Validity of hypothesis that [>>t] is compatible withA-tRNAGlx but notN-tRNAGlx
Since our primary focus here is to understand the role played by the H8-L -H9 motif in recognizing/discriminating tRNAGlx D-helix (augmented versus non-augmented), the genomic tRNAGln and tRNAGlusequences of all 13 bacterial species were examined (Figure S5) for the presence/absence of augmented/non-augmented D-helix in tRNAGlx. The results are summarized in Table S2. The genomic tRNAGlx is annotated with A (augmented) or N (non-augmented) corresponding to each bacterial species.
We first test our hypothesis that the rigid proteobacterial α-[>>t ]-H8-L -H9 conformational motif favorably interacts withA -tRNAGlx and unfavourably withN -tRNAGlx on three proteobacterial species whose GluRSs display the α-[>>t ] conformation — ECO(D-), SML(D-) and XOP(D-) — all associated withA -tRNAGlu andN -tRNAGln. In absence of gatCAB (all are D-), these three GluRSs must strictly discriminate againstN -tRNAGln and favorably interact withA -tRNAGlu. Our hypothesis matches with the strict requirement of the GluRSs to beN -tRNAGln-discriminatory. The hypothesis was then tested for other GluRSs (see below).
Two proteobacterial species BTE(D+) and PAE(D+) contain gatCAB in their genomes suggesting that there is no strict requirement forBte (D+)-GluRS and Pae (D+)-GluRS to discriminate against tRNAGln. Both genomes are associated withA -tRNAGlu andN -tRNAGln and both display the [>>t ] conformation (α- and β-). In absence of experimental data, our hypothesis suggests that both must be tRNAGln-discriminatory. The other proteobacterial species HPY (A -tRNAGlu;N -tRNAGlu) contain twin GluRSs. Among the two,Hpy (T1)-GluRS displayed the dynamic α-[>>t ] conformation, compatible with A - but not N -tRNAGlx.Hpy (T1)-GluRS has been experimentally shown to recognizeA -tRNAGlu and notN -tRNAGln, which is compatible with our hypothesis. Hpy (T2)-GluRS has been experimentally shown to be non-functional and therefore is not included in this discussion.
Two non-proteobacterial ND-GluRSs (Mtu -GluRS andBbu -GluRS; being ND they must recognize both tRNAGlu and tRNAGln) displayed the [>>t] motif. Both tRNAGlu and tRNAGln areA -type in the MTU genome, compatible with the presence ofA -tRNAGlx-recognizing [>>t ]-H8-L -H9 motif. In contrast, the BBU genome containsA -tRNAGlu/N -tRNAGln. The [>>t ]-H8-L -H9 motif inBbu -GluRS is thus compatible withA -tRNAGlu but not withN -tRNAGln, although as ND-GluRS it must recognize both. It is possible that with a large addendum to its C-terminal domain and the presence of K in place of D at the tip of the turn that protrudes into tRNA, the geometry of Bbu -GluRS with tRNA may be non-canonical and not the same as the structural models considered here.
All other (four) GluRSs originating from six bacterial species (Table S2) display the [t>> ]-H8-L -H9 motif, compatible with both A - and N -type of RNA. Of these, TEL (A -tRNAGlu; N -tRNAGlu) contain ND-GluRS which must recognize both tRNAGlu and tRNAGlu, compatible with our hypothesis. TMA (A -tRNAGlu;N/A -tRNAGlu) genome contains twin GluRSs.Tma (T1)-GluRS has been experimentally shown to recognize both tRNAGlu and tRNAGln, also compatible with our hypothesis. TTH (A -tRNAGlu;A -tRNAGlu) and EMG (A -tRNAGlu; A -tRNAGlu) genomes contain GlnRS and are devoid of gatCAB, making the corresponding GluRSs discriminatory towards A -tRNAGln. Since the [t>> ]-H8-L -H9 motif, displayed by these two proteins, is favorably disposed to interact withA -tRNAGln, the origin ofA -tRNAGln-discrimination for these two non-proteobacterial GluRSs must arise from interactions between the GluRSs and some non-D-helix element in tRNA. For Tth -GluRS, it has been experimentally shown that tRNAGln-discrimination is mediated by R358 at the anti codon-binding C-terminal domain of Tth -GluRS (26).
Correlation between H8-L-H9 sequence and genomic A/N-tRNAGlnin proteobacteria
The above discussion, where experimentally determined bacterial GluRS crystal structures were used to probe their role, especially the H8-L -H9 motif, in recognizing or discriminating againstA /N -tRNAGlx, identified certain sequence/structural features of the H8-L -H9 motif to be responsible for A -tRNAGlx recognition andN -tRNAGlx discrimination. This was especially found to be true for the proteobacterial class whose genomes are known to mostly contain A -tRNAGlu andN -tRNAGln. Experimental evidence forA -tRNAGlu recognitionN -tRNAGln discrimination by proteobacterial GluRSs also point towards the importance of the A/N -feature of tRNAGlx in tRNAGlx-specificity of proteobacterial GluRSs, for example the role of augmented versus non-augmented D-helix of tRNAGlx in E. coli(18), H. pylori (10) and A. ferroxidans (11). Here we probed the importance of the H8-L -H9 motif in proteobacterial GluRSs, extending the analysis to cases for which GluRS structures are not available.
The analysis is restricted to a small subset of T1/T2 or D/ND probacterial GluRS from three classes (γ-,ε- and α-). Specifically, we focus on three class-specific groups of proteobacteria: i) ε-proteobacterial GluRS(T1/T2) pairs from 11 species (CJR, HPY, WSU, ABU, NIS, TDN, NAM, SKU, SDL, SUN, NSA), ii) γ-proteobacterial GluRS(T1/T2) pairs from 7 species (AFE, MCA, HHA, AEH, NOC, CBU, TGR), iii) α-proteobacterial GluRS(D+) from four species (BJA, OCA, NHA, RPD) and α-proteobacterial GluRS(ND) from 9 species (SME, ATU, RET, LAS, AEX, HCI, CCR, PZU, PUB). Sequence variations of the H8-L-H9 motif were analyzed in parallel with the corresponding D-helix features (either ‘augmented’ or ‘non-augmented’) of tRNAGln and tRNAGlu isoacceptors.
All ε-proteobacterial species considered were associated with aA -tRNAGlu, aN -tRNAGln1 isoacceptor and the complete absence of the tRNAGln2 isoacceptor (34CUG36) and (Figure 6A). For all GluRS(T1), the length of the loop between H8 and H9 helices is 12-residues long while in all GluRS(T2) it is 13-residue long. The 12-residue long loop exhibited the ‘H(N)-G-D-Q(D)-E’ signature motif (except HPY, which exhibits a ‘Y-Q-D-K-E’ motif) with a conserved Arg at position 266 (Fig. 6A and Figure S6). We had already shown that the YQDKE sequence stretch of HYP forms a dynamic α-[>>t ] conformation, incompatible with N -tRNAGln, where YQDK formed a Type II’ -turn. The central QD stretch in the -turn in HPY is replaced by an equally turn-compatible GD stretch in all other ε-proteobacterial GluRS(T1) proteins, and therefore we predict that, likeHpy -GluRS, all other ε-proteobacterial GluRS(T1) proteins considered here will beN -tRNAGln-discriminatory. On the other hand since structural comparison showed that a 13-residue long loop, without the HGD(Q/A) motif at the center, gave rise to the [<<t ]-H8-L -H9 motif, compatible with A -tRNAGlx, we also predict that if H8-L -H9 motif is the sole basis for tRNAGlx-specificity, then GluRS(T2) will charge bothN -tRNAGln1 andA -tRNAGlu.
Unlike ε-proteobacterial species, the seven γ-proteobacterial genomes considered here possess two tRNAGln isoacceptors (Figure 6B and Figure S7): N -tRNAGln1(34UUG36) andA -tRNAGln2(34CUG36). All H8-L -H9 motifs of GluRS(T1) displayed a 12-residue long loop with a central sequence signature HGDQE (and Arg266), implying an α-[>>t ] conformation compatible with A - but not N -tRNAGlx. The only exception is TGR (Thioalkalivibrio sulfidiphilus ) that was an outlier and appeared in cluster B in Figure 3B; it displays the YEVDG sequence, compatible with a [<<t ] conformation (compatible with both A - andN -tRNAGlx). Canonical [>>t ] conformation-competent sequence motifs were also absent in all GluRS(T2), indicating that these also adopt the [<<t ] conformation. Therefore, our hypothesis predicts all GluRS(T2) andTgr -GluRS(T1) to be tRNAGln-non-discriminatory while all other GluRS(T1) to beN -tRNAGlx-discriminatory (X = Glu/Gln). This is consistent with experimental results (11) on AFE (A. ferrooxidans ) where it was shown that GluRS(T1) discriminates againstN -tRNAGln1(34UUG36) while GluRS(T2) is non-discriminatory.
The genomes of all four α-proteobacterial GluRSs(D+) considered here contain two tRNAGln isoacceptors (Figure 6C and Figure S8): N -tRNAGln1 andA -tRNAGln2. All GluRSs also display the α-[>>t ] conformation promoting sequence motif HGDQE (and Arg266), implying that all are expected to discriminate N -tRNAGln1 overA -tRNAGln2 orA -tRNAGlu.
Among the genomes of six (SME, ATU, RET, LAS, AEX, HCI) α-proteobacterial GluRSs(ND) considered here, two (AEX and HCI) contain two one and the rest contain two tRNAGln isoacceptors (Figure 6D and Figure S9): N -tRNAGln1 andA -tRNAGln2, and,A -tRNAGlu. The GluRSs do not contain sequence motifs compatible with the [>>t ] conformation implying that all are expected to be non-discriminating type. This is consistent with their functional requirement of compulsory glutamylation of both tRNAGln and tRNAGlu, in absence of GlnRS in their genomes.
The genomes of other three (CCR, PZU, PUB) α-proteobacterial GluRSs(ND) considered here (Figure 6E and Figure S9) contain only one tRNAGln isoacceptor (A -tRNAGln1) andA -tRNAGlu. The GluRSs show the α-[>>t ] conformation promoting sequence motif HGDDE (CCR, PZU) or YQDKE (PUB) along with Arg266; all three appear in “tRNAGln-discriminatory” clusters in Figure 3B (CCR/PZU in cluster F and PUB in cluster E). This implies that all are expected to discriminate N -tRNAGlx overA -tRNAGlx. AlthoughN -tRNAGlx discriminatory, since their genomes do not contain N -tRNAGlx, essentially the GluRSs are non-discriminatory, consistent with their functional requirement.