In order to highlight the potential of tRFs to enhance therapeutic small RNA development, it is important to understand their endogenous cellular roles. The first report of a biological function for tRFs was reported for tRF-1001 (Lee et al., 2009). This tRF was highly abundant in cancer cell lines and was found to be required for cell proliferation. Of note, no detectable miRNA-like function was found for tRF-1001 in this study. The mechanism by which tRF-1001 regulates cellular proliferation still remains to be determined, but a miRNA-like mechanism is unlikely.

The lack of miRNA-like functions for tRF-1s was also established in another study (Haussecker et al., 2010). Transfection of a reporter and an oligonucleotide antisense to cand45 (tRFdb ID tRF-1001) did not derepress the luciferase reporter. Surprisingly, the reporter was further repressed by anti-tRF-1001 introduction. These results may be explained by a mechanism in which addition of an antisense oligo stabilizes tRF-1001, rather than interfering with its function. Interestingly, a more global meta-analysis of tRFs associated with AGO proteins revealed a striking absence of tRF-1s with AGO1-4 (Kumar et al., 2014). Both results make it clear that tRF-1s generally lack base-pair mediated repression of target mRNAs under basal conditions.

In the one case that tRF-1s are associated with miRNA-like repression, the tRF-1 is generated by DICER1 cleavage. DICER1 cleavage of tRNAs can occur when the precursor tRNA maturation protein, SSB/La, is depleted (Hasler et al., 2016). Under these conditions, RNase Z cannot cleave the precursor trailer sequence, allowing the tRNA to possibly form a hairpin, a substrate amenable to DICER1 cleavage. Overexpression of the tRF-1, either via mimic transfection, shRNA, or parental tRNA mediated overexpression led to repression of a luciferase reporter. Although these data clearly support a role for tRF-1 mediated repression, it is unclear if the effects are mediated by the SSB/La-DICER1 regulatory axis, as the parental tRNA overexpression followed by luciferase reporter was not conducted in SSB/La or DICER1 deficient cells.

Unlike tRF-1s, tRF-3s have been shown to readily enter RISC and repress gene expression in a base-pairing specific manner. As alluded to previously, overexpression of parental tRNAs leads to a direct upregulation of tRF expression. Parental tRNA overexpression leads to tRF-3001a, -3003a, and -3009a generation and entry into AGO for efficient repression of luciferase reporters (Kuscu et al., 2018). Haussecker et al. (2010) also showed that type I tRFs (tRF-3s) can behave like miRNAs because transfection of an antisense oligo derepressed a luciferase reporter. In the context of cancer, CU1276 (tRF-3027b) was found to be absent from germinal center derived lymphomas, present in normal germinal centers, and can downregulate RPA1 in a sequence specific manner (Maute et al., 2013). For tRF-3 the evidence is strong that the repression does not require DICER1, because tRNA overexpression mediated repression of tRF-3 reporters continues in DICER1 knock-out cells (Kuscu et al., 2018). Importantly, the repression of tRF-3 targets when tRNAs are overexpressed requires AGO and requires seed-match with the 3’ UTR of the target (Kuscu et al., 2018).

The difference between the ability of tRF-1s and tRF-3s to repress target gene expression is unclear. The answer may lie in the inherent differences between the two classes: 1) different biogenesis enzymes (Vogel et al., 2005); 2) tRF-1s are likely to be single stranded (Lee et al., 2009), while the biogenesis of tRF-3s may involve a double stranded intermediate; or 3) undescribed sequence or modification difference; 4) differential association with RISC accessory proteins (Haussecker et al., 2010).

Another report found that two tRF-3s and one tRF-5 produced by rhizobial bacteria are enriched in soybean root nodules (Ren et al., 2019). These tRFs appeared to regulate soybean host genes in a miRNA-like manner because rhizobial tRF targets were predicted using plant miRNA target rules, overexpression of mimic rhizobial tRFs repressed predicted targets, and depletion of rhizobial tRFs resulted in an increase in predicted targets. A tRF-5 was also identified to regulate gene expression post-transcriptionally via base-pairing in the context of respiratory syncytial virus (RSV) infection (Deng et al., 2015). In this case, a tRF-5 derived from GluCTC downregulated expression of APOER2, which enhanced RSV replication. These findings highlight the ability of tRFs to behave like bona fide miRNAs, function in viral pathogenesis, and function across the kingdoms of life.

Perhaps one of the most important roles for tRFs may be in the function of stem cells and germ cells, hinting at a fundamental and primordial function for this class of small RNAs derived from tRNAs. tRFs have been found to be highly expressed in mouse stem cells where they suppress retrotransposition of endogenous retroviruses (Schorn et al., 2017). The 18 nucleotide tRF-3as accomplish this by blocking reverse transcription. 22 nucleotide tRF-3bs suppress retrotransposition via post-transcriptional silencing of retrotransposon gene expression. tRF’s ability to interact with a retroelement may not come as a surprise, since retroviruses usurp host cell tRNAs for priming in reverse transcription (Mak and Kleiman, 1997). In another example of tRF’s roles in early embryogenesis and germ cells, tRFs were found to be highly abundant in mature mouse sperm (Sharma et al., 2016; Sharma et al., 2018). tRF-5 from tRNA-GlyGCC (tRF-GG) was determined to be a repressor of MERVL via regulation of histone protein level and Cajal-body dependent noncoding RNAs (Boskovic et al., 2020). Finally, it has been shown that tRFs derived from tRNA primers can inhibit HIV reverse transcription (Yeung et al., 2009), or enhance HTLV-1 reverse transcription (Ruggero et al., 2014).

Interestingly, most studies that determine tRF mediated repression, either endogenously or via tRNA or tRF mimic overexpression, observe at most a 40–60% reduction in target luciferase reporter expression (Haussecker et al., 2010; Maute et al., 2013; Kuscu et al., 2018). This is perhaps due to the fact that most tRFs appear to enter AGO1, 3 and 4, rather than AGO2 which has slicer activity. Overexpression of either AGO2 (Haussecker et al., 2010; Maute et al., 2013) or knockout of DICER1 (Kuscu et al., 2018) in these systems leads to enhanced repression. These data suggest that availability of AGO2 binding is essential for miRNA-like efficiency in repression. This makes potential tRF-like therapeutics less useful if maximal repression of the target gene is required, which may lead to questioning of the utility of tRF mediated post-transcriptional gene silencing as a therapeutic approach. However, the moderate attenuation of gene expression mediated by tRFs may prove useful in disease contexts in which gene expression needs to be knocked down, but some degree of expression is still required for normal function. For example, Beckwith-Wiedemann syndrome is an imprinting disorder in which IGF2, among other genes, expression is too high (Wang et al., 2019; Borjas Mendoza and Mendez, 2021). IGF2 is a necessary gene for intrauterine growth and development, so too much repression of this gene would be detrimental. In fact, loss of IGF2 expression leads to Russell-Silver syndrome, a congenital growth syndrome which leads to failure to thrive (Saal et al., 2002). Thus, the correct balance of IGF2 expression needs to be achieved, and a tRF-like mechanism may be more amenable to this sort of repression, rather than siRNA or miRNA-like approaches.

One of the newest functions of tRFs is in nascent RNA silencing (Di Fazio et al., 2022). DICER1 dependent tRFs were found to target introns of genes in the nucleus in an AGO2 dependent manner. Nuclear functions of RISC have been reported (Gagnon et al., 2014), however, this is the first time tRFs have been attributed with this function. One potential reason tRFs may be uniquely suited to regulate nuclear gene expression either co- or post-transcriptionally may be their single stranded nature. David Corey and colleagues reported that single stranded small RNAs can readily bind nuclear AGO2, whereas double stranded siRNAs can only bind cytoplasmic AGO2, likely due to the nucleus missing accessory double stranded RNA loading factors.

There may be several reasons that tRFs are able to engage in nascent RNA silencing besides their single strandedness. One underexplored feature may be the modifications present on tRFs. It is possible that certain modifications allow tRFs to enter the nucleus and regulate gene expression. Although exciting, a more thorough understanding of nascent RNA silencing machinery and biochemistry will be necessary if it is to become a potential therapeutic modality.

tRF-3s, and under certain conditions, tRF-1s, have been shown to regulate gene expression post-transcriptionally via base-pairing (Haussecker et al., 2010; Maute et al., 2013; Deng et al., 2015; Kuscu et al., 2018; Ren et al., 2019). Sobala and Hutvagner suggested that tRF-5s do not function in this way, and in fact are capable of reducing mRNA translation in the absence of base-pairing (Sobala and Hutvagner, 2013). In particular, a “GG” dinucleotide appears to be necessary, but not sufficient, to mediate repression of a luciferase reporter (Sobala and Hutvagner, 2013). These tRFs repress translation by associating with polysomes. This work is in line with previous work showing that tRNA fragments and halves derived from the 5′ end of tRNAs generally inhibit translation across the domains of life (Yamasaki et al., 2009; Zhang et al., 2009; Ivanov et al., 2011; Gebetsberger et al., 2012). tiR-5s can be produced under stress conditions. For example, stress induced by arsenite treatment, heat shock, and UV radiation induced cleavage of tRNAs via ANG, which led to translation inhibition in a phospho-eIF2⍺ independent manner (Yamasaki et al., 2009). In a follow-up study, tiRs were found to work with YB-1 to displace eIF4G/A and eIF4F from uncapped mRNAs and m7G caps, respectively (Ivanov et al., 2011). Four to five guanines, or 5′ terminal oligo guanine (5′-TOG) motifs, were also found to be important for translation inhibition in this study, as tiR-5s from tRNA-Ala and tRNA-Cys with 5′-TOG motifs inhibit translation much more efficiently than tiRs without the motif. In sum, tiR-5s and tRF-5s seem to reduce global translation, and multiple guanines seem to be important for this mechanism.

As opposed to tiR-5s, much less seems to be known about tiR-3 function. A tiR-3 derived from tRNA-ThrAGU was found to be upregulated in starvation conditions in Trypanosoma brucei, one of the parasitic protozoans that causes the neglected disease sleeping sickness (Fricker et al., 2019). In contrast to tRF-5s and tiR-5s, the tiR-3 stimulated translation during starvation recovery (Fricker et al., 2019). Perhaps targeting this tiR-3 and preventing translation stimulation in post-starvation protozoa could be an alternative therapeutic approach for sleeping sickness caused by this subspecies.

The ability of tRFs and tiRs to regulate global translation sets up a potential paradigm for designing a new class of small RNA therapeutics. tRF-5s and tiR-5s may be designed to inhibit translation in disease states where an aberrant upregulation in translation is essential for pathogenesis. For example, mRNA translation is an essential process in rapidly proliferating cancer cells (Malina et al., 2012). The mTOR inhibitor everolimus is used alone or as part of chemotherapeutics regimens in breast cancer, renal clear cell carcinoma, subependymal giant cell astrocytoma, and advanced neuroendocrine tumors (DRUGBANK, 2022). In the case of renal clear cell carcinoma, patients often develop resistance to mTOR inhibitors everolimus and temsirolimus (Voss et al., 2011). In the future, it will be interesting to evaluate the effectiveness of tRF-5/tiR-5 mediated global translation inhibition as a cancer therapy.

The benefit of tRF and tiR mediated translation inhibition as a therapeutic approach is that it does not require the target cell to contain or efficiently use the RNA interference machinery. This is especially important in the context of bacteria, which does not have a system analogous to RNA interference. Antimicrobial resistance is one of the top 10 global health threats to humanity according to the World Health Organization (WHO, 2022), therefore novel approaches are needed to address this issue. tRNA fragments were first described in E. coli (Levitz et al., 1990), but only recently has research into the global expression and biological roles of tRFs in microbes been conducted (Li and Stanton, 2021). Of note, several currently used antibiotics are small molecule inhibitors of translation (Chellat et al., 2016), positioning tRF and tiR mediated inhibition of translation as a viable approach. The development of antisense oligonucleotides as antibiotics is of great interest due to the ease of design, alterations in response to resistance, and clear targeting (Kole et al., 2012; Xue et al., 2018). Delivery of oligonucleotides is a grand challenge in eukaryotic oligonucleotide therapeutics (Hammond et al., 2021), and this challenge is no different in prokaryotes. However, conjugating oligonucleotides to peptides, vitamin B12, or encapsulation in nanoparticles has greatly improved the delivery of oligonucleotides into bacteria (Xue et al., 2018). It is also becoming clear that microbes (including Gram positive bacteria) are able to package RNAs and other cargo into extracellular vesicles (Domingues and Nielsen, 2017; Liu Y. et al., 2018; Toyofuku et al., 2019). These vesicles enable cross-talk between other bacteria as well as the host. As mammalian extracellular vesicles and other nanoparticles are being developed as vectors for carrying small RNA therapies (Bost et al., 2021), it is conceivable that bacterial vesicles may also function as drug delivery vehicles. This seems to be the case and has been an active area of study (Li and Liu, 2020). Most antibiotic oligonucleotide therapy development focuses on antisense technology, however, tRFs and tiRs that inhibit translation could also be conjugated or encapsulated into delivery vehicles and could provide another tool in the antibiotic oligonucleotide therapy toolbox.

Of note, one limitation of tRF/tiR antibiotics might be the role that tiRs appear to play in the regulation of the RNA repair operon (Hughes et al., 2020). The function of this operon is important for survival following DNA damage, and tiR-5s bind the CARF domain of RtcR, leading to oligomerization and activation of the RNA repair operon. Potentially giving a growth advantage to bacteria with DNA damage may be circumvented by utilizing tiR based therapies without a 3′ cyclic phosphate (3′ cP), as a 3′ cP was necessary for optimal CARF domain binding. The role of alternative 3′ ends in tRF/tiR mediated translation inhibition requires further study.

Delivery of oligonucleotides to their site of action is a major challenge (Dowdy, 2017; Roberts et al., 2020; Hammond et al., 2021). tRF based therapeutics would have many of the same delivery issues. Oligonucleotide delivery can be broadly classified into two steps: 1) tissue delivery, and 2) cytoplasmic delivery or endosomal escape. Tissue specific delivery of oligonucleotides is more amenable to organs that are involved in blood filtration, such as the liver and kidney. Tissue delivery is also currently more amenable to organ systems that can be directly accessed via injection, such as the eye or brain and spinal cord by intrathecal injection. Furthermore, Alnylam has had success with liver specific delivery of siRNA-based therapies that are conjugated to N-acetylgalactosamine (GalNAc) (Springer and Dowdy, 2018). Lipid nanoparticles (LNPs) also enable more targeted delivery into tissues. LNPs have been successfully used for delivery of COVID-19 mRNA vaccines and the recently approved PCSK9 siRNA for the treatment of hypercholesterolemia (Fitzgerald et al., 2017; Buschmann et al., 2021).

Delivery of small RNAs to tissues may also be enhanced by encapsulation in extracellular vesicles (O’Brien et al., 2020). In support of the use of extracellular vesicles as tRF and tiR delivery agents, tRFs and tiRs are enriched in T cell extracellular vesicles (Chiou et al., 2018). tRFs were packaged into extracellular vesicles and released from activated T cells. Reduced expression of tRFs that are normally packaged and released leads to enhanced T cell activation. Although the effect of endogenous extracellular vesicle tRFs and tiRs on recipient cells is underexplored, it is clear that overexpressed tiRs can be delivered to recipient cells and alter gene expression (Gámbaro et al., 2020). Mouse epididymosomes can also deliver tRF-5s to maturing sperm, which ultimately represses MERVL (Sharma et al., 2016).

Another challenge for tissue delivery is oligonucleotide stability in circulation (Paunovska et al., 2022). Much of this issue has been solved, either by encapsulation of the oligonucleotide in an LNP, or by installment of RNA modifications (Yu et al., 2019). Interestingly, tRNAs, and likely tRFs, are one of the most highly modified RNAs in the cell, which may confer some enhanced stability in the circulation (Pan, 2018).

The second challenge of delivery, endosomal escape, is still a major unresolved bottleneck in the oligonucleotide therapeutics field (Dowdy, 2017). Small molecules such as nigericin and chloroquine have been found to enhance endosomal escape (Heath et al., 2019; Orellana et al., 2019), however, these compounds are too toxic for clinical use (Brown et al., 2020). It does appear that LNP platforms are better suited for endosomal escape than GalNAc conjugated siRNA mediated delivery, but this comes at the expense of duration of action (Brown et al., 2020). Any tRF or tiR therapeutic would also need to overcome these barriers.

Although tRNA based therapies are outside the scope of this review, it is a burgeoning field of development because nonsense suppressor tRNAs can prevent the formation of truncated, dysfunctional proteins (Porter et al., 2021). It is clear, however, that overexpressed tRNAs produce tRNA fragments. These tRNA fragments may cause off-target effects and lead to unpredictable side effects. In order to circumvent these effects, the biogenesis of tRFs and tiRs must continue to be thoroughly investigated. For example, it is clear that certain modifications can affect the production of tRFs ((He et al., 2021; Nagayoshi et al., 2021) and reviewed in (Lyons et al., 2018)). Parental tRNA cleavage induced by reactive oxygen species can also be prevented in T cells by binding Schlafen 2 (SLFN2) (Yue et al., 2021) (Figure 1). Findings like these will enable rational design of stable tRNA therapeutics.