Human cytomegalovirus-induced host protein citrullination is crucial for viral replication

Abstract

Citrullination is the conversion of arginine-to-citrulline by protein arginine deiminases (PADs), whose dysregulation is implicated in the pathogenesis of various types of cancers and autoimmune diseases. Consistent with the ability of human cytomegalovirus (HCMV) to induce post-translational modifications of cellular proteins to gain a survival advantage, we show that HCMV infection of primary human fibroblasts triggers PAD-mediated citrullination of several host proteins, and that this activity promotes viral fitness. Citrullinome analysis reveals significant changes in deimination levels of both cellular and viral proteins, with interferon (IFN)-inducible protein IFIT1 being among the most heavily deiminated one. As genetic depletion of IFIT1 strongly enhances HCMV growth, and in vitro IFIT1 citrullination impairs its ability to bind to 5'-ppp-RNA, we propose that viral-induced IFIT1 citrullination is a mechanism of HCMV evasion from host antiviral resistance. Overall, our findings point to a crucial role of citrullination in subverting cellular responses to viral infection.

Fig. 1. HCMV infection induces protein citrullination.

a Protein lysates from HFFs infected with HCMV strain Merlin (HCMV) (MOI 1 PFU/cell) at 48 and 96 h post infection (hpi) or from uninfected HFFs (mock) were exposed to an Rh–PG citrulline-specific probe (left panel) and subjected to gel electrophoresis to detect citrullinated proteins. Equal loading was assessed by Coomassie blue staining (right panel). One representative gel of three independent experiments is shown. b mRNA expression levels of PADI isoforms by RT-qPCR of HCMV-infected (24 hpi) vs. uninfected (mock) HFFs were normalized to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase GAPDH and expressed as mean fold change ± SEM over mock-infected cells (n = 3 independent determinations; PADI2, PADI4, and PADI6 P < 0.001, two-way ANOVA followed by Bonferroni’s post test). c PADI2 and PADI4 mRNA levels in HCMV-infected HFFs at the indicated time points (hpi) were normalized to GAPDH mRNA and expressed as mean fold change ± SEM over mock-infected cells (n = 3 independent determinations; PADI2: mock vs. 8 hpi P < 0.001, mock vs. 16 hpi P < 0.001, mock vs. 24 hpi P < 0.001, mock vs. 32 hpi P < 0.001; PADI4: mock vs. 8 hpi P = 0.00375, mock vs. 16 hpi P < 0.001, mock vs. 24 hpi P < 0.001, mock vs. 32 hpi P = 0.04, one-way ANOVA followed by Bonferroni’s post tests). d Western blot analysis of protein lysates from uninfected (mock) or infected HFFs using antibodies against PAD2, PAD4, UL44, or α-tubulin (α-TUB). One representative blot and densitometric analysis shown of three independent experiments. Values are expressed as fold change in PAD2 and PAD4 expression normalized to α-tubulin. e Western blot analysis of protein lysates from uninfected (mock) or infected HFFs with HCMV wild-type (HCMV), or UV-inactivated HCMV (UV) (left panel), treated with 150 μg/ml CHX (left panel) at 24 hpi, or with 250 µM PFA (right panel) at 72 hpi or left untreated. Analysis was performed using antibodies against PAD2, IEA (recognizing IE1-72- and IE2-86 kDa), pp28, or ACTIN. One representative blot of three independent experiments is shown. f Western blot analysis of protein lysates from uninfected (mock), infected HFFs for 48 h with AdVIE1, AdVIE2, or AdVLacZ (MOI = 10) using antibodies against IEA (recognizing IE1-72- and IE2-86 kDa), PAD2, or ACTIN. One representative blot of three independent experiments is shown. g Western blot analysis of protein lysates from HFFs infected with wild-type AD169 HCMV (AD169) or AD169ΔIE1 (MOI 1 PFU/cell), the latter complemented with AdVLacZ or AdVIE1 (MOI 10 PFU/cell), at 24 hpi or from uninfected HFFs (mock) using antibodies against IEA (recognizing IE1-72- and IE2-86 kDa), PAD2, or ACTIN. One representative blot of three independent experiments is shown. Data are shown as the mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 2. Cl-A blocks HCMV replication.

a HFFs were infected with HCMV (MOI 0.1 PFU/cell) and then treated with increasing concentrations of Cl-A, which were given 1 h prior to virus adsorption and kept throughout the whole experiment. At 144 hpi, viral plaques were microscopically counted and expressed as a percentage relative to untreated controls. The number of plaques is plotted as a function of Cl-A concentration. The concentrations resulting in 50% plaque formation (IC₅₀) reduction are represented by the red line. The number of viable cells was determined for each Cl-A concentration using the MTT method (black line). Values are expressed as means ± SEM (error bars) of three independent experiments (untreated vs. 25 µM Cl-A P < 0.001, untreated vs. 50 µM Cl-A P < 0.001, untreated vs. 75 µM Cl-A P < 0.001, untreated vs. 100 µM Cl-A P < 0.001, untreated vs. 200 µM Cl-A P < 0.001, one-way ANOVA followed by Bonferroni’s post test). b To determine the number of viral DNA genomes in HCMV-infected HFFs, viral DNA was isolated at 144 hpi and analyzed by qPCR using primers amplifying a segment of the IE1 gene. GAPDH was used to normalize HCMV genome counts. Values are expressed as mean ± SEM of three independent experiments (vehicle vs. 25 µM Cl-A P < 0.001, vehicle vs. 50 µM Cl-A P < 0.001, vehicle vs. 75 µM Cl-A P < 0.001, vehicle vs. 100 µM Cl-A P < 0.001, vehicle vs. 200 µM Cl-A P < 0.001, one-way ANOVA followed by Bonferroni’s post test). c HFFs were infected with HCMV at increasing MOIs (0.01–1 PFU/cell) and then treated with 100 µM Cl-A or vehicle. Viral supernatants were collected at 144 hpi and analyzed by standard plaque assay. Values are expressed as mean ± SEM of three independent experiments (MOI 1: untreated vs. vehicle P < 0.001, untreated vs. Cl-A P < 0.001, two-way ANOVA followed by Bonferroni’s post test). d Protein lysates from uninfected (mock) or infected HFFs (48 and 96 hpi) with (+) or without (−) 100 µM Cl-A were subjected to immunoblotting using the anti-peptidylcitrulline F95 antibody to detect citrullinated proteins, anti-pp28 to assess HCMV infection, or anti-α-tubulin (α-TUB) to show equal loading. The densitometric analysis shown is representative of three independent experiments. Densitometry values of F95 are normalized to those of α-tubulin. e PAD enzymatic activity assay. Histone H3 was immobilized on a 96-well microtiter plate and incubated with protein lysates from HCMV-infected HFFs or uninfected (mock) at the indicated time points, in the presence (red line) or absence (untreated or vehicle alone, black and blue line, respectively) of Cl-A. The conversion of peptidylarginine to peptidylcitrulline was detected with an anti-H3 citrulline antibody. Detection of the bound antibodies was performed by ELISA. Values are expressed as means ± SEM (error bars) of three independent experiments (48 hpi: untreated vs. Cl-amidine P < 0.001, 72 hpi: untreated vs. Cl-A P < 0.001, 72 hpi: untreated vs. vehicle P < 0.05, two-way ANOVA followed by Bonferroni’s post tests). f Protein lysates from uninfected (mock) or infected HFFs (24, 48, or 72 hpi) at an MOI of 1 PFU/cell treated with or without Cl-A ( + ) or vehicle (−) were analyzed by immunoblotting for viral expression (IE1-72 kDa, UL44, and pp28) and normalized to α-tubulin (α-TUB). One representative blot of three independent experiments is shown. Data are shown as the mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 3. Effect of PAD2 and PAD4 gene knockout on HCMV replication.

a Knockout (KO) gene variants in HFFs for PAD2 (PAD2 KO) and PAD4 (PAD4 KO) were generated using CRISPR/Cas9 technology. The efficiency of PAD2 and PAD4 protein depletion at 48 hpi was assessed by immunoblotting using antibodies against PAD2, PAD4, or α-tubulin (α-TUB), for equal loading. An anti-UL44 antibody was used to verify HCMV infection. The western blot and relative densitometric analysis are representative of three independent experiments. Values are expressed as fold change in PAD2 and 4 expression normalized to α-tubulin (α-TUB). b HFFs KO cells were infected with HCMV at an MOI of 0.1 PFU/cell. Viral supernatants were collected at the indicated time points and analyzed by standard plaque assay. Values are expressed as means ± SEM of three independent experiments (WT vs. PAD2 KO P < 0.001, WT vs. PAD4 KO P < 0.001, one-way ANOVA followed by Bonferroni’s post test). c To determine the number of viral DNA genomes in HCMV-infected HFFs KO cells (MOI 0.1), viral DNA was isolated at 144 hpi and analyzed by qPCR with primers amplifying a segment of the IE1 gene. GAPDH was used to normalize HCMV genome counts. Values are expressed as mean ± SEM of three independent experiments (WT vs. PAD2 KO P < 0.001, WT vs. PAD4 KO P < 0.001, one-way ANOVA followed by Bonferroni’s post test). Data are shown as the mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 4. Characterization of the citrullinated proteome (citrullinome) of HCMV-infected cells.

a Volcano plot depicting the host (upper panel—circle) and viral (lower panel—triangle) citrullinated proteins of infected cells vs. mock-infected cells at 48 hpi (left panel) and 96 hpi (right panel). Cell lysates from uninfected (mock) or HCMV-infected HFFs (MOI 1) were exposed to a biotin-PG to isolate citrullinated proteins on streptavidin agarose. Bound proteins were then subjected to on-bead tryptic digestion and analyzed by LC-MS/MS—in the graph, every identified citrullinated protein corresponds to a dot. The x axis represents the ratio of citrullination between mock and infected cells at the indicated time points, while the y axis indicates the statistical significance. Both variables were plotted on a logarithmic scale (n = 3). b Volcano plot depicting the host citrullinated proteins in HCMV AD169- (left panel) and AD169ΔIE1- (right panel) vs. mock-infected HFFs at 48 hpi. Cell lysates from uninfected (mock) or HCMV-infected (MOI 1) HFFs were exposed to a biotin-PG to isolate citrullinated proteins on streptavidin agarose. Bound proteins were then subjected to on-bead tryptic digestion and analyzed by LC-MS/MS—in the graph, every identified citrullinated protein corresponds to a dot. The x axis represents the ratio of citrullination between mock and infected cells at the indicated time points, while the y axis indicates the statistical significance. Both variables were plotted on a logarithmic scale (n = 3).

Fig. 5. The antiviral role of IFIT1 and Mx1 against HCMV.

a HFFs were silenced for IFIT1 and Mx1 using specific siRNAs (siIFIT1, siMx1, respectively). As negative control cells were also similarly transfected with scrambled siRNA (siCTRL). At 24 hpt, cells were infected with HCMV at an MOI of 0.1 PFU/cell. Viral supernatants were collected at 144 hpi and analyzed by standard plaque assay. Values are expressed as means ± SEM of three independent experiments (siCTRL vs. siIFIT1 P = 0.0111, unpaired two-tailed t test; siCTRL vs. siMx1 P > 0.05, unpaired two-tailed t test). b HFFs were transduced with AdVIFIT1 or AdVLacZ at an MOI of 10 PFU/cell. Subsequently, cells were infected with HCMV at an MOI of 1. The extent of virus replication was measured at 144 hpi. Results are expressed as means ± SEM of three independent experiments (AdVLacZ vs. AdVIFIT1 P > 0.05, unpaired two-tailed t test). c Gel electrophoretic mobility shift assay of IFIT1 binding to PPP-7SKas RNA. 50 nM of in vitro transcribed 7SK 5’-ppp-RNA was incubated with only buffer (buffer), 5 µM of recombinant IFIT1, either untreated (UNT rIFIT1) or treated with PAD2 (PAD2’d rIFIT1) and loaded on a tris-glycine agarose gel. Data are representative of three experiments. d Long-chain DNA synthesis directed by UL54 in the presence or absence of purified UL44ΔC290 was assayed by measuring the incorporation of labeled [₃₂P]TTP with a poly(dA)-oligo(dT) primer template. DNA products were resolved on an alkaline agarose gel that was exposed to film and, for quantification, to a phosphorescence screen followed by scanning with a Typhoon scanner. The image shows products directed by UL54 alone, lane 1), or by UL54 in the presence of untreated (UNT) UL44ΔC290 (lane 2), in the presence of UL44ΔC290 treated with PAD2 or PAD4 (PAD2’d, PAD4’d; lanes 3 and 4, respectively), in the presence of UNT UL44ΔC290 but with 270 pM PAD2 (lane 5) or 370 pM PAD4 (lane 6) added to the DNA synthesis reaction, or in the presence of UL44 incubated in PAD reaction buffer, but without PADs (mock-PAD, lane 7). One representative gel of three independent experiments is shown. Data are shown as the mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001.