Upadacitinib re-establishes the balance of peripheral blood monocyte subsets in RA

To assess whether JAK inhibitors (JAKi) had an effect on myeloid cells in RA patients, we initially determine the relative levels of monocyte subsets in peripheral blood before and after a 3-month period of Upadacitinib (Upa) treatment. Analysis of 3 independent RA patients showed that Upa treatment results in significantly diminished levels of the classical CD14 + + CD16- monocyte subset and an augmented proportion of non-classical CD14 + CD16 + monocytes (Fig. 1A–B and Supplementary Fig. 1). Moreover, after three months of treatment, monocyte subsets levels of RA patients were similar to those found in normal donors (Fig. 1A). Since reduction of the non-classical CD14 + CD16 + monocyte subset characterizes peripheral blood from active RA patients [33], these results indicate that Upa restores the balance of monocyte subsets found in healthy individuals, a finding that agrees with the known therapeutic action of JAKi (Fig. 1A–B). Transcriptomic analysis of CD14 + monocytes isolated from RA patients before (Pre-Upa) and 3 months along Upa treatment (3mo-Upa) further confirmed these phenotypic changes. GSEA revealed that the transcriptome of 3mo-Upa has a very significant over-representation of the genes that define non-classical (CD16 +) monocytes as well as a significantly reduced expression of the genes that define classical (CD14 +) monocytes, using the genesets previously defined GSE25913 [34], GSE94497 [35], and GSE16836 [36] (Fig. 1C). In fact, leading edge analysis evidenced that, besides CD16 (FCGR3A), Upadacitinib upregulates numerous other genes that characterize CD16 + monocytes, including RHOC and ADA. Conversely, Upadacitinib downregulates not only paradigmatic CD14 + monocyte marker genes (CD14, CCR2) but also genes whose expression characterizes CD14 + classical monocytes (VNN2, S100A12, SERPINB2) (Fig. 1C). Specifically, comparison of the transcriptome of Pre-Upa and 3mo-Upa indicated the existence of 149 genes whose expression is significantly (p < 0.05) diminished by Upa treatment, including STAT-dependent genes like PIM1, IFIT1 and RSAD2 [35, 37, 38] (Fig. 1D, Supplementary Table S1). On the other hand, the 129 genes whose expression is augmented upon Upa treatment included several genes associated to anti-inflammatory functions like CD28 and CD127 [39, 40] (Fig. 1D, Supplementary Table S1). Altogether, these findings indicate that JAKi influence monocyte differentiation at the phenotypic and transcriptional level.

Fig. 1

Upadacitinib re-establishes the balance of monocyte subsets in circulation A Relative frequency of monocyte subsets in three rheumatoid arthritis (RA) patients before (Pre-Upa) and three months after Upadacitinib treatment (3mo-Upa) and in nine normal donors sex-matched (p values are indicated, linear mixed models were used to examine monocyte subset changes between Pre-Upa vs 3mo-Upa; student's t-test for independent groups was used to detect monocyte subset changes in Pre-Upa and 3mo-Upa compared to normal donors). B A representative plot of CD14 and CD16 expression in one RA patient before and three months after Upadacitinib treatment. C GSEA on the ranked comparison of the 3 month Upadacitinib (3mo-Upa) treated versus pre-treated (Pre-Upa) monocyte transcriptomes, using the genes significantly overexpressed in CD16 + and in CD14 + monocytes (GSE25913) as data set. Normalized Enrichment Score (NES) and False Discovery Rate (FDRq) are indicated. Leading edge analysis of the GSEA of the genes that define the CD16 + or CD14 + monocyte subsets (GSE94497, GSE25913, GSE16836) on the ranked comparison of the transcriptomes of 3mo-Upa versus Pre-Upa monocytes is shown in the bottom panel. In the heatmap, expression values are represented as colors, where the range of colors (red, pink, light blue, dark blue) shows the range of expression values (high, moderate, low, lowest). D Heatmap of the expression of genes significantly (p < 0.05) altered by Upa treatment. For each gene, mRNA expression level is represented after normalizing gene expression and k-means clustering using Genesis (http://genome.tugraz.at/genesisclient/). The group of genes whose expression is either up-regulated or down-regulated by 25% after Upadacitinib treatment in the three patients is shown. E GSEA on the ranked comparison of the GM-MØ versus M-MØ transcriptomes (left) and the ranked comparison of the 3 month upadacitinib (3mo-Upa) treated versus pre-treated (Pre-Upa) monocyte transcriptomes (right), using the genes preferentially expressed by RA-specific clusters of synovial tissue macrophage (E-MTAB-8322) as data set. NES and FDRq value are indicated (FDRq < 0.01, dark filled circle; FDRq > 0.250, empty circle). The intensity of color increases with the enrichment of the gene signature

JAKi ameliorate the signs and symptoms of RA and are currently used for the treatment of RA [20, 21]. Recent research has shed light on the transcriptome of subsets of synovial tissue macrophages (STMs) across various states: in healthy subjects, RA patients, and those in RA remission [3]. In light of this, we investigated the impact of Upa on the expression of gene clusters defining distinct STM subsets. Notably, the gene clusters that define the four pathogenic macrophage subsets in RA, and whose transcriptome resembles that of GM-MØ (Clusters 5–8, Fig. 1E), were extremely responsive to Upa treatment, as their expression was significantly diminished in 3mo-Upa (Fig. 1E). Conversely, the expression of the gene clusters that define four of the five STM subsets associated to homeostasis and/or remission (Clusters 0–4, Fig. 1E), which exhibit a transcriptome akin to anti-inflammatory M-MØ, were increased in 3mo-Upa (Fig. 1E). As a whole, treatment with Upa causes CD14 + monocytes from RA patients to exhibit reduced expression of genes associated to pathogenic STM subsets, and, concurrently, to acquire the expression of genes characterizing macrophages from RA patients in remission. These observations align well with the established therapeutic action of JAKi.

Upadacitinib promotes the generation of monocyte-derived macrophages with an anti-inflammatory transcriptional and functional profile

The phenotypic and transcriptional effects of Upa treatment on human peripheral blood monocytes led us to hypothesize that JAKi might also modify the differentiation of monocyte-derived macrophages, which drive pathogenesis in most inflammatory diseases [41]. To directly address this hypothesis, we evaluated the effect of JAKi on the differentiation of monocyte-derived macrophages in response to GM-CSF, a major pathogenic cytokine in RA [7, 10]. To that end, 10 nM or 100 nM Upadacitinib (Upa) was added each day along the GM-CSF-dependent monocyte-to macrophage (GM-MØ) differentiation process to generate 10Upa-GM-MØ or 100Upa-GM-MØ (Fig. 2A). These Upa concentrations, which fall within the range of Upa levels found in Upa-treated RA patients [29], had no effect on macrophage viability (Supplementary Fig. 2A), and drastically impaired the GM-CSF-induced JAK2-dependent STAT5 and ERK phosphorylation in monocytes [42] (Fig. 2A). RNAseq revealed that 100 nM Upa not only reduces the expression of STAT5-dependent genes like CISH and PIM1 [37, 38] (Supplementary Fig. 2B) but promotes a huge shift in the GM-MØ transcriptional profile. Specifically, 100Upa-GM-MØ exhibited significantly (|log2FC|> 1; adjp < 0.05) altered expression of 859 genes (347 genes upregulated, 512 genes downregulated) compared to control GM-MØ (Fig. 2B, C), whereas 10Upa-GM-MØ only showed 90 differentially expressed genes (Fig. 2B). Therefore, long-term Upa treatment modifies the acquisition of the transcriptional profile of GM-CSF-dependent monocyte-derived macrophages in a dose-dependent manner (Fig. 2B, D), an effect that concurs with its dose-dependent ability to inhibit the GM-CSF-induced intracellular signaling (Supplementary Fig. 2C). Importantly, gene ontology analysis of the transcriptome of 10Upa-GM-MØ and 100Upa-GM-MØ revealed that the presence of Upa significantly (FDRq = 0.0) diminishes the expression of the pro-inflammatory “GM-MØ-specific” gene set (GSE188278) [43] (Fig. 2E), which includes genes like INHBA (Fig. 2F), and reduces the production of INHBA-encoded activin A (Fig. 2G), whose expression is particularly high in synovial macrophages from RA patients [16, 17]. Therefore, Upadacitinib weakens the expression of genes that characterize GM-CSF-dependent pro-inflammatory macrophages.

Fig. 2

Upadacitinib promotes monocyte-derived macrophages with an anti-inflammatory gene expression and functional profile A Schematic representation of the experiments. Monocytes were exposed to 10–100 nM Upadacitinib daily during macrophage differentiation process with GM-CSF and the RNA levels were determined at day 7 on GM-MØ, 10Upa-GM-MØ and 100Upa-GM-MØ. Right, immunoblot analysis of pSTAT5, STAT5, pERK, ERK and pp38 by monocytes treated for 1 h to 100 nM Upadacitinib (Upa) and exposed to GM-CSF for the indicated time points. B Number of annotated genes whose expression is regulated in GM-MØ after 7d of Upadacitinib treatment (adjp < 0.05). C Volcano plot of RNAseq results showing the 100 nM Upadacitinib-induced gene expression changes in GM-MØ. D PCA analysis of GM-MØ, 10Upa-GM-MØ and 100Upa-GM-MØ.Three independent donors are identified as I, II and III. E GSEA on the ranked comparison of the GM-MØ versus 10Upa-GM-MØ and GM-MØ versus 100Upa-GM-MØ transcriptomes, using the genes significantly modulated by GM-CSF (GM-MØ-specific markers) and M-CSF (M-MØ-specific markers) as data set. Normalized Enrichment Score (NES) and False Discovery Rate (FDRq) are indicated. F Relative expression of the indicated genes as determined by RNA-sequencing on GM-MØ, 10Upa-GM-MØ and 100Upa-GM-MØ. Mean ± SEM of 4 independent donors are shown, with the indication of the Padj. G Production of activin A, IL-10 and LGMN by GM-MØ, 10Upa-GM-MØ and 100Upa-GM-MØ. Mean ± SEM of 8 independent donors are shown (*p < 0.05, **p < 0.01, one-way ANOVA with Tukey´s post hoc test; F = 92.63 for Activin A, F = 17.85 for IL-10, F = 33.39 for LGMN). H Immunoblot analysis of CD163 and FOLR2 (down) by GM-MØ, 10Upa-GM-MØ, 100Upa-GM-MØ and monocytes differentiated with M-CSF (M-MØ). In panels A-G, vinculin or GAPDH protein levels were determined as protein loading controls and a representative experiment of two independent donors is shown. I Phagocytic activity in GM-MØ, 100Upa-GM-MØ and M-MØ. Mean ± SEM of 5 independent donors are shown (*p < 0.05, one-way ANOVA with Tukey’s post hoc test, F = 13.74). J Production of TNFα, IL-6 and IL-10 by GM-MØ and 100Upa-GM-MØ challenged with LPS for 24 h, as determined by ELISA. Mean ± SEM of 7–8 independent donors are shown (*p < 0.05, **p < 0.01, paired t-test)

Noteworthy, GSEA revealed an additional and unexpected effect of Upadacitinib, as the transcriptome of 10Upa-GM-MØ and 100Upa-GM-MØ exhibited a very significant over-representation of the genes that define anti-inflammatory monocyte-derived macrophages, namely, the M-CSF-dependent “M-MØ-specific” gene set (GSE188278) [43] (Fig. 2E), including IL10, LGMN, CD163 and FOLR2 (Fig. 2F, Supplementary Fig. 2D). Moreover, this anti-inflammatory effect of Upadacitinib was evident at the protein level, since Upa-treated macrophages had increased expression of IL-10, Legumain (LGMN), CD163 and FOLR2 (Fig. 2G, H). When compared to control GM-MØ, 100Upa-GM-MØ exhibited higher phagocytic activity and, higher levels of IL-10 and lower production of TNFα and IL-6 after exposure to LPS (Fig. 2I, J) (Supplementary Fig. 2E). All these results demonstrate that Upadacitinib not only limits the pro-inflammatory nature of GM-CSF-dependent monocyte-derived macrophages, but prompts the acquisition of anti-inflammatory features. Since the effect of Upa resembles the pro-differentiation action of M-CSF on monocytes (Fig. 2E–H), we also checked whether Upa modifies the expression of genes specifically regulated during monocyte-to-M-MØ differentiation (GSE188278). As shown in Fig. 3A, the genes exclusively upregulated along monocyte-to-M-MØ differentiation (“M-MØ >  > Monocytes”) were significantly over-expressed in 100Upa-GM-MØ, whose transcriptome also showed a reduced expression of the genes exclusively downregulated along monocyte-to-M-MØ (“Monocytes >  > M-MØ”, Fig. 3A). These results confirm the link between Upadacitinib and M-CSF-driven responses, and suggests that Upadacitinib re-programs macrophages at the transcriptional and functional level.

Fig. 3

Dose-dependent effect of Upadacitinib on monocyte differentiation A GSEA on the ranked comparison of the GM-MØ versus 100Upa-GM-MØ transcriptome, using the genes significantly modulated along the monocyte-to-M-MØ differentiation as data set. NES and FDRq value are indicated. B Heatmap of the expression of genes significantly (|log2FC|> 1; p < 0.05) altered by Upadacitinib in 10Upa-GM-MØ and 100Upa-GM-MØ, as determined by RNAseq, data is represented as read counts standardized into z-score. C Comparison of genes differentially expressed in the indicated macrophage types. D Schematic representation of the dose-dependent effect of Upadacitinib on monocyte-to macrophage differentiation

The effect of upadacitinib on monocyte differentiation is dose-dependent

As 10Upa-GM-MØ and 100Upa-GM-MØ showed different protein levels of FOLR2, CD163, Legumain and IL-10 (Fig. 2G–H), we assessed whether the transcriptional effect of Upa on macrophages was dose dependent. Comparison of the gene profile of 10Upa-GM-MØ and 100Upa-GM-MØ identified 81 differentially expressed genes (Fig. 3B). Of note, 70% of the genes with higher expression in 100Upa-GM-MØ (14 out of 20) belong to the “M-MØ-specific” gene set, while 59% of the genes with higher expression in 10Upa-GM-MØ (36 out of 61) belong to the pro-inflammatory “GM-MØ-specific” gene set (Fig. 3C). Therefore, long-term Upadacitinib treatment dose-dependently allows for the acquisition of an M-CSF-dependent profile (Fig. 3D).

Upadacitinib modulates the expression of genes that define macrophage subsets relevant in RA and tissue-resident macrophages

Since previous results have identified a set of genes preferentially expressed by macrophages from the synovium of RA patients (RAMØ) (GSE10500) [44], we next checked whether Upa affects the expression of genes preferentially expressed by RAMØ. GSEA revealed that RAMØ-specific genes are under-represented in the transcriptome of both 10Upa-GM-MØ and 100Upa-GM-MØ (Fig. 4A), thus indicating the inhibitory effect of Upadacitinib on the expression of genes that characterize pathogenic macrophages in RA. We also took advantage of recently published information on synovial tissue macrophage (STM) from RA patients [3], and analyzed the expression of gene clusters that are specific for the distinct STM subsets in 100Upa-GM-MØ. The gene profile of 100Upa-GM-MØ showed an over-representation of the genes that define the homeostatic STM MERTKpos TREM2high and FOLR2high/LYVE1pos subsets, and also the tissue-infiltrating antigen presenting HLAhigh/CLEC10Apos subset (Fig. 4A) [3, 6]. Reinforcing their homeostatic profile, TREM2high and FOLR2high/LYVE1pos subsets very significantly overexpress the gene sets that define anti-inflammatory/reparative M-MØ (Fig. 1E). Of note, and although to a lower extent, this enrichment was also seen in 10Upa-GM-MØ (Fig. 4A), further emphasizing the dose-dependent action of Upa. Consequently, exposure to Upadacitinib impairs the acquisition of the transcriptional profile of pathogenic RAMØ and also prompts the expression of genes that define STM clusters primarily involved in homeostasis and resolution of inflammation. At the functional level, 100Upa-GM-MØ exhibited higher efferocytosis than GM-MØ (Fig. 4B–C, Supplementary Fig. 2F and video microscopy) an anti-inflammatory activity related to MERTKpos STM subsets [3] and the protective role of the MerTK pathway in joint pathology [45]. In agreement with these findings, analysis of the MoMac-VERSE (a resource that identifies conserved monocyte and macrophage states and global imprinting across human tissues) [46] (GSE178209) revealed that the transcriptome of 100Upa-GM-MØ is significantly enriched in the gene cluster that defines the macrophage HES1_Mac cluster 2, that exhibits a “long-term resident”-like macrophage signature (GSE188647), as well as genes that characterize tissue-resident macrophages from various tissues and organs, including LGMN, MS4A6A and MAFB [47] (Supplementary Fig. 3A–C).

Fig. 4

Upadacitinib modulates the expression of genes that define macrophage subsets relevant in RA A GSEA on the ranked comparison of the GM-MØ versus 10Upa-GM-MØ and GM-MØ versus 100Upa-GM-MØ transcriptomes, using the genes preferentially expressed by control or RA-specific macrophages (GSE10500) and RA-specific clusters of synovial tissue macrophage (E-MTAB-8322) as data set. NES and FDRq value are indicated (FDRq < 0.01, dark filled circle; FDRq > 0.250, empty circle). The intensity of color increases with the enrichment of the gene signature. B Efferocytosis (% positive cells and mean fluorescence intensity) of GM-MØ, 100Upa-GM-MØ and M-MØ as determined by flow cytometry using staurosporine-induced CellTrace Violet-labeled apoptotic Jurkat cells. Mean ± SEM of 5 independent donors are shown (**p < 0.01; ***p < 0.001, ****p < 0.0001, one-way ANOVA with Tukey´s post hoc test, F = 137.2 for % positive cells, F = 70.32 for MFI). C Above, representative images of macrophages (GM-MØ, 100Upa-GM-MØ or M-MØ) and apoptotic Jurkat cells (blue) after 1 h of co-culture, as indicated. Below, time-lapse of 100Upa-GM-MØ co-cultured with apoptotic Jurkat cells (blue) imaged for 120 min at 2-min intervals. Bright field images correspond to indicated time-lapse frames of co-culture. The macrophage shape is encircled in yellow to show its position and dynamics during the process. Bars, 20 µm

The macrophage re-programming effect of Upadacitinib relies on the MAFB transcription factor

To identify the molecular basis of the macrophage re-programming effect of Upadacitinib we initially used Discriminant Regulon Expression Analysis (DoRothEA) [31]. The transcriptome of 100Upa-GM-MØ showed a negative enrichment in STAT1 and STAT2 regulons (Fig. 5A), as expected from the inhibitory effects of JAKi, and also exhibited diminished expression of the HIF1A-regulon, a result corroborated by the lower lactate release from 100Upa-GM-MØ cells, which also exhibited diminished levels of HIF1-regulated genes like SLC2A1, EGLN3 and AQP3 (Supplementary Fig. 4A–B) [48]. Importantly, Upadacitinib modified metabolic parameters in GM-MØ, including both glycolytic capacity and maximal ATP production (Supplementary Fig. 4C–G), indicating that JAKi alters not only the inflammatory but also the metabolic state in macrophages.

Fig. 5

JAKi increases the expression of MAFB transcription factor in macrophages A Discriminant regulon expression analysis (DoRothEA) of 100Upa-GM-MØ compared with GM-MØ. Top 25 transcription factors are shown. B GSEA of genes downregulated by siRNA MAFB and by siRNA MAF on macrophages (GSE155719) on the ranked comparison of the transcriptomes of 100Upa-GM-MØ and GM-MØ transcriptomes. NES and FDRq value are indicated. C Relative expression of MAFB and MAF as determined by RNA-sequencing on GM-MØ, 10Upa-GM-MØ and 100Upa-GM-MØ. Mean ± SEM of 4 independent donors are shown, with the indication of the Padj. D Immunoblot analysis of MAFB and pGSK3S9 by GM-MØ, 10Upa-GM-MØ, 100Upa-GM-MØ and M-MØ. E Immunoblot analysis of MAFB and CD163 along the monocyte to macrophage differentiation in the presence of 100 nM Upadacitinib (Upa). In panels (D-E), GAPDH protein levels were determined as protein loading control and a representative experiment of two (E) and four (D) independent donors is shown. F Schematic representation of the experiments: short term-Upadacitinib treatment to mature macropahges (GM-MØ). Immunoblot analysis of MAFB and pGSK3S9 (G) and production of activin A (H) by GM-MØ exposed to 10–100 nM Upadacitinib for the last 48 h. GAPDH protein levels were determined as protein loading control. In (G) a representative experiment of three independent donors is shown. In (H) mean ± SEM of 5 independent donors are shown (*p < 0.05, F = 12.28). I Immunoblot analysis of MAFB in two independent preparations of differentiating GM-MØ (day 2) generated from monocytes exposed to DMSO (−), Upadacitinib (Upa, 100 nM), STAT5 phosphorylation specific inhibitor (ST5i, 50 µM) or MEK1/2 inhibitor (UO, 2,5 µM). Right, quantification of MAFB expression. J Immunoblot analysis of MAFB in two independent preparations of differentiating GM-MØ (day 2) generated from monocytes transfected with either siCNT or MAFB-specific siRNA (siMAFB) and exposed to DMSO (−) or 100 nM Upadacitinib (+). Right, quantification of MAFB expression. In panels I-J, mean ± SEM of the relative MAFB protein levels in the macrophage subtypes from four independent donors are shown (*p < 0.05, **p < 0.01). K Relative mRNA expression of the indicated MAFB-dependent genes in siCNT GM-MØ, siMAFB GM-MØ, siCNT Upa-GM-MØ and siMAFB Upa-GM-MØ (day 2). Mean ± SEM of four independent experiments are shown (*p < 0.05; **p < 0.01; ***p < 0.001, F = 3.7 for IL10, F = 6.8 for CMKLR1, F = 4.4 for LGMN, F = 56.89 for CD163, F = 9.24 for FOLR2, F = 8.34 for MS4A6A)

Conversely, DoRothEA revealed that the gene profile of 100Upa-GM-MØ is highly enriched in MAFB and MAF regulons (Fig. 5A), in agreement with the gene ontology data (Supplementary Fig. 3) and overexpresses the “M-MØ-specific” gene set [43] (Fig. 2E), whose M-CSF-driven acquisition is MAFB/MAF-dependent [10, 12, 49, 50]. Supporting the significance of these findings, the transcriptome of 100Upa-GM-MØ showed a high over-representation of MAFB- and MAF-dependent genes (Fig. 5B). As MAFB gene expression was also enhanced in 100Upa-GM-MØ (Fig. 5C), we next assessed MAFB protein level in macrophages generated in the presence of Upa. As shown in Fig. 5D, 100Upa-GM-MØ exhibited elevated MAFB protein levels as well as enhanced levels of inactive (Ser9-phosphorylated) GSK3β, whose active form limits MAFB protein and activity [51]. The higher expression of MAFB was observed at all time points along 100Upa-GM-MØ differentiation (Fig. 5E) and matched with a progressive enhancement of the expression of the MAFB-dependent protein CD163 (Fig. 5E). Therefore, the macrophage re-programming activity of Upadacitinib coincides with a dose-dependent increase in the expression of MAFB and MAFB-targets [12]. In fact, short-term exposure to Upa (two doses in the last 48 h of differentiation of GM-MØ) (Fig. 5F) sufficed to increase the expression of MAFB, enhance the inhibitory phosphorylation of GSK3β (Fig. 5G) and reduce activin A production (Fig. 5H), further reinforcing the link between the re-programming action of Upadacitinib and MAFB.

As blocking JAK2 activation with Upadacitinib results in impaired GM-CSF-induced STAT5 and ERK phosphorylation (Fig. 2A), we next compared MAFB expression in GM-CSF-primed monocytes exposed to Upa or known inhibitors of STAT5 and ERK. Interestingly, Upa treatment results in a stronger MAFB expression than STAT5 or ERK-activating inhibitors (Fig. 5I), suggesting that, besides STAT5 and ERK, JAK2 might trigger additional intracellular signaling directly involved in controlling the anti-inflammatory differentiation of human macrophages. In this regard, and to determine whether MAFB mediates the macrophage re-programming action of Upadacitinib, we evaluated the effect of Upa after siRNA-mediated MAFB knock-down in monocytes. Importantly, knock-down of MAFB (Fig. 5J) significantly reduced the positive effect of Upa on the expression of IL10, CMKLR, LGMN, CD163, FOLR2 and MS4A6A (Fig. 5K). Therefore, MAFB mediates the reprogramming action of Upadacitinib during the GM-CSF-dependent differentiation of human monocyte-derived macrophages.

Macrophage re-programming by other JAK inhibitors

To date, five different JAKi (Tofacitinib, Baricitinib, Upadacitinib, Peficitinib and Filgotinib) have been approved for the treatment of RA [20, 24]. Given the re-programming action of Upadacitinib, we next asked whether other JAKi also exhibit a similar effect on the GM-CSF-dependent monocyte-derived macrophages. After checking for minimal effects on cell viability (Supplementary Fig. 2A), monocytes were exposed to different concentrations of Baricitinib (Bari, JAK1/2 inhibitor), Tofacitinib (Tofa, JAK1-3 inhibitor), Peficitinib (JAK3 inhibitor), Filgotinib (JAK1 inhibitor) and the TYK2 inhibitor Deucravacitinib each day along the monocyte-to macrophage differentiation process. Like Upa, long-term Bari treatment enhanced expression of the anti-inflammatory “M-MØ-specific” gene set, diminished expression of the pro-inflammatory “GM-MØ-specific” gene set, and augmented the expression of genes that define the tissue-resident homeostatic TREM2high and FOLR2highLYVE1pos subsets, as well as tissue-infiltrating antigen presenting HLAhigh/CLEC10Apos subsets (Fig. 6A), from RA patients [