Elevated SP140 expression in inflammatory diseases and mucosal macrophages of CD patients
SP140 gene expression in a variety of cells and tissues was analyzed using an in-house GSK microarray profiler. SP140 was predominantly expressed in blood, CD8+ and CD4+ T cells, monocytes, and secondary immune organs such as lymph node and spleen (Fig. 1a). We then investigated SP140 expression in white blood cells (WBC) and intestinal tissue of IBD patients versus controls. While SP140 gene expression in WBC was comparable in IBD and healthy controls (Fig. 1b), SP140 gene and protein expression were found to be increased in inflamed colonic mucosa of both CD and UC patients (Fig. 1c, d). Notably, an increase in CD68+SP140+ and HLA-DR+SP140+ but not CD68+SP140− or HLA-DR+SP140− cells was observed in CD inflamed colonic mucosa compared to normal control mucosa or uninflamed tissue, respectively (Fig. 1e–h). Elevated SP140 expression was also found in other inflammatory conditions including in WBC of systemic lupus erythematosus and rheumatoid arthritis patients (Fig. 1b), and in inflamed tissues of appendicitis, sarcoidosis, psoriatic arthritis, rheumatoid arthritis, Hashimoto’s thyroiditis and Sjogren’s syndrome patients (Additional file 1: Fig. S1b). High expression of SP140 was also observed in chronic lymphocytic leukemia (Fig. 1b).
Next, we analyzed a publicly available single-cell RNA sequencing experiment [24] of inflamed and uninflamed ileal tissue (biopsies) obtained from CD patients. Unsupervised clustering analysis identified 22 cluster blocks (Additional file 1: Figs. S2a and b). Based on different cell markers, we identified cell types existing in each cluster (Fig. 2a, b). Expectedly, SP140 expression was predominantly observed in the immune cell compartment (Fig. 2c). Based on the expression of CD14, CD16 (FCGR3A), CD1C, CD64 (FCGR1A), CD68, CD163, CD206 (MRC1), HLA-DR, and CLEC4A (Fig. 2a and Additional file 1: Fig. S2c), we identified cluster 6 as containing monocytes, macrophages, and dendritic cells (Additional file 1: Fig. S2c and Fig. 2b), altogether representing mononuclear phagocytes (MNP). As per the observations made by Martin et al. [24], the abundance of the MNPs was significantly higher in inflamed ileal biopsies (Additional file 1: Fig. S2d), which accordingly manifested as an increased percentage of SP140-expressing MNPs in inflamed ileal biopsies (Fig. 2d, e). Interestingly SP140 expression did not differ significantly in other immune cells (including T and B cells) between inflamed and uninflamed tissues (Fig. 2e). Through subclassification analysis of the MNPs, we mapped the monocytes (CD14 and FCGR3A) along their developmental trajectory towards macrophages (CD68, CD163, and CD206) or dendritic cells (CD1C and CLEC4A) (Additional file 1: Fig. S2e). Projecting the expression of SP140 along the inferred trajectory suggests significant association with the developmental process (Additional file 1: Fig. S2f).
SP140 mediates inflammatory “M1” macrophage function
Given the high SP140 expression in CD mucosal macrophages, we investigated whether SP140 is associated with inflammatory activation. Human CD14+ monocytes or THP-1 cells were differentiated into macrophages and then polarized into “M1” and “M2” phenotypes using IFN-γ and IL-4 respectively or left without treatment (M0) (Fig. 3a and Additional file 1: Fig. S3b). The polarization was verified by measuring gene expression of CD64 and CCL5 (markers of “M1” macrophages) and CD206 and CCL22 (markers of “M2” macrophages) (Additional file 1: Fig. S3a and b). SP140 mRNA was expressed at significantly higher levels in “M1” compared to “M0” or “M2” macrophages derived from primary monocytes (Fig. 3b) or THP-1 cells (Additional file 1: Fig. S3c). Protein staining showed increased numbers of SP140 protein-containing speckles in the nucleus of “M1” compared to “M2” macrophages derived from both human monocytes (Fig. 3c) and THP-1 cells (Additional file 1: Fig. S3d). LPS induced an increase in SP140 gene expression in “M0” and “M1” macrophages, along with enhanced CCL5 and decreased CCL22 gene expression (Additional file 1: Fig. S3e). The expression levels of 12 other BCPs (SP140L, SP100, SP110, BRD2, BRD3, BRD4, BRD9, BAZ2A, BAZ2B, PCAF, EP300, and CREBBP) showed no increase in “M1” compared to “M0” macrophages (Additional file 1: Fig. S4a), indicating that the association between SP140 and inflammatory macrophages was not a common phenomenon among BCPs.
To investigate SP140 function, we initially used siRNA-mediated knockdown to reduce SP140 expression in “M1” macrophages, achieving an approximately 75% reduction in mRNA (Fig. 3e) and a clear decrease in the number of SP140 protein-containing speckles (Fig. 3f). Knockdown seemed specific to SP140, as expression of related family members SP110, SP100, and SP140L was not affected (Additional file 1: Figs. S4b, b and c and Additional file 2: Table S1 and Additional file 3: Table S2). After the knockdown, cells were stimulated with LPS or kept unstimulated (Fig. 3d). SP140 silencing led to a decrease in LPS-induced IL-6 and TNF mRNA and protein levels (Fig. 3g). After transcriptional profiling, principal component analysis (PCA) revealed that the variance was most associated with SP140 siRNA treatment in LPS-stimulated “M1” macrophages (Fig. 3h). Downregulation of some key CD-associated genes, notably CXCL9 [25], CEACAM1 [26], and JAK2 [27, 28], was apparent among SP140-siRNA-treated unstimulated macrophages, and IL2RA [29] and TRIM69 [30] among SP140-siRNA LPS-stimulated macrophages (Fig. 3i). SP140 silencing affected common inflammatory pathways, such as TNF signaling via NFKB, IFN-γ response, inflammatory response (Fig. 3j), and cytokine signaling (Fig. 3k).
The development of a selective inhibitor of SP140 (GSK761)
To identify selective SP140 binding compounds, we utilized encoded library technology to screen the GSK proprietary collection of DNA-linked small molecule libraries [31, 32]. Affinity selection utilizing a recombinant protein construct spanning the PHD and Brd domains of SP140 was carried out, leading to the identification of an enriched building block combination in DNA Encoded Library 68 (DEL68) (Fig. 4a). Representative compounds of the identified three-cycle benzimidazole chemical series were synthesized, which yielded the small molecule, GSK761 (Fig. 4b, c). More details describing GSK761 development are added in the “Methods” section. To quantify the interaction between GSK761 and SP140 (aa 687-867), the dissociation constant (Kd) was determined using a fluorescence polarization (FP)-binding approach. A fluorophore-conjugated version of GSK761, GSK064 (Fig. 4d), was prepared and subsequently used to determine a Kd value of 41.07 ± 1.42 nM for the interaction with SP140 (aa 687-867) (Fig. 4e). To further validate this interaction, competition studies were carried out using GSK761 in a FP-binding assay configured using GSK064 and SP140 (aa 687-867). Competitive displacement of GSK064 from SP140 (aa 687-867) by GSK761 was observed, subsequently leading to the determination of an IC50 value of 77.79 ± 8.27 nM (Fig. 4f). Prior to utilizing GSK761 for cell-based binding studies to endogenous SP140, the cell penetration capacity of the compound was assessed using mass spectrometry and the methodology described by [33]. Concentration measurements showed that GSK761 had a pΔCtotal value of 1.45 ± 0.21, indicating that GSK761 permeates human cells and accumulates intracellularly by an order of magnitude when compared to GSK761 free in solution.
To confirm the binding of GSK761 to full-length SP140, immobilized GSK761 was used to probe for SP140 in nuclear extracts from HuT78 cells and HEK293 cells transfected with Halo-tagged SP140. Biotinylated beads only and untransfected HEK293 cells were utilized as controls. Both endogenous and Halo-tagged SP140 were pulled down with biotinylated GSK761 and visualized via Western Blotting (Fig. 4g). Endogenous SP140 is observed as a doublet containing the four largest isoforms (predicted 86–96 kDa) [10] while Halo-tagged SP140 exhibits a single band. No bands were detected with beads only or in untransfected HEK293 cells. The specificity of GSK761 for SP140 was profiled using the BROMOscan® assay, in which DNA-tagged BCPs were incubated with an increasing concentration of GSK761 or DMSO. Binding is assessed by measuring the amount of bromodomain captured in GSK761 vs DMSO samples by using an ultra-sensitive qPCR. There was evidence of low affinity interaction between GSK761 and several tested BCPs, with binding detected at concentrations >21,000 nM (Additional file 4: Table S4). However, no binding was detected at concentrations ≤ 21,000 nM indicating a high degree of specificity of GSK761 for SP140 (Additional file 4: Table S4).
GSK761 reduces the inflammatory activation of macrophages and expression of pro-inflammatory cytokines
SP140 expression was selectively increased in “M1” compared to “M0” macrophages, prompting us to test whether SP140 is required for the polarization to an inflammatory macrophage phenotype. We first tested whether GSK761 demonstrated any toxicity in macrophages to determine the appropriate dose range. At ≤0.12 μM, GSK761 showed no cytotoxicity (Additional file 1: Figs. S5a and b). To investigate the effect of GSK761 on macrophage polarization to inflammatory phenotype, “M0” macrophages were treated with either DMSO or GSK761 for 3 days in presence of IFN-γ or IL-4 (during the polarization to “M1” and “M2” macrophages, respectively). GSK761 enhanced mRNA expression of the anti-inflammatory marker CD206 in both “M1” and “M2” macrophages and decreased the pro-inflammatory marker CD64 in “M1” macrophages (Fig. 5a). FACS analysis showed that GSK761-treatment prior to IFN-γ (“M1”) polarization reduced CD64+ cells and CD64 protein expression (Fig. 5b, c) and increased CD206+ cells and CD206 protein expression (Fig. 5d, e). Adding GSK761 to the cells during IFN-γ (“M1”) polarization lowered TNF (“M1” polarization marker) gene expression in these cells compared to the control (Additional file 1: Fig. S5d). Altogether, these data suggest that SP140 inhibition during “M1” polarization biased differentiation towards a regulatory “M2” phenotype.
We next assessed the effect of SP140 inhibition on the response of “M1” polarized macrophages to inflammatory stimuli. LPS-stimulated “M1” macrophages pretreated with GSK761 showed a strong reduction in secretion of IL-6, TNF, IL-1β, and IL-12 (Fig. 5g) at concentrations where cytotoxicity was not observed (0.04 and 0.12 μM). When employing RNA transcriptional profiling using a customized qPCR array, SP140 inhibition was found to reduce the expression of many other pro-inflammatory cytokines and chemokines including GM-CSF, CCL3, CCL5, and CCL1 (Fig. 5f).
The marked anti-inflammatory effects of GSK761 on macrophages in vitro suggest that targeting SP140 may be an effective approach for IBD. Unfortunately, due to poor in vivo pharmokinetics (data not shown), GSK761 was not suitable to evaluate the effects of SP140 inhibition in in vivo animal models of colitis. To evaluate the impact of SP140 inhibition in human CD, CD14+ mucosal macrophages were isolated from CD anti-TNF refractory patients’ colonic mucosa and then cultured in vitro with either DMSO or GSK761 for 4 h. Spontaneous gene expression of TNF, IL6, and IL10 was significantly decreased in GSK761-treated tissue macrophages, demonstrating that GSK761 inhibited their immune reactivity (Fig. 5h). No changes in CD64 expression were observed. Previous studies have demonstrated that successful anti-TNF therapy is associated with an increase in CD206+ regulatory macrophages [34] and a decrease in pro-inflammatory cytokines [10] and SP140 expression [10] in CD mucosa. Thus, our data suggests that SP140 inhibition could be also beneficial in supporting the anti-TNF therapy. Thus, combining SP140 inhibition with anti-TNF may potentiate the anti-inflammatory effects in macrophages and possibly in IBD in general. Consistent with this idea, M1 macrophages treated in vitro with both GSK761 and Infliximab (anti-TNF) demonstrated an additive reduction in TNF release in response to LPS compared to those treated with GSK761 or infliximab alone (Additional file 1: Fig. S5e).
SP140 preferentially interacts with transcription start sites (TSS) and enhancer regions
SP140 possesses multiple chromatin binding domains, namely Brd, PHD, and SAND (SP100, AIRE-1, NucP41/75, DEAF-1) domains (Additional file 1: Fig. S1a), which may allow it to function as an epigenetic reader [10, 35]. To assess which histone peptides might be bound by SP140, we utilized an Active Motive histone peptide array, which showed strong binding of SP140 PHD-Brd to unmodified H31-19 and H326-45 peptides, while little binding to unmodified H2A, H2B, and H4 peptides was detected (Additional file 1: Fig. S6a). SP140 PHD-Brd was able to bind several modified H3 peptides bearing some methylation and acetylation marks including K27Ac, K14Ac, K4me3, and K9me3, and also to acetylated H4 peptides (Additional file 1: Fig. S6a).
To measure histone peptide binding in a more quantitative way, and to evaluate peptides for which binding is specifically dependent on the SP140 binding site of GSK761, we tested the ability of a range of peptides to compete for compound binding to SP140 PHD-Brd. We found that recombinant SP140 PHD-Brd protein bound to histone H3 peptides with sub-μM affinity (Additional file 1: Fig. S6b). The strongest binding was observed for an unmodified peptide corresponding to the N-terminal 21 aa, while methylation at lysine residue 4 (K4) led to substantially (~30-fold) reduced affinity (Additional file 1: Fig. S6c). K9 acetylation also resulted in lower affinity binding (4-5-fold), while SP140 binding was unaffected by either acetylation or methylation of K14 (Additional file 1: Fig. S6d). Similar binding affinity was observed for unmodified H31-21 vs H31-18, while reduced (4–5-fold) but still significant SP140 binding was measured for H31-9 (Additional file 1: Fig. S6d). Taken together, these data suggest that the SP140 PHD-Brd module can function as a reader for unmodified histone H3, with binding focused around the N-terminal 9 aa, and that GSK761 competes for this binding.
To evaluate whether native SP140 is also capable of binding to histones, we used immobilized histone H3 peptides with varying modifications to precipitate proteins from nuclear extracts of anti-CD3/CD28-stimulated HuT78 T cells and probed via Western blotting for the presence of SP140. SP140 was efficiently pulled down by unmodified H31-21 (Fig. 6a). Notably, H31-21 peptides bearing certain methylation and acetylation marks (K4me3, K9me3, K9ac, and K14ac) were also capable of capturing native SP140 (Fig. 6a), despite the lower measured affinity of some of these for recombinant SP140 PHD-Brd (see above). To evaluate whether SP140 interacts with histone H3 in a cellular setting, we utilized a NanoBRET system (Additional file 1: Fig. S6e). NanoBRET is a proximity-based assay that can detect protein interactions by measuring energy transfer from a bioluminescent protein donor NanoLuc® (NL) to a fluorescent protein acceptor HaloTag® (HT). This energy transfer was observed in the nuclei of HEK293 cells transfected with SP140-NL and Histone3.3-HT DNA, indicating close proximity of the two proteins (Additional file 1: Fig. S6e).
To assess whether SP140 associates with chromatin in macrophages and how this might be regulated in inflammation, we initially conducted ChIP-qPCR experiments in unstimulated or LPS-stimulated “M1” macrophages. First, the specificity of the antibody used for ChIP experiments to precipitate SP140 protein was verified (Additional file 1: Fig. S7a). Binding of SP140 to the TSS of TNF and IL6 genes was observed in unstimulated cells, and this was increased following LPS stimulation (Fig. 6b). We then evaluated chromatin occupancy of SP140 on a genome-wide level using ChIP-seq and tested the effects of GSK761 on both SP140 binding and gene expression (RNA-seq) in the context of LPS stimulation. In DMSO-treated, LPS-stimulated macrophages, epigenome roadmap scan analysis revealed that the majority of SP140 occupancy was at strong transcription associated regions, enhancers, and TSS regions (Fig. 6c). However, this occupancy was decreased when the cells were pretreated with GSK761 at 1 h (Additional file 1: Fig. S7c) and 4 h (data not shown) post LPS stimulation. Heatmap rank ordering of SP140 occupancy in unstimulated and LPS-stimulated “M1” macrophages showed a strong SP140 enrichment at the TSS (Fig. 6d). The top 20 enriched genes belonged mostly to those involved in the innate immune response, including TNF, ICAM4, IRF1, LITAF, TNFAIP2, and NFKBIA (Fig. 6d). This was confirmed when Metagene analysis showed that SP140 preferentially binds near the TSS of immune innate genes, while this binding was minimal at the TSS of non-immune genes (Additional file 1: Fig. S7b). However, the most SP140-enriched gene FLOT1 (Fig. 6d) has been reported to be strongly involved in tumorigenesis [36] and anti-fungal immunity [37]. We found SP140 also to occupy active enhancers, as marked by H3K27Ac in human macrophages by overlapping the publicly available H3K27Ac ChIP-seq data (GSE54972) with our SP140 ChIP-seq dataset at 1 h of LPS stimulation (Additional file 1: Fig. S7d). ChIP-qPCR showed a strong reduction of SP140 occupancy at the TNF-TSS in GSK761-treated “M1” macrophages, which was reduced to that of unstimulated cells (Fig. 6e), correlating with the previously observed reduced expression of TNF in GSK761-treated macrophages.
Following LPS stimulation, we observed a marked increase in binding at the TSS of the top 1000 genes occupied by SP140, reaching its maximum at 4 h (Fig. 6f). GSK761 treatment prior to LPS stimulation strongly reduced SP140 occupancy at the TSS (Fig. 6f) associated with an altered LPS-induced gene expression (Fig. 6g). PCA of RNA-seq data revealed a large separation in global gene expression between DMSO- and GSK761-treated macrophages at both time points post of LPS stimulation (4h and 8h) (Fig. 6g). Note that SP140 itself did not appear to be bound by SP140 protein, nor was its expression altered by GSK761 treatment (Additional files 5, 6 and 7: Table S5-7 and Additional files 12 and 13: Table S12-13).
Gene Ontology analysis of SP140-bound genes showed enrichment of genes that participate in the processes of cytokine response and immune response that were inhibited through GSK761 pre-exposure (Fig. 6h). To elucidate the regulatory pathways enriched by SP140 and affected by GSK761, we carried out hallmark pathway analysis for SP140 differentially bound genes (DBGs) (Fig. 6i) and DEGs (Fig. 6j) upon DMSO or GSK761 treatment at each time point of LPS stimulation. In DMSO-treated “M1” macrophages, we found that SP140 binding is predominantly enriched at many pathways typically defined as inflammatory such as TNF signaling via NFKB, and inflammatory response (Fig. 6i). However, this enrichment was no longer seen in GSK761-treated macrophages (Fig. 6i). Similar pathways were observed for DEGs and were downregulated in GSK761-treated macrophages, especially at 4 h (Fig. 6j). In addition, we observed an upregulation of MYC targets and oxidative phosphorylation gene sets in GSK761-treated macrophages, whereas these normally are downregulated because of the induction of aerobic glycolysis upon LPS stimulation of macrophages (Additional file 1: Fig. S8a) [38]. These data suggest SP140 as a critical regulator of genes involved in the inflammatory response and that GSK761 can inhibit this response through displacing SP140 binding to TSS and enhancer regions.
SP140 preferentially controls the expression of specific gene sets involved in the innate immune response
We next investigated the functional significance of inhibiting SP140 binding for gene expression. Heatmap (Top 100) and volcano plots of DEGs after 4 h LPS stimulation demonstrated a strong effect of GSK761 on gene sets that are involved in the innate immune response, such as the downregulation of TNFSF9, IL6, F3, CXCL1, CCL5, IL1β, TRAF1, IL23A, IL18, and GEM and the upregulation of PLAU and CXCR4 (Fig. 7a, b). We next explored the global DBGs in 1 h LPS-stimulated “M1” macrophages using R2 TSS plot (Fig. 7c) and R2 TSS-peak calling (Fig. 7d). Interestingly, the top DBGs belonged to TNF signaling via NFKB such as TNFAIP2, TNF, LTA, TRAF1, IRF1, and NFKBIA (Fig. 7c, d). GSK761 clearly reduced SP140 binding to those genes (Fig. 7d). Similarly, gene expression of TNFAIP2, TAF1, TNF, LTA, and IRF1 was reduced (Fig. 7e). By conducting a focused TNF signaling enrichment analysis (Additional file 1: Fig. S8b) and R2 TSS plot analysis (Additional file 1: Fig. S8c), we illustrated a new set of DBGs including TFs (TNFAIP3, CEBPB, ATF4, MAP3K8, MAP2K3, and JUNB), cytokines (IL1β, IL23A), and adhesion molecules (ICAM1). Most of these genes were DEGs (Additional file 1: Fig. S8d).
We next verified the impact of global reduced SP140 binding on gene expression by integrating the DBGs and the DEGs. Interrogating the top 1000 DBGs (at 1 or 4 h LPS) for their gene expression (at 4 or 8 h LPS) implicated genes that are involved in TNF signaling pathway (NFKBIA, SOD2, LITAF, SGK1, PNRC1, IRF1, and RNF19B) and cytokine-mediated signaling pathway (For example: IL10RA, NFKBIA, IRF1, SOD2, CAMK2D, IRF2, and ISG15) which showed the strongest concordant differential binding and expression (Fig. 7f and S9).
In addition to SP140 binding to chromatin, we speculated that SP140 may also interact with other TFs proteins to regulate gene expression. To this end, we performed homer known motif enrichment analysis (HKMEA) at DMSO_1h LPS (Fig. 7g). We also conducted a homo de novo motif discovery to define new DNA sequence motifs that are bound by SP140 (Fig. 7g). Interestingly, HKMEA suggests that SP140 may bind the same DNA sequence motifs recognized by certain TFs defined as key proteins in TNF signaling via NFKB (ATF3, FOSL2, FOSL1, NFKB-p65-rel, JUNB, and JUN-AP1) and of cytokine-mediated signaling pathway (BATF, NFKB-p65-rel, JUNB, IRF8, and IRF3) (Fig. 7g).
LPS stimulation strongly recruits SP140 to multiple human leukocyte antigen (HLA) genes, including HLA-A, HLA-B, HLA-C, HLA-F, HLA-DPA, and HLA-DPB (Additional file 1: Fig. S8e), binding which was strongly reduced by GSK761. This indicates a role of SP140 in regulating antigen presentation-associated genes. Notably, the Brd-containing protein 2 (BRD2) gene was occupied by SP140, and GSK761 reduced this binding and lowered BRD2 gene expression (Additional file 1: Fig. S8e and Additional file 5: Table S5). BRD2 has been reported to play a key role in inflammatory response in murine macrophages and in inducing insulin resistance [39].
Finally, we noticed that SP140 occupancy on TSS of chemokine activity genes was dramatically reduced by GSK761, inducing a reduced global chemokine activity at mRNA level (Fig. 7h). R2 TSS plots and R2 TSS-peak calling indicate a strong SP140 occupancy at several inflammatory macrophage-associated chemokines (CCL2, CCL3, CCL4, CCL5, CCL8, CXCL1, CXCL3, CXCL8 CXCL9, CXCL10, and CXCL11), chemokine ligands (CCL4L1, CCL3L1, CCL3L3, CCL4L2), chemokine receptors (CCR7), TFs (STAT1, JAK2, PIK3R5, RELA), and protein kinases (PRKCD, FGR, and HCR) (Additional file 1: Figs. S10a,b and c). This binding was reduced by GSK761, affecting expression of many genes involved in chemokine signaling (Additional file 1: Fig. S10d). However, SP140 binding was selective as it did not bind to the TSS of a range of other chemokines such as CXCL6, CXCL5, CCL11, and CCL7 (Additional file 1: Fig. S10c).