Activation of a Chimeric Rpb5/RpoH Subunit Using Library Selection

Rpb5 is a general subunit of all eukaryotic RNA polymerases which consists of a N-terminal and a C-terminal domain. The corresponding archaeal subunit RpoH contains only the conserved C-terminal domain without any N-terminal extensions. A chimeric construct, termed rp5H, which encodes the N-terminal yeast domain and the C-terminal domain from Pyrococcus furiosus is unable to complement the lethal phenotype of a yeast rpb5 deletion strain (Δrpb5). By applying a random mutagenesis approach we found that the amino acid exchange E197K in the C-terminal domain of the chimeric Rp5H, either alone or with additional exchanges in the N-terminal domain, leads to heterospecific complementation of the growth deficiency of Δrpb5. Moreover, using a recently described genetic system for Pyrococcus we could demonstrate that the corresponding exchange E62K in the archaeal RpoH subunit alone without the eukaryotic N-terminal extension was stable, and growth experiments indicated no obvious impairment in vivo. In vitro transcription experiments with purified RNA polymerases showed an identical activity of the wild type and the mutant Pyrococcus RNA polymerase. A multiple alignment of RpoH sequences demonstrated that E62 is present in only a few archaeal species, whereas the great majority of sequences within archaea and eukarya contain a positively charged amino acid at this position. The crystal structures of the Sulfolobus and yeast RNA polymerases show that the positively charged arginine residues in subunits RpoH and Rpb5 most likely form salt bridges with negatively charged residues from subunit RpoK and Rpb1, respectively. A similar salt bridge might stabilize the interaction of Rp5H-E197K with a neighboring subunit of yeast RNA polymerase and thus lead to complementation of Δrpb5.


Introduction
Archaea contain only one multi-subunit RNA polymerase (RNAP), whereas eukarya contain at least three RNAPs with different functions. Nevertheless, the overall structure of the archaeal and the eukaryal RNAPs is very similar and consists of about 11 or 12 subunits [1][2][3]. Based on sequence and structural comparison, the subunits of the RNAPs of all domains of life can be divided into three classes [2,4]. Class I subunits are conserved in all three domains whereas class III subunits are unique to each domain. Class II subunits are only present in eukaryal and archaeal RNAPs, but not in the bacterial domain. The following five eukaryal subunits with the corresponding archaeal subunits in brackets are the representatives of this class: Rpb5 (RpoH), Rpb6 (RpoK), Rpb8 (RpoG), Rpb10 (RpoN) and Rpb12 (RpoP) [2,4]. It is interesting to note that subunit RpoG was identified only recently and seems to be present only in crenarchaeota but not in euryarchaeota [3,[5][6][7].
The individual structures of these class II subunits as well as their interactions with other subunits are highly conserved between archaea and eukarya [6,8]. However, the eukaryal Rpb5 and the archaeal RpoH show a major difference in size [9,10]: The N-terminal domain of Rpb5 is missing in RpoH whereas the C-terminal domain is conserved between both subunits [9]. The N-terminal domain marks the far end of the DNA channel of the eukaryal RNAP subunit and probably accounts for the Rpb5/DNA contacts found ahead of the transcription fork in RNAP II [1,11,12]. This exposed position enables interaction with different transcription factors, for instance Rpa30 of the TFIIF complex or protein X of the hepatitis B virus [13][14][15]. The C-terminal domain of Rpb5 is in close contact with Rpb1 as indicated by the structure of RNAP II [1]. Yeast two hybrid experiments also demonstrated an interaction with Rpa190 and Rpc160, the corresponding RNAP I and III subunits [16,17]. Although the corresponding large RNAP subunit is split in archaea into two genes encoding subunits A9 and A0 [18], a strong interaction of A0 with RpoH has been demonstrated by Far-Western blot experiments with Pyrococcus (P.) furiosus components and with structural data from the Sulfolobus (S.) solfataricus RNAP [2,8].
In archaea it is possible to reconstitute an active RNAP from individual subunits [19,20], allowing for the functional analysis of individual subunits. Using an in vitro reconstituted archaeal RNAP lacking RpoP has revealed that this subunit is important for open complex formation and that the corresponding yeast Rpb12 subunit was able to rescue the in vitro activity of the DP enzyme [21]. Vice versa, yeast genetic experiments demonstrated that Pyrococcus rpoP under the control of a strong yeast promoter could complement the lethal phenotype of a rpb12 deletion mutant [21]. Similar experiments using a reconstituted enzyme lacking subunit H showed that this subunit is also important for open complex formation as well as initial transcription, and yeast Rpb5 was able to rescue the in vitro activity of a reconstituted DH enzyme [22]. In contrast to the in vitro results with Rpb5 and the successful complementation experiments with rpoP, a chimeric rp5H construct encoding the N-terminal yeast domain and the Cterminal domain from P. furiosus is unable to complement the lethal phenotype of a yeast rpb5 deletion strain [23 and unpublished data]. This negative complementation result in yeast with the archaeal RpoH domain is also in line with Rpb5 complementation experiments with the human counterpart. Rpb5 is the only common subunit in yeast which could not be substituted with the human homologue [24,25]. Only the C-terminal part of human Rpb5 could complement the corresponding sequence in yeast. To identify those amino acids which seem to be responsible for the negative complementation result, a more detailed mutational analysis was performed. This experiment revealed that only a small central segment (positions 121-146 in Saccharomyces cerevisiae) within the N-terminal domain could not be substituted with the corresponding human sequence [26]. These results prompted us to take a similar approach with the chimeric Rp5H construct. Using a combination of random mutagenesis and library selection in the Drpb5 strain we looked for mutants which activate the chimeric rp5H construct. We found that a single amino acid exchange within the archaeal domain of Rp5H is sufficient for complementation. Furthermore, in vivo and in vitro experiments with P. furiosus showed that this single exchange within RpoH does not influence the activity of the archaeal RNAP.

Strains, plasmids and general DNA manipulations
Strains and plasmids used in this study are listed in Table 1. Restriction enzymes and DNA polymerases for PCR reactions were purchased from NEB (Ipswich, USA). Genomic DNA was isolated using the ReliaPrep TM Kit from Promega (Mannheim, Germany).

Cloning of the chimeric construct rp5H
The chimeric construct rp5H was generated by overlapextension PCR [27] linking a DNA fragment encoding the Nterminal domain of Rpb5 (GenBank accession number NM_001178502; bp 1 -444) with rpoH of P. furiosus (PF1565). The oligonucleotides 59rpb5_EcoRI, 39rpb5_rpoH, 59rpb5_rpoH and 39rpoH_Not (sequences are shown as Table S1) were used for PCR amplification, digested with EcoRI and NotI, and ligated into the predigested E. coli-yeast-shuttle-vector pRS423 [28]. The resulting construct was termed pRS423_rp5H.
The resulting gene repertoire was ligated into pRS423 and used to transform electrocompetent E. coli XL1 Blue MRF' cells (Life Technologies GmbH, former Stratagene, Darmstadt, Germany). Colonies grown on SOB rich medium agar containing 20 mM glucose and 150 mg/ml ampicillin were swept off the plates, followed by the isolation of the plasmid library containing the rp5H mutants. As estimated from the number of grown colonies and the ligation efficiency tested by colony PCR, the library contained approximately 4.5610 5 independent mutants. Sequencing of 10 clones revealed an average number of 11+/27 nucleotide exchanges per gene, with a minimum number of two and a maximum number of 22 base substitutions. A bias for AT (71%) over GC (29%) exchanges and a ratio of transitions over transversion of 0.63 were observed.

Transformation and in vivo complementation of Saccharomyces cerevisiae YFN2
For complementation assays the yeast strain YFN2 was used [31]. Plasmid DNA from the gene library was used to transform YFN2 cells carrying a chromosomal deletion of rpb5 rescued by an rpb5/URA3 plasmid (pPL44-rpb5). The transformed Drpb5 cells were streaked out on synthetic dropout agar plates (without histidine and uracil) and incubated for three days at 30uC. In order to identify rp5H variants that could complement for rpb5 in vivo, the 4.6610 4 grown cells were replica plated on synthetic dropout medium (without histidine) containing 5-fluoroorotic acid (FOA; 1 gL 21 ; Toronto Research Chemicals, North York, Canada) and incubated up to ten days at 30uC [32]. Plasmid DNA was isolated from 104 single colonies and used to retransform Drpb5 cells, which were again plated on minimal media containing FOA. Out of the 53 positive colonies that grew after retransformation (true positive clones), the inserts of 12 individual clones were sequenced to identify amino acid exchanges allowing for complementation. All but one of the isolated pRS423_rp5H plasmids contained an rp5H insert coding for the E197K exchange, either alone or together with various combinations of other substitutions. The fastest growing colony contained an rp5H insert coding for the V23I+S68Y+M75K+E197K exchanges. The latter exchanges were cloned in different combinations in the background of the E197K exchange using megaprimer-PCR [33].

Constructs for the P. furiosus mutants
To perform selection for the Pyrococcus transformants with the antibiotic simvastatin, a resistance cassette was assembled using overlap extension PCR reactions [27]. It consists of the promoter region (2250 to 21) of the glutamate dehydrogenase gene (gdh, TK1432; primers SimR-Prom1-F and TK0914TK1431Pr-R), the coding region of the hydroxymethylglutaryl CoA reductase (TK0914; primers: TK0914-F and TK0914-R), and a sequence for termination of transcription downstream of the gdh gene (TK1432; primers: TK0914TK1431T-F and SimR-Term2-R) from Thermococcus kodakarensis. The fused DNA fragments were inserted into a SmaI hydrolyzed pUC19 vector by blunt end ligation.
Plasmid pMUR27 was created to introduce a hexahistidine/ streptavidin tag at the N-terminal region of the RNAP subunit RpoH for purification. For the construct, the following four fragments were combined: An upstream region of PF1565 (primers Hup-ERV-F and Fus-Hup-tkR-R), the rpoH gene and the downstream region of PF1565 (primers Fus-Strep-H-F and Hdwn-ERV-R), the resistance cassette mentioned above for the selection with simvastatin (primers SimR-Prom1-F and Fus-SimRpr-R) and the transcriptional promoter of rpoH (primers H-Prom-F and Fus-His-HPr-R). The purification tags were inserted by additional amplification of the rpoH fragment with an expanded forward primer (Fus-His-Strep-F). The fusion product was hydrolyzed with EcoRV and ligated into the SmaI site of pUC19 to establish pMUR27. This plasmid was employed to construct pMUR28 by introducing the single exchange E62K using the phosphorylated primers pf1565-Mut-E62K-F and pf1565-Mut-E62K-R.
The plasmids pMUR54 and pMUR43 are similar to pMUR27 and pMUR28, but also contain the DNA fragment encoding the N-terminal Rpb5 domain from yeast. This DNA fragment was amplified using the primers Rpb5-N-Domä ne F and Rpb5-N-Domä ne R and plasmid pRS423_rp5H(E197K) as template DNA. All DNA constructs were verified by sequencing.

Transformation of P. furiosus
Standard heat shock transformation of P. furiosus with linearized plasmid DNA was performed as described previously [34]. Transformants were enriched with 10 mM simvastatin in SMEstarch liquid medium at 85uC for 48 h and pure cultures were isolated after plating the cells on solidified medium in the presence of 10 mM simvastatin. The integration of the DNA fragments into the genome by double cross-over was verified by analyzing corresponding PCR products.
Growth conditions for P. furiosus strains P. furiosus was cultivated under anaerobic conditions at 85uC-95uC in SME medium with 0.1% starch, yeast extract and peptone, as described previously [34,35]. For solidification, gelrite was added to a final concentration of 1%. The antibiotic simvastatin (Toronto Research Inc., North York, Canada) was dissolved in ethanol and sterilized by filtration. For a more detailed analysis of the growth behaviour of Pyrococcus strains with modified rpoH genes, bottle flasks with 20 ml K SME-starch medium (supplemented with 10 mM simvastatin) were used and incubated at 95uC. Cell numbers were analyzed with a Thoma counting chamber (0.02-mm depth; Marienfeld, Lauda-Königshofen, Ger-many) using phase-contrast microscopy. All experiments were done in triplicate and the statistical mean value was used for plotting.

Purification of RpoH-Strep/His 6 tagged RNAP and in vitro transcription
Purification of RNAP containing a His 6 tag at the N-terminus of subunit H was performed as described previously for RNAP tagged at subunit D [34]. Cell extracts containing 20 mM imidazole were applied to NiNTA chromatography (1 ml His-Trap HP, GE Healthcare Europe GmbH, Freiburg, Germany), and bound protein was eluted with a 300 mM imidazole elution buffer. Further purification and desalting was obtained by gel filtration chromatography using a Superdex 200 column, equilibrated with 40 mM HEPES, pH 7.3, 250 mM KCl, 2.5 mM MgCl 2 , 0.5 mM EDTA, and 20% glycerol. Fractions were analysed for RNAP activity using in vitro transcription and SDS-PAGE analysis. In vitro transcription reactions were performed as described previously [36].

Results
The chimeric construct Rp5H is unable to complement a Drpb5 strain Fig. 1 illustrates the modular organization of Rpb5 consisting of a N-and a C-terminal domain. The N-terminal domain (orange part) is found only in eukarya, whereas the highly conserved Cterminal domain (red part) is also found in archaea (Fig. 1A). A superimposition using the structure of Saccharomyces cerevisiae and a model of P. furiosus (green part) shows a high similarity of the Cterminal domain between the yeast structure and the Pyrococcus model (Fig. 1A). Only the first few amino acids of the N-terminal region of RpoH do not fit into the superimposition (grey part). An alignment between both sequences confirmed the high degree of similarity of RpoH with the C-terminal domain of Rpb5 (Fig. 1B). These findings prompted us to reanalyze in more detail if this sequence and structural conservation enables a functional replacement of the C-terminal domain of Rpb5 with the corresponding archaeal domain. We used the yeast strain YFN2 with a chromosomal deletion of rpb5 [31]. This strain carries the rpb5 gene on plasmid pPL44-RPB5 encoding the URA3 marker. For complementation of the yeast strain in the presence of FOA we designed the shuttle vectors pRS423_rpb5 and pRS423_rp5H ( Fig. 2A). The positive control pRS423_rpb5 contains Rpb5 with an N-terminal Flag tag and the strong constitutive RPS28B promoter [37]. In construct pRS423_rp5H the conserved Cterminal domain of Rpb5 was substituted with the corresponding region of RpoH without the first amino acids of the N-terminal region (Fig. 1A, grey part). The results of the complementation experiments indicated that the positive control could complement the Drpb5 strain, but not the chimeric construct Rp5H with the archaeal RpoH domain replacing the yeast C-terminal domain ( Fig. 2A). These results confirmed previous experiments [23 and unpublished data].
A single amino acid exchange within RpoH of the chimeric construct Rp5H enables complementation To identify amino acid exchanges in the chimeric Rp5H construct that lead to complementation, we undertook a random mutagenesis approach using error-prone PCR in the presence of MnCl 2 . As we used specific PCR primers corresponding to the first eleven and the last nine amino acids of Rp5H these regions were excluded from mutagenesis (Fig. 2B). Amplified gene products were ligated into the shuttle plasmid pRS423 and propagated in E. coli. The analysis of the gene library revealed about 4.5610 5 individual clones containing an average of 11 nucleotide exchanges.
The plasmid library was used to transform the yeast strain YFN2 with the rpb5 deletion (Drpb5), and functional Rp5H variants were identified by the ability of their host cells to grow on synthetic dropout medium (without histidine) plates containing FOA. Positive clones were confirmed by retransformation and reassessed by in vivo complementation of Drpb5. Sequencing of 12 clones showed that all contained the exchange E197K in the Cterminal domain (stemming from RpoH) either alone or in combination with various exchanges in the N-terminal domain (stemming from Rpb5). The fastest growing colony contained an rp5H insert encoding the E197K exchange plus the V23I, S68Y, and M75K substitutions. In order to elucidate the effect of these N-terminal exchanges for complementation, they were cloned in isolation and in different combinations, always in the background of the E197K exchange (Fig. 3A). Following the transformation of Drpb5 with the different pRS423_rpb5 constructs, the growth of single colonies was compared in liquid minimal medium (synthetic dropout medium without histidine). Comparison of the growth behavior of the yeast strain with wild type Rpb5 and the strain with the single exchange E197K revealed that the maximum growth was reduced from OD 600 = 1.67 for the wild type to OD 600 = 1.27 for the mutant; the time needed to reach 50% of the maximum OD 600 value was increased from 23 to 44 hours for the mutant (Fig. 3A, compare lanes 1 and 9). The obvious growth disadvantage of the mutant containing the E197K exchange in comparison to wild type was reduced in the presence of additional substitutions in the eukaryotic-specific N-terminal domain (Fig. 3A). These exchanges (V23I, S68Y, M75K) stimulated the growth of the mutant containing E197K in an additive manner with M75K having the strongest influence. Growth experiments on solid medium at 24uC and 37uC confirmed the reduced growth rate of the Rp5H mutants compared to wild-type Rpb5 especially at higher temperatures ( Fig. 3B and data not shown). However, taken together, using a random mutagenesis approach it was possible to identify a single amino acid exchange within the archaeal domain of the chimeric construct Rp5H consisting of the N-terminal Rpb5 domain from yeast and the C-terminal archaeal RpoH domain which enables heterospecific complementation of the Drpb5 yeast strain.

The single E62K exchange within RpoH has no influence on the growth behavior of P. furiosus
To investigate if the identified exchange of a negatively for a positively charged residue influences the growth behavior of P. furiosus in the background of Rp5H and RpoH we used a recently developed genetic system for P. furiosus [34,35]. The plasmids pMUR27, pMUR28, pMUR43 and pMUR54 were used to establish the corresponding mutants within the Pyrococcus genome (Fig. 4). The plasmid pMUR28 is identical to pMUR27 except for the exchange E62K. This exchange is equivalent to the E197K exchange in the chimeric construct (Fig. 2). In the chimeric constructs pMUR54 and pMUR43 the eukaryotic N-terminal domain was fused with wild type RpoH or with RpoH-E62K (Fig. 4). For integration into the genome, the plasmids contain additionally 1000 bp upstream and downstream sequences of rpoH to enable double cross-over into homologous regions. For the Figure 1. Superimposition of the structure and sequence alignment of Rpb5 from S. cerevisiae and RpoH from P. furiosus. A: The N-terminal part of Rpb5 is shown in orange and the C-terminal part in red [1]. The RpoH model in green was obtained using the program Modeller in combination with the solved RpoH structures of Methanocaldococcus jannaschii, Methanobacterium thermoautotrophicus and S. solfataricus [2,9,44]. For superpositioning the program DaliLite (www.ebi.ac.uk/Tools/structure/dalilite) was used. The amino acids at the N-terminus of RpoH which do not fit with the Rpb5 structure are shown in grey. The N-termini and the C-termini of Rpb5 and RpoH are labeled. B: The Rpb5 sequence of yeast and the RpoH sequence of P. furiosus were analyzed using a webserver of the program ClustalW2 (www.ebi.ac.uk/Tools/msa/clustalw2). Conserved amino acids are shown in red colour. doi:10.1371/journal.pone.0087485.g001 selection of transformants a copy of the hmg-CoA reductase from Thermococcus kodakarensis (TK0914) was located upstream of the rpoH gene (Fig. 4). This gene provides resistance to the antibiotic simvastatin [34,38]. For purification of the corresponding RNAP an additional His 6 /Strep tag was fused to the N-terminal region of RpoH.
After sequence verification the plasmids were linearized and used for transformation of P. furiosus. Characterization of the transformants using PCR demonstrated that the Pyrococcus mutants with the chimeric constructs were unstable. A more detailed analysis using different PCR amplicons revealed that these mutants rapidly lost the eukaryal N-terminal domain. After about 40 generations the N-terminal domain was no longer detectable whereas the selection marker was still present (data not shown). In contrast, the Pyrococcus mutant with the RpoH-E62K subunit was stable. Growth experiments at 95uC with the wild type construct of RpoH MUR38Pf and the mutant MUR39Pf with the exchange showed a similar growth behavior than the wild type (data not shown).

In vitro transcription experiments indicate identical activities between wild type and mutated RNAP
Although the growth analysis of RpoH wild type and RpoH-E62K indicated no phenotypical difference, we additionally performed an in vitro transcription analysis to compare the activities of the purified RNAPs. To produce cell mass for RNAP purification we cultivated the wild type and the mutant strain in large scale (100 liter fermenter) and purified the RNAP as previously described [34]. Using this two-step procedure consisting of a Ni-NTA and a gel filtration chromatography, highly purified RNAP could be obtained. The degree of purification was compared by gradient SDS-Page and silver staining with purified RNAP with a His 6 tag located at subunit D (Fig. 5A, lanes 1 to 3). Both types of RNAP show for the most part a similar pattern of subunit stoichiometry, except that the electrophoretic mobilities of the modified subunits D (lane 1) and H (lanes 2 and 3) were increased due to the presence of the His 6 tags. The composition of the smaller subunits seems to be slightly modified in the case of the RNAP with RpoH-E62K, but the exact stoichiometry of these subunits was not analyzed in detail. Using identical amounts of purified RNAP, an in vitro transcription experiment revealed that both, RNAP with RpoH wild type sequence and the RNAP with the E62K exchange, show the same transcriptional activity as the

Discussion
In this paper we describe a random mutagenesis approach which converts the chimeric Rp5H construct consisting of the Nterminal Rpb5 domain from yeast and the C-terminal archaeal RpoH domain from P. furiosus from a yeast complementation inactive to a complementation active form. Specifically, we show that the single exchange E197K in the archaeal domain of the chimera is sufficient to functionally replace subunit Rpb5 of yeast RNAP.
Furthermore, using a genetic system for P. furiosus we could demonstrate that the corresponding substitution E62K in the RpoH subunit alone without the N-terminal extension seems to have almost no effect in vivo on the growth of P. furiosus. The establishment of the chimeric Rp5H construct in the Pyrococcus genome was not possible (Fig. 4). This finding is in contrast to in vitro data with a reconstituted DH RNAP enzyme of P. furiosus [22]. In this case yeast Rpb5 was able to rescue the activity of the reconstituted DH enzyme. This indicates that the additional presence of the N-terminal region in the chimeric Rp5H construct did not cause RNAP instability due to space problems. Along the same lines, a superimposition of the crystal structure of a crenarchaeal RNAP with the cryo-electron microscopy structure of the euryarchaeal RNAP from P. furiosus exhibited an easily accessible region in this part of the RNAP [39][40][41]. Furthermore, in some crenarchaeota an additional subunit of the RNAP, named Rpo13, is located at this position [7,39]. The most likely explanation for the failure to establish the chimeric construct in the Pyrococcus genome is the thermolabile nature of the N-terminal yeast domain. We assume that this part of the construct is sensitive to denaturation at the high growth temperature of P. furiosus, which might cause precipitation of the entire subunit. The in vitro experiment mentioned above was done at 70uC for 30 min [22] whereas the in vivo approach required long-term stability of the chimeric construct at 95uC.
The most interesting result from our data is that the single amino acid exchange E197K in the archaeal RpoH domain is sufficient to activate the chimeric subunit for complementation in yeast. To put this finding into a broader context, we analyzed the corresponding regions of subunit RpoH in several archaeal species using a multiple sequence alignment (Fig. 6A). The results of the sequence comparison indicate that mainly three different amino acids are present at this position: The negatively charged amino acid E is only found in the order of Thermococcales, Halobacterium and Aeropyrum pernix, whereas the positively charged amino acids R and K are present in all other archaea, examples are shown from the Sulfolobales order and the methanogens (Fig. 6A). A detailed analysis of the corresponding position in the crystal structure of the S. shibatae RNAP revealed that R64 of subunit H most likely forms a salt bridge with D12 of subunit K ( Fig. 6B; [3]). The distance between both amino acids is less than 4 Å , which is in perfect agreement with an ionic interaction (Fig. 6B, close-up; [41]). In this context it is interesting to note that the region containing D12 in subunit K of S. shibatae is missing in the K subunit of Pyrococcus (data not shown), which might allow for the negatively charged residue E62 in RpoH without generating electrostatic repulsion. As the crystal structure of the Pyrococcus RNAP is unavailable, a putative interaction partner of amino acid E62 is presently unknown.
At the position in subunit Rpb5 of Eukarya which corresponds to E62 in RpoH almost always a positively charged amino acid (R or K) is found (data not shown). The known structures of yeast RNAP contain only a part of Rpb6, which is the eukaryal subunit corresponding to RpoK [1,43]. It is therefore not possible to decide whether the lysine residue in Rpb5 and, by inference the newly introduced K197 residue in Rp5H, is engaged in a salt bridge interaction with a negatively charged residue from Rpb6 as found for the RpoH-RpoK pair in Sulfolobus RNAP. A RNAP structure with the complete Rpb6 molecule would be required to answer this question. On the other side, it has been shown that the region around E197K of Rpb5 is in close contact with subunit Rpb1 [26]. A detailed analysis of the yeast RNAP structure identifies amino acid D1013 of Rpb1 as a putative interaction partner for amino acid K197 of Rpb5, although the distance between the nitrogen of the lysine and the oxygen of the aspartate with almost 6 Å is relatively large for an ionic interaction ( Fig. 6C; [42,43]). Nevertheless, the strong demand for a positive charge at this position of the Rp5H chimera for the formation of an active yeast RNAP indicates that this amino acid is most likely involved in the formation of a salt bridge either between D1013 of Rpb1 or with a negatively charged amino acid of subunit Rpb6 in a similar way found for the Sulfolobus RNAP. Therefore, the presented data . Schematic drawing of the plasmids used for the construction of the Pyrococcus mutants. The plasmids were assembled by overlapping PCR. Plasmid pMUR27 contains the wild type sequence of RpoH and plasmid pMUR28 comprises RpoH with the single exchange E62K. Plasmids pMUR54 and pMUR43 contain the chimeric constructs. SimR is an additional copy of the hydroxymethylglutaryl CoA reductase from Thermococcus kodakarensis (TK0914) and provides resistance against simvastatin. For purification, a His 6 and a Strep tag is available at the N-terminal region of all constructs. At the right side of the figure the stability of the corresponding Pyrococcus mutants is indicated. doi:10.1371/journal.pone.0087485.g004 clearly demonstrate that the approach of random mutagenesis and library selection is a suitable tool to identify important amino acids which play a role during the evolution of RNAPs from eukarya and archaea.

Supporting Information
Table S1 Oligonucleotides used in this study. (DOCX) Figure 5. Purification and functional analysis of the RNAP. A: Silver-stained 10 to 20% gradient Tris Tricine SDS gel. Each lane contains 3 mg of purified RNAP. The analyzed RNAP fractions were purified using a His 6 tag at subunit D (lane 1) or subunit H (lanes 2 and 3). B: In vitro transcription with the purified RNAP fractions. Identical amounts of RNAPs were used to transcribe the gdh template in the presence of the archaeal transcription factors TBP and TFB. The transcription assays were performed as described previously [36]. doi:10.1371/journal.pone.0087485.g005 Figure 6. Sequence alignment and structural analysis of RpoH. A: Part of a multiple alignment of several RpoH sequences. For the alignment a webserver of the program ClustalW2 was used (www.ebi. ac.uk/Tools/msa/clustalw2). Conserved amino acids are shown in red and the positions corresponding to E62 of P. furiosus are shadowed in grey. B: Overall structure of the archaeal RNAP from S. shibatae (ID code in the Protein Data Bank: 2WAQ). Subunits A0, H and K are shown in space filling mode with different colors. Residue R64 (E62 in P. furiosus) of subunit H is shown in blue, and the putative salt bridge interaction partner D12 of subunit K is colored in red. In the close-up the distances between the nitrogen atoms of the arginine side chain and the oxygen atoms of the glutamate side chain are indicated in Å . The figures were drawn using the program PyMOL (www.pymol.org). C: Close-up of a putative salt bridge between K197 of Rpb5 and D1013 of RpbA from S. cerevisiae (ID code in the Protein Data Bank: 4BBR). The distances between the nitrogen atom of the lysine side chain and the oxygen atoms of the glutamate side chain are indicated in Å . doi:10.1371/journal.pone.0087485.g006