Accurate Initiation by RNA Polymerase II in a Whole Cell Extract from

We have developed a simple procedure for isolating a transcriptional extract from whole yeast cells which obviates the requirement for nuclear isolation. Detec- tion of accurate mRNA initiation by RNA polymerase II in the extract requires the use of a sensitive assay, recently that involves ac- tivation by a GAL4-VP16 fusion protein and a tem- plate lacking guanosine residues in the coding strand. The extract is prepared from fresh or frozen yeast cells by disruption with glass beads and fractionation of proteins by ammonium sulfate precipitation. The (Y- amanitin-sensitive transcripts synthesized in the assay were identical to those produced in a parallel assay using a yeast nuclear extract. The activity of the whole cell extract is lower per mg of protein than a nuclear extract but proportional to the volume of the nucleus relative to the whole cell. The optimal ranges for sev- eral reaction components including template, mono- and divalent cations, and nucleotide substrate concen- tration were determined.

We have developed a simple procedure for isolating a transcriptional extract from whole yeast cells which obviates the requirement for nuclear isolation.
Detection of accurate mRNA initiation by RNA polymerase II in the extract requires the use of a sensitive assay, recently described by Kornberg and co-workers (Lue, N. F., Flanagan, P. M., Sugimoto, K., and Kornberg, R. D. (1989) Science 246, 661-664) that involves activation by a GAL4-VP16 fusion protein and a template lacking guanosine residues in the coding strand. The extract is prepared from fresh or frozen yeast cells by disruption with glass beads and fractionation of proteins by ammonium sulfate precipitation. The (Yamanitin-sensitive transcripts synthesized in the assay were identical to those produced in a parallel assay using a yeast nuclear extract. The activity of the whole cell extract is lower per mg of protein than a nuclear extract but proportional to the volume of the nucleus relative to the whole cell. The optimal ranges for several reaction components including template, monoand divalent cations, and nucleotide substrate concentration were determined. Under optimal conditions the whole cell extract produced a maximum of approximately 1 X lo-' transcripts/template molecule in 30 min. Although it has been clear for many years that the structure of RNA polymerase II has been conserved throughout the eukaryotic kingdom (l), more recent work has demonstrated the conservation of other mRNA transcription factors. Several gene-specific transcriptional activators function both in yeast and animal cells: the yeast GAL4 protein (2, 3); mouse c-fos (4); avian v-jun (5); and rat glucocorticoid receptor (6). General transcription factors from yeast can substitute for the HeLa cell factors TFIIA' (7) and TFIID (8,9)  transcription reactions. This complementation has been used to advantage to circumvent the difficulty of purifying the HeLa cell factors (10, 11). The yeast TFIID equivalent has been highly purified (12, 13), and molecular genetic methods have been applied to clone the yeast gene and map it to a locus known to affect transcription (14)(15)(16). Differences between transcription in yeast and in multicellular eukaryotes, such as in the spacing between the "TATA box" and the transcription initiation site (17-19), exist as well. Elucidation of these similarities and differences by comparative biochemistry of the mammalian and yeast transcription machineries is now possible due to the development of a yeast nuclear extract capable of accurate transcription (20). The levels of transcription in the extract were originally quite low (20), but Kornberg, Ptashne, and their co-workers (21) recently increased the activity of the in vitro reactions by using a GAL4-VP16 fusion protein as a strong activator of a template containing a GALuAs element. This innovation increased the amount of RNA made nearly loo-fold to approximately lo-' transcripts/template molecule (21). In addition, the assay procedure has been simplified by the creation of a template containing the GALu*s upstream of the G-minus cassette originally described by 23).
The remaining difficulty in undertaking a biochemical analysis of yeast transcription lay in the nuclear isolation procedure, which required large scale spheroplasting using lytic enzymes and density gradient fractionation of nuclei. Besides being time consuming, spheroplasting introduces proteases to the extract, and the procedure is strain-and growth stage-dependent.
In addition, the nuclear fractionation is difficult to carry out on the large scale required for protein purification.
We report that prior isolation of nuclei is not necessary to obtain an extract capable of accurate in uitro transcription of mRNA-encoding templates. was diluted with an equal volume of the same buffer, and cells were broken with glass beads in a Dyno-Mill (Glen Mills) with a Neslab cooling unit set at -10 "C. By alternating three 1-min rounds of high speed grinding with 1 min of slow mixing, the temperature of the lysate was kept between 1 and 4 "C. One functional extract was made with an ice-cooled Bead-Beater (Biospec Products), alternating four 30-s bursts of grinding with 1 min of slower mixing; the temperature of the lysate rose to 10 "C. were omitted from a 5-min preincubation at room temperature (approximately 23 "C). Reactions were begun by addition of the remaining components in a 5-~1 volume. After 30 min at 23 "C, transcription was stopped and RNA digested by the addition of 120 ~1 of 10 mM Tris-HCl, pH 7.5, 300 mM NaCl, 5 mM EDTA, 0.05 unit/p1 RNase Tl (Sigma) and a further incubation of 10 min. Sodium dodecyl sulfate was then added to 0.5% and proteinase K to 0.25 mg/ml. After 20 min at 37 "C, the samples were precipitated with isopropyl alcohol, separated by electrophoresis on urea/polyacrylamide gels, and analyzed by autoradiography.

RESULTS
Characterization of the Reaction Products-As demonstrated in Fig. 1, both nuclear and whole cell extracts produced two transcripts in reactions with the pGAL4CG-template. The conditions used for these reactions were essentially those described by Lue et al. (( 23) see legend to Fig. 1) except that the DNA concentration was raised from 3 to 25 rg/ml. In each case, both transcripts were sensitive to the presence of 10 pg/ml Lu-amanitin, demonstrating that they were products of RNA polymerase II. The sizes of the transcripts, measured in this and other experiments as 350 and 375 nucleotides, correspond to those reported by Lue et al. (23) 1425, 821, 516, 506, 396, 344, 298, 245, 221, 220, and 219 nucleotides. nuclear extract, fractions (1 and 5%, Fig. 1) of the nuclear extract reaction were compared with the entire whole cell extract reaction. From these data, measurements of protein concentrations, and calculations of transcription efficiencies, we estimate that the whole cell extract was 25-40-fold less active than the nuclear extract in terms of the number of transcripts per pg of protein.
Optimization of the Reaction Conditions-The original assay conditions were based on those of Lue et al. (23) which were derived from those described by Tyler and Giles (28) for Neurospora whole cell extracts. We have varied several parameters of the assay, the most important being template and extract concentrations, and we have increased the overall activity of the reaction approximately 50-fold relative to the conditions used in Fig. 1. Preliminary tests revealed optima in the vicinity of 20-30 kg/ml template and 3-9 mg/ml protein. In addition, each preparation of the GAL4-VP16 fusion protein had to be titrated in the reactions. Optimal amounts of GAL4-VP16 protein stimulated transcription in the whole cell extract lo-fold, as reported for the nuclear extract (23). The level of stimulation held constant over a further lo-fold increase in GAL4-VP16 fusion protein.
We optimized conditions further using saturating amounts of one fusion protein preparation at approximately 80 ng of protein/reaction and one whole cell extract at 4.7 mg/ml. Template concentration had only a small effect within the range of 20-50 pg/ml (data not shown). Transcription was significantly reduced when template was below 20 pg/ml. In Fig. 2 we show the result of varying the magnesium concentration; the optimal range was between 15 and 19 mM magnesium acetate. This range is higher than that reported for the nuclear extract (20)  to DNase and to 10 pg/ml cr-amanitin (see Fig. 1).
At optimal extract, template, and magnesium concentrations, the potassium concentration exhibited a broad optimum. Tests of various counterions indicated that potassium acetate from 50 to 100 mM and potassium glutamate from 70 to 120 mM were virtually indistinguishable. Potassium chloride above 20 mM inhibited the reaction, and ammonium sulfate could not substitute for a potassium salt (data not shown). In all of the reactions shown in this work, the labeled nucleotide (UTP) was at 1 pM. Analysis of early time points by densitometry of autoradiographs has revealed continued increases in transcription in response to increasing UTP concentrations, with 50 pM UTP (the apparent K,,, for the selective transcription reaction) supporting approximately 30fold higher rates of synthesis than 1 PM UTP (data not shown).
Time Course of the Transcription Reaction-A comparison of reaction kinetics under the original and optimal mono-and divalent cation concentrations is shown in Fig. 3. Incorporation was linear for 15 min and reached a maximum by 20-40 min. The approximately 50-100 attomol of transcripts produced were largely stable for an additional 50 min. A third transcript of approximately 400 nucleotides was also visible in the later time points of Fig. 3A. This transcript was large enough to have traversed the entire G-minus cassette. Appearance of this transcript was variable but sensitive to 10 pg/ml oc-amanitin. Lue et al. (23) report that a transcript of similar size seen in reactions with nuclear extract is suppressed by inclusion of the chain terminator 3'-O-methyl GTP, suggesting readthrough from an upstream initiation site.
Extract Production Is Not Strain-dependent-We originally made a whole cell extract from a WHIl-1 mutant (BF338-9b), reasoning that the reduced cytoplasmic volume of these cells (30) might improve the recovery of transcription factors. We later tested wild-type WHIl+ cells, hoping to produce extracts from currently available mutants without requiring introduction of the WHIl-1 allele. Indeed, extracts made in parallel from the wild type and the mutant were both func-mRNA Transcription by a tional (Fig. 4). Although we observed stronger signals from extracts of a protease-deficient pep4 strain, extracts from a PEP4+ strain were also functional (Fig. 4), and we concluded that there was no absolute strain dependence in our ability to produce active extracts. DISCUSSION Using sensitive assays for transcription, it is possible to demonstrate initiation in vitro by RNA polymerase II in a yeast nuclear extract (20,21,23). We have found that extracts from whole yeast cells will also support accurate mRNA synthesis. The extracts are understandably less active per milligram of protein than nuclear extracts since the transcription apparatus is diluted by proteins from the cytoplasm, but the level of activity (approximately 3% that of the nuclear extract) is in proportion to the ratio of nuclear to total cell volume (approximately 5% (31)). The production of these extracts is not dependent on a particular yeast strain, and it can be carried out with apparatus readily available in most laboratories. Two particularly practical features include the ability to use stored cells and the fact that the final extract is stable to repeated cycles of freezing.
The optimal conditions for transcription are different for the whole cell extract relative to the reported conditions for the nuclear extract, which probably reflects the presence of cytoplasmic proteins and RNA. In particular, the optimal template concentration is nearly lo-fold higher than that seen for the nuclear extract (21,23). The high template concentration may be required to titrate nonspecific DNA binding proteins present in the extract. Since the template concentrations are lo-fold higher, the ratio of transcript to template is proportionally lower at a given extract concentration. Despite this numerical problem, we have found that under optimal conditions (including higher levels of the labeled nucleotide substrate), the whole cell extract will produce 10V2 transcripts/ template molecule, nearly identical to the value reported for the nuclear extract (21).
The ability to readily scale up the isolation procedure and the availability of some highly purified yeast transcription factors including RNA polymerase II (l), the TATA box binding factor (12, 13), and the GAL4 and GAL80 transcriptional regulators (32) sets the stage for a detailed biochemical and genetic characterization of eukaryotic mRNA synthesis.