Plant tRNA functions beyond their major role in translation

Abstract

Transfer RNAs (tRNAs) are well known for their essential function as adapters in delivering amino acids to ribosomes and making the link between mRNA and protein according to the genetic code. Besides this central role in protein synthesis, other functions are attributed to these macromolecules, or their genes, in all living organisms. This review focuses on these extra functions of tRNAs in photosynthetic organisms. For example, tRNAs are implicated in tetrapyrrole biosynthesis, mRNA stabilization or transport, and priming the reverse transcription of viral RNAs, and tRNA-like structures play important roles in RNA viral genomes. Another important function of tRNAs in regulating gene expression is related to their cleavage allowing the production of small non-coding RNAs termed tRNA-derived RNAs. Here, we examine in more detail the biogenesis of tRNA-derived RNAs and their emerging functions in plants.

Genome organization, mRNA stabilization, organelle, photosynthetic organisms, regulation of gene expression, short interspersed nuclear elements, tetrapyrrole biosynthesis, tRNA-derived RNA

Introduction

Transfer RNAs (tRNAs) are likely among the most ancestral and the best-known small non-coding RNAs. With a few exceptions, where they are shorter (Wende et al., 2013), tRNAs have a size of 76–93 nucleotides (nt). They have a characteristic cloverleaf secondary structure (Fig. 1) and adopt an L-shaped tertiary structure (Schimmel, 2018). In eukaryotes, the CCA triplet added post-transcriptionally at the 3ʹ end is the site of aminoacylation. Aminoacylated tRNAs can then play their major role, the incorporation of amino acids into the nascent peptide chains within the ribosome during protein synthesis. In plants, protein synthesis takes place in three compartments: the cytosol, the plastid, and the mitochondrion. In land plants, the plastidial genome encodes a full set of tRNA genes (e.g. 37 tRNA genes corresponding to 30 isoacceptor tRNAs have been identified in Arabidopsis thaliana) to ensure translation in this organelle. Mitochondrial DNA lacks several tRNA genes (e.g. there are 22 tRNA genes corresponding to 17 isoacceptor tRNAs in A. thaliana), thus several nucleus-encoded tRNAs have to be imported from the cytosol into the mitochondria to achieve an active mitochondrial protein synthesis process. In the nuclear genome, the number of tRNA genes varies among plant species. For instance, the A. thaliana nuclear genome has 585 tRNA genes that code for 46 isoacceptor tRNAs (Michaud et al., 2011; Cognat et al., 2022). With the exception of 124 clustered tRNA genes that are epigenetically silenced in A. thaliana (Hummel et al., 2020), it is believed that the others are expressed and active in protein synthesis in this model plant species.

Fig. 1.

Overview of additional functions, besides protein synthesis, of tRNAs in the plant cell. The typical cloverleaf secondary structure of a tRNA, with its D-, T-, and A- regions, is depicted in the center. (A) Plastidial tRNA^Glu is required for the biosynthesis of 5-aminolevulinic acid (ALA), the precursor of tetrapyrolles. (B) A tRNA is used as a primer for the reverse transcription (RT) of RNA of pararetroviruses. The 3ʹ end of the tRNA sequence is complementary to the primer binding site (PBS) located at the 5ʹ end of the viral RNA. (C) The different roles of tRNA-like structures (TLSs) found at the 3ʹ end of various plant viral RNA genomes. TLSs can be charged with valine (V), histidine (H), or tyrosine (Y). (D) In mitochondria, tRNAs (as illustrated here in Cyanophora paradoxa) or ‘t-elements’ deriving from tRNAs present at the 3ʹ end of mRNAs can prevent their degradation by exonucleases. (E) Discistronic mRNA:tRNA transcripts circulate over long distances via the phloem vasculature. (F) Short interspersed nuclear elements (SINEs) are ancestrally derived from tRNA genes and transcribed by polymerase III due to their internal promoters, the A and B boxes. In a few cases, the classical tRNA cloverleaf structure is conserved.

It is becoming increasingly evident that, besides the canonical role of tRNAs in the translation process, these macromolecules, whether aminoacylated or not, are important players in many other biological processes. This has already been emphasized in excellent reviews (e.g. (Raina and Ibba, 2014; Katz et al., 2016; Schimmel, 2018). To give a non-exhaustive list, charged tRNAs are implicated in the formation of bacterial cell walls, the biosynthesis of antibiotics, the aminoacylation of membrane lipids in bacteria and fungi (Yakobov et al., 2020), and the targeting of proteins for degradation (Kwon et al., 1999; Rai and Kashina, 2005; Graciet et al., 2006). Uncharged tRNAs have a regulatory role during amino acid starvation in yeast and mammals (Hinnebusch, 2005; Avcilar-Kucukgoze and Kashina, 2020) and during cell death. Finally, another crucial regulatory function of tRNAs is the role played by tRNA fragments, which have been termed tRNA-derived RNAs (tDRs) (Holmes et al., 2022, Preprint). First believed to be non-functional molecules that are degradation by-product, tDRs are now largely accepted as essential non-coding RNAs (Liu et al., 2021; Chen et al., 2021). For instance, they were shown to inhibit translation in humans under stress conditions (Yamasaki et al., 2009), to regulate the RNA decay pathway in Tetrahymena thermophila (Couvillion et al., 2012), to regulate apoptosis in mice (Saikia et al., 2014), to be engaged in the RNA silencing pathway to degrade specific mRNAs in various organisms (Kumar et al., 2014; Kuscu et al., 2018), and to be used as primers for viral reverse transcriptase (Ruggero et al., 2014). The implication of tDRs in these molecular processes likely has important biological consequences such as the regulation of haematopoiesis (Goncalves et al., 2016), cancer cell proliferation (Honda et al., 2015; Yu et al., 2020), or epigenetic inheritance (Chen et al., 2016; Sharma et al., 2016). Many more examples of tDR functions are emerging regularly and represent a very rapidly evolving field whatever the organism studied, including plant species.

This review highlights our current knowledge of the functions of tRNAs or tRNA-like structures, apart from the translation process, in photosynthetic organisms. The first part concerns the extra functions of tRNAs or tRNA-like structures, and the second part focuses on tDRs.

Roles of tRNAs and tRNA-like structures apart from translation

Production of organic compounds

With the discovery of the genetic code and the key role played by tRNAs in decoding it, the understanding of tRNA biology was thought to be well covered. Unexpectedly, novel functions outside translation have been attributed to certain tRNAs. Some of these functions, namely the production of tetrapyrroles and cytokinins, are crucial for plant growth and are detailed below.

Plant tetrapyrroles (heme, chlorophyll, siroheme, and phytochromobilin) are essential signaling molecules (Larkin, 2016). They are biosynthesized from the same precursor, 5-aminolevulinic acid (ALA). The pathway of ALA synthesis is located in the stroma of plastids and involves glutamyl tRNA (tRNA^Glu) charged with glutamate (GlutRNA^Glu) (Fig.1A). The plastidial genome of land plants encodes a single tRNA^Glu(TTC). After its aminoacylation with glutamate, GlutRNA^Glu is not only used for the incorporation of glutamate into proteins during plastidial translation but was shown, 35 years ago (Schon et al., 1986), to also be the substrate of glutamyl-tRNA reductase (GluTR). This enzyme catalyzes the conversion of Glu into Glu-1-semialdehyde, which is then transformed into ALA by a second enzyme, Glu-1-semialdehyde aminotransferase. Recent work has, interestingly, demonstrated that GluTR activity is inhibited by unprocessed tRNA^Glu (Agrawal et al., 2020).

Plant hormones such as auxin and cytokinin are essential for plant growth and development. Their homeostasis is thus tightly regulated (Hrtyan et al., 2015). Cytokinins are derivatives of adenine with either an isoprene or an aromatic side chain attached at the N⁶-terminus. The presence of isoprene-cytokinin-containing tRNAs has been observed in different organisms, from bacterial to mammals (Juhling et al., 2009). Cytokinins found in plants, zeatin and isopentenyladenine residues, are located at position 37 of tRNAs. Although de novo biosynthesis of cytokinins is the most common pathway, the prenylation of tRNAs by tRNA isopentenyltransferases followed by their degradation also allows the release of cytokinins and represents a minor pathway for cytokinin biosynthesis (Miyawaki et al., 2006; Kurakawa et al., 2007; Kieber and Schaller, 2014).

These two examples perfectly illustrate how tRNAs or amino acid-tRNAs can be hijacked from the translation process to produce metabolites. Although knowledge is so far restricted to a very few examples in plants, it is tempting to speculate that other alternative functions may exist. In particular, diverting aminoacyl-tRNAs to aminoacylate membrane lipids is a well-known process in bacteria (Katz et al., 2016) and has been recently described in fungi (Yakobov et al., 2020). A similar role in plants remains speculative but worth deciphering.

tRNAs as primers for initiation of reverse transcription in plant pararetroviruses

Surprisingly, there is an astonishing relationship between tRNAs and viruses. The two following sections describe different aspects of this relationship: the first relates to the use of specific host tRNAs to prime the reverse transcription of viral RNAs (Fig. 1B), while the second one is linked to tRNA mimicry with the finding of tRNA-like structures (TLSs) in viral genomes (Fig. 1C).

In 1970, the discovery of an RNA-dependent DNA polymerase (thus termed reverse transcriptase) in two retroviruses, murine leukemia virus (Baltimore, 1970) and Rous sarcoma virus (Temin and Mizutani, 1970), demonstrated that genetic information can be transferred from RNA to DNA. Since then, numerous reverse transcriptases have been described in retroviruses. These reverse transcriptases require specific tRNAs, such as tRNA^Trp for murine leukemia virus or tRNA^Trp for Rous sarcoma virus, as primers to initiate DNA synthesis (Marquet et al., 1995). Of note, the reverse transcription of mobile elements, the retrotransposons, also requires tRNAs as primers. In plants, some double-stranded DNA viruses were found to replicate via an RNA intermediate and were named pararetroviruses. The most well-known representative of these is cauliflower mosaic virus, a virus infecting Brassicaceae. Forty years ago, the involvement of a reverse transcription step in the replication of the virus was demonstrated, and the implication of a tRNA was strongly suspected (Pfeiffer and Hohn, 1983). Indeed, as for retroviruses and retrotransposons, the host initiator tRNA^Meti initiates the replication (Turner and Covey, 1984; Hull et al., 1987) (Fig. 1B). Retroviruses, retrotransposons, and plant pararetroviruses share another feature: the specificity of the reverse transcription priming is ensured by complementarity between the 3ʹ end of the tRNA and the sequence of the primer binding site (PBS). For instance, in cauliflower mosaic virus, there is a 14 nt sequence of perfect homology between the 3ʹ end of tRNA^Meti and the PBS so that a strict initiation of reverse transcription occurs.

As a whole, primer tRNA appears to be essential for the life cycle of plant pararetroviruses. More generally, a small set of specific tRNA species serve as primers of reverse transcription even in distantly related retrotransposons, retroviruses, and pararetroviruses, and we can wonder why tRNAs were selected rather than other abundant RNA species. Is this related to a close evolutionary origin between viruses and tRNAs, and/or to the need to highly regulate the reverse transcription process? These questions remain unanswered to date.

Roles of tRNA-like structures of viral RNA genomes

A number of positive-strand viral RNA genomes lack a poly(A) tail at their 3ʹ end, an important feature for their translation and stability. To cope with this lack, many of them possess a TLS at their 3ʹ end (Fig. 1C). As explained in detailed reviews (Wu et al., 2021; Dreher, 2009), a TLS presents some criteria of a true tRNA molecule: for instance, folding in a cloverleaf secondary structure, recognition by tRNA-specific enzymes (e.g. RNases P or Z, tRNA-nucleotidyl transferase, aminoacyl-tRNA synthetases), and formation of a tertiary complex with GTP and an elongation factor. The first TLS to be identified was in the RNA genome of turnip yellow mosaic virus (Yot et al., 1970). Since then, various TLSs have been found in many plant viral genomes. With the exception of tobraviruses, plant viral TLSs perfectly mimic tRNAs and are aminoacylated by either valine, histidine, or tyrosine. Several important but widely variable functions have been attributed to viral TLSs. For instance, valylation of the turnip yellow mosaic virus TLS is essential to support infectivity, whereas the TLS of tomato aspermy cucumovirus is not even aminoacylated. TLSs are indeed also implicated in the replication step of viruses, in particular as the minus strand initiation start. In some cases, translation enhancement by TLSs has been observed in a manner that resembles a poly(A) tail (Miras et al., 2017). In the brome mosaic virus, the presence of TLSs, by being implicated in RNA recombination, could be an advantage for the virus. In addition, in this virus, TLSs are involved in the encapsidation of the tripartite viral genomes. It has also been suggested that some plant viral RNAs use TLSs as carriers to be transported in phloem sieve tubes (Tolstyko et al., 2020). Finally, the role of TLSs in the stabilization of viral genomic RNAs is also well known. The integrity of the CCA triplet (which is repaired by the host tRNA-nucleotidyl transferase) present at the 3ʹ end of TLS sequences is important in this regard.

tRNAs or tRNA-like structures as stabilizing structures

In the same vein, the presence of tRNA at the 3ʹ end of organellar mRNAs also appears to be an ancestral event. Recent work showed that mitochondrial-encoded tRNAs from the freshwater alga Cyanophora paradoxa have two functions within the organelle: on the one hand, as expected, they are key players of mitochondrial translation and, on the other hand, they are involved in a peculiar punctuation processing mechanism leading to mRNA–tRNA hybrids with a CCA triplet at their 3ʹ end (Fig. 1D) (Salinas-Giege et al., 2021). The presence of highly folded tRNA structures at the 3ʹ end of mRNAs likely impedes exonuclease activities and thus stabilizes the mRNAs until they reach the mitochondrial ribosomes to be translated. Cyanophora paradoxa, a glaucophyte representing the most basal group of Archeplastida, is considered to be a living fossil among photosynthetic organisms. The existence in this organism of mitochondrial tRNA genes, located just downstream of protein-coding genes, serving as stabilizing factors is reminiscent of the presence of numerous tRNA pseudogenes or TLSs (also called t-elements) located close to RNA ends in the mitochondria of higher plants such as Arabidopsis (Forner et al., 2007) and potato (Varre et al., 2019). It is tempting to speculate that such ‘t-elements’ are derived from true tRNA genes; they progressively became pseudogenes and lost their ability to be correctly processed and active in translation in the mitochondria of higher plants but still kept another essential function in mRNA stabilization.

Transport of tRNA:mRNA hybrids

Similar to viral RNAs and organellar mRNAs, the presence of tRNAs or TLSs as part of plant cytosolic mRNAs represents an additional example of their key role in gene expression regulation and signaling. Indeed, the mobility of mRNAs represents an important aspect of cell-to-cell communication. The mobility of RNAs and their importance as signaling molecules has been described in detail (Maizel et al., 2020). Grafting experiments in Arabidopsis demonstrated that mRNA:TLS transcripts can move bidirectionally in phloem sieve elements and that the TLS seems to be determinant for the long-distance transport of mRNAs between tissues (Fig. 1E) (W. Zhang et al., 2016). These hybrids appear to be enriched in the population of graft-mobile mRNAs. Interestingly, the authors also showed that, once transported, these mRNAs can be translated. The presence of 19 dicistronic tRNA:mRNA transcripts was also demonstrated in grapevine (Fabres et al., 2021). The authors suggest that the use of TLSs as mobile signals for mRNA transcripts to move over long distances is linked to the appearance of plant vasculature. The functional role of these mobile mRNAs in distant tissues and the mechanism allowing the transport of mRNA:TLS transcripts remain to be further addressed.

Collectively, it is striking to note the large number of cases in which TLSs are found associated with either viral RNAs or mRNAs, and the number of functions in which they may be implicated (e.g. replication, stabilization, transport, translation). Is this reminiscent of the world of RNA and why is it so well conserved during evolution? These intriguing questions could be related to the origin of life.

Short interspersed nuclear elements

The last example illustrating the importance of tRNAs, apart from their role in protein synthesis, is their implication in the emergence of non-autonomous retrotransposons, the short interspersed nuclear elements (SINEs). Again, this strongly suggests that primordial ancestral tRNAs likely already had the potential to be drivers of evolution in the ‘RNA world’.

SINEs are short (100–600 bp) non-autonomous retroelements transcribed by RNA polymerase III (Pol III) (Kramerov and Vassetzky, 2011). They are found in a wide variety of eukaryotic genomes and rely on long interspersed elements for their propagation. The best-known family is the Alu family. In most cases, SINEs are ancestrally derived from tRNA genes. SINEs derived from tRNAs have a composite structure (Fig. 1F). They comprise a TLS at their 5ʹ end, with A and B boxes that are characteristic of Pol III internal promoters, followed by a tRNA-unrelated portion and a poly(A)-rich sequence at the 3ʹ end. Since the discovery of the first SINEs in plants, in the Waxy gene of Oryza sativa (Umeda et al., 1991), many other SINEs have been identified in photosynthetic organisms (Seibt et al., 2016). For example, various families have been identified in A. thaliana, in Brassica species (Pelissier et al., 2004), and in other plants (Gadzalski and Sakowicz, 2011). A family of 35 SINE members has been found in Populus trichocarpa (Michaud et al., 2011), and more than 200 copies of such elements have been identified in the unicellular green alga Chlamydomonas reinhardtii (Cognat et al., 2008). Remarkably, whereas in most cases the typical tRNA cloverleaf secondary structure is difficult to retrieve, in both C. reinhardtii and P. trichocarpa the SINE tRNA-related sequences have retained mostly perfect cloverleaf structures. Compared with humans, in which SINEs represent more than 10% of the genome, they usually correspond to less than 0.3% of plant genomes. While the roles of SINEs have been well investigated in humans (Zhang et al., 2021), only limited information is available about the functions of plant SINEs. In Solanaceae, Seibt et al. (2016) demonstrated that SINE retrotransposons are important contributors to gene and genome evolution: SINEs were shown to be involved in an increased length of untranslated regions and introns, to be responsible for the formation of splice sites, exons, and start and stop codons, or to form tandem-like structures. The genome-wide impact of the Au SINE family was extensively studied in wheat species (Ben-David et al., 2013). This work supports the existence of Au SINE activity throughout the evolution of wheat and its polyploidization, followed by epigenetic regulation via cytosine methylation. The Au SINE family has thus likely contributed to the diversification of wheat species by inducing genetic and epigenetic changes. As a whole, tRNA-related SINE retroelements are likely important players in the evolution and organization of the plant nuclear genome, but overall this remains largely poorly understood and undoubtedly other regulatory functions of plant SINEs remain to be deciphered.

The emerging world of tRNA-derived RNAs

High-throughput small RNA sequencing technologies have greatly expanded the repertoire of small non-coding RNAs. Among them, the number of known RNA fragments derived from tRNAs has grown considerably in recent years. Produced through enzymatic cleavage of tRNAs, these small RNAs have been described in all domains of life, including plants, as novel actors in the regulation of genome expression (Megel et al., 2015; Kumar et al., 2016; Magee and Rigoutsos, 2020; Alves and Nogueira, 2021). Many names and acronyms have been assigned to the small RNAs derived from tRNAs: tRFs, tsRNAs, tiRNAs, and tRNA halves, among others. Recently, a consensus nomenclature has been proposed with the name tDRs, for ‘tRNA-derived RNAs’ (Holmes et al., 2022, Preprint). A bioinformatic tool, tDRnamer (http://trna.ucsc.edu/tDRnamer/index.html), allows standardization of the names assigned to tDRs: the name takes into account the size of the tDR, its position relative to its corresponding tRNA, and the nature of the tRNA. A suffix indicates whether the RNA sequence can correspond to several tRNA transcripts and their number, as well as the possible modifications of the tRNA carried by the tDR. Here, we have adopted the novel abbreviation tDR and we present our current knowledge of the population, biogenesis, and functions of tDRs in photosynthetic organisms.

The population of tDRs

To characterize tDR populations, classical high-throughput small RNA sequencing methods are commonly used. They usually require 5ʹ-monophosphate (5ʹ-P) and 3ʹ-hydroxyl (3ʹ-OH) for the ligation of adapter molecules, but recent data show that other types of ends (e.g. 5ʹ-OH, 5ʹ-ppp, 3ʹ-p, 2ʹ3ʹ-cyclic phosphate, or 3ʹ-aminoacyl), not compatible with such classical ligation procedures, may exist (see e.g. Wang et al., 2021). In addition, nucleotide modifications such as methylation strongly impede the reverse transcription step and result in stops or misincorporations. Thus, it is important to note that, due to these sequencing biases (e.g. blocked extremities, modifications) and limitations in the length of the analyzed RNAs, part of the tDR population is likely missing in current analyses. Novel approaches such as ARM-seq (Cozen et al., 2015), PANDORA-seq (Shi et al., 2021), CPA-seq (Wang et al., 2021), and RTcB sRNA-seq (Gu et al., 2022) represent important tools to enlarge the repertoire of tDRs.

Nevertheless, the data obtained so far have already provided interesting insights regarding tDR pools. tDRs can be divided into two groups according to the region of the tRNA from which they derive and according to their size (Fig. 2A). Short tDRs (~15–28 nt) originate from the 5ʹ end of mature tRNA after cleavage in the D region (tDR-5D) or from the 3ʹ end after cleavage in the T region (tDR-3T). Long tDRs (~30–40 nt) correspond to mature tRNA halves generated through cleavage in the anticodon region (tDR-5A and tDR-3A).

Fig. 2.

Overview of the biogenesis of tRNA-derived RNAs (tDRs). (A) The major classes of short (tDR-5D and tDR-3T) and long (tDR-5A and tDR-3A) tDRs originate from nucleus (N)-encoded, plastid (P)-encoded, and mitochondrion (M)-encoded tRNAs. (B) tDR populations fluctuate under stress conditions (e.g. phosphate deprivation, UV, or heat) or depending on the tissue or developmental stage (e.g. roots, leaves, flowers, and siliques). (C) In pollen, the Dicer-like ribonuclease DCL1 has been shown to be involved in the production of short tDRs. (D) The RNases T2 (RNS1–3) are the major enzymes acting in the biogenesis of short and long tDRs in various tissues.

In plants studied so far, tDRs originating from mature tRNAs represent more than 99.8% of the population of tDRs. tDRs derived from precursor tRNAs (in particular from the 3ʹ trailer sequences released after cleavage by RNase Z) are an exception (Cognat et al., 2017; Alves et al., 2017), unlike in mammals, in which they are more abundant (Liao et al., 2010). Interestingly, the population of tDRs is derived from only a part of different tRNA species. For instance, only 12 of 46 nuclear and 10 of 30 plastidial isoacceptor tRNAs are predominantly cleaved to form tDR-5Ds (Cognat et al., 2017). Among the short tDRs, those of 19 nt and 20 nt are the most abundant. The tDR-5Ds result mostly from cleavages in the D-loop and to a lesser extent in the D-arm. The opposite is true for tDR-3T, where cleavages are more abundant in the T-arm than in the T-loop. Zahra et al. (2021) analyzed small RNA libraries from six angiosperm species (A. thaliana, Solanum lycopersicum, Cicer arietinum, Medicago truncatula, O. sativa, and Zea mays) and obtained similar results. A few internal tDRs were also retrieved.

In Arabidopsis, tDRs from nuclear, plastidial, and mitochondrial tRNAs were found. Three-quarters of the identified tDRs are from nuclear tRNAs, about one-quarter are of plastidial origin, and less than 1% are from mitochondrial tRNAs (Cognat et al., 2017). The presence of the different classes of plastidial tDRs were also found in Brassica rapa (Wang et al., 2011).

Within different tissues (seedlings, leaves, and flowers), the tDR population slightly varies. In contrast, the population in roots shows notable differences. This difference is due not only to changes in the plastidial population of tDRs (because of the absence of chloroplasts in roots compared with photosynthetic tissues) but also to fluctuations in the nuclear tDR population (Cognat et al., 2017; Wang et al., 2011; Alves et al., 2017). Globally, few tDR-3Ts are found in roots. Concerning tDR-5Ds, four of them represent more than 75% of the tDR-5D population in roots [Arg (ACG), Arg (TCG), Cys (GCA), and Gly (TCC)]. An accumulation of tDR-5Ds has also been observed in mature pollen grains (Martinez et al., 2017).

Importantly, it is now well described that the accumulation of several tDRs is induced by various stresses (Fig. 2B). For instance, oxidative stress triggers the accumulation of a variety of short and long tDRs (Thompson et al., 2008; Alves et al., 2017). Variations in the population of tDRs were also observed in Arabidopsis, wheat, rice, and turnip under various abiotic stresses (e.g. salt, drought, cold, heat, or UV) (Loss-Morais et al., 2013; Alves et al., 2017; Cognat et al., 2017). Up- or down-regulation of specific tDRs was observed, and the variations were not always the same depending on the plant species. Whether this reflects different biological functions or biases due to the various approaches used remains to be studied in more detail. In addition, Byeon et al. (2017, 2019) have demonstrated a transgenerational response to heat stress implicating a differential accumulation of tDRs in Brassica rapa. They first observed a down-regulation of glutamyl tDR-5D and an up-regulation of aspartyl tDR-5D in different organs upon heat stress. Then, they showed that, in the progeny of the stressed plants, there was still an accumulation of aspartyl tDR-5D, whereas other tDRs (Ala, Arg, and Tyr) were reduced. The authors propose that tDRs could contribute to the improvement of plant responses to future stress via the regulation of several molecular pathways, but this remains to be established. Accumulation of specific tDRs (in particular Gly and Asp tDRs) was also found during phosphate starvation in Arabidopsis (Hsieh et al., 2009, 2010; Megel et al., 2019), changes in tDR abundance depending on phosphorus availability was also observed in barley (Hackenberg et al., 2013), and a global decrease in tDRs was obtained following treatment of tomato leaves with the phytohormone abscisic acid (Luan et al., 2020). Finally, biotic stresses were also shown to influence the population of tDRs. For example, in wheat, tDRs deriving from tRNA Glu (CTC), Lys (CTT), and Thr (CGU) are highly expressed during Fusarium graminearum infection (Sun et al., 2022).

Biogenesis of tDRs

The production of tDRs requires the cleavage of tRNA molecules by endoribonucleases. Many enzymes with this function have been identified in organisms ranging from bacteria to eukaryotes during the past decade (e.g. Megel et al., 2015; Chen et al., 2021). A non-exhaustive list of these includes Colicin in Escherichia coli, the RNases T2 in Saccharomyces cerevisiae and Tetrahymena thermophila, and a multitude of nucleases in mammals, such as ELAC2, RNase L, the RNase A Angiogenin, the RNase III Dicer, and nucleases of the Schlafen family (Yang et al., 2018). Most of these nucleases cleave tRNAs in the anticodon region to generate long tDRs; very few data are available concerning the biogenesis of short RNAs after cleavage in the D- or T- regions. This complexity shows the difficulty in defining precisely which enzyme does which job, and much remains to be done to address this question. This is particularly true in plants (Fig. 2C, D). While Dicer-like 1 (DCL1) has been proposed to generate specific tDRs in pollen grains (Martinez et al., 2017), other data in which plant DCLs are not involved in tDR biogenesis have presented by two research teams (Alves et al., 2017; Megel et al., 2019). Rather, both teams demonstrate that the biogenesis of both short and long tDRs in leaves, flowers, seed coats, and endosperms relies on the activity of endoribonucleases belonging to the RNase T2 family, named RNSs. Indeed, in Arabidopsis grown under phosphate starvation conditions, the induction of RNS1 is correlated with the accumulation of specific tDRs previously observed by Hsieh et al. (2010). How can we explain such a discrepancy? Each Arabidopsis RNS has a specific expression profile but together all the members of the family cover the whole plant, including flowers and pollen, thus we expect a rather broad and constitutive activity of tRNA cleavage by this family of endoribonucleases. DCLs, which are also highly expressed in pollen, may be responsible for only a very limited, but essential, number of specific tRNA cleavages, especially to produce specific tDRs such as those of 19 nt. To generate tDR-5Ds of 19 nt in length, tRNA cleavage occurs in the D-loop. As DCL enzymes usually cleave double-stranded RNAs, what also remains to be elucidated is the mode of action of these enzymes to produce tDRs. How DCLs, and also RNSs, interact with tRNAs remains unknown and awaits structure/function analysis.

Another essential parameter to take into account for the biogenesis of tDRs is the presence of numerous modifications on eukaryotic tRNAs (Suzuki, 2021). These modifications, made by either promoting or inhibiting tRNA cleavages, will affect the abundance of tDRs (Lyons et al., 2018; Chen et al., 2021). For instance, in Drosophila melanogaster, under heat stress, the presence of a methylated cytosine at position 38 (m⁵C) of tRNA^Asp, tRNA^Gly, and tRNA^Val reduces the production of tDR-5A (Durdevic et al., 2013). By contrast, in humans, decreased methylation of cytidines 48 and 49 (m⁵C) increases tDR-5A production. This is due to an increased affinity between Angiogenin and unmodified tRNAs (Blanco et al., 2014). In plants, tRNAs are also modified, and these modifications may vary under stress conditions (Wang et al., 2017a) or during plant development (Pfitzinger et al., 1990). However, as yet, no data are available on the roles played by tRNA modifications in the production of tDRs in these organisms. It is tempting to speculate that, as for the other organisms studied so far, their presence or absence will impact tRNA cleavage in plants. This is still an open question that could be addressed by analyzing Arabidopsis knockdown mutant lines of modifying enzymes.

Roles of tDRs

In all organisms, the population of tDRs appears to be complex and comprises thousands of small non-coding RNAs. Many of these tDRs are likely to be degradation by-products related to tRNA turnover or tRNA quality control (Megel et al., 2015). This is also an efficient way to recycle nutrients and deliver phosphate and nitrogen to the cells. Nevertheless, there is now growing evidence that tDRs have additional functions that have important biological consequences (e.g. in cancer, neurological diseases, viral infection, and epigenetic inheritance), and several reviews have listed the multiple roles they have been attributed so far and the underlying molecular mechanisms (Schimmel, 2018; Magee and Rigoutsos, 2020; Su et al., 2020; Chen et al., 2021). In plants, the functions of tDRs are still poorly understood (Fig. 3).

Fig. 3.

Overview of the functions of plant tDRs. (A) tDRs can associate with the Argonaute protein AGO1 in the RISC complex to target and cleave RNAs such as transposable elements (TEs). (B) Through their ability to interact with AGO1, bacterial tDRs target and repress genes that prevent symbiosis in Fabaceae. (C) tDRs can modulate the translation process by their ability to associate with polyribosomes.

Specific tDRs were found associated with Argonaute (AGO) proteins (Loss-Morais et al., 2013; Alves et al., 2017; Cognat et al., 2017; Ren et al., 2019; Trolet et al., 2019; Gu et al., 2022), thus suggesting their implication in the regulation of gene expression via the RNA-induced silencing complex (RISC), similar to a microRNA-like pathway. Indeed, Martinez et al. (2017) observed an accumulation of tDR-5D of 19 nt in Arabidopsis pollen grains; this accumulation was also found in the reproductive organs of rice, maize, and the moss Physcomitrium patens. The Arabidopsis 19 nt long tDRs were found incorporated on to AGO1. Bioinformatics analysis to search for target mRNAs with sequence complementarity with the tDRs showed that they could target mostly protein-coding mRNAs but also a smaller proportion of transposable elements (TEs). Finally, the authors demonstrated that a tRD-5D derived from the initiator tRNA^Met is able to target the Athila6A transcript, a member of the Gipsy family of retroelements, and to induce its cleavage (Fig. 3A). As germinative cells lose heterochromatin during epigenetic reprogramming, the use of a tDR as a microRNA to target and degrade TEs may be a way to avoid their reactivation and maintain the stability of the genome. More recently, it was proposed that a tDR-5D derived from tRNA^Ala associates with AGO1 and targets the mRNA coding for the cytochrome 450 71A13, which is involved in the biosynthesis of camalexin, a toxin for pathogens. Upon infection of Arabidopsis with Botrytis cinerea, the abundance of the tDR would be reduced and the production of camalexin increased to derepress the antifungal defense (Gu et al., 2022). In wheat, upon infection with Fusarium graminearum, there is a very high accumulation of tDRs, in particular three deriving from tRNA^Lys(CTT). Potential targets including genes implicated in replication, translation, protein degradation, and resistance to pathogens have been suggested. As yet, there is no experimental evidence demonstrating a role of these tDRs in the regulation of gene expression via RNA silencing, but this could represent another interesting example. Finally, a very interesting discovery comes from a study focusing on the interaction between the rhizobial bacterium Bradyrhizobium japonicum and soybean (Glycine max) during root nodule formation (Fig. 3B). Ren et al. (2019) showed the presence in the nodules of bacterial tDR-3Ts with sizes ranging between 18 nt and 24 nt. More specifically, they demonstrated that three tDRs from bacterial tRNA^Val(CAC), tRNA^Gly(TCC), and tRNA^Gln(CTG), found loaded on to AGO1, target three genes, GmRHD3, GmHAM4 and GmLRX5, associated with nodule initiation and development. The authors then demonstrated that, through their ability to interact with AGO1, tDRs target and repress genes that prevent symbiosis. Here, by acting as positive regulators of nodulation, the rhizobial tDRs represent a nice example of cross-kingdom regulation between bacteria and plants.

While a few tDRs have been found associated with AGO proteins, many of them are not. A few studies have shown that RNAs deriving from tRNAs are able to inhibit translation independently of any RISC (Fig. 3C). In pumpkin (Cucurbita maxima), tDRs, among other small non-coding RNAs, accumulate in the phloem sap (Zhang et al., 2009). The authors demonstrated that this complex population of small RNAs triggers translation inhibition in vitro and suggested that tDRs were the best candidates to play this role and act as long-distance signaling molecules. To address the question of whether plant tDRs can act as regulators of protein synthesis, in vitro experiments were performed with Arabidopsis tDRs. Both long (from tRNA^Ala, tRNA^Leu, or tRNA^Cys) and short (from tRNA^Ala or tRNA^Asn) tDRs are capable of inhibiting translation in vitro (Nowacka et al., 2013; Lalande et al., 2020). In mammals, translation repression by tDR-5A derived from tRNA^Ala was also demonstrated and shown to be dependent on the presence of the four guanine residues located at its 5ʹ end that enable G-quadruplex formation (Ivanov et al., 2014). By contrast, Arabidopsis alanine tDR-5D also contains four guanines at the 5ʹ end but this is not a prerequisite for translation inhibition. Analysis of Arabidopsis ribosomal fractions showed that alanine tDR can associate with active polyribosomes and modulate translation, but how this is achieved remains to be elucidated (Lalande et al., 2020). So far, the inhibition of protein synthesis by tDRs has been demonstrated only in vitro. It will be essential to provide evidence of the involvement of tDRs in the regulation of translation in vivo and to characterize the biological processes impacted, such as stress responses or plant development.

Concluding remarks

tRNA biology appears to be much more complex than first envisaged. With the discovery of the genetic code and the indispensability of tRNAs to transform genetic information into proteins during the translation process, it seemed that the circle was closed. This loop was reopened with the discovery of novel functions attributed to these macromolecules. We have presented some of them in this review dedicated to the plant kingdom, but these are likely only the tip of the iceberg. During evolution, nuclear tRNA genes have been duplicated, multiplied, and dispersed in eukaryotic genomes, such as those of the plant world. Point mutations have progressively generated tRNA isoacceptors (i.e. different anticodons for the same amino acid) and tRNA isodecoders (i.e. the same anticodon but with sequence differences). Moreover, the transition from the prokaryotic to the eukaryotic world is correlated with a significant increase in both the number and the complexity of base modifications found on tRNAs (Boccaletto et al., 2022). Altogether, the result is an immense and complex reservoir of tRNAs that opens the door to the emergence, in the course of evolution, of various and unanticipated roles that for the most part have yet to be determined. Identifying these different functions, whether related or not to translation, remains a challenge. The development of new high-throughput technologies should facilitate this work. The study of tRNA epitranscriptome variation during cell development or during the response or adaptation to stresses could lead to the characterization of certain tRNAs involved in specific processes. This is particularly true for plants, which, as sessile organisms, must quickly and continuously adapt to the environment. In humans and yeast, variation in tRNA modification levels were observed during the cell cycle and in response to environmental stresses (e.g. Chan et al., 2010; Huber et al., 2019; Chen et al., 2021). In plants, such changes were also shown under different stress conditions such as drought, salt, or cold (Wang et al., 2017b) but variations during cell development have not been studied yet. Many tRNA modifications are essential for efficiently decoding the corresponding codons, and dynamic changes in tRNA modification levels in response to stress conditions can favor the selective translation of subpopulations of mRNAs (Ignatova et al., 2020; Lin et al., 2018). Variations in tRNA modification levels can also affect the stability of tRNAs (Huang and Hopper, 2016; Torrent et al., 2018). In terms of the canonical role of tRNAs, protein synthesis efficiency will then be affected. In addition, as described above, the biogenesis of tDRs can be impacted by changes in tRNA modification, thereby potentially influencing the regulatory functions of tDRs. Importantly, it is worth mentioning here that, in addition to its effect on tDR biogenesis, altered modification status may also influence the function of the resulting tDRs (X. Zhang et al., 2016). The importance of the many tRNA modifications represents a large biological field to be discovered. tRNA epitranscriptomic comparative studies could provide new clues as to which modifications and which tRNA isoacceptors and/or isodecoders are differentially affected during cell development or stress responses, representing a first step towards understanding their various roles in central biological processes.

The number of tRNA genes and also their localization on the genome are aspects to be taken into consideration regarding their potential functions. Very few studies in this field exist. It is known that the environment and the three-dimensional structure of chromatin can affect the transcription of tRNA genes by RNA Pol III, for example, in S. cerevisiae (Thompson et al., 2003). Conversely, the spatial organization of tRNA genes in a genome could influence the three-dimensional structure of the chromatin. Indeed, in S. cerevisiae, Hamdani et al. (2019) have shown that tRNA genes and RNA Pol III transcription elements contribute to chromatin organization and local domain formation. Whether this is a widespread phenomenon, and whether it occurs in photosynthetic organisms, remains to be elucidated. The annotation of plant tRNA genes (PlantRNA, http://plantrna.ibmp.cnrs.fr/; PtRNAdb, http://14.139.61.8/PtRNAdb/index.php) and the use of recent technologies including high-throughput chromatin conformation capture and derivatives allowing local- and long-distance chromatin interactions (Domb et al., 2022) could help to demonstrate the role of tRNA genes in shaping genome structure.

The discovery of tRNA-derived RNAs with multiple functions, in particular as regulators of genome expression, has completely changed the paradigm. Indeed, tRNA-derived RNAs are no longer considered to be simple degradation products; in fact, they have greatly expanded the tRNA world, even though they remain quite enigmatic. Many unanswered questions remain: for example, which ribonuclease is required for which tDR biogenesis? Among the vast repertoire of RNA fragments, which tDR is involved in which biological process? Which molecular processes and which tDR interactors play roles in which biological process? In plants, the situation is even more complex as tDRs derived from organellar (plastidial and mitochondrial) tRNAs have been observed. Plant organellar tDRs are not generated within the organelles (Cognat et al., 2017). Whether these organellar tDRs play a role in retrograde signaling is another open question.

Finally, among the outstanding questions, it is still debated how, through the process of evolution, the present-day tRNA structure arose (Schimmel, 2018). Among the various hypotheses, it has been suggested that a combination of tRNA halves is at the origin of modern tRNAs (Zuo et al., 2013). This is a ‘chicken and egg’ issue, and both possibilities—the emergence of modern tRNAs via the fusion of smaller RNAs at the beginning of life to ensure protein synthesis, and the biogenesis of a diversity of small non-coding RNAs through the cleavage of tRNAs over the course of evolution to acquire new ways of regulating gene expression—are not exclusive. At this point, it is worth recalling that tDRs are present in the three domains of life, Archaea, Bacteria, and Eukarya, meaning that they are among the oldest classes of small non-coding RNAs. Similarly, tRNAs are considered to be among the oldest and most conserved non-coding RNAs, and their complex relationship with viruses, as described above, is intriguing from an evolutionary point of view. As suggested by Chen et al. (2021), while tRNAs are now considered to be major players in protein translation, another primary ancestral role may have been to be drivers of early life by providing the possibility, through fragmentation, to produce a large set of tDRs, opening up the potential to acquire new regulatory functions. This attractive hypothesis will need to be addressed in the future.

Author contributions

MC and LD contributed to the writing of the manuscript and the drawing of the figures.

Conflict of interest

No potential conflict of interest was reported by the authors.

Funding

This work was supported by the Interdisciplinary Thematic Institute IMCBio, as part of the ITI 2021–2028 program of the University of Strasbourg, CNRS and Inserm, supported by IdEx Unistra (ANR-10-IDEX-0002), and by the SFRI-STRAT’US project (ANR 20-SFRI-0012) and EUR IMCBio (ANR-17-EURE-0023) under the framework of the French Investments for the Future Program.