Cytomegalovirus-mediated activation of pyrimidine biosynthesis drives UDP–sugar synthesis to support viral protein glycosylation Stefanie Renee DeVitoa, Emilio Ortiz-Riañob, Luis Martínez-Sobridob, and Joshua Mungera,1 Departments of aBiochemistry and Biophysics and bMicrobiology and Immunology, University of Rochester Medical Center, Rochester, NY 14642

Human cytomegalovirus (HCMV) induces numerous changes to the host metabolic network that are critical for high-titer viral replication. We find that HCMV infection substantially induces de novo pyrimidine biosynthetic flux. This activation is important for HCMV replication because inhibition of pyrimidine biosynthetic enzymes substantially decreases the production of infectious virus, which can be rescued through medium supplementation with pyrimidine biosynthetic intermediates. Metabolomic analysis revealed that pyrimidine biosynthetic inhibition considerably reduces the levels of various UDP–sugar metabolites in HCMVinfected, but not mock-infected, cells. Further, UDP–sugar biosynthesis, which provides the sugar substrates required for glycosylation reactions, was found to be induced during HCMV infection. Pyrimidine biosynthetic inhibition also attenuated the glycosylation of the envelope glycoprotein B (gB). Both glycosylation of gB and viral growth were restored by medium supplementation with either UDP–sugar metabolites or pyrimidine precursors. These results indicate that HCMV drives de novo-synthesized pyrimidines to UDP–sugar biosynthesis to support virion protein glycosylation. The importance of this link between pyrimidine biosynthesis and UDP–sugars appears to be partially shared among diverse virus families, because UDP–sugar metabolites rescued the growth attenuation associated with pyrimidine biosynthetic inhibition during influenza A and vesicular stomatitis virus infection, but not murine hepatitis virus infection. In total, our results indicate that viruses can specifically modulate pyrimidine metabolic flux to provide the glycosyl subunits required for protein glycosylation and production of high titers of infectious progeny. cytomegalovirus

| pyrimidine | glycosylation | metabolism | UDP–sugar

We have previously demonstrated that HCMV infection is responsible for numerous changes to the host cell metabolic network (14, 15). These changes include induction of many branches of central carbon metabolism, including glycolysis and the tricarboxylic acid cycle (15). Additionally, HCMV infection results in an expansion of pyrimidine metabolite pools (14). De novo pyrimidine biosynthesis is the main source of pyrimidines during cellular replication, whereas the pyrimidine salvage pathway provides a smaller amount of pyrimidines to quiescent cells and cells in G0 (16, 17). The de novo pathway is primarily regulated through its rate-limiting enzyme, carbamoyl phosphate synthetase–aspartate transcarbamylase–dihydroorotase (CAD). CAD catalyzes the first three steps of the pathway, including the first committed step (18, 19). CAD is a ∼250-kDa protein that possesses three enzymatic activities and multimerizes in vivo (20, 21). CAD is heavily regulated by posttranslational modifications, which alter the sensitivity by which CAD is allosterically activated and inhibited and, in turn, induce or inhibit de novo pyrimidine biosynthesis, respectively (16, 22). De novo pyrimidine biosynthesis also provides pyrimidines for synthesis of UDP–sugars, which are widely used as substrates to feed cellular glycosylation reactions. The UDP–sugars, including UDP–glucose and UDP–N-acetyl-glucosamine (UDP–GlcNAc), along with GDP–mannose, are required for building the necessary precursor oligosaccharide structure that forms immediately before N-linked glycosylation. Additionally, UDP–GlcNAc, but not UDP–glucose, is required for the formation of O-linked glycosyl groups (23). The HCMV viral envelope contains a number of glycoproteins that are critical for HCMV replication Significance

A

variety of evolutionarily divergent viruses have been shown to activate specific metabolic activities upon infection (1–3). These virally induced metabolic activities can be targeted for antiviral therapy. The most common metabolic-based antivirals include those targeting divergent nucleotide metabolism and are used to treat hepatitis B virus, HIV, human cytomegalovirus (HCMV), and herpes simplex virus infections (4, 5). Increasing evidence has identified additional nonnucleotide metabolic activities that are both specifically induced by viral infection and important for viral replication (6–9). Despite the importance of these activities, in most cases, little is known about how viruses induce these activities or how they contribute to viral infection. HCMV, a member of the betaherpesvirus family, is a widespread pathogen that causes serious disease in immunosuppressed individuals, including cancer patients, transplant recipients, and AIDS patients (10). Additionally, congenital HCMV infection occurs in 1–2% of all live births (11) and can result in multiple system abnormalities, including central nervous system damage (12). HCMV is a double-stranded DNA virus that contains a ∼235-kb genome that encodes >200 ORFs. The genome is encapsulated in a protein capsid that is surrounded by a tegument protein layer. Collectively, this structure is enclosed in a phospholipid envelope, which contains a number of viral glycoproteins that mediate virus attachment and entry (13). www.pnas.org/cgi/doi/10.1073/pnas.1415864111

Viruses use the host cell to provide the energy and molecular subunits to assemble viral progeny. The progeny of a variety of viral families possess envelope glycoproteins that are essential for viral infection. The production of these functional glycoproteins requires an ample supply of UDP–sugar subunits that serve as the substrates for glycosylation reactions. Our results indicate that human cytomegalovirus induces a viral metabolic program that activates pyrimidine biosynthesis to drive UDP– sugar biosynthesis. This metabolic activation is important for viral protein glycosylation and high-titer viral replication. Further, our results suggest that this metabolic link between pyrimidine and UDP–sugar biosynthesis is shared between evolutionarily diverse viral families, which may provide novel avenues for antiviral therapeutic intervention. Author contributions: S.R.D., E.O.-R., L.M.-S., and J.M. designed research; S.R.D. and E.O.-R. performed research; S.R.D., E.O.-R., L.M.-S., and J.M. analyzed data; and S.R.D., E.O.-R., L.M.-S., and J.M. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1

To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1415864111/-/DCSupplemental.

PNAS Early Edition | 1 of 6

MICROBIOLOGY

Edited by Thomas Shenk, Princeton University, Princeton, NJ, and approved November 13, 2014 (received for review August 18, 2014)

Inhibition of Pyrimidine Biosynthesis Attenuates the Production of HCMV Viral Progeny. Given that HCMV induces pyrimidine bio-

Fig. 1. HCMV infection induces host de novo pyrimidine biosynthesis. (A) MRC5 cells were mock- or HCMV-infected (MOI = 3.0) and harvested at 48 hpi for LC-MS/MS–based measurement of UTP concentrations (mean ± SEM, n = 6). *P < 0.05. (B) Schematic of UTP labeling from U–13C-glucose. 13C, red; 12 C, black; N, blue; O, gray. Certain atoms were omitted for clarity. (C) MRC5 cells infected as in A were labeled at 48 hpi for 0, 1, 2, or 6 h with U–13Cglucose and analyzed by LC-MS/MS to measure the fraction of 12C–UTP present (12C–UTP/total UTP). Values are representative of four biological replicates. (D) UTP flux estimates based on the data in A and C were calculated as indicated in SI Materials and Methods (mean ± SEM, n = 4).

(24), but little is known about how HCMV infection may impact the cellular glycosylation machinery. Here, we show that de novo pyrimidine biosynthetic flux is induced upon HCMV infection and that inhibition of de novo pyrimidine biosynthesis reduces HCMV replication, indicating that induction of pyrimidine biosynthesis is necessary for hightiter viral replication. Further, we find that HCMV-infected cells require pyrimidine biosynthesis to maintain UDP–sugar pools and proper glycosylation of the gB virion protein. This link between pyrimidine biosynthesis and the UDP–sugars is important to HCMV infection, because UDP–GlcNAc or UDP–glucose supplementation rescues viral growth in the face of pyrimidine biosynthetic inhibition. Further, the importance of this metabolic link was also observed during influenza A and vesicular stomatitis virus (VSV) infection, suggesting that it could be a common feature of viral infection.

synthesis, we hypothesized that pyrimidine biosynthetic inhibition might attenuate viral replication. We chose to target de novo pyrimidine biosynthesis using N-phosphonacetyl- L aspartate (PALA), a CAD inhibitor (Fig. 2A), and Leflunomide, a dihydroorotate dehydrogenase (DHODH) inhibitor (Fig. 2A), which has been shown to have antiviral effects (26). Human fibroblasts were infected with HCMV [multiplicity of infection (MOI) = 3] and treated with DMSO or various concentrations of PALA (Fig. 2B) or Leflunomide (Fig. 2C). Samples were harvested at 96 hpi for quantification of viral progeny production. We found that both PALA (Fig. 2B) and Leflunomide (Fig. 2C) treatment caused a decrease in the production of infectious viral progeny. Both inhibitors also blocked the HCMV-induced expansion of UTP and CTP pools (Fig. S1C). Because Leflunomide has been reported to possess additional activities unrelated to DHODH inhibition (27), we wanted to test whether the associated viral attenuation resulted specifically from pyrimidine biosynthetic inhibition. Toward this end, we supplemented viral growth medium with uridine, which enables UMP formation through the salvage pathway downstream of the biosynthetic enzymes targeted (Fig. 2A). Addition of exogenous uridine fully rescued viral replication during PALA treatment, providing evidence that PALA inhibition is specific for pyrimidine biosynthesis (Fig. 2D). In contrast, the inhibitory effect of Leflunomide treatment was only partially rescued by uridine supplementation (Fig. 2E), indicating the possibility of offtarget effects. To test the toxicity of PALA and Leflunomide on human cells, a live/dead assay was performed on human fibroblasts treated with DMSO, either 100 μM PALA or 100 μM Leflunomide, or 1% ethanol as a control. This assay stains live cells green based on their esterase activity and dead cells red

Results HCMV Infection Induces Pyrimidine Biosynthetic Flux. To gain a more thorough understanding of how HCMV impacts pyrimidine biosynthesis, we used liquid chromatography and tandem mass spectrometry (LC-MS/MS) to measure the concentration of pyrimidine pools in mock or HCMV-infected human fibroblasts. HCMV infection induced a ∼2.5-fold increase in UTP concentration at 48 hours postinfection (hpi) (Fig. 1A). Before 48 hpi, at 6 and 24 hpi, there was no change in UTP concentrations upon infection (Fig. S1A). The HCMV-mediated induction of UTP levels appears to be dependent on viral gene expression, but not viral DNA replication, because infection with UV-irradiated HCMV or treatment with cycloheximide blocked induction of UTP pools, as opposed to treatment with phosphonoacetic acid (PAA), an inhibitor of viral DNA synthesis, which did not (Fig. S1B). Molecular flux through metabolic pathways can be estimated through the analysis of pool turn over after a pulse with an isotopically labeled metabolic tracer (15, 25). To determine how HCMV impacts pyrimidine biosynthetic flux, mock and HCMV-infected fibroblasts were pulsed with medium containing a U–13C–glucose isotope tracer, which resulted in labeling of the UTP ribose moiety (Fig. 1B). HCMV infection substantially induced the turnover of the unlabeled UTP pool upon 13 C-glucose labeling (Fig. 1C). Measurement of pool concentrations (Fig. 1A) and pool-labeling kinetics (Fig. 1C) enables estimation of UTP biosynthetic flux (25). HCMV infection induced an approximately fourfold increase in UTP metabolic flux at 48 hpi (Fig. 1D). These results indicate that HCMV infection strongly induces pyrimidine biosynthetic flux and increases UTP levels upon infection. 2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1415864111

Fig. 2. Impact of pyrimidine biosynthetic inhibition on HCMV replication. (A) Schematic of de novo pyrimidine biosynthesis. PRPS1, phosphoribosyl pyrophosphate synthetase 1; UMPS, UMP synthetase; UK, uridine kinase. (B and C) MRC5 cells were HCMV-infected (MOI = 3.0), treated with DMSO, PALA (B), or Leflunomide (C) at the indicated concentrations, and harvested at 96 hpi for analysis of viral progeny (mean ± SEM, n = 4). *P < 0.05. (D) Cells were infected as in B, treated with PALA (100 μM) and uridine at 0, 5, or 50 μM as indicated, and harvested at 96 hpi for analysis of viral progeny (mean ± SEM, n = 4). *P < 0.05. (E) Cells were infected as in B, treated with Leflunomide (100 μM) and uridine at 0, 0.5, or 1 mM as indicated, and harvested at 96 hpi for analysis of viral progeny production (mean ± SEM, n = 4). (F) MRC5 cells were treated with 100 μM PALA, 100 μM Leflunomide, DMSO, or 1% ethanol (+ control) for 24 h. The toxicity of PALA and Leflunomide was measured by using a live/dead assay, in which green fluorescence indicates the presence of esterase activity associated with viable cells, and red fluorescence indicates a loss of cellular membrane integrity associated with cell death.

DeVito et al.

based on nucleic acid staining that is dependent on membrane integrity breakdown. Neither PALA nor Leflunomide induced significant toxicity at the tested concentrations (Fig. 2F). CAD is a rate-controlling enzyme of pyrimidine biosynthesis (20, 21). We analyzed CAD protein levels and found that HCMV had little impact on CAD protein abundance (Fig. 3A). Phosphorylation of CAD at Thr-456 results in increased sensitivity to allosteric activators, and decreased sensitivity to allosteric inhibitors (19, 28–30). HCMV infection increased Thr-456 phosphorylation, most notably at 48 hpi (Fig. 3A), which is consistent with increased CAD activity and consequent activation of pyrimidine biosynthesis. To further test the impact of pyrimidine biosynthetic inhibition on HCMV replication, we targeted CAD mRNA expression with siRNA. Transfection with the CAD siRNA, but not with a nonspecific control siRNA, successfully reduced CAD protein levels (Fig. 3B). Transfection with CAD-specific siRNA, but not control siRNA, attenuated the production of infectious viral progeny to levels comparable to PALA treatment (Fig. 3C). These results indicate that pyrimidine biosynthesis and CAD expression are important for high-titer HCMV replication. To explore the impact of pyrimidine biosynthetic inhibition on the viral life cycle, we examined the accumulation of viral proteins that represent the major kinetic classes of viral gene expression: IE1 for immediate early genes; UL26 and UL44 for early genes; and pp28 as a true late gene (13). PALA treatment did not substantially impact the accumulation of IE1, UL26, UL44, or pp28 (Fig. 4A), indicating that PALA treatment did not dramatically impact the coordinated expression of HCMV genes. Analysis of viral DNA abundance indicated that PALA treatment decreased viral DNA abundance by ∼3.3-fold (Fig. 4B). Surprisingly, the addition of exogenous uridine to PALA-treated infected cells did not rescue the observed reduction in the viral DNA pool (Fig. 4B). Given that uridine supplementation fully rescues the growth defect associated with PALA treatment (Fig. 2D), this finding suggests that the reduced level of viral DNA replication that occurs with PALA treatment is still sufficient to enable the production of wild-type levels of viral progeny. This notion is supported by the results indicating that PALA treatment does not impact pp28 levels (Fig. 4A), because pp28 is a true late gene whose production is dependent on viral DNA synthesis. To investigate the impact of pyrimidine biosynthetic inhibition on the host-cell metabolic network, we used LC-MS/MS to examine the relative changes in metabolites between mock samples, HCMV-infected samples, and HCMV-infected samples treated with PALA. As expected, we found that a number of

Fig. 3. siRNA-mediated targeting of CAD attenuates HCMV replication. (A) MRC5 cells were mock- or HCMV-infected (MOI = 3.0) and harvested at 6, 24, and 48 hpi for Western analysis of CAD, CAD phospho–Thr-456 (P-CAD), or α-tubulin abundance (data are from a representative experiment of four). (B) Subconfluent MRC5 cells were transfected with a CAD-specific or control siRNA. At 24 h after transfection, cells were serum-starved for 24 h and HCMV-infected (MOI = 3.0), before Western analysis at 24 and 96 hpi for the indicated antibodies (data are from a representative experiment of two). (C) Cells transfected and infected as in B were harvested for viral titration at 96 hpi (mean ± SEM, n = 4). *P < 0.05.

DeVito et al.

MICROBIOLOGY

The Impact of Pyrimidine Biosynthetic Inhibition on HCMV Infection.

Fig. 4. The impact of pyrimidine biosynthetic inhibition on HCMV infection. MRC5 cells were HCMV-infected (MOI = 3.0) and treated with DMSO, PALA (100 μM), or 50 μM uridine as indicated. (A) Cells were harvested at 24 and 72 hpi for Western analysis of IE1, α-tubulin, UL26, UL44, or pp28 abundance. Data are from a representative experiment of four. (B) Viral DNA was extracted from cells at 72 hpi and analyzed by quantitative PCR (qPCR) for viral DNA abundance (mean ± SEM, n = 3). (C) Cells were harvested at 48 hpi for metabolite analysis by LC-MS/MS. LC-MS/MS–derived ion counts were normalized to mock-infected samples and plotted as a heat map relative to mock-infected cells (n = 6).

pyrimidine end products were substantially decreased by PALA treatment (Fig. 4C, black boxes). PALA, a CAD inhibitor, also completely abolished the HCMV-induced increases in N-carbamoyl-L-aspartate, the product of the CAD-catalyzed reaction (Fig. 4C, dashed black box). Additionally, our metabolomic analysis revealed that PALA treatment substantially reduced UDP–sugar levels—both UDP–glucose and UDP–GlcNAc (Fig. 4C, red boxes), which serve as the essential sugar donors for cellular glycosylation reactions. A more thorough analysis of these metabolites indicated that HCMV infection substantially expanded UDP–glucose and UDP–GlcNAc pools (Fig. 5 A and B, respectively). PALA treatment completely blocked these increases, reducing the size of these pools relative to mockinfected cells (Fig. 5 A and B). The UDP–sugar pools of mockinfected cells were completely resistant to PALA treatment, suggesting that the sensitivity of these pools to pyrimidine biosynthetic inhibition is HCMV-specific. Given the changes observed in UDP–sugar pools, we wanted to explore the impact of HCMV infection on UDP–sugar biosynthetic flux. Glucose carbon feeds UDP–sugar biosynthesis through both the UTP ribose and the conjugated sugar (Fig. 5C). To explore the impact of HCMV infection on UDP–sugar biosynthesis, we pulsed cells with 13C-glucose and measured UDP–sugar pool turnover to estimate metabolic flux. HCMV infection induced a modest increase in UDP–glucose flux and a larger increase in UDP– GlcNAc flux (Fig. 5 D and E, respectively). These data indicate that HCMV specifically targets UDP–sugar metabolism, inducing UDP–sugar biosynthesis and increasing UDP–sugar pool sizes, which, in contrast to mock-infected cells, are dependent on pyrimidine biosynthesis. PNAS Early Edition | 3 of 6

Fig. 5. HCMV increases UDP–sugar biosynthesis and pool sizes. (A and B) MRC5 cells were mock- or HCMV-infected (MOI = 3.0) and treated with DMSO or 100 μM PALA. At 48 hpi, cell were harvested and processed for LCMS/MS analysis to quantify UDP–glucose (A) or UDP–GlcNAc (B) (means ± SEM, n = 6). *P < 0.05. (C) Schematic of 13C–glucose labeling of UDP–sugars (13C–glucose, red). Some atoms were omitted for clarity. (D and E) MRC5 cells were infected as in A. At 48 hpi, cells were labeled with 13C–glucose for 0, 1, 2, or 6 h and processed for LC-MS/MS to measure the fraction of 12C–UDP– glucose (C) or 12C–UDP–GlcNAc (D). Flux estimates were calculated as indicated in SI Materials and Methods (mean ± SEM, n = 4).

CD44S, a glycosylated host protein, was not impacted by treatment with PALA for 72 h (Fig. 6D), whereas control treatment of these extracts with PNGase resulted in an electrophoretic mobility shift of CD44S, consistent with its deglycosylation (Fig. 6D). Although strong conclusions cannot be made after analysis of a single cellular protein, these results are suggestive that the link between pyrimidine biosynthetic activation and glycosylation may be specific for viral infection. It has been observed that inhibition of UDP–GlcNAc transferases upon tunicamycin treatment attenuates HCMV replication (33). Similarly, we found that tunicamycin treatment reduced the production of infectious virions by ∼100-fold (Fig. 6E), further highlighting the importance of glycosylation reactions for HCMV infection. Combined, our results indicate that HCMV infection requires pyrimidine biosynthesis to maintain high UDP–sugar levels and viral protein glycosylation, which are important for production of infectious viral progeny. If pyrimidine biosynthesis inhibits glycosylation through a decrease in UDP–sugar levels, medium supplementation with UDP–sugars could be predicted to restore gB glycosylation. We tested this possibility and found that supplementation with UDP–GlcNAc, but not UDP–glucose, restored gB glycosylation to non–PALA-treated levels at 72 hpi (Fig. 6F). Interestingly, medium supplementation with UDP–glucose did restore gB

Pyrimidine Biosynthetic Inhibition Attenuates Viral Protein Glycosylation.

UTP is required for the formation of UDP–glucose and UDP– GlcNAc, which are the essential subunits that drive glycosylation reactions. It is known that HCMV encodes many viral proteins that are glycosylated and are necessary for tegument and envelope formation, and host cell entry (13). This viral dependence on glycosylated proteins suggests that HCMV may rely on increased pyrimidine biosynthesis to provide the glycosyl groups necessary for protein glycosylation. To address this possibility, we examined the accumulation of the gB protein, a highly abundant, essential HCMV glycoprotein. The gB protein goes through several posttranslational modifications. The nascent protein (∼105 kDa) is first glycosylated, giving rise to an ∼170kDa product, before being transported to the Golgi, where it is cleaved to form the mature protein, which consists of two fragments, one of ∼116 kDa and the other 55 kDa (31, 32). The two cleavage products form a functional heterodimer (31, 32). The different stages of gB processing are distinguishable through differential SDS/PAGE migration (31, 32). We examined gB accumulation at 72 hpi in the presence of DMSO, PALA, and uridine and found that PALA treatment resulted in a decrease in the mature 116-kDa glycosylated form and an increase in the unglycosylated 105-kDa form (Fig. 6A). These PALA-induced changes were reversed with uridine medium supplementation (Fig. 6A). Quantification of the different gB isoforms indicated that the mature glycosylated isoform—i.e., the 116 kDa isoform—was substantially reduced relative to the total amount of gB upon PALA treatment, a decrease that was rescued with uridine supplementation (Fig. 6B). The total amount of gB was also reduced with PALA treatment (Fig. 6A), which likely reflects the well-established necessity of protein glycosylation to ensure the stability of freshly translated glycoproteins (23). Consistent with this notion, PALA treatment had no effect on the accumulation of gB mRNA levels, indicating that PALA does not impact gB transcription (Fig. 6C). The glycosylation of 4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1415864111

Fig. 6. Pyrimidine biosynthesis is important for glycosylation of gB. (A and B) MRC5 cells were HCMV-infected (MOI = 3.0), treated with DMSO, PALA (100 μM), or uridine (50 μM) as indicated, and harvested at 72 hpi. The abundances of gB and GAPDH were measured by Western analysis. The data are from a representative experiment of three. The quantification of percent mature gB from these experiments is shown in B (glycosylated 116 -kDa gB/total gB × 100; n = 3). (C) Cells were infected as in A, harvested at 24, 48, and 72 hpi, and processed for mRNA analysis by qPCR using gB-specific primers and normalization to GAPDH-specific primer amplification (mean ± SEM, n = 3). (D) MRC5 cells were infected as in A, treated with DMSO or 100 μM PALA, and harvested at 72 hpi. Samples in Right were treated after harvest with PNGase as indicated. The abundance of CD44S was measured by Western analysis. (E) MRC5 cells were infected with HCMV (MOI = 3.0), treated with DMSO or tunicamycin (10 μg/mL), and harvested at 96 hpi for assessment of viral titers (mean ± SE, n = 2 biological and 4 technical replicates). (F) MRC5 cells were HCMV-infected as in A and treated with PALA (100 μM) in the presence of uridine (50 μM), UDP–glucose (5 μM), or UDP–GlcNAc (100 μM) as indicated. Cells were harvested at 72 and 144 dpi and processed for Western analysis of gB and GAPDH abundance (n = 3). (G) Cells were HCMV-infected (MOI = 3.0), treated with DMSO, 100 μM PALA, UDP–glucose, or UDP–GlcNAc as indicated, and harvested at 96 hpi for the quantification of viral progeny production (mean ± SEM, n = 4). *P < 0.05. (H). Cells were HCMV-infected (MOI = 1.0) and treated with DMSO or 100 μM PALA. Supernatant virions were partially purified at 144 hpi and analyzed for associated viral genomes by real-time PCR or pfus by plaque assay. The pfu/genome ratio is indicated relative to wild-type infection (mean ± SEM).

DeVito et al.

The Metabolic Link Between Pyrimidine Biosynthesis and UDP–GlcNAc Is Important for the Replication of Diverse Viruses. Diverse envel-

oped viruses contain membrane-embedded glycoproteins that are critical for their respective infections. The link between pyrimidine biosynthesis and UDP–sugar production could therefore be important for the replication of evolutionarily disparate viruses. To explore this possibility, we tested whether a pyrimidine biosynthesis inhibitor, A3—which targets DHODH and has recently been shown to possess antiviral activity (34, 35)—could attenuate infection of diverse viral families and whether the attenuation could be rescued with medium supplementation of various pyrimidine metabolites or UDP–GlcNAc. Similar to PALA, A3 attenuated HCMV replication. HCMV replication in A3-treated cells could be rescued with the addition of uridine, UDP–GlcNAc, or orotate (Fig. 7A). A3 treatment also reduced influenza A growth, which could be rescued with medium supplemented with orotate or UDP–GlcNAc (Fig. 7B). Similarly, VSV infection was attenuated by A3 and was rescued with orotate or UDP–GlcNAc (Fig. 7C). In contrast, orotate, but not UDP–GlcNAc, rescued viral replication in A3-treated cells infected with mouse hepatitis virus (MHV). These results suggest that the importance of the link between pyrimidine biosynthesis and UDP–GlcNAc is maintained between divergent viral families including HCMV, a herpesvirus; influenza A, an orthomyxovirus; and VSV, a rhabdovirus; but not MHV, a coronavirus. Discussion Our results indicate that HCMV infection induces pyrimidine biosynthesis, which is required for induction of UDP–sugar pools, proper viral protein glycosylation, and high titer infection. This activation of pyrimidine biosynthesis is likely mediated in part by induction of CAD activity resulting from the phosphorylation of CAD at Thr-456 (Fig. 3), which has been shown to induce CAD activity (30). CAD phosphoregulation is complex. MAPK kinase activation has been shown to induce CAD Thr-456 phosphorylation (30), and the mTORC1–S6k complex has also been shown to activate CAD through Ser-1859 phosphorylation (36). In contrast, PKA-mediated CAD phosphorylation has been reported to inhibit CAD activity (28). HCMV is known to modulate these pathways (reviewed in refs. 37 and 38), raising the possibility that they play a role in HCMV-mediated CAD phosphorylation and activation and, consequently, HCMV-mediated induction of pyrimidine biosynthesis. Inhibition of pyrimidine biosynthesis lowered the levels of viral DNA accumulation. Surprisingly, uridine supplementation rescued viral growth without rescuing viral DNA abundance. These results suggest that exogenously added uridine does not have immediate access to purine deoxynucleotide pools. Further, given that uridine rescues infectious virion production, these results suggest that viral DNA is normally produced in substantial excess of what is required for the production of infectious DeVito et al.

Fig. 7. The link between pyrimidine and UDP–sugar metabolism is important for the replication of evolutionarily diverse viruses. (A) MRC5 cells were infected with HCMV (MOI = 3.0) and treated with DMSO, A3 (2.5 μM), orotate (2.5 mM), uridine (50 μM), or UDP–GlcNAc (100 μM) as indicated. Cells were harvested at 96 hpi for analysis of the production of viral progeny (mean ± SEM). (B) A549 cells were infected with influenza A/Puerto Rico/8/34 NS1–GFP (MOI = 0.005). Cells were treated with A3 (2.5 μM) in the presence of orotate (2.5 mM) or UDP–GlcNAc (100 μM) as indicated. At 48 hpi, the supernatants were collected for viral titration in MDCK cells (mean ± SEM). (C) A549 cells were infected with rVSV–GFP (MOI = 0.0001). Cells were treated with A3 (2.5 μM) in the presence of orotate (2.5 mM), or UDP–GlcNAc (100 μM) as indicated. At 48 hpi, the supernatants were collected for viral titration in Vero cells (mean ± SEM). (D) 293T–mCEACAM1 cells were infected with rMHV–GFP (MOI = 0.0005). Cells were treated with A3 (2.5 μM) in the presence of orotate (2.5 mM) or UDP–GlcNAc (100 μM) as indicated. At 48 hpi, the supernatants were collected for viral titration in L929 cells (mean ± SEM).

viral progeny. Our findings also indicate that pyrimidine biosynthetic inhibition dramatically reduced UDP–sugar pools in infected, but not in uninfected, cells. This virally specific reduction could reflect an increased utilization of UDP–sugars during infection that drains these pools in the absence of an increased pyrimidine supply. Alternatively, these results could indicate that viral infection is specifically funneling pyrimidine end products into UDP–sugar biosynthesis. If so, this finding would join a recent and growing body of literature indicating that viruses induce specific host-cell metabolic activities to facilitate infection (39–44). It was recently shown that HCMV infection induces the production of specific long-chain fatty acids for incorporation into the viral envelope, which was subsequently found to be important for high-titer HCMV replication (7). Results such as these suggest that viruses may require specific anabolic precursors and their associated synthetic enzymes for virion construction. The metabolic requirements for these specialized precursors may reflect advantageous biophysical properties that they convey to newly produced virions. To the extent that these metabolic activities are specific for viral infection, they may represent attractive targets for antiviral therapeutic intervention. An inherent advantage of targeting virally induced host-cell metabolic enzymes, as opposed to viral factors, would be the higher evolutionary barrier that viruses would have to overcome to acquire drug resistance. As a class, glycosylated envelope proteins negotiate the initial encounters of enveloped viruses with their hosts, including cellular binding and fusion. As such, glycosylated envelope proteins are critical determinants of tropism and pathogenicity (45). Virion glycan epitopes are also major determinants of immune recognition and have been found to be critical viral determinants that mediate escape from antibody neutralization (46). Despite their importance, the mechanisms through which viruses modulate the metabolic processes responsible for specific glycosyl linkages are not known. With further mechanistic elucidation of these processes, these areas may represent potential intervention points to alter the outcome of viral infection. Our data indicate that the metabolic links between pyrimidines and UDP–sugars are important to diverse viral families. These results highlight the possibility that the mechanisms through which PNAS Early Edition | 5 of 6

MICROBIOLOGY

glycosylation when examined later during infection, at 144 hpi (Fig. 6F). To determine whether UDP–sugar supplementation also restored viral growth in the presence of pyrimidine biosynthetic inhibition, we treated cells with DMSO or PALA in presence or absence of UDP–GlcNAc or UDP–glucose. As shown in Fig. 6G, supplementation with either UDP–GlcNAc or UDP–glucose rescued PALA-mediated inhibition of HCMV replication. To explore how inhibition of pyrimidine biosynthesis impacts viral particles, we partially purified secreted virions and assessed their infectivity. As expected, PALA treatment substantially reduced the secretion of infectious virus (Fig. S2A). The amount of genomes per pfu was substantially higher upon PALA treatment (Fig. 6H), indicative of an increase in noninfectious particles. PALA treatment reduced the secretion of viral particles containing genomes by ∼2.5-fold, a small change in comparison with the larger reduction in secreted pfu upon PALA treatment (>2-log) (Fig. S2). Combined, our results indicate that the link between pyrimidine biosynthesis and UDP–sugar production is important for production of infectious HCMV virions.

viruses target this link may prove to be a vulnerability shared by diverse viral types.

Flux Analysis. Flux estimations were largely performed as described (25, 48), with details indicated in SI Materials and Methods.

Materials and Methods

Statistical Analysis. All statistical analysis was performed with Origin (Version 8.0) as described in SI Materials and Methods.

Cell Culture and Biological Reagents. Cells and viruses were maintained under standard conditions specifically described in SI Materials and Methods.

LC-MS/MS Analysis. LC-MS/MS analysis was performed largely as reported (47), with details as described in SI Materials and Methods.

ACKNOWLEDGMENTS. This work was supported by National Institute of Allergy and Infectious Diseases Grant R01AI081773 (to J.M.). S.R.D. was supported by NIH Ruth L. Kirschstein National Research Service Award Training Grants GM068411 and AI049815-13. E.O.-R. is supported by a postdoctoral fellowship for Diversity and Academic Excellence from the University of Rochester Office for Faculty Development and Diversity. Research in the L.M.-S. laboratory is supported by NIH Grants R01 AI077719 and R03AI099681-01A1, National Institute of Allergy and Infectious Diseases Centers of Excellence for Influenza Research and Surveillance Grant HHSN266200700008C, and University of Rochester Center for Biodefense Immune Modeling Grant HHSN272201000055C.

1. Bradley PL (1957) Metabolism of pyruvate and alpha-ketoglutarate in virus-infected mouse brain. Nature 180(4599):1418–1419. 2. Kaplan AS, Ben-Porat T (1961) The action of 5-fluorouracil on the nucleic acid metabolism of pseudorabies virus-infected and noninfected rabbit kidney cells. Virology 13:78–92. 3. Pearson HE, Winzler RJ (1949) Oxidative and glycolytic metabolism of minced day-old mouse brain in relation to propagation of Theiler’s GD VII virus. J Biol Chem 181(2): 577–582. 4. Goodrich JM, et al. (1993) Ganciclovir prophylaxis to prevent cytomegalovirus disease after allogeneic marrow transplant. Ann Intern Med 118(3):173–178. 5. Lee WA, Martin JC (2006) Perspectives on the development of acyclic nucleotide analogs as antiviral drugs. Antiviral Res 71(2-3):254–259. 6. Heaton NS, Randall G (2010) Dengue virus-induced autophagy regulates lipid metabolism. Cell Host Microbe 8(5):422–432. 7. Koyuncu E, Purdy JG, Rabinowitz JD, Shenk T (2013) Saturated very long chain fatty acids are required for the production of infectious human cytomegalovirus progeny. PLoS Pathog 9(5):e1003333. 8. McArdle J, Moorman NJ, Munger J (2012) HCMV targets the metabolic stress response through activation of AMPK whose activity is important for viral replication. PLoS Pathog 8(1):e1002502. 9. Yamane D, et al. (2009) Inhibition of sphingosine kinase by bovine viral diarrhea virus NS3 is crucial for efficient viral replication and cytopathogenesis. J Biol Chem 284(20): 13648–13659. 10. Gerna G, Baldanti F, Revello MG (2004) Pathogenesis of human cytomegalovirus infection and cellular targets. Hum Immunol 65(5):381–386. 11. Andrei G, De Clercq E, Snoeck R (2008) Novel inhibitors of human CMV. Curr Opin Investig Drugs 9(2):132–145. 12. Burny W, Liesnard C, Donner C, Marchant A (2004) Epidemiology, pathogenesis and prevention of congenital cytomegalovirus infection. Expert Rev Anti Infect Ther 2(6): 881–894. 13. Mocarski ES, Shenk T, Pass RF (2006) Cytomegaloviruses. Fields Virology, ed Knipe DM (Lippincott-Williams and Wilkins, New York), pp 2701–2757. 14. Munger J, Bajad SU, Coller HA, Shenk T, Rabinowitz JD (2006) Dynamics of the cellular metabolome during human cytomegalovirus infection. PLoS Pathog 2(12):e132. 15. Munger J, et al. (2008) Systems-level metabolic flux profiling identifies fatty acid synthesis as a target for antiviral therapy. Nat Biotechnol 26(10):1179–1186. 16. Evans DR, Guy HI (2004) Mammalian pyrimidine biosynthesis: Fresh insights into an ancient pathway. J Biol Chem 279(32):33035–33038. 17. Fairbanks LD, Bofill M, Ruckemann K, Simmonds HA (1995) Importance of ribonucleotide availability to proliferating T-lymphocytes from healthy humans. Disproportionate expansion of pyrimidine pools and contrasting effects of de novo synthesis inhibitors. J Biol Chem 270(50):29682–29689. 18. Thoden JB, Holden HM, Wesenberg G, Raushel FM, Rayment I (1997) Structure of carbamoyl phosphate synthetase: A journey of 96 A from substrate to product. Biochemistry 36(21):6305–6316. 19. Huang M, Graves LM (2003) De novo synthesis of pyrimidine nucleotides; emerging interfaces with signal transduction pathways. Cell Mol Life Sci 60(2):321–336. 20. Zhang Y, et al. (2008) Model-based analysis of ChIP-seq (MACS). Genome Biol 9(9): R137. 21. Goecks J, Nekrutenko A, Taylor J, Galaxy T; Galaxy Team (2010) Galaxy: A comprehensive approach for supporting accessible, reproducible, and transparent computational research in the life sciences. Genome Biol 11(8):R86. 22. Sigoillot FD, Berkowski JA, Sigoillot SM, Kotsis DH, Guy HI (2003) Cell cycle-dependent regulation of pyrimidine biosynthesis. J Biol Chem 278(5):3403–3409. 23. Braakman I, Bulleid NJ (2011) Protein folding and modification in the mammalian endoplasmic reticulum. Annu Rev Biochem 80:71–99. 24. Yu D, Silva MC, Shenk T (2003) Functional map of human cytomegalovirus AD169 defined by global mutational analysis. Proc Natl Acad Sci USA 100(21):12396–12401. 25. Yuan J, Bennett BD, Rabinowitz JD (2008) Kinetic flux profiling for quantitation of cellular metabolic fluxes. Nat Protoc 3(8):1328–1340.

26. Waldman WJ, et al. (1999) Inhibition of cytomegalovirus in vitro and in vivo by the experimental immunosuppressive agent leflunomide. Intervirology 42(5-6):412–418. 27. Manna SK, Mukhopadhyay A, Aggarwal BB (2000) Leflunomide suppresses TNFinduced cellular responses: Effects on NF-kappa B, activator protein-1, c-Jun N-terminal protein kinase, and apoptosis. J Immunol 165(10):5962–5969. 28. Sigoillot FD, Evans DR, Guy HI (2002) Growth-dependent regulation of mammalian pyrimidine biosynthesis by the protein kinase A and MAPK signaling cascades. J Biol Chem 277(18):15745–15751. 29. Kotsis DH, et al. (2007) Protein kinase A phosphorylation of the multifunctional protein CAD antagonizes activation by the MAP kinase cascade. Mol Cell Biochem 301(1-2):69–81. 30. Graves LM, et al. (2000) Regulation of carbamoyl phosphate synthetase by MAP kinase. Nature 403(6767):328–332. 31. Britt WJ, Vugler LG (1989) Processing of the gp55-116 envelope glycoprotein complex (gB) of human cytomegalovirus. J Virol 63(1):403–410. 32. Britt WJ, Vugler LG (1992) Oligomerization of the human cytomegalovirus major envelope glycoprotein complex gB (gp55-116). J Virol 66(11):6747–6754. 33. Rasmussen L, Nelson M, Neff M, Merigan TC, Jr (1988) Characterization of two different human cytomegalovirus glycoproteins which are targets for virus neutralizing antibody. Virology 163(2):308–318. 34. Ortiz-Riaño E, et al. (2014) Inhibition of arenavirus by A3, a pyrimidine biosynthesis inhibitor. J Virol 88(2):878–889. 35. Hoffmann HH, Kunz A, Simon VA, Palese P, Shaw ML (2011) Broad-spectrum antiviral that interferes with de novo pyrimidine biosynthesis. Proc Natl Acad Sci USA 108(14): 5777–5782. 36. Robitaille AM, et al. (2013) Quantitative phosphoproteomics reveal mTORC1 activates de novo pyrimidine synthesis. Science 339(6125):1320–1323. 37. Buchkovich NJ, Yu Y, Zampieri CA, Alwine JC (2008) The TORrid affairs of viruses: Effects of mammalian DNA viruses on the PI3K-Akt-mTOR signalling pathway. Nat Rev Microbiol 6(4):266–275. 38. Yurochko AD (2008) Human cytomegalovirus modulation of signal transduction. Curr Top Microbiol Immunol 325:205–220. 39. Aizaki H, et al. (2008) Critical role of virion-associated cholesterol and sphingolipid in hepatitis C virus infection. J Virol 82(12):5715–5724. 40. Heaton NS, et al. (2010) Dengue virus nonstructural protein 3 redistributes fatty acid synthase to sites of viral replication and increases cellular fatty acid synthesis. Proc Natl Acad Sci USA 107(40):17345–17350. 41. McArdle J, Schafer XL, Munger J (2011) Inhibition of calmodulin-dependent kinase kinase blocks human cytomegalovirus-induced glycolytic activation and severely attenuates production of viral progeny. J Virol 85(2):705–714. 42. Spencer CM, Schafer XL, Moorman NJ, Munger J (2011) Human cytomegalovirus induces the activity and expression of acetyl-coenzyme A carboxylase, a fatty acid biosynthetic enzyme whose inhibition attenuates viral replication. J Virol 85(12): 5814–5824. 43. Delgado T, Sanchez EL, Camarda R, Lagunoff M (2012) Global metabolic profiling of infection by an oncogenic virus: KSHV induces and requires lipogenesis for survival of latent infection. PLoS Pathog 8(8):e1002866. 44. Yu Y, Maguire TG, Alwine JC (2014) ChREBP, a glucose-responsive transcriptional factor, enhances glucose metabolism to support biosynthesis in human cytomegalovirus-infected cells. Proc Natl Acad Sci USA 111(5):1951–1956. 45. Böttcher-Friebertshäuser E, Garten W, Matrosovich M, Klenk HD (2014) The hemagglutinin: A determinant of pathogenicity. Curr Top Microbiol Immunol 385:3–34. 46. Drummer HE, et al. (2013) Allosteric modulation of the HIV-1 gp120-gp41 association site by adjacent gp120 variable region 1 (V1) N-glycans linked to neutralization sensitivity. PLoS Pathog 9(4):e1003218. 47. AuCoin DP, Smith GB, Meiering CD, Mocarski ES (2006) Betaherpesvirus-conserved cytomegalovirus tegument protein ppUL32 (pp150) controls cytoplasmic events during virion maturation. J Virol 80(16):8199–8210. 48. Yuan J, Fowler WU, Kimball E, Lu W, Rabinowitz JD (2006) Kinetic flux profiling of nitrogen assimilation in Escherichia coli. Nat Chem Biol 2(10):529–530.

Protein and Nucleic Acid Analysis. Analysis of proteins and nucleic acids was performed by using standard procedures specifically described in SI Materials and Methods. Chemical Reagents. All chemical reagents are described in SI Materials and Methods.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1415864111

DeVito et al.