Mice were housed in accordance with the directives and guidelines of the Swedish Board of Agriculture, the Swedish Animal Protection Agency, and the Karolinska Institutet (djurskyddslagen 1988:534; djurskyddsförordningen 1988:539; djurskyddsmyndigheten DFS 2004:4). The study was performed under approval of the Stockholm North Ethical Committee on Animal Experiments permit number 1374–2020 and 2476-2025.

Bone marrow cells were flushed from tibia and femurs with PBS, filtered through a 70μm cell strainer, resuspended in DMEM supplemented with 10% FCS and 30% L929 cell-conditioned medium (as a source of macrophage-colony stimulating factor) and incubated for 6 days at 37°C, 5% CO₂. Bone marrow-derived macrophage (BMM) cultures were then washed with PBS, detached with 1%trypsin, and 5x10⁵ cells were seeded per well in 24-well plates. BMM were further incubated for 24 h at 37°C before infections or treatment with diverse compounds. BMM were F4/80+, CD11b+, CD11c- and SiglecF- (23).

M. tuberculosis Harlingen or H37Rv carrying the green fluorescent protein (GFP)-encoding pFPV2 plasmid were grown in Middlebrook 7H9 (Difco, Detroit, MI) supplemented with albumin, dextrose and catalase and quantified by densitometry. BMM were infected with sonicated bacteria at a multiplicity of infection (MOI) of 2 unless otherwise indicated. After 4 h, cells were washed twice with PBS to remove extracellular bacteria and further incubated for 1 to 5 days. The infected BMM were then detached using trypsin-EDTA at different times after infection, incubated with live/dead stain (LIVE/DEAD™ Fixable Yellow Dead Cell Stain, Invitrogen), washed with FACS buffer (PBS containing 0.5% FCS and 0.5 mM EDTA] and fixed with 4% formaldehyde (Sigma-Aldrich) at room temperature for 10 min. Data were acquired on a Sony ID7000 spectral cytometer and analyzed with FlowJo software (Tree star Inc., Ashland, OR). Cells within the mononuclear gate were ≥ 95% viable, and ≥ 95% of live cells are CD11b+ F4/80+. This procedure allowed the exclusion of Mtb associated with dead cells and enable determination of the frequency of infected cells and the relative levels of intracellular bacteria (23).

siRNA transfections were performed as previously described (24), using sequences listed in the Supplementary Table 2. BMM were plated at 1 × 10⁶ cells per ml in 24-well plates overnight. On the day of transfection, the media was replaced with 450 μl DMEM without penicillin/streptomycin or L-929 cell supernatants. For each target gene, two tubes were prepared. Optimem (25 μl per well) was added to each tube. Lipofectamine RNAimax (2.5 μl per well; Invitrogen) was added to one tube and siRNA (50 nM per well, unless otherwise indicated) was added to the second tube. The siRNA was then combined with the vial with RNAimax, mixed well by pipetting and incubated for 15 min. Then, 50 μl of the mix was incubated with the BMM cell culture. Twenty-four hours after transfection, cells were infected or incubated with different molecules.

Total RNA was extracted from lung samples or culture cells using Trizol (Sigma Aldrich) and cDNA was obtained by reverse transcription. Transcripts were quantified by real time PCR as described (25). Transcripts were quantified using hprt as a control house-keeping gene to calculate the ΔCt values for individual samples. The relative number of transcripts was calculated using the 2^-(ΔΔCt) method. The primer sequences used are listed in Supplementary Table 1. These values were then used to calculate the relative expression of mRNA in the different conditions (infection and/or treatment) used in tissues and cells.

The concentration of IL-1β in BMM supernatants were quantified by ELISA, according to the manufacturer’s instructions (BioLegend ELISA MAX™ Standard Set Mouse IL-1β).

RNA was extracted from BMM, infected or not with M. tuberculosis and treated with the USP18 inhibitor 4 h before infection. RNA was isolated using the miRNeasy micro kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions and processed for sequencing at bioinformatics and expression analysis at Novogene Biotech Co (Cambridge, UK). The RNA quality was assessed on a 2200 TapeStation Instrument (Agilent, Santa Clara, CA). PolyA RNA selection was performed using the Illumina TruSeq RNA Sample Preparation Kit according to the manufacturer’s protocol. RNA-seq libraries were prepared and sequenced on the Illumina HiSeq 2000 platform. All the FASTQ files that passed QC were quantified by Salmon (26) using GRCm39 as the reference transcriptome. Differential gene expression analysis was performed with R (version 4.3.3) using the Limma package (27). Genes with adjusted p-value less than 0.05 were and fold change larger than 2 were identified as differentially expressed. All the FASTQ files that passed QC were quantified by Salmon using GRCm39 as the reference transcriptome (26). Differential gene expression analysis was performed with R (version 4.3.3) using the Limma package (27). Genes with adjusted p-value less than 0.05 were and fold change larger than 2 were identified as differentially expressed genes (DEGs). Gene Ontology enrichment analysis (Biological Process, Molecular Functions) or KEGG pathway enrichment was performed with WebGestalt (http://www.webgestalt.org) using default parameters. The raw and processed RNA sequencing data can be found at Gene Expression Omnibus (GEO) repository.

The selection and development of the small molecular inhibitors of USP18 used 2K04 and BB7 have been recently described (28). Briefly, 2K04 was selected from a chemical library of candidate DUBs screened for the inhibition of the USP18 biochemical activity. BB7 is an optimized inhibitor derived by medicinal chemistry and NC is a stereoisomer of BB7 that shows no USP18 inhibition. The inhibitors formed covalent interactions with USP18, and were cell penetrating. The compounds demonstrated an exceptional specificity for murine USP18 across 41 deubiquitinases including 22 USPs (Ubiquitin Specific Proteases) from the related USP family of DUBs (28). Both compounds inhibited mUSP18 enzymatic activity with IC₅₀ at the nanomolar range and inhibited the engagement of cellular USP18 with ISG15 at 1 μM. No BMM toxicity was observed at the concentrations used in our study.

BMMs were lysed in RIPA buffer (25mM Tris-Cl pH 7.5, 50 mM NaCl, 0.5% NP40, 0.1% SDS, 0.5% DOC, 1mM DTT, 20mM NEM supplied with fresh Protease and Phosphatase inhibitors). A syringe with 29G needle was used to mechanically disrupt genomic DNA. Lysates were clarified at 10000 g for 15 min at 4°C. Protein concentration (DC Protein Assay Kit, Bio-Rad) was determined on the clear supernatants. Desired concentrations of cell lysates were denatured for 10 min at 95°C in loading buffer (NuPage 4X, Reducing Agent 10X, Invitrogen, Carlsbad, CA) and loaded in acrylamide Bis-Tris 4-12% gradient gels (Invitrogen, Carlsbad, CA). After transfer onto PVDF membranes (Merck Millipore, Bedford, MA) for 60 min at 0.34 A, the filters were blocked in TBS-T solution (50 mM Tris-Cl, 150 mM NaCl, 0.1% Tween-20 and 5% non-fat milk, pH 7,6), and incubated with specific primary antibodies (polyclonal rabbit anti-ISG15, monoclonal rabbit anti-phospho-IκBα pSer32, monoclonal rabbit anti-total IκBα), overnight at 4°C, following an incubation with the appropriate horseradish peroxidase-conjugated secondary antibodies for 1 h at room temperature. Proteins were visualized by chemiluminescence (ECL, GE Healthcare, Uppsala, Sweden), detected by the ChemiDoc MP Imaging System (Bio-Rad Laboratory, CA), and chemiluminescence intensity was analyzed only for non-saturated bands using the corresponding imaging analysis software, ImageJ version 5.0.

The coding sequences for the full-length and the cap free (removing the C-terminal hexapeptide cap) mouse ISG15 protein (Uniprot Acc. Nr Q64339) were obtained as synthetic genes (Thermo Fisher, GeneArt). The coding sequences were cloned in the pNIC28Bsa4 expression vector (GenBank Acc. Nr. EF198106). The expression constructs contain an N-terminal hexahistidine tag and a tobacco etch virus (TEV) protease recognition site for affinity tag removal; the resulting products carry one extra N-terminal serine residue. The His₆-tagged proteins were expressed in E. coli BL21(DE3). The cells were harvested by centrifugation, resuspended in 25 mM Tris-HCl, 300 mM NaCl, 10 mM imidazole (pH 8.0), and lysed by 40 µg/mL lysozyme, 6 µg/mL DNase-I, and sonication. The clarified lysates were loaded onto a 0.5 mL Ni-NTA column (Thermo Scientific). The His₆-tagged proteins were eluted by an increasing imidazole gradient.

The imidazole was removed by a PD10 desalting column (Cytiva). The protein preparations were treated with TEV protease for affinity tag removal. The His₆-tagged TEV protease and un-cleaved proteins were removed using a Ni-NTA column. The tag-free proteins were concentrated using the Amicon concentrator device (Merck) with 3 kDa molecular weight cut-off and loaded on size exclusion chromatography column (Superdex-75, GE Healthcare). The mouse ISG15-capped (immature) was made available at 545 µM concentration, and that ISG15-uncapped (mature) at 1453 µM. The protein preps were aliquoted and stored at –80°C until use. The recombinant mouse ISG15 protein preparations were analyzed in SDS-PAGE and the by mass-spectrometry to control purity and identity (Supplementary Figure 4). The sequence coverage of the tryptic peptides from the preparation was 98% for the capped construct and 94,8% for the uncapped product, strongly indicating proper biological functions.

All in vitro assays were performed at least twice, each time using biological triplicates in each assay. Differences in transcript and protein levels and % infected cells were analyzed by unpaired Student’s t-test considering unequal variances (Welch’s test) or by one-way ANOVA using the Welch’s adjustment when comparing three or more groups. When more than two parameters (i.e infection and treatments) were analyzed a two-way ANOVA was applied. Statistical analyses were performed in Prism (GraphPad, La Jolla, CA).

Type I IFN responses, including several regulators of ISGylation, are increased in TB patients as well as in individuals who progress to active disease (4). In addition to being strongly induced by IFN-I, ISG15 has also been shown to be upregulated by viral and bacterial infections, LPS and retinoic acid (29–31). Therefore, we asked whether M. tuberculosis infection influences ISGylation in bone marrow-derived macrophages (BMM). Notably, upon infection, BMM showed a clear upregulation of the expression genes involved in the ISGylation pathway including isg15, usp18 and ube2l6, (Figures 1A–C). Similarly, analysis of previously published transcriptomic datasets revealed increased expression of ISGylation-related genes in M. tuberculosis-infected human monocyte derived macrophages, and alveolar macrophages, as well as bone marrow-derived macrophages (Supplementary Figure 1).

The effect of IFN-I supplementation on ISGylation responses of M. tuberculosis-infected BMM was evaluated. Stimulation with IFN-β increased the expression of the ISGylation-related transcripts in uninfected BMM and further increased their levels in M. tuberculosis-infected cells. Transcripts encoding for il1b, inos and tnf, immune mediators that contribute to intracellular M. tuberculosis control, were increased in infected BMM (Figures 1D–F). The addition of IFN-β further enhanced the levels of these transcripts in infected, but not in uninfected, BMM.

In contrast to the stimulatory effects of IFN-β on ISGylation-related transcripts, stimulation of infected BMM with IFN-γ or IL-1β cytokines that improve control of M. tuberculosis infection (2, 23, 32, 33), did not further increase isg15, usp18 and ube2l6 mRNA levels beyond those induced by the infection alone (Figures 1G–J, L-N). Both IFN-γ and IL-1β increased levels of inos and il1b mRNA expression (Figures 1K, O, P).

Stimulation of BMM with various concentrations of IFN-β resulted in higher intracellular levels of M. tuberculosis, as measured by an increased frequency of infected BMM and a greater bacteria density per infected BMM (Figures 2A, B).

ISG15 is synthesized as 161 aminoacid-long precursor that is rapidly processed into its mature 155 aminoacid form by proteolytic cleavage, thereby exposing the conjugation site at its C-terminus (34). The immature (capped) ISG15 its not conjugated to substrates but may upon secretion, act extracellularly in a cytokine-like manner (12, 21). We produced both mature and capped ISG15 and tested whether these could influence M. tuberculosis control in BMM. The intracellular bacterial levels in BMM treated with different concentrations of mature or immature ISG15 were similar (Figures 2C, D).

Next, we investigated whether USP18 modulates macrophage responses to M. tuberculosis infection. For this purpose, two USP18 inhibitors, BB7 and 2K04, were used. These compounds were selected from a small-molecule compound library through a deISGylase assay-based high-throughput screening (2K04) and further optimized through medicinal chemistry (BB7) (28). BMM were treated with BB7 or 2K04 along with a structurally related inactive control compound (NC), before and during M. tuberculosis infection. Treatment of infected BMM with either 1 or 10 μM of BB7 or 2K04 but not with NC resulted in increased ISGylation (Figures 2E, F). Both inhibitors increased ISGylation in uninfected BMM (Figures 2G, H). M. tuberculosis infection resulted in increased ISGylation in BMM relative to uninfected controls (Figures 2F, H). In some experiments incubation with NC showed slightly increased ISGylation levels, although to a lesser extent than BB7 or 2K04 (Figures 2F, H). Incubation with the inhibitors did not reduce BMM viability (Supplementary Figure 2A), consistent with previous observations in cell lines at higher concentrations (28).

We then tested whether USP18 inhibition by BB7 and 2K04 affected the control of M. tuberculosis infection in BMM. Cells treated with either BB7 or 2K04 showed improved control of intracellular M. tuberculosis when compared to BMM NC- treated controls, as indicated by the reduced frequency of infected BMM and the lower bacterial density per cell (Figures 3A, B). Coincubation of M. tuberculosis in axenic cultures with 10 μM USP18 inhibitors did not reduce the bacterial growth (Supplementary Figure 2B). Treatment of M. tuberculosis-infected BMM with either BB7 or 2K04 increased the levels of the IFN-I regulated usp18, isg15 and cxcl10 mRNA (Figures 3C–E), as well as il1b and inos (Figures 3F, G), which are transcriptionally regulated by NF-κB. The transcriptional activity of NF-κB is tightly repressed by binding to the IκBα inhibitor. Upon stimulation, IκBα is phosphorylated, leading to its the degradation and the release of NF-κB to activate gene transcription (35). Infection of BMM with M. tuberculosis increased IκBα phosphorylation, and treatment with the USP18 inhibitors further enhanced phospho-IκBα levels in infected BMM, demonstrating strong NF-κB activation by these inhibitors (Figures 3H, I).

We next compared the transcriptomic profiles of M. tuberculosis-infected BMM treated with BB7 or NC. A differential clustering among uninfected, M. tuberculosis (Mtb) and M. tuberculosis-BB7-treated BMM was revealed by PCA analysis and Pearson’s correlation, highlighting the variation in gene expression between the diverse groups (Figure 4A, Supplementary Figure 3A). We identified 710 differentially expressed genes (DEGs) between uninfected and Mtb BMM and 243 DEGs comparing Mtb-BB7 vs Mtb BMM, 129 of these DEGs were upregulated (Figures 4B, C, Supplementary Figure 3B).

Gene ontology (GO) analysis revealed enrichment of pathways related to antiviral responses, innate immune activation and inflammasome signaling when comparing Mtb-BB7 vs Mtb BMM (Figure 4D). Conversely, genes associated with stress responses, hypoxia and antioxidant activity were downregulated in Mtb-BB7 (Figure 4E). A detailed analysis of individual transcripts within these pathways is shown in Figures 4F–I and Supplementary Figures 3C, D. ISGylation-related genes usp18, ube2l6 and isg15 were detected among transcripts of GO “responses to virus” that were upregulated by BB7 treatment. Additionally, the genes inos, igtp, gbp2 and gbp5 which have been shown to contribute to the intracellular control of M. tuberculosis in macrophages (2, 36–38) were also increased in BB7-treated BMM (Supplementary Figure 3C). In contrast, expression of arg1, socs3, and nr4a1 mRNA, which might impair anti-M. tuberculosis responses by macrophages was downregulated following BB7 treatment (Supplementary Figure 3D).

Because the inflammasome pathway was upregulated in BB7-treated, M. tuberculosis-infected BMM, mature IL-1β levels in culture supernatants were measured as a readout of inflammasome activation. IL-1β levels were stimulated by the M. tuberculosis infection but not by BB7 treatment alone. However, the addition of BB7 to M. tuberculosis-infected BMM increased IL-1β levels as compared to untreated controls, indicating that the USP18 inhibitor enhances the inflammasome pathway during infection (Figure 4J).

Silencing of USP18 by siRNA transfection, reduced the accumulation of usp18 mRNA in M. tuberculosis-infected BMM (Figure 5A). Conversely, levels of the ISGylation related transcript isg15 and ube2l6, as well as the IFN-I regulated genes ifit1 and cxcl10, were increased following usp18 silencing (Figures 5B–E). Similarly, usp18-silenced infected BMM displayed increased il1b and inos mRNA levels (Figures 5F, G), recapitulating the effect of pharmacological USP18 inhibition.

Then, we investigated whether IFN-I signaling contributes to the role of USP18 in M. tuberculosis infection of BMM. Silencing ifnar1, which encodes the IFN-I receptor α chain, reduced the accumulation of ifnar1 mRNA (Figure 5H) and decreased the expression of IFN-I regulated genes usp18, isg15, ube2l6 and ifit1 as compared with control siRNA-transfected BMM (Figures 5I–L), indicating that the IFN-I signal was hampered by the ifnar1 siRNA. Levels of il1b and inos mRNA were increased in ifnar1 siRNA transfected BMM as compared to those in BMM transfected with control siRNA (Figures 5M, N).

Next, we examinate whether ifn1ar silencing hindered the ability of USP18 inhibitors to control M. tuberculosis. Treatment with 2K04 reduced both the proportion of infected BMM and bacterial density per infected cell compared with controls (Figures 5O, P). Blocking IFN-I signaling through ifna1r siRNAs similarly decreased the intracellular bacterial load, and combining 2K04 treatment with ifna1r silencing did not alter the enhanced M. tuberculosis control mediated USP18 inhibition (Figures 5O, P). These results indicated that the improved control of the intracellular bacteria after USP18 inhibition is independent of IFN-I signaling.

The effect of IFN-β supplementation in combination with USP18 inhibition in the responses of M. tuberculosis-infected BMM was then evaluated. The levels of isg15 and usp18 mRNA in M. tuberculosis-infected BMM increased after the stimulation with IFN-β. These effects were further amplified upon IFN-β and 2K04 co-treatment (Figures 6A, B). Levels of il1b and inos transcripts were also higher in IFN-β-treated M. tuberculosis-infected BMM exposed to 2K04 than in NC-treated controls (Figures 6C, D). However, IFN-β stimulation partially impaired the improved bacterial control conferred by the USP18 inhibitor (Figures 6E, F). Thus, high levels of IFN-I combined increased the susceptibility to M. tuberculosis in BMM with impaired USP18 regulation.

Finally, we assessed whether stimulation with the TLR3 agonist polyIC, a potent inducer of IFN-I expression, would mimic the effects of IFN-β addition on USP18 inhibitor-treated BMM. PolyIC induced the accumulation isg15, usp18, ifit1 and cxcl10 transcripts in both uninfected and M. tuberculosis-infected BMM (Figures 6G–K). PolyIC also increased the levels of il1b and inos mRNA in both uninfected and infected BMM (Figures 6K, L), while IFN-β-induced the accumulation of i1b and inos mRNA in infected but not in uninfected BMM (Figures 1D–F). Moreover, the addition of polyIC reduced the intracellular density of M. tuberculosis as compared to untreated controls (Figures 6M, N). Combined treatment with both polyIC and 2K04 further decreased the intracellular bacterial levels in BMM relative to either treatment alone (Figures 6M, N).