Loss of threonyl-tRNA synthetase-like protein Tarsl2 has little impact on protein synthesis but affects mouse development

Aminoacyl-tRNA synthetases (aaRSs) are essential components for mRNA translation. Two sets of aaRSs are required for cytoplasmic and mitochondrial translation in vertebrates. Interestingly, TARSL2 is a recently evolved duplicated gene of TARS1 (encoding cytoplasmic threonyl-tRNA synthetase) and represents the only duplicated aaRS gene in vertebrates. Although TARSL2 retains the canonical aminoacylation and editing activities in vitro, whether it is a true tRNA synthetase for mRNA translation in vivo is unclear. In this study, we showed that Tars1 is an essential gene since homozygous Tars1 KO mice were lethal. In contrast, when Tarsl2 was deleted in mice and zebrafish, neither the abundance nor the charging levels of tRNAThrs were changed, indicating that cells relied on Tars1 but not on Tarsl2 for mRNA translation. Furthermore, Tarsl2 deletion did not influence the integrity of the multiple tRNA synthetase complex, suggesting that Tarsl2 is a peripheral member of the multiple tRNA synthetase complex. Finally, we observed that Tarsl2-deleted mice exhibited severe developmental retardation, elevated metabolic capacity, and abnormal bone and muscle development after 3 weeks. Collectively, these data suggest that, despite its intrinsic activity, loss of Tarsl2 has little influence on protein synthesis but does affect mouse development.

Aminoacyl-tRNA synthetases (aaRSs) are essential components for mRNA translation. Two sets of aaRSs are required for cytoplasmic and mitochondrial translation in vertebrates. Interestingly, TARSL2 is a recently evolved duplicated gene of TARS1 (encoding cytoplasmic threonyl-tRNA synthetase) and represents the only duplicated aaRS gene in vertebrates. Although TARSL2 retains the canonical aminoacylation and editing activities in vitro, whether it is a true tRNA synthetase for mRNA translation in vivo is unclear. In this study, we showed that Tars1 is an essential gene since homozygous Tars1 KO mice were lethal. In contrast, when Tarsl2 was deleted in mice and zebrafish, neither the abundance nor the charging levels of tRNA Thr s were changed, indicating that cells relied on Tars1 but not on Tarsl2 for mRNA translation. Furthermore, Tarsl2 deletion did not influence the integrity of the multiple tRNA synthetase complex, suggesting that Tarsl2 is a peripheral member of the multiple tRNA synthetase complex. Finally, we observed that Tarsl2-deleted mice exhibited severe developmental retardation, elevated metabolic capacity, and abnormal bone and muscle development after 3 weeks. Collectively, these data suggest that, despite its intrinsic activity, loss of Tarsl2 has little influence on protein synthesis but does affect mouse development.
Aminoacyl-tRNA synthetases (aaRSs) are ubiquitously expressed housekeeping proteins that are critical for catalyzing the ligation of tRNAs with cognate amino acids, thus providing the aminoacyl-tRNA building blocks for ribosomal protein synthesis (1,2). In addition to tRNA aminoacylation, approximately one-half of extant aaRSs have evolved with a proofreading (editing) function, enabling them to remove tRNAs that are improperly charged (such as Ser-tRNA Thr ) (3,4). Thus, aminoacylation and editing activities are two conserved canonical functions of aaRSs.
In vertebrates, due to presence of cytoplasmic and mitochondrial protein synthesis, there are two mRNA translation systems, requiring two sets of aaRSs, in either cytoplasm or mitochondria (5)(6)(7). In mammalian cells, two sets of aaRSs are usually encoded by two separate aaRS genes, except KARS1 and GARS1, in both cases one gene encodes two different protein isoforms via alternative mRNA splicing (8,9) or translational reinitiation (10). Accordingly, there are in total 37 aaRS genes in human cells (5). One typical feature of mammalian aaRSs is the existence of a multiple tRNA synthetase complex (MSC) comprising nine aaRS activities (RARS1, LARS1, IARS1, MARS1, EPRS1, QARS1, DARS1, and KARS1) and three nonenzymatic auxiliary factors (p43/ AIMP1, p38/AIMP2, and p18/AIMP3) (11)(12)(13). The evolution of MSC formation is regarded as a progressive process with the appearance of simple entities in lower eukaryotes and complicated entities in higher eukaryotes. AaRSs from higher eukaryotes often have a N-or C-terminal extension than their bacterial or lower eukaryotic counterparts, including the glutathione S-transferase (GST) domain, the WHEP domain and other aaRS-specific ones (14). The primary sequences and topological structures of these extensions differ remarkably. It is generally found that these extensions are not involved in substrate binding and catalysis. However, they participate in protein-protein interaction, as in MSC assembly (14,15). Interaction patterns between several components of MSC have been clarified by various methods including crosslinking, yeast two hybrids, or crystal structure determination (15)(16)(17). For example, the N-terminal extension of RARS1 forms two long conserved α-helixes each containing a leucine-zipper (LZ), which promotes its hydrophobic interaction with the N-terminal α-helix of p43. Meanwhile, the N-terminal extension of ‡ These authors contributed equally to this work. * For correspondence: Pengfei Fang, fangpengfei@sioc.ac.cn; En-Duo Wang, edwang@sibcb.ac.cn; Xiao-Jian Sun, xjsun@sibs.ac.cn; Xiao-Long Zhou, xlzhou@sibcb.ac.cn.
RARS1 is also involved in interaction with QARS1 (17). In addition, mammalian aaRSs frequently perform additional functions distinct from tRNA charging and editing. These noncanonical functions include mTOR signaling, mRNA translation regulation, inflammation, gene transcription, virus replication, among others (18,19). In line with these observations, one of the rationales of MSC formation is to coordinately integrate both canonical and noncanonical functions of aaRSs. Once released from the MSC, aaRSs are ready to fulfill their regulatory roles (2,18,19). Cytoplasmic and mitochondrial threonyl-tRNA synthetases (ThrRSs) in eukaryotic cells are encoded by TARS1 and TARS2, respectively (5,20). Interestingly, a TARSL2 (TARS3) gene derived from the duplication of the TARS1 gene and encoding a third ThrRS-like protein has been identified in vertebrates (20)(21)(22)(23)(24)(25). The most striking difference between TARS1 and TARSL2 is the distinct N-terminal extensions of the two proteins. The extension of TARSL2 is larger (161-aa in length) than that of TARS1 (82-aa in length), the former carrying sequences from the N-terminal extensions of RARS1 and TARS1 (22,23). The N-terminal extension of TARSL2 readily facilitates TARSL2 incorporation into the MSC. We have previously demonstrated that the two LZs in the RARS1homologous peptide of the N-terminal extension are critical for MSC incorporation. By performing both yeast two-hybrid and coimmunoprecipitation assays, we revealed that TARSL2 interacted with both RARS1 and p43 in the MSC (23). Notably, the N-terminal extension of RARS1 interacted with p43 in the MSC (17). Although the structure of Escherichia coli ThrRS was resolved more than 2 decades ago (26), the structure of full-length eukaryotic ThrRS has never been elucidated (27). To date, only half of the eukaryotic cytoplasmic and mitochondrial ThrRSs, including the aminoacylation domain and the tRNA-binding domain, have been characterized (27,28). The three-dimensional structures of the N-terminal extensions of TARS1 and TARSL2 remain unknown. However, the catalytic domains of TARS1 and TARSL2, including the N1 (editing regulation) (29), N2 (editing), aminoacylation, and Cterminal tRNA-binding domains, are highly conserved (22). We have previously reported that the aminoacylation and editing sites are active in TARSL2 and that they charge tRNA Thr and edit mischarged Ser-tRNA Thr in vitro (22). However, one of the most fundamental questions about TARSL2, namely whether TARSL2 has evolved to perform canonical functions in vivo, is unclear. In other words, due to presence of TARSL2 in vertebrate, whether TARS1 is an essential gene for normal development and growth is an open question.
In this work, we modeled the structures of TARS1 and TARSL2, and revealed that their N-terminal extensions adopt different architectures, while the structures of other domains are very similar. We also showed that Tars1 is essential for embryonic development in mice and thus cannot be knocked out; in contrast, Tarsl2 can be deleted. The absence of Tarsl2 in mice had no effect on the abundance or aminoacylation level of tRNA Thr or other noncognate tRNAs, but a disruption of growth was observed with alterations in metabolism, in bone and muscle development. Similarly, Tarsl2 can also be deleted without exerting an obvious effect on tRNA Thr aminoacylation, development, or angiogenesis in zebrafish. In summary, our data suggest that the recently emerged Tarsl2 is not involved in tRNA aminoacylation under physiological conditions and therefore cannot replace the essential function of Tars1. The precise function of Tarsl2 awaits further investigation.

Results
TARSL2 and TARS1 have a similar structure, except for their N-terminal extensions TARSL2 results from the duplication of the TARS1 gene during evolution. TARSL2 and TARS1 have very similar domains/regions, including an N-terminal extension and two N1 and N2 domains related to editing activity, an aminoacylation catalytic domain, and an anticodon binding domain (Fig. 1A). However, the TARSL2 N-terminal extension differs from that of TARS1 in that the former is twice as long, the additional region carrying two LZs (Fig. 1A). Due to the significant level of accuracy achieved by a deep learning-based approach to predict the 3D structure of proteins (30), we were able to analyze the structures of TARS1 and TARSL2 predicted by AlphaFold 2 even without the full-length crystal structures (31). Except for the disordered regions (residues 1-51 of TARS1 and residues 96-127 of TARSL2), which showed low model confidence, the models of most structures, including the helix in the C-end of the extension region and the other four domains, were predicted with a high degree of confidence and the two LZs of TARSL2 were predicted with a medium degree of confidence ( Fig. 1, B-D). The TARS1 and TARSL2 models could be superimposed on the crystal structure of the E. coli ThrRS-tRNA complex (Fig. 1D), consistent with our previous finding that TARSL2 exhibits intrinsic tRNA aminoacylation and editing activities (22). The two LZs of TARSL2, which are absent in TARS1, were predicted to form a hairpin structure and be attached to the editing domain ( Fig. 1, C and D). We have previously shown that the two LZs were very similar to the N terminus of RARS1, which interacts with p43 and QARS1 (Fig. 1E), and that TARSL2 was incorporated into the MSC through these two LZs (23). When the two LZs from the TARSL2 N-terminal extension region are assembled into the MSC, conformational changes are likely to occur, allowing TARSL2 interactions with p43, QARS1, or RARS1 (Fig. 1E). These findings indicate that TARSL2 carries an additional extension and that this extension binds the MSC, which is the only structural difference between TARSL2 and TARS1.
Tars1 is an essential gene and cannot be replaced by Tarl2 in vivo To understand whether Tars1 is an essential gene or can be substituted with Tarsl2 in vivo, we initially tried to knock out the Tars1 gene in the NIH3T3 cell line using the CRISPR/Cas9 method. However, despite extensive efforts, we were unable to obtain a null allele, implying that Tars1 may be essential for cell viability. To directly explore a potential pivotal role for Tars1 in vivo, we purchased commercially available heterozygous Tars1 gene-KO mice (C57BL/6N-Tars em1cyagen ) from Cyagen. Crossing the mice revealed that only WT (Tars1 +/+ ) and heterozygous (Tars1 +/− ) mice, but not homozygous (Tars1 −/− ) Tars1-deletion mice, were viable and they were obtained at a 1:2 proportion ( Fig. 2A). The growth and body weight of Tars1 +/+ and Tars1 +/− mice were comparable (Fig. 2, B-D). No significant difference in the protein level of Tars1 was observed in muscle tissue between Tars1 +/+ and Tars1 +/− mice (Fig. 2E). Similarly, the steady-state protein abundance of Tarsl2 in Tars1 +/+ and Tars1 +/− mice did not differ (Fig. 2E), suggesting that Tarsl2 gene expression was not stimulated. The steady-state amounts of tRNA Thr (AGU) and tRNA Thr (CGU) and their charging levels in muscle tissue were comparable between Tars1 +/+ and Tars1 +/− mice (Fig. 2, F and G). All these data showed that Tars1 is an essential gene for cell viability and mouse survival in vivo and deletion of one copy of Tars1 did not lead to haploinsufficiency. Furthermore, the results suggest that Tars1 cannot be replaced by Tarsl2 in vivo.

Tarsl2 is not critical for canonical tRNA aminoacylation in vivo
To explore the function of Tarsl2 in vivo, we successfully constructed homozygous Tarsl2 gene-KO (Tarsl2 −/− ) mice using CRISPR/Cas9-mediated gene editing. The survival of the Tarsl2 −/− mice suggests that Tarsl2 is not essential in vivo, consistent with our previous findings obtained by constructing Figure 1. The topology of TARSL2 in the N-terminal extension differs from that of TARS1. A, the domains of TARS1 and TARSL2. B, the TARS1 structure model was predicted by AlphaFold 2. C, the TARSL2 structure model was predicted by AlphaFold 2. In (B and C), the structures are colored on the basis of the confidence of each model prediction. D, the structural models of TARS1 and TARS2 are superimposed onto the E. coli ThrRS-tRNA structure (PDB ID 1QF6). TARS1 is light blue. TARSL2 is wheat; however, the leucine zippers in one chain are highlighted in orange and magenta. E. coli ThrRS is cyan, and tRNA is pink. E, the leucine zippers of RARS1 are highlighted with a red dashed line in the RARS1-QARS1-p43 complex structure (PDB ID 4R3Z). ABD, anticodon binding domain; CD, catalytic domain; LZ, leucine zipper; N-ext, N-terminal extension; ThrRS, threonyl-tRNA synthetase.  16-week-old Tars1 +/+ (n = 5) and Tars1 +/− (n = 4) female mice. E, protein abundance of the 8-week-old Tars1 and Tarsl2 in Tars1 +/+ and Tars1 +/− mouse muscle as detected by Western blot. Gapdh was used as the loading control. F, levels of tRNA Ala (AGC), tRNA Thr (AGU), and tRNA Thr (CGU) in muscle tissues of the 8-week-old Tars1 +/+ and Tars1 +/− mice. 5S RNA was as the loading control. G, aminoacylation levels of tRNA Thr (AGU) and tRNA Thr (CGU) in muscle tissues of the 8-week-old Tars1 +/+ and Tars1 +/− mice. *, a nonspecific band. De: tRNA deaminoacylation control. Error bars in (C-G) represent the mean ± SD with the p values indicated (two-tailed Student's t test). Figure 3. Loss of Tarsl2 exerts no effect on the steady-state or charging of tRNA Thr s in vivo. A, schematic showing Tarsl2-mutant mouse construction by CRISPR/Cas9 gene editing technology. Two sgRNAs were designed around exon 3 to delete a sequence fragment. Primers 1 and 2 and Primers 3 and 4 were used to detect DNA mutations. B, PCR detection of the Tarsl2 +/+ , Tarsl2 +/− , and Tarsl2 −/− mouse genotypes using Primers 1 and 2 and Primers 3 and 4. C, sequencing results of Tarsl2 +/+ and Tarsl2 −/− mice. In the mutant mice, Tarsl2 mRNA translation was prematurely terminated at nt 342. D, Tarsl2 mRNA levels in the muscle of the 6-week-old Tarsl2 +/+ and Tarsl2 −/− mice, as determined by RT-PCR with two pairs of primers. E, protein abundance of Tarsl2 in the 6-week-old Tarsl2 +/+ and Tarsl2 −/− mouse muscle, as detected by Western blotting. Gapdh was used as the loading control. F and G, levels of tRNA Thr (UGU), tRNA Thr (CGU), and tRNA Thr (AGU) in the muscle tissues of the 6-week-old Tarsl2 +/+ and Tarsl2 −/− mice. tRNA Gly (CCC) was included as the loading control. H and I Aminoacylation levels of tRNA Thr (UGU), tRNA Thr (CGU), and tRNA Thr (AGU) in the muscle tissues of the 6-week-old Tarsl2 +/+ and Tarsl2 −/− mice.

Tarsl2 in vertebrates
the Tarsl2-deleted NIH3T3 cell line (23). In Tarsl2 −/− mice, a genomic fragment including exon 3 of Tarsl2 was deleted (Fig. 3A), leading to a truncated allele, which was confirmed by two sets of primers (Fig. 3B). Premature termination of translation of the mRNA transcribed from this allele occurrs at position 342, encoding a truncated protein of 113 amino acids (Fig. 3C). RT-quantitative PCR (qPCR) analyses using two different primer sets showed that Tarsl2 mRNA was significantly downregulated compared to WT mice, suggesting its degradation by nonsense-mediated mRNA decay (Fig. 3D). In addition, Tarsl2 protein was absent in Tarsl2 −/− mice (Fig. 3E).
Tarsl2 is most highly expressed in mouse muscle and heart tissue (22). We have previously proposed that, given its nuclear localization, Tarsl2 may be involved in quality control of tRNA Thr in the nucleus and thus contribute to the amount of cytoplasmic tRNA Thr (22,23). However, a Northern blot analysis showed that, in muscle, the steady-state levels of three cytoplasmic tRNA Thr isoacceptors [tRNA Thr (UGU), tRNA Thr (CGU), and tRNA Thr (AGU)] was not changed after Tarsl2 depletion (Fig. 3, F and G). Furthermore, the aminoacylation levels for tRNA Thr (UGU), tRNA Thr (CGU), and tRNA Thr (AGU) in muscle tissues from WT and Tarsl2 −/− mice were examined and compared in acidic polyacrylamide/urea gel system. The results showed little difference in the aminoacylation levels of the three tRNA Thr s between the two mice ( Fig. 3, H and I). These results strongly coincided with our previous hypothesis (23) suggesting that despite its ability to catalyze canonical aminoacylation in vitro, Tarsl2 is not critical for mRNA translation in vivo.

Tarsl2 is likely a peripheral member of the MSC
We have already reported that TARSL2 is a bone fide component of the MSC by interacting with RARS1 and p43 (23). Thus, we measured the levels of both proteins and assessed their incorporation into the MSC in Tarsl2 −/− mice. We found little difference in the steady state protein levels of several mouse cytoplasmic aaRSs (including Eprs1, Rars1, Kars1, Qars1, and Tars1) or that of two cofactors (p38 and p43) between WT and Tarsl2 −/− mice (Fig. 4A). We then sought to identify a role for Tarsl2 in maintaining the integrity of the MSC, notably through the incorporation of Rars1 and p43. For this purpose, muscle tissue lysate was fractionated by gel filtration chromatography. The data showed that after depletion of Tarsl2, several MSC components, including Rars1, p43, Qars1, and Eprs1, were copurified with the fractions containing the MSC, suggesting that Tarsl2 does not play a direct role in MSC assembly or in the incorporation of Rars1 and p43 in the MSC (Fig. 4B). These data suggested that, despite a direct interaction between Tarsl2 and p43 or between Tarsl2 and Rars1, the absence of Tarsl2 does not influence the incorporation of Rars1 or p43 into the MSC, which is consistent with the accepted role of p38 as the core scaffolding protein of the MSC (32) and the fact that Rars1 or p43 interacts with multiple proteins in the MSC (17). We further studied the steady-state abundance of several cytoplasmic tRNAs, including tRNA Lys (CUU), tRNA Ile (AAU), tRNA Gln (CUG), tRNA Arg (ACG), and tRNA Leu (AAG), which are aminoacylated by various MSC components. No significant difference was observed between WT and Tarsl2 −/− mice (Fig. 4, C and D). Furthermore, the aminoacylation levels of tRNA Lys (CUU) and tRNA Ile (UAU) were unchanged (Fig. 4, E and F). The global protein synthesis rate, as measured by puromycin incorporation assay, was comparable between WT and Tarsl2 −/− mice (Fig. 4, G and H), suggesting that the deletion of Tarsl2 has little effect on the rate of translation in vivo. Altogether, these data suggest that Tarsl2 does not impact on the integrity of MSC in vivo and is likely to be a peripheral but not a core member of the MSC.
Tarsl2 knockout affects the growth and metabolism of mice Tarsl2 +/+ , Tarsl2 +/− , and Tarsl2 −/− mice displayed comparable body weights at birth (Fig. 5A), suggesting that Tarsl2 is not involved in embryonic development. However, weight and size differences between Tarsl2 +/+ and Tarsl2 −/− mice gradually increased, becoming obvious (Fig. 5, B and C) after 3 weeks and maintained throughout the mouse lifespan (Fig. 5D). These results suggested that the lack of Tarsl2 leads to developmental defects in mice.
Our previous study showed that Tarsl2 was most highly expressed in mouse muscle; therefore, we performed an RNAseq analysis with Tarsl2 +/+ and Tarsl2 −/− mouse muscle tissue. The results showed that the expression of metabolism-related genes in the muscle was higher in the Tarsl2 −/− mice than in the Tarsl2 +/+ mice (Fig. S1). Indeed, a RT-qPCR analysis confirmed the upregulation of several well-established metabolism-related genes, including Igf2, Ppara, Pdk4, Ucp3, Pgc1a, and Pgc1b (33-35) (Fig. 5E). These results suggested that Tarsl2 may be involved in metabolism. Therefore, the effects of Tarsl2 on glucose homeostasis and insulin sensitivity were further determined by intraperitoneal glucose and insulin tolerance tests. We found that a lack of Tarsl2 led to glucose and insulin tolerance (Fig. 5, F and G). Normal and fasting plasma glucose levels were measured and the results showed that their levels were obviously lower in the Tarsl2 −/− mice than in the Tarsl2 +/+ mice (Fig. 5H). Subsequently, the triglyceride, cholesterol, LDL, and HDL were determined after isolating the plasma blood and all of these compounds were lower in the Tarsl2 −/− mice than in the Tarsl2 +/+ mice (Fig. 5, I  and J). Moreover, a body fat analysis showed that the Tarsl2 −/− mice were leaner than the Tarsl2 +/+ mice (Fig. 5K). Collectively, these results indicated a functional requirement for Tarsl2 in mouse growth and metabolism.

Tarsl2 deletion causes abnormal development of bone and muscle
Tarsl2 deficiency leads to a decrease in mice size during development (Fig. 5C). Therefore, we assessed bone tRNA Gly (CCC) was included as the control. De: tRNA deaminoacylation control. ****p < 0.0001. Error bars in (G and I) represent the mean ± SD with the p values indicated (two-tailed Student's t test).
Tarsl2 in vertebrates development in 8-week-oldTarsl2 +/+ and Tarsl2 −/− mice. H&E staining showed ossified and clear bone trabeculae in the distal femur in the Tarsl2 −/− mice but not in the Tarsl2 +/+ mice (Fig. 6A). This meant that bone development was abnormal in the Tarsl2-deficient mice. To determine the function of Tarsl2 in the skeletal system, a microquantitative computed tomography analysis was performed to compare changes in bone-related elements in the long bones of the Tarsl2 −/− mice and Tarsl2 +/+ mice (Fig. 6B). Eight-week-old Tarsl2 −/− mice showed increased bone mass per tissue volume (BV/TV) compared to the Tarsl2 +/ + mice (Fig. 6C). Further analysis revealed a difference in the trabecular separation of cancellous bone (Tb.Sp) and trabecular number (Tb.N) (Fig. 6, D and E), but there was no significant difference in trabecular thickness (Tb.Th), bone mineral density, or cortical bone thickness (Ct.Th) (Fig. 6, F-H). These results show that Tarsl2 deletion influences bone development.  Based on the high expression of Tarsl2 in muscle, which is closely connected to metabolism, we further measured muscle development in Tarsl2 −/− mice and compared the results to those obtained with WT Tarsl2 +/+ mice. Notably, the sizes of the tibialis anterior (TA) and soleus muscle (Sol) were significantly smaller in the Tarsl2 −/− mice than in the Tarsl2 +/+ mice old Tarsl2 +/+ (n = 3) and Tarsl2 −/− (n = 3) mice. K, body fat percentage of the 6-week-old Tarsl2 +/+ (n = 5) and Tarsl2 −/− (n = 5) mice, measured on the basis of fat/body weight (left) and lean body weight (right). ****p < 0.0001, ***p < 0.001, **p < 0.01 and *p < 0.05. The error bar represents the mean ± SD with the p values indicated (two-tailed Student's t test).  6). I, comparisons of the tibialis anterior muscle and soleus muscle between the 6week-old Tarsl2 +/+ and Tarsl2 −/− mice. The scale bar represents 1 cm. J, the weight of the TA and soleus muscle of the Tarsl2 +/+ and Tarsl2 −/− mice in (I) was measured. K, H&E staining of soleus cross sections from 6-week-old Tarsl2 +/+ and Tarsl2 −/− mice. L, area of the soleus cross section from Tarsl2 +/+ and Tarsl2 −/− mice. M, H&E staining of TA cross sections from 6-week-old Tarsl2 +/+ and Tarsl2 −/− mice. N, area of TA cross sections of Tarsl2 +/+ and Tarsl2 −/− mice. ****p < 0.0001, ***p < 0.001, and *p < 0.05. The error bar represents the mean ± SD with the p values indicated (two-tailed Student's t test). TA, tibialis anterior.
Tarsl2 in vertebrates (Fig. 6, I and J). Subsequently, we examined whether the smaller muscle in Tarsl2-null mice was due to fiber size differences. Quantification of a cross-sectional myofiber area showed that the fiber size in both the TA and Sol was significantly smaller in the Tarsl2 −/− mice than in the Tarsl2 +/+ littermates (Fig. 6, K-N). Thus, Tarsl2 plays a crucial role in muscle development.
The tRNA Thr aminoacylation level remains unaltered in Tarsl2-deleted zebrafish Zebrafish Tarsl2 displays high sequence similarity (73.1%) with mouse Tarsl2. To understand whether Tarsl2 has evolved with tRNA charging capacity and to explore the role of tarsl2 in lower vertebrates, we used the CRISPR-Cas9 system to establish a tarsl2-null (tarsl2 −/− ) zebrafish line that carries a 1 bp deletion (c.1356delG), which causes a frameshift after leucine codon 475 (L475) (Fig. S2A). This mutant Tarsl2 protein (p.L475fs*23) was nonfunctional, as the catalytic domain (core) and the tRNA-binding domain were missing due to the premature stop codon. The RT-qPCR results showed that the mRNA expression of tarsl2 was markedly reduced in tarsl2 −/− embryos 5 days postfertilization (Fig. S2B), suggesting that the mutated mRNA was degraded by the nonsense-mediated mRNA decay pathway. We first compared the steady-state abundance of tRNA Thr between WT and tarsl2 −/− fish. The amount of tRNA Thr (AGU) was significant but somewhat lower in the tarsl2 −/− fish, similarly to noncognate tRNA Gly (GCC); however, comparable tRNA Thr (AGU) and tRNA Thr (UGU) levels were observed (Fig. 7A). Moreover, tRNA Thr charging between the two fish lines was similar, as determined by acidic Northern blot (Fig. 7B). The tarsl2 −/− zebrafish were viable and showed normal fertility. The body length of the tarsl2 −/− zebrafish was also similar to that of WT and heterozygous zebrafish in the early stage of embryonic development (Fig. 7, C-E) and in adults (Fig. 7F). Given that knocking out tars, the paralog of tarsl2, led to the acquisition of a stress-induced angiogenic phenotype (36), we next questioned whether this phenotype was acquired in our tarsl2 −/− embryos. We crossed the tarsl2 −/− mutant with a transgenic (Tg) line (flk1:EGFP) to visualize the development of blood vessels in vivo. As expected, tars KO but not tarsl2 KO induced the acquisition of an ectopic angiogenic phenotype compared to WT embryos (Fig. 7, G). Taken together, these data demonstrate that, similar to the Tarsl2-null mouse, the tarsl2 −/− zebrafish line appeared phenotypically normal. These results suggest that Tarsl2 is not a tRNA synthetase, at least in physiological conditions, since its evolutionary emergence in vertebrates.

Discussion
Although a chimeric protein from the N-terminal domain of Tars1 and the main body of Tarsl2 catalyzes tRNA Thr aminoacylation and Ser-tRNA Thr editing in vitro (22), our genetic data clearly proved that loss of Tarsl2 did not influence the total or aminoacylated tRNA Thr levels in vivo. In addition, Tars1 is an essential gene despite the emergence of Tarsl2. These observations suggest that Tarsl2 did not evolve for tRNA Thr aminoacylation in physiological conditions. Failure of Tarsl2 in supporting survival at the absence of Tars1 is probably attributable to its very low amount compared to Tars1 and other aaRSs in vivo (22,24). Indeed, the mRNA level of Tars1 is 3.4-to 100-times higher than that of Tarsl2 in various mouse tissues (22). MS analyses have also shown that protein abundance of TARSL2 is much lower than that of TARS1 (24). In addition, its nuclear distribution due to an nuclear localization signal (NLS) in the C terminus may also prevent the charge of tRNA Thr , whose localization is rather cytoplasmic (22). Despite being bound by Rars1 and p43 (23), loss of Tarsl2 did not influence the incorporation of the two proteins in the MSC, suggesting that Tarsl2 is a peripheral component. The integrity of the MSC remains unchanged, which is consistent with the fact that the amount of tRNA and the charging levels of noncognate tRNAs are not changed. These results suggest that Tarsl2 plays little role in protein synthesis, at least in physiological conditions. We cannot exclude the possibility that, during specific stimuli or in stress conditions, Tarsl2 is able to charge tRNA Thr for protein synthesis. Alternatively, Tarsl2 perhaps participates in other pathways of tRNA metabolism, such as processing or modification.
Although Tarsl2 plays little role in tRNA aminoacylation, it is clearly essential for the growth of mice. Specifically, an elevated metabolism rate was observed after the loss of Tarsl2. In line with its high expression in muscle (22), muscle development in Tarsl2-KO mice was impaired. Lars1 negatively regulates myogenic differentiation and injury-induced skeletal muscle regeneration mediated by the Rag-mTORC1 pathway (37). Similarly, TARS1 has been recently found to be a key negative regulator in muscle development by inhibiting c-Jun N-terminal Kinase (JNK) signaling. The N-terminal extension of TARS1 has been found to be essential in the regulation of myoblast differentiation (38). Notably, Tarsl2 forms a heterodimer with Tars1 in vivo (23) but their N-terminal extensions differ. To understand whether Tarsl2 affects the JNK pathway similar to Tars1 or via Tars1, we measured the phosphorylation of JNK after Tarsl2 expression in C2C12 cells and we found that the total and phosphorylated JNK levels were unaltered before and after the induction of muscle differentiation (data not shown). Thus, the precise function of Tarsl2 in muscle development needs further exploration.
In addition to the involvement of TARS1 in muscle development, at least two other noncanonical functions of TARS1 have been described. TARS1 is secreted during inflammation to stimulate endothelial cell migration and angiogenesis (39). Considering the high similarity between TARS1 and TARSL2, TARSL2 may be released from cells, but evidence for this remains to be determined. In addition, TARS1 participates in translational initiation regulation to positively regulate mRNA translation in vertebrates (40). In this context, the N-terminal extension of TARS1 is required for the TARS1 interaction with 4EHP. Considering the completely distinct architectures of the two extensions, the possibility that TARSL2 is involved in mRNA translation initiation is low.

Tarsl2 in vertebrates
We have recently reported that a nonsense mutation in zebrafish Tars1, leading to an inactive truncated Tars1, causes abnormal angiogenesis (36). Obviously, we found that angiogenesis keeps unchanged in the absence of Tarsl2, clearly showing that Tarsl2 is not involved in angiogenesis in zebrafish. In addition, the body length of adult zebrafish between WT and Tarsl2-deleted species is comparable, in contrast to the obvious growth impairment of Tarsl2 KO mice. Regarding the seemingly different phenotypes between the mouse and zebrafish models, it has been widely noticed and accepted that many metabolic stress-related phenotypes in zebrafish appear relatively mild compared to those in the mouse (41). Possible mechanistic explanations for this phenomenon may include (1): early development of zebrafish embryos relies more on the maternally deposited mRNAs and proteins, because their zygotic genes are not expressed until 1000-cell stage, equivalent to 2-cell stage of mice (2, 42) zebrafish embryos can obtain nutrients from their yolk, and this process can be a passive diffusion independent of forming a complicated vasculature system (43); and (3) zebrafish can tolerate a higher degree of starvation stress, as it has been shown that zebrafish larvae can live up to 17 days postfertilization without any food supply (44).
Tarsl2 is the only aaRS duplicated gene in vertebrates. Tarsl2 encompasses a range of functions and roles that include the catalytic functions of aminoacylation and editing (22) and the ability to be incorporated in the MSC (23). Tarsl2 is also essential for growth, metabolism, and muscle development, however, its precise molecular function in vivo is still unclear and needs to be further explored. The ability of Tarsl2 to bind amino acid and nucleic acids (22) suggests that it may sense amino acid levels like LARS1 in mTOR signaling pathway (45,46) or bind RNA or DNA in cytoplasm and/or nucleus, like other tRNA synthetases (47,48). In addition, with two conserved LZs in the N-terminal extension, Tarsl2 probably interacts with other yet unidentified proteins by hydrophobic interactions.

Structural modeling
Full-length monomeric structure models of TARS1 and TARSL2 were obtained from the AlphaFold Protein Structure Database (30,31). Two monomeric models were superimposed on the human TARS1 structure (PDB ID 4P3N) to generate TARS1 and TARSL2 dimer models (27). Then, two tRNA molecules were docked onto the dimer models in reference to the E. coli ThrRS-tRNA complex structure (PDB ID 1QF6) (26).

Gel filtration chromatography of cell lysates
Mouse tissue extract was applied to a Superose-6 column for HPLC and eluted at a flow rate of 0.5 ml/min by using a buffer containing 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 1 mM PMSF, and 1 mM DTT. Fractions were collected for immunoblotting (Western blot).

Western blot analysis
Different proteins in whole-cell lysates were separated by 10% SDS-PAGE and then transferred to PVDF membranes. The PVDF membrane was then cropped to blot different proteins in the same lane. After blocking with 5% (w/v) nonfat dried milk, the membranes with targeted proteins were incubated with the corresponding primary antibodies overnight at 4 C. The membranes were then washed three times using PBS supplemented with 0.05% Tween-20 (PBST) (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , and 0.5‰ Tween-20) and incubated with the corresponding horse radish peroxidase-conjugated secondary antibodies at room temperature for 30 min. After washing three times with PBST, the membranes were treated with a chemiluminescent substrate and imaging was performed using an Amersham imager 680 system (GE).

Generation of the Tarsl2-deletion mice
Tarsl2-deletion mice were established using CRISPR/Cas9mediated gene editing technology by Shanghai Model Organisms. Briefly, two guide RNAs (gRNAs) (5 0 ACCCTT GGTATCCCGAGAACTGG3 0 and 5 0 AGACAACATTAA TTGTATACGGG 3' (PAM sequences are underlined) were designed to target two introns around exon 3 of the Tarsl2 transcript (ENSMUST00000032728.8). Cas9 (New England Biolabs, M0646T) and gRNAs (obtained by in vitro transcription) were then injected into C57BL/6J mouse zygotes to obtain F0 generation mice, which were subsequently crossed with C57BL/6J mice to obtain F1 generation mice. Gene editing led to the deletion of a gene fragment harboring exon 3 and premature termination of mRNA translation at the Val and tarsl2 −/− embryos two dpf. The scale bar represents 100 μm. **p < 0.01 and ***p < 0.001. The error bar represents the mean ± SD with the p values indicated (two-tailed Student's t test).

Real-time PCR quantification of mRNA
Total RNA was isolated from mouse muscle or zebrafish using TRIzol reagent as described in the Methods. Complementary DNA (cDNA) was synthesized following the procedures of the PrimeScript RT reagent kit protocol. Realtime qPCR was then performed using 2 × T5 Fast qPCR Mix. The sequences for qPCR quantification are listed in Table S1.

Northern blot analysis
Total RNA extracted from muscle of 6-week-old mice was isolated using TRIzol reagent, and 5 μg of total RNA was electrophoresed through a 15% polyacrylamide-8 M urea gel in Tris-borate-EDTA buffer at room temperature at 150 V for 1.5 h. For aminoacylation assays, total RNA was extracted and resolved with 0.1 M NaAc (pH 5.2). To separate the charged and uncharged tRNAs, 5 μg of total RNA was electrophoresed through an acidic (pH 5.2) 10% polyacrylamide-8 M urea gel at 4 C with 8 W for 16 h. The RNA was then transferred onto a positively charged nylon membrane at 4 C under 250 mA for 30 min. After UV-crosslinking (8000 × 100 μJ/cm 2 ), the nylon membrane was blocked with prehybridization solution (4 × saline-sodium citrate, 1 M Na 2 HPO 4 , 7% SDS, 1.5 × Denhardt's solution, 0.4 mg/L fish sperm DNA) at 55 C for 2 h. The blocking solution was substituted for hybridization solution with digoxin-probes (the probe sequences are listed in Table S2) for specific tRNAs and 5S RNA and incubated at 55 C overnight. The membrane was washed twice with wash buffer (0.1 M maleic acid and 0.15 M NaCl, pH 7.5) for 15 min each time, and then, 1 × blocking reagent (0.1 M maleic acid, 0.15 M NaCl, and 10% blocking reagent, pH 7.5) was added and incubated for 30 min at room temperature. Next, the cells were incubated with anti-AP buffer (1 × blocked reagent and antidigoxigenin-AP antibody diluted 1:10,000) for 1 h and washed twice with washing buffer. The membrane was treated with CDP-Star, and imaging was performed using an Amersham imager 680 system (GE).

Puromycin incorporation assay
Primary mouse embryonic fibroblasts (MEFs) were prepared after isolation from WT and Tarsl2 −/− mice according to a standard protocol (50) and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS). After reaching approximately 95% confluence, fresh Dulbecco's modified Eagle's medium (supplemented with 10% FBS and 1.5 μg/ml puromycin) was used to culture the cells for 2 h. The cells were then collected and lysed, and the newly synthesized proteins with incorporated puromycin were detected with an antipuromycin antibody by Western blotting.

RNA-sequencing
Total RNA was extracted from muscle in 6-week-old Tarsl2 +/+ and Tarsl2 −/− mouse and isolated using TRIzol reagent. The RNA-seq libraries were prepared using Illumina TruSeq Stranded mRNA Library Prep Kit Set A (RS-122-2101, Illumina), starting with 2 μg of total RNA as input material. cDNA library preparation and sequencing were performed according to the Illumina's standard protocol. Library size selection was performed using Agencourt AMPure XP beads (Beckman Coulter, A63882), resulting in an average library size of 400 bp. The libraries were sequenced using Illumina NovaSeq 6000 with paired-end 150 bp reads. Sequencing quality was assessed by FastQC v.0.11.4. All reads were mapped to the reference genome of Illumina iGenomes UCSC mm10 using HISAT2 v.2.05.0. Differential expression analysis was implemented using DESeq2 R package (1.20.0).

Histology
The 8-week-old Tarsl2 +/+ and Tarsl2 −/− mouse femurs were fixed in 4% paraformaldehyde for 48 h and specimens were decalcified until soft and pliable. Samples were incubated in 15% diethyl pyrocarbonate-EDTA (pH 7.8) for decalcification. Then, specimens were embedded in paraffin and sectioned at 5 μm. Muscle tissues were embedded with optimal cutting temperature (O.C.T.) Compound (Sakura) and frozen in liquid nitrogen for 1 min. The samples were preserved at −80 C. Then, the samples were sectioned at 10 μm thickness and stained with H&E.

Myofiber cross-section area measurement
The cross-sectional area of the myofibers section images were obtained from TA muscles. Three-hundred fibers were chosen randomly from each mouse and were measured. The myofiber cross-section area was measured by Image J software (https://imagej.nih.gov/ij/).

μCT analysis
The 8-week-old Tarsl2 +/+ and Tarsl2 −/− mouse femurs were collected. Soft tissues were removed and the remaining tissues were fixed in 70% ethanol. Scanning was performed with the microquantitative computed tomography SkyScan 1176 System. The mouse femurs were scanned at an 8.96 μm resolution for quantitative analysis. The interesting region segmented by a fixed threshold was reconstructed into three-dimensional Tarsl2 in vertebrates images. The properties of Tb.Th, Tb.N, BV/TV, C.Th were then quantified.

Angiogenic phenotype assay
Live Tg (flk1:EGFP) transgenic embryos were anesthetized with 0.03% tricaine (Sigma; A5040) and then mounted on dishes with 1% low-melting agarose (Sangon Biotech; A600015). Fluorescence images were captured under a scanning confocal microscope (Nikon A1). Blood vessels in the region of the trunk encompassing 7-9 intersomitic vessels (ISVs) were analyzed to quantify the angiogenic phenotype. The number of ectopic branching points per ISV (52) was statistically calculated.

Animal experiments
The animal experiments with zebrafish were approved by the Committee of Animal Use for Research at Shanghai Jiao Tong University School of Medicine. Animal experiments with mice were approved by the Institutional Animal Care and Research Advisory Committee at the Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. All methods were performed in accordance with the relevant guidelines and regulations of the Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences.

Data availability
Raw RNA-seq data were deposited in the NCBI Gene Expression Omnibus database with accession number GSE218318. All of the study data are included in the article and/or supporting information.
Supporting information-This article contains supporting information.