The HIF transcription network exerts innate antiviral activity in neurons and limits brain inflammation

Summary

Pattern recognition receptors (PRRs) induce host defense but can also induce exacerbated inflammatory responses. This raises the question of whether other mechanisms are also involved in early host defense. Using transcriptome analysis of disrupted transcripts in herpes simplex virus (HSV)-infected cells, we find that HSV infection disrupts the hypoxia-inducible factor (HIF) transcription network in neurons and epithelial cells. Importantly, HIF activation leads to control of HSV replication. Mechanistically, HIF activation induces autophagy, which is essential for antiviral activity. HSV-2 infection in vivo leads to hypoxia in CNS neurons, and mice with neuron-specific HIF1/2α deficiency exhibit elevated viral load and augmented PRR signaling and inflammatory gene expression in the CNS after HSV-2 infection. Data from human stem cell-derived neuron and microglia cultures show that HIF also exerts antiviral and inflammation-restricting activity in human CNS cells. Collectively, the HIF transcription factor system senses virus-induced hypoxic stress to induce cell-intrinsic antiviral responses and limit inflammation.

Graphical abstract

Highlights

• •HSV-1 and -2 disrupt the hypoxia-inducible factor (HIF) network in permissive cells
• •HIF activation induces autophagy, which exerts anti-HSV activity in neurons
• •Neuronal HIF activation regulates infection and inflammation in the infected brain

Using transcriptome analysis of disrupted transcripts in herpes simplex virus-infected cells, Farahani et al. identify the hypoxia-inducible factor gene network to possess antiviral activity through induction of autophagy. This contributes to antiviral defense and regulation of inflammation during infection in the CNS.

Introduction

Viral infections can cause severe diseases, and efficient immune responses are required to control the invading pathogens. The innate immune system exerts the first line of defense and also primes activation of the adaptive response.^1,2 In the absence of rapid and efficient innate immune responses, the subsequent immune activities mediated by sustained innate mechanisms and adaptive immune responses may amplify inflammation and promote disease.³ The innate immune response to viral infections is believed to be mainly driven by mechanisms activated downstream of pattern recognition receptors (PRRs), which sense viral molecules, primarily nucleic acids, to induce host defense and inflammation.⁴ This includes the DNA-activated cyclic-GMP-AMP synthetase (cGAS)- stimulator of interferon genes (STING) pathway, which signals through the kinase TBK1 and is important for host defense against many viruses, such as herpes simplex virus types 1 and 2 (HSV-1 and -2).^5,6,7 The best-described PRR-driven antiviral program is the type I interferon (IFN) system of cytokines acting in auto- and paracrine manners to induce expression of IFN-stimulated genes (ISGs) with antiviral and immune-stimulatory functions.⁸ However, viruses efficiently evade and suppress immune responses to allow for the establishment and maintenance of infection.⁹ For instance, most viruses, including influenza A virus, HSV-1, and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), inhibit type I IFN expression and function.^10,11,12,13

In recent years, it has emerged that in addition to the potent type I IFN response, a number of other effector mechanisms also contribute to innate antiviral defense.¹⁴ Many of these are constitutive and stress-induced immune mechanisms activated by specific viral molecules or processes induced by viral replication. For instance, several infections lead to activation of autophagy,^15,16,17,18 and it was recently reported that loss of function in autophagy genes is associated with susceptibility to infections with HSV-2.¹⁶ In addition, numerous restriction factors, such as the dNTP triphosphohydrolase SAMHD1, exert early control of virus infections.^19,20 These antiviral constitutive and stress-induced immune mechanisms act often independently of PRRs and the IFN system,¹⁴ thus providing an immediate layer of host defense not associated with strong inflammation.

Hypoxia-inducible factor 1 (HIF1) is a transcription factor complex activated in response to hypoxic conditions.²¹ The active complex is composed of an α subunit, of which there are three, HIF1α–3α, and a β subunit. The HIF1α protein is rapidly turned over under normoxic conditions, driven by prolyl hydroxylases, which trigger pVHL-mediated ubiquitination and proteasomal degradation of hydroxylated HIF1α. Under hypoxic conditions, hydroxylation of HIF1α is prevented, allowing dimerization with HIF1β and nuclear translocation of the HIF complex, thus promoting a transcriptional program leading to glycolysis, mitophagy, and pro-survival activities.²¹ HIF signaling has been extensively studied in immunity and inflammation, particularly the impact of hypoxia and metabolic shifts on the activity of immune cell populations.^22,23,24 In relation to infection immunology, HIF1α signaling has been demonstrated to contribute to protection against several bacterial infections, mechanistically involving phagocytosis, granule proteases, and antimicrobial peptides.^25,26,27,28 By contrast, the role of the HIF pathway in antiviral defense is still not well understood.^29,30

HSV-1 and -2 are alphaherpesviruses, which productively infect epithelial cells and neurons.³¹ In addition, HSV can activate immune cells, and essential antiviral activities are actively modulated by the viruses.^11,32,33 HSV infections can give rise to a number of severe diseases, including herpes simplex encephalitis, HSV-2 meningitis, and recurrent genital and oral herpes.^34,35 A requirement for development of these diseases is productive viral replication in permissive cells and viral modulation of host immune responses.² Transcriptomics represents a powerful tool to uncover response patterns and mechanisms in infectious diseases. Recent work has revealed that HSV infection not only modulates an abundance of specific gene transcripts but also causes disruption of transcriptional termination.³⁶ Thus, in-depth analysis of functional and disrupted cellular transcripts in virus-infected cells may lead to identification of novel host mechanisms of importance for antiviral defense.

In this work, we have analyzed the pattern of viral disruption of host transcripts as an approach to identify novel antiviral immune mechanisms. This analysis revealed that HSV infection in permissive neurons and epithelial cells lead to disruption of the HIF transcription network. Use of different approaches to stabilize HIF1α in vitro, as well as HSV-2 infection of neuron-specific Hif1/2a-deficient mice, demonstrated that this pathway has broad antiviral activity and is essential for protection. Mechanistically, HIF exerts antiviral activity by activation of autophagy in productively infected cells. Moreover, HIF-mediated antiviral defense reduced not only viral load but also PRR signaling and inflammatory cytokine expression in the brain, thereby limiting the potentially pathological inflammatory response. Thus, the HIF system activates antiviral activity in productively infected cells and controls viral replication in a manner that limits the inflammatory response, hence sparing the brain from unnecessary immunopathology.

Results

HSV infections disrupt transcription in neuron-like cells and primary human neurons

HSV productively infects epithelial cells and neurons but also interacts with immune cells. We aimed to examine how HSV infection in cell lines originating from these three cell types affected the transcriptome, thus aiming to identify novel viral modes of host immune modulation (Figure 1A; Data S1–S5). For this purpose, we used the neuroblastoma cell line SH-Sy5y, the keratinocyte cell line HaCaT, and the monocyte leukemia cell line THP-1, which reflected primary cells relatively well with respect to permissiveness to HSV-1 and -2 infection (Figures S1A and S1B). The cell lines were infected with HSV-1 and -2 for different time intervals, and RNA was isolated and sequenced (Figures S1C–S1L). In the SH-Sy5y cells, the fraction of reads mapping to the virus genome increased over time, illustrating efficient viral replication (Figure S2A). To quantify host shutoff in SH-Sy5y cells, transcripts for each gene were mapped to exon and exon-exon junctions and expressed as the ratio to the maximum transcripts per million value for each gene. This showed that HSV-1 or -2 infection of SH-Sy5y cells led to extensive reduction in the levels of transcripts, with proper composition at the later stage of infection (Figure S2B).

Figure 1.

HSV-2 disrupts HIF transcription networks in neuron-like cells and primary human neurons

(A) Graphical illustration of the rationale behind the project and experimental setup. Viral disturbance of cellular pathways at the transcriptional level has been used to identify antiviral mechanisms. Image was generated with BioRender.

(B) Number of up-regulated (red) and down-regulated/disrupted (dark and light blue) genes in SH-Sy5y cells infected with HSV-2 (MOI 1) after satisfying the threshold of false discovery rate (FDR) p value ≤0.05 and |log₂(fold change)| ≥1.5 as well as in-ratio fold change <3 and read through-ratio fold-change <5 compared to mock.

(C) Pathway enrichment analysis of genes shown in (B). Pathways significantly enriched (p < 0.05) in at least 3 databases in EnrichR at 4, 12, and 24 h are shown. Enrichment at 12 and 24 h for the top altered pathways was similar, and hence they were grouped together.

(D) Heatmap showing read-in, readthrough, and intron/exon transcripts per million (TPM) ratio to the maximum TPM in IFN-related genes relative to mock in HSV-2 infection in SH-Sy5y cells over the different time points.

(E) Heatmap showing read-in, readthrough, and intron/exon TPM ratio to the maximum TPM in HIF target genes relative to mock over the different time points in HSV-2 infection in SH-Sy5y cells.

(F and G) Heatmap showing read-in and readthrough TPM ratio to the maximum TPM in IFN-related genes and HIF target genes relative to mock after 24 h in HSV-2 infection in hESC-derived neurons.

(H) Graphical illustration of the experimental design to examine for transcription termination disruption by RT-PCR. Image was generated with BioRender.

(I and J) Total RNA from SH-Sy5y cells infected for 12 h with HSV-2 333 (MOI 1) or clinical isolates of HSV-1 and -2 were subjected to RT using either poly(dT) or gene-specific primers, followed by PCR. Data are presented as ratio of PCR product from the RT priming with poly(dT) versus gene-specific primer.

For fold change calculations in (B), Wald statistical test was used in order to test the differences between all test samples versus mock. FDR p value was determined using Benjamini-Hochberg correction for multiple testing. The data in (I) and (J) are shown as means ± SD. n = 5 biological replicates. Data are representative of at least three independent experiments.

See also Figures S1–S3 and Data S1–S6.

When analyzing transcriptionally induced genes, we observed an increase in intronic reads, which was detectable from 4 h post-infection (hpi). Similarly, a dramatic increase in reads mapped to intergenic regions was observed, with the strongest effect in HSV-2-infected cells (Figure S2C). To test whether this was due to disruption of transcription termination and readthrough transcription as reported,³⁶ we quantified the number of read-in and readthrough for each gene. The identified read-in and readthrough, as well as the intron-exon ratio, increased over time for both HSV-1 and -2 (Figure S2D). These data confirm that transcripts not undergoing appropriate termination accumulated in the HSV-infected cells.

HSV infections in neuron-like cells suppress HIF transcription networks

In order to explore whether the disrupted transcripts were enriched for specific cellular functions, we calculated the fold change of the read-in and readthrough ratios in infected versus mock-treated cells. Conservatively, genes with a false discover rate p value ≤0.05 and |(log₂ fold change)| ≥1.5 satisfying (1) in-ratio fold change <3 and (2) readthrough ratio fold change <5 were considered as undisrupted. Based on scatterplots of the in-ratio and out-ratio values at 4 hpi, thresholds were chosen such that most genes remained below these values (Figure S3A). Next, we performed a pathway enrichment analysis based on up-regulated, non-disrupted genes as well as the combined list of transcriptionally disrupted and down-regulated, non-disrupted genes. Interestingly, antiviral sensing and IFN signaling pathways were identified to be highly enriched among the disrupted genes in both HSV-2 and -1 infection (Figures 1B, 1C, and S3B). To further corroborate these findings, we observed that a large fraction of the ISGs were transcriptionally disrupted upon HSV infection in SH-Sy5y cells (Figures 1D and S3C). In agreement with the transcriptome data, we observed reduced IFNβ-induced phosphorylation of STAT1 in HSV-infected cells (Figures S3D and S3E).

Besides IFN signaling pathways, we noted that the HIF1 transcription network was suppressed in SH-Sy5y cells upon HSV-1 and -2 infection (Figures 1B, 1C, and S3B). To explore this observation further, we investigated HIF target genes in more details (Data S6). This analysis revealed that a very large fraction of the HIF-inducible genes, including HIF1A itself, were transcriptionally disrupted at later times post-infection (Figures 1E and S3F–S3H) and to an extent far exceeding the disruption of classical housekeeping genes (Figures S3I and S3J). To test whether the observations in SH-Sy5y cells recapitulate phenomena in human neurons, we generated human embryonic stem cell (hESC)-derived neurons. These cells were infected with HSV-1 and -2, and we performed transcriptome analysis. Consistent with published data from human fibroblasts³⁶ and brain organoids,³⁷ transcription termination was also disrupted in ESC-derived neurons and resulted in transcriptional shutoff. Importantly, we found that expression of IFN signaling pathways as well as HIF target genes was disrupted upon HSV infection (Figures 1F–1G, S3K, and S3L). The effect was consistently more profound upon HSV-2 infection compared to HSV-1. To corroborate the transcriptome-based findings with data from an alternative approach, we performed RT-qPCR using either poly(dT) or gene-specific RT primers (Figure 1H). In agreement with the transcriptome data, we observed that the ratio of PCR products from the poly(dT) versus gene-specific primed RT decreased dramatically in the HSV-2-infected cells (Figure 1I). Using RT-qPCR, we also confirmed the early up-regulation of a subset of HIF-regulated genes (Figure S3M), as suggested by the pathway enrichment analysis (Figure 1C). Finally, the disruption of transcripts from HIF-regulated genes was observed when testing clinical isolates of HSV-1 and -2 (Figure 1J), thus supporting the physiological relevance of the finding. Collectively, these data demonstrate that HSV infection in the neuronal cells disrupts the HIF transcription network.

HIF activation inhibits HSV-1 and -2 replication

In order to explore the function of the HIF pathway in the context of HSV infection, we subjected the cells to hypoxia, which potently induced stabilization and nuclear translocation of HIF1α (Figures 2A and 2B). Importantly, these conditions led to significant reduction in the accumulation of HSV transcripts (Figures 2C and 2D). Likewise, activation of the HIF pathway through depletion of VHL expression with CRISPR-Cas9 technology also reduced HSV-2 gene expression (Figures 2E and 2F). Third, we treated cells with a panel of small-molecule HIF activators. For instance, the HIF1 activator ML228 led to stabilization, nuclear translocation, and induction of HIF-induced genes (Figures 2G, 2H, S4A, and S4B) and did not lead to overt toxic effects on the cells at the concentrations used (Figure S4C). Treatment with ML228 potently blocked replication of both HSV-1 and -2, measured by either virus yield or transcription of the late viral gene gB (Figures 2I, 2J, S4D, and S4E). The ability to block HSV-1 and -2 replication was also observed for other HIF1α activators, including CoCl₂, dimethyloxallyl glycine, and diethyl-succinate (Figures S4F–S4M). Additionally, we confirmed the inhibition of HSV-2 replication upon treatment with the HIF1α activators MK-8617, which is in preclinical testing, and vadadustat, which has been clinically approved (Figures S4N–S4P). The observed antiviral activities were indeed dependent on the HIF pathway, as depletion of HIF1α using CRISPR-Cas9 technology severely compromised the antiviral actions of hypoxic conditions or treatment with HIF activator (Figures 2K–2N).

Figure 2.

HIF activation restricts HSV-1 and -2 replication

(A) Illustration of experimental setup for infection (HSV-1, MOI 0.3; HSV-2, MOI 0.1) under hypoxic conditions (1% O₂) of SH-Sy5y cells.

(B) Nuclear translocation of HIF1α upon normoxia and 1% O₂ in SH-Sy5y cells, measured by immunoblotting.

(C and D) HSV-1 and -2 gB mRNA transcripts measured by RT-qPCR following infection during normoxia and hypoxia (1% O₂); gB levels were normalized against 18S rRNA. Normalized ratios (NRs) of infected cells are shown relative to uninfected cells. n = 6 biological replicates.

(E) Nuclear translocation of HIF1α in cells treated with VHL and AAVS1 single guide RNA (sgRNA)/Cas9 ribonucleoproteins (RNPs) following ML228 treatment. ML228 treatment was used as a positive control relative to untreated condition.

(F) HSV-2 gB mRNA transcripts measured by RT-qPCR; gB expression in VHL and AAVS1 gRNA-treated cells were normalized against 18S rRNA. NRs of VHL relative to AAVS1 are shown. n = 4 biological replicates.

(G) Nuclear translocation of HIF1α following ML228 treatment (0.5 μM, 24 h) in SH-Sy5y cells, evaluated by immunoblotting.

(H) Illustration of experimental setup for infection (HSV-1, MOI 0.3; HSV-2, MOI 0.1) and ML228 treatment (0.5 μM) of SH-Sy5y cells.

(I and J) HSV gB mRNA levels measured by RT-qPCR. Data were normalized against 18S rRNA, and NRs are shown. n = 5 biological replicates.

(K) HIF1A was depleted using sgRNA/Cas9 RNPs. Immunoblotting of HIF1α in cells upon 1% O₂ condition.

(L) HSV-2 titers in supernatants from ML228-treated cells treated with sgRNA/Cas9 RNPs targeting AAVS1 or HIF1A. n = 4 biological replicates.

(M and N) HSV-1 and -2 gB transcript levels were measured by RT-qPCR in HIF1A-depleted cells subjected to normoxic or hypoxic (1% O₂) conditions or treated with ML228 versus DMSO. Data were normalized against 18S rRNA, and NRs are shown. n = 4–5 biological replicates.

(O) Analysis of HSV-2 transcripts (MOI 1, 24 h) in SH-Sy5y cells treated with ML228 and DMSO as control. Volcano plot is presented as –log₁₀(p values) versus log₂(fold change). Red dots: adjusted p value < 0.05. Vertical lines indicate ±1 log₂(fold change). The names of the 20 most significant hits are indicated. The data are shown as means ± SD. Individual values are represented by dots.

Data are representative of at least three independent experiments. All the statistical analyses were performed by Student’s t test, two-tailed, parametric distribution to identify statistical significance.

See also Figure S4 and Data S7 and S8.

When comparing the antiviral activity induced by treatment with ML228 versus IFNα, we found that the HIF activator blocked viral replication with a potency comparable to IFNα treatment, at least with the concentrations used in our experiments (Figures S4Q–S4U). The antiviral activity of ML228 was due to a global blockage of viral transcripts (Figures 2O and S4V; Data S7 and S8), suggesting an antiviral mechanism acting upstream or at the level of viral gene transcription. In agreement with this, ML228 treatment reduced the accumulation of HSV-1 capsids from infecting virions at the nuclear membrane of the neuron-like cells (Figure S4W). Collectively, these data demonstrate that the HIF pathway has antiviral activity against HSV-1 and -2 in neuroblastoma SH-Sy5y cells.

HSV-1 infection disrupts the HIF transcription network in epithelial cells

To investigate whether the identified transcriptional termination disruption and inhibition of the HIF network was also observed in other cell types, we analyzed the transcriptomes from infected HaCaT cells (human epithelioid cells) (Figures S1G and S1H) and PMA-differentiated THP-1 cells (macrophage-like) (Figures S1I–S1L). In HaCaT cells, the fraction of reads mapping to the virus genome increased over time, illustrating efficient viral replication in this cell line (Figure S5A), and we observed pronounced transcriptional shutoff, as seen by the increasing fraction of reads mapping to intronic and intergenic regions (Figures S5B–S5D). This demonstrated transcriptional termination disruption in the HSV-infected HaCaT cells. Pathway enrichment analysis revealed that the HIF transcription network was down-regulated in the infected cells (Figures 3A, 3B, and S5E). Activation of the HIF pathway by either hypoxia or treatment with ML228 inhibited replication and HSV gene expression of both HSV-1 and -2 (Figures 3C–3H).

Figure 3.

HSV-1 infection disrupts the HIF transcription network in epithelial cells

(A) Number of up-regulated (red) and down-regulated/disrupted (dark and light blue) genes in HaCaT cells infected with HSV-1 (MOI 3). The criteria for disruption was FDR p value ≤0.05, |log₂(fold change)| ≥1.5, read-in-ratio fold change <3, and read-through-ratio fold change <5 compared to mock.

(B) Pathway enrichment analysis of genes shown in (A). Pathways significantly enriched (p < 0.05) in at least 3 databases in EnrichR at 4, 12, and 24 h are shown.

(C) HSV titers evaluated by plaque assays in supernatants from HaCaT cells infected with HSV-1 (MOI 0.3) or -2 (MOI 0.1) and cultures under normoxic or hypoxic conditions (1% O₂) for 24 h. n = 4 biological replicates.

(D) Nuclear translocation of HIF1α following ML228 treatment (0.5 μM, 24 h) in HaCaT cells, evaluated by immunoblotting.

(E–H) HSV-1 and -2 replication measured by plaque assay on supernatants and gB RT-qPCR on total RNA. PCR data are shown as NRs against 18S rRNA. n = 4 biological replicates.

In (A), for fold change calculations, the Wald statistical test was used. FDR p value was determined using Benjamini-Hochberg correction for multiple testing. Illustrations in (A) were generated with BioRender. In panels (C) and (E)–(H), data are representative of at least three independent experiments. The data are shown as means ± SD from at least three biological replicates per group. Individual values are represented by dots. Data are representative of at least three independent experiments. All the statistical analyses were performed by Student’s t test, two-tailed, parametric distribution to identify statistical significance.

See also Figures S5 and S6.

When analyzing transcriptome data from THP-1 cells, which were rather non-permissive to HSV replication (Figures S1A, S1B, and S6A), very limited transcriptional shutdown occurred (Figure S6B). Similarly, the fraction of reads mapping to the intronic and intergenic regions as well as read-in and readthrough ratios did not increase at the later stages of infection relative to mock (Figures S6C and S6D). When subjected to pathway enrichment analysis, IFN signaling pathways were identified to be highly enriched among the up-regulated genes in PMA-differentiated THP-1-derived macrophages infected with HSV-1 and -2 (Figures S6E–S6H). In contrast to what was seen in SH-Sy5y and HaCaT cells, no selective reduction in HIF-induced genes was observed (Figures S6I and S6J). Collectively, in the cell types tested, permissiveness to HSV replication enables disruption of the HIF transcription network, which possesses antiviral activity.

HIF activation exerts broad antiviral activity in neuron-like cells

To examine whether the antiviral effect of HIF activation in neurons was limited to HSV-1 and -2, we examined a number of other viruses with the ability to cause central nervous system (CNS) diseases in humans and animals. The viruses tested differ greatly with respect to genome type and mechanisms of replication, including modes of entry, sub-cellular sites of replication, and mechanisms of virion assembly and egress. Similar to HSV, replication of vesicular stomatitis virus, encephalomyocarditis virus, SARS-CoV-2, and influenza A virus were inhibited significantly upon HIF activation using ML228 (Figures 4A–4D). By contrast, measles virus replication was not restricted (Figure 4E). Collectively, these data demonstrate a broad antiviral activity of the HIF pathway against viruses that can cause disease in the CNS.

Figure 4.

HIF activation exerts broad antiviral activity in neuron-like cells

(A and B) Virus titer of vesicular stomatitis virus (VSV) and encephalomyocarditis virus (EMCV) in culture supernatants from SH-Sy5y cells treated with ML228 (0.5 μM) and infected for 19 h with EMCV (MOI 0.1) or VSV (MOI 0.001). n = 3 biological replicates.

(C) SARS-CoV-2 N2 RNA levels evaluated by RT-qPCR in total RNA from SH-Sy5y cells treated with ML228 (0.5 μM) and infected for 48 h with SARS-CoV-2 (MOI 1). Data were normalized against 18S rRNA. n = 5 biological replicates.

(D) Influenza A virus M2 RNA levels evaluated by RT-qPCR in total RNA from SH-Sy5y cells treated with ML228 (0.5 μM) and infected for 24 h with influenza A virus (MOI 0.1). Data were normalized against 18S rRNA. n = 5 biological replicates.

(E) Measles virus titer in culture supernatants from SH-Sy5y cells treated with ML228 (0.5 μM) and infected for 36 h with Measles virus (MOI 0.1). n = 6 biological replicates.

Data are representative of at least three independent experiments. The data are shown as means ± SD. Individual values are represented by dots. For (A), Mann-Whitney was applied to identify statistical significance. For (B)–(E), the statistical analyses were performed by Student’s t test, two-tailed, parametric distribution to identify statistical significance. ns, not significant. Illustrations of viruses were generated with BioRender.

The antiviral activity of the HIF pathway is dependent on autophagy

To start addressing the mechanism involved in HIF-driven antiviral activity in neuronal cells, we first mapped the transcriptome analysis in ML228-treated SH-Sy5y cells and compared cells pretreated with sgRNAs against HIF1A versus AAVS1. As expected, the differential expression analysis showed that HIF target genes were the top differentially expressed genes in ML228-treated control versus HIF1A-depleted cells (Figure 5A; Data S9). In addition, pathway enrichment analysis confirmed that HIF transcriptional activity was the top up-regulated pathway (Figure 5B). We specifically checked a list of HIF target genes induced upon ML228 treatment in control cells and found their expression significantly attenuated in HIF1A-depleted cells (Figure 5C). Interestingly, HIF agonist treatment did not induce expression of type I IFN genes or inflammatory cytokines such as CXCL10, CXCL8, and interleukin-6 (Figures S7A–S7C). Accordingly, depletion of the type I IFN receptor chain 2 did not compromise the antiviral activity of ML228 (Figures S7D–S7F). Moreover, TLR3 has been reported to be important for antiviral activity in neurons,³⁸ but disruption of HIF1A did not alter the ability of cells to respond to the double-stranded mimic, and TLR3 agonist, poly(I:C) (Figure S7G).

Figure 5.

HIF activation mediates restriction of HSV replication through autophagy

(A) Transcriptome analysis of ML228-stimulated SH-Sy5y cells (0.5 μM, 8 h) treated with HIF1A-/AAVS1-targeting gRNA Cas9/sgRNA RNPs. Volcano plots are presented as –log10(p values) versus log2(fold change). Red dots: adjusted p value < 0.05. Vertical lines indicate ±1 log2(fold change). The names of the 20 most significant hits are indicated.

(B) Pathway enrichment using genes satisfying the applied cutoffs. Up-regulated (in red) and down-regulated gene (in blue) pathways, shown in at least 3 databases in EnrichR.

(C–E) Differential expression of the indicated classes of genes in AAVS1- and HIF1A-depleted SH-Sy5y cells treated with ML228 (0.5 μM, 8 h): heatmaps showing (C) HIF target genes, (D) glycolysis genes, and (E) autophagy genes.

(F) Immunoblots for BNIP3, LC3B, and vinculin (VCL) in lysates harvested at different time points after treatment with ML228 (0.5 μM).

(G) LC3, P62, and DAPI staining of fixed SH-Sy5y cells treated with ML228 for 24 h and visualized by confocal microscopy. Scale bar, 20 μm.

(H) Quantification of LC3 foci per cell. n = 30 cells per group.

(I) Immunoblot for LC3B and VCN with lysates treated with ML228 for 4 h following pretreatment with 2-deoxy-D-glucose (2-DG; 5 mM) for 2 h.

(J) Immunoblots for BNIP3, LC3B, HIF1α, and VCL in lysates harvested after 24 h of incubation at 1% O₂.

(K) ATG5 depletion by CRISPR-Cas9 technology. AAVS1 was used as control. Lysates were immunoblotted for ATG5 and VCL.

(L and M) ATG5-depleted and control cells were cultured at normoxic or hypoxic (1% O₂) conditions for 24 h before infection with HSV-1 (MOI 0.3) and HSV-2 (MOI 0.1). Cultures were washed after 1 h, and total RNA harvested 20 h post-infection for determination of levels of HSV-1 and -2 gB transcripts by RT-qPCR. gB expression was normalized against 18S rRNA. n = 5–6 biological replicates.

The data in (H), (L), and (M) are shown as means ± SD. Individual values are represented by dots. Data are representative of three independent experiments. All the statistical analyses were performed by Student’s t test, two-tailed, parametric distribution to identify statistical significance.

See also Figure S7 and Data S9.

In order to understand the underlying antiviral mechanism of the HIF pathway, we investigated the genes induced upon HIF activation. In particular, we noted elevated expression of the autophagy-related genes BNIP3L and BNIP3³⁹ as well as the number of glycolytic enzymes (Figure 5A). There is accumulating evidence showing that HIF activation promotes autophagy through a process driven by induction of autophagy-related genes and glycolysis.^40,41,42 Autophagy, in turn, has been reported to exert potent antiviral activity, particularly in the CNS.^15,16,18,43 When investigating glycolysis- and autophagy-related genes more globally, we observed many genes in these pathways to be dependent of HIF1A expression in the SH-Sy5y cells (Figures 5D and 5E). The induction of autophagy-related proteins was confirmed by immunoblotting for BNIP3 (Figure 5F). This was accompanied by HIF-dependent activation of autophagy as measured by conversion of LC3B-I to LC3B-II, and formation of LC3 II and P62 punctae, as markers of autophagosome formation (Figures 5F–5H). Blockage of glycolysis by 2-deoxy-D-glucose reduced activation of autophagy (Figure 5I). BNIP3 expression and LC3B-I-to-LC3B-II conversion were also induced upon hypoxia (Figure 5J). To examine whether the HIF-induced autophagy response was mediating the antiviral activity, we inhibited autophagy in SH-Sy5y cells by disrupting the ATG5 gene (Figure 5K). Importantly, the antiviral activity against HSV-1 and -2 evoked by hypoxia, and which was dependent on HIF1A (Figures 2M and 2N), was significantly compromised in ATG5-deficient cells (Figures 5L and 5M). Likewise, the antiviral activity of ML228 was significantly impaired in ATG5-deficient cells (Figure S7H). Collectively, these data suggest that the HIF pathway induces antiviral activity in neuroblastoma cells through the induction of autophagy.

HIF deficiency in neurons renders mice susceptible to HSV infection in the CNS

To examine whether the observed antiviral activity of the HIF pathway in neurons was indeed of physiological relevance, we used a murine cornea HSV-2 infection model, leading to brain infection through the neuronal route (Figure 6A). First, we stained brain sections from mice infected with HSV-2 with pimonidazole, which detects hypoxia. Importantly, in the HSV-2-infected brains, we detected punctae pimonidazole stainings, primarily in cells with neuron soma morphology (Figure 6B). Next, we tested mice with neuron-specific deletion of Hif1a and Hif2a genes (nHif1a/Hif2a^ΔΔ/ΔΔ). Neuron-specific Hif1a/Hif2a knockout mice were generated by crossing animals harboring two floxed alleles of Hif1a and Hif2a genes with transgenic mice expressing Cre recombinase under the control of the calcium/calmodulin-dependent protein kinase IIa (Camk2a) gene promoter.⁴⁴ The Camk2a-gene-promotor-driven Cre/loxP recombination was shown to be restricted to post-mitotic neurons,⁴⁵ resembling the expression pattern of the endogenous Cam2ka gene (Figure S8A). When nHif1a/Hif2a^ΔΔ/ΔΔ mice were infected with HSV-2 in the cornea, expression of the HIF-induced genes Eno1, Bnip3, and Vegfa were reduced in the brains but not in the spleen (Figures 6C and S8B–S8F), and activation of autophagy was impaired as measured by LC3 turnover (Figure S8G). Importantly, the nHif1a/Hif2a^ΔΔ/ΔΔ mice exhibited significantly elevated weight loss, development of disease symptoms, and reduced survival after infection (Figures 6D–6F). This was accompanied by elevated viral load in the nHif1a/Hif2a-deficient brain stem (Figure 6G). Likewise HIF1α/2α-deficient neurons failed to promote hypoxia-induced antiviral activity in vitro (Figure S8H). Higher viral load could lead to elevated activation of PRRs such as the cGAS-STING pathway, which induces phosphorylation of TBK1 at serine 172 to activate downstream inflammation. Accordingly, in the HSV-2-infected nHif1a/Hif2a-deficient brain stems, we found elevated levels of phosphorylated TBK1 (Figure 6H). In addition, the levels of inflammatory cytokine transcripts in the brain stem were higher following infection in nHif1a/Hif2a^ΔΔ/ΔΔ mice (Figures 6I, 6J, and S8I). This was confirmed at the protein level for the chemokine CXCL10 (Figure 6K). Collectively, these data demonstrate an essential role for the HIF pathway in neurons in early protection against HSV-2 infection and in control of PRR-driven inflammation in the brain.

Figure 6.

HIF deficiency in neurons renders mice susceptible to HSV-2 infection in the CNS

(A) Illustration of experimental setup for HSV-2 infection of mice. Image was generated with BioRender.

(B) C57BL/6 mice were infected in the cornea with 400 PFU/mL HSV-2. After 5 days, mice were injected with pimonidazole 2 h before they were sacrificed, and brain stem sections were monitored for staining. Scale bars: 100 μm (top row) and 20 μm (bottom row).

(C) nHif1a/Hif2a^ff/ff and nHif1a/Hif2a^ΔΔ/ΔΔ mice were infected in the cornea with 400 PFU/mL HSV-2. Brain stems were harvested 5 days later for measurement of Eno1 mRNA by RT-qPCR. Data were normalized to 18S rRNA and are shown as means ±SD. n = 4–5 mice per group.

(D–F) Weight change, clinical scores, and survival were monitored measured from the day of infection. n = 9 mice per group.

(G–K) Brain stems were isolated 5 days post-infection, and homogenates were analyzed for (G) levels of infectious HSV-2 by plaque assay, (H) phosphorylation of TBK1 at S172 by immunoblotting (lysates were pooled in groups, 4 mice per group), (I and J) Cxcl10 and Tnfa mRNA levels by RT-qPCR, and (K) CXCL10 protein by ELISA.

For data in (D), (I), and (J), n = 20 mice per group. For data in (K), n = 20 mice per group. Data are shown as means ±SD. p values were calculated using two-way ANOVA with Bonferroni’s post hoc test.

See also Figure S8.

HIF activation exerts antiviral activity in hESC-derived neurons

Finally, we wanted to explore whether HIF activity also blocks HSV replication in hESC-derived neurons (Figure 7A) and its impact on inflammatory activity, e.g., in microglia. First, we observed that human neurons do not express STING or evoke IFN-inducing signaling in response to HSV-1 infection (Figure 7B), hence suggesting cell-autonomous antiviral mechanisms in neurons to be independent of the cGAS-STING pathway and PRR signaling. As described above, HSV-1 and -2 infections in ESC-derived neurons disrupted HIF-induced transcripts, including BNIP3 and ENO1 (Figures 1G and S3K). Consistent with the observations in SH-Sy5y cells, activation of HIF1α potently blocked HSV-1 and -2 replication in hESC-derived neurons, as measured by the accumulation of viral transcripts (Figures 7C, 7D, S9A, and S9B) and viral protein (Figure 7E), and by virus production (Figures S9C and S9D). This was accompanied by the induction of the canonical HIF-induced genes ENO1 and BNIP3 (Figures S9E and S9F) and conversion of LC3I to LC3II (Figure 7E). To test the impact of the hypoxic antiviral program on the activation of brain inflammatory responses, we transferred supernatants from HSV-2-infected hESC-derived neuron cultures under normoxic and hypoxic conditions to hESC-derived microglia and evaluated the responses (Figure 7F). The supernatants from the infected hypoxic neurons contained about ten times lower levels of virus than the supernatants from infected normoxic neurons (Figure 7G). Importantly, microglia treated with supernatants from infected hypoxic neurons exhibited less phosphorylation of STING and TBK1 (Figure 7H) and reduced induction of inflammatory gene expression (Figures 7I, 7J, and S9G). Collectively, these data suggest that HIF activation induces antiviral defense in neurons by activation of autophagy, thus limiting viral load and subsequent inflammatory responses in brain immune cells.

Figure 7.

Activation of the HIF pathway induces antiviral activity in human neurons

(A) Image of hESC-derived neurons. Scale bar, 100 μm.

(B) Neurons were infected with HSV-1 (MOI 1) for the 24 and 48 h, and lysates were isolated and immunoblotted for the indicate antigens. We used PMA-differentiated THP1 macrophage-like cells stimulated with cGAMP (4 μg/mL) for 2 h as positive control for the STING pathway. VCN, vinculin.

(C and D) HSV transcripts (gB) were measured by RT-qPCR analysis of total RNA from hESC-derived neurons isolated 48 h after infection with HSV-1 and -2 (MOI 1) in the presence and absence of treatment with ML228 (1.4 μM). Transcript levels were normalized against 18S rRNA and are shown as NRs. n = 2 biological replicates.

(E) Levels of HSV-1 and -2 VP5 protein, LC3B, and VCL in hESC-derived neurons 48 h after infection with HSV-1 and -2 (MOI 1) cultures in the presence or absence of treatment with ML228 (1.4 μM).

(F) Illustration of experimental setup for testing of effect of neuronal hypoxia on microglia activation. Image was generated with BioRender.

(G) HSV-2 gB mRNA transcripts measured by RT-qPCR following infection in ESC-derived neurons during normoxia and hypoxia (1% O₂). n = 5 biological replicates.

(H) Immunoblot for signaling proteins in microglia lysates treated for 24 h with supernatants (1:5) from neurons infected for 48 h (MOI 1) with HSV-2 under normoxic or hypoxic (1% O₂) conditions.

(I and J) Total RNA was isolated from microglia treated as described in (H) and examined for levels of CXCL10 and TNFA. n = 4 biological replicates. Transcript levels were normalized against 18S rRNA and shown as NRs.

The data in (C), (D), (G), (I), and (J) are shown as means ± SD. Individual values are represented by dots. For (C) and (D), Student’s t test, two-tailed, parametric distribution was used to identify statistical significance. For (G), (I), and (J), Mann-Whitney was applied to identify statistical significance.

See also Figure S9.

Discussion

The immune system has evolved to exert host defense against infections with minimal loss of fitness. Viruses, on their side, possess mechanisms to evade immune responses, thus allowing establishment and maintenance of infection. For instance, the type I IFN system is inhibited by practically all pathogenic viruses.^2,9 In addition, many viruses target intrinsic and stress-induced antiviral mechanisms, thus suggesting these to be important for host defense.^19,46,47,48 Thus, studies on viral modulation of host responses represent a powerful tool to identify host mechanisms important for host defense. In this work, we report that HSV-1 and -2 target the HIF pathway in permissive epithelial and neuronal cells, and we demonstrate this pathway to restrict virus replication in neurons, through the activation of autophagy. Finally, we report that HIF1α/2α deficiency in neurons significantly impairs control of HSV-2 infection in the mouse brain in vivo and leads to elevated inflammatory response.

A key finding of this work is that HSV-1 and -2 infections disrupt termination of transcription in permissive neuron-like SH-Sy5y cells and epithelial HaCaT cells, as well as in hESC-derived neurons, and that this disruption was enriched for HIF-induced transcripts. In non-permissive THP-1 macrophage-like cells, the transcription termination disruption was not observed. Previous work has shown that HSV-1 disrupts termination of transcription in human foreskin fibroblasts.^36,49 Next, we observed that HIF activation through small-molecule HIF1α activators, hypoxic conditions, or depletion of the HIF regulator VHL all led to reduced replication of HSV-1 and -2 in neuronal cells. Based on these data, we conclude that HSV disrupts the HIF transcription network in permissive cells to evade an innate antiviral mechanism. Regarding the mechanism of antiviral activity, we identified that HIF activation in neuronal cells potently induced expression of autophagy- and glycolysis-related genes and activated autophagy, in agreement with previous findings in other cell types.^40,41,42 Deletion of the genes encoding the essential autophagy gene ATG5, but not deletion of the type I IFN receptor chain 2, reduced the HIF-driven anti-HSV activity. These data suggest that the autophagy pathway plays an important role in the antiviral activity downstream of HIF1α activation in neurons. Autophagy has been demonstrated previously to be important for anti-HSV activity in neurons in vivo,⁴³ and there are several reports showing that autophagy contributes to antiviral activity against RNA viruses in multiple cell types including neurons.^15,18 This group of authors recently reported loss-of-function mutations in autophagy genes in patients with HSV-2 meningitis and poliomyelitis^16,18 and demonstrated causal links between the clinical phenotype and impaired control of viral replication. The exact antiviral effector mechanism of autophagy remains to be fully uncovered, but our data suggest that it involves direct degradation of incoming or progeny virions.¹⁵ Work by others has shown that HIF1 induces expression of DNAse I, which exerts antiviral activity against hepatitis B virus,³⁰ and HIF agonists have also been shown to exert antiviral activity against SARS-CoV-2 through a mechanism involving down-regulation of ACE2.^29,50 Thus, the HIF1 pathway induces a number of antiviral actions, and here, we suggest that notably autophagy is important in neuronal cells.

Previous work has identified the HIF pathway to promote host defense, mainly against bacterial infections. For instance, HIF1α activation is induced by bacterial infection, even under normoxia, and was reported to be essential for optimal phagocytosis and bactericidal capacity of phagocytes.²⁶ Likewise, HIF1α agonist treatment induces innate defense against a panel of bacterial skin infections,²⁸ a process likely dependent on HIF-dependent production of antimicrobial peptides.²⁷ In the context of bacterial infections and upon stimulation with bacterial products, HIF1α has been reported to both promote and counteract inflammatory responses. For instance, hypoxia induces expression of TLR4,⁵¹ and HIF1α promotes TLR4 signaling.⁵² We found that HIF activation alone did not induce expression of type I IFNs or inflammatory cytokines, that double-stranded RNA induced IFNB expression independent of HIF1A, and that the antiviral activity of the HIF1α activator used was independent of the type I IFN receptor. This suggests that the antiviral activity of HIF is independent of the IFN system. With respect to the physiological importance of the HIF pathway during virus infections, we found that neuron-specific HIF1α/2α deficiency renders mice highly susceptible to HSV-2 infection in the CNS and leads to elevated signaling through the cGAS-STING pathway. Likewise, mice deficient in HIF1α selectively in alveolar type II epithelial cells exhibit elevated lung influenza A virus load and an augmented inflammatory response, resulting in aggravation of disease.⁵³ Thus, the HIF pathway can exert both direct antimicrobial activity and modulate inflammatory responses, with the latter occurring in a context-dependent manner dependent on the spectrum of PRRs activated during specific infections.

We recently suggested that constitutive and stress-induced immune mechanisms contribute to antimicrobial defense independent of PRRs and that they act in a manner that leads to limited activation of inflammation.¹⁴ Our present data show that the HIF activation in neuronal cells did not induce expression of ISGs and inflammatory cytokines in neuronal cells. In fact, mice with neuron-specific deficiency of HIF1α/2α exhibited elevated activation of TBK1 and expression of ISGs and inflammatory cytokines in the HSV-2-infected brain. Similar data were reported in an influenza A virus infection model.⁵³ Mechanistically, HIF could limit inflammation both through restriction of virus replication and autophagy-mediated direct antagonism of signaling, as has been shown for several PRRs.^54,55,56 This suggests that the antiviral activity of the HIF pathway is independent of PRRs and classical inflammatory activities and even counteracts the inflammatory response, either through a direct mechanism or indirectly by reducing the level of viral immune-stimulatory molecules. In agreement with this, we also observed that human microglia respond with lower signaling through the cGAS-STING pathway and reduced expression of inflammatory genes following exposure to supernatants from HSV-2-infected neurons cultured under hypoxic versus normoxic conditions.

The above-described results suggest a regulatory role for HIF in inflammation. However, HIF1α is mostly described to have the capacity to promote inflammatory responses, e.g., in Th17 cells and in M1 macrophages in the mature stages of immune responses.^57,58 However, several anti-inflammatory activities of the HIF system have also been reported, including, for instance, inhibition of release of neutrophil extracellular traps and associated cell death.^59,60,61 Therefore, we propose that the HIF pathway can exert both homeostasis-protecting antimicrobial defense and amplification of inflammation, with the function being dependent on the context. The factors governing the activity of HIF in the microenvironment remain to be fully characterized. It is tempting to speculate that HIF-induced gene expression early during infection predominantly induces homeostasis-protecting activities, such as autophagy and microbial restriction, whereas sustained signaling leads to profound metabolic reprogramming and inflammation driven by activation of macrophages, and T cells. In support of this, hypoxia and increased expression of HIF target genes were observed in infiltrating immune cells in the cornea during HSV-1 keratitis in mice, and blockage of HIF1 activation decreased influx of CD4 T cells and non-granulocytic myeloid cells.⁶²

Collectively, in this work, we report that HSV infection in neurons disrupts the HIF pathway to block a potent antiviral activity and hence obtains a replication advantage in the arms race between viruses and the immune system. This illustrates that early host responses to infection involve stress-sensing immune mechanisms, which, together with PRR-driven responses, contribute to control of viral replication. Harnessing the HIF pathway may have therapeutic potential.

Limitations of the study

Much of the mechanistic data of this work were from neuron-like cell lines. Therefore, although we do provide data from mice and hESC-derived neurons, we do not know to what extent the phenomenon and proposed mechanism contribute to host defense in humans. For instance, this work does not address whether hypoxia-activated autophagy contributes to antiviral defense and control of inflammation in the human brain. Also, the disruption of host transcripts in vivo was not characterized. Moreover, the authors did not provide as extensive a comparison of our data on antiviral activity of HIF activators with other studies as we would have liked.^63,64 This was partly due to toxic effects of some compounds in the SH-SY5Y cells used in this study. Finally, the translational potential of the findings was not examined in detail, thus leaving open the question of whether HIF activation has potential for treatment of viral brain infections.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
HIF-1alpha (D2U3T)	Cell Signaling	#14179S; RRID:AB_2622225
THOC1	Sigma-Aldrich	#HPA019687-25UL; RRID:AB_1857980
Vinculin	Sigma-Aldrich	#V9131-0.2ML; RRID:AB_477629
LC3B	Cell Signaling	#2775; RRID:AB_915950
ATG5 (D5F5U)	Cell Signaling	#12994; RRID:AB_2630393
VHL	Cell Signaling	#68547; RRID:AB_2716279
BNIP3	Cell Signaling	#44060; RRID:AB_2799259
STAT1	Cell Signaling	#9172; RRID:AB_2198300
pSTAT1	Cell Signaling	#7694S; RRID:AB_10950970
IFNAR2	Abcam	#ab193410
TBK1	Cell Signaling	#38066; RRID:AB_2827657
pTBK1	Cell Signaling	#5483; RRID:AB_10693472
V5	Abcam	#ab27671; RRID:AB_471093
IRF3	Cell Signaling	#4302; RRID:AB_1904036
pIRF3	Cell Signaling	#37829; RRID:AB_2799121
STING	Cell Signaling	#13647; RRID:AB_2732796
pSTING	Cell Signaling	#50907; RRID:AB_2827656
peroxidase-conjugated F(ab)2 donkey anti-mouse IgG (H + L)	Jackson ImmunoResearch	#PRID: AB_2340774
peroxidase-conjugated F(ab)2 donkey anti-rabbit IgG (H + L)	Jackson ImmunoResearch	#PRID: AB_2340590
LC3	Nordic Biosite	#PM036; RRID:AB_2274121
P62	Progen	#GP62-C; RRID:AB_2687531
ICP5/VP5	Abcam	Ab6508
β-Tubulin	Cell Signaling	#2146; RRID:AB_2210545
Alexa 488	Thermo Fischer	#A-21206; RRID:AB_2535792
Alexa A568	Thermo Fischer	#A-11075; RRID:AB_2534119
DAPI	Santa Cruz	#sc-3598
goat anti-mouse IgG1 biotin	SouthernBiotech	#1071-08; RRID: AB_2794427
Bacterial and virus strains
HSV-1 (KOS) strain	ATCC	# VR-1493
HSV-2 (333) strain	Kristina Eriksson	Iversen et al.6
Clinical isolates HSV-1 2762	Kristina Eriksson	Iversen et al.6
HSV-2 B4327UR	Kristina Eriksson	Iversen et al.6
VSV (Indiana) strain	ATCC	#VR-1238
EMCV (EMC) strain	ATCC	#VR-129B
Influenza virus A (H1N1), PR8 strain	ATCC	#VR-1469
Measles virus (Edmonston) strain	ATCC	#VR-24
SARS-CoV-2 (MZ314997/B.1.1.7 variant)	Arvind Patel, University of Glasgow	SARS-CoV-2 B1.1.7
Chemicals, peptides, and recombinant proteins
CoCl2	Sigma-Aldrich	#14726-62-6
Diethyl succinate	Sigma-Aldrich	#123-25-1
Dimethyl sulfoxide (DMSO)	Sigma-Aldrich	#67-68-5
Vadadustat	MedChemExpress	#HY-101277
MK-8617	MedChemExpress	#HY-101023
Dimethyloxallyl Glycine	MedChemExpress	#HY-15893
ML228	Tocris	#4565
Poly(I:C) LMW	InvivoGen	#tlrl-picw
2-Deoxy-D-glucose	Sigma-Aldrich	#D8375
Alt-R® S.p. cas9 Nuclease V3, 500 μg	Integrated DNA Technologies	#1081059
Critical commercial assays
Hypoxyprobe-1™Plus Kit	Hydroxyprobe	HP2-200Kit
High Pure RNA Isolation kit	Roche	#11828665001
NE-PER Nuclear and Cytoplasmic Extraction Reagent Kit	Thermo Fisher Scientific	#78833
Brilliant III Ultra-Fast SYBR Green QRT-PCR Master Mix kit	Agilent Technologi	#600886
RNA-to-Ct-1-Step kit	Applied BioSYSTEMS	#4392653
TaqPath™ qPCR Master Mix	Applied Biosystems	#A16245
CXCL10 ELISA kit	Bio-techne	#DY466
KAPA mRNA Hyper Prep Kit	Roche	N/A
SMARTer Stranded Total RNA-Seq Kit v2- Pico Input	Takara	N/A
SMARTer Stranded Total RNA Sample Prep Kit- HI Mammalian	Takara	N/A
Deposited data
RNA-seq	This paper	European Nucleotide Archive. Accession Number:PRJEB60497
Experimental models: Cell lines
Human neuroblastoma SH-Sy5y	ATCC	#CRL-2266
human HaCaT keratinocytes	ATCC	# HB-241
Human monocytic leukemia THP1	ATCC	#TIB-202
Vero cells	ATCC	#CCL-81
H9	WiCell	WA09
VeroE6 cells expressing human TMPRSS2 (VeroE6-hTMPRSS2)	Gift from Professor Stefan Pöhlmann, University of Göttingen	N/A
Experimental models: Organisms/strains
nHif1a/Hif2aΔΔ/ΔΔ mice	Stefan Kunze	Barteczek et al.44
Camk2a-Cre mice	Stefan Kunze	Barteczek et al.44
Oligonucleotides
Synthetic guide RNA for HIF1A	This study	Table S1
Synthetic guide RNA for ATG5	This study	Table S1
synthetic guide RNA for AAVS1	This study	Table S1
synthetic guide RNA for VHL	This study	Table S1
synthetic guide RNA for IFNAR	This study	Table S1
gB HSV1/2 Forward primer	This study	Table S1
gB HSV1/2 Reverse primer	This study	Table S1
gB HSV1 Forward primer	This study	Table S1
gB HSV-2 Forward primer:	This study	Table S1
gB HSV1/2 Reverse primer	This study	Table S1
HSV-1 probe:	This study	Table S1
HSV-2 probe	This study	Table S1
nucleocapsid (N2) SARS-CoV-2 Forward primer	This study	Table S1
nucleocapsid (N2) SARS-CoV-2 Reverse primer	This study	Table S1
nucleocapsid (N2) SARS-CoV-2 probe.	This study	Table S1
Influenza matrix 2 (M2). Forward primer	This study	Table S1
Influenza matrix 2 (M2). Reverse primer	This study	Table S1
Influenza matrix 2 (M2). Reverse primer	This study	Table S1
Gene-specific primer BNIP3	This study	Table S1
Gene-specific primer HK2	This study	Table S1
Gene-specific primer PFKFB2	This study	Table S1
Taqman primers and probe BNIP3-FAM	Thermo Fisher	#Hs04187525_g1
Taqman primers and probe ENO1-FAM	Thermo Fisher	#Hs00361415_m1
Taqman primers and probe EPAS-FAM	Thermo Fisher	#4331182 Hs01026149_m1
Taqman primers and probe 18S rRNA	Thermo Fisher	#Hs99999901_s1
Taqman primers and probe ENO1	Thermo Fisher	#Mm01619597_g1
Taqman primers and probe VEGFA	Thermo Fisher	#Mm00437306_m1
Taqman primers and probe CXCL10	Thermo Fisher	#Mm00445235_m1
Taqman primers and probe IL6	Thermo Fisher	#Mm00446190_m1
Taqman primers and probe TNFA	Thermo Fisher	#Mm00443258_m1
Taqman primers and probe BNIP3	Thermo Fisher	#Mm00833810_g1
Software and algorithms
QIAGEN CLC Workbench	QIAGEN	N/A
GraphPad Prism version 8.4.3	Graph Pad	N/A
ChemiDoc Imaging System	BioRad	N/A
STAR RNA-seq read aligner	anaconda.org/bioconda	N/A
Other
Dulbecco’s Modified Eagle Medium (DMEM)	Sigma-Aldrich	#D5796
10% heat inactivated fetal calf serum (FCS)	Sigma-Aldrich	#F4135
1% penicillin/streptomycin	Sigma-Aldrich	#S1277
L-Glutamine	Sigma-Aldrich	#56-85-9
RPMI media	Sigma-Aldrich	#R8758
phorbol 12-myristate 13-acetate (PMA)	Sigma-Aldrich	#16561-29-8
non-essential amino acids	Life Technologies	#11140035
knockout serum replacement	Life Technologies	#10828028
FGF-2	Peprotech	#AF-100-18C
Activin A	R&D systems	#338-AC
insulin/transferrin/selenium	Life Technologies	51300044
N2	Life Technologies	17502–048
retinol-free B27	Life Technologies	12587010
glucose	Sigma-Aldrich	G7021-1KG
SB431542	Tocris	1614
LDN-193189	Stemgent	04–0074
poly-L-ornithine	Sigma-Aldrich	P3655-10MG
fibronectin	Sigma-Aldrich	F1141-1MG
laminin	Sigma-Aldrich	L2020-1MG
Glutamax	Thermo Fischer	35050038
Y-27632	Tocris	1254
ascorbic acid	Sigma-Aldrich	A4403-100MG
DAPT	Tocris	2634
Embryonic body medium contained mTeSR1	STEMCell	#85850
ROCK inhibitor	Tocris	# Y-27632
BMP-4	Peprotech	#120-05
SCF	Peprotech	#300-07
VEGF-121	Peprotech	#100-20A
X-VIVO 15	Lonza	#02-053Q
β-mercaptoethanol	Life Technologies	#21985023
M-CSF	Peprotech	#300-25
IL-34	Peprotech	#200-34
GM-CSF	Peprotech	#300-24
Target Retrieval Solution	Dako	#GV804
Haematoxylin Ortho	CellPath	#RBA-4213-00A
SDS-PAGE on 4–20% Criterion TGX precast gradient gels	BioRad	5671095
PVDF membranes	BioRad	1704275
Skim milk powder	Sigma Aldrich	70166-500G
Normal human immunoglobulin	Octapharma	N/A
Thiazolyl Blue Tetrazolium Bromide	Sigma	M2128-1G
Prolong Gold mounting medium	Thermo Fischer	#P36934

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Søren R Paludan (srp@biomed.au.dk).

Materials availability

All materials generated in this study are available from the lead contact upon request.

Data and code availability

• •The full NGS dataset are available at ENA (European Nucleotide Archive) with the identifier ‘ena-STUDY-AARHUS UNIVERSITY 12-12-2018-17:12:31:528-124’, under accession number ‘PRJEB60497’.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details

Cell lines, reagents, compounds, and culture conditions

Human neuroblastoma SH-Sy5y and human HaCaT keratinocytes were cultured at 37°C in Dulbecco’s Modified Eagle Medium (DMEM) (Sigma-Aldrich) supplemented with 10% heat inactivated fetal calf serum (FCS)(Sigma-Aldrich), 1% penicillin/streptomycin (Sigma-Aldrich) + L-Glutamine (Sigma-Aldrich). Human monocytic leukemia (THP-1) were cultured in RPMI media (Sigma-Aldrich) with L-glutamine supplemented with 10% heat inactivated FCS, 1% penicillin/streptomycin. THP1 cells were differentiated into macrophage cells using phorbol 12-myristate 13-acetate (PMA), 150 nM overnight. Vero cells were cultured in DMEM supplemented with 10% heat inactivated FCS, 1% penicillin/streptomycin + L-Glutamine. ML228 (Tocris), Dimethyloxallyl Glycine (Sigma-Aldrich), MK-8617 (MedChemExpress) and Vadadustat (MedChemExpress) dissolved in Dimethyl sulfoxide (DMSO) (Sigma-Aldrich), Diethyl succinate (Sigma-Aldrich), 2-Deoxy-D-glucose (Sigma-Aldrich), Poly(I:C) LMW (InvivoGen), and CoCl₂ (Sigma-Aldrich) dissolved in DMEM, were applied in the experiments.

Human embryonic stem cell culture

H9 (WA-09, WiCell) cell lines were cultured on irradiated human foreskin fibroblasts (HFF) in KSR media composed of DMEM/nutrient mixture F-12, completed with β-mercaptoethanol 0.1mM, non-essential amino acids 1%, glutamine 2mM, penicillin 25 U/ml, streptomycin 25 μg/ml, and knockout serum replacement 20% (all from Life Technologies), supplemented with FGF-2 10 ng/mL (Peprotech) and Activin A 10 ng/mL (R&D systems). All cells were cultured at 37°C 5% CO₂. Colonies were mechanically dissected and transferred to freshly prepared HFF every 7 days. Media was changed every second day.

Viruses

HSV-1 KOS strain, HSV-2 333 strain, and clinical isolates HSV-1 2762 and HSV-2 B4327UR were propagated in Vero cells and purified using ultra-centrifugation and titrated via standard plaque assay. Indiana strain VSV and EMC strain EMCV virus were treated similarly and applied in the experiments. Influenza virus A (H1N1) strain PR8, was propagated in SPF eggs in the allantoic cavity. The allantoic fluid was layered on sucrose after concentration and suspension in Hepes-Saline. The interface band was diluted, pelleted, and re-suspended in Hepes-Saline. Antigen was tested for protein concentrate of 2 mg of protein per Ml using a Bio-Rad colorimetric protein assay. Measles virus (Edmonston strain) was obtained from the American Type Culture Collection (VR-24). It was passaged on Vero cells and titrated on Vero cells.⁵⁸ B.1.1.7 SARS-CoV-2 (alpha variant, Kent, UK, isolate) was provided under MTA by Professor Arvind Patel, University of Glasgow. The viruses used are clinical isolates. The B.1.1.7 variant is in the database as MZ314997. The VeroE6 cells expressing human TMPRSS2 (VeroE6-hTMPRSS2), were kindly provided by Professor Stefan Pöhlmann, University of Göttingen, and were used for virus propagation. Briefly, VeroE6-hTMPRSS2 cells were infeted with a multiplicity of infection (MOI) of 0.05, in DMEM (Gibco) + 2% FCS (Sigma-Aldrich) + 1% Pen/Strep (Gibco) + L-Glutamine (Sigma-Aldrich) (From here, complete medium). 72h post infection supernatant, containing new virus progeny, was harvested, and concentrated on 100kDa Amicon ultrafiltration columns (Merck) by centrifugation at 4000 xg for 30 min. Virus titers were determined by TCID_50% assay and calculated by Reed-Muench method. The viral replication was validated by RT-qPCR (Taqman based) against SARS-CoV-2 genome.

Mice

Mice were housed in the Animal Facility of Aarhus University, Health, in a 12/12-h light/dark cycle and given ad libitum food and water access. The nHif1a/Hif2a^ΔΔ/ΔΔ mice were generated by crossing of Hif1a^f/f and Hif2a^f/f mice to generate Hif1a/Hif2a^ff/ff mice, and subsequent breeding with Camk2a-Cre mice.⁴⁴ For infection experiments in vivo, age-matched 7-week-old male and female were used. The experiments involving animals were approved in advance by the Animal Ethics Committee at the Danish Veterinary and Food Administration (Stationsparken 31–33, 2600 Glostrup, Denmark) and were carried out in accordance with the Danish Animal Welfare Act for the Care and Use of Animals for Scientific Purposes.

Method details

Differentiation and culture of stem cell-derived brain cells

Neuron differentiation; hESCs were mechanically dissected into pieces ∼0.5mm in diameter and transferred to vitronectin-coated organ culture plates in N2B27 medium containing 1:1 mixture of neurobasal medium with DMEM/F12 medium, supplemented with insulin/transferrin/selenium 1%, N2 1%, retinol-free B27 1%, glucose 0.3%, penicillin 25 U/ml, and streptomycin 25 μg/ml (all from Life Technologies) for 13 days. From day 0 to day 6 the cells were grown in N2B27 media with SB431542 10 μM (Tocris) and LDN-193189 100 nM (Stemgent). From day 6 to day 13 the cells were grown in N2B27 media with FGF-2 20 ng/mL (Peprotech). At Day 13, 100,000 of the cells were re-plated on 24 well plate coated with poly-L-ornithine, fibronectin, and laminin (all from Sigma-Aldrich) and grown in NBM containing neurobasal media supplemented with B27 1%, penicillin 25 U/ml, streptomycin 25 μg/ml, and Glutamax 0.5%. From day 13 to day 15 the cells were grown in NBM with Y-27632 10 μM (Tocris), ascorbic acid (AA) 200 μM (Sigma-Aldrich) and DAPT 2.5 μM (Tocris). From day 15–21 the cells were grown in NBM with AA 200 μM and DAPT 2.5 μM. The media was changed every second day and at day 21 the cells were stimulated.

Microglia differentiation: H9 cells (4 × 10⁶) were seeded per Aggrewell 800 in 24-well plates (StemCell Technologies) in embryonic body medium. Cells were cultured for 4 days to form embryonic bodies with half media change every day. Embryonic bodies were harvested using an inverted cell strainer (40 μm), and ∼15 embryonic bodies were plated per 6 wells in hematopoietic medium (HM). Two mL media were replaced every 7 d by fresh HM. After ∼30 days, primitive macrophage precursors were harvested during the media change and plated in microglia medium (MiM) into 48 wells at a density of 100,000 cells/cm². Cells were differentiated in MiM for another 7–10 d with a full media change every second day. Embryonic body medium contained mTeSR1 supplemented with 10 μM ROCK inhibitor, 50 ng/mL BMP-4, 20 ng/mL SCF, and 50 ng/mL VEGF-121 (all from Peprotech). HM contained X-VIVO 15 (Lonza) supplemented with 2 mM Glutamax, 100 U/ml penicillin, 100 μg/mL streptomycin, 55 μM β-mercaptoethanol, 100 ng/mL M-CSF (Peprotech), and 25 ng/mL IL-3 (Cell Guidance Systems). MiM contained X-VIVO 15 (Lonza) supplemented with 2 mM Glutamax, 100 U/ml penicillin, 100 μg/mL streptomycin, 55 μM β-mercaptoethanol, 100 ng/mL IL-34 (Peprotech), and 10 ng/mL GM-CSF (Peprotech).

Viral infection experiments in vitro

4x 10⁵ SH-Sy5y cells were seeded in three replicates in 24 well plate. The next day, cells were pre-treated with ML228, DMOG, CoCl_2, Diethyl succinate, MK-8617 and Vadadustat at concentrations of 0.5 μM, 5μM, 150 μM, 20mM, 10nM and 40 μM for 2 h. DMSO with the same concentration was applied for the compounds dissolved in DMSO. HSV-1 at MOI of 0.3 and HSV-2 at MOI of 0.1, were used to infect the cells. 1 h following the infection, the virus was removed, and the cells were incubated with activators and DMSO for 19 h. In the samples infected with HSV-1/2, the supernatant was used for virus plaque assay and the cells were lysed for RNA-extraction and RT-qPCR. Similarly, 4x 10⁵ SH-Sy5y cells were seeded in three replicates in 24 well plate. The next day, cells were pretreated with ML228 and DMSO at concentration of 0.5 μM. After 2 h pretreatment EMCV, at MOI of 0.1 were added to the cells. 1 h following the infection, the virus was removed, and the cells were incubated with activators and DMSO for 19 h. In these samples, supernatant was used for virus titration using TCID50% assay. Similarly, 4 x 10⁵ SH-Sy5y cells were seeded in three replicates, in 24 well plate. 24 h later, the cells were pretreated with ML228 and DMSO at concentration of 1 μM and diethyl-succinate at concentration of 20mM, for 2 h. The cells were infected with VSV at MOI of 0.001. After 1 h infection, the virus was removed, and the cells were incubated with activators for 19 h. In these samples, supernatant was used for virus titration using plaque assay.

Additionally, 4 x 10⁵ SH-Sy5y cells were seeded in five replicates, in 24 well plate. 24 h later, the cells were pretreated with ML228 and DMSO at concentration of 0.5 μM for 2 h. The cells were infected with SARS-CoV-2 at MOI of 1. Two hours following the infection, the virus was removed, and the cells were incubated with ML228 and DMSO (0.5 μM) for 48 h. Subsequently, the cells were lysed for RNA-extraction and RT-qPCR. Similarly, 4 x 10⁵ SH-Sy5y cells were seeded in five replicates, in 24 well plate. 24 h later the cells were pretreated with ML228 and DMSO at concentration of 1 μM for 2 h. The cells were infected with Influenza virus A at MOI of 0.1. Two hours after the infection, the virus was removed, and the cells were incubated with ML228 and DMSO (1 μM) for 24 h. Subsequently, the cells were lysed for RNA-extraction and RT-qPCR. For infection with measles virus, 1x 10⁶ SH-Sy5y cells were seeded in six replicates, in 12 well plates. 24 h later the cells were pretreated with ML228 or DMSO at 1 μM for 2 h. The cells were infected with measles virus at MOI of 0.1. After 2h of adsorption, leftover viruses were washed off and the cells were incubated with ML228 and DMSO (1 μM). At 30 p.i., the supernatant was harvested for titration on a plaque assay.

The same producers as mentioned above were applied for SH-Sy5y cells treated with diethyl succinate at concentration of 20 mM and infected with HSV-1, HSV-2 and VSV. Similarly, In HSV-1 and HSV-2 infection samples, supernatant was used for virus titer. In VSV infected samples, supernatant was used for virus titration using TCID50% assay. All the conditions were maintained at 37°C and 5% CO₂ in a standard culture incubator. For evaluation of the antiviral activity of ML228 in HaCat and THP-1 cells, they were pretreated with ML228 and DMSO at concentration of 0.5 μM for 2 h and subsequently infected with HSV-1 at an MOI of 0.3 and HSV-2, at an MOI 0.1. After 19 h, supernatants were used for virus plaque assay and the cells were lysed for RNA-extraction and RT-qPCR.

For the hypoxia experiments, 4 x 10⁵ SH-Sy5y cells were seeded. The following day the cells were exposed to 24 h hypoxia prior infection, using a hypoxia incubator chamber (STEMCELL, #27310) set to 37C, 5% CO₂ and 1% O₂. After 24 h, cells were infected with HSV-1 at MOI of 0.3 and HSV-2 at MOI of 0.1 for 1 h. Viruses were washed off after 1 h and cells were incubated in the hypoxia chamber incubator for 19 h. After 19 h, cells were lysed for RNA-extraction and RT-qPCR. hESC-derived neurons were pretreated with ML228 and DMSO at concentration of 1.4 μM for 2 h and were infected with HSV-1 and HSV-2, at an MOI of 1. After 48 h, the supernatant was used for virus plaque assay and the cells were lysed for RNA-extraction and RT-qPCR.

Animal infection experiments

For infection experiments in vivo, age-matched, 7-week-old male and female mice were anesthetized with i.p. injection of a mixture of ketamine (100 mg kg-1 body weight) and xylazine (10 mg kg-1 body weight). Both corneas were scarified in a 10 × 10 crosshatch pattern with a 25-gauge needle and mice inoculated with 400 pfu/eye of HSV-2 (strain 333) in 5 μL of infection medium (DMEM containing 200 IU mL-1 penicillin and 200 mg mL-1 streptomycin), or mock-infected with 5 μL of infection medium. Mice were scored for disease and weighed at the indicated times after infection. The scoring was performed as a blinded study, with head swelling scored as 0, no signs of disease; 1, minor head swelling; 2, major head swelling or purulent lesion; 3, very bad general condition.

Hypoxyprobe-1 staining and immunohistochemistry

Intra-peritoneal injection of either saline or pimonidazole hydrochloride (Hypoxyprobe-1Plus Kit, Hydroxyprobe) was given to uninfected, and HSV-2-infected mice on 5-day p.i. Mice received pimonidazole at the concentration of 60 mg/kg mouse body weight. The drug was dissolved in 0.9% of normal saline. Ninety minutes after injection, mice were anastized and perfused with PBS and 4% formaldehyde fixative through the circulatory system and postfixed with phosphate-buffered 4% formaldehyde and brain sections were cut at 9-μm thickness and every section with a gab of 250 μm was stained. The antibody retrieval was performed at 80°C for 30 min with Target Retrieval Solution (Dako). The sections were treated with methanolic H₂O₂ (3% H₂O2 and 1% absolute methanol) for 30 min to inactivate endogenous peroxidase activity. The sections were then blocked with 0.5% bovine serum albumin (BSA) in TBS for 1 h at room temperature (RT), and incubated overnight at 4°C with anit-pimonidazole antibody (1:50)(Hypoxyprobe-1Plus Kit, Hydroxyprobe) in tris-buffered saline (TBS) containing 0.1% BSA, 0.3% Triton X-, followed by several washes in TBS containing 0.3% Triton X- TBS and staining with and the secondary antibody, goat anti-mouse IgG1 biotin (1:200) (1071-08, SouthernBiotech) for 1 h at RT. Horseradish peroxidase-labelled antibodies were developed with chromagen diaminobenzidine. Finally, the slides were rinsed in water, stained with Haematoxylin Ortho (CellPath), dehydrated, and prepared for imaging. The imaging was performed on Zeiss AX10, Scope A1 microscope using ×10 and ×40 objectives.

CRISPR/Cas9-mediated genome editing

For genome-editing of HIF1A, VHL, IFNAR2, ATG5 and AAVS1, we used synthetic guide RNA sequences. See Table S1 for Oligonucleotides sequences, targeting HIF1A, VHL, IFNAR2, ATG5 and AAVS1. The sgRANA guides as well as Cas9 nuclease were transfected into the wild type SH-Sy5y cells using electroporation and nucleofactor technology (Lonza). After nucleofection, cells were grown in DMEM (Sigma-Aldrich) supplemented with 10% heat inactivated FCS (Sigma-Aldrich), 1% penicillin/streptomycin (Gibco) + L-Glutamine (Sigma-Aldrich). After a few days, these cells were extended to larger cultures and were screened for the absence of protein by immunoblotting.

Multiplexed RNA-sequencing

SH-Sy5y, HaCaT, and THP1 cells infected with HSV-1 and HSV-2

15x 10⁴ SH-Sy5y and HaCaT cells were seeded in 24-well culture plates. After 24 h, cells were washed and infected with HSV-1 at MOI of 3 and HSV-2 at MOI of 1 for 4 h, 12 h and 24 h. Supernatants were discarded post infection, and the cells were used for RNA isolation and mRNA-seq. Similarly, a total of 3x10³ THP1 cells were seeded in 24 well culture plates. Cells were differentiated to macrophage using 150 nM PMA, for 24 h. After 24h, cells were washed and infected with HSV-1 at MOI of 3 and HSV-2 at MOI of 1 for 4 h, 12 h and 24 h.

Two biological replicates for each experiment were included in the study. Total RNA was isolated using an RNA isolation kit (Roche) according to the manufacturer instructions. mRNA-seq from total RNA were extracted and each cell type infected with different viruses at different time points were uniquely barcoded using KAPA mRNA Hyper Prep Kit (Roche). In total 42 libraries were pooled in one sample and were sequenced on Illumina’s NovaSeq-S1 sequencing platform using a paired-end protocol.

SH-Sy5y cells treated with HIF activator and infected with HSV-1 or HSV-2

40x10⁴ SH-Sy5y cells, were seeded in 24-well culture plates. After 24 h, cells were washed and pre-treated with ML228 (0.5 μM) and DMSO (0.5 μM) for 2 h. After 2h, HSV-1 (MOI 3)

and HSV-2 (MOI 1) were added to the cells. After 24 h incubation of the virus with ML228 and DMSO, supernatants were discarded, and the cells were used for RNA isolation. Samples were rRNA depleted and were prepared for sequencing using SMARTer Stranded Total RNA Sample Prep Kit- HI Mammalian (Takara). Two biological replicates were included for each condition in the study. Each library was barcoded uniquely, the libraries were then pooled and sequenced as 150 bp paired-end reads on an Illumina HiSeq sequencer.

SH-Sy5y cells depleted for HIF1A by genome-editing and subjected to infection

40x10⁴ SH-Sy5y cells, depleted for HIF1A or AAVS1 through sgRNA/Cas9 RNP transfection, were seeded in 24-well culture plates. After 24 h, cells were washed and treated with ML228 and DMSO (0.5 μM) for 8 h. Supernatants were discarded post treatment, and the cells were used for RNA isolation (according to manufacturer’s instructions). Samples were rRNA depleted and prepared for sequencing using SMARTer Stranded Total RNA Sample Prep Kit- HI Mammalian (Takara). For each condition, three biological replicates were included in the study. Each library was barcoded uniquely, and the libraries were pooled and sequenced as 150 bp paired-end reads on an Illumina HiSeq sequencer.

hESC-derived neurons infected with HSV-1 and HSV-2

At day 21, the cells were infected with HSV-1 (MOI 0.3) and HSV-2 (MOI 0.3) for 48 h. Mock was used as negative control. hESC-derived neurons were prepared for sequencing using SMARTer Stranded Total RNA-seq Kit v2- Pico Input (Takara). Three biological replicates for each experiment were included in the study. Each library, was barcoded uniquely and was pooled and sequenced as 150 bp paired end reads on Illumina HiSeq sequencer.

Mapping of RNA-seq data

RNA sequencing reads mapping was performed using QIAGEN CLC Workbench. For the following conditions sequenced later, namely (1) SH-Sy5y cells infected with HSV-1 and HSV-2 upon ML228 treatment, (2) HIF1A- and AAVS1 sgRNA-transfected cells treated with ML228, (3) hESC-derived neurons infected with HSV-1 and HSV-2 and Mock, STAR was used as read mapper. Overall, each condition yielded at least 44 million paired reads, which passed the FASTQC quality control. Collectively, in conditions infected with viruses, reads were first mapped against HSV-1 genome strain KOS (downloaded from NCBI with accession code of JQ673480.1) and HSV-2 genome strain HG52 (downloaded from NCBI with accession code of NC_001798.2) with 99% genomic sequence identity with strain 333.⁶⁵ Next, the reads not mapped to the virus genome were mapped against human genome (hg38) and the BAM files were exported. The same procedures were applied for hESC-derived neurons infected with HSV-1 and HSV-2. The resulting BAM files were exported for further processing.

Quantification of readthrough and read-in

For further investigations, a BED file was constructed which delineates upstream and downstream regions for every gene following the protocol in ref. 36. Briefly, for all protein coding genes in the human genome (hg38, 21463 genes) a window downstream of the 3′ end and upstream of the 5′-end was created. The window size was set to 5kbp if the adjacent intergenic region was 15kbp or longer, or to a third of the length of the intergenic region otherwise. In case the resulting length of the window was below 100bp, the neighboring genes were excluded from the analysis. Any pair of genes on the same strand with overlap between their annotated regions or adjacent regions were also removed (3578 genes), resulting in a bed file with 17885 genes and their adjacent upstream and downstream regions each. The resulting BED file was used to calculate the coverage (read counts) from the BAM files using the bedtools utilities.⁶⁵ Thereby we quantified the number of reads mapped against each gene as well as its upstream (read-in) and downstream region (readthrough). Then, transcripts per million (TPM) values for the genes and the window of downstream of the 3′ -end or upstream of the 5′-end of each gene were calculated for all conditions. The ratio of TPM read-in or readthrough versus the TPM of the associated gene was used to quantify the extent of readthrough or read-in. The resulting read-in and readthrough ratios in all conditions were normalized versus the TPM of Mock, resulting in relative TPM values. If the fold change of this relative TPM value was <3 for read-in and <5 for readthrough (Extended Data Figure 2A), the gene is classified as non-disturbed, that is no drastic increase in transcriptional termination disruption downstream or upstream of the gene during virus infection.

Gene expression analysis—Identification of up- and down-regulated genes

From the read-mapping, in SH-Sy5y, HaCat and THP1 cells infected with HSV-1 and HSV-2, expression values for each time point relative to the Mock as the control group were calculated for every gene and transcript (a total of 21474 genes) with CLC Differential Expression RNA-seq tool, using the Wald statistical test to test the differences between all test samples versus Mock. The fold changes are calculated based on generalized linear model (GLM) method, which corrects differences across samples with different library sizes as well as confounding factors effect. The threshold for the false discovery rate (FDR) p value was determined using Benjamini-Hochberg correction for multiple testing. In this study, thresholds of the FDR p value ≤0.05 and |log₂(FoldChange)|≥ 1.5 were used to define significant differentially expressed genes for further functional analysis. In SH-Sy5y cells infected with virus upon ML228 treatment, differential expression analysis was performed using DESeq2 in R for human and viral gene expression profiles. Since the global expression level of viral genes differs substantially between different groups, the viral differential expression was done in the context of human gene expression. This was done by merging feature counts outputs from human and viral gene expressions before DESeq2 analysis. FDR p value ≤0.05 and |log₂(FoldChange)|≥ 1 were used to define significant differentially expressed genes for further functional analysis. Similarly, DESeq2 was used for differential expression analysis in hESC-derived neurons and SH-Sy5y cells depleted for HIF1A and AAVS1 sgRNA transfected. A threshold of |log₂(FoldChange)|≥ 1 was used to define significant differentially expressed genes for further functional analysis. Similarly, in SH-Sy5y cells depleted for HIF1A or AAVS1 (as control), DESeq2 in R was used for the analysis of up- and down-regulated genes. FDR p value ≤0.05 and |log₂(FoldChange)|≥ 1 were used to define significant differentially expressed genes for further functional analysis.

Functional enrichment analysis

Biological functions of transcriptionally disrupted combined with down-regulated genes as well as up-regulated genes in each cell type were identified through pathway enrichment analysis using Enrichr.⁶⁶ Only pathways with a p value <0.05 (using Fisher exact test) which have appeared in at least three different databases in Enrichr are represented.

Sub-cellular fractionation and immunoblotting

4x10⁵ SH-Sy5y cells were seeded in 24 well plates. The day after cells were treated with ML228 and DMSO at concentration of 0.5 μM, for overnight. Subsequently, supernatant was discarded, and the cells were lysed on ice using sub-cellular fractionation (NE-PER Nuclear and Cytoplasmic Extraction Reagent Kit, Thermo Fisher Scientific). Following sub-cellular fractionation, HIF1α protein was assessed in these cells, in cytoplasmic and nuclear fractions through immunoblotting. On day 21, hESC-derived neurons were incubated for 24 h with ML228 (1.4 μM). Next, whole cell lysate were used for immunoblotting. Immunoblotting was performed as described.⁴⁷ Briefly, samples were separated by SDS-PAGE on 4–20% Criterion TGX precast gradient gels (BioRad). Each gel was run at 70 V. Afterward, proteins transfer onto PVDF membranes (BioRad) was done using Trans-Blot Turbo Transfer system for 7 min. Membranes were blocked for 1 h using 5% skim-milk (Sigma Aldrich) at room temperature in PBS supplemented with 0.05% Tween 20 (PBST). Membranes were cut in smaller pieces (according to the proteins molecular kDa) and probed overnight at 4 °C with primary antibodies in BSA as follow: anti- HIF1α (D2U3T, Cell Signaling 1:1000), anti-THOC1 (HPA019687-25UL, Sigma-Aldrich 1:1000), anti-Vinculin (V9131-0.2ML, Sigma-Aldrich, 1:10000), anti-ATG5 (D5F5U, Cell Signaling 1:1000), anti-LC3B (2775, Cell Signaling, 1/1000), anti-VHL (68547, Cell Signaling, 1/1000), anti-BNIP3 (44060, Cell Signaling, 1/1000), anti-STAT1 (9172S, Cell Signaling, 1/1000), anti-pSTAT1(7694S, Cell Signaling, 1/1000), anti IFNAR2 (ab193410, Abcam, 1/1000), anti-TBK1 (38066, Cell Signaling, 1/1000), anti-pTBK1 (5483, Cell Signaling, 1/1000), anti-V5 (ab27671, Abcam, 1/1000), anti-IRF3 (4302, Cell Signaling, 1/1000), anti-pIRF3 (37829, Cell Signaling, 1/1000), anti-STING (13647, Cell Signaling, 1/1000), anti-pSTING (50907,Cell Signaling, 1/1000).

THOC1 and Vinculin primary antibodies were used as loading control for nucleus and cytoplasm respectively. Membranes were washed three times in PBST. Subsequently, secondary antibodies, peroxidase-conjugated F(ab)2 donkey anti-mouse IgG (H + L) (1:10000) or peroxidase-conjugated F(ab)2 donkey anti-rabbit IgG (H + L) (1:10,000) (Jackson ImmunoResearch) were added to the membrane in PBST 1% milk for 1 h at room temperature. After three times washing, all membranes were exposed by either the SuperSignal West Pico PLUS chemiluminescent substrate or the SuperSignal West Femto maximum sensitivity substrate (ThermoScientific) and an ChemiDoc Imaging System (BioRad). Uncropped immunoblots are shown in Data S10.

RNA isolation and reverse transcription quantitative PCR

High Pure RNA Isolation kit (Roche) was used to extract total RNA according to the manufacturer’s instructions. Nanodrop spectrometry (Thermo Fisher) was used for RNA quality and concentration measurment. Viral replication was assessed through measuring RNA level, against HSV-1/HSV-2 Glycoprotein B (gB) using Brilliant III Ultra-Fast SYBR Green QRT-PCR Master Mix kit (Agilent Technologi). See Table S1 for Oligonucleotides sequences. Additionally, we confirmed HSV-1 and HSV-2 viral replication against Glycoprotein B (gB) using Taqman with customized primers and probe. See Table S1 for Oligonucleotides sequences. For SARS-CoV-2 replication, RNA level against nucleocapsid (N2) protein was assessed using qPCR (Taqman).⁴⁷ See Table S1 for Oligonucleotides sequences. For influenza virus A, levels of viral matrix 2 (M2) RNA were measured by using RTqPCR (Taqman) with the customized primers (M2 Seq 7).⁶⁷ Gene expression assays for BNIP3 (Hs04187525_g1) and ENO1 (Hs00361415_m1) EPAS1 (4331182 Hs01026149_m1), and 18S rRNA (Hs99999901_s1) was done by RT-qPCR based on Taqman primers. Gene expression was measured using premade TaqMan assays and the RNA-to-Ct-1-Step kit according to the manufacturer’s recommendations (Applied BioSYSTEMS). ΔCT was defined as the difference between the Target gene CT and the 18S rRNA CT. ΔΔCT values were calculated for each replicate in different conditions as follows: (Drugs ΔCT – Vehicle ΔCT) and (hypoxia/normoxia-infected ΔCT – Uninfceted ΔCT). In each group, fold change in the target gene mRNA expression was expressed as 2^–ΔΔCT. To distinguish between disrupted and non-disripted RNA by RT-qPCR, two micrograms of RNA was subjected to reverse transcription using oligo(dT)₁₅ or gene-specific primers for BNIP3, HK2 and PFKFB2 See Table S1 for Oligonucleotides sequences. The cDNAs were amplified by PCR using the TaqPath qPCR Master Mix (Applied Biosystems).

Gene expression assays in mice were done by RT-qPCR based on Taqman primers: ENO1(Mm01619597_g1), VEGFA (Mm00437306_m1), CXCL10 (Mm00445235_m1), IL6 (Mm00446190_m1), TNFA (Mm00443258_m1) and BNIP3 (Mm00833810_g1). ΔCT was defined as the difference between the Target gene CT and the 18S rRNA CT. The mRNA expression was expressed as 2^–ΔCT.

ELISA

Protein levels of murine CXCL10 were measured in cleared brain stem homogenates using an ELISA kit (Bio-techne) and following the instructions from the manufacturer.

Virus plaque assays

Supernatants collected from HSV-infected SH-Sy5y cells, HaCat, THP-1 and hESC-derived neurons were subjected to plaque assay. For this purpose, 1.2x10⁶ vero cells/Petri dish (diameter 60 mm), in DMEM were seeded and to settle and to form the monolayer, were left overnight. The next day, 100 μL of supernatants in appropriate serial dilutions were used to infect the cells. The cells were incubated for 1 h in 37°C. Subsequently, 5 mL DMEM supplemented with 0.2% normal human immunoglobulin (Octapharma) were added to the cells. The plates were incubated for 2–3 days in 37°C and stained with 0.03% methyl blue. The plaques were counted, and the results were represented as PFU (Plaque-forming units) per mL. For measurement of VSV titers by plaque assay, Vero cells were seeded in a 6-well plate at a density of 0.7x10⁶ cells per well in DMEM and were allowed to settle and form a monolayer overnight. The next day, 100 μL of virus-containing supernatants in appropriate serial dilutions were used to infect the cells. The cells were incubated for 1 h in 37°C. Subsequently, the virus-containing media was replaced with 1% methyl cellulose in DMEM. The plates were incubated for 2 days at 37°C and stained with a 2% crystal violet solution. The plaques were counted, and the results were represented as PFU (Plaque-forming units) per mL.

TCID50 assay

Supernatants collected from SH-Sy5y cells treated with ML228 and DMSO and infected with EMCV, were used for virus titer. Vero cells were seeded at a density of 37,500 per well in DMEM in flat-bottom 96-well plates. The following day, 10 μL of a 10-fold serial dilution of the samples were added to the cells. One full plate was used for each sample replicate, and each dilution was repeated 8 times on a plate. The plates were incubated for 48 h at 37°C and stained with 0.03% methyl blue. Virus-mediated cytopathic effect was assessed by TCDI50% calculated by Reed-Muench method.

MTT assay

Cells were seeded in 96-well plates at a density of 10 x 10³ cells/well and incubated overnight with DMEM. The next day, cells were treated with 90 μL DMSO and ML228 for 24 h, at concentration of 0.1, 0.5 and 1.4 μM in triplicates. In the following day, 10 μL 98% Thiazolyl Blue Tetrazolium Bromide (Sigma) was added to each well, and cells were incubated for 4 h. After 4h incubation, the medium was removed, and the insoluble formazan was dissolved in DMSO. The plates were shaken for 10 min and were read at 570 nm wavelength. The assay should be protected from the light. Cell viability was calculated based on the following formula: (ML228 treated cells – blank/DMSO treated cells – blank) x 100. DMEM was used as blank.

Confocal microscopy and image analysis

40 x 10⁴ SH-Sy5y cells, were seeded in 24-well culture plates on coverslips. Cells were treated as specified for the specific experiments. Cells were fixed with 4% formaldehyde (VWR 9713.1000) for 15–20 min at RT, and permeabilized with methanol for 10 min at −20°C and blocked in PBS 3% donkey serum for 1 h at RT. The fixed cells were stained with rabbit anti-LC3 (Nordic Biosite, PM036, dilution 1:2000), guinea pig anti-p62 (Progen, GP62-C, dilution 1:200), rabbit anti-β3-tubulin (Cell Signaling, 2146, 1:100) for 1 h at RT. The cells were washed three times with PBS and incubated for 1 h with a secondary antibody of donkey anti-rabbit Alexa 488 antibody (Thermo Fischer, A-21206, dilution 1:1000) for LC3, a secondary goat anti-guinea pig Alexa A568 (Thermo Fischer, A-11075) for P62 and DAPI. Antibodies were diluted in PBS 1% donkey serum. After three washes with PBS, coverslips were mounted using Prolong Gold mounting medium (Thermo Fischer P36934). Images were acquired on a Zeiss LSM 800 confocal microscope and processed with the Fiji software package. LC3 foci were counted manually in 25 randomly selected cells per condition. HSV1 capsids associated with the nuclear rim were quantified over 100 nuclei.

Quantification and statistical analysis

All the experimental data were analyzed using GraphPad Prism version 8.4.3 (GraphPad, San Diego, CA, (USA). p values <0.05 were considered significant. The data are shown as means ± st. dev. When the data exhibited normal distribution, two-tailed Student’s t-test (parametric) was used to determine the statically significance. When the dataset did not pass the normal distribution test, two-tailed Mann-Whitney was used to show the statistical significance. two-way ANOVA with Bonferroni’s post hoc test were used for in vivo studies. The majority of experiments were performed at least 3 times. A few experiments were performed 2 times. Results from statistical analysis can be found in the figures, and information on statistical tests used, etc can be found in the figure legends.

Published: February 15, 2024