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) and 2 mml-glutamine (Cytiva Europe GmbH, Uppsala, Sweden). HT-29 cells (HTB-38; American Type Culture Collection, ATCC, Manassas, VA, USA) were cultured in McCoys 5 A medium (Merck KGaA, Darmstadt, Germany) supplemented with 2 mml-Glutamine and 10% heat-inactivated FBS. All cell culture media was free of antibiotics. Cells were cultured at 37°C, 5% CO₂ and passaged when they reached confluency. Mycoplasma testing was performed, and cells were confirmed mycoplasma negative.

Cells were plated to reach a confluence level of 50–70% on day 2 post seeding. Cells were then exposed to either baricitinib (Aviva Systems Biology, CA, USA) or deucravacitinib (BMS-986165; MedChemExpress, NJ, USA), both dissolved in DMSO, at the concentrations indicated. Control cells were exposed to the solvent DMSO at corresponding concentrations. In some experiments, cells were later exposed to IFNs or buffer alone (see below) in the continued presence of JAK1/JAK2 or TYK2 inhibitor. Cells were then collected for mRNA extraction, FACS staining, or infected by Coxsackie B virus (CVB) virus (see below).

Cells were exposed to IFNα (IntronA, Paranova Läkemedel AB, Solna, Sweden), IFNλ1 or IFNλ2 (PeproTech, NJ, USA) or buffer alone (0.1% BSA, Sigma-Aldrich, Merck KGaA, Darmstadt, Germany). Cytokines were used at the concentrations indicated in each results section (IFNα 0–3,000 U/mL, IFNλ1 and IFNλ2 0–300 ng/mL, 0.1% BSA ≤1:500) and stimulated for the indicated length of time (6–72 h).

Upon termination of the experiment, cells were collected by trypsinization, then unspecific binding was blocked with 10% FCS. Cells were stained for viability assessment with Live/Dead Fixable Far-Red stain (ThermoFisher Scientific, MA, USA) and HLA-ABC expression (Antibody clone W6/33, BioLegend, CA, USA). Cells were acquired by the BD Accuri C6 (BD Biosciences, Stockholm, Sweden) and analysed using FlowJo v10 software (BD Biosciences, OR, USA).

CVB3-V13 was originally obtained as previously described (Stone, 2021), propagated and titered in RD cells. Pilot experiments were carried out to optimize infection conditions (data not shown). Cells were plated in 96-well plates at a concentration of 5 × 10³ cells/well (Caco-2) or 1 × 10⁴ cells/well (HT-29). Infections were performed using the protocol described in [21]. In brief, at the day of infection, cells in at least two individual wells were counted to calculate the average number of cells/well. Cells were then washed in PBS followed by infection by CVB3-V13 diluted to the indicated multiplicity of infection (MOI) using serum free medium. Control cells were mock infected (using serum free medium). One hour later, the cells were washed 3 times in PBS and incubated for the indicated time in complete media. If present prior to infection, JAK/STAT and TYK2 inhibitors and/or IFNs were not included during the 1 h infection period, but after this, they were re-added to the culture medium at similar concentrations as before infection. After the indicated time periods, the media was removed, and the cells were fixed using modified Carnoy’s reagent (25% Glacial Acetic Acid in Methanol) followed by staining with 0.2% crystal violet. Optical density was measured at 595 nm using a Bio-Rad xMark^™ Microplate Spectrophotometer (Bio-Rad Laboratories AB, Solna, Sweden). In every experiment (biological replicate), each condition was performed in at least triplicate (technical replicates) and a mean of the technical replicates was calculated.

Total RNA was extracted using the Qiagen RNeasy mini kit (Qiagen, Sollentuna, Sweden). Briefly, cells were washed in ice-cold PBS followed by the addition of Buffer RLT. Retrieved cell suspensions were homogenized using QIAshredder columns, loaded onto RNeasy spin column and on-column DNase treatment was performed. RNA was eluted in RNase free water. RNA concentration and purity were measured using a NanoDrop ND-1000 spectrophotometer (Saveen Werner AB, Limhamn, Sweden).

Total RNA (2 μg per sample) was reverse transcribed to cDNA using the High Capacity cDNA Reverse Transcription Kit from Applied Biosystems (Life Technologies Fisher Scientific, Göteborg, Sweden) according to the manufacturer’s instructions. The mRNA expression of ISG15, MX1 and GAPDH was quantified by real-time PCR using QuantiTect primers (Qiagen) (MX1: QT00090895; ISG15: QT00072814; GAPDH: QT00079247) and RT² SYBR Green qPCR Mastermix (Qiagen). Samples were analysed with the QuantStudio^™ 5 Real-Time PCR System (Applied Biosystems, Life Technologies Fisher Scientific, Göteborg, Sweden). The cycle detection threshold for both MX1 and ISG15 was set to a Ct value of 35, samples below the detection level was considered gene not expressed. Gene expression levels were normalized to the expression of GAPDH. The data are presented as either 2^−ΔCt or ΔΔCt normalized to control sample, as indicated in each figure legend.

Total RNA from 2 to 6 individual experiments per setup was mixed in equal concentration to gain 3 µg total RNA. cDNA synthesis was performed using the RT² first strand kit according to manufacturer’s instructions (Qiagen). RT² profiler PCR arrays for human interferon and receptors (PAHS-064Z-D) (Qiagen) were performed on each cDNA using RT² SYBR Green qPCR Master Mix in the Bio-Rad CFX96 real-time PCR system. The manual Ct threshold was set to 300 across all arrays, quality control on RTC and gDNA Ct-values pass was confirmed. Array relative quantity normalization was performed using the five housekeeping genes (ACTB, B2M, GAPDH, HPRT1, RPLP0) and normalized gene expression value was calculated for all genes compared to the housekeeping genes in each array. Analysis was performed using the online RT² Profiler PCR Data Analysis (https://dataanalysis2.qiagen.com/pcr/analysis). Data are presented as differential gene expression compared to control cells with 2-fold primer efficiency across all genes (2^DGE), or as relative quantity (2^−ΔCt) compared to the geometric mean of the housekeeping genes.

Statistical analysis was performed using the GraphPad Prism 9 software (GraphPad Software, La Jolla, CA, USA). Comparisons between two groups were performed using the Student’s t-test. Comparisons between groups in studies including two independent variables were done using two-way ANOVA with Šídák's multiple comparisons test. A p value <0.05 was considered statistically significant. Data are presented as the mean ± standard deviation (S.D.).

We have previously demonstrated that IECs respond to both type I and III IFNs by expressing genes encoding ISGs (e.g., ISG15, MX1, PKR, and OAS2) [21]. By assessing the induction of ISG15 and MX1 mRNA expression following a 6 h exposure of Caco-2 cells to increasing concentrations of IFNs, we confirmed that a robust induction of MX1 mRNA expression was reached after exposure to 100 ng/mL IFNλ1 or IFNλ2, and 1,000 U/mL IFNα (online suppl. Fig. S1; for all online suppl. material, see https://doi.org/10.1159/000538571). We also noted a clear increase in ISG15 mRNA expression after stimulation with increasing concentrations of all three IFNs (online suppl. Fig. S1). In contrast to MX1, the basal expression level of ISG15 was very low (Ct-values in control cells between 35 and 37 compared to 26 and 28 for MX1) and as the accuracy of the PCR-based quantification is poor at low expression levels, ISG15 expression was not assessed in further studies. We next analysed the expression of a larger set of ISGs and related genes in Caco-2 cells treated with baricitinib, a JAK1/2 inhibitor, or deucravacitinib, a selective TYK2 inhibitor [12], at concentrations commonly used for in vitro studies on IFN signalling [2, 24, 25] followed by exposure to type III IFN (IFNλ1). Consistent with the previous studies [21], the expression of MX1, ISG15 and numerous other ISGs were induced by IFNλ1. However, this expression was reduced or completely prevented by both inhibitors (Fig. 1a; online suppl. Table S1). Furthermore, cell surface expression of MHC class I (HLA-ABC) was upregulated in both Caco-2 cells after 24 h exposure to IFNλ and this was attenuated by both baricitinib and deucravacitinib (Fig. 1b). Similar results were obtained when studying IFNλ-induced MHC class I expression in HT-29 cells (Fig. 1b). These studies confirmed that ISG expression is increased in both Caco-2 and HT-29 cells following IFN stimulation and provide new data, demonstrating that JAK1/2 and TYK2 inhibitors prevent this upregulation.

Treatment with JAK1/2 and TYK2 inhibitors may not have an impact on IFN-induced expression of ISGs alone but may also affect the basal expression of ISGs and other IFN-related genes in IECs. In line with this, the basal expression levels of several IFN-related genes were decreased in Caco-2 cells treated with either baricitinib or deucravacitinib alone (Fig. 2a; online suppl. Table S2). The relative mRNA expression of MX1 (shown in Fig. 2b) was reduced to around 60% compared to that in untreated control cells (Caco-2 + baricitinib: 0.66 ± 0.25; Caco-2 + deucravacitinib: 0.51 ± 0.13; HT-29 + baricitinib: 0.61 ± 0.11; HT-29 + deucravacitinib: 0.63 ± 0.19).

Lowered basal ISG expression caused by JAKi treatment may result in a weakened antiviral defence in the gut and increase permissiveness to viruses infecting via the gastrointestinal tract. CVB belongs to the enterovirus genus and infects via the intestine. Infections cause the common cold and occasionally more severe and life-threatening illnesses (e.g., meningitis, myocarditis, and pleurodynia). Infections with this type of virus have also been associated with the development of type 1 diabetes [26]. To address if exposure to baricitinib or deucravacitinib increases the permissiveness of IECs to CVB infection, IECs pre-treated with inhibitors or mock treated were infected with CVB serotype 3 (CVB3) at increasing MOIs. Two days later, cytopathic effect (CPE) was assessed by measuring the optical density of the cell monolayer. CPE was visible in mock-treated cells following CVB infection at MOIs of 10⁻⁵ and 10⁻⁶ or higher for Caco-2 and HT-29 cells, respectively (online suppl. Fig. 2, and data not shown). The CPE was not different in cells preincubated with baricitinib or deucravacitinib. Taken together, these observations suggested that although basal ISG expression levels were reduced, permissiveness to infection per se was not increased following the blockage of either JAK1/2 or TYK2.

During early viral infection, IFNs released by parenchymal and immune cells have an essential function in acting on uninfected cells, stimulating the upregulation of genes involved in antiviral defence, thereby lowering permissiveness to infection [15, 16]. In a previous study, we demonstrated that type III IFNs protect IECs from CVB3 infection [21]. To examine whether inhibition of JAK1/2 or TYK2 signalling has a negative impact on the ability of IECs to enter an antiviral state following type III IFN stimulation, we pre-treated IECs with inhibitors prior to stimulation with IFNλ1 and then infected the cells with CVB3 at different MOIs. Cells exposed to type I IFN (IFNα) were included for comparison. Pre-treatment with either type III (IFNλ1) or type I (IFNα) IFN provided strong protection from infection (Fig. 3). Indeed, CPE was seen at MOIs of 10⁻⁵ and 10⁻⁶ or higher for untreated Caco-2 and HT-29 cells, respectively, while for the IFN treated cells, the virus dose had to be increased by one to two orders of magnitude (10–100 times higher infectious dose) to cause CPE. The protection afforded by IFN treatment was completely lost when cells were treated with baricitinib or deucravacitinib prior to IFN treatment and subsequent infection (Fig. 3), suggesting that infection risk is increased during treatment with JAK1/2 and TYK2 inhibitors.

Given the key role of type III IFNs in protecting the intestinal tract from infection [16, 17, 20‒23], we hypothesized that a preserved type III IFN response in the intestine may be beneficial for individuals treated with JAKis. Therefore, we next asked whether the JAK1/2 or TYK2 inhibitors have any difference in impact on type III IFN-induced antiviral defence. To this end, Caco-2 and HT-29 cells were treated with increasing concentrations of either baricitinib or deucravacitinib, including those corresponding to plasma concentrations at approved treatment doses [27]. Two hours later, the cells were stimulated with type III (IFNλ1) IFN for 6 h, after which they were collected for gene expression analysis. For comparison, separate wells of cells were stimulated with type I (IFNα). Both baricitinib and deucravacitinib inhibited IFNλ1-and IFNα-induced MX1 expression in a dose-dependent manner (Fig. 4). At all concentrations tested, baricitinib (JAK1/2 inhibition) was equally efficient in blocking IFNλ1-or IFNα-induced MX1 gene expression. In contrast, deucravacitinib (TYK2 inhibition) was much less effective in inhibiting IFNλ1-than IFNα-induced MX1 gene expression at equimolar concentrations below 0.25 μm (Fig. 4), indicating that at clinically relevant concentrations, a TYK2 inhibitor better preserves type III IFN-induced host antiviral defence pathways in IECs than a more broadly acting JAK1/2 inhibitor.

Type III IFNs activate intracellular signalling via JAK/STATs and have an essential role in protecting the intestinal mucosa from infection [16, 17]. In this study, we found that drugs which target and inhibit JAK1/2 or TYK2 signalling have an impact on the autonomous antiviral defence in IECs. At high concentrations, the inhibitors lowered steady-state ISG expression; however, this occurred without increasing permissiveness to a common human enteric virus, CVB3. Importantly, our study revealed a difference in the efficiency of a JAK1/2 and a TYK2 inhibitor in blocking type III IFN-induced ISG expression at clinically relevant drug concentrations. While the JAK1/2 inhibitor blocked both type I and III IFN-induced ISG expression at such concentrations, the more selective TYK2 inhibitor only showed efficient suppression of type I IFN-induced ISG expression. These data suggest that TYK2 inhibition at drug concentrations relevant to clinical practice prevents type I IFN-mediated signalling while preserving the type III IFN response, known to be critical for intestinal antiviral barrier protection.

Type I and III IFNs are produced early following viral infection [15]. They have unique and complementary activities which contribute to controlling and clearing viral infections [15, 16]. While the type I IFN receptor is ubiquitously expressed, the type III IFN receptor is mainly expressed on cells of epithelial origin and a few immune cell types. This in part explains the different actions of the two IFN families; both induce the expression of ISGs (many of which have antiviral activity), but their primary roles are in different body compartments during infection. There is extensive literature demonstrating that type III IFNs are of greater importance for the protection of mucosal sites than type I IFNs, which in turn have a more prominent role in regulating the systemic spread of infection and are also more commonly involved in excessive inflammatory responses. For example, type III IFNs but not type I IFNs are important in controlling rotavirus infection in both human IECs [28] and in the intestine of mice [29, 30]. Human IECs lacking the type III IFN receptor showed increased permissiveness for SARS-CoV-2 infection, whereas there was no difference in permissiveness to infection between control cells and cells depleted of the type I IFN receptor [31]. Furthermore, type III IFNs have a key role in limiting enterovirus replication in the intestinal epithelium [23].

There is a well-documented increased risk of infections associated with the use of JAKis [10]. Nasopharyngitis, upper respiratory tract infections and diarrhoea are commonly reported AEs, and, with several JAK1/2/3 inhibitors, reactivation of latent infections (e.g., varicella zoster and herpes simplex virus) has been described. The risk of infections varies between different JAKis (e.g., [32]) and is also dependent on the dose administered [10]. This suggests a correlation between the magnitude of inhibition and the risk of infection. The aim of the current study was to investigate and compare the effect of JAK1/2 (baricitinib) and TYK2 (deucravacitinib) inhibitors on intestinal antiviral defence mechanisms, with a particular focus on type III IFNs.

Using concentrations of baricitinib and deucravacitinib comparable to those studied previously [2, 24, 25], we observed a clear reduction in baseline ISG expression in human IECs cultured with either baricitinib or deucravacitinib. This suggests that IECs have a basal expression of proteins involved in antiviral defence which is dependent on JAK1/2 and TYK2 signalling and is in line with studies demonstrating homoeostatic IFN-stimulated gene expression in the intestinal epithelium [17, 33]. For example, Van Winkle et al. elegantly showed that bacterial microbiota stimulates homoeostatic ISG expression in mouse intestinal epithelium via immune cell derived IFNλ. Furthermore, when this type III IFN-induced antiviral response is absent, mice show dramatically increased susceptibility to rotavirus infection [33]. Our experimental in vitro model system lacks the influence of microbiota and immune cells and whether the low constitutive expression of ISGs in human IEC can be explained by other cell-intrinsic factors remains to be understood. However, we found that although expression of host antiviral defence in IECs was attenuated, the inhibitors did not increase susceptibility to infection by a common enteric virus per se.

Early during viral infection, cells at the local site of infection respond by producing and secreting type I and III IFNs. Through autocrine and paracrine signalling, IFNs induce an antiviral state that protects uninfected cells from infection and prevents unrestricted virus replication in cells that are already infected [15, 16]. As previously shown [21], treatment with IFNs reduced the permissiveness of IECs to CVB infection. This protection was abolished if the cells were treated with a high concentration of inhibitor, showing that although JAKi treatment does not increase permissiveness to infection per se, it could reduce critical IFN-induced intestinal antiviral defence leading to rampant virus replication.

Up to this point, our studies utilized drug concentrations previously employed to investigate the effects of baricitinib and deucravacitinib on IFN signalling in vitro [2, 24, 25]. However, the concentrations we used exceed both the anticipated plasma levels at approved treatment doses [27] and the IC50 for the drugs. Consequently, we conducted dose-response studies to explore more relevant drug concentrations. These analyses revealed an interesting and significant observation: JAK1/2 and TYK2 inhibitors exhibit distinct effects in blocking type III IFN-mediated antiviral defence in the intestine. At drug concentrations similar to those estimated in patients receiving treatment [27], the TYK2 inhibitor deucravacitinib did not block type III IFN-induced ISG expression as efficiently as type I IFN-induced ISG expression. This observation was made in both IEC cell lines and has several implications. First, the results suggest that in IECs, type III IFN is less dependent on TYK2 for signalling than type I IFN. To the best of our knowledge, this is the first time that such a difference has been uncovered in human IECs. These findings are in line with recent studies highlighting a low or negligible role for TYK2 in type III IFN signalling [16, 18, 34, 35] and add human IECs to the list of cell types where this has been observed. Second, our studies imply that while a JAK1/2 inhibitor affects both type I and III IFN signalling, a TYK2 inhibitor is more effective in blocking type I IFN than type III IFN signalling at clinically relevant concentrations. Given the critical importance of type III IFNs in the local antiviral defence of the gut [16, 17], these data may be of relevance for the treatment of certain IMIDs. Indeed, when JAKis are used in the treatment of diseases associated with cytokines signalling via JAK1/2- and TYK2-coupled receptors (e.g., type I IFNs, IL-6, IL-12, IL-22, and IL-23), a TYK2 inhibitor may be preferable to a JAK1/2 inhibitor, reducing risk of excessive infections by viruses infecting via the gut while still strongly blocking the activity of cytokines with proinflammatory and tissue-damaging effects. Indirect support for such a scenario comes from studies demonstrating that TYK2 variants with reduced function are linked to decreased or loss of function of type I IFN, IL-12, and IL-23 signalling and protection from autoimmune disease without, however, resulting in general immunodeficiency (e.g., [36‒39]).

It has been suggested that a few IMIDs have a viral trigger. For example, Epstein Barr virus infection has been implicated in multiple sclerosis [40]. Epstein Barr virus infects epithelial cells of the oropharynx and parotid glands and causes latent infection. Furthermore, enterovirus infections have been linked to the appearance of islet autoantibodies and subsequent development of type 1 diabetes [41, 42]. Most enteroviruses infect via the intestine and are rapidly cleared; however, continuous faecal shedding of enteroviruses was recently associated with a risk of developing islet autoimmunity [43]. Considering the heightened risk of viral infections and reactivation of latent virus infections associated with the use of JAKis [9‒11], the preservation of antiviral defence mechanisms in mucous membranes may be of importance when using JAKis to treat diseases with a suspected viral aetiology.

Both type I and II (IFNγ) IFNs are proposed to play important roles in the pathogenesis of type 1 diabetes. Type I IFN may be of importance both in the early, prediabetic stages of the disease and at clinical disease onset, while IFNγ seem to mainly contribute to pancreatic beta cell destruction mainly in the late stage (e.g., [44, 45]). Preclinical studies using JAKis for the prevention of autoimmune diabetes have demonstrated positive outcomes [46, 47]. Due to this, initiatives have been taken to investigate the efficacy of baricitinib in reducing beta cell loss after clinical disease presentation, stage 3 [8]. A new notion that IFNγ dampens the proliferation of autoreactive T cells has however called into question the appropriateness of blocking this cytokine [45]. Given that studies have shown an important role for type III IFNs in attenuating enterovirus infection of IECs (present study and [20‒23]) and in clearing chronic viral infections of the gut [48, 49], the recent approval of deucravacitinib (SOTYKTU^™) in the USA and Japan [12] now opens up the possibility of using a more targeted inhibition of type I IFN signalling (but not IFNγ) that may simultaneously preserve type III IFN-mediated intestinal defence against enteroviruses.

In summary, our study shows that antiviral defence mechanisms of the intestinal epithelium are affected by JAKis. We also show that at clinically relevant drug concentrations, a TYK2 inhibitor effectively block the activity of proinflammatory type I IFN but has less negative impact on critical type III IFN-induced antiviral defence mechanisms in the intestine. The information gained from these studies may help inform strategies to treat various diseases of inflammatory and autoimmune origin with greater precision and fewer side effects.

The authors thank Mrs. Selina Parvin for excellent technical support.

The cell lines used in this study were obtained from Dr. Aspenström Fagerlund, the Swedish National Food Agency, Stockholm, Sweden (Caco-2), and American Type Culture Collection, ATCC, Manassas, VA, USA (HT-29). Ethical approval for the use of these is not required in accordance with local/national guidelines.

The authors have no conflicts of interest to declare.

This work was supported by Grants from Karolinska Institutet, Sweden, including the Strategic Research Programme in Diabetes, the Swedish Child Diabetes Foundation, the Swedish Diabetes Foundation, the Swedish Heart and Lung foundation, and the Swedish Research Council. Martha A. Castro was supported by a Fulbright Fellowship. The funders had no role in the design of the study; the collection, analyses, or interpretation of data; writing the manuscript; or the decision to publish the results.

M.A.C.: experimental work, data interpretation, data curation, formal analysis, critical revision of the article, and final approval. E.E.R.: conceptualization and supervision, experimental work, data interpretation, data curation, formal analysis, critical revision of the article, and final approval. V.M.S.: supervision, experimental work, critical revision of the article, and final approval. A.P.: experimental work, final approval of article. M.F.T.: conceptualization and supervision, funding acquisition, data interpretation, drafting of the article, and final approval.

Raw data and analyses of the RT² profiler PCR arrays are available as online supplementary Tables 1 and 22. Further enquiries can be directed to the corresponding author.