SARS-CoV-2 ORF8 Mediates Signals in Macrophages and Monocytes through MyD88 Independently of the IL-17 Receptor

This article is featured in Top Reads, p. 167

Severe acute respiratory syndrome coronavirus 2 surfaced as a novel virus responsible for the ongoing COVID-19 pandemic, infecting >750 million individuals to date according to the World Health Organization. Although the immune responses that govern disease outcome have not been completely elucidated, severe infection is often associated with a systemic “cytokine storm,” characterized by hyperactivation of immune cells and overproduction of proinflammatory cytokines (1).

SARS-CoV-2 open reading frame 8 (ORF8) encodes a secreted multifunctional protein that is highly expressed in severe disease, and patients infected with ORF8-deleted viral mutants exhibit milder symptoms (2, 3). ORF8 has been described to play multiple roles in the course of infection, including immune modulation by regulation of MHC expression, acting as a histone mimic, and antagonizing a type I IFN response (3). Two recent studies reported unexpectedly that ORF8 activates downstream cytokine responses in murine macrophages and human monocytes through the IL-17 receptor, and thus it was concluded that ORF8 is a viral “mimic” of IL-17 (IL-17A) (4, 5). This observation carries important implications, both for understanding the pathogenesis of coronavirus-induced pulmonary pathology and for strategies that may be employed to mitigate pathogenic inflammation such as biologic intervention (6).

IL-17 is the founding member of a distinct subclass of cytokines and protects against an assortment of microbes, especially extracellular bacteria and fungi (7). Production of IL-17A and its closest homolog, IL-17F, is limited to type 17 cells, most famously CD4⁺ Th17 cells, but also subsets of CD8⁺ cells, γδ-T cells, some NK cells, and group 3 innate lymphoid cells (8). Whereas IL-17 is made by lymphocytes, most studies have demonstrated that IL-17–responsive cells are nonhematopoietic, typically epithelial and mesenchymal cell types. This conclusion has been drawn from diverse approaches and model systems, including cultured primary and immortalized cells, bone marrow chimeras, and conditional IL-17R–knockout mouse lines (9–14). Sensitivity to IL-17 appears to be rate limited by restricted expression of the IL-17RC subunit (15–19). In contrast, the IL-17RA subunit is widely expressed in both the hematopoietic and nonhematopoietic compartments and is a coreceptor for several members of the IL-17 ligand family, including IL-17C, IL-25 (IL-17E), and IL-17F (15, 16, 19–21). Although IL-17 can direct myeloid cell recruitment to tissue during inflammation, this is generally believed to be mediated indirectly through the induction of chemokines and cytokines in epithelial or mesenchymal cell types (7).

Given our longstanding interest in IL-17 signaling mechanisms, we were intrigued by the suggestion raised in these and other reports that myeloid cells potentially have the capacity to respond to IL-17, and moreover that SARS-CoV-2 ORF8 may exploit the IL-17R pathway to promote pathologic inflammation in the lung (4, 5). In this study, we reproduce findings that ORF8 potently induces inflammatory cytokines in myeloid cells, including murine macrophages (RAW 264.7 cells, primary bone marrow–derived macrophages [BMDMs]) and in primary human monocytes. However, in our hands, IL-17 did not detectably induce these events in macrophage/monocyte cell types. We show that IL-17RA but not IL-17RC is expressed in macrophage/monocyte cells isolated from SARS-CoV-2–infected lungs isolated from humans with severe COVID-19 disease or from infected K18-hACE2 mice. Knockout of IL-17RA or Ab blockade of IL-17RA and IL-17RC did not impair ORF8 signaling. Rather, ORF8 relies on the TLR/IL-1 family adaptor MyD88, and these observed effects are not due to contaminating endotoxin. Thus, ORF8-induced monocyte and macrophage responses appear to occur independently of the IL-17R pathway.

Wild-type (WT), Myd88^−/− (no. 009088), Il1r1^−/− (no. 003245), and K18-hACE2 (no. 034860) mice (on the C57BL/6 background) were from The Jackson Laboratory (Bar Harbor, ME) or generated in-house. Il17ra^−/− mice were from Amgen. Experiments were performed on males (6–10 wk of age) under Institutional Animal Care and Use Committee protocols approved by the University of Pittsburgh.

BMDMs were generated by 6-d culture of femoral bone marrow in DMEM (10% FBS, 1% penicillin/streptomycin, 2 mM l-glutamine, 20 mM HEPES, 60 ng/ml M-CSF) and replated 1 d before stimulation. RAW 264.7 cells (American Type Culture Collection) were grown in IMDM (10% FBS [Life Technologies], sodium pyruvate, MEM nonessential amino acids, 1% penicillin/streptomycin). ST2 cells were cultured in MEM (10–12% FBS, 1% penicillin/streptomycin). Human monocytes were isolated with an EasySep human monocyte isolation kit (STEMCELL Technologies) and cultured in X-VIVO 15 media (Lonza). Human immortalized keratinocytes (KCs; N/TERT-2G [22]) were grown in KC SFM (serum-free medium; Thermo Fisher Scientific) with 30 µg/ml bovine pituitary extract, 0.2 ng/ml epidermal growth factor, and 0.3 mM CaCl₂. Murine or human IL-17A (PeproTech) was used at 50–200 ng/ml, Escherichia coli LPS (Sigma-Aldrich) at 1 μg/ml, and His-Tag ORF8 (Thermo Scientific) at 1 μg/ml. Rat anti–IL-17RA, rat IgG2A, goat anti–IL-17RC, goat IgG (Bio-Techne, 5 µg/ml), or polymyxin B (Sigma-Aldrich, 30 µg/ml) were applied 1 h prior to cytokine stimulation.

RNA was isolated with RNeasy mini kits (Qiagen), cDNA was synthesized with an iScript cDNA synthesis kit, and quantitative real-time PCR (qPCR) was performed with SYBR Green Supermix on a CFX Opus 96 (Bio-Rad). Primers were from QuantiTect (Qiagen).

Raw fastq data were from public datasets of lungs of 4-d SARS-CoV-2–infected K18-hACE2 mice (accession no. GSE175996, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE175996) (23) and six severe COVID-19 patients at Shenzhen Third People’s Hospital (accession no. GSE145926, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE145926) (24). Raw 3′ single-cell RNA sequencing (scRNA-seq) data were processed with Cell Ranger v6.1.2 (10x Genomics). Transcripts were aligned to a customized reference genome in which ORF8 (MN985325.1, https://www.ncbi.nlm.nih.gov/nuccore/MN985325.1/) was added to human and mouse reference genomes GRCh38 and mm10-2020. An entry summarizing the ORF8 gene was appended to GRCh38 and mm10-2020 annotation gtf files and the genome was indexed using cellranger_mkref. Loupe Cell Browser (10x Genomics) was used to define clusters. Barcodes were filtered to exclude high counts of unique molecular identifiers, representing likely doublets. A threshold unique molecular identifier count of 50,000 and 60,000 per barcode on a linear scale was used for humans and mice, respectively. Seurat suite v4.0.6 was used for quality control and analysis.

Cell suspensions were blocked with Fc receptor Abs (BioLegend) in 5% goat serum (Jackson ImmunoResearch) and stained with CD11c-BV421, CD11b-allophycocyanin/Cy7 (BioLegend) and CD45-FITC (Invitrogen). Data were collected on an Aurora (Cytek) and analyzed with FlowJo.

Datasets were tested for normality using the Shapiro–Wilk normality test. Data were analyzed by 1- or 2-way ANOVA with Bonferroni’s multiple comparisons test (parametric data) or Kruskal-Wallis tests with Dunn’s multiple comparisons test (nonparametric data) in GraphPad Prism. A p value <0.05 was considered significant, and values are expressed as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

A paradigm in the IL-17 field is that IL-17–mediated signaling is limited to nonhematopoietic cells. Given reports that ORF8 may bind to and activate the IL-17R complex in macrophages and monocytes (4, 5), we felt compelled to re-examine the capacity of IL-17 to induce inflammatory signals in myeloid cells. Accordingly, we treated BMDMs from C57BL/6 WT mice and RAW 264.7 macrophage cells with LPS (1 μg/ml) or recombinant murine IL-17 (200 ng/ml) for 3 and 6 h (Fig. 1). Cytokine genes ascribed to IL-17 and/or ORF8 signaling were assessed by qPCR (Il6, Ilb, Cxcl1, Cxcl2) normalized to Gapdh. These early time points were used based on prior experience showing that IL-17–sensitive cells (e.g., fibroblasts, epithelial cells, lymph node stromal cells) reliably upregulate target genes within this time frame (25–29). IL-17 did not detectably upregulate these cytokine transcripts in BMDMs, whereas LPS potently induced their expression (Fig. 1A). Similarly, LPS but not IL-17 enhanced target gene expression in RAW 264.7 cells (Fig. 1B).

Cell line contamination (e.g., HeLa cells) is a widespread but often unacknowledged problem in cell culture studies (30). Therefore, we authenticated cells used in the current study by staining for the hematopoietic marker CD45 and monocyte and macrophage markers CD11b and CD11c. ST2 stromal cells, an IL-17–responsive line routinely used to evaluate IL-17 signaling (29), served as a control (Fig. 1C). The high expression levels of CD45, CD11b, and CD11c on BMDMs and RAW 264.7 cells used in these studies were consistent with bona fide macrophage identity.

Because ORF8 was suggested to serve as an IL-17R agonist (4, 5, 31), we next attempted to reproduce findings that ORF8 upregulates expression of cytokine genes in BMDMs, RAW 264.7 cells, and primary human monocytes isolated from PBMCs of healthy donors. Cells were treated with recombinant ORF8 (1 μg/ml), IL-17 (200 ng/ml), or LPS (1 μg/ml) for 3 h. ORF8 induced inflammatory transcripts in all cell types tested at a magnitude similar to LPS, whereas IL-17 did not enhance gene expression (Fig. 2A–C). Accordingly, in keeping with most prior observations, macrophages/monocytes are sensitive to LPS but do not detectably respond to IL-17.

Unlike myeloid cells, KCs respond robustly to IL-17 (28, 32). We reasoned that if ORF8 can indeed activate the IL-17R, KCs would respond to ORF8 in a manner similar to IL-17. Accordingly, we treated immortalized human KCs (22) with IL-17 or ORF8 for 3 h. As expected, KCs showed elevated levels of IL6, IL1B, CXCL1, and CXCL2 in response to IL-17, but ORF8 did not significantly activate these genes (Fig. 2D), further arguing against an IL-17R–mediated mechanism of action for this viral protein.

Although the predominance of literature suggests that IL-17RC expression is limited to hematopoietic cells, there are reported exceptions (18, 33–36). To visualize IL-17RC and IL-17RA in the setting of SARS-CoV-2 infection, we interrogated public scRNA-seq datasets from lung of K18-hACE2 transgenic mice (37) infected with SARS-CoV-2 or in humans with severe COVID-19 disease (23). In mice, Il17ra was ubiquitously expressed, whereas Il17rc mRNA was highly expressed in fibroblasts but undetectable in monocyte/macrophage clusters (Fig. 3A). Likewise, in human bronchoalveolar lavage fluid from patients with severe COVID-19, IL17RA was found in both fibroblasts and monocyte/macrophages whereas IL17RC was not observed in the myeloid compartment (Fig. 3B). ORF8 mRNA was detectable in both cell types in mice and humans, confirming that samples were virally infected (Fig. 3A, 3B). Hence, given the absence of IL-17RC, ORF8 is unlikely to be able to activate the IL-17R pathway in monocytes/macrophages.

To determine whether the IL-17R is required for ORF8-driven responses, Il17ra^−/− or WT BMDMs were treated with IL-17 and ORF8 and cytokines were evaluated by qPCR. The response to ORF8 was fully preserved in Il17ra^−/− BMDMs but markedly impaired in Myd88^−/− cells (Fig. 4A). We also evaluated the capacity of blocking Abs against IL-17RA or IL-17RC to inhibit ORF8 signaling. In ST2 stromal cells, these Abs efficiently blocked IL-17 induction of Il6, a canonical IL-17–induced target gene (Supplemental Fig. 1A). However, anti–IL-17RA/C Abs did not diminish ORF8-induced signals in RAW 264.7 macrophages (Fig. 4B). Taken together, these data support a model in which ORF8 signals independently of IL-17RA and IL-17RC.

In the same BMDM experiments, we evaluated the ability of ORF8 to signal in the absence of MyD88, given reported structural similarities of ORF8 to IL-1 (38). Notably, MyD88 is dispensable for IL-17 signaling (27). In contrast to both WT and Il17ra^−/− BMDMs, the response to ORF8 was markedly impaired in Myd88^−/− cells. As expected, LPS responses were largely mitigated in the absence of MyD88 but not IL-17RA (Fig. 4A). The modest signals in Myd88^−/− BMDMs upon LPS treatment are likely mediated through TRIF (Toll/IL-1R domain-containing adapter inducing IFN-β), a MyD88-independent adaptor activated by TLR4 but also not required for IL-17 signaling (27).

Given that the IL-1 receptor is upstream of MyD88 as well as the predicted structural similarities between ORF8 and IL-1 (37), we evaluated the role of IL-1 signaling in ORF8-induced inflammatory responses. To this end, Il1r1^−/− or WT BMDMs were treated with IL-17 and ORF8, and cytokines were evaluated by qPCR. The response to ORF8 was preserved in the absence of IL-1R (Fig. 5), indicating that ORF8-induced responses are not reliant on the IL-1R in monocytes/macrophages.

Based on similarities between ORF8 and LPS signaling responses, we questioned whether residual LPS might contaminate the commercial ORF8 preparation (made in E. coli) and whether trace levels of endotoxin capable of eliciting signaling might account for cytokine induction induced by this protein (39), especially given the exquisite sensitivity of monocytes/macrophages to bacterial LPS. To rule this out, we pretreated RAW 264.7 cells with polymyxin B, which inactivates endotoxin and inhibits LPS-induced MyD88 activation (40), and then treated cells with LPS or ORF8. As shown, ORF8 induced Il6, Il1b, Cxcl1, or Cxcl2 regardless of polymyxin B, but polymyxin B effectively blocked LPS signaling (Fig. 6). Thus, the activities mediated by ORF8 are unlikely to be a result of endotoxin contamination.

Many viruses encode cytokine mimics, particularly in the IFN and TNF pathways (41). In fact, mammalian IL-17 was initially recognized based on homology to an ORF in Herpesvirus saimiri. The advantage to a virus in encoding an IL-17 agonist is unclear, although a few studies implicate IL-17 in antiviral immunity (25, 42, 43). The potential for SARS-CoV-2 ORF8 to be a viral agonist of the IL-17R was intriguing.

We were motivated to pursue the present analyses because our extensive experience with IL-17 signaling has shown that its functions are limited to mesenchymal and epithelial cell types (13, 14), and not to myeloid cells as described in the studies of ORF8 and the IL-17R. Only minimal studies during the past few decades have observed IL-17 to act on monocytes or macrophages, and in some cases alternative explanations for such phenomena, such as HeLa contamination or indirect effects, were not ruled out rigorously. Also, often in these settings only investigation of IL-17RA was performed, which is a shared subunit for other cytokines in the IL-17 ligand family (44). Given the global significance of the COVID-19 pandemic, we thought it was important to explore this issue further.

As noted, most reports using bone marrow chimeras or conditional IL-17R deletions have not observed roles for IL-17 signaling in the hematopoietic system, although some studies credibly describe IL-17 activity on immune cells. For example, in activated T cells, one of the initial studies of IL-17 showed it could mediate costimulation of T cell proliferation, in a similar manner to Herpesvirus saimiri gene 13 (HVS13), a viral homolog of this cytokine. IL-17 was shown to act in an autocrine manner in Th17 cells to limit pathogenicity (25, 33). In NK cells, it was reported that IL-17RA deficiency causes cells to be hyporesponsive to microbial stimuli (45), and in preosteoclasts (which are from a hematopoietic lineage), IL-17RC deficiency abolishes IL-17F–induced osteoclast differentiation (18).

Understanding regulation of inflammatory cytokines in the context of severe COVID-19 is clinically important, given the mortality arising from the cytokine storm. For example, blockade of IL-1 and IL-6 improves disease outcomes in severe infection settings (46–49). A recent study revealed an intricate relationship between airway epithelial cells and myeloid cells, in which bystander myeloid cells sense extracellular genomic DNA and mitochondrial DNA released by SARS-CoV-2–infected epithelial cells (50) and trigger inflammasome signaling. The resulting release of mature IL-1β from cocultured epithelial cells and PBMCs in turn promotes IL-6 release, amplifying inflammation (50). IL-17 is also thought to participate in the COVID cytokine storm (51). In the lung, cellular targets of IL-17 signaling are airway epithelial cells (13); in this regard, it is not known whether ORF8 can act directly on lung epithelial cell types, although in human KCs we saw no activation of a cytokine gene panel in response to ORF8 treatment.

In this study, we reproduce observations that ORF8 induces inflammatory cytokines in monocytes and macrophages (4, 5). Although IL-17 effectively activated gene expression in stromal cells and immortalized KCs, as expected, we saw no IL-17–induced cytokine induction in BMDMs, RAW 264.7 cells or primary human monocytes, in keeping with another recent report (52). The cytokine gene targets tested were selected based on prior publications, but we did not do more extensive transcriptomic analyses that might have uncovered other potential target genes. Although we are not certain why our findings differ from prior reports with respect to ORF8 versus IL-17, a widespread challenge is cell line misidentification, especially cross-contamination of HeLa cells (30). HeLa cells are cervical epithelial cells that are very responsive to IL-17 (53, 54). It will be valuable to determine whether discordant results at least in some cases could be explained by HeLa contaminants or other technical considerations.

ORF8 was concluded to bind to IL-17RA based on coimmunoprecipitation and proximity ligation assays (4, 5, 31), but there are important caveats in interpreting results based on the systems used (e.g., overexpression in HEK293 cells, NIH-3T3 fibroblasts). Coimmunoprecipitation and proximity ligation assay approaches do not distinguish well between protein–protein interactions occurring at the cell surface versus intracellularly, and excess IL-17RA upon transfection accumulates at high levels within the endoplasmic reticulum/Golgi, potentially resulting in nonspecific protein aggregation (55). No studies to our knowledge have directly demonstrated cell surface interactions of ORF8 with IL-17RA (or IL-17RC), which are needed to support a bona fide ligand–receptor relationship. A recent follow-up report indicates that ORF8 is secreted in both glycosylated and nonglycosylated forms, but only nonglycosylated ORF8 can coimmunoprecipitate with IL-17RA (56). However, our studies use E. coli–derived ORF8, which is not glycosylated, and we still observed no induction of NF-κB–dependent cytokines in human or murine monocytes/macrophages. Additionally, this new study (56) uses NF-κB activation as a readout of IL-17 activity, and only nonglycosylated ORF8 was found to activate this transcription factor. However, NF-κB was assessed at an inappropriately late time point (12 h), and more specific markers of IL-17 signaling activation such as Act1 phosphorylation or proximal NF-κB/IκBξ or C/EBPβ/δ induction (57) were not tested. Thus, although ORF8 is clearly secreted and capable of activating monocytes and macrophages, the role of the IL-17R in mediating these activities is open to question.

Canonical IL-17 signaling requires IL-17RA and IL-17RC (15, 16, 19), although other configurations of the receptor have been proposed, for example, IL-17RA/RD heterodimers and IL-17RC homodimers (58–60). Our studies with Il17ra^−/− BMDMs and IL-17RA/RC blocking Abs suggested that ORF8 responses are not mediated through the IL-17R complex, but instead they occur through a MyD88-dependent receptor. It is not evident that ORF8 has three-dimensional homology with IL-17. Rather, its structural properties bear resemblance to IL-1β and IL-1RA agonists (38), in keeping with its dependence on MyD88. Although our studies using Il1r1^−/− BMDMs suggest that this does not occur via IL-1R, the nature of which receptor engaged by ORF8 remains unknown. We speculate that this may be a member of the TLR family, but further studies are warranted in this regard.

In summary, we reproduce findings that the multifunctional SARS-CoV-2 ORF8-secreted protein provokes cytokine secretion in monocytes and macrophages, which could contribute to the deadly cytokine storm seen in severe manifestations of COVID-19. However, our data argue against ORF8 mediating effects through the IL-17R. We suggest that secreted ORF8 is more likely to act on TLR family receptors that signal through MyD88. Defining the function of ORF8 and its ability to modulate cytokine expression in key immune cells is important for probing effective ways to combat the consequences of SARS-CoV-2 severe disease.

The authors have no financial conflicts of interest.

We thank Amgen for Il17ra^−/− mice, and B.M. Coleman and S. Sarkar for valuable input.