Archaeal Tuc1/Ncs6 Homolog Required for Wobble Uridine tRNA Thiolation Is Associated with Ubiquitin-Proteasome, Translation, and RNA Processing System Homologs

While cytoplasmic tRNA 2-thiolation protein 1 (Tuc1/Ncs6) and ubiquitin-related modifier-1 (Urm1) are important in the 2-thiolation of 5-methoxycarbonylmethyl-2-thiouridine (mcm5s2U) at wobble uridines of tRNAs in eukaryotes, the biocatalytic roles and properties of Ncs6/Tuc1 and its homologs are poorly understood. Here we present the first report of an Ncs6 homolog of archaea (NcsA of Haloferax volcanii) that is essential for maintaining cellular pools of thiolated tRNALys UUU and for growth at high temperature. When purified from Hfx. volcanii, NcsA was found to be modified at Lys204 by isopeptide linkage to polymeric chains of the ubiquitin-fold protein SAMP2. The ubiquitin-activating E1 enzyme homolog of archaea (UbaA) was required for this covalent modification. Non-covalent protein partners that specifically associated with NcsA were also identified including UbaA, SAMP2, proteasome activating nucleotidase (PAN)-A/1, translation elongation factor aEF-1α and a β-CASP ribonuclease homolog of the archaeal cleavage and polyadenylation specificity factor 1 family (aCPSF1). Together, our study reveals that NcsA is essential for growth at high temperature, required for formation of thiolated tRNALys UUU and intimately linked to homologs of ubiquitin-proteasome, translation and RNA processing systems.


Introduction
Posttranscriptional modifications of RNA are widespread and varied amongst all evolutionary lineages with over 100 different modifications of the four canonical RNA nucleotides [1][2][3]. RNA modifications play critical roles in cell metabolism and RNA structural stability [3]. In particular, modification of tRNA is important for proper codon-anticodon base-pairing and decoding [4,5]. One such tRNA modification is the 2-thiomodification of the wobble uridine of tRNAs specific for lysine (tRNA Lys UUU ), glutamate (tRNA Glu UUC ), and glutamine (tRNA Gln UUG ) [6], which was recently reported to enhance translational efficiency by increasing ribosomal A-site binding and peptide bond formation based on in vitro study of yeast [7].
The 2-thiomodification of wobble uridine tRNAs in yeast relies on a series of enzymes for the activation and incorporation of sulfur into the tRNA. In the early stages, the thiosulfate sulfurtransferase homolog Tum1/YOR251c is found to stimulate and accept persulfide sulfur from the cysteine desulfurase Nfs1 [8]. The ubiquitin-related modifier 1 (Urm1) and E1-like enzyme Uba4 intersect this Nfs1-Tum1-mediated sulfur relay [8]. The Cterminal a-carboxylate of Urm1 is activated as an acyl-adenylate and thiocarboxylated by Uba4 through Nfs1-Tum1 sulfur transfer [8]. The Urm1 thiocarboxylate can be utilized in subsequent in vitro reactions for the 2-thiolation of wobble uridine tRNAs presumed to be adenylated by a thiouridylase complex of Ncs6 (Tuc1) and Ncs2 (Tuc2) [8].
While thiolated tRNA is identified in Archaea [2,9,10], the source and incorporation of this sulfur into the tRNA is not well studied. A recent report suggests sulfide can act as a sulfur donor for 4-thiouridine biosynthesis in Methanococcus maripaludis tRNA [11]. Biosynthesis of 2-thiouridine in tRNA of the haloarchaeon Haloferax volcanii has also been suggested from study of small archaeal modifier proteins (SAMPs) [12]. SAMP2 and the E1-like ubiquitin-activating homolog, UbaA, are found important in not only posttranslational protein modification but also in the formation of thiolated tRNA Lys UUU indicative of 2-thiolation of wobble uridine tRNAs [12]. A Tuc1/Ncs6 homolog (HVO_0580, named NcsA), predicted to be associated with 2-thiouridine formation, was also found to co-immunoprecipitate with SAMP2 suggesting that NcsA is covalently attached to SAMP2 and that sampylation may regulate tRNA modification [13].
Here we report the characterization of Hfx. volcanii NcsA. NcsA was found important for the cellular pools of thiolated tRNA Lys UUU and growth at elevated temperatures. NcsA was covalently modified by apparent polySAMP2 chains through an UbaA-dependent mechanism and was non-covalently associated with homologs of the eukaryotic ubiquitin-proteasome and exosome systems. Taken together, our results suggest the haloarchaeal Ncs6 (Tuc1) homolog, NcsA, is important for 2thiolation of wobble uridine tRNAs and is intimately linked with post-translational systems including ubiquitin-like protein modification, proteasomes, translation and RNA processing.

Results
NcsA and its haloarchaeal homologs form a distinct subgroup within the adenine nucleotide a hydrolase (ANH) superfamily and have conserved tRNA thiolase active site residues Hfx. volcanii HVO_0580 (NcsA) is a member of the adenine nucleotide a hydrolase (ANH) superfamily (cd01993) and is predicted to be involved in tRNA thio-modification based on Gene Ontology annotation (GO:0034227) and sequence similarity to tRNA modification enzymes such as Ncs6 (Tuc1). In this study, hierarchical clustering was used to further understand the relationship of NcsA to members of the ANH protein superfamily ( Figure S1 in File S1). Hfx. volcanii NcsA was found to form a tight cluster with uncharacterized ANH superfamily members from other haloarchaea. Proteins of the haloarchaeal-specific ANH cluster were related to eukaryotic Ncs6 (Tuc1) and relatively distinct from the other bacterial and archaeal members that have been characterized including: Salmonella enterica serovar Typhimurium TtcA [14] Thermus thermophilus TtuA [15][16][17], and Pyrococcus horikoshii Ph0300 [17]. These protein sequence relationships suggested that new insight would be provided through biochemical and genetic study of NcsA.
We next determined whether Hfx. volcanii NcsA had conserved active site residues common to ANH superfamily members using Phyre2-based homology modeling and multiple amino acid sequence alignment (Figure 1, Figure S2 in File S1). By this approach, NcsA was found to have a conserved 3D-structural fold and residues common to Ncs6 and TtuA of the TtcA family group II including the five C-X 2 -[C/H] motifs and the PP motif (P-looplike motif in a widespread ATP pyrophosphatase domain; SGGXDS, where X is any amino acid residue) [14,[17][18][19]. Based on recent study of TtuA by in vivo site-directed mutagenesis and xray crystallography, the first and second C-X 2 -[C/H] motifs form an N-terminal zinc finger (ZnF1), the third C-X 2 -C forms the putative catalytic active site and the C-terminal zinc finger (ZnF2) is formed by the fourth and fifth C-X 2 -C motifs [17]. Thus, Hfx. volcanii NcsA is predicted to have conserved residues of the cysteine-rich-and PP-motifs that mediate the binding, adenylation and thiolation of tRNA.
NcsA is necessary for thio-modification of lysine tRNA with a wobble uridine We next used a genetic approach to investigate the role of NcsA in the thiolation of tRNA. Hfx. volcanii strains with a markerless deletion of the ncsA (hvo_0580) gene and its trans-complement (that expressed NscA with a C-terminal StrepII-tag, NcsA-StrepII) were generated from parent strain H26 and confirmed by Southern blotting, PCR and DNA sequence analyses ( Figure S3 in File S1). Total RNA was purified from these strains and analyzed for thiolation of wobble uridine tRNA by use of acryloylaminophenylmercuric chloride (APM) gel electrophoresis coupled with Northern blotting using a probe specific for tRNA Lys UUU ( Figure 2). The tRNA Lys UUU probe was chosen based on the presence of a uridine nucleoside in the wobble position of its anticodon specific for lysine tRNAs. By this experimental approach, a fraction of tRNA Lys UUU in the parent and transcomplement strains was found to be thio-modified ( Figure 2, lanes 1 and 3). By contrast, only non-thiolated tRNA Lys UUU was detected in the DncsA mutant strain (Figure 2, lane 2). Taken together, these results revealed NcsA is required for the thiomodification of the wobble uridine of the tRNA UUU specific for lysine, similarly to what has been previously observed for the ubiquitin-fold SAMP2 and E1-like UbaA [12]. Thus, UbaA, SAMP and NcsA may function like the eukaryotic Uba4, Urm1 and Ncs6 in the thiol-modification of wobble uridine tRNAs (i.e., tRNA Lys UUU , tRNA Glu UUC and tRNA Gln UUG ).
NcsA is necessary for growth at elevated temperature We next examined whether NcsA is necessary for optimal growth at elevated temperature similar to SAMP2 and UbaA [12]. Hfx. volcanii DncsA mutant and its trans complement were compared to H26 parent and Dsamp2 mutant for growth at 50uC. Growth at this elevated temperature was compared to growth at 42uC, a temperature within the range for optimal growth of Hfx. volcanii [20]. When cultured at 42uC, all four Hfx. volcanii strains were found to have comparable growth rates and cell yield under all conditions tested ( Figure S4 in File S1). By contrast, when cells were grown at 42uC and transferred to 50uC, a slow-growth phenotype was observed for the DncsA and Dsamp2 mutant strains compared to the parent and ncsA trans-complement ( Figure 3A). To examine whether this slow-growth phenotype may be attributed to suppressor mutation(s), the four Hfx. volcanii strains were grown to stationary phase at 50uC, inoculated into fresh medium, and monitored for growth at 50uC ( Figure 3B). By this experimental approach, the DncsA and Dsamp2 mutant strains were found to display no detectable growth at 50uC compared to the robust growth and cell yield detected for the parent and trans-complement strains ( Figure 3B). Similar results were observed by rich medium agar plate assay ( Figure 3C). Thus, the slow growth phenotype of the DncsA and Dsamp2 strains was not due to a suppressor mutation but instead is likely due to a component(s) present in the 42uC inoculum that was relatively active for initial batch culture at 50uC but not functional for long-term growth at this elevated temperature.
UbaA mediates covalent and non-covalent associations of NcsA with SAMP2 To further characterize NcsA, the C-terminal StrepII tagged variant of NcsA (NcsA-StrepII) that complemented the DncsA mutant for growth and tRNA thiolation was purified by StrepTactin chromatography. Strains that co-expressed NcsA-StrepII with either N-terminal Flag-tagged (Flag-) SAMP1 or Flag-SAMP2 were included in the purifications. Flag-SAMP1 served as a negative control, as SAMP1 is not needed for tRNA Lys UUU thiolation [12] nor predicted to interact with NcsA based on previous MS analysis [13]. The DncsA mutant with the empty vector control, pJAM202c, was also used to detect any non-specific proteins that may co-purify by StrepII affinity chromatography. Expression of StrepII-and Flag-tagged proteins was confirmed by . Conserved residues are highlighted in red, grey and black, with the conserved residues in red of the ATP pyrophosphatase signature PP-motif (SGGXDS) involved in ATP binding (Bork and Koonin, 1994) as well as motifs CXXC and GHXXDD (which act to recognize RNA) present in the TtcA protein family (Jager et al., 2004). Zinc fingers are highlighted in blue boxes, ubiquitin-fold modified lysine residues are in red boxes, and conserved catalytic cysteine residues are indicated by a star. Secondary structural elements predicted for HVO_0580 based on Phyre2 3D homology modeling are highlighted with blue arrows (b-sheets) and green cylinders (a-helices) above the amino acid sequence. doi:10.1371/journal.pone.0099104.g001 immunoblotting (IB) cell lysate with antibodies specific for each tag prior to pull-down assay. By contrast to a-Flag, StrepII tag affinity purification is fully suited for inclusion of 2 M salt in buffers to maintain the integrity and activity of haloarchaeal proteins (as previously demonstrated for proteasomes) [21,22]. The molar levels of salt correlate well with the intracellular environment of haloarchaea where K+ is a prominent ion at 1.9 to 5.5 M [23].
Once purified, NcsA-pull down fractions were analyzed by reducing SDS-PAGE followed by total protein staining (Coomassie Blue, CB) and immunoblotting (IB) ( Figure 4A). Based on total protein staining ( Figure 4A, upper panel), no proteins were detected in the empty vector control (lane 1) revealing the NcsApull down assay was relatively specific for NcsA and NcsAassociated proteins. The NcsA-pull down fractions (lanes 2-4) were found to be primarily composed of a 36 kDa protein (of molecular mass consistent with the 37.4 kDa theoretical mass of NcsA-StrepII) with additional protein bands of varying abundance also detected. In particular, an ,20 kDa protein was found to be relatively abundant in fractions purified from the strain expressing Flag-SAMP2 in trans (lane 3). Further examination of these samples by anti-StrepII IB to probe for NcsA-StrepII ( Figure 4A, middle panel) revealed the majority of NcsA was in the 36 kDa form with high molecular mass bands of .50 kDa also detected. Use of anti-Flag IB to probe for the Flag-SAMPs ( Figure 4A, lower panel) revealed SAMP2 to be associated with NcsA in both free (,20 kDa) and high molecular mass (.50 kDa) forms. This association appeared specific for SAMP2, as SAMP1 was not detected in NcsA-pull down fractions from strains co-expressing Flag-SAMP1 with NcsA-StrepII ( Figure 4A, lower panel, lane 4). Based on these results, NcsA is covalently modified and co-purifies with a variety of protein partners including the Ub-fold SAMP2, which is linked to the NcsA protein network by covalent and noncovalent bonds.
We next performed experiments to determine whether SAMP2 encoded from the genome would influence the covalent modification of NcsA and whether this modification required the E1-like enzyme, UbaA. In brief, NcsA-StrepII was affinity purified from H26 (wild-type, parent), Dsamp2, and DubaA strains, and fractions were analyzed by anti-StrepII IB (Figure4B). NcsA isolated from wild-type cells was found as a mixture of isoforms including the predominant 36 kDa protein and minor species of ,60-125 kDa ( Figure 4B, lane 1). By contrast, NcsA was only detected in its 36 kDa form when purified from Dsamp2 and DubaA strains, providing evidence that SAMP2 and UbaA are required for the modified isoforms of NcsA detected by this assay. These results suggested that NcsA is isopeptide linked to SAMP2 by an UbaAdependent mechanism.
To further investigate the isopeptide linkage of SAMP2 to NcsA, Flag-SAMP2 and NcsA-StrepII were co-expressed in wildtype and DubaA mutant strains and analyzed by NcsA pull-down assay (with expression of all tagged proteins confirmed by IB of cell lysate prior to assay) ( Figure 4C). UbaA was found to be required for detection of the covalently modified (.60 kDa) forms of NcsA by anti-StrepII IB of pull down fractions ( Figure 4C, lanes 1-2). Likewise, the covalently linked and free forms of SAMP2 detected in the NcsA pull-down fractions were significantly reduced in the DubaA mutation ( Figure 4C, lane [3][4]. These results support the model that NcsA is isopeptide linked to SAMP2 by an UbaAdependent mechanism and that UbaA stimulates the non-covalent association of SAMP2 with NcsA. A Flag-SAMP2 specific protein of ,60 kDa was found associated with NcsA in pull-down fractions of the DubaA mutant (vs. empty vector control) that was not NcsA-StrepII. The molecular details of this finding are not known, as formation of ubiquitin-fold protein conjugates by a mechanism that is independent of UbaA has not been reported for Hfx. volcanii or other archaea.

HvJAMM1 cleaves NcsA-SAMP2 conjugates
To further investigate the SAMP2-modified forms of NcsA, NcsA-StrepII pull-down fractions from an DncsA mutant coexpressing Flag-SAMP2 were treated with the desampylating enzyme, HvJAMM1, and analyzed by IB ( Figure 5). HvJAMM1 is a Zn 2+ -dependent metalloprotease of the JAB1/MPN+/MOV34 superfamily that hydrolyzes isopeptide and linear linkages of SAMP1-3 to target proteins and is inhibited by the metal chelator EDTA [24,25]. Thus, samples incubated in the presence of EDTA (lanes 2 and 5) or in the absence of HvJAMM1 (lanes 1, 4 and 8) served as controls. By this approach, HvJAMM1 was found to hydrolyze the SAMP2-modified forms of NcsA but not the unmodified form of NcsA detected at 36 kDa by anti-StrepII IB (lane 3). Likewise, HvJAMM1 hydrolyzed the majority of SAMP2conjugates detected by anti-Flag IB in NcsA pull down fractions of the DncsA strain with exception of two Flag-SAMP2 specific bands of ,60 and 75 kDa (lane 6). The SAMP2-specific bands of comparable migration (,60 and 75 kDa) detected in NcsA pull down fractions of the DubaA mutant were also resistant to cleavage by HvJAMM1 ( Figure 5, lanes 7-8) suggesting a small subset of proteins are linked to SAMP2 by a mechanism that is independent of UbaA and not reversed by HvJAMM1. However, the majority of NcsA isoforms were collapsed to a single species of 36 kDa by the desampylase HvJAMM1, providing further evidence that NcsA is covalently bound to SAMP2 and suggesting NcsA modification may, in part, be regulated by HvJAMM1. NcsA Lys204 is isopeptide linked to SAMP2 With evidence that NcsA is covalently linked to SAMP2 through a mechanism that requires UbaA, we next sought to determine whether this linkage was an isopeptide bond and to identify the site(s) of modification. NcsA-StrepII was purified, subjected to trypsin digest and analyzed for diglycine footprints by LTQ-Orbitrap liquid chromatography (LC) tandem MS (MS/ MS) as previously described [13]. Observed spectra (1775 total) were of high coverage (78%) and mapped to residues spanning D9 to K311 of the deduced sequence of NcsA (321 amino acids total). NcsA Lys204 was identified as a site of missed cleavage by trypsin with an increase in molecular mass of +114 kDa, indicating the presence of a diglycine footprint derived from the C-terminal tail of SAMP2 ( Figure S5 in File S1). An almost full series of y-and btype ions were detected, and the isopeptide linkage was uniquely identified in the MS/MS spectral analysis of NcsA-StrepII purified samples (not in the vector alone control). Based on these results, NcsA Lys204 is isopeptide linked to SAMP2.

NcsA is covalently modified by poly-SAMP2 chains
Ub/Ubl proteins can form polymeric chains on target substrates, such as the Lys48-linked Ub chains that serve as signals for degradation of proteins by proteasomes in eukaryotic cells [26]. Lys58-linked SAMP2 chains have been identified in Hfx. volcanii [13]. However, it is not known whether these chains are anchored to protein targets or whether the second lysine residue (SAMP2 Lys64) can serve as a site for chain formation [13]. In an effort to determine whether NcsA is modified by SAMP2 chains, NcsA-StrepII was co-expressed with a Flag-SAMP2 variant devoid of lysine residues (K58R and K64R, named K.R) in wild-type, Dsamp2, and DubaA backgrounds. NcsA was subjected to pull down assays and probed with anti-Flag and anti-StrepII antibodies ( Figure 6A and B, respectively). With this approach, SAMP2 and its K.R variant were found functional in formation of robust levels of protein conjugates in wild type and Dsamp2 strains ( Figure 6A, lanes 1-2 and 4-5) vs. the DubaA mutant strain (lanes 3 and 6). However, only a single modified form of NcsA (,50 kDa) was detected when SAMP2 K.R was expressed in the Dsamp2 background ( Figure 6B, lane 5) compared to wild type cells (lane 4) or cells expressing SAMP2 (lane 1-2). While the NcsA that is linked to the single moiety of SAMP2 K.R is at low levels, the conjugate is detected and may be limited due to editing by a desampylase similarly to deubiquitylases of eukaryotic cells [27]. The physiological role of poly-SAMP2 chain formation on NcsA is unclear. However, our results provide evidence that poly-SAMP2 Figure 3. NcsA is required for growth of Hfx. volcanii at an elevated temperature (506C). Hfx. volcanii H26 (wt, parent), Dsamp2, DncsA, and trans complemented DncsA strains were grown in ATCC 974 medium. Freshly isolated colonies were inoculated into 3 ml medium (in 136100 mm culture tubes) and thrice subcultured at 42uC. Cells grown to logarithmic phase at 42uC from these subcultures were used as inoculum for monitoring growth at 50uC as presented in panel A. Cells grown to stationary-phase from these 50uC cultures were used as an inoculum for monitoring long-term growth at 50uC as presented in panel B. Inoculum was at 0.02 OD 600 with rotary shaking (200 rpm) in 20 ml medium in 250 ml baffled flasks for the growth assays presented in panels A and B. For panel C, cell cultures, as indicated above each plate, were diluted to 0.1 OD 600 and then spot-plated on solid agar ATCC 974 medium in serial dilutions as indicated. Plates were incubated at 50uC. Control experiments performed at 42uC are presented in Figure S4 in File S1. See Methods section for details. doi:10.1371/journal.pone.0099104.g003 chains are anchored on NcsA and that these chains do not form on NcsA when SAMP2 lysines are modified to arginine residues.
NcsA is associated with homologs of ubiquitinproteasome, RNA processing, and translation systems based on MS/MS By an MS/MS approach, NcsA protein partners were identified. In brief, NcsA-StrepII was purified from DncsA strains (with and without expression of Flag-SAMP1/2) and compared to an empty vector control. Protein samples were subjected to trypsin digest and analyzed by LC-MS/MS as described for mapping the site of SAMP2 modification [13]. By this approach, NcsA was found to co-purify with protein partners that could be identified at .95% probability and .25% coverage by MS/MS (Table 1). NcsA protein partners included UbaA, SAMP2 and the AAA+ ATPase proteasome-activating nucleotidase A/1 (PAN-A/1). In addition, NcsA was found associated with translation elongation factor aEF-1a (HVO_0359) and archaeal cleavage and polyadenylation specificity factor 1 group of the b-CASP ribonuclease family of proteins (aCPSF1; HVO_0874) [28]. We note that our previous MS/MS studies readily detect SAMP1-3 in protein samples [13,25]. However, in this study only SAMP2 was found to co-purify with NcsA (Table 1). Thus, NcsA appears to be specifically associated with SAMP2 and not SAMP1/3 and is linked to components of sampylation, proteasome function, translation and RNA processing.

NcsA is associated with UbaA and PAN-A/1 based on immunoblotting
To further analyze NcsA protein partners, an IB approach was used in which NcsA-pull down fractions were probed with polyclonal antibodies raised against the E1-like UbaA and   proteasomal ATPase PAN-A/1. By this approach, the UbaA protein band (of 36 kDa) was detected in NcsA pull-down fractions from wild-type but not DubaA mutant or empty vector control strains ( Figure S6A in File S1). Likewise, PAN-A/1 of ,50 kDa was readily detected in NcsA pull-down fractions of wild-type cells but not in fractions similarly isolated from wild type cells with the empty vector control ( Figure S6B in File S1). Thus, NcsA was found to associate with archaeal homologs of the eukaryotic ubiquitin-proteasome systems.

NcsA and aCPSF1 form a complex
To further investigate the physical association of NcsA and aCPSF1 detected by our MS-analysis, the Hfx. volcanii aCPSF1 homolog (HVO_0874) was N-terminally Flag-tagged and expressed with and without NcsA-StrepII in a DncsA mutant background. aCPSF and NcsA synthesis were confirmed by anti-Flag and -StrepII IB of cell lysate, respectively ( Figure 7A). Interestingly, when aCPSF1 was expressed alone, a ,100 kDa protein was readily detected in cell lysate by a-Flag immunoblot that migrated similarly (although larger) than the Flag-aCPSF1 of 73.2 kDa theoretical mass based on genome sequence ( Figure 7A, lane 3). By contrast, multiple aCPSF1 bands of 50-200 kDa were detected when NcsA was included in the expression strain ( Figure 7A, lane 4), suggesting that, when NcsA is present in the cell, the aCPSF1 undergoes covalent modifications that alter its molecular mass with the lower molecular masses most likely due to proteolytic cleavage.
To determine whether or not NcsA co-purifies with aCPSF1, NcsA-StrepII pull-down assays were performed with the Flag-aCPSF1 and NcsA-StrepII strains and appropriate controls. With this approach, the ,100 kDa isoform of aCPSF1 was found to specifically co-purify with NcsA ( Figure 7B, lane 3). Interestingly, while co-expression of NcsA and aCPSF1 appeared to promote the generation of multiple isoforms of aCPSF1 from 50-200 kDa ( Figure 7A, lane 4), only the 100 kDa isoform of aCPSF1 was found to co-purify with NcsA ( Figure 7B, lane 3). Likewise, the NcsA isoform detected after pull down assay appeared limited to the non-sampylated ,36 kDa form of NcsA in strains expressing Flag-aCPSF1 ( Figure 7B, lane 3). Further analysis of the NscA-pull down fractions by total protein (CB) stain revealed aCPSF1 as the major protein associated with NcsA when co-expressed in trans in the DncsA background ( Figure 7C). Thus, a complex of nonsampylated NcsA and 100 kDa aCPSF was found to dominate the NcsA pull down fractions when both proteins were expressed in trans. While NcsA was not associated in a stable complex with the other isoforms of aCPSF, detection of these isoforms in cell lysate was dependent upon NcsA expression.

Discussion
Here we demonstrate that tRNA thiolase homologs of the adenine nucleotide a hydrolase (ANH) superfamily are widespread and important for the thiolation of wobble uridine tRNA in haloarchaea. In particular, the ANH superfamily tRNA thiolase homolog NcsA of the halophilic archaeon Hfx. volcanii was found to be required for the thiolation of tRNA Lys UUU and growth at elevated temperature similarly to the tRNA thiolases of bacteria (T. thermophilus TtuA [29]) and fission yeast (Schizosaccharomyces pombe Ncs6 (Tuc1 or Ctu1) [30]). Based on recent work demonstrating the importance of wobble uridine thiolation in structuring the anticodon of human tRNA for efficient and accurate recognition of cognate and wobble codons [31], we speculate that the thermosensitive phenotype detected for the DncsA mutant was due to the lack of the 2-thiomodification of wobble uridine tRNA that would otherwise promote efficient translation and structural stability of tRNA. Overall, we suggest that NcsA and its close homologs are important for 2-thiolation of tRNAs with wobble uridine, including those specific for lysine (tRNA Lys UUU ), glutamate (tRNA Glu UUC ), and glutamine (tRNA Gln UUG ) in diverse haloarchaea. Based on this study, we propose a model in which NcsA is covalently modified at lysine residue(s) by isopeptide linked chains of the ubiquitin-fold SAMP2, through a mechanism that requires the ubiquitin-activating E1 enzyme homolog UbaA ( Figure 8A). While polymeric chains of SAMP2/3 were previously detected in the Hfx. volcanii proteome by MS/MS [13,25], it was unclear whether or not these chains were anchored to protein targets. The prokaryotic ubiquitin-like protein (Pup) of mycobacteria is not known to form polymeric chains [32]. Thus, NcsA is the first example of a target protein that is isopeptide-linked to polymeric chains of an ubiquitin-like protein modifier in prokaryotic cells. While the lysine residue (Lys204) of NscA that was found modified by poly-SAMP2 is not highly conserved among archaea and appears restricted to species of Haloferax and Natrialba, Lys204 is predicted to be in a unstructured region of NcsA (highlighted in Figure 1) that has undergone large amino acid insertions in some species (e.g., Haloarcula vallismortis GI:490650014). Interestingly, the tRNA thiolase of T. thermophilus, TtuA, was recently demonstrated to be isopeptide linked at multiple lysine residues (including Lys137, 226 and 229; highlighted in Figure 1) to the ubiquitin-like protein modifier TtuB [16]. While two of these modified lysine residues (Lys226 and Lys229) are unique to TtuA, the third (Lys137) is conserved with P. horikoshii Ph0300 and is located within a disordered 'hole-forming' region near the disulfide bond forming cysteine residues proposed to function in catalysis of tRNA thiolation [17]. Thus, ubiquitin-like modification of TtuA at Lys137 is thought to alter enzyme activity [17]. TtuA Lys137 is not conserved with NcsA or Tuc1/Ncs6 homologs. However, the modified Lys204 of the Hfx. volcanii NcsA is located in close proximity to conserved active site cysteine and PP-motif residues based on 3D-homology modeling ( Figure S2 in File S1). Thus, the ubiquitin-like modification of NcsA by SAMP2 may regulate its tRNA thiolase activity. Alternatively, poly-samp2ylation of NcsA may act as a signal for degradation by proteasomes and/or interaction with other proteins based on its location in a flexible loop and in analogy to the eukaryotic ubiquitin-proteasome system (UPS) ( Figure 8B). Clearly further work is needed to determine the biological role of the observed polysampylation of NcsA.
In this report, we also provide evidence that NcsA associates non-covalently with ubiquitin-proteasome (UbaA, SAMP2 and PAN-A/1), translation (aEF-1a) and RNA processing (aCPSF1) system homologs ( Figure 8B). Interactions of the E1-like UbaA and ubiquitin-fold protein SAMP2 with NcsA are consistent with the proposed mechanism of sulfur transfer to tRNA ( Figure 8A) and in analogy to the interactions detected for the eukaryotic E1 Uba4 with the ubiquitin-fold Urm1 and Ncs6 by MS and yeast-two hybrid analyses [33]. Our finding that the non-covalent association of NcsA with SAMP2 is stimulated by UbaA is consistent with the model that UbaA-mediated adenylation and thiocarboxylation of SAMP2 precedes the non-covalent binding of SAMP2 to NcsA ( Figure 8A). The association of NcsA with aEF-1a is also in line with the tRNA thiolation pathway based on the need for this translation elongation factor to deliver aminoacyl-tRNAs to the ribosome, where the tRNAs are presumably 2-thiolated prior to association with aEF-1a to enhance translation fidelity ( Figure 8B). The biological role(s) of NcsA association with the proteasomeassociated AAA+ATPase PAN-A/1 and RNA processing homolog aCPSF1 are less clear than the other protein partners identified in this study. The close association of the Ub-fold SAMP2 and E1like UbaA with sulfur mobilization and protein modification may account for the binding of NcsA to PAN-A/1. Whether the PAN-A/1 associates with NcsA to catalyze protein remodeling and/or proteasome-mediated proteolysis is not known. Interestingly, when aCPSF1 is expressed in cells with NcsA, the aCPSF1 protein is found in high and low molecular mass forms that are not detected in DncsA mutant cells suggesting NcsA promotes the posttranslational modification of aCPSF. Whether or not the archaeal ubiquitin-proteasome system homologs (UbaA, SAMP2 and PAN-A/1) associate with NcsA to promote the covalent modification and/or proteolysis of aCPSF1 remains to be determined. We note that DncsA strains expressing aCPSF1 in trans display poor growth and are not stable (data not shown) suggesting the posttranslational modification of aCPSF1 is important to cell function when expressed in trans. Interestingly, only the 100 kDa species of aCPSF1 was found to co-purify with NcsA suggesting this form of aCPSF1 has high affinity for NcsA compared to the other aCPSF1-specific bands migrating between 50-200 kDa.
Our finding that NcsA interacts with aCPSF1 is the first identification of a protein partner for a putative RNA cleavage enzyme of the b-CASP family in archaea. Members of the b-CASP family are found in all three evolutionary lineages and include the eukaryotic cleavage and polyadenylation specificity factor (CPSF) shown to cleave the 39 ends of newly synthesized pre-messenger RNA (pre-mRNA) during transcription [34]. The archaeal b-CASP protein of Methanothermobacter thermautotrophicus is  Whether additional factors are needed to provide specificity to the sampylation system is unclear, as E2 and E3 homologs are not predicted based on genome sequence. Panel B, NcsA is found isopeptide linked to SAMP2 and non-covalently associated with various proteins, as noted by dotted red lines. NcsA partners include the E1-like UbaA and Ub-fold SAMP2 of the tRNA thiolation and sampylation pathways. NcsA is also found associated with EF-1a that binds aminoacylated-tRNAs and mediates translation elongation, PAN-A/1 (an AAA+ATPase associated with energy-dependent proteolysis by proteasomes (20S core particles or CPs) and protein remodeling, and the b-CASP ribonuclease homolog of the aCPSF1 family thought to cleave mRNA and/or tRNA. RNAP, RNA polymerase. doi:10.1371/journal.pone.0099104.g008 characterized as a metalloenzyme which binds RNA at U-rich regions [35] and is hypothesized to act as a nuclease in the turnover of RNA including mRNAs encoding proteins targeted to the proteasome for degradation [35,36]. More recently, a Pyrococcus abyssi b-CASP protein (Pab-aCPSF1) is demonstrated to have an RNA endoribonucleolytic activity that preferentially cleaves at single-stranded CA dinucleotides and to have a 59-39 exoribonucleolytic activity that acts on 59 monophosphate substrates [37]. In Hfx. volcanii, mRNA polyadenylation is not observed under standard growth conditions [38,39] making it difficult to predict the biological role and biochemical function of the b-CASP protein aCPSF1. Hfx. volcanii does synthesize an RNaseZ, which enzymatically cleaves tRNA precursors from the 39-end, suggesting aCPSF1 is not needed for this type of activity [40]. Finding the close association of aCPSF1 with NcsA leads us to speculate that this RNase regulates NcsA-mediated thiol modification of wobble uridine tRNA by cleaving the tRNA substrate or the mRNAs actively translated by ribosomes associated with these tRNAs and, thus, provide a mechanism to modulate cell growth. Overall, our results provide the first characterization and evidence of an archaeal tRNA thiolase enzyme (NcsA), which is associated with an aCPSF1 homolog, modified by the sampylation system and important for mediating 2-thiouridine formation in halophilic archaea.

Strains, media and plasmids
Strains, plasmids and primers used for cloning are summarized in Tables S1-S2 in File S1. Liquid cultures were aerated at 200 rpm. Escherichia coli Top10 was used for routine recombinant DNA experiments. Hfx. volcanii strains were transformed [41] with plasmid DNA isolated from E. coli GM2163. E. coli strains were grown at 37uC in Luria-Bertani medium, and Hfx. volcanii strains were grown at 42uC or 50uC in ATCC 974 complex medium. Media were supplemented with novobiocin (0.2 mg?ml 21 ), uracil (50 mg?ml 21 ) and ampicillin (100 mg?ml 21 ) as needed. For growth curves, Hfx. volcanii strains were grown in biological triplicate in ATCC 974 complex medium to exponential phase or stationary phase. Cells were grown from isolated colonies from plates into 3ml cultures (136100 mm tubes at 42uC). After three subcultures at exponential phase, a main culture of 20 ml ATCC 974 medium in 250 mL Erlenmeyer baffled flasks was inoculated with cells to an OD 600 of 0.02 and incubated at 42uC or 50uC, as indicated. For spot-dilutions, cell cultures were diluted to 0.1 OD 600 and spotplated on ATCC 974 agar-medium in a ten-fold dilution series. Optical density at 600 nm was measured to monitor growth.

DNA purification and electrophoresis
Hfx. volcanii genomic DNA was prepared as previously described [42]. Plasmid DNA was isolated by use of the QIAprep spin miniprep kit (Qiagen, Valencia, CA). Polymerase chain reaction (PCR) products were purified by MinElute (Qiagen) prior to modification by restriction enzymes or T4 DNA polynucleotide kinase. For rapid PCR screening, template DNA was extracted from Hfx. volcanii and E. coli colonies as described previously [43]. DNA was separated by electrophoresis using 0.8% or 2% (wt/vol) agarose gels in 16TAE electrophoresis buffer (40 mM Tris acetate, 2 mM EDTA, pH 8.5), stained with ethidium bromide at 0.5 mgNml 21 and photographed with a Mini visionary imaging system (Fotodyne, Hartland, WI). Sizes of the DNA fragments were estimated using Hi-Lo DNA molecular weight markers (Minnesota Molecular, Minneapolis, MN).

DNA sequencing
Specificity of all PCR products including DNA cloned into plasmids listed in Table S1 in File S1, was confirmed by Sanger automated DNA sequencing using an Applied Biosystems model 3130 genetic analyzer (ICBR Genomics Division, University of Florida).

Generation of mutant strain
The ncsA gene corresponding to HVO_0580 (GI: 292654746) was deleted from the Hfx. volcanii genome by use of a markerless pyrE2-based pop-in/pop-out strategy [42]. In brief, the hvo_0580 deletion plasmid pJAM1910 was created by the insertion of two separate DNA fragments, corresponding to 500 bp flanking regions of ncsA (generated by PCR with primer pairs P1/P2 and P3/P4, Figure S1 in File S1), into plasmid vector pTA131. The chromosomal deletion was confirmed by PCR using P7/P8 primer pairs, DNA sequencing of the PCR product, and Southern blotting.

Southern blotting
Genomic DNA, isolated from the Hfx. volcanii parent H26 and DncsA mutant MH105 strains, was subjected to restriction enzyme digestion with PciI and analyzed by Southern blotting as previously described [43]. In brief, a 29-deoxyuridine-59-triphosphate coupled by an 11-atom spacer to digoxigenin (DIG-11-dUTP) was used to label a dsDNA probe generated by PCR with primer pairs P1/P2. Hybridization between the immobilized dsDNA probe and its target was detected by chemiluminescence using an alkaline phosphatase (AP)-linked anti-digoxigenin antibody and chloro-5-substituted adamantyl-1,2-dioxetane phosphate (CSPD) as recommended by supplier (Roche Applied Science, Indianapolis, IN).

RNA isolation
For tRNA thiolation assays, strains were grown in ATCC 974 medium (37.5 mL of 100 mL culture in a 500-mL flask; 42uC at 200 rpm) and total RNA was extracted from log-phase grown wild-type and Dhvo_0580 (DncsA) mutant strains as previously described [45] with modification [12]. Isolated RNA was precipitated in 0.25 M sodium acetate (pH 5.0) with two volumes of 95% ethanol (270uC, 15 min) and washed with 70% ethanol.
The air-dried RNA pellet was resuspended in 30 ml DEPC-treated water with a typical yield of 100 to 150 mg RNA. RNA integrity was assessed by agarose gel electrophoresis, and RNA concentration was determined by A 260 nm.
Assay for tRNA thiolation RNA was separated by 12% urea polyacrylamide gels supplemented with 30 mg APM per ml, and tRNA Lys UUU was detect by Northern blotting as previously described [12]. The probe for Northern blotting (tRNA-Lys-UUU; Table S2 in File S1) was 59 end-labeled using T4 polynucleotide kinase and [c-32 P]ATP [12].

Protein purification
For large-scale protein purification, Hfx. volcanii strains (detailed in Table S1 in File S1) were grown to stationary phase in ATCC 974 medium with Nv (0.2 mg?ml 21 ) (1 L in 2.8-L Fernbach flask). Cells were harvested, washed once with ice-chilled Tris-salt buffer (20 mM Tris, 2 M NaCl, pH 8.0) and stored at 280uC until used. Cell pellets were resuspended in Tris-salt buffer (1.5 ml per 1 g wet wild-type cells) and lysed using a French Press (36, 2000 psi). Soluble extracts were obtained by centrifugation (20 min at 50006g and 4uC) followed by filtration (0.8 mm). Samples were applied to a 1 ml column volume Strep-Tactin column (Qiagen) pre-equilibrated in Tris-salt buffer. The column was washed with 30 column vol Tris-salt buffer, and proteins were eluted with Trissalt buffer supplemented with 2.5 mM D-desthiobiotin. Purified protein fractions were pooled and buffer-exchanged by dialysis with a 3.5 kDa cutoff SnakeSkin dialysis tubing (ThermoScientific, Rockford IL) into Tris-salt buffer overnight at 4uC. Strep-Tactinpurified proteins were applied to an Amicon Ultra 4 ml Filter with a 3.5 kDa cutoff (Millipore, Billerica, MA) and concentrated by centrifugation at 3000 RCF for 40 min at 4uC in a Allegra X-22R swinging bucket rotor (Beckman Coulter, Indianapolis IN). Purified protein fractions were stored at 4uC. For small-scale purification, 50 ml cultures were grown to stationary phase in 250 ml Erlenmeyer baffled flasks, harvested, lysed, and purified by chromatography as described earlier using 50 ml Strep-Tactin resin.

Protein quantification
Protein concentration for purified proteins was determined by bicinchoninic acid (BCA) assay according to Manufacturer's protocols (Thermo Scientific, Rockville, IL). Bovine serum albumin (Bio-Rad Life Science, Hercules, CA) served as a standard and absorbance was read at absorbance 562 nm using a Synergy HT microplate reader (BioTek, Winooski, VT).

Protein electrophoresis
Protein samples were boiled (15 min) in freshly prepared SDSreducing buffer (1x concentration: 2% wt/vol SDS, 4% vol/vol glycerol, 40 mM Tris-HCl pH 6.8, 0.01% wt/vol bromophenol blue and 2.5% vol/vol b-mercaptoethanol or 10 mM dithiothreitol). Proteins were separated by 10% SDS-PAGE using a Mini-Gel apparatus with Tris-glycine-SDS buffer system according to supplier (Bio-Rad Life Science). Dilute protein samples were precipitated on ice using 10% trichloroacetic acid (TCA), washed twice with ice-chilled acetone to remove excess salt, and air dried prior to boiling in SDS-reducing buffer.

Desampylation assay
Desampylation of NcsA-associated SAMP2 conjugates was assayed by incubation with HvJAMM1 protease [24]. Assay reactions included 5 mM HvJAMM1, 1 mg SAMP conjugates, and 500 mM ZnCl 2 in HEPES-salt buffer (20 mM HEPES, 2 M NaCl, pH 7.5). Negative controls were performed either in the presence of 50 mM EDTA, or in the absence of HvJAMM1. All reactions were incubated (4 h, 50uC). Free and conjugated forms of Flag-and StrepII-tagged proteins were detected by immunoblotting.

Mass spectrometry
Proteins purified by Strep-Tactin chromatography were analyzed from in-gel and in-solution samples using a mass spectrometry (MS)-based approach. For in-gel analysis, proteins were separated by 10% SDS-PAGE, visualized by staining with Bio-Safe Coomassie (Bio-Rad), destained in double deionized water, and excised in gel slices. Proteins in gel were treated with 45 mM dithiothreitol (DTT) and 100 mM iodoacetamide (IAA). To minimize IAA carryover prior to trypsin digest, liquid was removed from the treated gel pieces and samples were dehydrated by treatment with acetonitrile, aspiration and brief air drying. For identification of sampylation sites, protein samples were treated in solution with 10 mM DTT at 95uC for 5 min. The temperature was lowered to 55uC for 30 min. Samples were cooled to room temperature, 40 mM IAA was added, and the samples were incubated in the dark at room temperature for 45 min. Alkylation was quenched by addition of 40 mM DTT at room temperature for 45 min to prevent lysine akylation [47]. Samples were treated with trypsin (1 mg to 50 mg protein) at 37uC for 15 h.
Tryptic peptides were injected onto a capillary trap (LC Packings PepMap) and desalted for 5 min with 0.1% vol/vol formic acid at a flow rate of 3 ml?min 21 prior to loading onto an LC Packing C18 Pep Map nanoflow high performance liquid chromatography (HPLC) column. The elution gradient of the HPLC column started at 3% solvent A (0.1% vol/vol formic acid, 3% vol/vol acetonitrile, and 96.9% v/v H 2 O), 97% solvent B (0.1% vol/vol formic acid, 96.9% vol/vol acetonitrile, and 3% vol/vol H 2 O) and finished at 60% solvent A, 40% solvent B using a flow rate of 300 ml?min 21 for 30 min. LC-MS/MS analysis of the eluting fractions was carried out on an LTQ Orbitrap XL mass spectrometer (ThermoFisher Scientific, West Palm Beach, FL). Full MS scans were acquired with a resolution of 60,000 in the Orbitrap from m/z 300-2000. The ten most intense ions were fragmented by collision induced dissociation (CID).
Raw data were analyzed using Mascot (Matrix Science, London, UK; version 2.2.2) against a Hfx. volcanii database and a target decoy database including the proteome set of Hfx. volcanii and a set of reversed sequences generated by Mascot. Mascot was searched with a fragment ion mass tolerance of 0.8 Da and a parent ion tolerance of 15 ppm. Iodoacetamide derivative of Cys was indicated as a fixed modification while deamidation of Asn and Gln, oxidation of Met, and isopeptide linkage to Gly-Glywere specified as variable modifications. Scaffold (Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications, where protein probabilities were assigned by the Protein Prophet algorithm and peptide probabilities were assigned by the Peptide Prophet algorithm [48,49].

Supporting Information
File S1 Supporting Information. File S1 includes: Table  S1. Strains, plasmids used in this study; Table S2. Oligonucleotide primers used in this study; Figure S1. Dendrogram analysis of Haloferax volcanii NcsA and homologs of the a hydrolase (ANH) superfamily from archaea, eukaryotes and bacteria; Figure S2. 3D-structural model of Haloferax volcanii NcsA; Figure S3. Organization of ncsA and its targeted deletion on the genome of Haloferax volcanii; Figure S4. Growth of Haloferax volcanii DncsA mutant compared to parent strain H26 at optimum growth temperature; Figure S5. Lys204 residue of NcsA is found isopeptide linked to SAMP2; Figure S6. Detection of E1-like UbaA and PAN-A/1 ATPase in NcsA-StrepII pull-down fractions. (PDF)