Two complementing in vivo selection systems based on CCA-trimming exonucleases as a tool to monitor, select and evaluate enzymatic features of tRNA nucleotidyltransferases

Abstract

tRNA nucleotidyltransferase represents a ubiquitous and essential activity that adds the indispensable CCA triplet to the 3'-end of tRNAs. To fulfill this function, the enzyme contains a set of highly conserved motifs whose coordinated interplay is crucial for the sequence-specific CCA polymerization. In the human enzyme, alterations within these regions have been shown to lead to the manifestation of disease. Recently, we developed an in vivo screening system that allows for the selection and analysis of tRNA nucleotidyltransferase variants by challenging terminal AMP incorporation into tRNA during induced RNase T-catalyzed CCA-decay. Here, we extend this method for screening of full CCA-end repair by utilizing the CCA-trimming activity of exonuclease LCCR4. To demonstrate the combined potential of these two in vivo selection systems, we applied a semi-rational library design to investigate the mode of operation of catalytically important motifs in the human CCA-adding enzyme. This approach revealed unexpected requirements for amino acid composition in two motifs and gives new insights into the mechanism of CCA addition. The data show the potential of these RNase-based screening systems, as they allow the detection of enzyme variations that would not have been identified by a conventional rational approach. Furthermore, the combination of both RNase T and LCCR4 systems can be used to investigate and dissect the effects of pathogenic mutations on C- and A-addition.

Keywords: CCA-adding enzyme; CCA-trimming exonuclease; RNase LCCR4; RNase T; in vivo screening of CCA-adding activities; tRNA nucleotidyltransferase.

Graphical abstract

Figure 1.

Coexpression of CCA-trimming exonucleases can be used to monitor the CCA-adding activity of recombinantly expressed tRNA nucleotidyltransferases in vivo. (A) Principle of the in vivo system is based on a pETDuet-derived vector enabling dual expression of an RNase (brown) and a tRNA nucleotidyltransferase of interest (cyan), subjecting the tRNA pool in E. coli JM109(DE3) Δcca to CCA-end hydrolysis and restoration. (B) E. coli growth was examined on LBamp plates in an IPTG gradient from 0 to 200 µM, resulting in a gradual increase of tRNA-CCA end hydrolysis mediated by recombinant co-expression of either E. coli RNase BN (left) or T. brucei LCCR4 (right). Only enzymes with full CCA-adding activity give rise to selective bacterial growth. Removal of CCA-ends by RNase BN results in efficient growth of cells coexpressing the wt CCA-adding enzyme. Co-expression of the inactive enzyme variant EcoCCA (AxA) or the A-adding enzyme BhaA also lead to cell growth, although at a clearly restricted efficiency. This is an indication that under RNase BN expression, a considerable number of tRNAs still carry a complete or partial CCA-end, interfering with a selective growth depending on CCA-adding activity. EcoCCA, E. coli CCA-adding enzyme; DxD, wildtype enzyme with catalytically active carboxylates; AxA, inactive enzyme variant with catalytic carboxylates replaced by alanine residues; BhaA, B. halodurans A-adding enzyme.

Figure 2.

HsaCCA R105A is severely impaired in full CCA addition. (A) As previously hypothesized [10,47], a salt bridge could be formed between R105 in the flexible loop region and E164 in motif D to switch the enzyme’s specificity from CTP to ATP incorporation. The presented model of the protein region including the flexible loop is based on the crystal structure of HsaCCA (pdb entry 1OU5) according to Hoffmeier et al. [10]. (B) Complementation of HsaCCA R105A (2) does not promote growth when selecting for A- or cca-repair activity, while the wildtype (wt) enzyme (1) rescues the growth defect.

Figure 3.

In vivo selection of position 105 in HsaCCA. Codon 105 was randomized in the open reading frame of HsaCCA. The resulting pool was subjected to selection in both RNase T-based (A) as well as LCCR4-based selection system (B). Under both conditions selecting for A- or CCA-addition, only arginine codons could be retrieved, indicating that this residue is invariant and essential for complete CCA-addition. As the NNK (K = G, T) randomization in the RNase T system includes the arginine codons CGT, CGG and AGG, all these codons could be selected. In the LCCR4 system, the 22c-trick mutagenesis was applied. This includes only one (CGT) arginine codon. Hence, only this codon appeared in the selection. To avoid any artificial arginine retrieval, the starting sequence for randomization contained an alanine codon GCA at this position (red). Hence, a selection bias due to residual wt sequences can be excluded. For both selections, a representative number of individual clones is shown.

Figure 4.

Semirational library design uncovers HsaCCA enzyme variants with altered active site residues capable of efficiently repairing tRNA 3′ends. (A) Amino acid template in motif D of class II CCA-adding enzymes and its specific interaction with ATP (green). The model is based on the co-crystal structure of GstCCA, the CCA-adding enzyme from Geobacillus stearothermophilus (1MIW) [6]. ATP and relevant amino acid residues are presented as stick models, hydrogen bonds are shown as black dashed lines; the structural surrounding was omitted for reasons of clarity. In the library selection with randomized positions 105 and 164, asp (D), glu (E) and gln (Q) appeared at position 164 (Figure S4). These residues (with their respective frequency at which they were retrieved) were placed by the PyMOL mutagenesis wizard and were further highlighted in dot representation. (B) In the same selection, glycine (G) appeared at position 164. In this situation, the templating side chain of R168 has a considerably increased positional freedom (indicated by the double arrow) that reduces the specificity of the NTP binding pocket. (C) After introducing a DNA library with tailored codon diversity at the three conserved template positions 164, 165 and 168 in HsaCCA motif D, the retrieved sequences from the RNase T selection system – and to a much lower extent from the LCCR4 system – show alternative amino acid side chains only at position 164 (Figure S5). Letter heights in the sequence logos correlate with frequency at which they were retrieved after selection. As the randomized library contained between 5 and 6% E, D and G codons, respectively, this result represents a clear codon enrichment based on enzyme function. Sequence logos were created using the WebLogo generator [49]. (D) Growth of E. coli JM109(DE3) Δcca complemented with the retrieved HsaCCA variants was assessed on LBamp in an IPTG gradient from 0 to 100 µM, resulting in a gradual increase of RNase T-mediated tRNA end hydrolysis. A representative series of gradient plates out of 3 independent experiments is shown. Cells expressing the control enzyme EcoCCA wt tolerate very high RNase T stress, resulting in colonies over 77.1 ± 4.3% of the IPTG gradient distance. HsaCCA wt is less active, but still results in considerable RNase T tolerance, allowing cell growth up to 44.8 ± 1.3% of the gradient. Variants E164D and E164G are viable at low RNase T expression levels (E164D: growth distance of 22.2 ± 1.2%; E164G: 21.5 ± 0.7%), while variant E164A is not viable at all, even at the lowest RNase T levels.

Figure 5.

Catalytic properties of HsaCCA wt and variant GDxxR (E164G). (A) in vitro nucleotide incorporation analysis with recombinant enzymes. Upper panel: complete CCA-addition on a tRNA substrate lacking the CCA-terminus. When all four NTPs (N) or only CTP (C) are offered, both enzymes show efficient incorporation activity. A partial incorporation activity in the presence of UTP (U) can be detected for both enzymes. Full length reaction products were isolated and analyzed by sequencing; numbers indicate the amount of correct CCA ends in the analysis of 13 and 17 clones, respectively (see also Table 1). Lower panel: A-addition on tRNA-CC. Both enzyme versions readily add the terminal A if NTPs (N) or only ATP (A) is offered. M, mock incubation in the absence of enzyme. (B) Timeresolved product formation in the presence of 0.5 and 1.0 ng enzyme per 20 µl reaction shows a mild decrease in catalytic activity for variant E164G compared to the wildtype situation. M, mock incubation without enzyme.

Figure 6.

NTP binding of HsaCCA wt and HsaCCA E164G. The DRaCALA assay shows unambiguously that both enzymes interact with CTP (left panel), while only the E164G variant is also able to bind ATP (right panel). Free radioactively labelled NTPs diffuse on the nitrocellulose membrane, leading to a grey circle. Enzyme-bound NTPs remain at the position where the protein is immobilized, resulting in a dark small circle in the centre of the spot. Binding was quantified by relative signal densities of the bound versus the unbound NTP. While the wt enzyme only binds CTP, HsaCCA E164G interacts with CTP as well as ATP, and the existence of both binding-competent states reduces the efficiency of CTP recognition, resulting in reduced relative signal intensity compared to the wt enzyme. As both proteins showed increasing precipitation at higher concentrations, dissociation constants could not be determined.

Figure 7.

Growth behaviour of E. coli expressing HsaCCA variants on RNase T gradient plates. Depicted gradient plates represent one assay of a triplicate analysis (n = 3). Compared to HsaCCA wt, HsaCCA D139A does not result in a reduced cell viability in the presence of RNase T and grows up to a distance of 43.6 ± 0.5%. As the in vitro A-adding activity of HsaCCA D139A is 15-fold reduced, this unchanged growth behaviour is an indication for an efficient endogenous A-addition backup system by poly(A) polymerase and PNPase. HsaCCA E164G, however, leads to a reduced growth of 21.5 ± 0.7%, indicating that this mutation is more detrimental, and the backup system is not sufficient to restore wt growth. The double variant HsaCCA D139A E164G shows a further growth reduction of down to 14.0 ± 1.5%, suggesting that the combination of both amino acid replacements leads to a further saturation of the A-addition backup.

Figure 8.

Growth behaviour of E. coli host cells expressing HsaCCA carrying disease-causing mutations. Representative plates of a triplicate analysis are shown (n = 3). Left panel: monitoring of A-addition in the RNase T system. wt HsaCCA expression leads to a growth distance of 43.1% of the total plate distance, while pathogenic variants R70W and D134V reduce the growth distance down to 36.9% and 24.3%, respectively. A combination of both completely abolishes cell growth. The equally pathogenic mutation M129V shows almost no effect, as the cells grow up to a distance of 41.2%, nearly identical to that of cells expressing the wt enzyme. Right panel: monitoring full CCA-addition in the LCCR4 system. Due to the increased stringency, R70W, D134V and R70W/D134V enzyme variants cannot restore viability in E. coli, indicating that C-addition is strongly affected. Expression of HsaCCA M129V, however, leads to considerable growth to up to 36.4%. Here, a slight reduction compared to the wt enzyme is visible. Hence, in this variant, A-addition is similar to the wt level, while C-addition is affected, but not as severe as in the other tested pathogenic variants.