The human cytomegalovirus protein pUL13 targets mitochondrial cristae architecture to increase cellular respiration during infection

pUL13 Localizes to the Mitochondria and Is Necessary for Productive HCMV Replication.

We previously demonstrated that the uncharacterized viral protein pUL13 has a temporal localization to the mitochondria during HCMV infection (16). To better define the temporality of pUL13 localization, we infected fibroblasts with a virus strain expressing pUL13 tagged with yellow fluorescent protein (YFP) and analyzed its localization using live-cell fluorescence microscopy at timepoints spanning the HCMV replication cycle (Fig. 1A). In agreement with our previous findings, pUL13 localized to the plasma membrane early in infection (12 hpi) (16). However, we also observed mitochondrial localization throughout HCMV replication, demonstrating that pUL13 localizes to the mitochondria earlier than anticipated. By 48 hpi, pUL13 gained a predominantly mitochondrial localization, which was retained for the remainder of the HCMV replication cycle. Mitochondrial morphological and metabolic changes have been reported to begin at 48 hpi with HCMV, raising the question of whether pUL13 is involved in modulating these processes during infection (4–7, 14).

To investigate the requirement of pUL13 for efficient HCMV replication, we used a mutant HCMV strain (UL13M) with disrupted pUL13 expression via a transposon insertion in the UL13 locus (Fig. 1B) (19). By measuring supernatant viral titer collected from fibroblasts infected with either wild-type HCMV (WT, AD169) or UL13M, we observed that UL13M resulted in a significant, 6.4-fold reduction in viral titer relative to WT (Fig. 1C). Given the potential of the transposon-mutagenized pUL13 to retain partial function, we also generated an HCMV strain with a full deletion of the UL13 locus (∆UL13) (Fig. 1B). We confirmed decreased pUL13 levels during infection with WT, UL13M, and ∆UL13 HCMV using targeted mass spectrometry (MS) to specifically monitor the levels of signature pUL13 peptides (Fig. 1D). Low-abundance pUL13 peptides were still detected during UL13M infection, whereas levels were completely ablated during ∆UL13 infection. Furthermore, infection with ∆UL13 resulted in a more significant, 10.9-fold decrease in supernatant viral titer relative to WT infection (Fig. 1E). Therefore, we utilized the ∆UL13 strain in subsequent experiments.

To pinpoint when the defect during ∆UL13 replication occurs, we analyzed temporal viral protein levels, replication of the viral genome, and generation of infectious viral progeny as parameters corresponding to the sequential stages of virus replication. We analyzed the abundance of the IE, DE, and L viral protein markers, IE1, pUL26, and pUL99, respectively, using Western blotting (Fig. 1F). While IE1 levels were similar at all time points during WT and ∆UL13 infections, pUL26 and pUL99 levels were decreased during ∆UL13 infection, beginning at 48 hpi. We measured cell-associated viral genome abundance during infection using qPCR and observed an over twofold decrease in viral genome abundance during ∆UL13 relative to WT infection by 72 hpi (Fig. 1G). To measure the production of new infectious viral progeny, we quantified cell-associated viral titers and observed a threefold decrease during ∆UL13 compared to WT infection, suggesting a defect in virus production (Fig. 1H). We additionally quantified the particle:plaque-forming unit (PFU) ratio of extracellular virus as a parameter of viral particle infectivity (i.e., the ability to form plaques) (Fig. 1I). ∆UL13 resulted in an ∼eightfold (120 hpi) and ∼fivefold (144 hpi) increase in the particle:PFU ratio, indicating that the virus produced during ∆UL13 infection is less infectious. As DE proteins regulate the replication of the viral genome, which is then required for the expression of L genes and viral assembly, it is likely that the observed defects in genome replication, virus production, and infectivity during ∆UL13 infection are tied to the decrease in DE protein levels. Together, these results demonstrate that pUL13 serves an important function during HCMV infection and that loss of pUL13 negatively impacts stages of the viral life cycle subsequent to the expression of IE genes.

The Temporal Profile of Viral Protein Levels Is Dysregulated during ∆UL13 Infection.

To further investigate the importance of pUL13 during HCMV replication, we explored the temporal changes in the virus and host proteomes during WT and ∆UL13 infections using tandem mass tag (TMT) labeling and MS quantification (SI Appendix, Fig. S1A). Infected fibroblasts were collected throughout the HCMV replication cycle (12, 24, 48, 72, and 96 hpi), thereby capturing time points at which the replication defect was observed during ∆UL13 infection. Prior to TMT labeling, a fraction of each sample from each replicate (30 samples total) was pooled to form an internal reference channel. This allowed for internal reference scaling and reduced variation between replicates (SI Appendix, Fig. S1 B–D, Dataset S1). A total of 9,250 proteins were identified, with 5,177 host proteins and 95 viral proteins passing our quality and quantification criteria.

As expected, pUL13 levels were robustly decreased at all time points during ∆UL13 relative to WT infection, further verifying the deletion of the UL13 locus (Fig. 2A and SI Appendix, Fig. S2A). Additionally, we observed a marked disruption in the temporal profile of viral protein abundances during ∆UL13 infection. Congruent with our Western blot results (Fig. 1F), DE and L viral proteins were more impacted by ∆UL13 than IE proteins (Fig. 2A and SI Appendix, Fig. S2A). By 48 hpi, the vast majority (81/95) of detected viral proteins were decreased in abundance during ∆UL13 relative to WT infection. The decrease in viral protein abundances became more pronounced at later time points of infection (i.e., after 48 hpi). In agreement, a principal component analysis (PCA) of viral protein abundances showed a high correlation between WT and ∆UL13 at 12 and 24 hpi (Fig. 2B). The virus strains diverged beginning at 48 hpi, becoming less clustered as infection progressed.

To investigate the possibility that the decrease in viral protein levels observed during ∆UL13 infection stems from a temporal delay, we performed a distance matrix analysis (SI Appendix, Fig. S2B). Prior to 96 hpi, ∆UL13 was most similar to the corresponding WT time point. At 96 hpi, ∆UL13 became more similar to the 72 hpi WT time point. This demonstrates that the decreased protein levels seen during ∆UL13 infection, up to 96 hpi, are not due to a kinetic delay but rather a global decrease in protein levels.

Notably, the abundances of three viral proteins (pUL14, pUL15A, and pUL17) significantly increased during ∆UL13 infection relative to WT infection (SI Appendix, Fig. S2 A and C). These currently uncharacterized proteins are encoded directly downstream of the UL13 locus (SI Appendix, Fig. S2D). We validated increased pUL14 and pUL17 levels by targeted MS during both ∆UL13 and UL13M infections (SI Appendix, Fig. S2C). We also attempted to detect pUL15A but were not successful, likely due to the small mass of this protein (11 kDa) and the few tryptic peptides produced from its digestion. As there are no known regulatory regions in the UL13 locus, one possibility is that there is a compensatory function for these currently uncharacterized viral proteins during infection. Our previous finding that pUL15A localizes to the mitochondria during infection supports this possibility (16). Altogether, our TMT-MS analysis demonstrates that ∆UL13 significantly disrupts the temporal cascade of viral gene expression, supporting a critical role for pUL13 during HCMV replication.

Metabolic and OXPHOS Proteins Decrease in Abundance during ∆UL13 Infection.

Given our finding that pUL13 is important for productive HCMV replication, we next sought to understand its function. Using our TMT-MS proteome dataset, we monitored the levels of cellular proteins, assessing temporal protein profiles and enriched gene ontology (GO) terms in each cluster (Fig. 2C). Proteins functioning in the regulation of immune response and extracellular matrix organization were increased during ∆UL13 relative to WT infection, while proteins involved in cell division and DNA replication and repair were decreased (Fig. 2C). Notably, metabolic pathways were among the enriched GO terms for both temporally increased and decreased clusters, suggesting that pUL13 might function to regulate cellular metabolism during infection. Evident were the increase in proteins involved in the activation of protein kinase A (PKA) activity and oligosaccharide catabolism (clusters 1 and 2, respectively) and the decrease in glycogen and lipid biosynthetic proteins during ∆UL13 infection.

Functional network analysis of proteins that passed our fold-change and P-value significance cutoffs (1.25-fold change and P value < 0.05) pointed to a temporally regulated decrease in components of the TIM/TOM complex (TOMM5, TIMM8A, and TIMM13), which transports OXPHOS proteins across the mitochondrial membranes (Fig. 2 C, Right and SI Appendix, Fig. S2E). At the same time point (48 hpi), subunits of the ETC decreased in abundance (1.3- to 1.9-fold). We plotted the log₂ abundances of all ETC protein components detected in our TMT-MS dataset (n = 81 proteins belonging to complexes I through IV of the ETC and ATP synthase) (Fig. 2 D and E). The decrease in proteins was complex specific: subunits of complex I, complex IV, and ATP synthase were decreased, while no significant change was observed for complex II or III subunits (Fig. 2E). Altogether, the differential alterations of the cellular proteome during infections with ∆UL13 and WT HCMV suggest a role for pUL13 in regulating cellular metabolism and OXPHOS.

pUL13 Is Necessary for Increased Oxidative Phosphorylation during HCMV Infection.

OXPHOS is known to increase during HCMV infection and to be required for virus replication (14). Given our finding that ETC proteins decrease in abundance during ∆UL13 infection, we hypothesized that pUL13 contributes to modulating OXPHOS during infection. Given that OXPHOS accounts for the majority of cellular oxygen consumption, we used a live-cell Seahorse assay to measure oxygen consumption rates (OCRs) in WT and ∆UL13-infected cells (Fig. 3A). Mock, WT, or ∆UL13-infected fibroblasts were provided with a respiratory substrate, and the OCR of the cells was measured using a series of respiratory modulators. We assessed several functional ETC parameters, including basal respiration, ATP-linked respiration, maximal respiration, proton leak, spare respiratory capacity, and nonmitochondrial respiration. Glycolysis and glutaminolysis, two major pathways for energy production, are increased during HCMV infection to promote viral replication (Fig. 3B) (5–7). Therefore, either glutamine or pyruvate were provided to cells as respiratory substrates during the Seahorse assay to determine if a particular metabolic pathway was preferentially utilized for respiration during infection.

Infection with WT or ∆UL13 increased OCRs over mock at 48 and 72 hpi for both glutamine and pyruvate (Fig. 3C and SI Appendix, Fig. S3A). This reveals that both pyruvate and glutamine are utilized as fuel sources for supporting increased cellular respiration during HCMV infection. Additionally, this finding uncovers that HCMV induces increased cellular respiration by 72 hpi relative to 48 hpi. However, ∆UL13 decreased OCR relative to WT infection at 48 and 72 hpi, supporting a function for pUL13 in increasing cellular respiration during HCMV infection. We converted the raw OCR traces into parameters of ETC function to investigate the impact of pUL13 on cellular respiration (Fig. 3D). We observed similar levels of nonmitochondrial respiration for the two virus strains, supporting the idea that pUL13 specifically alters mitochondrial OCR. In agreement with previous literature (14), HCMV infection resulted in increased basal, maximal, and ATP-linked respiration relative to uninfected cells. However, infection with ∆UL13 resulted in reduced basal, maximal, and ATP-linked respiration on both tested respiratory substrates. The dysregulation of cellular respiration during ∆UL13 infection was more pronounced at 72 compared to 48 hpi, suggesting a progressive defect.

The Seahorse assay media was unbuffered, allowing for quantification of extracellular acidification rate (ECAR), another measurement of metabolic activity. Two major contributors to ECAR are lactate produced from pyruvate and CO₂ produced from TCA cycle reactions (20). Glutaminolysis contributes primarily to TCA cycle–derived acidification, as glutamine bypasses glycolytic fermentation and enters directly into the TCA cycle. Pyruvate contributes to both lactate and TCA cycle–derived acidification. We found no change in ECAR between infected and uninfected cells at 48 hpi (Fig. 3E). However, by 72 hpi, there is a substantial increase in ECAR of infected cells relative to uninfected cells. This indicates that flux through pyruvate fermentation and TCA-cycle pathways are up-regulated during infection, in agreement with previous findings (6, 7). ∆UL13 did not substantially alter ECAR relative to WT levels when glutamine was provided as the fuel source. However, ∆UL13 resulted in reduced ECAR following oligomycin treatment when pyruvate was provided. Oligomycin inhibits ATP synthase, thereby shifting cellular metabolism toward glycolytic ATP production and pyruvate fermentation. Therefore, this finding suggests that ∆UL13 results in decreased capacity for metabolic fuel switching between mitochondrial respiration and glycolysis.

To validate our findings regarding decreased cellular respiration and ATP production during ∆UL13 infection, we performed a liquid chromatography mass spectrometry (LC-MS)–based metabolite profiling experiment to monitor the levels of metabolites related to ATP metabolism in uninfected, WT, and ∆UL13-infected cells at 48 and 72 hpi (Fig. 3F). We used the quantified abundances of ATP, ADP, and AMP to calculate the ATP/ADP ratio and energy charge of uninfected and infected cells as indices of bioenergetic status (21). The ATP/ADP ratio was significantly decreased in infected cells relative to uninfected cells (Fig. 3F). Given that HCMV infection results in increased cellular respiration, this finding suggests that ATP consumption is also increased during infection. Similarly, the energy charge of mock-infected cells was 0.6, within the typical range observed for confluent fibroblasts (22). HCMV infection resulted in a significant decrease in energy charge to below 0.5, indicating a relative increase in AMP abundance during infection. This previously unrecognized decrease in energy charge during infection could reflect that, despite increased respiratory activity, cells still do not keep up with energy demand during HCMV infection. Interestingly, ∆UL13 did not impact the ATP/ADP ratio or energy charge of cells relative to WT infection, pointing to similar energetic status. We observed ATP levels to be slightly elevated during ∆UL13 infection relative to WT infection, suggesting decreased ATP consumption during ∆UL13 relative to WT infection. Additionally, we found an over twofold increase in hypoxanthine levels during ∆UL13 relative to WT infection at 72 hpi. When ATP synthesis is decreased, AMP is degraded to hypoxanthine to maintain the cellular energy charge (23). Altogether, these findings support that ATP metabolism is dysregulated, both in production and usage, during ∆UL13 relative to WT infection.

It has previously been shown that another HCMV protein, pUL37 × 1, contributes to increased cellular respiration during infection by increasing mitochondrial biogenesis (24). Therefore, we investigated the possibility that pUL13 functions similarly during infection. We quantified mitochondrial DNA (mtDNA) abundance during WT and ∆UL13 infection (SI Appendix, Fig. S3B). mtDNA levels increased during infection, in line with previous studies (24), and we found no difference in mtDNA levels between the two viral infections. We additionally analyzed mitochondrial morphology using confocal microscopy during WT and ∆UL13 infection (SI Appendix, Fig. S3C). We observed similar morphologies during infection with either virus. Together, these results demonstrate that pUL13 increases mitochondrial respiration, and unlike pUL37 × 1, this occurs without alteration of mitochondrial biogenesis.

pUL13 Interacts with Cristae-Shaping Proteins at 48 hpi.

Having shown that pUL13 localizes to the mitochondria, impacts the abundance of ETC proteins, and increases OXPHOS without increasing mitochondrial biogenesis, we next investigated how pUL13 can accomplish this function. Given the temporally distinct localization of pUL13 during HCMV infection, we assessed its localization-dependent functions by analyzing its temporal interactions using IP-MS. The use of the UL13-YFP virus strain provided an affinity tag for isolating pUL13. By comparing protein markers of virus replication during UL13-YFP and WT infections, we observed similar levels of IE (IE1), DE (pUL26), and L (pUL99) virus proteins (SI Appendix, Fig. S4A). These observations support the idea that the tag does not affect the temporality of virus replication and therefore is suitable for interaction studies. We infected cells with UL13-YFP HCMV and immunopurified pUL13 at 24, 48, 72, 96, and 120 hpi (Fig. 4A). Using the Significance Analysis of INTnteractome (SAINT) algorithm (25, 26) for filtering specific protein interactions, 117 proteins were predicted to be pUL13 interactions. We visualized their temporal dynamics using our Inter-ViSTA computational platform (27) (Fig. 4B and Movie S1). As expected, pUL13 associated primarily with proteins localizing at the plasma membrane and mitochondria, with additional interactions observed with proteins localized at the endoplasmic reticulum (ER)/Golgi/assembly complex. K-means clustering highlighted four temporal trends of pUL13 interactions (SI Appendix, Fig. S4 B–C). In agreement with our microscopy analyses, many mitochondrial-localized proteins were found to interact with pUL13 beginning 48 hpi (cluster 1, SI Appendix, Fig. S4 B and C).

Functional analysis of its entire temporal interactome implicated pUL13 in a range of cellular processes, including cell adhesion, cell polarity, protein folding, and calcium regulation (Fig. 4C). In agreement with our TMT-MS proteome results, many pUL13 interactions pointed to a likely function in the regulation of cellular metabolism during infection. We found pUL13 to interact with calcium/calmodulin-dependent protein kinase type II (CAMK2) and its downstream effector MYO1C, an unconventional myosin (Fig. 4 C and E). These proteins stimulate glucose uptake through the GLUT4 transporter, thereby implicating pUL13 in regulating glucose transport, which is known to increase during HCMV infection (28, 29). Also in line with our TMT-MS results, which implicated pUL13 in regulating PKA signaling, we found pUL13 to interact with the PKA anchoring protein AKAP8L, which is involved in PKA nuclear shuttling (30).

At 48 hpi, when pUL13 is predominantly localized in the mitochondria, we observed increased interaction with components of complexes II, III, and IV of the ETC (Fig. 4C). pUL13 also interacted with the prohibitin complex, an inner mitochondrial membrane (IMM) complex that associates with and is necessary for optimal ETC function, as well as with LRPPRC, which is necessary for translation of several ETC components (31, 32). Further suggesting a role in regulating OXPHOS during infection, pUL13 formed interactions with IMMT and CHCHD3, core components of the MICOS (33). The MICOS complex bridges the mitochondrial intermembrane space and regulates the architecture of cristae, which house ETC complexes (34, 35) (Fig. 4D). Dysregulation of cristae structure via perturbation of the MICOS complex is linked to impaired cellular respiration (36, 37).

To validate these temporal associations, we designed a targeted MS method for more accurate quantification of pUL13 interactions (Fig. 4E). A UL13-YFP IP was performed, and targeted MS was used to quantify the abundance of proteins of interest from our initial untargeted IP-MS assay, which included proteins involved in mitochondrial structure, metabolism, and calcium signaling (Fig. 4C). Targeted MS by parallel reaction monitoring (PRM) offered a more accurate quantification of the relative abundance of pUL13-interacting proteins at time points spanning the HCMV replication cycle. To further validate the pUL13–MICOS interaction, we performed reciprocal isolations of endogenous IMMT following infection with UL13-YFP (SI Appendix, Fig. S4D). Additionally, to ensure that the interaction does not occur through the YFP tag, endogenous IMMT IPs were performed during WT HCMV infection (Fig. 4F). Quantification of either pUL13 or pUL13-YFP abundances using targeted MS confirmed association with IMMT and further refined the temporal profile of the interaction, revealing that pUL13 most abundantly interacts with IMMT at 24 and 48 hpi.

To further validate our IP-MS findings, we investigated whether pUL13 is a soluble or integral membrane protein by performing sodium carbonate extraction analysis of mitochondria purified from UL13-YFP–infected cells at 48 hpi (SI Appendix, Fig. S4E). pUL13-YFP was detected almost exclusively in the membrane pellet fraction, indicating that it is an integral membrane protein. This submitochondrial localization supports the interaction of pUL13 with the membrane-spanning MICOS complex known to be embedded in the inner mitochondrial membrane (33). Altogether, the pUL13 interactome highlighted the likely multifunctional nature of this viral protein, in agreement with its diverse temporal localizations. Furthermore, it uncovered an interaction between pUL13 and the MICOS complex, which may underlie the ability of pUL13 to increase mitochondrial bioenergetics during infection.

pUL13 Is Sufficient to Alter Cristae Architecture and Increase Mitochondrial Bioenergetics.

Our finding that pUL13 interacts with the MICOS complex suggested that pUL13 is positioned to modulate cristae architecture, which could facilitate pUL13-dependent increases in OXPHOS. However, we also observed and confirmed an interaction between pUL13 and pUL37 × 1, a viral protein known to modulate mitochondrial functions (4, 24, 38). Therefore, to investigate whether pUL13 alone is able to regulate mitochondrial structure and bioenergetics, we generated a fibroblast line stably expressing pUL13 tagged with 3xFLAG and a control YFP-3xFLAG cell line and verified construct expression using Western blotting (SI Appendix, Fig. S5A). Confocal microscopy showed that pUL13 localized to the mitochondria in the stable cell lines, illustrating that other viral factors are not required for its mitochondrial localization (Fig. 5A).

To further confirm that the pUL13 construct is functional, we investigated whether pUL13 expression can rescue the defect in viral titer observed during ∆UL13 infection (Fig. 5B). In line with our previous results, infection of the YFP control cell line with ∆UL13 resulted in decreased viral titer relative to WT. However, infection of the pUL13 cell line with ∆UL13 partially rescued the defect in viral titer at all analyzed infection time points. This result supports the functionality of the pUL13-3xFLAG construct and the fact that the defect in replication of ∆UL13 stems from the absence of pUL13. The partial rescue of viral titer likely stems from the tightly regulated temporality of pUL13 localization during infection, which is lost in the pUL13-expressing stable lines. We next considered whether a rescue in the levels of pUL14 or pUL17, the viral proteins found to be increased during ∆UL13 infection, contributes to this increase in viral titer. Using targeted MS, we quantified pUL14 and pUL17 levels during ∆UL13 infection of the pUL13 or YFP cell lines. We found that, although pUL13 expression partially rescues virus titer during ∆UL13 infection (Fig. 5B), we continue to see similar levels of pUL14 and pUL17 during ∆UL13 infection in both YFP- and pUL13-expressing cells (SI Appendix, Fig. S5B). This suggests that the partial rescue in viral titer is not due to a return of pUL14 and pUL17 abundances to WT infection levels but is instead due to the partial recovery of pUL13 expression.

We next investigated mitochondrial ultrastructure in the stable cell lines using stimulated emission depletion (STED) microscopy (Fig. 5A). The cristae were visualized using an anti-COXIV antibody. This probe was chosen because COXIV, a component of ETC complex IV, preferentially localizes to cristae, and the presence or absence of pUL13 did not change COXIV abundance in our TMT-MS dataset. We observed that pUL13 expression resulted in more heterogeneity in cristae shape and spacing along the mitochondria. Quantification of cristae morphology supported this observation, revealing that cristae are larger and more spherical in pUL13-expressing cells relative to control (Fig. 5C). These data align with our IP-MS findings, supporting that pUL13 modulates mitochondrial ultrastructure.

The modulation of cristae architecture and partial rescue of viral titer resulting from pUL13 expression led us to hypothesize that pUL13 is sufficient for regulating OXPHOS. Using the live-cell Seahorse assay, we measured ECAR and OCR in the pUL13 stable cell line relative to control. By monitoring ECAR during the Seahorse assay, we observed that pUL13 expression results in increased ECAR when pyruvate is provided as the fuel source. This suggests that pUL13 is capable of increasing glycolytic flux (SI Appendix, Fig. S5C). pUL13 expression also increased OCR on glutamine and pyruvate respiratory substrates (Fig. 5D). pUL13 expression did not significantly alter nonmitochondrial respiration, basal respiration, or ATP-linked respiration on any respiratory substrate (Fig. 5E). Excitingly, pUL13 expression resulted in significantly increased maximal respiration and the ability of the ETC to respond to increased energy demand (spare respiratory capacity). These increases were most significant when cells were fed glutamine, the predominant respiratory substrate during HCMV infection (7). pUL13 expression also resulted in a significant decrease in proton leak when glutamine was utilized as the respiratory substrate. These observations demonstrate that pUL13 does not increase basal levels of respiration in cells, but instead functions to prime the ETC for increased energy demand. Energy demand is increased during HCMV infection (14), providing a biological context for this pUL13 function. Together, these results establish that pUL13 modulates cristae architecture and is sufficient to support increased bioenergetic output during HCMV infection.

pUL13 Modulates Cristae Structure to Increase OXPHOS.

The cristae create a distinct bioenergetic microdomain within the mitochondria, where ETC complexes are localized and the proton gradient is concentrated (35, 40). Several proteins and complexes have been shown to modulate cristae architecture. Of these, the MICOS complex is the key factor involved in regulating cristae dynamics and junction formation (18, 34, 41–44). We observed that, at 48 hpi, pUL13 interacts with the core MICOS subunits, IMMT and CHCHD3 (33), which we validated using targeted MS and reciprocal IP. Cristae architecture, and therefore cristae-shaping complexes, are intricately linked to mitochondrial bioenergetics and ETC function. Knockdowns of CHCHD3 and IMMT (MIC60) have been shown to substantially reduce mitochondrial respiration (18, 41, 42). Several mechanisms contribute to these MICOS functions. By controlling cristae junction formation, MICOS concentrates the proton gradient within cristae, restricting proton diffusion away from ATP synthase (44, 45). Additionally, deletion or knockdown of MICOS components has been shown to decrease the abundance and assembly of inner mitochondrial membrane complexes (41, 43, 46). We observed significant decreases in the abundance of IMM proteins during ∆UL13 infection, including TIMM8A-13 complex components and ETC proteins (Fig. 2C). Interestingly, the complex IV subunit COX7A2L was the most substantially decreased ETC subunit during ∆UL13 infection (Fig. 2E). COX7A2L is known to regulate the formation of ETC supercomplexes, supramolecular assemblies that function to increase ETC efficiency (47).

We propose that pUL13 stabilizes MICOS during infection, potentially by functioning as a chaperone (Fig. 6). Moreover, our superresolution microscopy suggests that this interaction functions to maintain cristae integrity and promote tight cristae-junction formation, thereby providing a means for increasing the proton gradient and ETC assembly to promote ETC efficiency and ATP production. This model is supported by several lines of evidence: 1) The ETC has already been demonstrated to be more efficient during HCMV infection (14, 24), suggesting that HCMV has evolved a mechanism to effectively couple electron transport and proton pumping to ATP production; 2) pUL13 does not increase mitochondrial biogenesis during infection, as evidenced by the similar mtDNA abundance during WT and ∆UL13 infection (SI Appendix, Fig. S3B), supporting the idea that pUL13-mediated induction of OXPHOS stems from modulation of ETC function rather than increased mitochondrial content; 3) pUL13 interacts with the core MICOS subunits, IMMT and CHCHD3, during infection; 4) The mitochondrial ultrastructure is visibly perturbed upon expression of pUL13 (Fig. 5A); and 5) Expression of pUL13 alone is sufficient for increasing the maximal respiration rate and spare respiratory capacity of the ETC (Fig. 5E). These findings establish that pUL13 serves to prime the mitochondria, promoting formation of a cristae architecture that is capable of supporting increased OXPHOS in response to the energy demand driven by HCMV replication.

Our alignment of the pUL13 amino acid sequence across the laboratory-adapted strain AD169 and the clinical strains MERLIN and TB40/E shows a high level of homology (SI Appendix, Fig. S5D). The pUL13 sequence between TB40/E and AD169 shares 99.4% of its amino acid sequence homology, with only three amino acid substitutions and no insertions or deletions, and the MERLIN and AD169 share 97.9% of their homology. The overall similarity in sequences across these strains suggests that pUL13 functions may be conserved during infections with other HCMV strains.

Viral perturbation of ETC function has been typically studied in the context of viral modulation of reactive oxygen species generation, apoptosis, and innate immune sensing (48). Our findings contribute to an emerging narrative in which viruses, particularly those with long replication cycles, target the ETC to increase mitochondrial bioenergetics (49). By uncovering a function for pUL13 in increasing OXPHOS during HCMV infection, we establish a mechanism through which viruses can increase cellular respiration via modulation of MICOS function and IMM architecture.

Cell Culture and Viral Infection.

Human fibroblasts (HFs) were cultured under standard conditions in a complete growth medium. HFs stably expressing pUL13-3xFLAG or YFP-3xFLAG were generated via lentivirus transduction followed by puromycin selection.

Stocks of HCMV AD169, AD169-GFP, AD169 UL13-YFP, AD169 UL13M, and AD169 ∆UL13 were produced from bacterial artificial chromosomes in HFs and titered using tissue culture infectious dose (TCID50).

Infections were carried out at 37 °C for 1 h, followed by a phosphate-buffered saline (PBS) wash and addition of growth media. Infectious units were quantified by using cell-culture supernatant to infect a reporter plate of HFs. At 24 hpi, the reporter plate was stained with anti-IE1 antibody and DAPI. IE1-positive nuclei were quantified using Operetta imaging.

Immunofluorescence Microscopy.

Immunofluorescence microscopy was conducted using a Nikon Ti-E confocal microscope equipped with a spinning disk module. A Nikon A1R confocal microscope equipped with an STED detector was utilized for confocal and STED analysis of pUL13-3xFLAG and YFP-3xFLAG stably expressing HFs.

qPCR.

Viral, mitochondrial, and nuclear genomes were quantified by qPCR using the SYBR green PCR master mix and the ViiA7 Real-Time PCR System. Primers were specific to the virus IE1 gene, mtDNA 16s ribosomal RNA gene, or the nuclear β2-microglobulin gene. To quantify extracellular virus particles, the cell-culture supernatant was treated with DNase I to degrade nonpackaged viral DNA prior to qPCR sample preparation.

TMT-MS.

HFs infected with WT or ∆UL13 HCMV in biological triplicate were collected at 12, 24, 48, 72, and 96 hpi. Equal protein concentrations for each sample were reduced, alkylated, precipitated, and digested with trypsin. Prior to TMT labeling, equal volumes of all samples from each of the three replicates were pooled and then split into three aliquots to be used as internal reference channels. Samples were labeled using the TMT10plex and TMT11-131C Isobaric Label Reagents and analyzed via nano liquid chromatography–tandem mass spectrometry (LC-MS/MS).

IP-MS.

pUL13-YFP IPs, along with the matched GFP-virus control, were performed in biological duplicate at each time point (24, 48, 72, 96, and 120 hpi) for discovery and targeted experiments. For anti-GFP IPs, preconjugated GFP-Trap magnetic beads were used. For reciprocal endogenous IMMT IPs, anti-IMMT antibodies were conjugated to protein A/G magnetic beads. Bound proteins were eluted from the beads and reduced, alkylated, and digested with trypsin. Samples were analyzed via LC-MS/MS, and the total spectral count data were analyzed by SAINT (26) using the Resource for Evaluation of Protein Interaction Networks (REPRINT) (25) interface to determine interaction specificity.

Targeted MS.

Targeted MS samples were analyzed via LC-MS/MS with targeted MS2 scans controlled by a peptide inclusion list containing two to four unique peptides for each targeted protein. Data analysis was performed using the Skyline software (50). A summed area under the curve of three transitions per peptide was used for the quantitation.

Quantification of Cellular Respiration.

Cellular respiration was quantified using the Seahorse XF Cell Mito Stress Test Kit according to manufacturer specifications. The first injection contained the carbon source (either 5 mM pyruvate or 5 mM glutamine), and subsequent injections consisted of 2 μM oligomycin, 1 μM FCCP, and 0.5 μM rotenone/antimycin A. OCR results were normalized to the total protein abundance.

Metabolite Profiling.

Cellular metabolites were extracted using ice-cold 80% MeOH containing 1 mM N-ethylmaleimide and analyzed by LC-MS. Data analysis was performed using EL-MAVEN software, and metabolite abundances were normalized to the total protein abundance. The cellular energy charge was calculated as (ATP+1/2ADP)/(ATP+ADP+AMP).