Zinc stoichiometry of yeast RNA polymerase II and characterization of mutations in the zinc-binding domain of the largest subunit.

Atomic absorption spectroscopy demonstrated that highly purified RNA polymerase II from the yeast Saccharomyces cerevisiae binds seven zinc ions. This number agrees with the number of potential zinc-binding sites among the 12 different subunits of the enzyme and with our observation that the ninth largest subunit alone is able to bind two zinc ions. The zinc-binding motif in the largest subunit of the enzyme was investigated using mutagenic analysis. Altering any one of the six conserved residues in the zinc-binding motif conferred either a lethal or conditional phenotype, and zinc blot analysis indicated that mutant forms of the domain had a 2-fold reduction in zinc affinity. Mutations in the zinc-binding domain reduced RNA polymerase II activity in cell-free extracts, even though protein blot analysis indicated that the mutant subunit was present in excess of wild-type levels. Purification of one mutant RNA polymerase revealed a subunit profile that was wild-type like with the exception of two subunits not required for core enzyme activity (Rpb4p and Rpb7p), which were missing. Core activity of the mutant enzyme was reduced 20-fold. We conclude that mutations in the zinc-binding domain can reduce core activity without altering the association of any of the subunits required for this activity.

Zinc is an integral component of both bacterial and eukaryotic DNA-dependent RNA polymerases (RNAPs) 1 and is essential for their function (1). The importance of zinc to the function of RNAP is not well understood; however, studies have suggested that it is required to maintain structure that is important for interactions with the other polymerase subunits and/or with DNA/RNA at the active site.
Bacterial RNAP binds two zinc ions that are thought to play a structural role in the formation of the active polymerase (2).
One of these two zinc ions is bound in the RNAP largest subunit (␤Ј) by a non-canonical zinc-binding motif ( 814 CX 73 CX 6 CX 2 C 898 ) (3), and the second zinc is thought to be bound by a zinc-binding motif in the amino-terminal end of the ␤Ј subunit. Mutations in the non-canonical domain inhibit the ability of denatured ␤Ј to bind zinc and fold in vitro into a state that is able to form a functional RNAP when added to the other subunits (3). The coordination of this zinc was confirmed by the recent crystal structure of Thermus aquaticus RNAP (4). The second zinc-binding motif in the amino terminus of the ␤Ј subunit has been implicated along with the carboxyl terminus of the second largest subunit (␤) in maintaining the stability of the DNA-RNA-RNAP ternary complex in its elongating phase (5). Unfortunately, the analogous 74-residue segment in the ␤Ј subunit of T. aquaticus RNAP lacks electron density in the crystal structure (4) so the coordination of zinc by this domain cannot be confirmed.
The ␤Ј and ␤ subunits of Escherichia coli RNAP are orthologs of the first and second largest subunits of eukaryotic RNAP polymerases I, II, and III (reviewed in Refs. 6 -8). Although the largest subunit of eukaryotic RNA polymerases has no equivalent to the ␤Ј non-canonical zinc-binding motif, they do have an extended form of the bacterial ␤Ј amino-terminal motif (see below) that has similarly been suggested to play a structurerelated role. For instance, genetic evidence has implicated the zinc-binding motifs of the largest and second largest subunits of yeast RNAP I in mediating a functional interaction between these two subunits (9), and a mutation in the zinc-binding motif of the largest subunit of RNAP III alters the association of three of the smaller subunits with the core enzyme (16).
The number of zinc ions bound to eukaryotic RNAP is uncertain. In a zinc-blotting assay, zinc binds to 6 of the 12 different subunits of RNA polymerase II (RNAP II) from the yeast Saccharomyces cerevisiae (10,11). However, the number of zinc ions associated with this enzyme was measured by atomic absorption spectroscopy to be only 1 or 2 (12,13). Here, we attempt to reconcile this apparent contradiction by measuring the zinc content of an affinity purified form of yeast RNAP II. We show that the 9th largest subunit (Rpb9p) alone is able to bind two zinc ions.
Each of the subunits thought to bind zinc in yeast RNA polymerase II (see Refs. 10 and 11 and Table I) contains a zinc-binding motif of the structural type that forms a secondary structure, which in turn interacts with protein or with nucleic acids. These types of motifs include at least four potential coordinating amino acids, predominantly cysteines, and to a lesser extent histidines, and these potential ligands occur within a relatively small stretch of amino acids (14,15). One such motif ( 67 CX 2 CX 6 CX 2 HX 26 CX 2 C 110 ) occurs in the largest subunit of RNA polymerase II (Rpo21p; also called Rpb1p) and is conserved in the largest subunits of all three RNA polymerases from a variety of eukaryotes (16). Part of this motif is also conserved in the amino terminus of the bacterial ␤Ј subunit (17). Amino acids Arg-47 to Asn-119 of Rpo21p containing this motif were expressed as a fusion protein in E. coli and were shown to bind zinc in a zinc-blotting assay (10).
As a first step toward understanding the role of zinc in the largest subunit of RNAP II, we mutated those residues in the Rpo21p zinc-binding domain (ZBD) that are both well conserved and capable of coordinating zinc. We determined the effect of conservative substitutions in these residues on viability, the ability of the ZBD to bind zinc in vitro, and the activity and stability of RNAP II core enzyme. We conclude that mutations in the Rpo21p-ZBD are tolerated better than mutations in the analogous ZBD of the RNAP III largest subunit (16). We also show that one mutation in the Rpo21p-ZBD has an effect on the basal activity of core RNAP II without altering association of any of the subunits required for this activity.

EXPERIMENTAL PROCEDURES
Atomic Absorption Spectroscopy and Amino Acid Analysis-RNA polymerase II was prepared using heparin-Sepharose, DEAE-Sephacel, affinity chromatography using 8WG16 monoclonal antibody (18), and high pressure liquid chromatography fractionation essentially as described previously (19) with the following changes. Polymerase was prepared from 450 g of bakers' yeast cake stored at Ϫ70°C. Protease inhibitors (1 mM phenylmethylsulfonyl fluoride and 2 mM benzamidine) and 10 mM DTT were included throughout the purification. The heparin-Sepharose column was washed with 5 column volumes of buffer A containing 150 mM KCl followed by 10 column volumes containing 200 mM KCl. DE52 resin was used in place of DEAE-Sephacel to remove nucleic acids. Flow-through from the DE52 column was applied directly to 8WG16-Sepharose beads in 500 mM ammonium sulfate and incubated for 4 h with one change of beads. The protein was eluted from the 8WG16-Sepharose beads at 23°C and was purified further by high pressure liquid chromatography on an Amersham Pharmacia Biotech Mono-Q column using a gradient of ammonium sulfate from 70 mM to 1 M in 50 mM Tris-HCl (pH 7.5), 1 mM EDTA, 10% glycerol, and 10 mM DTT (no protease inhibitors were included). RNA polymerase II lacking subunits 4 and 7 (RNAP II⌬4/7) was a gift from Aled Edwards that was prepared from yeast strain rpb-4 as described previously (19,20).
RNA polymerase II was prepared for atomic absorption spectroscopy and amino acid analysis by dialysis against three changes of 2 liters of buffer D (20 mM HEPES (pH 7.5), 10 mM DTT, and 0.5 or 10 mM EDTA) over 6 h at 4°C. This was followed by dialysis against three changes of 1 liter of metal-free buffer (100 mM ammonium bicarbonate and 1 mM DTT made metal-free by passage over a Chelex 100 ion exchange membrane (Bio-Rad)) over 12 h at 4°C. SDSpolyacrylamide gel electrophoresis and Coomassie staining revealed that RNA polymerase II subunits represented greater than 99% of the protein present (Fig. 1).
Zinc content was determined by Zeeman graphite-furnace atomic absorption spectroscopy at the Best Institute Trace Elements Laboratory (all RNAP II⌬4/7 samples) or in Dr. J. E. Coleman's laboratory by Matthew Junker and Kevin Gardner (all RNAP II samples) using an Instrumentation Laboratory IL157 spectrophotometer. Zinc content was determined from a standard absorption curve for 1-15 M zinc. Post-column, phenylisothiocyanate amino acid analysis of samples was performed at the Best Institute HSC/Amersham Pharmacia Biotech center with 25 nmol of norleucine added to assay recovery of the sample after hydrolysis. The concentration of protein was then calculated from the concentration of arginine, alanine, and leucine given the molar ratio of these amino acids to RNAP II or to RNAP II⌬4/7 (see Refs. 8 and 21 and references therein).
His-tagged Rpb9p, Rpb9p-Zn1 (amino acids Met-1 to Glu-47), and Rpb9p-Zn2 (amino acids Thr-55 to Ser-112) were prepared from E. coli by Sally Hemming (22). Each of the proteins retained three amino acids (Gly, Ser, and His) at the amino terminus after removal of the His tag. Zinc stoichiometry was determined for these proteins essentially as described above.
S. cerevisiae strains used in this study are derivatives of W303-1a (see Table II). The plasmid shuffling scheme (see below) was based on yeast strain YF1703 (constructed by D. Jansma) in which the portion of the RPO21 open reading frame extending from the AvrII site (nucleotide (nt) ϩ166; where the A of the initiating methionine codon is ϩ1) to the SnaBI site (nt ϩ5200) has been replaced by a BamHI fragment containing the HIS3 gene. Viability of this strain is maintained on galactose by the maintenance plasmid, pYF1577 (a pAPS2-based (25), URA3, CEN/ARS plasmid constructed by J. Archambault possessing a copy of RPO21 whose endogenous promoter was replaced with the glucose-repressible/galactose-inducible GAL10 promoter (26)). Strain YF1733, constructed by D. Jansma (27), was used to introduce mutant alleles of RPO21 into the chromosome; the chromosomal copy of RPO21 in this strain has been disrupted by insertion of the ADE2 gene between the BstE2 (Ϫ722) and SpeI (ϩ658) sites of RPO21.
Plasmids Used in the Mutagenic Analysis of the Rpo21p-ZBD-The plasmid pYF1560 is derived from pYF1361. The plasmid pYF1361 (constructed by M. Drebot) is a 7.2-kb HindIII genomic fragment of the yeast S. cerevisiae containing the open reading frame of RPO21 as well as 1.5 kb of upstream sequence subcloned from pJAY36 (constructed by J. Archambault) to the polylinker of a pBluescript II KS(ϩ) vector (Stratagene) whose SpeI site had been previously destroyed by endfilling. Two unique sites in pYF1361 (SalI in the pBluescript II KS(ϩ) polylinker and an SpeI site downstream of the zinc-binding motif) encompassed a 2.1-kb fragment containing about 1.5 kb of sequence upstream of RPO21 and the 5Ј end of the RPO21 open reading frame encoding amino acids Met-1 to Ser-221. This 2.1-kb, SalI-SpeI fragment was subcloned from pYF1361 to the polylinker of pBluescript II KS(Ϫ) to generate pYF1560. pYF1513 (constructed by D. Jansma) was made by subcloning the 7.2-kb, HindIII, yeast genomic fragment from pYF1361 to pFL39 (TRP1 CEN6 ARS) (30)) whose unique EcoRI site had been destroyed. This left the EcoRI and SpeI sites (nt Ϫ312 and ϩ658 respectively; where the A of the initiating methionine codon is ϩ1) in the yeast genomic fragment unique to the plasmid. All DNA manipulations were performed essentially as described (31).
Site-directed Mutagenesis of the Rpo21p-ZBD and Phenotypic Analysis-Antisense, single-stranded DNA was prepared from pYF1560 in E. coli strain CJ236 using the helper bacteriophage R408 as described (32). Second-strand synthesis using the Klenow fragment of DNA polymerase I (33) was primed using a mutagenic oligonucleotide designed a The zinc stoichiometry for each of the subunits (except Rpb9) is assumed to be at least 1 from the ability of these subunits to bind zinc in vitro (10,11 (21) before it was resolved from a co-migrating subunit Rpb11p (54) and so may have been overestimated.
j Zinc stoichiometry determined in this study. k Also known as ABC10␤ and RPB10␤ (11,55). l Stoichiometry of Rpb10p was determined before it was resolved from Rpb12p. The stoichiometry of both is assumed here to be 1 for the purposes of calculation.
to introduce the amino acid change(s) in Rpo21p listed in Fig. 2. The double-stranded product was introduced into JF1754 (dut ϩ ung ϩ ) to select against the wild-type product. Dideoxynucleotide chain termination method sequencing (34) was used to confirm the presence of only the expected mutation(s) between the EcoRI and SpeI sites. These 1-kb EcoRI-SpeI fragments were then subcloned to pYF1513 to replace the wild-type fragment and generate plasmids pYF1547 to pYF1559 carrying mutant alleles rpo21-27 to rpo21-39, respectively. These plasmids, expressing mutant alleles of RPO21 from the endogenous promoter, were introduced (28) into the RPO21 gene disruption yeast strain, YF1703, where the conferred phenotypes were assayed by plasmidshuffling analysis (29); transformants were grown at 23 or 30°C on SC-glucose medium supplemented with 5-FOA to select for loss of the URA3 maintenance plasmid, pYF1577. Finally, 5-FOA r colonies were streaked to YPD or SC-glucose medium lacking histidine and tryptophan at the following four different temperatures: 16, 23, 30, and 37°C. These experiments were performed in parallel with pYF1513 (positive control) and pFL39 (negative control) transformants of YF1703. The resulting strains are YF2155 (pYF1513: RPO21) and YF2142-YF2154 (rpo21-27 to rpo21-39).
Mutant alleles of RPO21 were introduced into the chromosome by transforming yeast strain YF1733 with plasmids pYF1547 to 1553 and pYF1513 cut with HindIII. Transformants able to grow on glucose medium were screened for an Ade Ϫ (pink colored) phenotype. Finally, transformants were selected in which the pYF1577 maintenance plasmid was lost on 5-FOA medium indicating those transformants (YF2063 to YF2070) in which the chromosomal rpo21::ADE2 allele had been replaced by the allele carried on the transforming plasmid DNA (rpo21-27 to rpo21-33 and RPO21), respectively. At least two transformants of each allele were tested for growth phenotypes at 23, 30, and 37°C on YPD medium.
Fusion Proteins-The sequence that encodes amino acids Arg-47 to Asn-119 of Rpo21p was amplified as a polymerase chain reaction fragment flanked by BamHI and PstI sites using Taq polymerase and the primers 5Ј-d(CCCCCGGATCCAGAGCGAAAATTGGTGGTC)-3Ј and 5Ј-d(CCCCCCTGCAGCTAATTATGTTCATCCAGTAATAG)-3Ј and either pYF1513 (RPO21), pYF1556 (rpo21-36), pYF1557 (rpo21-37), or pYF1559 (rpo21-39) as template. These fragments were then subcloned in frame to the 3Ј end of the coding region for the maltose-binding protein (MBP) in the bacterial expression vector pMAL-c2 (New England Biolabs). The resulting fusion proteins were expressed in TB1 E. coli and purified by affinity chromatography on agarose-maltose beads (New England Biolabs) as described by the manufacturer. Each of the protein preparations was analyzed by SDS-polyacrylamide gel electrophoresis (35); in each case, a single, Coomassie Blue-stained band was observed corresponding to the expected fusion protein product of 52 kDa. The ability of these proteins to bind zinc was assayed using an in vitro zinc-blotting assay previously described (10). Negative controls from New England Biolabs included maltose-binding protein alone (MBP2*) and MBP fused to the ␤-galactosidase-␣ domain (MBP-LacZ).
Protein Blot Analysis-Preparation of whole-cell yeast extracts and protein blot analysis were performed essentially as described (27,36). Two antibodies were used to detect Rpo21p. The first was the mouse monoclonal antibody 8WG16 (gift of Richard Burgess (37)) which recognizes the hypo-phosphorylated carboxyl-terminal domain of Rpo21p (36). The second was a rabbit polyclonal antibody raised against a fusion of glutathione S-transferase and the phosphorylated carboxylterminal domain (gift of Susan McCracken and David Bentley). Rpb2p was detected with polyclonal antibody anti-B150 (gift of Andre Sentenac). Secondary goat anti-IgG rabbit or goat anti-IgG mouse conjugated to horseradish peroxidase was from Life Technologies, Inc. Secondary antibody was detected using ECL reagent as described by the manufacturer (Amersham Pharmacia Biotech). Images were scanned and quantified using NIH Image software.
Preparation of Transcription Extract-Whole-cell transcription ex-tracts were prepared from yeast cells grown at 30°C in YPD medium to an A 600 of 1.5-2.0 in log phase. Harvested cells were ground under liquid nitrogen as described (38)  All reactions were performed in either the presence or absence of 10 g/ml ␣-amanitin (Sigma). Reactions were started by adding 5 l of extract (typically containing 20 g of protein) or purified RNA polymerase II in TDB buffer and allowed to proceed at 23°C for 5-30 min. Reactions were stopped by adding 10 l of stop buffer (80 mM EDTA, 2% SDS, and 2 g/l proteinase K) and incubating at 37°C for 20 min. The reaction mix was spotted (25 l) to a 2.4-cm diameter DE81 filter (Whatman) that was washed three times in P buffer (350 mM Na 2 HPO 4 , 10 mM sodium pyrophosphate) for 5 min each, followed by 3 times in water for 2 min each, and finally by a rinse in ethanol. Filters were dried and counted in scintillation fluid using an LKB 1217 liquid scintillation counter. All reactions were performed in triplicate. Error between identical reactions was generally less than 5%. Counts per min in no-DNA control reactions represented less than 5% of counts in DNA-containing reactions.
Promoter-specific transcription assays were performed essentially as described (39). Reactions were initiated by adding 7 l of extract (typically containing 80 g of protein) in TDB to the reaction mix that included 300 ng of circular template DNA: pGAL4CG- (41). Reactions were incubated at 23°C for 1 h. Processed reactions were separated by electrophoresis on a 6% polyacrylamide, 7 M urea gel. The dried gel was exposed to a PhosphorImager screen for 6 -12 h. Quantification of the data was done using Image-Quant software (Molecular Dynamics).

RESULTS
Zinc Stoichiometry of RNA Polymerase II-It has been shown that six of the yeast RNAP II subunits are capable of binding zinc in an in vitro zinc-blotting assay (see Refs. 10 and 11 and Table I). However, it has been estimated from atomic absorption analysis that only one or two zinc ions are bound by this enzyme (12,13). In an attempt to resolve this apparent contradiction, we prepared an affinity purified form of RNAP II (see "Experimental Procedures" and Fig. 1) and determined its zinc content. The highly purified nature of this enzyme allowed us to determine protein concentration by amino acid analysis. Zinc content was determined by atomic absorption spectroscopy for three independent preparations. The molar ratio of zinc to polymerase was found to be 7.26 Ϯ 0.27 following dialysis of the enzyme against 10 mM EDTA at 4°C followed by metal-free buffer (see "Experimental Procedures" and Table III). These a W303-1a and W303-1b were obtained from R. Rothstein. b Constructed by David Jansma.
samples bound comparable amounts of zinc when dialyzed against only 0.5 mM EDTA. In addition, we used a preparation of RNAP II⌬4/7, which lacks subunits Rpb4p and Rpb7p. Neither of these two subunits contains a zinc-binding motif nor are they able to bind zinc in an in vitro zinc-blotting assay. Furthermore, this polymerase is indistinguishable from RNAP II in promoter-independent transcription assays. The molar ratio of zinc to RNAP II⌬4/7 was calculated to be 7.0 Ϯ 0.22 (Table   III). We conclude that the number of zinc ions bound by RNAP II is one more than the number of subunits that are able to bind zinc in an in vitro zinc-blotting assay. Rpb9p has two potential zinc-binding motifs, and therefore, it could account for at least two of the zinc ions bound to RNAP II. We measured the zinc bound independently by Rpb9p and to the two halves of Rpb9p (Rpb9p-Zn1 and Rpb9p-Zn2; each containing one zinc-motif) produced in E. coli. We found that these proteins bound 2.3 Ϯ 0.25, 0.7 Ϯ 0.08, and 0.7 Ϯ 0.07 zinc ions, respectively (see "Experimental Procedures" and Table  III).
In conclusion, approximately 7 zinc ions are bound by RNAP II, two of which are accounted for by Rpb9p alone. This value falls within the expected range of potential zinc-binding ability of RNAP II based on the number of zinc-binding motifs present and the stoichiometry of each of the subunits (see Table I). It seems likely, from the data presented above, that each of these motifs may in fact bind zinc and that the zinc content of RNAP II was originally underestimated (see "Discussion").
Zinc-binding Domain Mutagenic Analysis-The zinc-binding domain of Rpo21p (hereafter referred to as the Rpo21p-ZBD) consists of amino acids Arg-47 to Asn-119. This domain was expressed as a fusion protein to the maltose-binding protein of E. coli and was shown to bind zinc in an in vitro zinc-binding assay (10). This domain contains a zinc-binding motif that is conserved among the largest subunits of all three eukaryotic RNA polymerases from a variety of organisms as well as in viral and archaebacterial RNA polymerases (16). Six amino acids in the Rpo21p-ZBD have been identified as being both conserved and potentially zinc-coordinating: Cys-67, Cys-70, Cys-77, His-80, Cys-107 and Cys-110 (16). Mutations encoding individual substitutions in each of these residues were constructed. Each of the mutant alleles was introduced on a low copy number plasmid into a strain of yeast lacking a chromosomal copy of RPO21 but carrying a wild-type RPO21 copy on a maintenance plasmid. Loss of the maintenance plasmid was selected for by growth on medium containing 5-FOA, and the phenotype conferred by the remaining mutant rpo21 allele was assayed at various temperatures. The results of the phenotypic analysis are summarized in Fig. 2.
The first four potentially zinc-coordinating amino acids of the zinc-binding motif all conferred temperature-sensitive phenotypes at 37°C when individually mutated; C67S(rpo21-27) and C70S(rpo21-28) conferred slow growth phenotypes at 37°C, whereas C77S(rpo21-29) and H80Y(rpo21-30) were unable to support growth at 37°C. These results were not altered by expressing the mutant alleles from low copy number plasmids since the phenotypes were no more severe when the alleles were expressed from single copies on the chromosome (data not shown). The most dramatic phenotypes were observed for amino acid changes C107S and C110S; the rpo21-34 (C107S) allele fails to support growth since the plasmid expressing this allele did not permit the loss of the RPO21 maintenance plasmid on 5-FOA medium. Also, the plasmid expressing the rpo21-35 (C110S) allele permitted loss of the maintenance plasmid on 5-FOA medium only at 16 and at 23°C. The resulting FOA r strain grew slowly at 16 and 23°C and failed to grow at 30 or 37°C. Each of these phenotypes was visible only when the RPO21 maintenance plasmid was removed, indicating that the mutant alleles were recessive. These results suggest that the conserved residues of the zincbinding motif are important to the function of RNA polymerase and therefore to the viability of the cell, especially at higher temperatures. We constructed substitution mutations in other cysteine and histidine residues of the zinc-binding domain that were not as well conserved in order to strengthen this hypoth-  esis. These mutant strains (H83Y(rpo21-31), C103S(rpo21-32), and C105S(rpo21-33)) were all without phenotype under the conditions assayed, suggesting that this domain is not generally sensitive to substitutions unless they lie within the conserved zinc-binding motif. We reasoned that if each of these conserved residues is involved in a similar function (i.e. in binding zinc) then the effect of one mutation that prevents proper zinc binding in this domain should not be exacerbated by the presence of another. Multiple mutations were constructed that substituted whole groups of potentially zinc-coordinating residues in the Rpo21p-ZBD at once. In general, these multiple mutations (M1 and M2) conferred phenotypes that were only slightly more severe than those conferred by the individual substitutions of which they were composed (Fig. 2). In combination, substitutions C67S and C70S (M1) prevented growth at 37°C, whereas each of these individually only conferred slow growth at 37°C. In addition, both multiple mutations M1 and M2 (C77S, H80Y, and H83Y) conferred growth phenotypes at 16°C that were not seen for the individual substitutions. The observation that multiple mutations in this domain are not synergistically lethal argues that each of the mutated residues plays a role in some common function (such as coordinating zinc). Additional structural changes in the multiple mutants may explain the appearance of the new or more severe phenotypes that do not appear for the single mutants.
In Vitro Zinc Binding Analysis-Since substitutions of amino acids that are both highly conserved and capable of coordinating zinc in the Rpo21p-ZBD confer growth phenotypes, it is possible that these same substitutions might reduce the ability of this domain to bind zinc. The Rpo21p-ZBD (amino acids Arg-47 to Asn-119) was produced in E. coli as a fusion to the MBP. In addition, three mutant versions were constructed.
Wild-type MBP-ZBD fusion protein was able to bind zinc in a zinc-blot assay (Fig. 3B, lane 4). This binding activity was attributed to the ZBD portion of the fusion protein since neither MBP2* (lane 2) nor MBP-lacZ␣ (lane 3) bound detectable levels of zinc in this assay. Zinc binding was quantified and normalized to protein by staining the membrane with Amido Black. This revealed that the zinc bound by mutant fusion proteins MBP-M1, MBP-M2, and MBP-M4 was reduced to 40, 60, and 45%, respectively, compared with that bound by the wild-type MBP-ZBD. This decrease is consistent with a reduced affinity for zinc conferred by the ZBD mutations. We hypothesize that a deficiency in zinc binding underlies the decreased RNAP II activity (see below) and the growth phenotypes conferred by these same mutations.
Effect of ZBD Mutations on Rpo21p Steady-state Levels and RNAP II Stability-We examined one of the most severe growth phenotypes conferred by a mutation in the ZBD. Fig. 4 shows the following: 1) the growth rate of the C110S substitution mutant begins to slow 2 h after shift to the non-permissive temperature, and 2) the cell mass ceases to increase by 24 h after shift. In addition, we assayed the ability of these cells to grow on solid medium at various times after shift. Cell viability began to decrease after 5 h of shift to the non-permissive temperature (data not shown). This pattern was similar for both mutants M1 and M2, although mutant M1 took longer to cease dividing, and shifted cells could recover when transferred to solid medium at the permissive temperature even after 30 h of shift. In general, mutations in the zinc-binding domain confer a slow shut-off phenotype when shifted to the non-permissive temperature. Such a phenotype is consistent with a weak defect in Rpo21p stability, in core RNAP II stability/assembly and/or in RNAP II activity.
It has been demonstrated that underproduction of the larg- Growth phenotypes on solid medium at four different temperatures, as determined by plasmid-shuffling experiments, are described: FOA s indicates that the strain will not lose the maintenance plasmid at any temperature; ϩ indicates growth indistinguishable from the strain transformed with wild-type RPO21 (pYF1513); Ϫ indicates an absence of growth, and s indicates slow growth at the given temperature. a Two temperature-sensitive mutants previously isolated (57) in this same domain. b rpc160 mutants (analogous to those in this study) described in Ref. 16 and expressed extrachromasomally are shown, where (Ϫ) indicates a lethal phenotype and (ϩ) indicates wild-type growth. est subunit of RNAP II can confer temperature-sensitive growth phenotypes (27). Therefore, we hypothesized that the steady-state level of the largest subunit might be reduced by substitutions in the zinc-binding domain. The data shown in Fig. 4C suggest that a stability defect is unlikely; protein blot analysis of extracts prepared from mutant and WT cultures grown at the permissive temperature showed that C110S had, if anything, increased levels of Rpo21p compared with WT even though this mutant grows more slowly than the WT strain at 23°C. Furthermore, this increase persisted at least 2 h after temperature shift (data not shown). Even at 6 h after shift the level of Rpo21p in C110S was comparable to that of WT grown at the permissive temperature (compare lanes 1 and 4). Mutants M1, M2, and H80Y also had increased Rpo21p levels at the permissive temperature (30°C) (Fig. 5C) which persisted for at least 2 h after shift to the non-permissive temperature (data not shown). Steady-state levels of both Rpo21p and Rpb2p did decrease in the C110S substitution mutants at 12 and 24 h (Fig. 4C, lanes 5 and 6); most likely, this reflected an overall decrease in RNAP II levels as the cells began to die.
We considered the possibility that increased levels of Rpo21p as detected by 8WG16 antibody might indicate increased levels of hypo-phosphorylated Rpo21p (36) rather than overall increased levels of the subunit since the monoclonal antibody 8WG16 preferentially recognizes the hypo-phosphorylated form of the carboxyl-terminal domain of Rpo21p. We probed protein blots with a polyclonal antibody raised against the phosphorylated carboxyl-terminal domain of Rpo21p, and we confirmed that levels in mutants were indeed increased with respect to that found in WT extracts (data not shown). We conclude that the temperature-sensitive defects conferred by mutations in the zinc-binding domain are not due to an overall decrease in the steady-state levels of Rpo21p.
In order to confirm that the ZBD mutants conferred an RNAP II defect, we prepared whole-cell extracts from WT and mutant strains (M1, M2, and H80Y) grown at the permissive temperature of 30°C. Each of the mutant extracts showed about a 2-fold reduction in RNA polymerase II activity in as-says that used heat-denatured calf thymus DNA as template (Fig. 5A). This defect was specific to RNAP II activity since RNA polymerase I plus polymerase III activity in these same mutant extracts was at least 90% that present in WT extracts (Fig. 5A). We also assayed these extracts in promoter-specific transcription assays (39). Mutants were able to produce fulllength, accurately initiated transcripts (Fig. 5B). However, there was again a similar decrease in overall activity indicating that the mutations likely affect the promoter-independent activity of RNA polymerase and have no specific effect on initiation. No increase in the defect was seen with increasing temperatures (data not shown) indicating that the growth phenotype is likely a result of increased sensitivity to reduced RNA polymerase II activity at higher temperatures rather than to a temperature-sensitive gene product.
Mutations in the zinc-binding domain of the RNAP III largest subunit were reported previously to alter the association of smaller subunits with the core enzyme (16). We hypothesized that mutations in the analogous RNAP II ZBD also alter asso-ciation of other subunits with the core enzyme. In order to investigate this possibility, we prepared enzyme from one of the mutants (M1; C67S, C70S) in exactly the same manner described for preparing WT RNAP II for the zinc stoichiometry determinations. Fig. 6 shows that each of the RNAP II subunits is present in the M1 preparation with the exception of Rpb4p and Rpb7p. Each of the subunits was quantified from a scan of a Coomassie-stained gel using NIH Image software. The intensity of bands corresponding to subunits Rpb2p, Rpb3p, Rpb5p, Rpb8p, and co-migrating subunits Rpb9pϩ11p indicated that these subunits were present in stoichiometries (relative to Rpo21p) identical to those for WT RNAP II. Rpb6p stains poorly with Coomassie Blue and could not be quantified; however, it was found to be present by silver staining (Fig. 6B). Both Rpb9p and Rpb11p were detected and resolved by silver staining. A band corresponding to co-migrating subunits FIG. 5. Transcription activity of ZBD mutant extracts. A, equivalent amounts of whole-cell transcription extract (20 g) from WT (YF2155) and each of the mutants M1 (YF2151), M2 (YF2152), and H80Y (YF2145) were assayed for RNA polymerase activity in a transcription assay using calf thymus DNA as template for 20 min at 23°C. The amount of extract and the time of reaction used were both shown to be non-saturating. Total RNA polymerase activity or RNA polymerase I plus III activity (stippled bars) was measured, respectively, in the absence or presence of ␣-amanitin. RNA polymerase II activity was calculated as the difference between the two (black bars). Activity was expressed as a percentage of that found in WT extracts. Each of the bars represents the average of three trials with an error of less than 5%. The experiment was carried out with two sets of extracts. B, equivalent amounts of whole-cell transcription extract (80 g) made from the same strains were also assayed for promoter-specific activity at 23°C using template DNA that has a G-less cassette. Transcripts initiated from this promoter do not incorporate guanosine and so they are not degraded in the presence of RNase T 1 . Transcripts (labeled with [␣-32 P]UTP) from two separate start sites of 375 and 350 nt were detected. Numbers below each lane represent promoter-specific transcription product relative to WT. C, protein-blot analysis of Rpo21p levels in WT and mutant extracts. The whole-cell extracts used above were adjusted to equivalent protein concentrations (as determined by Bradford assays). Rpo21p levels in increasing volumes of each extract were examined by protein-blot analysis with 8WG16 antibody. Numbers below lanes indicate signal strength with respect to signal from an equivalent volume of WT extract. Rpb10p and Rpb12p was present in both the mutant and WT enzymes, but the two subunits were not resolved. Despite the presence of each of these subunits, the M1 RNAP II was 20-fold less active in transcription assays using denatured calf thymus DNA as template (Fig. 6C). The fact that this decrease is greater than observed for the M1 whole-cell extracts may reflect the absence of other factors present in the extract that would not be present in this assay with core polymerase alone. This decrease in activity cannot be explained by the absence of Rpb4p and Rpb7p since these subunits are required only for promoter-directed transcriptional activity (20). Alternatively, Rpb4p and Rpb7p may be required in the presence of a mutant ZBD; however, this seems unlikely since recombinant Rpb4p and Rpb7p added to a M1 RNAP II preparation in a 30-fold molar excess did not restore transcription (data not shown). These observations demonstrate that mutations in the ZBD can decrease basal activity of the enzyme without observably altering the presence of any subunits required for this function.
We cannot conclude that the M1 ZBD mutation causes the absence of Rpb4p and Rpb7p. The association of these subunits with the polymerase is usually substoichiometric and can vary with the growth phase of the cells used to prepare the polymerase (42). It is estimated that the stoichiometry of Rpb4p in RNAP II is 0.2 in log phase cells and that this increases to 1 in stationary phase cells (42). The M1 RNAP II was prepared from cells grown to late log phase and the WT polymerase from yeast cake (stationary phase). This might account for part of the absence of Rpb4/7p from the M1 preparation, but it cannot completely explain it. Mutations in the zinc-binding domain might make minor modifications to the structure of RNAP II that cause Rpb4p and Rpb7p to dissociate more easily during purification. We note that at least one other mutation in the largest subunit has been shown to alter association of these subunits with the polymerase (43). DISCUSSION The evidence presented here supports the results of zinc-blot analysis showing that 6 of the 12 different RNAP II subunits bind zinc. One of the subunits (Rpb9p) binds 2 equivalents of zinc. This is the first study to show a correlation between the number of zinc ions bound by RNAP II in vitro and the number of zinc-binding motifs present in the core enzyme.
The three largest subunits and the three smallest subunits of RNAP II bind zinc in a zinc-blot assay (10,11). Given that Rpb9p binds two zinc ions (this study), there is a potential for RNAP II to bind at least seven zinc ions (see Table I). The actual number of zinc ions bound may be more or less than this depending on the stoichiometry of Rpb9p, Rpb10p, and Rpb12p (see Table I footnotes). Indeed, we found that RNAP II was able to bind 7.26 Ϯ 0.27 zinc ions. In addition, we found that 7.0 Ϯ 0.22 zinc ions were associated with RNAP II⌬4/7. These results are more consistent with the number of RNAP II subunits able to bind zinc in an in vitro blotting assay than with the earlier, independent estimations of 1 or 2 zinc ions (12,13).
The large difference between our result and those published previously may be due to a combination of factors including the method of RNAP II purification and preparation for atomic absorption spectroscopy. In this study, RNAP II was dialyzed against either 0.5 mM EDTA or 10 mM EDTA at 4°C with no apparent difference in the number of zinc ions bound. This higher concentration of EDTA was shown previously to remove half of the zinc associated with the enzyme after a 4-h dialysis at room temperature (13). This raises the possibility that binding of zinc to RNAP II may be less stable at higher temperatures. Our zinc determination differs in one other important respect from others; we were able to determine protein concentration from amino acid analysis. This is essential to obtaining an accurate, absolute value that cannot be determined reliably from relative protein concentration methods.
Other eukaryotic RNA polymerase II enzymes have been reported to bind various numbers of zinc ions (reviewed in Ref. 1 ranging from 2.2 for Euglena gracilis (44) to 6 for wheat germ RNAP II (45)). These differences may reflect a different number of zinc ions bound by each polymerase. Alternatively, each polymerase may bind a similar number of zinc ions that are not always retained depending on the method used to purify the polymerase and prepare it for atomic absorption spectroscopy.
This work is an initial attempt to understand the role of one of these RNAP II zinc motifs using mutational analysis (see Fig. 2). Growth phenotypes conferred by changes to potential zinc-coordinating amino acids of the Rpo21p-ZBD demonstrate that the conserved amino acids of the ZBD are important for the function of RNA polymerases II. This also has been demonstrated for the analogous domain in the largest subunit of RNAP III (Rpc160p) (16). In contrast, however, substitutions of any of the conserved residues in Rpc160p conferred a much more severe phenotype (lethal) compared with the Rpo21p substitutions, of which only one conferred a lethal phenotype (Fig.  2). These differences could be explained by the use of conservative amino acid substitutions in Rpo21p (this study) compared with the non-conservative changes made in Rpc160p (16). However, identical substitutions (H80Y and G79D) were made in the ZBDs of both Rpo21p and of Rpc160p, and these conferred a lethal phenotype at 30°C only in the RNAP III largest subunit. In addition, non-conservative substitutions G79D and G111E in Rpo21p both conferred temperature-sensitive phenotypes that were similar to conservative changes in this same region (Fig. 2). Finally, substituting any of the conserved histidine or cysteine residues in the Rpc160p-ZBD for other residues capable of coordinating zinc also conferred a lethal phenotype (16). Taken together, these data suggest that the function of the Rpo21p-ZBD is generally much more tolerant to mutations than the Rpc160p-ZBD.
Substitutions C107S and C110S conferred phenotypes that were much more severe than those made in any of the other conserved Rpo21p-ZBD residues. The meaning of this is uncertain without additional structural data, including an identification of those residues that directly coordinate zinc. Cys-107 and Cys-110 are not conserved in the ␤Ј subunit zinc motif of E. coli RNA polymerase (17); either E. coli RNA polymerase does not require these residues or their function is provided elsewhere in the molecule.
In considering how mutations in the ZBD might affect RNAP II activity, we addressed three possibilities. The ZBD mutations might cause a defect in the amount of the Rpo21p subunit, in core RNAP II stability/assembly, and/or in RNAP II core activity itself. The possibility that the ZBD mutations simply caused a decrease in the amount of Rpo21p seems unlikely based on an analysis of Rpo21p steady-state levels in one ZBD mutant (Fig. 4). The fact that Rpo21p levels were actually increased in mutant extracts (Figs. 4C and 5C) may be the result of an Rpo21p feedback mechanism. The existence of such a mechanism has been demonstrated (27), and decreased RNAP II levels or activity in ZBD mutants might trigger this mechanism and lead to an excess of the subunit.
Next, we considered the possibility that ZBD mutations might alter the association of other subunits in the core RNA polymerase. However, we found no significant alterations in the stoichiometry of subunits associated with Rpo21p in a preparation of mutant M1 RNAP II with the exception of subunits Rpb4p and Rpb7p which were missing (Fig. 6). This mutant M1 polymerase was reduced in basal transcriptional activity to 5% of WT RNAP II levels. Since Rpb4p and Rpb7p are not required for this activity (20), these observations demonstrate that mutations in the ZBD can decrease basal activity of the enzyme without altering the presence of any subunits required for this function. The possibility that Rpb4p and Rpb7p are required for basal activity only in the presence of a ZBD mutation seems unlikely since adding a 30-fold excess of these subunits back to the mutant polymerase had no effect on activity.
Studies with E. coli RNA polymerase point to a role for zinc that is closely related to the activity of core polymerase. A double cysteine to serine substitution mutant in the ␤Ј subunit of E. coli RNAP (analogous to the M1 mutant investigated here) was shown to be inactive in vivo and the mutant polymerase to be less processive than the wild-type in vitro (5). In addition, cross-linking studies identified a domain that included the ␤Ј zinc-binding motif along with conserved region I of the ␤ subunit as being associated with a 7-9-nucleotide stretch of double-stranded DNA downstream of the RNAP active center (5). It was shown further that these interactions are responsible for conferring the salt resistance of the elongating ternary complex. The purified E. coli mutant RNAP produces a number of early-terminated transcripts in an in vitro transcription assay. We looked for evidence of a similar effect in the yeast system; however, an extract made from the corresponding M1 mutant in this study showed no signs of early-terminated transcripts in promoter-specific assays ( Fig. 5B and data not shown).
The association of three subunits (Rpc82p, Rpc34p, and Rpc31p) with RNAP III is decreased in the presence of a mutation in the ZBD of Rpo31p (16). These three subunits interact with one another to form an RNAP III subcomplex (46), and mutations in either RPC34 or RPC31 decrease in vitro promoter-directed transcription by RNAP III but not nonspecific transcription activity from poly[d(A-T)] templates (47,48). This in vitro phenotype is reminiscent of RNAP II lacking the Rpb4p-Rpb7p subcomplex; the absence of these subunits decreases RNAP II promoter-directed transcription but not nonspecific transcription (20). This raises the possibility that the Rpc82p-Rpc34p-Rpc31p subcomplex is a functional homolog of Rpb4p/ Rpb7p and that the interaction of either with its RNAP is dependent upon a functional ZBD. Determining whether the Rpb4 -7p subcomplex functionally interacts with the Rpo21p-ZBD may be resolved by further genetic experiments and structural elucidation of RNAP II.