Itaconate and derivatives reduce interferon responses and inflammation in influenza A virus infection

Itaconate has recently emerged as a metabolite with immunomodulatory properties. We evaluated effects of endogenous itaconate and exogenous itaconate, dimethyl-, and 4-octyl-itaconate on host responses to influenza A virus infection. Infection induced ACOD1 (the enzyme catalyzing itaconate synthesis) mRNA in monocytes and macrophages, which correlated with viral replication and was abrogated by itaconate treatment. Pulmonary inflammation and weight loss were greater in Acod1-/- than wild-type mice, and ectopic synthesis of itaconate in human epithelial cells reduced infection-induced inflammation. The compounds induced different recruitment programs in infected human macrophages, and transcriptome profiling revealed that they reversed infection-triggered interferon responses and modulated inflammation in cell lines, PBMC, and lung tissue. Single-cell RNA sequencing of PBMC revealed that infection induced ACOD1 exclusively in monocytes, whereas treatment silenced IFN-responses in monocytes, lymphocytes, and NK cells. Viral replication did not increase under treatment despite the dramatically repressed IFN responses, but 4-octyl itaconate inhibited viral transcription in PBMC. The results reveal dramatic reprogramming of host responses by itaconate and derivatives and their potential as adjunct treatments for hyperinflammation in viral infection.


Introduction
ACOD1 mRNA induction was seen in M2 cells and undifferentiated PBMC, and the weakest in 140 M1 cells and dTHP-1 cells (Fig. 2B-E), even though the latter two cell types did support a 141 significant degree of viral transcription ( Fig. 2A,D). As in IAV-infected mouse lung (Fig. 1), 142 TNFAIP3 and ACOD1 transcriptional correlated with each other (Fig. 2C). Infection of dTHP-1 143 cells with three IAV (H1N1) strains of differential replication efficiency (Petersen, Mostafa et al., 144 2018b) showed the strongest ACOD1 and TNFAIP3 induction by the two strains with the highest 145 and the weakest by the strain with the lowest replication efficiency (Fig. 2D,E). Thus, the extent 146 of ACOD1 induction after IAV infection differed significantly among the cell types tested, was 147 strongest in M2 primary cells, was accompanied in all cases by an increase in TNFAIP3 mRNA, 148 and correlated with viral replication. 150 RNA replication. We then tested whether these compounds could modulate cellular 151 inflammation due to IAV infection. In infected dTHP-1 cells, treatments with itaconate or DI did 152 not affect viral hemagglutinin (HA) mRNA expression, but both reduced induction of CXCL10, a 153 key pro-inflammatory chemokine during influenza virus infection, with DI being several-fold 154 more potent (Fig. 3B,C, Fig. S1). The 12 h time point was chosen because our previous work had  Fig. 3G). Strikingly, addition of both 162 itaconate and DI led to global down-regulation of IFN-regulated genes. This was accompanied by 163 upregulation of a substantial number of other transcripts, indicating that the compounds also 164 modulated other cellular processes (Fig. 3H,I). A principle component analysis (PCA) indicated 165 that effects of the compounds on the cells were so pronounced at these concentrations that treated 166 infected and treated uninfected cells clustered closely together, but far away from untreated 167 infected or uninfected cells (Fig. S2). Indeed, a hierarchical clustering analysis revealed an 168 across-the-board normalization of a clade of IFN-and inflammation-related transcripts by the compounds, but also clusters of DEGs that were upregulated by either or both compounds with 170 respect to both uninfected and infected cells, further indicating effects upon cell homeostasis in 171 general (Fig. S3). mRNA and protein expression may be differentially affected by IAV infection 172 due to virus-induced "host-cell shut-off". We therefore assessed release of inflammation-related 173 polypeptides in supernatants from the same experiment (Fig. 3J). This analysis confirmed the 174 downregulation of pro-inflammatory factors (e.g., IP-10, MCP-1) by itaconate and DI, but also 175 showed increased IL-8 (DI only) and minor increases of IL1B (both IA and DI) in uninfected 176 cells. Of note, both compounds also increased levels of the anti-inflammatory IL1 receptor 177 antagonist (IL1RA) in supernatants from both infected and uninfected cells. A comprehensive 178 view of the protein targets that were detected by this assay revealed that the two compounds 179 effected substantially different changes in the target cells, as several factors were induced even in 180 uninfected cells specifically by DI or itaconate (Fig. S4). Thus, itaconate and DI apparently act 181 on macrophages to release different recruitment programs that modify local inflammatory cell 182 populations. 183 Analysis of enriched GO terms revealed that IFN-I signaling was strongly induced in the infected 184 cells, which was prevented by treatment with itaconate and DI (Fig. 4). This analysis also 185 revealed that both compounds activated metabolic processes, whereas itaconate also modulated 186 processes relating to differentiation and membrane signaling. A KEGG pathway analysis 187 additionally revealed induction of classic pro-inflammatory pathways such as TNF signaling, 188 TLR receptor signaling, and chemokine signaling by infection, all of which were depleted by 189 itaconate and DI treatment (Fig. S5). Taken together, these results revealed strong anti- IFN and 190 anti-inflammatory effects of itaconate and DI, but also suggested that they exert other important 191 effects on cell metabolism and differentiation, some of which differ between the compounds.  (Fig. 5A), which was accompanied by appearance of substantial concentrations of IA (Fig. 5B). While there was no effect on viral HA mRNA expression, 200 transfection with the ACOD1-expressing vector led to a marked reduction in infection-associated 201 CXCL10 induction (Fig. 5C,D). A preliminary treatment experiment showed that (as opposed to 202 THP-1 cells) DI was cytotoxic at 1.0 mM in A549 cells, and the 0.5 mM concentration was 203 therefore used. Treatment with DI resulted in a modest reduction of HA expression early in 204 infection and a marked reduction of CXCL10, IL6, and IFNB expression (Fig. 5E-H). Exogenous 205 itaconate, too, reduced CXCL10 mRNA, but less efficiently (Fig. S6). Taken together, these 206 results obtained with A549 cells suggest that immunomodulatory effects of itaconate and DI are 207 independent of the ability of the cell type to express endogenous ACOD1, and that similar effects 208 on IFN-directed gene expression can be obtained with ectopic, endogenous synthesis of itaconate 209 and exogenously added itaconate and DI. 210 Next, we assessed effects of itaconate and DI on cellular transcriptomes in IAV-infected A549 211 cells. PCA revealed the expected normalization of gene expression due to DI treatment, but also 212 increasing effects of the compounds on overall cell homeostasis (Fig. S7). As in THP-1 cells, 213 hierarchical clustering analysis revealed clear clustering of each of the four groups and the close 214 relationship of the uninfected and DI-treated infected cells (Fig. S8). However, the anti-215 inflammatory effect of itaconate was weaker than that of DI, and at the higher (40 mM) 216 concentration, the itaconate signature became so dominant that the itaconate-treated infected 217 group now formed its own clade (Fig. S9). IAV infection enriched GO terms related to IFN-I and 218 -II responses as well as other antiviral and pro-inflammatory terms, which could all be reduced to 219 near baseline by treatment with DI or itaconate (Fig. 6). Visualization of differentially expressed 220 IFN-related mRNA also showed that, in spite of its overall lower effect on pro-inflammatory 221 signaling than DI, itaconate treatment at the lower (20 mM) concentration did have a 222 considerable impact on IFN-I expression (Fig 6B). In addition, itaconate uniquely inhibited 223 NFKB signaling and fatty acid metabolism. KEGG-pathway analysis additionally revealed 224 depletion of other pro-inflammatory pathways, including TNF signaling and necroptosis, by the 225 treatments (Fig. S10).  (Fig. S11), whereas ACOD1 mRNA levels did 231 not change significantly. Adding itaconate and DI reduced expression of CXCL10 mRNA in 232 tissue and IP10 protein in culture supernatant, and itaconate additionally reduced ISG15 mRNA 233 in tissue, whereas viral titers in the supernatant were unaffected (Fig. 7). Re-analysis of a 234 published dataset from a model using a different IAV strain (H3N2) and histologically normal  inflammatory response both at the mRNA and protein level (Fig. 8A,B). The induction of 246 ACOD1 mRNA expression in peripheral blood in IAV infection was also corroborated by 247 reanalysis of a published dataset from whole blood from patients with moderate and severe 248 influenza (Fig. S13). The effects of itaconate were mixed in that there was no significant change 249 in ACOD1, IFNB1, and TNF mRNA expression in cells or of IL1B protein in supernatants. In 250 contrast, DI and 4OI led to a marked reduction of IFN and pro-inflammatory cytokine expression 251 at the mRNA and protein level, as well as of ACOD1 mRNA. As exemplified by the effect of DI 252 and 4OI on TNF and ACOD1 expression, the treatments reduced expression of some mRNA 253 targets even in uninfected cells, suggesting that they also reduced some baseline inflammation.

254
Of note, 4OI in addition led to a pronounced (95%) reduction in HA transcription (p=<0.0001), 255 indicating inhibition of viral RNA replication. PCA of bulk PBMC transcriptomes revealed 256 strong effects of the compounds on cell transcriptomes in general, which led to one outlying DI-257 treated sample and two itaconate-treated samples, whereas all four 4OI-treated samples formed a 258 clearly discernible group (Fig. S14). The most significant transcriptome changes were driven by 4OI, which had major effects on many genes that were not regulated by IAV infection or 260 itaconate or DI treatments (Fig. S15). Nonetheless, induction of HMOX1 by DI in IAV-infected 261 samples was evident. Functional enrichment analysis of GO terms based on the transcriptomes 262 revealed that infection induced the expected antiviral responses, including response to IFN-I, as 263 well as several other inflammation-related terms (Fig. 8C). At least one of the three compounds 264 prevented activation of each of the infection-associated terms. However, consistent with its lack 265 of effect on IFNB expression (Fig. 8), itaconate did not affect response to IFN-I or influenza A. 266 Several terms that were depleted by the compounds in infected PBMC were not enriched by 267 infection alone, suggesting that the compounds diminished also a baseline activation of the 268 PBMC. Interestingly, 4OI treatment strongly enriched terms relating to chromatin conformation, 269 which turned out to be driven by upregulation of histone deacetylase epression. A KEGG 270 pathway analysis essentially confirmed the findings of the GO term analysis (Fig. S16) (Fig. 9A,B). Monocytes constituted the least numerous cell type, and DI treatment of 282 control and infected cells led to a further reduction (Fig. S17). It was not possible to discern 283 whether this was due to decreased survival or a technical artefact. Monocytes were essentially the 284 only cell type expressing viral RNA, as only negligible amounts of viral transcripts were detected 285 in lymphocytes and NK cells. DI treatment appeared to result in a slight increase in viral RNA 286 expression in monocytes (Fig. 9C). IAV infection upregulated expression of ACOD1 exclusively, 287 and CXCL10 predominantly, in monocytes, whereas TNFAIP3, IFIT1, and ISG15 were 288 upregulated also in the NK, T and B cell compartments (Fig. 9D). DI treatment markedly reduced expression of ACOD1, IFIT1, and CXCL10, but -interestinglyit reduced TNFAIP3 and ISG15 290 (albeit weakly) expression only in monocytes. Consistent with the observation that they 291 constituted the main PBMC host cell, IAV infection triggered the most vigorous transcriptomic 292 host response in monocytes, which was driven by the expected IFN responses (Fig. 9E). 293 However, discernable differential expression (mostly of IFN-related RNAs) was also observed in 294 the other cell types. CD8+ T cells are shown as a representative example in Fig. 9F, the other cell preferential effect on monocytes became apparent (Fig. 9G,H). 302 GO term enrichment analysis of the monocyte scRNAseq data confirmed the major enrichment of 303 IFN-related processes by infection and their depletion by DI treatment, which had been seen in 304 all models in this study. In addition, infection broadly dampened RNA metabolism, ribosome 305 assembly, and protein synthesis and export, which was consistent with the cytostatic effects of 306 type I IFN. DI treatment reversed all these effects and also stimulated some in uninfected cells 307 (Fig. 10A). A KEGG enrichment analysis of these monocyte data confirmed the depletion of 308 IFN-related pathways (influenza, measles, cytosolic DNA sensing pathway) by DI treatment, but 309 also of other pro-inflammatory pathways (TLR signaling, cytokine-cytokine receptor interaction, 310 NF-kB signaling) in infected and uninfected cells (Fig. S19). Of note, DI treatment apparently 311 stimulated ribosome function in both infected and control monocytes.

312
The Venn diagrams in Fig. 10B illustrate that monocytes are the cell type in which the most 313 biological processes were activated by infection and depleted by DI treatment, but reduction of 314 antiviral and pro-inflammatory responses was a common theme in the other cell types as well 315 (Fig. 10C). Indeed, GO term analysis identified four major IFN-related processes as a central 316 functional network that was upregulated by infection and downregulated by DI treatment during 317 infection in all five cell types (Fig. 10B). A marked stimulation of protein synthesis and export 318 by DI was also apparent in the uninfected lymphocytes and NK cells. Finally, in order to identify a common transcriptomic network we performed GO and KEGG pathway enrichment analyses of 320 the 81 transcripts that were commonly differentially expressed in IAV infection and regulated in cell activation by DI (Fig. S20).

328
In this first dedicated study on the impact of itaconate and derivatives on host responses to IAV 329 infection, we found that (i) the endogenous ACOD1/itaconate axis limits pulmonary   effects, but also broad effects on processes relating to, e.g., chromatin structure, the potential 354 effects of which on cell homeostasis and cell-pathogen interactions will require further study.   Infection-independent effects on target cells. By using relatively high doses of itaconate and DI 385 and assessing effects of the compounds on uninfected cells or their effects on processes that were 386 not affected by infection alone, our study also provides a first comprehensive look at global 387 effects of the compounds on target cells. In particular, itaconate and 4OI exerted pronounced, 388 broad effects on gene expression in A549 cells and PBMCs, respectively. Clearly, it will be of 389 great interest to investigate how these effects relate to differences in toxicities on different cell 390 types, or to differences in their effects on virus-host interactions.  applications. Considering the more reliable results obtained with DI as compared to unmodified 420 itaconate, the lower required concentrations of DI and 4OI, but also the striking anti-viral effects 421 of 4OI, it appears that chemical variants of itaconate, rather than its native form, will take the 422 lead in further translational development to the bedside.           Software (ThermoFisher Scientific). Array QC and data normalization was performed using the where all the cDNA generated from an individual cell share a common 10x Barcode. In order to 565 identify the PCR duplicates, Unique Molecular Identifier (UMI) was also added. The GEMs were incubated with enzymes to produce full length cDNA, which was then amplified by PCR to 567 generate enough quantity for library construction. Quality was checked using the Agilent 568 Bioanalyzer High Sensitivity Assay.

569
Library construction and quality control. The cDNA libraries were constructed using the 10x    a tendency towards increased TNFAIP3 expression are seen. As opposed to the strong induction of both genes in the mouse model (Fig. 1A), HMOX1 and HMOX2 expression is unchanged. 839 *p<0.05; **p<0.01; ***p<0.001 (pairwise t-tests with pooled SD).