“Transfer” of power: The intersection of DNA virus infection and tRNA biology

Abstract

Transfer RNAs (tRNAs) are at the heart of the molecular biology central dogma, functioning to decode messenger RNAs into proteins. As obligate intracellular parasites, viruses depend on the host translation machinery, including host tRNAs. Thus, the ability of a virus to fine-tune tRNA expression elicits the power to impact the outcome of infection. DNA viruses commonly upregulate the output of RNA polymerase III (Pol III)-dependent transcripts, including tRNAs. Decades after these initial discoveries we know very little about how mature tRNA pools change during viral infection, as tRNA sequencing methodology has only recently reached proficiency. Here, we review perturbation of tRNA biogenesis by DNA virus infection, including an emerging player called tRNA-derived fragments (tRFs). We discuss how tRNA dysregulation shifts the power landscape between the host and virus, highlighting the potential for tRNA-based antivirals as a future therapeutic.

1. INTRODUCTION

Viruses have evolved unique mechanisms to invade hosts, alter cellular pathways, and redirect host factors for viral processes. One commonly hijacked pathway is protein translation. Viruses manipulate translation to prioritize production of viral peptides and block the host innate immune response. Transfer RNAs (tRNAs) are emerging as a ubiquitous viral target of the translational machinery. tRNAs are adaptor molecules which decode the sequence of a messenger RNA (mRNA) during protein translation and supply the correct amino acid to the growing polypeptide. In addition to their role in protein synthesis, tRNAs function in cellular stress signaling, apoptosis, innate immune sensing, and gene silencing[1]. As tRNAs are absolutely required for protein translation, it is not surprising that viruses rely on sustained or elevated tRNA expression for fitness. For example, Influenza A virus and vaccinia virus modulate local (ribosome-associate) pools of host tRNAs within infected cells to promote translation of viral proteins [2]. Many tumor viruses (human polyomavirus, human papillomavirus, gamma-herpesvirus) enhance RNA Polymerase III (Pol III) activity and tRNA expression (Table 1 and references). Viral and host oncogenes frequently target tRNAs and their expression is correlated with tumorigenesis and cancer progression [3]. There are non-canonical uses of tRNAs by viruses as well; for example, retroviruses use specific host tRNAs as the primer for reverse transcription of their RNA genome to dsDNA[4]. In addition to enriched pools of tRNAs, some DNA viruses encode their own Pol III genes (described in section 2) and might logically benefit from enhanced Pol III activity during infection to express viral non-coding RNAs. tRNA modulation has recently been reviewed in the context of RNA virus infection [5], so here we will discuss modulation of tRNAs by mammalian DNA viruses. Excitingly, the advancement of tRNA sequencing technologies is paving the way towards a better understanding of how tRNA regulation impacts infection. Our understanding of host-virus interactions and our ability to recognize potential tRNA-based therapeutics will improve as more studies dissecting tRNA function and regulation during viral infection are executed.

Table 1.

Tumor viruses that impact Pol III activity

Virus family	Viruses studied	Mechanism(s) of Pol III stimulation proposed
Adenoviridae	Adenovirus 2 (Ad2)	E1A oncoprotein drives increased expression and activity of TFIIIC[55–59]
	Adenovirus 5 (Ad5)
		E1A interacts with Rb, relieving repression of TFIIIB[60]
		During infection, E1b and E4 may also play a role in addition to E1A[54]
Papillomaviridae	Human Papillomavirus (HPV)	HPV E6 and E7 oncoproteins bind Rb, relieving repression of TFIIIB[61–63]
Polyomaviridae	Simian virus 40 (SV40), polyomavirus (Py)	SV40 and Py transformed cells show increase in TFIIIC levels and phosphorylation[46, 64, 65]
		SV40 and Py transformed cells have elevated BDP1, subunit of TFIIIB[65]
		Large T antigen (SV40 and Py) binds Rb, relieving repression of TFIIIB[63, 65]
		Middle T antigen (Py) contributes to Pol III induction[65]
Herpesviridae	Alpha-Herpes Simplex Virus-1 (HSV-1)	HSV-1 transcript accumulation triggers tRNA upregulation[49]
		POLR2A binding correlates with upregulated tRNAs[49]
	Beta-Human cytomegalovirus (HCMV)	HCMV transcript accumulation and/or an early-late/late factor drive chromatin changes and enhanced Pol III elongation at tRNA genes[50]
	Gamma-Murine gammaherpesvirus	Increased Pol III abundance at tRNA genes upon MHV68 infection[66]
	68 (MHV68), Epstein-Barr Virus (EBV)	MHV68 ORF36 inhibits HDAC activity, leading to increased Pol III transcription[67]
		MHV68 ORF45 signals through ERK/MAPK, stimulating Brf1 and Pol III transcription[68]
		Host shutoff via MHV68 ORF37/muSOX blocks the turnover of pre-tRNAs[66]
		EBV transformation and EBNA1 increase TFIIIC RNA and protein levels, ATF-2 and cMyc transcription factors[52, 53]
		EBV lytic infection increases tRNA expression[69]

2. tRNA gene expression and maturation

There are three DNA-dependent RNA Polymerases (RNA Pol). RNA Pol I, II, and III are comprised of 14, 12, and 17 core subunits respectively (Fig. 1A) [6]. While five subunits (POLR2E, POLR2F, POLR2H, POLR2K, POLR2L) are shared amongst them, each RNA Polymerase possesses unique subunits and distinct transcription accessory machinery. Pol III is uniquely adapted for transcription of short (100–500 nt), abundant, noncoding RNAs (ncRNAs)—including 5S ribosomal RNA (rRNA), transfer RNA (tRNA), Alu elements, 7SL, 7SK, and U6 snRNA. Pol III also transcribes ncRNA for various DNA viruses, including VAI and VAII RNAs of adenoviruses, EBER1 and EBER2 of Epstein-Barr Virus (EBV), and the tRNA-miRNA encoding RNAs (TMERs) of murine gammaherpesvirus-68 (MHV68) [7–10]. Pol III transcription requires its own unique set of general transcription factors and have been classified into three different promoter types (Fig. 1B) [11]. Each of these promoters consists of distinct combinations of internal cis-acting sites (A, B, C, IE, TATA box) which recruit a complement of basal transcription factors. In this review, we will be focusing on tRNAs which are Pol III-dependent Type 2 genes.

Figure 1. tRNA gene expression.

A) There are three DNA-dependent RNA Polymerases (RNA Pol). RNA Pol I, II, and III are comprised of 14, 12, and 17 core subunits respectively. The largest of these is Pol III which is responsible for transcribing short (100–500 nt), highly abundant, noncoding transcripts. B) Pol III transcripts can be separated into three classes based on core cis- and trans-acting elements. This is in sharp contrast to the promoter complexity of Pol II-dependent transcripts, which have a wide array of distal and proximal sequences elements (DSE or PSE). Pol III has a unique set of core general transcription factors which includes TFIIIA, TFIIIB, and TFIIIC. Only TATA-binding protein (TBP) is shared between the Pol III and Pol II machinery. C) tRNA gene expression occurs through a classical three step mechanism including initiation, elongation, and termination. Pol III is uniquely capable of recycling, or proceeding directly from termination to initiation at the same template gene.

tRNA gene expression uses a canonical three step mechanism including initiation, elongation, and termination (Fig. 1C). tRNA promoters are first bound by TFIIIC, a complex containing GTF3C1, GTF3C2, GTF3C3, GTF3C4, GTF3C5, and GTF3C6 [12, 13]. The GTF3C5 and GTF3C1 subunits bind the A- and B-box elements within the tRNA gene, respectively [6, 14, 15]. TFIIIC recruits binding of a TFIIIB complex containing BDP1, TBP, and BRF1 [16]. While TBP is a component of the TFIIIB complex, most mammalian tRNA promoters do not depend on the presence of TATA-box elements [17]. The BDP1 subunit of TFIIIB contacts the dsDNA major groove and is stabilized by interactions with TFIIIC [6]. TFIIIB is responsible for recruiting Pol III through interactions with the POLR3E and POLR3F subunits. Unlike Pol II, Pol III does not exhibit promoter proximal pausing. Instead, TFIIIC may regulate Pol III elongation rates. Recent work from Male et al. 2015 proposes dissociation of the TFIIIC B-box binding lobe after initiation. ChIP-Seq studies analyzing Pol III occupancy of tRNA genes found a bimodal distribution with peaks at the A and B-boxes, suggesting transient pausing at these elements [18]. This data suggests TFIIIC occupancy after initiation negatively impacts the rate of Pol III elongation. This model requires additional support as there is conflicting evidence regarding whether TFIIIC remains bound to tRNA genes after elongation [12, 19, 20]. Pol III termination occurs when the polymerase contacts a string of 4 to 5 deoxythymidine nucleotides (poly-dT) [21, 22]. Contact with poly-dTs results in a Pol III conformational change and elongation pause. Subsequently, the Pol III complex dissociates and can be directly transferred from the termination site to the promoter of the same template. TFIIIB-DNA complexes are stable once formed and facilitate the rapid rates of Pol III reinitiation [12, 23–25]. Rapid recycling of Pol III in this way is purported to be a major factor behind the high rates of Pol III gene transcription [23].

Pol III transcription is subject to additional negative and positive regulators (Fig. 1C). These include MAF1 which binds to Pol III preventing interaction with TFIIIB and initiation complex formation [26, 27]. The tumor suppressor gene, Rb, also acts as a negative regulator of Pol III activity. Rb binds to TFIIIB preventing recruitment to tRNA genes and blocking initiation [28]. On the other hand, the proto-oncogene Myc promotes Pol III activity by enhancing transcription of the core machinery (i.e. BRF1, TBP, GTF3C3, GTF3C4, GTF3C5) or promoting initiation [29]. To add another layer of complexity, recent studies have demonstrated crosstalk between Pol II and Pol III. Highly expressed Pol III transcripts are located in regions of open chromatin adjacent to Pol II promoters [30]. Additionally, Pol II binding is observed at highly expressed Pol III promoters [31], and Pol III occupancy frequently scales with nearby levels of Pol II [11, 32]. Recent evidence also suggests a role for chromatin in modulating Pol III transcription, or vice versa [33]. There is still much unknown regarding the significance of chromatin signatures and Pol III-dependent gene expression.

Generation of mature-tRNA requires a number of post-transcriptional processing steps including i. trimming of leader (5’) and trailer (3’) sequences, ii. splicing of introns for a small subset of pre-tRNAs, iii. 3’-CCA addition, and iv. nucleoside modifications [34] (Fig. 2). The study of RNA modifications, or the epitranscriptome, has risen to the forefront lately, with evidence for over 170 distinct modifications (reviewed in [35]). The most abundant such modification for mRNA is N⁶-methyladenosine (m⁶A) which globally impacts splicing, degradation, and epigenetic regulation. Compared to other transcript classes, tRNAs have the highest concentration of nucleoside modifications—an average of 13 modifications per molecule [36]. Over 100 different nucleoside modifications have been identified for tRNAs, the most ubiquitous of these are pseudouridine (Ψ), inosine (I), dihydrouridine (D), 4-thiouridine (s⁴U), 1-methyladenosine (m¹A), 3-(3-amino-3-carboxypropyl) uridine (acp³U), and queuosine (Q). These nucleoside modifications are essential to tRNA stability and codon recognition, ultimately impacting translation rates.

Figure 2. Alteration of tRNA expression and maturation by DNA viruses.

Diverse DNA viruses modulate tRNA expression, most commonly by increasing transcriptional output of RNA Pol III. Various mechanisms, including increased expression of TFIIIC transcription factors and interaction with Retinoblastoma (Rb), have been described (see text). There are likely other steps of tRNA maturation that are perturbed during infection that are yet to be identified. HCMV= human cytomegalovirus, HPV= human papilloma virus, HPyV= human polyoma virus, AdV= adenovirus, EBV= Epstein-Barr Virus, HSV-1= Herpes Simplex Virus-1, MHV68= murine gammaherpesvirus

Mature-tRNAs also undergo aminoacylation, an ATP dependent reaction which covalently links a charged amino acid moiety to the 3’ of the cognate tRNA [37]. Only when these steps are completed properly can tRNAs function as adaptor molecules to decode the sequence of a mRNA during protein translation. However, many of these modifications cause problems when performing tRNA sequencing. When building cDNA libraries, many tRNA nucleoside modifications block the progression of reverse transcriptases used in cDNA synthesis. Charged 3’ ends can preclude ligation of sequencing adaptors. Additionally, removal of the only loci-specific sequences (5’ leader, 3’ trailer, and/or intron) during tRNA maturation makes it highly difficult to accurately map tRNA sequencing reads. These technical hurdles in tRNA sequencing have resulted in many gaps in knowledge regarding mammalian tRNA expression and regulation. For example, the driver(s) of differential tRNA expression, tRNA quality control, and non-canonical functions of tRNAs and tRNA-derived transcripts are among the underdeveloped areas in tRNA biology.

3. Technological advances in tRNA studies

As mentioned above, tRNAs have historically been hard to study due to their secondary structure and high number of nucleoside modifications. Most reverse transcriptase (RT) enzymes are unable to copy the full length of tRNA transcripts leading to artificial read fragments and biased sampling. tRNA detection was limited to Northern blots—capable of detecting and delineating between pre- and mature-tRNA species, but severely limited in sensitivity or scale—and RT-qPCR which is only suitable for intron-containing pre-tRNAs. Since 2015, a number of techniques have arisen which address these complications and facilitate high-throughput sequencing (HTS). These technologies leverage demethylase enzymes and sophisticated reverse transcription strategies to produce full length tRNA RT products. Demethylase-thermostable group II intron RT tRNA sequencing (DM-tRNA-seq) and AlkB-facilitated RNA methylation sequencing (ARM-seq) target methylated RNAs to allow for efficient sequencing of methyl-modified RNAs [38, 39]. Similarly, panoramic RNA display by overcoming RNA modification aborted sequencing (PANDORA-seq) also utilizes enzymes such as AlkB and T4PNK to remove modifications that interfere with reverse transcription and adaptor ligation [21]. A recent study combined DM-tRNA-Seq with RNA mass spectrometric analyses for high-confidence analysis of tRNA modifications [40]. While these techniques enable tRNA HTS, RT processivity and adaptor ligation probe bias still complicates discrimination of tRNA species. Ordered Two-Template Relay (OTTR)-Seq leverages a novel bioengineered RT that efficiently synthesizes full copies of tRNA species, and thus pre-tRNA, mature-tRNA, and tRNA-derived fragments (tRFs) can be confidently assessed [41]. Analysis of tRNA sequencing data also requires a custom-tailored method for mapping and quantitation. This is due, at least in part, to the way tRNA genes are present in the genome. The human genome contains approximately 500 tRNA genes, some of which are gene duplicates spread across multiple chromosomes [42]. Furthermore, discrimination of pre- and mature-tRNA is not as simple as in studies of mRNA. Only a few tRNA genes contain introns, making exon-exon junctions an unreliable distinguishing feature. Recent work from Holmes et al. 2021 generates a bioinformatic pipeline (https://github.com/UCSC-LoweLab/tRAX) which enables efficient quantitation of pre- and mature-tRNA species, as well as tRFs. This pipeline adds to a growing list of bioinformatic tools for tRNA and tRF mapping[43–45]. These technological advances now enable global analysis of tRNA species, abundance, and modifications.

4. DNA viruses enhance RNA polymerase III activity and tRNA expression

Since the 1970s, we have known that DNA virus infection increases the production of Pol III transcripts, including tRNAs (Table 1). Dramatic increases in expression are seen with exogenous Pol III templates (Ad5 VAI or E. coli tRNA) incubated with extracts from adenovirus, polyomavirus, or herpesvirus-infected cells (see references in Table 1). In contrast, minor (or no) increases of endogenous Pol III transcripts, including tRNAs, are seen upon DNA virus infection or transformation with viral oncoproteins [46–48]. The differences in magnitude of Pol III stimulation on the genome versus exogenous templates could be explained by several mechanisms, one being differences in chromatin accessibility. It seems more likely, however, that changes in nascent tRNA transcription are difficult to detect above the background of pre-existing, highly abundant tRNAs. For example, a nuclear run-on assay to measure nascent Pol III transcription revealed increased tRNA expression in response to SV40 transformation[46] where bulk steady state measurement did not[47]. Furthermore, recent studies measuring nascent Pol III transcription (4SU-seq[49], Pro-seq[50], and bioinformatic mapping strategies to parse precursor and mature transcripts[49, 51]) with herpesviruses in fact show striking changes in endogenous tRNA expression upon viral infection. In this section, we present proposed mechanisms for how tRNAs are transcriptionally upregulated by DNA viruses.

4.1. Many DNA viruses upregulate Pol III transcription factor levels and/or activity

Several viral strategies have evolved to stimulate Pol III activity during infection (Table 1), with a common strategy involving increases in the levels or activity of the Pol III transcription factor complexes, TFIIIB and TFIIIC, by viral oncoproteins. These viral oncoproteins, including E1A from adenoviruses (Ad2 and Ad5), T antigen proteins from the polyomavirus SV40, and E6/E7 proteins from human papillomavirus (HPV), influence Pol III activity by two different mechanisms (references in Table 1). First, these proteins can increase the activity or steady state levels of TFIIIC subunits. Second, these proteins interact with Retinoblastoma (Rb), disrupting the interaction of Rb with TFIIIB and thus releasing TFIIIB for transcription. The EBNA1 protein expressed in Epstein-Barr Virus (EBV)-transformed cells similarly increases TFIIIC RNA and protein levels and boosts the expression of ATF-2 and cMyc transcription factors that support Pol III activity[52, 53]. It is important to note that many of these studies were performed using transient transfection of viral proteins, or with virus-transformed cells, and may not be reflective of what happens during active viral replication. For example, use of adenovirus mutant strains revealed that E1A was not necessary for Pol III activation, and other viral products (E1b and E4) might also contribute[54]. Altogether, increasing the concentration of active Pol III transcription factors is a common viral strategy for enhancing tRNA transcription.

4.2. Herpesviruses-driven host shutoff requires a different strategy to upregulate Pol III

While enhancing transcription factor concentration is a common strategy used by DNA viruses for upregulating tRNA expression, this is not the case during alpha- and gamma- herpesvirus infection. These viruses dramatically downregulate host mRNA expression and translation via an effect called ―host shutoff,‖ driven by expression of viral endonucleases (HSV-1 vhs, KSHV SOX, MHV68 muSOX, EBV BGLF5) that cleave mRNAs genome-wide and lead to decreased protein synthesis [70, 71]. Alpha-herpesviruses further decrease host transcription by sponging Pol II from host mRNA promoters in a ICP4-viral DNA dependent fashion [72–75]. Global downregulation of host transcripts during lytic infection is not compatible with the general strategy used by other DNA viruses to increase expression of Pol III and associated transcription factors. However, we considered the possibility that the Pol III machinery is resistant to host shutoff. To examine this, we have compiled proteomics data across all three sub-families of herpesviruses (Fig. 3) [6, 76–78]. In terms of both transcript and protein abundance all components of the tRNA gene expression machinery were decreased during HSV-1 infection [49]. KSHV lytic infection induced a global decrease in Pol III machinery abundance, with the positive regulator MYC being the only holdout. During EBV lytic infection there was a subtle increase in protein abundance for TFIIIC, BDP1, and TBP; consistent with prior work in EBV transformed cells [79]. Interestingly, the negative tRNA regulator, MAF1, was also decreased during lytic EBV infection. This observation requires follow-up to assess whether it may contribute to increased tRNA expression. Of note, TFIIIB subunits are missing in several of these proteomics studies, likely due to low expression levels. However, in a study of HSV-1 infection in primary fibroblasts, BRF1 and TBP levels were measured by western blot, again revealing unchanged or decreased levels after infection [49].

Figure 3. Protein levels of tRNA transcription machinery.

Quantitative proteomic analysis for herpesvirus infection models. HSV-1 data is from immortalized human keratinocyte cells (HaCaT) infected at an MOI of 10 pfu/cell [78]. CMV data is from human foreskin fibroblasts (HFFs) infected with strain Merlin at an MOI of 10 pfu/cell [6]. KSHV data is from immortalized human umbilical vein endothelial cells (HuAR2T) harboring latent KSHV transduced with RTA to induce lytic reactivation and FACS enriched [76]. EBV data is from 4-HT-induced P3HR1-ZHT/RHT cells sorted to select for fully lytic (gp350+) cells [77].

In contrast, infection with HCMV—which lacks a viral endonuclease—leads to dramatic increases in all components of the tRNA transcription machinery at early times post infection (24 hours) (Fig 3). This upregulation of the tRNA transcription machinery may contribute to upregulated tRNA expression. A recent report found altered tRNA gene elongation rates as early as 24 hours after infection, with global changes by 48 hours [50]. Unlike the general increase in tRNA expression following HSV-1 infection [49], HCMV appears to dramatically upregulate select tRNAs and repress others. This data suggests HCMV shifts the tRNA pool during infection, however more evidence is necessary as mature tRNAs have not been assessed. Considering tRNAs have half-lives around 3 days, but the HCMV replicative cycle is on the scale of 5 to 7 days; one can see how a shift in the tRNA pool could be particularly advantageous for HCMV. Additionally, HCMV—unlike other herpesviruses—does not inhibit host translation [80] and must compete with host mRNAs for the translation machinery. Shifting the host tRNA pool may be a mechanism by which HCMV promotes viral translation.

In summary, alpha- and gamma- herpesviruses exhibit a general downregulation of TFIIIB/C and Pol III, except for subtle (<2-fold) increases in TFIIIC subunits during EBV replication. Ultimately these data suggest alpha- and gamma-herpesviruses employ a distinct strategy to alter tRNA expression that is yet to be defined.

4.3. Host factors participating in enhanced Pol III transcription during herpesvirus infection are unknown

This leaves an unanswered question as to what signal is stimulating Pol III transcription in alpha- and gammaherpesvirus infected cells. The mechanism driving tRNA induction during MHV68 and HSV-1 infection has been explored recently, yet definitive, mechanistic insight remains limited [49, 66, 81]. While transcriptional upregulation of tRNAs is associated with increased recruitment of POLR3A (independent of the Pol III regulator, Maf1) following MHV68 infection in mouse fibroblasts [66], no changes in tRNA gene occupancy for a panel of Pol III factors (BRF1, TBP, GTF3C5, and POLR3A) was found by ChIP-Seq following HSV-1 infection in human fibroblasts [49]. tRNA induction during HSV-1 infection might involve potential cross-talk with Pol II, as induced tRNAs had increased POLR2A binding. This finding was particularly notable, as POLR2A occupancy is drastically reduced on mRNA targets during HSV-1 infection. Whether POLR2A binding impacts tRNA transcription in this context remains untested [49]. We posit that increased POLR2A binding may be a ―symptom‖ of euchromatic markers at tRNA loci. In this way, tRNA expression remains upregulated in a nuclear environment generally unsupportive to host transcription. Other potential mechanisms involve a change in the activity state of TFIIIB/C or Pol III (rather than changes in protein levels), or involvement of Rb or other Pol III regulators like p53 or cMyc. To date, host factors driving tRNA transcriptional increases during herpesvirus infection remain unknown.

4.4. Herpesviral proteins that promote tRNA upregulation

Though host factors remain unknown, several viral gene products contribute to elevated tRNA expression during gammaherpesvirus infection. Infection with MHV68, a murine model gammaherpesvirus, triggers increased pre-tRNA levels which are attributed to both transcriptional upregulation and decreased turnover/processing [66]. At least three MHV68 proteins contribute to increased levels of pre-tRNAs: ORF36 [67], a conserved gammaherpesvirus kinase, ORF45 [68], a conserved virion tegument protein known to interact with ORF36, and ORF37/muSOX, the host shutoff endonuclease [66]. Both ORF36 and ORF45 are sufficient to upregulate Pol III transcription of tRNAs and B2 retrotransposons [67, 68], which are derived from an ancestral tRNA [82]. ORF36 likely does this through inhibiting histone deacetylation activity [67], while ORF45 activates MAPK/ERK signaling and can increase Brf1 expression when expressed outside the context of infection [68]. In contrast, muSOX is not sufficient for tRNA upregulation (unpublished data) and is hypothesized to affect downstream turnover or processing of pre-tRNAs, leading to increased stability of pre-tRNAs in infected cells [66]. While ORF36 and ORF45 are sufficient to induce tRNA expression, experiments using ORF36-deleted or kinase-dead MHV68 (MHV68 ORF36S/KN), WT MHV68 in the presence of shRNA targeting ORF45, or muSOX-mutant MHV68 (MHV68.R443I) show only partial reduction in pre-tRNA levels [66–68]. These results echo previous work with Ad5, suggesting that E1A, while sufficient for Pol III upregulation, is not fully necessary in the context of infection [54]. Similar experiments using an extensive panel of HSV-1 mutants were unable to identify any one protein that was fully necessary for tRNA upregulation [49]. It was concluded that substantial viral gene expression, independent of sequence or product, may be what initiates Pol III activation in the case of HSV-1; however, it is possible that the right combination of HSV-1 proteins has not been assessed. Together, this suggests that the increase in tRNA expression upon MHV68 and HSV-1 infection is driven by multiple viral factors and signaling events that affect Pol III recruitment and the downstream stability of pre-tRNAs.

5. Functional consequences of Pol III stimulation and tRNA expression changes during infection

5.1. Pol III activation stimulates, but is not sufficient for, cellular transformation

Because DNA virus infection and cancer both involve cellular transformation[83–85], it has been frequently hypothesized that elevated Pol III activity and/or tRNA expression increases might drive transformation. Similar to DNA virus infection or transformation by viral oncoproteins, cancer is associated with elevated Pol III transcription through inhibition of tumor suppressors like Rb and p53 and the resulting release of TFIIIB (references in [83]). Additionally, some cancers are associated with increased expression of TFIIIB and TFIIIC subunits[52, 86]. Several studies have explored whether increased Pol III expression drives cellular transformation. The HPV E7 oncoprotein drives cellular transformation, while a single mutation in E7 (called the PRO2 mutant) abolishes this activity. However, both wild type and the PRO2 versions of the E7 protein can trigger Pol III upregulation, suggesting that Pol III upregulation is not sufficient for cellular transformation. Beyond studies involving viral oncoproteins, two studies have shown that Brf1 overexpression alone is not sufficient to initiate cellular transformation in fibroblasts or tumorigenesis in mice[87, 88]. Together, this suggests that there are multiple mechanisms used by viral oncoproteins to incite cellular transformation, and supports the conclusion that Pol III activation, while contributing to, and in some cases necessary for transformation, is itself not sufficient[87, 88].

5.2. Influence on translational efficiency

Translation is a central battleground in the host-virus arms race, as nicely reviewed by Stern-Ginossar et al. [89]. Viruses rely exclusively on host translation machinery to make proteins, prompting one of the main consequences of the innate immune response—translational control. One (of many) example includes the phosphorylation of eukaryotic initiation factor (eIF)2α by activated pattern-recognition receptor, protein kinase R (PKR) to globally block translation. Conversely, given the need to scavenge host translational machinery for protein expression, viruses have evolved creative ways to favor the translation of viral proteins in the infected cell. These include general viral strategies to inhibit innate immune pathways and outcompete host mRNAs for translation.

Although there are numerous reports of enhanced tRNA expression during DNA virus infection, we know exceedingly little about how these changes affect translation and/or the progression of infection (Fig 4). The primary outcome following changes in tRNA expression might be a shift in the translational pool of mature tRNAs. This could allow for increased translational efficiency for viral genes, which often have different codon biases than their host. Accordingly, the GC content of DNA viruses is highly variable (30–80%[90]), meaning that some DNA viruses, including poxviruses and HSV-1, have different codon usage than their host. There is some evidence that GC content is related to viral tissue tropism and the tRNA-based translational efficiencies of target tissues[91]. However, depending on the scenario, differential codon usage theoretically provides leverage for the host or virus to take advantage of altered tRNA pools during infection. Finally, there are reports of host tRNAs associated with herpesvirus virions, which is especially interesting to consider given the altered repertoire in infected cells [92, 93].

Figure 4. Downstream effects of virus induced tRNA dysregulation.

A number of DNA viruses alter tRNA pools by shifting the species profile or increasing total abundance. Changes to mature tRNA repertoires alters protein translation and has been shown in some cases to favor viral protein synthesis. DNA viruses also perturb tRNA maturation creating a build-up of intermediate tRNA isoforms (pre-tRNAs) or tRNA fragments (tRFs). Pre-tRNAs and tRFs may induce a RIG-I dependent innate immune response, as they possess an uncapped 5’-ppp. tRFs may also function like other noncoding RNAs (i.e., miRNA, lncRNA) to alter host gene expression. This has been shown to occur through RNA binding protein (RBP) sponging or via DICER1-processing to convert tRFs to canonical miRNAs.

tRNA nucleoside modifications alter their rigidity and structure, ultimately impacting translation specificity and rates. Recent reports have found that DNA viruses, including AdV, HSV-1, HCMV, and EBV modulate the m⁶A machinery during infection [94–97]. Additionally viral transcripts have themselves been shown to contain nucleoside modifications, such as m⁶A and inosine (reviewed in [98]). While the m⁶A modification is not a major player in tRNA biology, it highlights another potential route by which viruses can impact the tRNA machinery. Recent technological advances now enable global assessment of tRNA nucleoside modifications [99], and we expect this to facilitate new and exciting findings regarding how viruses alter this crucial aspect of tRNA biology.

5.3. Schlafen proteins: antiviral tRNA endonucleases

Studies of the Schlafen gene family support the notion that tRNA repertoire is fine-tuned in the infected cell (reviewed in [100]). The Schlafen gene family consists of nine Schlafen genes in mice (Sfln), six in humans (SLFN), and are considered interferon-stimulated genes. The size and domain structure of Schlafen proteins vary, with many of the members shown to bind and cleave tRNAs through the N-terminal Schlafen domain[101], although some members can also bind and cleave other nucleic acids. The first report of antiviral action of this family of proteins came from work with HIV-1, where human SLFN11 was shown to counteract changes in tRNA pools upon infection[102, 103]. SLFN11 was shown to bind and cleave tRNAs, leading to reduced translation of non-optimized host and viral transcripts[102, 104, 105]. There is also evidence that DNA viruses can be restricted by Schlafen proteins, including SLFN5/11 (HSV-1, HCMV), SLFN14 (VZV), and Slfn2 (MCMV)[6, 106–108]. There has yet to be a clear dissection of whether the antiviral activity of Schlafen proteins relies on tRNA endonuclease activity. In fact, SLFN5 was shown to inhibit HSV-1 transcription through its binding to the viral genome[107]. Without this knowledge, the role of tRNA modulation during Schlafen antiviral restriction remains correlative. Interestingly, Schlafen domains can be found in orthopoxvirus genomes, meaning that Schlafen protein activity is useful enough to have been co-opted by viruses. Whether viral Schlafens interact with tRNAs has not yet been explored.

5.4. Non-canonical tRNAs as potential modulators of DNA virus infection

One surprise emerging as tRNA sequencing technologies develop is the production of tRFs following diverse viral and bacterial infections. In fact, the first report of tRNA cleavage was following infection of E. coli with T4 bacteriophage, suggesting this may be an evolutionary conserved response[109]. Interestingly, tRNA cleavage by Angiogenin is inducible by NFκB signaling upon infection with Mycobacterium, although this has yet to be demonstrated with viral infection[110]. The tRF cleavage products produced from this endonucleolytic event can have measurable function as gene expression control and/or signaling molecules in mammalian cells. tRFs have been shown to play diverse antiviral roles in the context of infection by multiple RNA viruses (reviewed in [5]). During hepatitis C virus infection, fragments derived from 3’ trailers of pre-tRNAs bind La protein, blocking its access to viral mRNAs which depend on La chaperone activity during translation[111]. A tRF complementary to the primer-binding site of HIV was shown to be expressed in HIV-infected cells, bound by the RISC complex, and capable of blocking reverse transcription of the viral genome[112]. Respiratory syncytial virus infection is also associated with increased tRF expression, including a proviral 5’ tRF that can bind and silence a host antiviral mRNA[113]. Another interesting mechanism described for tRFs formed during Mycobacterium infection is their secretion from cells in exosomes, leading to activation of TLR signaling in nearby cells[110]. Though the described functions of tRF are diverse, they ultimately play a role in fine-tuning the translational landscape during infection. Although tRFs have been reported in response to DNA virus infection[66], systematic analysis and tRF functionality in the context of DNA virus infection has not yet been explored.

During the innate immune response to viral infection, viral components are sensed by receptors that activate the type I interferon response and antiviral genes. One of these receptors, RIG-I, is an RNA sensor recognizing key structural features, including 5’ triphosphates and double stranded stretches, abundant during RNA virus infection. Interestingly, DNA viruses can also be sensed by RIG-I, leading several groups to identify RIG-I ligands during DNA virus infection[114, 115]. Surprisingly, identified ligands include host Pol III transcripts, including 5S pseudogenes and vault RNAs, that are not properly processed or shielded by protein interactors during infection. While tRNAs have not been found to engage with RIG-I, it remains possible that abundant misprocessed pre-tRNAs or tRFs could amplify the innate immune response to viral infection.

6. Conclusions- tRNA regulation by DNA virus infection is a field ripe for discovery

Given the fundamental roles of tRNAs in gene expression (both canonical and emerging non-canonical activities), the observation that tRNA transcription is enhanced during DNA virus infection is a significant and underexplored outcome of infection. There are many remaining questions in the field:

1. What is the full extent of tRNA and tRF expression changes during viral infection?

2. Are there virus-induced changes in tRNA functionality that might be invisible by sequencing? i.e., charge or modification status?

3. What are the functional outcomes of tRNA and tRF upregulation, including both translational output and a potential role in perpetuating the innate immune response?

These questions are a jumping off point in the field as we continue to explore the intersection between DNA viruses and tRNAs. Technological advances, primarily in tRNA sequencing, offer promising tools to study the expression of tRNAs and tRFs following DNA virus infection and the impact viral infection has on translational output. Continued exploration of this intersectionality will allow for functional characterization of viral infection-induced tRNAs and tRFs as drivers of an innate immune response and give insight into the potential of tRNA-based antivirals.

Abbreviations:

tRFs

tRNA-derived fragments

Pol III

RNA polymerase III