Substitutions in conserved regions preceding and within the linker affect activity and flexibility of tRNase ZL, the long form of tRNase Z

The enzyme tRNase Z, a member of the metallo-β-lactamase family, endonucleolytically removes 3’ trailers from precursor tRNAs, preparing them for CCA addition and aminoacylation. The short form of tRNase Z, tRNase ZS, functions as a homodimer and is found in all prokaryotes and some eukaryotes. The long form, tRNase ZL, related to tRNase ZS through tandem duplication and found only in eukaryotes, possesses ~2,000-fold greater catalytic efficiency than tRNase ZS. tRNase ZL consists of related but diverged amino and carboxy domains connected by a flexible linker (also referred to as a flexible tether) and functions as a monomer. The amino domain retains the flexible arm responsible for substrate recognition and binding while the carboxy domain retains the active site. The linker region was explored by Ala-scanning through two conserved regions of D. melanogaster tRNase Z: NdomTprox, located at the carboxy end of the amino domain proximal to the linker, and Tflex, a flexible site in the linker. Periodic substitutions in a hydrophobic patch (F329 and L332) at the carboxy end of NdomTprox show 2,700 and 670-fold impairment relative to wild type, respectively, accompanied by reduced linker flexibility at N-T inside the Ndom- linker boundary. The Ala substitution for N378 in the Tflex region has 10-fold higher catalytic efficiency than wild type and locally decreased flexibility, while the Ala substitution at R382 reduces catalytic efficiency ~50-fold. These changes in pre-tRNA processing kinetics and protein flexibility are interpreted in light of a recent crystal structure for S. cerevisiae tRNase Z, suggesting transmission of local changes in hydrophobicity into the skeleton of the amino domain.


Introduction
Transfer RNA (tRNA) is central to translation [1]. Sequencing of the first tRNAs established a canonical secondary structure (cloverleaf) which arises from intramolecular base pairing, and a conserved L-shaped tertiary structure. D-T loop pairing forms the elbow, with the anticodon and acceptor stem at opposite ends. CCA at the 3' end is universally conserved. C 74 C 75 of P-site tRNA H-bond with large subunit rRNA, positioning it for peptidyl transfer; the 2'OH of A 76 on the peptidyl tRNA participates critically in catalysis.
tRNAs are transcribed as precursors and processed by endonucleolytic removal of a 5' leader by RNase P, first characterized as a ribozyme and later shown to be a protein-only enzyme in mitochondria and chloroplasts (for reviews, see [2], [3]). Some tRNAs are transcribed with introns that are removed by splicing and all tRNAs undergo extensive post-transcriptional nucleoside modification. The 3' trailer is removed by a combination of endo-and exonucleases in E. coli, in which -CCA 76 is transcriptionally encoded. In eukaryotes, -CCA 76 is not transcriptionally encoded; CCA-addition is thus required in eukaryotic nuclei and plastids. tRNase Z provides the principal mechanism for endonucleolytic removal of eukaryotic 3' trailers, leaving the discriminator base (N 73 ) with a 3'-OH ready for CCA addition. This pathway may be complemented by exonucleases in S. cerevisiae ([4], [5]).
The tRNase Z function and a gene encoding the enzyme are widely conserved [6]. A short and long form (tRNase Z S and tRNase Z L , respectively) both endonucleolytically cleave pre-tRNA 3' end trailers, however the two forms are unevenly distributed among the domains of life. Bacteria and archae exclusively possess tRNase Z S . While tRNase Z S is found in some eukaryotes, tRNase Z L is more widespread (for example, tRNase ZS is absent from S. cerevisiae, C. elegans and D. melanogaster).
tRNase Z is a member of the β-lactamase family of metal-dependent hydrolases, characterized by an αβ/βα sandwich fold with the active site located at the interface between the domains [7]. Motifs I-V are conserved, including seven residues (His and Asp) that coordinate binding of two Zn ++ ions which direct H 2 O in general in-line acid-base catalysis, four of them in the signature His cluster (HxHxDH; Motif II).
tRNase Z S functions as a homodimer of identical subunits. In addition to Motifs I-V, a unique flexible arm [8], [9] protrudes from the globular core of tRNase Z and binds the elbow of tRNA, directing the acceptor stem including the scissile bond into the active site of the enzyme. The flexible arm in subunit 1 thus positions the 3' end of the substrate in the active site of subunit 2.
Sequence and structural studies show that tRNase Z L emerged as a tandem duplication of tRNase Z S with subsequent divergence of the amino and carboxy domains (first suggested by [10] and subsequently supported by numerous studies, reviewed in [11]). The amino domain retained the flexible arm but lost the key His and Asp residues from the catalytically important motifs otherwise related by sequence, while the carboxy domain retained functional motifs required for catalysis and lost the flexible arm. The resulting enzyme is better adapted for pre-tRNA 3' end processing, based on~2,000-fold higher catalytic efficiency of H. sapiens tRNase Z L than that of tRNase Z S [12]. tRNase Z L is a monomer in solution based on size exclusion chromatography [13] and in the recently solved crystal structure of S. cerevisiae tRNase Z [11] (a tRNase Z L ).
A 62-85 residue flexible linker joins the conserved, relatively stable amino and carboxy domains in tRNase Z L [13]. The boundary between the linker and the carboxy domain is delineated by homology between the carboxy domain of tRNase Z L and the amino end sequence of tRNase Z S . Interestingly, the linker spans the protein surface like a flexible strap [11]; the interface between the amino and carboxy domains of tRNase Z L is much like the dimer interface of tRNase Z S .
Within the amino domain of tRNase Z L , sequences align with the carboxy domain of tRNase Z L and with tRNase Z S up to and including the flexible arm. In the second half of the amino domain, homology blocks identifiable in tRNase Z L s are less clearly related to sequences in the carboxy domain.
We used previously developed methods ( [14]; [13]) to investigate the function and flexibility in regions preceding and within the linker. Ala scans (substitution of alanine for each wild type residue) with processing kinetics were performed followed by flexibility analysis of selected variants. Results were interpreted based on local changes in hydrophobicity in light of the newly available S. cerevisiae tRNase Z structure [11].

Ala scanning mutagenesis
Conserved regions were selected for Ala scanning mutagenesis, one just before the flexible linker and two within the linker. N dom T prox consists of 19 residues in the last homology block in the amino domain on the amino side of the linker (H315 -G 333 ). T flex consists of 9 residues from the most flexible conserved internal region of the linker (M 376 -R 384 ). The PEEY region, glutamate rich and less conserved, consists of 9 residues further toward the carboxy end of the linker (P 397 -H 405 ). These 37 residues were individually substituted with alanine by replacing the wild type codon at each position with a GCC triplet using A, B amplification and A-B segment joining by PCR and overlap extension PCR, as previously described [18].~40-mer oligonucleotides were typically used with the 1, 2 or 3 nt substitution in the middle and with a GCrich cluster at the 3' end for stability of primer annealing. The AflII site (nt 1077-1082) subcloning forward primer combined with the reverse mutagenesis primer were used to amplify the A segment using a wild type template. The coding strand (forward) mutagenesis primer combined with the SacI site (nt # 1527-1532) subcloning reverse primer were used to amplify the B segment. A and B segments were gel purified and joined by overlap extension and amplification using the AflII forward and SacI reverse primers. Joined segments were gel purified, recovered, double digested, recovered, and ligated into the AflII-SacI digested vector from which the 454 bp wild type segment had been removed. Plasmids that passed the RE screen were sequenced (Macrogen) to confirm presence of each intended GCC codon and absence of any other sequence changes. The FastBacHT (Invitrogen) transfer vectors with variant tRNase Z cDNAs were transposed into bacmids using DH10Bac (Invitrogen). Large true white colonies produced by successful transformation and transposition were selected for bacmid DNA isolation and transfection into insect Sf9 cells using Cellfectin 2 reagent (Invitrogen).

Baculovirus expression and affinity purification
Amplified baculoviruses with variant D. melanogaster tRNase Z cDNAs were used to infect insect Sf9 cells for 72 h using Hyclone SFX insect cell medium supplemented with 0.5% FBS to minimize degradation of recombinant proteins by endogenous proteases. Cells were lysed with NP40, expressed proteins were affinity purified using Ni-NTA Sepharose (Qiagen) and the 6XHis tag was cleaved overnight at 4˚C with AcTEV protease as previously described [18].

tRNase Z reaction kinetics
Nuclear encoded pre-tRNA Arg transcript was prepared with T7 RNA polymerase and cleaved using a cis-acting hammerhead leaving a 5'-OH at +1 of the tRNA as previously described [19]. Kinasing with γ-32P-ATP by polynucleotide kinase was performed at +1 of the tRNAs, followed by gel purification and recovery. The processing reaction buffer (PB) consisted of 25 mM Tris-Cl pH 7.2, 2.5 mM MgCl 2 , 1 mM freshly prepared DTT, and 100 μg/ml BSA. Unlabeled substrate concentration was varied over a range of 4-100 nM with a fixed trace amount of 5'-labeled substrate. tRNase Z stocks were adjusted to 25 μM before use from which a dilution series was prepared. Analytical lanes were run with known concentration standards for both input tRNase Z and unlabeled tRNA and the enzyme and unlabeled substrate concentrations used in each experiment was corrected accordingly. Reactions at 28˚C were sampled after 5, 10 and 15 min, and quenched with formamide-marker dye mix on ice. Electrophoresis of the samples was carried out on a 6% polyacrylamide gel containing 8 M urea. Gels were dried and exposed overnight using a phosphor screen, which was scanned using a Typhoon 9410 imager and analyzed with IQTL v8.1. Each lane trace yielded a % product and the time course results were converted to % product/min using Excel, equivalent to 0.01 X V/[S], then converted to V X 10 −11 M/min by multiplying by nM [S], and further analyzed using the single ligand binding function in SigmaPlot. k cat was obtained by dividing V max by [E]. Concentration of each variant enzyme was adjusted as necessary depending on the impairment factor observed in previous kinetic experiments. The processing experiments with each variant were repeated until acceptable standard errors were achieved.

Flexibility of wild type and variant tRNase Z analyzed by limited proteolysis and protein electrophoresis
Wild type and selected variant tRNase Zs were proteolyzed with trypsin at 1 μg/ml in PB at 28˚C and reactions were sampled after 0, 3, 10 and 30 min reaction. Limited proteolysis reactions were analyzed on 1D SDS polyacrylamide gels or using a 2D system (BioRad) with isoelectric focusing in 0.75 mm diam 1 st dimension tube gels and SDS electrophoresis in the 2 nd dimension as previously described [13]. Protein bands and spots were detected by staining with Sypro Orange and scanning with a Typhoon 9410 and quantitated using IQTLv8.1.

Results
A local hydropathy plot [15] provides a useful extension to PsiPred for interpretation of tRNase Z structure and flexibility (Fig 1). For example, pronounced hydrophobicity troughs found close to both ends of the protein, typical of globular proteins in aqueous solution, coincide with flexible regions (cf [13]). N-T and T flex , the two most flexible regions in the linker, are also predicted local hydrophobicity troughs.
The carboxy domain of tRNase Z L is homologous to tRNase Z S , including the active site. Similarly, the flexible arm (FA) in tRNase Z L is related to one of the three branches of flexible arms [9], and the sequence that precedes it is also related, in agreement with the evolution of tRNase Z L from a tandem duplication of tRNase Z S followed by divergence of the amino and carboxy domains. Less is known about the flexible linker of tRNase Z L , however. The S. cerevisiae tRNase Z linker closely follows the exterior contours of the protein as it joins the amino and carboxy domains ( [11] ; Fig 2). A multiple sequence alignment (Fig 2A; see [20]) combined  [13] which occur at flexible, hydrophilic regions. Directly above the predicted secondary structure, a hydropathy plot (created with ExPASy using the Wolfenden scale [15]) depicts the relative hydrophobic and hydrophilic character of the corresponding regions, the dashed red line indicating approximate neutrality. Little structural information was available on the amino domain and linker until the recent publication of a S. cerevisiae tRNase Z structure [11]. The basic structure of the S. cerevisiae tRNase Z amino domain is an αβ/βα sandwich fold, like that of the carboxy domain. The flexible arm, located between two strands of β twisted sheet, is extruded from the body of the amino domain. The tRNase Z linker spans the globular core of the enzyme like a strap (Fig 2B; [11]). The linker is an adjunct to, not a substitute for, the domain interface between the amino and carboxy domains, which is much like that observed in the tRNase Z S homodimer ( [11]; cf [8]). N dom T prox (within the N domain, proximal to the linker) is the last such homology block preceding the linker [19], [13]. Based on the S. cerevisiae tRNase Z crystal structure [11], the 1 st half of N dom T prox has little secondary structure, followed by a short β strand and an α helix (α8) with high local hydrophobicity. A flexible hydrophilic patch located on the linker side of the amino domain-linker boundary designated N-T, less conserved than N dom T prox , which in S. cerevisiae tRNase Z consists of a short helix followed by a β strand (β13), gives rise to the limited proteolysis species C dom 1 [13]. Another conserved flexible hydrophilic region designated T flex , found~35 residues within the linker, gives rise to the C dom 2 family of proteolysis products.
The regions subjected to single residue Ala substitution and kinetic analysis include 19 residues in N dom T prox and 13 residues in T flex . A short sequence further into the linker is characterized by contiguous glutamates (PEEY region). The goal of an Ala scan is to discover residues of sufficient importance that, when replaced by Ala, cause a significant effect on enzyme activity. Such effects were not observed within the PEEY region, which will therefore not be discussed further. The N-T region is generally conserved in location and hydrophilicity but does not align well and was therefore not examined. Once results of processing kinetics were available, flexibility of selected variants with suggestive functional impairments were studied by limited proteolysis with trypsin and protein gel electrophoresis as in [13].

Substitutions in two bulky hydrophobic N dom T prox residues close to the N dom -linker boundary greatly impair processing and also reduce flexibility in the N-T region
The Ala scan processing results in the 1 st half of N dom T prox , suggested by PsiPred to be in αhelix, are unremarkable. Alanine substitutions in two bulky hydrophobics spaced three residues apart close to the carboxy end of N dom T prox , Phe329Ala and Leu332Ala, strikingly impair processing with impairment factors of 2,700X and 700X relative to wild type (Figs 3 and 4). In the example illustrated (Fig 3), it was necessary to use the Phe329Ala variant at a >1,000-fold higher concentration than wild type enzyme to obtain a comparable series of processing time courses over the range of unlabeled substrate concentrations used in kinetic experiments.
These substitutions for bulky hydrophobic residues on the carboxy side of N dom T prox were selected for further examination for limited proteolysis with trypsin and protein gel electrophoresis (Fig 5 and data not shown). Phe329Ala demonstrated a marked change in the ratio of stable C dom products produced upon trypsin cleavage compared to WT tRNase Z (similar results were obtained from Leu332Ala, not shown). The N dom T prox region is proximal to the preferred trypsin N-T cleavage site at K 348 /K 351 which produces stable C dom 1species (accompanying schematics at bottom of Fig 5) that differ slightly in size and charge depending on cleavage at clustered basic residues ( [14]; Fig 1). The T flex site further into the linker at R 384 / K 385 gives rise to the smaller C dom 2 species. The C dom 1 to C dom 2 ratio in WT tRNase Z is 2:1; in the F 329 variant this ratio decreases to 0.33:1, showing that the alanine substitution at F 329 locally reduces N-T site flexibility.

Effects of T flex region substitutions on processing kinetics and local flexibility
Of the nine T flex alanine variants expressed and analyzed with processing kinetics, Arg382Ala at the carboxy end of the T flex region showed the greatest impairment factor, an approximately 50X reduced processing efficiency relative to WT tRNase Z (Fig 6). Multiple sequence alignment shows this to be a conserved residue (Fig 2). Asn378Ala, a substitution in a non-conserved residue near the amino boundary of T flex , unexpectedly showed a tenfold increase in processing efficiency (Figs 6 and 7). Additionally, the Asn378Ala substitution markedly reduces local flexibility as shown by limited proteolysis (Fig 8). The T flex region includes the trypsin cleavage site at R 384 /K 385 which gives rise to the stable C dom 2 species. In WT tRNase Z the spot intensity ratio of C dom 1 to C dom 2 is 1.   value obtained in Fig 5). For the Asn378Ala variant this ratio increases to 4:1. Alanine substitution at N 378 thus causes a dramatic decrease in C dom 2 species seen after trypsin digestion due to a local decrease in flexibility.

A subdomain defined by interior hydrophobicity arises from interactions across the amino domain-Flexible linker boundary
The greatest impairment of tRNase Z activity obtained in the Ala scan through the N dom T prox region was observed with substitution of bulky hydrophobics spaced three residues apart (F 329 , L 332 ) toward the carboxy end of the region (Figs 3 and 4). The most closely corresponding residues in S. cerevisiae tRNase Z are Y 361 and F 364 in α8 (Fig 2A). If the backbone in this region is α-helical or helix-like (in the D. melanogaster sequence a proline at 330 would be expected to interrupt an α-helical path ; Fig 1), these bulky R-groups would point in roughly the same direction, with potential to collaborate in formation of a hydrophobic cluster. Such a local structural subdomain inflated with high hydrophibicity would not be located deep within the protein considering that the flexible linker spans the enzyme surface (Fig 2B). Based on the recent structure 5MTZ [11], the best candidate hydrophobic partners are I 391 and I 393 in β13 of S. cerevisiae tRNase Z (Fig 9A). α8 in N dom T prox is the last homology block at the carboxy end of the amino domain before the start of the flexible linker. β13 is on the carboxy side of the  A and B). Additionally, 10 minute reactions were electrophoresed using 1 st dimension isoelectric focusing followed by 2 nd dimension SDS-PAGE (right panels in A and B). C dom 1 species are marked in blue ( on the 1D SDS-PAGE and enclosed in a dashed ellipse in the accompanying 2D gel); C dom 2 species in green ( on the 1d SDS-PAGE and enclosed by a green dashed ellipse in the 2D gel). Cleavage at clusters of basic residues accounts for the multiple C dom 1 and C dom 2 species. C dom 1/C dom 2 ratios determined with IQTL are shown at lower right of the 2D gel panels in A, B. The schematic diagrams below illustrate the location of N dom T prox with respect to the N-T and T flex sites. N-T hydrophilic patch that marks the amino boundary of the flexible linker, corresponding to the flexible region sensitive to trypsin (K 348 KTKL) in D. melanogaster tRNase Z which gives rise to the C dom 1 species (Figs 5 and 8, cf [14]). Corresponding hydrophilic residues in S. cerevisiae tRNase Z (E 387 KDN; blue in Fig 9) are in a short helical element with R-groups facing solvent. Bulky hydrophobic pairing partners for D. melanogaster F 329 and L 332 in the N-T region of the flexible linker cannot be identified due to imperfect alignment (Fig 2A).
Internal subdomains are apparently created by juxtaposition of several bulky hydrophobic R groups shielded from solvent, producing a micellar spherule inflated like a beach ball ( Fig  9A). Substitution of either of the identified bulky hydrophobic R groups in N dom T prox with the single methyl group of alanine (white in Fig 9B and 9C) leads to hydrophobicity collapse (illustrated with dashed ellipses and arrows). The Y 361 side chain -OH also makes a polar contact The Flexible Linker of tRNase Z L collapse in α8/β13 could thus be transmitted into the twisted β sheet in the amino domain on the carboxy side of the flexible arm, as suggested by the enlarged view in Fig 10B. β13 is a member on one edge of a 7-stranded β sheet, in which H 392 makes backbone Hbonds with the carboxy group of H 315 and the amino group of H 317 in β12 (Fig 10A and  10B). The first four strands from β13 (13/12/11/10) are parallel; the last two strands (10/9/8/ 7) are antiparallel, and β9,10 ascend to and descend from the flexible arm, respectively. The collapse (deflation) of the α8-β13 spherule arising from substitution of the specific bulky hydrophobic residues in N dom T prox with Ala (Fig 9B and 9C) damages the overall fold of tRNase Z, explaining the 2,700-fold and 700-fold impairment of tRNase Z activity (Figs 3  and 4). This also reduces local flexibility (Fig 5) by occluding the N-T site that produces the C dom 1 family of spots relative to T flex , which produces C dom 2. In some ways, these longrange effects of changes in internal subdomain hydrophobicity resemble those of the L187A substitution at the flexible arm-hand boundary in the ascending stalk of D. melanogaster tRNase Z, which causes a close to 100-fold impairment in enzyme activity due to increased K M [14], accompanied by increased flexibility [13].
T flex coincides with a short β strand (β15), one of two short antiparallel β strands in the linker (β15-14) which join a twisted sheet (β1-β6) on the amino side of the flexible arm through backbone H-bonds between β14-β1 (Fig 10C). Concerning the strongest impairment observed in the region with the R382A substitution (Figs 6 and 7), ionized residues on the surface of the protein such as E 419 and D 422 in the S. cerevisiae tRNase Z β15-α9 loop face the polar solvent as expected for T flex . Replacement with a small hydrophobic residue could lead to structural eversion in which the substituted residue buries itself in a partially exposed hydrophobic patch, like the effects of the HbS substitution on hemoglobin structure and function. associates here with the twisted β sheet which forms half the skeleton of the amino domain, on the amino side of the flexible arm (Fig 10C).
Based on the alignment in Fig 2A, N 415 in S. cerevisae tRNase Z is the most similar residue in position and identity to N 378 in D. melanogaster tRNase Z. Replacement of N 415 in β15 with a small hydrophobic residue would locally reduce linker flexibility by strengthening skeletal architecture of the amino domain preceding the flexible arm. The reduced linker flexibility arising from the N378A substitution in D. melanogaster tRNase Z (Figs 6 and 7) could thus improve catalytic efficiency by stiffening the skeleton of β structure on the amino side of the flexible arm. Also noteworthy in this regard, the conservative substitution Leu423Phe in H. sapiens tRNase Z L (ELAC2) associated with mitochondrially based cardiac hypertrophy [21] is located at the start of β15.

Conclusion
A biochemical exploration of little-understood regions of D. melanogaster tRNase Z through Ala scanning mutagenesis followed by processing kinetics was aided by analysis of flexibility using limited proteolysis and two-dimensional protein electrophoresis. This approach, informed by interpretation of a recent crystal structure of the S. cerevisiae homolog, uncovered a previously unknown hydrophobic subdomain formed across the amino domain-linker boundary, leading us to suggest that peripheral substitutions affect the skeleton of twisted β sheets in the amino domain on both sides of the flexible arm.