Endosomal fusion of pH-dependent enveloped viruses requires ion channel TRPM7

Validation of TRPM7 knockdown in SVG-A cells

TRPM7 was depleted in SVG-A cells, a human fetal astroglial cell line, using siRNA. Compared to control (Ctl) cells which were treated with non-targeting siRNA, the knockdown (M7KD) SVG-A cells were depleted of TRPM7 mRNA to ~20% of levels in control cells (Fig. 1a). Whole-cell patch clamp electrophysiology showed that siRNA-mediated depletion of TRPM7 greatly reduced the typical outwardly rectifying TRPM7 currents (I_M7) that are readily detected in control cells. The current-voltage (IV) relationships seen in Ctl and M7KD cells are shown overlaid (Fig. 1b, left panel). Quantification of peak current densities observed at +100 mV is shown (Fig. 1b, right panel). These analyses confirm the efficient depletion of TRPM7 in M7KD cells.

Assessment of infection by eGFP-expressing VSV particles

VSV particles engineered to express eGFP-tagged Matrix (M) protein (Fig. 1c, top panel schematic) were used to infect the SVG-A cells. Three stages of infection can be observed using this approach, as shown in the bottom panel of Fig. 1c and schematized in Supplementary Fig. 1a. In the first stage, the internalization of the viral particles yields endosomes that contain the fluorescent virus, and these can be visualized as internalized puncta. In this stage, the virus has not yet escaped from the endosomes, and from the standpoint of productive infection, these cells are still counted as “negative” for infection. In the second stage, successful fusion of these viral particles enables the virus to escape from the endosome and into the cytosol. Once the virus penetrates the cytoplasm, the M-protein dissociates and coats the nuclear membrane of the host cell, and this is visualized as nuclear rings³⁷. These cells are counted as “Nuclei+”. Finally, in the third stage of the infection, the virus replicates, and the amplified MeGFP protein fills the cytosol (Cytosol+). Thus, the latter two stages, Nuclei+ and Cytosol+, represent successful penetration of the virus across the endosomal membrane which is indicative of infection. The outlines of the cells are determined based on staining with fluorescently labeled Wheat Germ Agglutinin (WGA) – representative images are shown (Supplementary Fig. 1b).

TRPM7-depleted SVG-A cells are resistant to VSV-chimeras of Ebola, Lassa, and LCMV viruses

The VSV genome can be further manipulated to generate VSV chimeras in which the VSV glycoprotein is substituted by the molecular analogs belonging to other enveloped viruses. The glycoprotein is responsible for binding and fusion of enveloped viral proteins to the cell, and thus chimeric viruses are an excellent tool to study virus entry into the cell³⁷. We infected SVG-A cells with VSV chimeras expressing the glycoproteins specific to VSV (indicated as VSV-G), Rabies, Ebola, Lassa, and LCMV. Cells were imaged before budding of viruses could occur, ensuring that cells were only infected by the initial one-hour virus incubation. As shown in Fig. 1d, TRPM7-depleted SVG-A cells (M7KD) were protected to varying extent against infections by most VSV chimeras. We quantified these infections in terms of the infection stages described earlier (Fig. 1f). The TRPM7-depleted cells were only mildly resistant to VSV-G and had no resistance to VSV-Rabies infection but were significantly more resistant to VSV-Ebola, VSV-Lassa and VSV-LCMV infections. This inhibition was seen across two multiplicities of infection (MOIs) of the virus (Supplementary Fig. 2). Since these VSV chimeras differ only in the molecular components needed for membrane fusion, these effects are not due to differences in the replication and translation of the virus genomes, which would present as cells only being nuclei+. Loss of TRPM6, a close homolog of TRPM7, had no effect on virus infection in SVG-A cells (Fig. 1d, f).

TPC channels, host factors for Ebola infection, do not regulate infection by several other VSV chimeras

TPC1 and 2 are endosomal cation channels known to be host factors for Ebola, Middle East Respiratory Syndrome coronavirus (MERS), and SARS-CoV-2 infection^38,39,40. To assess the selectivity of the TRPM7 effect, we tested whether TPC channels also regulated enveloped virus infections in a manner similar to TRPM7. As expected, the loss of TPC1 or TPC2 completely abrogated the infection by VSV-Ebola (Fig. 1e, g). However, the loss of TPC1 only had a modest effect on infection by other VSV-chimeras tested, and the loss of TPC2 had no significant effect on LCMV and only a modest effect on Lassa. These differences suggest that TRPM7 and TPC channels execute different functions during enveloped virus infection.

Loss of TRPM7 decreases plaque formation

To supplement the imaging-based infection assays, we performed virus plaque assays (Supplementary Fig. 1c). Unlike the imaging assay, which utilizes a higher multiplicity of infection (MOI) of virus and assesses the first round of viral infection, plaque assays integrate infections resulting from many rounds of virus replication. Control (Ctl) and TRPM7-knockdown (M7KD) SVG-A cells were infected with VSV-Lassa (MOI = 0.01) and incubated for 48 hours prior to harvesting for plaque assay. Plaque assays of VSV-G, VSV-Lassa, and VSV-LCMV in Vero cells show that depletion of TRPM7 by siRNA significantly decreases the infection by these three viruses. Note that due to the exponential expansion of the virus during the 48 h infection period, smaller differences, like the ~20% decrease seen in the imaging-based infection assay of VSV-G, are amplified further in the plaque assay.

TRPM7 knockout in HeLa cells prevents infection from VSV-Lassa and various pseudoviruses

Next, we used HeLa-crM7 cells in which TRPM7 is deleted by CRISPR-Cas9 technique. As expected, the HeLa-crM7 cells do not display I_M7 (Fig. 2a). The HeLa-crM7 cells are strongly resistant to VSV-Lassa (Fig. 2b), demonstrating a similar response to the SVG-A knockdown. To ensure that these observations were not an artifact of the VSV-chimera system, we infected HeLa cells with pseudoviruses produced using a lentivirus system⁴¹. These constructs are lentivirus particles, which express the glycoprotein of various enveloped viruses (Fig. 2c). These particles also encode luciferase in the viral genome, which is expressed in the host cell following successful infection. Control HeLa cells were readily infected with all pseudoviruses. HeLa-crM7 cells were infected with VSV and rabies pseudotyped viruses, but were resistant to infection by Ebola, Lassa, or LCMV constructs (Fig. 2d). These results are consistent with those obtained with VSV chimeras.

a Representative whole cell patch clamp recordings shown as current-voltage (IV) relationship of HeLa (gray) and HeLa-crM7 (blue) wherein Trpm7 was deleted by Crispr-cas9. Quantitative summaries of peak current densities at +100 mV are shown to the right (HeLa n = 13, HeLa-crM7 n = 8). P values determined by two-tailed Student’s t test. Median, 25th, and 75th quartiles are shown, as well as maximum and minimum. b (Left) Representative 20 μm maximum intensity projection of HeLa and HeLa-crM7 cells infected with VSV-G and VSV-Lassa. (Right) Quantification of nuclei+ and cytosol+ infected cells. P values determined by two-tailed Student’s t test of total positive cells. Each experiment is plotted as the average percent infected across multiple fields of view, n = 3. Mean and SEM are plotted. c Schematic of lenti-pseudovirus expressing various enveloped virus glycoproteins. Each pseudovirus genome encodes luciferase. d Relative luciferase luminescence of lenti-pseudovirus infected HeLa cells. Graph shows data points from three independent experiments with five well replicates of HeLa (gray) and HeLa-crM7 (blue). Data points were normalized to average control HeLa luciferase intensity per experiment. P values determined by two-tailed Student’s t test, n = 15. Mean and standard deviation are plotted. e (Left) Representative images of infection Trpm7^fl/fl and Trpm7^−/− BMDMs with VSV-Lassa. WGA outlines are shown. (Right) Quantification of average percent infected from n = 3 independent experiments. P values determined by two-tailed Student’s t test of total positive cells. Mean and SEM are plotted. f (Top) Schematic of ebola virus-like particle (VLP) expressing membrane-bound VP40 tagged with β-lactamase. (Bottom) Cryo-EM of ebola VLP. White arrows indicate one VLP. The cryo-EM image shown is representative of multiple VLPs observed in two independent sample preparations. g (Left) Spectrum of β-lactamase intensity of WT (gray) and TRPM7 KO (blue) BMDMs at 400 nm excitation. Mean and SEM are plotted. β-lactamase tagged VP40 from cytosolic-penetrated VLPs can cleave CCF-2 AM dye, decreasing FRET 520 nm signal and increasing 450 nm signal. (Right) Average ratio of 450 nm to 520 nm emission is plotted. Two-tailed Student’s t test was used to determine p values. Mean and standard errors are shown. n = 6.

Trpm7^−/− primary mouse macrophages are resistant to VSV-Lassa and Ebola virus-like particles

We then sought to extend these results to primary cells. Since Lassa and Ebola initially infect human macrophages during the course of their systemic infection⁴², we tested whether TRPM7 was necessary for such infection in murine bone marrow-derived macrophages (BMDMs). We found that VSV-Ebola does not infect murine BMDMs, but VSV-Lassa infects BMDMs with high efficiency. Trpm7^−/− BMDMs, derived from Trpm7^fl/fl LysM-Cre mice, were highly resistant to infection by VSV-Lassa (Fig. 2e). Lastly, we used Ebola virus-like particles (VLPs) which retain the shape and membrane composition of ebola in a non-replicative vector⁴³. This particular construct expresses β-Lactamase tagged to VP40, a protein lining the inner leaflet of the ebola membrane (Fig. 2f). Viruses that successfully escape the endosome release VP-40-β-lactamase into the cytosol, which can then cleave CCF2-AM dye. CCF2 consists of 2 fluorophores connected by a β-lactam ring that FRET pair when connected but are separated by β-lactamase. Therefore, the ratio of CCF2 that emits at 450/520 reflects the efficiency of virus escape into the cytosol. Bone marrow-derived macrophages were incubated with ebola VLPs for 3 hours and then loaded with CCF2-AM dye at room temperature before detecting emission. Compared to WT BMDMs, M7KO BMDMs exhibited a significant decrease in 450/520 nm emission, indicating a decrease in cytosolic viral penetration.

Overall, the results obtained using multiple approaches (siRNA, genetic deletion, imaging-based virus infection assays and plaque assays), different viral platforms (VSV chimeras, Lentivirus-based pseduviruses, and ebola VLPs), and different cell types (SVG-A, HeLa, primary BMDMs) reveal that TRPM7 is expendable for VSV and Rabies infection but has a crucial role in infection by other clinically relevant enveloped viruses (i.e. Lassa, LCMV, and Ebola).

TRPM7 promotes, but is not necessary for, the endocytosis of VSV chimeras

We first tested the hypothesis that TRPM7 regulates the endocytosis of the enveloped viruses. To test this, we used SVG-A cells that were gene-edited to incorporate fluorescent tags in endogenous alleles of proteins that mark early and late endosomes³⁷. In these cells, mScarlet was fused to the N-terminus of the early endosome marker EEA1 (mScarlet-EEA1), and Halo was fused to the C-terminus of the late endosome marker NPC1 (NPC1-Halo). Halo is a modified haloalkane dehalogenase that can be covalently conjugated with synthetic, cell-permeable fluorescent dyes. Halo was fluorescently labeled with JFX647 for 30 minutes before exposure to virus. We infected these gene-edited SVG-A cells with VSV-Lassa-MeGFP. The cells were imaged at 30-, 45- and 60-minutes post-infection and individual viruses were determined to localize to the plasma membrane, early endosome, late endosome, dual labeled compartments, or free in the cytosol. The number of endocytosed viral particles was estimated based on the average fluorescent intensity of a single viral particle and the point spread function of fluorescent spots in the cells (Supplementary Fig. 3a–d). In total, 3,791 viruses in control cells and 2,258 viruses in knockdown cells were analyzed for the VSV-Lassa-treated cells. The endocytosed viral particles co-localized with early endosomes (EEA1) and late endosome/lysosomes (NPC1) in both control and TRPM7-depleted cells (M7KD) (Fig. 3a). We found that compared to control cells, the cells depleted of TRPM7 contained a reduced number of VSV-Lassa particles (Fig. 3b), but these particles progressed from early to late endosomes in both cell populations (Fig. 3c and Supplementary Fig. 3e). While the rate of progression was slowed in cells depleted of TRPM7, by 60 minutes the same proportion of viruses had progressed to NPC1⁺ stage (Fig. 3d). We also evaluated VSV-G and VSV-LCMV colocalization at the 60-minute time point. Of the viruses endocytosed, a similar proportion of viruses localized to early and late endosomal compartments (Fig. 3e, f, Supplementary Fig. 3f). VSV-LCMV also trended towards a decrease in the number of viruses endocytosed. These results indicate that although TRPM7 promotes the endocytosis of VSV chimeras, it is not essential for this process, and the striking resistance to VSV-Lassa and VSV-LCMV infection is not adequately explained by this aspect of TRPM7 function.

a Representative images of VSV-Lassa-MeGFP infected Control (Ctl) and TRPM7 knockdown (M7KD) SVG-A cells. Early endosomes are tagged with EEA1-mScarlet (magenta), late endosomes/lysosomes with NPC1-Halo-JFX647 (blue). Cells were incubated with VSV-Lassa-MeGFP for 20 minutes, washed, and imaged at 30-, 45-, and 60-minutes post-infection. Maximum projection of 1.08 μm is shown. b Box and whisker plot of total number of virus particles detected per cell at each time point in Control (gray) and M7KD (blue) cells at 30-, 45-, and 60-minutes post-infection. Median, quartiles, maximum, and minimum are shown. P values determined by two-way ANOVA with Sidak multiple comparisons test. For control-treated cells at 30, 45, and 60 minutes, N = 10, 9, 10. For all M7KD timepoints, N = 10. Same cells are used for analysis in (c, d). c Average percent of viruses localizing to the cell surface (gray), early endosome (red, EEA1), late endosome (blue, NPC1), or free in cytosol (green) per cell at 30-, 45-, and 60-minutes post-infection. Some viruses localized to endosomes that were EEA1⁺ and NPC1⁺ (both, purple). Mean and SEM plotted. d Average percent of viruses per cell in early (red) and late (blue) endosomes. Control cells are represented by solid lines, while M7KD cells are represented with dashed lines. Mean and SEM are shown. P values determined by two-way ANOVA with Sidak multiple comparisons test. e (Left) Box and whisker plot of VSV-G particle uptake at 60 minutes post-infection in Control (gray) and M7KD (blue) cells. Median, quartiles, maximum, and minimum are shown. P values determined by two-tailed Student’s t test. (Right) Subcellular localization in Control and M7KD SVG-A cells at 60 minutes post-infection. n = 10 cells per condition. Mean and SEM plotted. f (Left) Box and whisker plot of total VSV-LCMV viral uptake at 60 minutes post-infection in Control (gray) and M7KD (blue) cells. Median, quartiles, maximum, and minimum are shown. P values determined by two-tailed Student’s t test. (Right) Subcellular localization in Control and M7KD SVG-A cells at 60 minutes post-infection. n = 10 cells per condition. Mean and SEM plotted.

TRPM7 is recruited to dense virus-laden endosomes

To determine if TRPM7 localizes to virus-containing endosomes, we used density gradient centrifugation to fractionate the intracellular vesicular compartments (Supplementary Fig. 3h). In this experiment, the MeGFP expressed by the VSV-Lassa-MeGFP enables the identification of fractions that contain the virus-laden endosomes. In uninfected cells, Rab5, a marker of early endosomes is found predominantly in the lighter fractions (fraction 3 and 4). However, in infected cells, the early endosomes that contain the viruses are dense and hence Rab5 is detected in the denser fractions 8 and 9. In uninfected SVGA cells, TRPM7 is seen in vesicular fractions of various densities. However, in VSV-Lassa-MeGFP infected cells, TRPM7 is found predominantly in the dense fractions (fractions 8 and 9) and depleted from less dense fractions. These data suggest that TRPM7 is a component of virus-laden endosomes.

TRPM7-deficient cells are particularly resistant to viruses that depend on low pH for endosomal fusion

The viral glycoproteins we tested undergo pH-dependent conformational changes to trigger membrane fusion. The envelope glycoproteins of VSV and Rabies are estimated to undergo a conformational change at a pH of ~6.2 and ~6.15, respectively^44,45. The Lassa and LCMV glycoproteins require a pH lower than 5.5 for viral escape^46,47,48. The Ebola glycoprotein requires proteolytic processing by acid-activated host cell cathepsin in the endolysosomal network^11,47,49. Since cathepsins are optimized for the low pH found in the lysosome, Ebola escapes in the later stages of the endolysosomal pathway⁵⁰. We noted that TRPM7-depleted cells exhibited significantly higher resistance to viruses that require low pH for membrane fusion. To test the hypothesis TRPM7 regulates the fusion of viruses that require a low endosomal pH, we focused on VSV-G. While VSV binds to its receptor at the plasma membrane and is endocytosed like other viruses, VSV does not need the receptor once internalized and only requires a modest decrease in endosomal pH to trigger fusion³. It has been shown that a substitution of a single amino acid (V269H) in the VSV glycoprotein allows the virus to get internalized into the host cell normally but requires a lower pH (37. We here refer to this VSV variant as VSV-pH. We tested VSV-G and VSV-pH for infection in control and TRPM7-depleted SVG-A cells. As seen previously, depletion of TRPM7 conferred a modest resistance to VSV-G, but the infection by VSV-pH was completely inhibited (Fig. 4a, b). Since VSV-G infects HeLa cells readily, we also tested VSV-pH infection in HeLa-crM7 cells wherein endogenous TRPM7 is deleted. Again, infection by VSV-pH was robustly inhibited in the absence of TRPM7 (Supplementary Fig. 4a). These results substantiated our hypothesis that TRPM7 activity is required for infection by enveloped viruses that depend on low endosomal pH for membrane fusion. This also indicates that changes in receptor availability or endocytosis are not the main factors of infection resistance in TRPM7-depleted cells. Loss of other endosomal ion channels, TPC1 and TPC2, did not significantly inhibit VSV-pH infection, nor did the loss of TRPM6 (Supplementary Fig. 4b), which is only expressed at very low levels.

a Representative maximum projections of Control (Ctl) and TRPM7 knockdown (M7KD) SVG-A cells infected by parental VSV-G or the VSV pH-sensitive mutant, VSV-V269H. Cells were fixed 6 hours post-infection. b Quantification of VSV-G and VSV-pH sensitive mutant infection. Percent of Nuclei+ cells are shown in light green and Cytosol+ cells in dark green. P values determined from two-tailed Student’s t test of total positive cells from three independent infection assays. Mean and SEM are plotted. c Schematic of pHrodo-tagged VSV-chimera viruses expressing MeGFP. d Representative images of MeGFP and pHrodo signal time course in Ctl and M7KD SVG-A cells. VSV-Lassa-MeGFP or VSV-LCMV-MeGFP is shown in green. pHrodo signal is shown pseudo-colored based on the intensity spectrum, with black/purple being the lower intensity and yellow being the highest intensity. Images are 2.43 μm maximum projections. e Single particle quantification of the viral endosome pH at 30-, 45-, and 60 minutes following virus incubation. Control (Ctl) cells are shown in gray and TRPM7 knockdown (KD) cells in blue. Each data point represents the average pH of all virus particles detected across a single cell. Violin plots display median (thick dotted line) as well as 25th and 75th quartiles (thin dotted lines). P values are determined by two-way ANOVA with Sidak multiple comparisons test. Inset graphs show mean pH per cell over time with SEM. At 30, 45, and 60 minutes VSV N = 15, 18, 14; Lassa N = 25, 22; 16; LCMV N = 17, 19, 11 for control cells and VSV N = 15, 15, 19; Lassa N = 24, 23, 20; LCMV N = 19, 20, 18 for M7KD cells.

TRPM7 is required for the acidification of virus-laden endosomes

To assess the acidification of virus-containing endosomes, we conjugated the envelope proteins of VSV-MeGFP with the pH-sensitive dye pHrodo, which reports increases in fluorescence intensity as pH decreases (Fig. 4c). While the pHrodo tag was exposed to the endosomal lumen, the eGFP-labeled M-protein located inside the envelope of the virus is protected from acid-induced quenching by endosomal acidification. This configuration enables ratiometric assessment of the change in endosomal pH. A decrease in endosomal pH is reported by an increase in the ratio of pHrodo:eGFP. We calibrated these measurements to absolute pH by imaging the virus particles alone in varying pH buffers (Supplementary Fig. 4c). Across time points and treatments, 4633 VSV-G, 9430 VSV-Lassa, and 1885 VSV-LCMV particles were analyzed, and the average particle pH was reported for each cell. As expected, the control SVG-A cells infected with pHrodo-labeled VSV-Lassa-MeGFP display a progressive decrease in pH over 60 minutes. In contrast, the TRPM7-depleted SVG-A cells (M7KD) display reduced acidification (Fig. 4d, left). A similar difference is also seen in cells infected with VSV-LCMV (Fig. 4d, right). A quantitative summary of these data shows that depletion of TRPM7 results in a substantial and significant defect in acidification of the virus-loaded endosomes, allowing a pH range of 6.8–6.5 at one hour while controls exhibit a pH range of 5.8–5.4 (Fig. 4e). Infection of HeLa and HeLa-crM7 cells showed a very similar inhibition of acidification (Supplementary Fig. 4d). Addition of 50 nM Bafilomycin, a potent inhibitor of the V-ATPase, completely abrogated endosomal acidification. We conclude from these data that TRPM7 is necessary for the acidification of virus-loaded endosomes.

TRPM7 does not regulate the acidification of all endosomes

If TRPM7 regulates the pH of all endosomal compartments, we would expect drastic effects on cell proliferation and survival, as nutrient uptake and autophagy are highly dependent on endosomal acidification. However, HeLa-crM7 cells grow and proliferate readily (Supplementary Fig. 5a). To test whether loss of TRPM7 induced global changes to the pH of highly acidified compartments, cells were loaded with Lysosensor, a dye that accumulates in acidic compartments and increases in fluorescence as pH decreases (Supplementary Fig. 5c). Compared to parental HeLa cells, there were no robust changes of fluorescence in HeLa-crM7 cells. Bafilomycin treatment led to a robust loss of Lysosensor fluorescence, indicating that intracellular compartments in HeLa-crM7 were substantially acidified. Lastly, we tested whether TRPM7 regulates the degradation of the epidermal growth factor receptor (EGFR). EGF binding to EGFR triggers the complex to be endocytosed and trafficked to lysosomes for pH-mediated degradation that can be inhibited by Bafilomycin⁵¹. Immunoblot of cell lysates treated with EGF demonstrates that loss of TRPM7 in SVG-A or HeLa cells does not impede EGFR degradation (Supplementary Fig. 5b). These results indicate that TRPM7 regulates the acidification of a subset of endosomes (including virus-containing endosomes), but not all endosomes.

The ion channel activity of TRPM7 is necessary for infection by pH-dependent VSV-pH

TRPM7 contains both a functional ion channel and a kinase domain. To determine which of these activities was salient for endosomal pH regulation, SVG-A cells were transfected with mouse TRPM7 variants wherein either the ion channel pore or the kinase activity is inactivated by site-directed mutagenesis (Fig. 5). The unmodified wild-type TRPM7 (TRPM7-wt), the pore mutant (TRPM7-pm) or the kinase inactivated variant (TRPM7-ki) were reported previously³⁴. Tetramerization of TRPM7 monomer is required to form a functional channel and thus, incorporation of TRPM7-pm monomers into the tetrameric complex renders the entire ion channel pore dysfunctional. Each of these variants was ectopically expressed in SVG-A cells and evaluated for infection by VSV-G and VSV-pH. The ectopic expression of TRPM7-pm in SVG-A cells had the expected dominant-negative effect on endogenous TRPM7 currents (I_M7) (Fig. 5b). Immunoblotting of membrane fractions for mouse TRPM7 demonstrated successful expression of all three variants. TRPM7-wt and TRPM7-ki had lower expression, most likely because high expression of the functional ion channel can lead to cell death and therefore favoring the survival of cells with moderate expression³⁴. Compared to cells transfected with TRPM7-wt, the TRPM7-pm transfected cells were modestly deficient in infection by VSV-G. However, TRPM7-pm expression resulted in complete abrogation of infection by the more pH-dependent VSV-pH virus (Fig. 5c, d) and inhibited the acidification of the virus-laden endosomes (Supplementary Fig. 6a). These results established that the ion channel activity of TRPM7 is required for endosomal acidification. In contrast, the cells ectopically expressing TRPM7-ki displayed only a modest resistance to infection by VSV-pH (Fig. 5c, d). The role of TRPM7 kinase activity cannot be ruled out due to differences in expression levels, but the available evidence indicates that it plays a modest role in the regulation of viral endosome acidification. Since TRPM7 channel activity is required for VSV-pH infection, we tested the effect of two structurally distinct small molecule TRPM7 channel inhibitors on VSV-pH infection. Based on previous reports, we evaluated FTY720 at 5 μM⁵² and NS8593 at 30 μM^53,54. At these concentrations, both compounds inhibited the native I_M7 in SVG-A cells (Supplementary Fig. 6b). Cells treated with either compound exhibit more resistance to VSV-pH than to VSV-G (Fig. 5e, f). Infections of TRPM7 knockdown cells in the presence of TRPM7 inhibitors reveal that FTY720 further inhibits the infection while NS8593 does not (Supplementary Fig. 6c). These data argue that NS8593 is a more selective inhibitor of TRPM7 and that other molecular targets of NS8593 are not factors in virus infection of SVG-A cells. These findings establish a salient function for TRPM7 ion channel activity in the acidification of virus-laden endosomes and in the infection by pH-dependent viruses.

a Diagram of TRPM7 domains with annotations of pore mutant (pm) and kinase-inactive (ki) mutations. b (Left) Whole-cell patch-clamp recordings shown as current-voltage (IV) relationship of mouse TRPM7 constructs expressing SVG-A cells. The IV of non-transfected (black), wild-type (TRPM7-wt, dark gray), pore mutant (TRPM7-pm, blue), and kinase-inactive (TRPM7-ki, orange) transfected SVG-A cells are overlayed. (Right) Immunoblot of membrane fraction of non-transfected SVG-A cells (ctl) and cells transfected with mouse TRPM7-wt, -pm, and -ki constructs. Immunoblotting is done with an anti-mouse TRPM7 antibody which does not detect human TRPM7. A non-specific band seen at 45 kDa is used as a loading control. The immunoblot results were reproduced in two independent experiments. c Representative maximum projection (20 μm) images of VSV-G and VSV-pH infections (green) in SVG-A cells ectopically expressing the TRPM7 variants: wild-type (TRPM7-wt), pore mutant (TRPM7-pm), or kinase-inactive (TRPM7-ki). d Quantification of infection in SVG-A cells ectopically expressing TRPM7-wt, TRPM7-pm, and TRPM7-ki variants. Cells were infected with either parental VSV-G or VSV-pH. Percent cytosolic+ (dark green) and nuclei+ (light green) cells are shown. Averages are derived from three independent infections. P values were determined from total cells infected by one-way ANOVA with Tukey’s multiple comparison test. Mean and SEM are plotted. e Representative maximum projection (20 μm) images of VSV-G and VSV-pH infections (green) in SVG-A cells treated with TRPM7 inhibitors FTY720 and NS8593 at 5 µM or 30 µM respectively. The inhibitors were added to the cells 15 minutes prior to infection and are present throughout infection. f Quantification of VSV-G and VSV-pH infection of SVG-A cells in the presence of TRPM7 inhibitors FTY720 (5 μM) and NS8593 (30 μM). Percent cytosolic+ (dark green) and nuclei+ (light green) are shown. Averages are derived from three independent infections. P values were determined from total cells infected by one-way ANOVA with Tukey’s multiple comparison test. Mean and SEM are plotted.

Endosomal proton-pumping cannot be sustained without TRPM7

A substantial pool of TRPM7 is located intracellularly in vesicular compartments²⁵. We reasoned that intracellular TRPM7 may have an important role in early and late endosomes as an ion channel mediating the cationic countercurrent to the proton-pumping activity of V-ATPases. To test this hypothesis, we used an optogenetic tool that can manipulate the intraluminal pH on command. Arch3-CD63-Halo (A3CH), a light-driven proton pump generated by fusing Archaerhodopsin 3 (Arch3) with the late endosome/lysosome protein CD63 (known as lysophoenix)⁵⁵, was modified to contain a Halo-tag and was stably expressed in SVG-A cells. A3CH co-localized to endosomes containing VSV-pH (Supplementary Fig. 7b). When V-ATPase is inhibited by Bafilomycin, endosomes fail to acidify, and the cells are resistant to infection by viruses that require an acidified endosome for membrane fusion and infection. However, Arch3 is not sensitive to Bafilomycin. Therefore, exposure of cells to light activates the Arch3 pump to acidify the endosome and allows viral fusion even in the presence of Bafilomycin (Fig. 6a). We demonstrate that blue light-mediated activation of A3CH can compensate for Bafilomycin-mediated inhibition of V-ATPase in control cells, thereby enabling infection by VSV-G, which requires mild acidification to a pH of approximately 6.2⁴⁴ (Fig. 6b). Similarly, activation of A3CH in Bafilomycin treated TRPM7 knockdown cells is also sufficient to restore VSV-G infection. A quantitative summary of these data is shown (Fig. 6b, right panel). However, infection by VSV-pH requires a significantly lower pH (37). As expected, VSV-pH infection is abrogated by Bafilomycin, and it can be restored by activation of A3CH in control cells (Fig. 6c). However, VSV-pH infection cannot be restored by optogenetic acidification in TRPM7-depleted cells at the same dose of light stimulation, arguing that TRPM7 is intrinsically necessary to reach the optimal endosomal pH for VSV-pH infection. A calibration curve of light exposure to VSV-G and VSV-pH treated control and M7KD cells shows that sustained light stimulation in M7KD cells can restore a small amount of virus infection in VSV-pH treated cells (Supplementary Fig. 7c). This is likely because the photonic energy provided by prolonged stimulation is sufficient to overcome the energy barrier of the unfavorable electrical gradient that is created by the accumulation of endosomal protons. Light stimulation of cells expressing the Arch3 pump inactive mutation (D95N) does not restore VSV-pH infection demonstrating that this effect requires proton-pumping activity and is not a result of membrane damage inflicted by photons (Supplementary Fig. 7d).

a Diagram showing the use of the optogenetic tool Arch3, a light-activated proton pump that can be used to acidify endosomes in the presence of Bafilomycin. Arch3 fused to CD63 is targeted to late endosome/lysosomes. Normally, the endogenous V-ATPase pumps protons to acidify the endosomal lumen and this enables the virus to escape the endosome (left). When cells are treated with V-ATPase inhibitor Bafilomycin (50 nM), the endosome fails to acidify, and the virus stays in the endosomes (middle). But even in the presence of Bafilomycin, light-mediated activation of Arch3 acidifies the endosome and enables the virus to enter the cytosol (right). b Maximum projections of Arch3-CD63 expressing, Ctl (gray) and TRPM7 knockdown (M7KD, blue) SVG-A cells infected with VSV-G. Cells were treated with Bafilomycin (50 nM) and/or exposed to 15 µWatt 488 nm laser (60 seconds). Quantification of cells positive for infection (green) from three independent infections is shown in the right panel. P values determined by one-way ANOVA with Tukey’s multiple comparison analysis. Mean and SEM is plotted. c Maximum projections of Arch3-CD63 expressing Ctl (gray) and M7KD (blue) SVG-A cells infected with VSV-pH mutant with or without Bafilomycin (50 nM) or light stimulation (15 µWatt 488 nm, 60 seconds). Quantification of total cells positive for infection (green) from three independent infections is shown in the right panel. P values determined by one-way ANOVA with Tukey’s multiple comparison analysis. Mean and SEM is plotted.

TRPM7 ion channel activity is necessary for infection by SARS-CoV-2 and other coronaviruses

Next, we investigated whether TRPM7 regulates infection by VSV-SARS-CoV-2 variants expressing the SARS-CoV-2 Spike glycoprotein from the Wuhan, Delta, and Omicron strains of SARS-CoV-2. All chimeras also include a genetically encoded eGFP that is transcribed upon infection (Fig. 7a). Previous studies have shown that SARS-CoV-2 enters the cells via endocytosis and requires proteolytic processing in acidified endosomes for escape into the cytosol^56,57,58,59. The Spike glycoprotein of SARS-CoV-2 requires cleavage by low pH-activated cathepsins to reach its fusion competent conformation, except in TMPRSS2-expressing cells. The protease TMPRSS2 cleaves the spike protein earlier in the endocytic pathway and permits virus fusion at a less acidic pH than that necessary for cathepsin activation^56,58,60. Vero cells which express endogenous ACE2 were readily infected by VSV-SARS-CoV-2 variants with or without the ectopic expression of TMPRSS2. Depletion of Trpm7 by siRNA inhibited VSV-SARS-CoV-2 infection by all variants in both parental Vero cells as well as Vero cells over-expressing TMPRSS2 (Fig. 7b–d). Expression of TMPRSS2 did slightly increase the number of cells infected by the Wuhan, Delta, and Omicron variants in the absence of TRPM7, likely due to the change in pH dependence of virus fusion through TMPRSS2-cleavage of the spike protein. To check if VSV-SARS-CoV-2 is endocytosed, we used a construct of VSV-SARS-CoV-2 (Wuhan spike) with an eGFP tag on the P protein of VSV and imaged them in SVG-A cells expressing endogenously tagged EEA1 and NPC1 and ectopically expressing ACE2. The fluorescent viral particles were found co-localized with EEA1 and NPC1 in these cells with or without TRPM7 (Supplementary Fig. 8d). These data indicates that, like VSV-Lassa in Fig. 3c, loss of TRPM7 permits endocytosis of VSV-SARS-CoV-2 but still suppresses infection, likely by preventing the acidification of the virus-laden endosomes. This model is supported by the observation that pharmacologic inhibition of TRPM7 ion channel activity is sufficient to prevent infection by VSV-SARS-CoV-2. We subjected Vero and Vero-TMPRSS2 cells to VSV-SARS-CoV-2 in the presence of TRPM7 inhibitors FTY720 and NS8593. Both inhibitors blocked infection (Figs. 7e, f). Like TRPM7-depleted cells, expression of TMPRSS2 enabled a moderate rescue of the infection in cells treated with NS8593. FTY720, a sphingosine analog, can be phosphorylated by some cells to yield FTY720-phosphate which is a potent agonist of S1P receptor, with EC₅₀ less than 10 nM^61,62. The ability to block SARS-CoV-2 infection however requires considerably higher concentration of FTY720 ( ~ 1.5 μM) (Supplementary Fig. 8e).

a Schematic of VSV-coronavirus chimera construct with genetically encoded eGFP that is transcribed upon infection. Different variant-specific spike proteins were expressed as the envelope glycoprotein. b Maximum projection of cells infected with VSV-SARS-CoV-2-eGFP variants (green). Vero cells +/- TMPRSS2 were transfected with Control (Ctl) and TRPM7 siRNA (M7KD). Cells were then infected with VSV-SARS-CoV-2-eGFP expressing the Spike protein of the Wuhan, Delta, or Omicron variants. Cells were incubated with virus for 8 hours before fixation and imaging. The scale bar at the bottom left of the images is representative of 50 µm for all images across each row. c Quantification VSV-SARS-CoV-2-eGFP variant infections in control (gray) and M7KD (blue) Vero cells. Cells were infected with VSV pseudotyped with Wuhan, Delta, and Omicron spike variants. P values determined by two-tailed Student’s t test. n = 3 independent experiments. Mean and SEM plotted. d Quantification of infection in Vero cells stably overexpressing TMPRSS2 with VSV-SARS-CoV-2-eGFP variants in control (gray) and M7KD (blue) cells. P values determined by two-tailed Student’s t test. n = 3 independent experiments. Mean and SEM is plotted. e Maximum projection (20 μm) of Vero and Vero-TMPRSS2 cells infected with VSV-SARS-CoV-2-eGFP Wuhan strain in the presence of 5 μM FTY720 or 30 μM NS8593. Cells were pretreated with FTY720 and NS8593 for 15 minutes before exposure to virus and throughout the infection assay. f Quantification of VSV-SARS-CoV-2-eGFP infection in Vero and Vero-TMPRSS2 cells in the presence of 5 μM FTY720 (purple) or 30 μM NS8593 (teal) or without drug (gray). P values were determined by one-way ANOVA with Tukey’s multiple comparison test. n = 3 independent experiments. Mean and SEM plotted. g Quantification of infection by VSV-MERS, VSV-SARS-CoV-1, and VSV-SARS-CoV-2 Wuhan variant in Control (Ctl), TRPM7 (M7KD), TPC1 (TPC1KD), and TPC2 (TPC2KD) knockdown Vero cells. Infection assays were performed in parental Vero cells and Vero cells stably overexpressing TMPRSS2. P values determined by one-way ANOVA with Tukey’s multiple comparison test. n = 3 independent experiments. Mean and SEM plotted.

The endosomal ion channel TPC2 has previously been shown to inhibit infection by several coronaviruses: the MERS virus and SARS-CoV-2^39,40. We directly compared the effect of depleting TRPM7, TPC1, and TPC2 on VSV-eGFP chimeras of MERS, SARS-CoV-1, and SARS-CoV-2 infection in Vero cells, with or without the ectopic expression of TMPRSS2 (Fig. 7g). Loss of TRPM7, TPC1, or TPC2 led to a complete inhibition of infection in parental Vero cells infected with either VSV-MERS, VSV-SARS-CoV-1, or VSV-SARS-CoV-2. However, in Vero-TMPRSS2 cells, we observed differences. In TPC1 and TPC2 depleted cells, presence of TMPRSS2 led to full recovery of VSV-MERS and VSV-SARS-CoV-2 infection and a mild recovery of VSV-SARS-CoV-1 infection. Conversely, presence of TMPRSS2 in TRPM7-depleted cells only led to a slight recovery of infection across VSV-MERS, -SARS-CoV-1, and -SARS-CoV-2 (

Targeting the ion channel activity of TRPM7 inhibits influenza infection

To extend our findings to a live, unmodified enveloped virus, we assessed the role of TRPM7 in infection by the Influenza strain A/Puerto Rico/8/1934. The fusion of Influenza to host membranes requires a pH of ~5.8–5.0 for viral entry, depending on the isoform of Hemagglutinin, the influenza fusion protein. A/Puerto Rico/8/1934 fuses at a pH of ~5.1⁶³. In plaque assays of influenza infection, we see a robust inhibition of plaque formation in M7KD Vero cells (Fig. 8a). We then assessed Influenza entry in control (Ctl) and M7-depleted (M7KD) Vero cells by staining the cells for the accumulation of Influenza A Nucleoprotein (NP), a viral structural protein that is highly expressed in infected cells⁶⁴. The M7KD Vero cells were highly resistant to infection by both trypsin-cleaved and non-trypsin cleaved influenza (Fig. 8b, c). As trypsin cleavage increases the efficiency of influenza entry, less virus is needed to achieve infection. Notably in cells treated with influenza that was not treated with trypsin, virus particles can be seen accumulating in M7KD intracellular compartments, consistent with the observation that TRPM7 is necessary for efficient viral egress from the endosome. FTY720 and NS8593 also inhibited Influenza infection of Vero cells (Supplementary Fig. 9). The virus particles can also be seen in the intracellular compartments of Vero cells treated with these inhibitors during infection. Finally, we tested the effect of NS8593 in a mouse model of Influenza A infection. Mice were inoculated intranasally with PBS (40 µL) containing 250 µM NS8593 or DMSO along with 150 PFUs of Influenza (A/Puerto Rico/8/1934). After infection, the mice were monitored and weighed for the following 14 days as weight loss is representative of the severity of influenza infection in mice⁶⁵. NS8593 treatment prevented significant weight loss (Fig. 8e). The onset of weight loss was also delayed in NS8593 treated mice, suggesting that the drug inhibited the early stages of infection. To assess the overall pathology in the lung tissue, we checked the degree of immune infiltration. Three mice from each cohort were used for histological analysis of the lungs by hematoxylin and eosin stain as well as the Masson’s Trichrome strain. Lungs of the NS8593 treated mice showed a trend towards less immune infiltration, correlating with the decreased illness exhibited by NS8593 treated mice (Fig. 8f).

a Plaque assay of Influenza A/Puerto Rico/8/1934 at two different dilutions. Viruses were propagated in Control (Ctl, gray) or TRPM7 knockdown (M7KD, blue) cells, and plaque assays were performed with parental Vero cells. P values determined by two-tailed Student’s t test. n = 6 independent plaque assays. Mean and standard deviation are plotted. b (Left) First round trypsin-cleaved influenza infection in control and M7KD Vero cells stained for influenza nucleoprotein (NP, green). Individual cells are identified by DAPI staining (Magenta). (Right) Quantification of percent cells infected in Ctl and M7KD vero cells. P values determined by two-tailed Student’s t test. n = 3 independent experiments. Mean and SEM plotted. c (Left) Maximum projection of first round non-trypsin-cleaved influenza infection in control and M7KD Vero cells stained for influenza nucleoprotein. Inset shows viral nucleoprotein retained in intracellular compartments after 8-hour incubation. (Right) Quantification of infection in Ctl and M7KD Vero cells with p values determined by two-tailed Student’s t test. n = 3 independent experiments. Mean and SEM plotted. d (Left) Representative maximum projection images of non-trypsin cleaved influenza infection in Vero cells treated with 5 µM FTY720 or 30 µM NS8593. (Right) Quantification of infection in Vero cells treated with FTY720 or NS8593. P values determined by one-way ANOVA with Tukey’s multiple comparison test. n = 3 independent experiments. Mean and SEM plotted. e Weight of mice treated with 150 Plaque Forming Units (PFU) of Influenza A/Puerto Rico/8/1934. Control (DMSO) mice are shown in gray, and NS8593 treated mice in teal. Virus was nasally administered in 40 µL of PBS containing DMSO or 250 µM NS8593. P values determined by Repeated Measures two-way ANOVA with Sidak multiple comparisons test. f (Left) Histology of lungs from mice treated with Influenza +/− NS8593. Hematoxylin and eosin (H&E) stain as well as Mason’s trichrome stain are shown. (Right) Quantification of tissue area with immune infiltration from H&E staining in Control and NS8593 murine lungs. P values determined by two-tailed Student’s t test. n = 3 lungs from individual mice. Mean and standard deviation are plotted.