Specific Transcription of Homologous Class I11 Genes in Yeast-soluble Cell-free Extracts*

Cell-free extracts prepared from whole yeast cells carry out selective and accurate transcription, in vitro, of purified yeast class III genes. Both 6 S rRNA and tRNA genes are specifically transcribed by DNA-de-pendent RNA polymerase III present in these whole cell extracts. These extracts also appear to carry out nucleolytic processing of the in uitm synthesized transcripts. Optimal conditions for specific class m gene transcription in vitro are defined. Initial fractionation of the yeast extract has indicated that multiple chro-matographically separable factors (fractions) are re- quired, in addition to RNA polymerase m, for specific in vitro transcription of class III genes. "genetics," regulatory sequence class I11 genes transcribed in DNA-dependent RNA polymerase S rRNA tRNA internal control initiation 5'-region genes. Fractionation soluble Xenopus ovary 37,000) short control

in vitro transcription of class III genes.
The utility of soluble in vitro transcription systems for the study of eucaryotic transcription has recently been amply documented (reviewed in Ref. 1). When in vitro transcription studies were coupled with in vitro "genetics," very surprising results were uncovered concerning the regulatory sequence elements of class I11 genes (genes transcribed in vivo by DNAdependent RNA polymerase 111). 5 S rRNA (2-4), tRNA (5-8), and adenovirus virus-associated RNAl (9,10) genes all appear to have internal control regions which serve to direct transcription initiation at the 5'-region of these genes. Fractionation of soluble Xenopus laevis ovary extracts has led to the identification and purification of a protein (Mr = 37,000) (11) which specifically interacts with the short intragenic control region of the homologous Xenopus 5 S gene (11)(12)(13)(14). This protein, termed TFIIIA in concert with Xenopus RNA polymerase I11 and (at least) two other transcription factors are both necessary and sufficient to reconstitute specific transcription of the Xenopus 5 S gene in vitro. It was our goal to conduct similar experiments with class I11 gene systems in the lower eucaryote, Saccharomyces cerevisiae, to exploit the unique biochemical and genetic advantages of this organism. In this report, we describe the preparation and properties of a totally soluble transcription system derived from S. cereuisiae. We also describe preliminary chromatographic fractionation studies of these extracts and present data which suggests that there are multiple distinct class I11 transcription factors required for tDNA and 5 S DNA transcription. Recently, very similar soluble transcription systems have been described.' Growth of Cells and Preparation of Cell-free Extracts-Cells were grown in YEPD media (1% w/v yeast extract, 2% w/v peptone, 2% w/ v dextrose) to a density measured by absorbance at 650 nm of 5-10. Cells were harvested by centrifugation, washed twice with distilled H20, and resuspended in solubilization buffer (200 mM Tris-C1 (pH &I), 10% v/v glycerol, IO mM MgC12, IO mM &mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride) at 2 ml of buffer/g of cells, wet weight. The cell slurry was either used directly for extract preparation or was stored at -80 "C. The cell suspension (-250 ml) was placed in the chamber of a Bead Beater cell disruptor (Biospec Products, Bartlesville, OK). To the chamber was added an approximately equal volume of 0.445-mm diameter acid-washed glass beads. Sufficient beads were added to exclude all air from the chamber. The unit was then assembled and placed in an ice jacket containing a CaCb-ice slurry. Cells were lysed by homogenization for 4-5 min total time, in 30-s bursts with 2-3 min cooling time between bursts. The lysate was decanted from the glass beads, and phenylmethylsulfonyl fluoride and P-mercaptoethanol were each re-added to 1 and 10 mM final concentrations, respectively. The lysate was made 0.4 M in ammonium sulfate by the addition of 4 M ammonium sulfate (pH 7.9). The extract was allowed to sit on ice for 10-15 min before centrifugation at 1 0 0 , O O O X g for 60 min. The proteins in the supernatant from this centrifugation (typically a volume of 150-175 ml) were precipitated with ammonium sulfate (0.35 g/ml) and harvested by centrifugation. The precipitated proteins were dissolved in a minimal volume (15-20 m l ) of Buffer C (20 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (pH 7.9), 20% v/v glycerol, 0.2 mM EDTA, 10 mM P-mercaptoethanol, 0.1 mM phenylmethylsulfonyl fluoride) and dialyzed 8-10 h against 100-200 volumes of Buffer C containing 100 mM NaCl. This concentrated extract (average yield -25 ml/100 g of cells, 20-30 mg of protein/ml) was divided into aliquots and stored at -80 "C. Occasionally, residual nucleic acids were removed by passing the extracts through a DEAE-cellulose column at high salt (0.3 M ammonium sulfate) in Buffer C.
RNA Polymerase Purification-RNA polymerase 111 was purified through the ion exchange chromatography steps in the procedure of Valenmela et al. (19).
Fractionation of Extracts-Extracts were absorbed to phosphocellulose (Whatman P-11) columns (10-15 mg of protein/ml of bed volume) equilibrated with Buffer C + 100 mM NaC1. Bound proteins were step-eluted with Buffer C containing 600 and 1500 mM NaC1. Absorbance was monitored at 280 nm throughout the procedure. Fractions containing protein were pooled, and the samples were desalted into Buffer D (Buffer C + 10 mM MgC12) + 100 mM NaC1.
Following desalting, the phosphocellulose 600 mM NaCl step fraction, which contained both transcription factors and endogenous RNA polymerase 111, was applied to a DEAE-Sephadex A-25 (Pharmacia) column (1-2 mg of protein/ml of bed volume) previously equilibrated with Buffer D + 100 mM NaC1. Bound proteins were step-eluted with Buffer D containing 250 and 1000 mM NaCl. Protein concentration was monitored by absorbance, and appropriate fractions were pooled and desalted prior to assaying for transcription factors. In all the chromatographic steps, fractions equivalent to IO-15% of the column volume were collected. Samples were desalted either by dialysis (8-10 h) or gel filtration on columns of Sephadex G-25 (medium). Following the desalting step, appropriate column fractions were stored at -80 "C in small aliquots.
Transcription Assays-Assays were conducted in 50-pl volumes, of which up to 60% of the assay volume was contributed by the extract or column fractionated material. The standard assay contained 12 mM Hepes' (pH 7.9), 150 mM NaCl, 10 mM MgCL, 12% (v/v) glycerol, 6 mM /3-mercaptoethanol, 600 p~ each of ATP, CTP, and UTP, 25 p~ [LY-~*P]GTP (10 Ci/mmol), and 25 pg/ml of supercoiled plasmid DNA. When column fractions were assayed, the reactions were supplemented with 20-50 units (as defined in Ref. 18) of yeast RNA polymerase 111. Reactions were incubated 30-60 min at 20 "C. Transcription was terminated, and RNAs were pugled, fractionated on native or denaturing (20) polyacrylamide gels, and detected by autoradiography as described (18,21). Fingerprint and Oligonucleotide Analysis-"P-labeled RNAs were synthesized in vitro in scaled-up reactions (500 pl) containing the a-32P-labeled NTP at 100-200 Ci/mmol. Labeled RNAs were resolved on denaturing polyacrylamide gels and localized by autoradiography of the wet slabs. Discrete 32P-RNAs were eluted from the appropriate gel slices (22) and digested either with pancreatic RNase A or RNase TI. The oligonucleotides in the digests were subjected to two-dimensional fractionation by ionophoresis on cellulose acetate strips (2 h X 3000 V), followed by homochromatography on DEAEcellulose thin layer plates (23), using homomix C-15 of Brownlee (24). Double-digestion and nearest-neighbor analyses on oligonucleotides eluted from the thin layer plates were performed as described by Volckaert and Fiers (25,26). End-group analyses were carried out following published procedures (27,28), using commercial unlabeled polyphosphate (guanosine and adenosine) markers. Some [a-"PI GTP-labeled TI-oligonucleotides were directly analyzed using the partial enzymatic degradation methodology of Randerath et al. (29). Yeast 5 S rRNA (either unlabeled or labeled in vivo with '*P) was isolated and gel purified as described by Rubin (30). This 5 S [32P] RNA was subjected to RNA-fingerprint sequence analyses as described above for the in vitro transcripts.

RESULTS
On the assumption that a nuclear extract would be enriched for all the components required for directing selective transcription, we initially used isolated yeast nuclei to prepare a soluble in vitro transcription system. Purified nuclei (31) were lysed, and proteins were solubilized by sonication in the presence of 0.4 M ammonium sulfate. The lysate was subjected to high speed centrifugation and the proteins in the supernatant were concentrated by ammonium sulfate precipitation. These concentrated nuclear extracts (10-20 mg of protein/&) directed the transcription of discrete low molecular weight RNAs when programmed with either tRNA or 5 S rRNA genes.3 However, because total yields of extract were quite low and the method for nuclei preparation was rather tedious and time-consuming, and since our ultimate goal was to fi-actionate the extracts, we explored alternate methods of extract preparation. Whole cell extracts were prepared under conditions where significant amounts of RNA polymerases I1 and I11 would be solubilized (32). Cells were lysed in an RNA polymerase solubilization buffer (33), modified as described by Dezelee et al. (34) to decrease proteolysis of polymerases. Insoluble material and ribosomes were removed by high speed centrifugation, and solubilized proteins were concentrated by ammonium sulfate precipitation. The resultant extracts were very concentrated (20-30 mg of protein/ml) and contained significant amounts of RNA polymerase activity (-50% of the total cellular RNA polymerase, c/ Valenzuela et a2. (19)). These whole cell extracts proved to be much more active than the nuclear extracts in directing the synthesis of discrete-sized low molecular weight RNAs in response to purified 5 S and tRNA genes. Additionally, total yields of extract were much greater. We therefore proceeded to characterize the in vitro synthesized RNAs in more detail. These whole cell extracts were used in all the studies to be described below.
Synthesis of Discrete RNAs by RNA Polymerase 111 in Vitro-The two homologous templates that we have used in these studies are the yeast 5 S rRNA gene (in the plasmid pSc90) and the serine-ochre suppressor tRNA gene, tRNA?LA (in the plasmid pPml6). Yeast 5 S rRNA is a very abundant and well characterized molecule. The nucleotide sequence of both the RNA (35,36) and the gene coding for it (37,38) have been determined. Mature 5 S rRNA is 121 nucleotides long and contains a polyphosphate terminus at its 5'-end. The structure of the primary transcription product of the 5 S rRNA gene is not entirely certain since molecules containing 3'-extensions of 6-8 nucleotides have been reported in vivo (39)(40)(41) in yeast and mammalian in vitro systems (42,43). However, no molecules larger than these have ever been observed in vivo or in vitro. Additionally in vitro mapping experiments (2,3,44) have defined the 5 S rRNA gene quite precisely to a region 1130 nucleotides in length.
The serine ochre suppressor tRNAs?LA is present in vivo in only trace amounts. The sequence of the mature tRNA has been determined, and including the -CCA terminus which is added post-transcriptionally, it is 86 bases long (ck Ref. 17 and references therein). Putative precursor transcripts of this gene 102 and 120 nucleotides long have been detected in vivo (45). The primary transcript of this gene has not yet been characterized though. Approximately 200 base pairs of the plasmid pPml6, which contains tDNAFLA, has been sequenced. The sequence includes the tRNA coding region and 5'-and 3"flanking sequences. The nucleotide sequence indicates that this tRNA gene does not contain an intervening sequence. Hence, transcription of this gene in vitro should result in a fairly simple transcript pattern, and this is one of the reasons that this gene was chosen for detailed studies. Fig,  1 shows an autoradiogram of a gel which displays the discrete low molecular weight RNAs synthesized in the yeast soluble whole cell extracts in response to the addition of purified DNA templates containing a variety of class I11 genes (lanes 1, 5, 10, 11, and 12). The synthesis of these RNAs is both DNA-dependent and actinomycin D-sensitive (Fig. I , Lanes  4, 8, and 9). Consistent with transcription being mediated by DNA-dependent RNA polymerase 111, RNA synthesis is not inhibited by low (50 pg/ml, Lanes 2 and 6) or high (1000 pg/ ml, Lanes 3 and 7) concentrations of a-amanitin. These concentrations of amanitin inhibit yeast RNA polymerases I1 and I, respectively, but have no effect on enzyme I11 (46, 47). This resistance to a-amanitin has been observed for both soluble and endogenous nuclear/chromatin bound polymerase I11 (22,31).
Three RNA species of approximately 95,90, and 85 bases in length are observed when transcription is carried out in the presence of the serine tRNA gene template (pPml6) (Fig. 1, Lanes I, 2, and 3; see also Figs. 3A, 4B, and 5A). These RNA species appear to be derived from a common precursor which does not accumulate to any appreciable extent in the in vitro transcription reaction (see below). In contrast to this gene, transcription of the pSup4 tyrosine tRNA gene, which contains an intervening sequence, leads to a much more complex  Refs. 48,49).
We observe the synthesis of only one major RNA species -120 bases in length ( Fig. 1, Lane 5) when a yeast 5 S rRNA gene is used to program RNA synthesis by the whole cell extracts. This species is the major 32P-RNA synthesized either at early (10 min) or late (30 min or longer) times of transcription (not shown). In addition, this RNA comigrates on a thin sequencing gel with in vivo 5 S RNA.4 This result is somewhat in contrast to what is observed in other class I11 transcription systems (42,43) where fairly large amounts of apparent precursor RNAs (3"extended forms with 6-8 additional residues) accumulate. In general, this is the 5 S DNA transcription profile seen with crude extracts (but see below and Fig. 4). Also shown in Fig. 1 are the RNAs synthesized by yeast extracts when transcription is programmed with heterologous DNA templates: X . laevis methionyl tRNA gene (Lane IO) and adenovirus-2 virus-associated RNA genes (Lane 12). These class I11 genes direct the synthesis of multiple discrete RNA species which co-migrate (on both denaturing and these nondenaturing gels) with the transcripts synthesized from the corresponding templates in KB cell S-100 extracts (18) and data not shown.' Fingerprint Analyses of in Vitro Transcripts-In order to confirm that the discrete-sized in vitro synthesized RNAs result from accurate transcription, we characterized the transcripts of the tRNAsr and 5 S rRNA genes in more detail by fingerprint analyses. Transcripts labeled in vitro via synthesis with either [a-"PIGTP, ATP, or UTP were resolved in denaturing polyacrylamide gels and purified as described under "Experimental Procedures." Gel-purified RNAs were digested with either RNase TI or pancreatic RNase A. The resulting oligonucleotides were fractionated by standard two-dimensional separation methodologies. Oligonucleotides were eluted from the chromatograms and further characterized by nearest-neighbor and double-digestion techniques.
' M. S. Klekamp and P. A. Weil, unpublished observations. Fig. 2A represents the autoradiogram of the gel fractionation of RNA synthesized in vitro from the 5 S rRNA gene with [a-"PIGTP as the labeled nucleotide. The RNA species marked with an arrow was eluted from this preparative gel and analyzed as described above. The TI oligonucleotide pattern is shown in Fig. 2B. For comparison, the TI oligonucleotide pattern of in vivo "'P-labeled 5 S rRNA is also shown (Fig. 2C). The arrows point to the only major difference between the fingerprint patterns of the in vivo and in vitro synthesized RNAs, spot 6 ( Fig. 2, compare B and C). This oligonucleotide represents the 3'-end of the mature 5 S rRNA and has the sequence C~A~A~U~C~U~H (35). It would not be present in a fingerprint of 5 S RNA labeled in vitro with [a-32P]GTP. However, this oligonucleotide is present on T I fingerprints of [a-32P]ATP-and UTP-labeled RNAs (Table I).
Consistent with the known 5'-end sequence (pppGpGp . . .) of 5 S rRNA, 5'-end group analyses of 5 S rRNA synthesized in vitro in the presence of [cx-~'P]GTP identified only "P-labeled pppGp, ppGp, and pGp. No other labeled polyphosphates were detectable: These data, in combination with the fingerprint-sequence data of Table I, indicate that the RNA synthesized by the endogenous RNA polymerase I11 in the yeast extracts clearly represents a properly initiated and (probably) terminated 5 S rRNA molecule. We cannot rule out, however, the possibility of very rapid nucleolytic processing of 3'-extended forms of 5 S RNA (cf Refs. [39][40][41][42][43]. The 32P-labeled products of a preparative in vitro transcription of the serine tRNA gene are shown in Fig. 3A. This transcription reaction was conducted for only 10 min. Three major species of RNA, labeled I (-95 nucleotides), I1 (-90 nucleotides), and I11 (-80 nucleotides) are observed in roughly equal amounts. RNAs I and I11 were analyzed in more detail.
They were eluted from the preparative gel and digested with either pancreatic or TI RNases. The TI oligonucleotide patterns of these RNAs (Fig. 3, B and D, respectively) are identical except for the (almost) total absence of a large TI oligonucleotide (spot 14; see Table I1 for oligonucleotide compositions indicated by the arrow in Fig. 30) from band I11 RNA. This suggests that band I11 is derived by processing of band I RNA. The kinetics of appearance of these RNAs is consistent with this idea (see below). All oligonucleotides were eluted and analyzed by nearest-neighbor and double-digestion techniques. Additionally, the larger [a-"PIGTP-labeled TI oligonucleotides (spots [10][11][12][13][14][15][16] were analyzed using partial enzymatic digestions coupled with sizing analyses using polyethyleneimine thin layer chromatography (29): The results of these analyses are summarized in Table 11. Comparison of the data obtained for band I11 RNA with the known sequence for the mature tRNA indicates that this RNA corresponds to the fully processed mature-length, in vivo transcript. Oligonucleotides from the 5'-end (spot I3), anticodon (spot 16 32P-labeled transcripts were then incubated in vitro with the yeast cell-free extract under standard synthesis conditions. No discrete RNAs were generated: These results clearly indicate that neither the discrete serine-tRNA gene transcripts nor the 5 S rRNA transcripts observed above are generated via specific nucleolytic processing of random transcripts. Taken together, all these results indicate that these two homologous class I11 genes are transcribed specifically in vitro by endogenous DNA-dependent RNA polymerase I11 present in the yeast soluble extracts to produce discrete RNAs. These RNAs are not generated via specific processing of random transcripts but rather are the result of accurate and specific initiation, elongation, and termination reactions conducted by the enzymes and factors (see below) present in these extracts. This is the first report of specific transcription of yeast class I11 genes in vitro in a totally soluble homologous transcription system.

Properties of the in Vitro Transcription
System-Extracts prepared as described above are quite stable when stored at -80 "C ( f u l l activity retained for greater than 6 months). They can survive several cycles of freeze-thawing with only small decreases in activity. However, storage in small aliquots is still recommended. We have examined the optimal conditions for specific transcription in vitro with the yeast soluble extract. When a crude extract is utilized for transcription, the optimal temperature is 20-25 "C (greater than at 15, 30, or 37 "C).
However, if extract passed through DEAE-cellulose at high salt (see "Experimental Procedures") or column-fractionated extract (see below) is used for transcription, the optimal In Vitro Transcription of Class 111 Genes

S DNA in vitro transcripts
Nucleotide sequences were deduced from the following data: mobility in fvst and second dimension of fingerprint analyses, molar ratios of spots, nearest-neighbor analyses, and known RNA fingerprints and DNA sequences.  I, 11, and 111. These three temperature is 30 "C, the temperature for both optimal growth of yeast cells and optimal activity of the purified enzymes (19). This is presumably due to removal of endogenous transcription inhibitors or nucleases. Preincubation of extract and DNA with or without rNTPs had little or no effect on either the rate or extent of specific transcription. Neither did the addition of an NTP-regenerating system. Optimal ionic conditions for specific transcription of all class I11 templates RNAs were eluted from the gel and subjected to RNase TI digestion, and the resulting oligonucleotides were separated in two dimensions.

Pancreatic RNase A end products from RNA labeled with T, oligonucleotide" [u-"P]CTP [a-"P]ATP
The fingerprints of RNAs I, 11, and 111 are shown in B, C, and D, respectively. The arrow in D marks the expected position of oligonucleotide 14 (discussed in text), which is missing in the TI fingerprint pattern of RNA 111. B indicates the mobility of the blue dye. First and second dimensions were run as indicated in Fig. 2. tested were similar, 120-150 mM NaCl, 10-15 mM MgC12. These optima are consistently higher than those observed for class I11 genes in other eucaryotic soluble transcription systems (18) and appear to be an inherent property of the yeast extracts. The extracts do not exhibit the strikingly sharp optima with regard to DNA concentration as seen in other cell-free systems. DNA concentration optima were quite broad, and even at higher concentrations of DNA (400-500

In Vitro Transcription
of Class 111 Genes a437 TABLE I1 Sequence of RNase TI oligonucleotides of tDNA8A in vitro transcripts Nucleotide sequences were deduced from various pieces of data: mobility of TI oligonucleotides in first and second dimensions of the fingerprint analyses, molar ratios of eluted oligonucleotides, nearest-neighbor analyses, partial enzymatic digestion of terminally ([Ix-~*P]GTP) labeled TI oligonucleotides. known DNA and RNA sequence information (cf. Olson et al. (17)). ' Predicted from known DNA/RNA sequence; bases in parentheses represent the nucleotide 3' to the indicated TI oligonucleotide.

Bands
Inconsistent (oligo)nucleotides, either present as extra oligonucleotides or missing from double digestion analysis. pg/ml), a fairly strong signal was still detectable. Specific transcription occurs linearly, with little or no lag, for approximately 30-45 min, at which time a steady state between synthesis and random degradation is attained. Under optimal conditions these extracts synthesize from 0.1-10 transcripts/ gene/h. The rate of specac transcription is variable, depending on the extract, but approaches that seen in other systems. Pulse-chase experiments indicate a half-life of -30 min for RNAs in the extract. The apparent processing of the tRNA gene band I transcript to the band I11 RNA occurs fairly rapidly; within 5-10 min, about 20-30% is processed. We have not yet been able to determine conditions (using whole cell extracts) where we could totally dissociate transcription from RNA processing. The most effective procedure to date has been to carry out transcription reactions in the presence of polyuridylic acid (100 pg/ml) (49). Under these conditions, RNA processing is 35-50% inhibited4, while transcription was essentially unaffected (see also Fig. 5A and discussion below). However, we still do not observe the accumulation of the presumed precursor to band I RNA under these conditions. Chromatographic Fractionation of Soluble Extracts-Mammalian and amphibian soluble class I11 transcription systems have been shown to contain several distinct factors involved in directing selective transcription (11,54 and references therein). For comparison, we have subjected the soluble yeast extract to ion exchange chromatography. Crude extracts were fractionated on columns of phosphocellulose (P-11) and DEAE-Sephadex by salt-step elution, as outlined in the flow diagram of Fig. 4A. Using phosphocellulose chromatography, three fractions were obtained by the following procedure: A BT fraction, a 600 mM NaCl step-eluted fraction, and a 1500 mM NaCl step-eluted fraction. Each of these fractions were then assayed alone and in combinations to see if specific class I11 gene transcription could be reconstituted (Fig. 4, B and C). The 600 mM step fraction alone was able to support the specific transcription of both tRNA and 5 S RNA genes (Fig.  4, B and C, Lanes 3). Both 5 S DNA and tDNA transcription reconstitute with high recovery (compare Lanes 1 and 3 in Fig. 4, B and 0. Two other class I11 templates, the tyrosine tRNA and adenovirus virus-associated RNA genes, are also transcribed by the components present in the 600 mM step fraction: RNA processing activity also appears to be present in this 600 mM phosphocellulose step fraction since maturesized serine tRNA transcripts (Fig. 4B) and partially processed tyrosine tRNA transcripts4 accumulate in these reactions. Transcription is severely inhibited when the P-11 BT fraction is added to the 600 mM fraction (Fig. 4, B and C, Lane 5). The BT fraction has no qualitative effects on the specificity of transcription, even when very small amounts of this fraction are added to large amounts of the P-11 600 mM fra~tion.~ As discussed above, 3'-extended forms of 5 S rRNA have been observed both in vivo and in vitro in higher eucaryotes and yeasts. We generally do not observe significant amounts of a 5 S RNA precursor accumulating in vitro when whole cell extracts are used for transcription. However, it is obvious from the autoradiogram shown in Fig. 4C (compare Lanes 1 and 3) that "unprocessed 5 S rRNA" apparently does accumulate when P-11 fractionated material is used for in vitro 5 S DNA transcription. This could possibly be due to the removal of inhibitory substances or concentration of a processing activity itself by P-11 chromatography or, alternatively, it could be due to an increase in the amount of incorrect chain termination (Le. read-through). This longer RNA species has not been characterized further.
A l l the assays depicted in Fig. 4, B and C, were conducted without the addition of exogenous RNA polymerase I11 (although the results are identical if exogenous enzyme is added to all the reactions4). Hence, it cannot be determined whether or not the phenomenon of specific transcription observed here is simply due to an "holoenzyme-like" form of RNA polym- erase I11 present in these extracts or due to transcription factors acting in concert with endogenous enzyme 111. We thus sought a procedure for fractionating the P-11600 mM fraction further in order to separate the putative transcription factors from the endogenous RNA polymerase 111. This was accomplished by chromatography of the P-11 fraction on DEAE-Sephadex as outlined in Fig. 4A. DEAE-chromatography yielded three fractions: a BT fraction, a 250 mM NaCl stepeluted fraction, and a lo00 mM NaCl step-eluted fraction. All of these fractions were assayed alone and in combinations to determine whether or not specific 5 S DNA and tDNA tran-NaCl eluted fraction was applied to an -20-ml DEAE-Sephadex column (2.6 X 3.6 cm) that was preequilibrated with Buffer D + 100 mM NaCI. Unabsorbed proteins were eluted with this same buffer. RNAs were analyzed in parallel on 12% acrylamide gels and exposed to the same sheet of x-ray film. Therefore, quantitative comparisons of band intensities can be made between all the lanes. scription could be reconstituted. In this set of assays, all reactions were supplemented with exogenous RNA polymerase I11 (with two exceptions, Fig. 4, D and E, Lane 9).

In Vitro
None of the DEAE-fractions alone are able to support the specific transcription of either tDNA or 5 S DNA (Fig. 4, D and E, . In contrast, the combination of DEAE-BT + 250 mM fractions reconstitutes specific transcription of both of these class I11 gene templates with reasonable recovery (Fig. 4, D and E, Lane 6). This reconstitution of specific transcription is absolutely dependent upon the addition of exogenous RNA polymerase I11 (not shown). The endogenous enzyme I11 binds very tightly to DEAE-Sephadex and is eluted in the 1OOO mM step fraction. This endogenous RNA polymerase I11 can also be used to reconstitute (albeit very poorly) specific transcription effected by the DEAE-BT + 250 m fractions, as shown in Lune 9 of Fig. 4, D and E. Simply adding exogenous enzyme I11 to this combination of components greatly stimulates the amount of specific transcripts (compare Lunes 9 and 10 in Fig. 4, D and E ) . These results imply that there are minimally three components required to effect specific transcription of yeast class I11 genes in uitro: DNA-dependent RNA polymerase I11 and (at least) two chromatographically separable factors (fractions).
Brown and Roeder and their co-workers (11-13, 55) have demonstrated the existence of gene-specific class I11 transcription factors. One factor termed TFIIIA has been shown to bind specifically to the 5 S DNA internal control region. In addition, this protein will bind specifically to the transcript of the gene 5 S rRNA (11-13, 55, 56). In fact, it has been postulated that this dual binding activity (DNA and RNA) could be involved in regulating 5 S RNA gene transcription in uiuo. Regardless of the extract mechanisms and role of the nucleic acid binding, this potential property of a 5 S DNA specific transcription factor is quite easily assayed for. If a TFIIIA equivalent exists in the yeast cell-free transcription extracts, then addition of increasing amounts of 5 S rRNA should selectively inhibit specific 5 S DNA transcription but not tDNA transcription. This experiment was conducted with our whole cell extracts and the results are shown in the autoradiogram depicted in Fig. 5A. tDNASer (Lunes 1-6) and 5 S DNA (Lunes [7][8][9][10][11][12][13] were transcribed in uitro in the presence of increasing concentrations of purified yeast 5 S rRNA (10-10,OOO ng of 5 S rRNA added/reaction). This represents a molar ratio of 5 S rRNA molecules of 5 S DNA molecules ranging from -0.7 to 700. Specific transcription of the 5 S rRNA gene is extremely sensitive to inhibition by 5 S RNA (Lunes 7-13), while tDNA transcription is totally unaffected (Lunes 1-6) over the concentration range tested. We have occasionally seen a slight increase in the amount of tDNA transcription in response to the addition of 5 S rRNA? Processing of tDNA transcripts is slightly inhibited by 5 S rRNA addition. This is consistent with results described above where polyuridylic acid inhibited tRNA processing. The results of this experiment are presented in quantitative fashion in Fig. 5B. Radioactive bands were excised from the dried gel and counted. The amount of specific transcripts as a function of 5 S rRNA concentration is plotted relative to control reactions, which contained no added 5 S rRNA. As seen in the autoradiogram, specific tDNA transcription is insensitive to 5 S rRNA addition, while 5 S DNA transcription is 50% inhibited at a level of -45 ng/reaction. This represents a molar ratio of 5 S rRNA to 5 S DNA "promoter" of about 3. We estimate that there is <5 ng of 5 S rRNA added to these reactions by the whole cell extract. Addition of yeast tRNA (at 10 pg/reaction) to tDNA or 5 S DNA transcription reactions has no effect on the amount of specific tran~cription.~ Thus, the 5 S rRNA-inhibition effect observed is quite specific and suggests that there does indeed seem to be a TFIIIA-like 5 S DNA-specific transcription factor in S. cereuisiue. Hence, there appears to be at least two (chromatographically) distinct transcription factors present in our yeast extracts. One of these factors is apparently tDNA-specific, while the other factor is 5 S DNA-specific. This notion is corroborated by a series of preliminary experiments which indicate that addition of DEAE-Sephadex BT fraction factor, but not DEAE-Sephadex 250 m step fraction factor (cfi Fig. 4), will relieve the inhibition of 5 S DNA transcription by 5 S rRNA? This suggests that the putative TFIIIA-like molecule is in the DEAE-BT fraction.
The fractionation patterns which we describe for yeast transcription factors are similar to the results of others obtained during the fractionation of transcription extracts from higher eucaryotes. Engelke et ul. (1 1) described the purification of TFIIIA from Xenopus ovary tissue. In this system, TFIIIA (one of the 5 S-DNA specific factors) binds quite tightly to cation exchange columns such as phosphocellulose.

In Vitro Transcriptio
Segall et al. (54) show that when S-100 extracts of human KB cells are applied directly to phosphocellulose at low salt, a 5 S DNA specific transcription factor (presumably TFIIIA-like) is found in the BT fraction. However, if nucleic acids are rigorously removed from this phosphocellulose BT fraction containing the KB 5 S DNA-specific factor, it will bind quite tightly to phospho~ellulose.~ We have occasionally observed this anomolous chromatographic behavior on phosphocellulose with the yeast 5 S DNA specific factor. We presume that this was due to unusually large amounts of nucleic acids in those particular extracts. The fractionation pattern described above in Fig. 4 is the only reproducible fractionation scheme we currently observe.

DISCUSSION
We have described the preparation and characteristics of a totally soluble transcription system derived from yeast. These extracts specifically transcribe both homologous and (probably) heterologous class I11 genes. This is the f i t description of such an in uitro DNA-dependent soluble transcription system derived from yeast. These extracts also contain enzymes that accurately process precursor tRNAs. In addition, we have presented preliminary chromatographic fractionation and transcription inhibition studies which demonstrate multiple distinct transcription factors (chromatographic fractions) that are required in addition to DNA-dependent RNA polymerase I11 for class I11 gene transcription in uitro. There are apparently different factors for 5 S DNA and tDNA transcription. All of the properties of the soluble yeast extracts are amazingly similar to other previously described soluble eucaryotic transcription systems. This perhaps could not have been predicted with certainty, especially for class I11 genes ( i e . tRNA and 5 S RNA), since the organization of the S. cereuisiae 5 S rRNA genes (interspersed with the large rRNA gene in an -9-kilobase repeating unit (38)) is so very different from that found in higher organisms.
In comparison to higher eucaryotes, yeasts offer many advantages for biochemical and genetic studies. Since many aspects of eucaryotic class I11 gene structure, transcription, metabolism, and function are highly conserved, we feel confident that detailed studies on class I11 gene transcription can profitably be pursued in this system. Forthcoming results utilizing soluble yeast transcription systems should therefore provide valuable insights into the mechanisms of regulated eucaryotic gene expression.

In Vitro Transcription
of Class 111 Genes