Accurate and specific polyadenylation of mRNA precursors in a soluble whole-cell lysate

Cell, Vol. 33, 595-605.

June 1983. CopyrIght

0092-8674/83/060595-i

0 1983 by MIT

1 $02.00/O

Accurate and Specific Polyadenylation of mRNA Precursors in a Soluble Whole-Cell Lysate James L. Manley Department of Biology Columbia University New York, New York 10027

Summary Conditions have been developed which allow for the efficient, accurate, and specific polyadenylation of exogenously added mRNA precursors in a wholecell lysate derived from HeLa cells. Precursors are prepared by in vitro transcription using linear DNA templates in a HeLa whole-cell lysate, purified, and added to another lysate. Under optimal conditions (which are quite precise with respect to several variables), 70% or more of the precursor molecules become polyadenylated, and the length of the poly(A) segment added is controlled much as it is in vivo, giving rise to 150-300~nucleotide long stretches of poly(A). Under suboptimal conditions, both the fraction of precursor RNA that becomes polyadenylated and also the length of the poly(A) segment added are reduced. The in vitro polyadenylation reaction is also remarkably specific: Only in vitro-synthesized pre-mRNAs that contain a 3’ end located at or slightly downstream from the corresponding in vivo mRNA 3’ end site can be efficiently polyadenylated in vitro. These results suggest that the poly(A) polymerase requires one or more protein (or RNA) factors in order to bring about accurate and specific polyadenylation in vitro, and that the poly(A) polymerase complex must interact with a specific recognition signal at the 3’ end of mRNA precursors in order to catalyze subsequent polyadenylation. Introduction In vitro transcription systems that are derived from mammalian cells and are capable of accurately initiating transcription of mRNA encoding genes (Weil et a/., 1979; Manley et al., 1980) have proven useful in a variety of different types of studies. For example, these systems have been invaluable to investigations of the structure and function of RNA polymerase II promoter sites, the mechanism of transcription initiation, and the regulation of gene expression at the level of transcription initiation (see for reviews: Manley, 1982; Heintz and Roeder, 1982; Manley, 1983). In addition, such cell lysates are proving useful as a source of starting material for the identification and ultimate purification of the factors required to bring about accurate and specific transcription initiation by RNA polymerase II (Matsui et al., 1980; Tsai et al., 1981; Dynan and Tjian, 1981; Samuels et al., 1982). Soluble, DNA-dependent transcription systems have to date been much less useful for studying the remaining reactions that are required to synthesize a mature mRNA in viva. Thus, although readily detectable levels of transcription initiation and very efficient capping and methyla-

tion occur, none (Handa et al., 1981; Cepko et al., 1981) or only very low levels (Weingartner and Keller, 1981; Kole and Weissman, 1982) of RNA splicing have been detected. No evidence that the formation of mRNA 3’ ends and/or polyadenylation can occur in these systems has been reported. Most attempts to detect mRNA processing in soluble cell lysates have demanded that transcription and processing occur simultaneously in the same reaction mixture (although see Goldenberg and Raskas, 1981). Thus, a typical approach is that a cloned mammalian gene is added to a HeLa whole-cell lysate under conditions that are required to obtain accurate transcription initiation, and the RNA synthesized is assayed to determine if processing has occurred. However, there are several inherent difficulties with this approach. First, if mRNA processing requires different experimental conditions than does transcription initiation, it may be that processing does not occur efficiently in vitro simply because conditions that allows transcription initiation to occur must be maintained. An example of this problem is that the RNA polymerase III contained in HeLa whole-cell lysates will initiate transcription in vitro from at least one promoter site only under conditions drastically different than those required for initiation by RNA polymerase II (Manley and Colozzo, 1982). A second problem is that in a coupled transcription-processing reaction, even if processing is suspected, it may prove very difficult to carry out many important experiments, such as demonstration of a precursor-product relationship, identification of intermediates and analysis of the kinetics of the processing reaction. Finally, it will ultimately be absolutely required that transcription and processing be uncoupled if the components required for processing are to be identified and purified. To circumvent these problems, we have begun to use a different approach to search for the enzyme activities responsible for mRNA processing. First, a 32P-labeled precursor RNA that should contain all the signals required for RNA processing is synthesized in a HeLa whole-cell lysate under conditions that give rise to maximal levels of specific transcription. This precursor is then purified and incubated in whole-cell lysate under a variety of different experimental conditions. The RNA is repurified, and its structure is analyzed. Using this approach, I have identified an enzyme activity in HeLa whole-cell lysates that catalyzes the efficient and accurate polyadenylation of in vitro-synthesized mRNA precursors. The reaction is highly specific: only premRNAs that have a 3’ end site (created by “run-off” transcription) that is located near the 3’ end site of the authentic in vivo mRNA serve as efficient substrates in the polyadenylation reaction.

Results Synthesis of a Pre-mRNA in Vitro and Characterization of the RNA after Incubation in a HeLa Whole-Cell Lysate The DNA template used for production of an mRNA precursor should satisfy several criteria. First, it should contain

Cell 596

a strong promoter so that sufficient amounts of the radioactive precursor can be readily obtained. Second, the size of the transcription unit should be relatively small (less than 5 kilobases [kb]) so that the entire transcription unit can be efficiently transcribed in vitro. Third, the transcription unit should be simple, ideally containing only one intron and one poly(A) addition site. Finally, it would be reassuring to know that the RNA can be efficiently processed in vivo in HeLa cells. The recombinant plasmid p’PCSVA (see Figure I), which contains the strong adenovirus late promoter fused to the SV40 T antigen-encoding sequences, fulfills all of these criteria. p’P-SVA was digested with BamHI, purified, and used as a template for in vitro transcription in a HeLa whole-cell lysate (Manley et al., 1980). This resulted in the synthesis of an approximately 2.7-kb run-off transcript (e.g. Figure 2, lane 1). RNA was extracted from the in vitro reaction mixture and aliquots were incubated in the presence of decreasing concentrations of different samples of the same extract. Following reextraction, the RNA was denatured with glyoxal and analyzed by electrophoresis through an agarose gel. Preliminary experiments indicated that, after incubation at low extract concentrations, the size of a fraction of the RNA was increased slightly (results not shown). This observation was further investigated, and the results of one such experiment are shown in Figure 2. Precursor RNA without further incubation is shown in lane 1. When another aliquot of this same RNA is incubated in a reaction mixture containing 20% lysate (transcription reactions typically contain 50-60% lysate) and 500 PM ATP, a substantial fraction (-70%) of the 2.7-kb run-off transcript disappears and another species, approximately 200-300 nucleotides larger, is detected. (This size estimate is based on comparisons with a number of RNA size markers; see below.) If ATP is omitted, appearance of this larger species is completely inhibited (lane 3). Results identical to those shown in lane 3 are obtained if any of the other three ribonucleotide triphosphates are substituted for ATP (results not shown). If lysate is omitted, processing is again completely blocked (lane 4). Conversion of the

Eall

(5171) lkll

hq

I

ml

1.471

A42

Figure 1. Map of the Recombinant

12.001

EdI

L26%3 hall

Ml

(2.42)

t267J

1225)

run-off RNA to a higher molecular weight shows a somewhat surprising dependence on the concentration of RNA in the reaction: if the amount of RNA in the reaction mixture is reduced by 50%, the afficiency of processing (i.e. polyadenylation; see below) is not significantly affected (compare lanes 2 and 5). However, if the concentration of added RNA is reduced by 80%, polyadenylation is much less efficient (lane 6). Note that both the fraction of the

123456

Figure 2. Characterization of the p’P4SVA BamHI-Generated after Incubation in a HeLa Whole-Cell Lysate

Run-off RNA

IV46

Plasmid pcP-SVA

p(P4-SVA contains adenoviral and SV40 DNA sequences inserted between the EcoRl and BamHl sites of pBR322. The heavy line represents adenovirus DNA sequences from -260 base pairs to +33 base pairs relative to the late mRNA start site, which is +I (Hu and Manley, 1981). The SV40 sequences extend from a Hindlll site at nucleotide 5171 to the BamHl site at 2533 (coordinates are SV numbers; Buchman et al., 1981). pBR322 sequences are indicated by the dashed line. P represents the adenovirus late promoter site and poly A indicates the SV40 early region polyadenylation site. The numbers in parentheses show the drstance (in kilobases) from the RNA start site to the indicated restriction endonuclease cleavage site.

RNA was synthesized In vitro, purified, and added to reaction mixtures as described in Experimental Procedures (except that creatine phosphate was not included rn the processing reaction). Following extraction and glyoxalation, RNA was analyzed by electrophoresis through a 1.2% agarose gel. Lane 1, run-off RNA without additronal incubation; lane 2, incubated as described in Experimental Procedure; lane 3, minus ATP: lane 4, minus lysate, plus an equal volume of dialysis buffer: lane 5, identical to lane 2, except that the amount of RNA added was reduced by 50%: lane 6, identical to lane 3, except that the amount of RNA added was reduced by 80%. In lanes I-4, the amount of RNA analyzed was equivalent to 20% of the RNA In a processing reaction; in lane 5, 40%. and in lane 6. lOO%, of the RNA was analyzed.

29’itro

Polyadenylation

RNA that becomes polyadenylated and also the apparent size of the poly(A) added are significantly reduced. A similar pattern is obtained if the concentration of extract in the reaction mixture is varied (unpublished results).

B

A 123

456

M

123

Analysis of the Processed RNA by Oligo(dT) Selection and Sl mapping The experiments just described strongly suggested that RNA synthesized from the plasmid pcP4SVA can be efficiently polyadenylated if incubated in vitro under the appropriate conditions. This notion was tested directly by determining the ability of the RNA to bind to oligo(dT)cellulose (Figure 3A). Before processing, the RNA quantitatively flows through a column of oligo(dT)-cellulose (lanes l-3); i.e. it is not polyadenylated. However, after incubation in a whole-cell lysate as described in Figure 2, lane 2, the RNA partitions into two fractions (lanes 4-6): the RNA that was not altered in size by incubation also quantitatively flows through the column. However, the fraction of RNA that was increased in size is quantitatively retained on the column until eluted with a low salt buffer (see Experimental Procedures). This bound fraction quantitatively rebinds to the oligo(dT)-cellulose if recycled through the binding procedure (e.g. see Fig. 7) or remains bound if the column is extensively washed with an intermediate salt buffer (0.125 M NaCI) before elution (results not shown). The experiments just described firmly establish that poly(A) sequences, approximately 200-300 nucleotides in length, are added to the in vitro-synthesized RNA, presumably to the 3’ end, when this RNA is incubated in a HeLa whole-cell lysate. An important question then arises: is the poly(A) added directly onto the RNA 3’ end created by run-off transcription, or is the RNA first cleaved at the authentic poly(A) addition site utilized in vivo, and then polyadenylated at this site? The natural SV40 early mRNA 3’ end site is located about 45 nucleotides upstream of the BamHl site used to generate the run-off transcript (Reddy et al. 1979; see Figure 1). That mRNA 3’ ends can be created by endonuclytic cleavage was first suggested by experiments on the adenovirus late transcription unit (Nevins and Darnell, 1978) and was subsequently confirmed, in this system, by direct biochemical analysis (Manley et al., 1982). Although there is evidence suggesting that the 3’ ends of SV40 late mRNAs are created by a similar mechanism (Lai 1978; Ford and Hsu, 1978) it remains a hypothesis that such cleavage is involved in formation of the SV40 early mRNAs. To determine whether cleavage occurs in the polyadenylation reaction described above, the 3’ ends of unprocessed and processed, oligo(dT)-selected RNA were localized by the Sl -mapping technique (Berk and Sharp, 1977). %P-labeled RNA was hybridized to the purified SV40 Pst B fragment (0.04-0.27 SV40 map unit), and treated with Sl nuclease. Following removal of the nuclease, the DNA-RNA hybrids were denatured with glyoxal, and the size of the protected RNA was determined by agarose gel electrophoresis. (Analysis of the RNA, rather than the DNA-RNA hybrids, allows for

2.7

64.

.67

.37*

.16Figure 3. Analysrs of Processed Sl Mapprng

RNA by Oligo(dT)

Selection and Nuclease

A, equal aliquots of RNA prror to processing (lanes 1-3) or after processrng (lanes 4-6) were analyzed by oligo(dT)-cellulose column chromatography (Manley et al., 1979) glyoxalation, and electrophoresrs. Lanes 1 and 4, total, unselected RNA; lanes 2 and 5, poly(A)- fraction; lanes 3 and 6, poly(A)+ fractron. B, equal amounts of RNA prior to processing (lanes 1 and 3) or following processing and selection on oligo(dT)-cellulose (lane 2) were analyzed by hybridizatron to the SV40 Pst B fragment (lanes 1 and 2) drgestron with Sl nuclease, glyoxalation, and electrophoresrs through a 1.6% agarose gel. Lane 3 is a control that did not contarn DNA rn the hybridization reaction. Lane M displays =P-RNA size markers obtained from runoff transcnption. Numbers Indicate RNA sizes (in kilobases)

an accurate estimate of the size of the RNA, by comparison with RNA size standards.) The results, shown in Figure 38, indicate that the unprocessed RNA and the polyadenylated RNA have the same 3’ ends. The size of the protected RNAs (0.67 kb) are the sizes expected if the RNAs are terminated at the BamHl site, and not modified by endonuclytic cleavage at the authentic in vivo mRNA 3’ end site. Thus, the poly(A) appears to be added onto the 3’ end of the RNA that was created by run-off transcription.

The in Vitro Poly(A) Addition Reaction Displays a Remarkable Substrate Specificity All of the experiments described above utilized the recombinant plasmid p(P4SVA linearized with BamHl to generate “run-off” transcripts that were then used as substrates for RNA processing. To determine whether the nature of the RNA precursor influences the efficiency of polyadenylation, a variety of different precursors have been tested. These include RNAs transcribed from different recombinant plasmids, endogenous HeLa RNAs, and transcripts produced from p’P4SVA that had been digested with restriction

cell 598

A 1234

56

78,

28 S-

18S-

., .*

barn Ftgure 4. Fate of Other RNAs followlng lncubatton

eco in HeLa Whole-Cell

Lysate

A, pBalE run-off RNAs. The recombinant plasmid pBalE (Manley et al., 1980) was digested with BamHl (lanes 1-4) or EcoRl (lanes 5-8) and aliquots were used as templates in standard preparative transcription reactions. Following Incubation in processrng reactions and purification, equal aliquots of the RNA were analyzed as follows: lanes 1 and 5, precursor RNA; lanes 2 and 6, processed RNA; lanes 3 and 7, processed RNA selected on oligo(dT)-cellulose, poly(A)- fraction; lanes 4 and 8. processed RNA selected on oligo(dT)-cellulose, poly(A)+ fractron. Numbers indicate RNA sizes in kilobases. B, HeLa ribosomal RNA. RNA from a processing reaction that had contained a precursor synthesized from p’P4-SVA digested with BamHl (Figure 7A. lanes 9-12) was analyzed exactly as rn Figure 6A, lanes 1-4, except that only the RNA visualized by ethrdrum bromide staining is shown. The only bands visible correspond to 18s and 28s rrbosomal RNA.

endonucleases other than BamHI. In Figure 4A, experiments which tested the ability of RNAs transcribed from the recombinant plasmid pBal E (Manley et al., 1980) to serve as substrates for polyadenylation are shown. This plasmid contains a 2.4-kb (14.7 to 21.5 mu.) fragment of adenovirus DNA inserted at the BamHl site of pBR322. This region contains the late promoter and encodes the first and second late RNA leader

segments but no known poly(A) addition sites (Flint and Broker, 1981). If this DNA is cleaved with BamHl or EcoRI, run-off transcripts 1.75 or 2.1 kb long, respectively, are obtained following in vitro transcription (Figure 4A, lanes 1 and 5). To determine the efficiency with which these transcripts can be polyadenylated in a subsequent processing reaction, RNA was analyzed by glyoxalation and gel electrophoresis before processing, following process-

In Vitro Polyadenylation 599

ing, and after fractionation by oligo(dT)-cellulose selection. If the RNA can serve as a substrate for polyadenylation, the amount of the run-off transcript should decrease following the processing reaction, and this decrease should be accompanied by the appearance of a new, higher molecular weight species (e.g. see Figure 2). RNA the size of the initial precursor should not bind to oligo(dT), but the higher molecular weight RNA should be quantitatively retained on such columns (e.g. see Figure 3). The data in Figure 4A indicate that the 1.75kb BamHI-generated runoff transcript is not detectably polyadenylated. Only a very small fraction (approximately 10%) of the Eco RI-generated transcripts appear to become polyadenylated. Unlike with the pP4SVA BamHI-generated run-off RNA, a discrete band corresponding to the polyadenylated RNA has never been observed. Rather, a smear extending slightly upwards from the position of the run-off RNA is observed (e.g. Figure 4A, lane 8). This finding suggests that not only is this precursor polyadenylated at only a very low efficiency, but also the size of poly(A) added is relatively small. A large amount of RNA other than the in vitro-synthesized RNA is added to the processing reaction (see above), and a large fraction of this is HeLa ribosomal RNA. The fate of this RNA following incubation in a HeLa whole-

A

12345678

Pst Figure 5. Processing

9101112

bcl

of RNAs Synthesized

barn from pq4-SVA

That Had Been Drgested

cell lysate reaction under conditions that brought about efficient polyadenylation of in vitro-synthesized p(PCSVA BamHl run-off RNA in the same reaction mixture is shown in Figure 4B. In this experiment, RNA was analyzed precisely as in Figure 4A, except that after electrophoresis, the RNA was deglyoxalated and visualized by ethidium bromide staining. The results show that neither the 18s nor 28s RNA is detectably increased in size, and that this RNA quantitatively flows through a oligo(dT)-cellulose column. It therefore appears that these RNA species are not substrates for polyadenylation. Identical results were obtained when the fate of 5s and tRNA were analyzed: no change in size can be detected, and no binding to oligo(dT)-cellulose is apparent (results not shown). Figure 5 shows the results of experiments designed to test whether transcripts of p(P4SVA terminated at sites other than the BamHl site can function as substrates in the polyadenylation reaction. Four run-off transcripts, ranging in size from 2.0 to 3.0 kb (see Figure 1) were synthesized, purified, and added to HeLa whole-cell lysates under conditions optimal for obtaining polyadenylation of the pcP4SVA BamHI-generated run-off RNA. In Figure 5A, in vitro transcripts were analyzed exactly as shown in Figure 4A. The BamHI-generated run-off transcript (lanes 9-12) is polyadenylated with the same high efficiency as shown

131415 16

B

12

34

M

sal wrth Different

Restrictron

Endonucleases

A, transcripts synthesized from p+WSVA that had been digested with Pstl (lanes I-4) Bell (lanes 5-8) BamHl (lanes 9-12), or Sall (lanes 13-16) were analyzed before and after processing. Lanes 1, 5, 9, and 13, precursor RNA; lanes 2, 6, 10, and 14, processed RNA; lanes 3, 7, 11, and 15, processed RNA selected on oligo(dT)-cellulose, poly(A)- fraction; lanes 4, 8, 12, and 16, processed RNA selected on olrgo(dT)-cellulose, poly(A)+ fraction. Numbers rndrcate RNA srzes in krlobases. B. processed RNA that had been synthesized from p‘+WSVA digested with Sall (lanes 1 and 2) or BamHl (lanes 3 and 4) were analyzed by olrgo(dT)-cellulose chromatography. Lanes 1 and 3. poly(A)- fractron; lanes 2 and 4, poly(A)+ fraction, selected twrce on oligo(dT); lane M. RNA markers Electrophoresis in a 1 2% agarose gel was for 5 hr at 7.5 V/cm.

Cdl 600

above. The 2.00&b Pstl- and 2.43-kb B&generated transcripts apparently do not serve substrates for polyadenylation (lanes l-8): No increase in the size of these transcripts has been detected, and the processed transcripts do not appear to bind to oligo(dT)-cellulose. (Note, however, that a small fraction [approximately lo%] of the processed, Bell-generated RNA is indeed found in the poly(A)+ fraction [lane 81. It is unlikely, however, that this RNA has been polyadenylated, since the size of the RNA is exactly the same as the unprocessed precursor [within the limits of resolution of this method] and the RNA does not efficiently rebind to oligo(dT)-cellulose [results not shown]. However, the possibility that a very short stretch of poly(A) is added to a small fraction of the molecules cannot be excluded.) A significant fraction of the RNA produced from the p(P4SVA Sall-digested template does, however, become polyadenylated (Figure 5A, lanes 13-16). Poly(A)+ and poly(A)- RNA obtained after processing of transcripts synthesized from this template, and from p+‘4-SVA BamHIcut DNA is analyzed in more detail in Figure 58. Specifically, the poly(A)’ RNA (lanes 2 and 4) was selected twice on oligo(dT)-cellulose, and the time of electrophoresis was significantly longer than in the experiments shown previously. Two important points are apparent. First, although both these transcripts function as substrates for polyadenylation, the efficiencies with which they become polyadenylated is quite different. Thus, approximately 25% of the Sall-generated transcript is polyadenylated (compare Figure 5B, lanes 1 and 2) while approximately 75% of BamHIgenerated RNA becomes polyadenylated (lanes 3 and 4). These estimates are obtained by first scanning the autoradiogram, and then comparing the intensities of the bands obtained from the poly(A)- run-off RNAs with the intensities of those which arose from the polyadenylated RNAs. Nearly identical estimates have also been obtained by a different method, described below. The second point is that the poly(A) added to BamHl run-off transcript is longer than the poly(A) added to the Sall-generated RNA. The Sall-poly(A)+ RNA (Figure 58, lane 2) forms a broad band which extends from a point just larger than the band from the poly(A)- run-off RNA (lane 1) up to a point corresponding to a size approximately 150 nucleotides larger than the run-off transcript. The BamHI-poly(A)+ RNA (Figure 5B, lane 4) also contains some RNA only slightly larger than the poly(A)- run-off transcript (lane 3). However, the bulk of the BamHI-poly(A)+ RNA gives rise to a broad, intense band that corresponds to a size roughly 200-300 nucleotides larger than the poly(A)- RNA. Thus, it appears that both the fraction of a particular RNA that can become polyadenylated and also the length of poly(A) added can vary, depending on the sequence (or structure) at the 3’ end of the RNA. The above results suggest that a sequence encoded between the Bell and BamHl sites of p’P4SVA is necessary to obtain efficient polyadenylation of an mRNA precursor in vitro. One possibility is that the 3’ end of the RNA

generated by run-off transcription must approximate the 3’ end of the authentic in vivo mRNA. (The length of the BamHl run-off RNA is only approximately 45 nucleotides longer than the corresponding in vivo mRNA would be; see Figure 1.) A prediction of this model is that restriction endonuclease cleavage of a given gene at a site very close to the authentic mRNA 3’ end site should lead to the synthesis, following in vitro transcription, of a run-off RNA that will function efficiently as a substrate for polyadenylation. An excellent test of this prediction is provided by the adenovirus polypeptide IX gene. The restriction endonuclease Bell cleaves this gene at a site only three to four nucleotides upstream of the polyadenylation site (Alestrom et al., 1980). The recombinant plasmid pAd6 (Hu and Manley, 1981) contains a 3.4-kb Hindlll fragment of adenovirus DNA which includes this gene, as well as the late promoter and 190 base pairs of downstream sequences, If this DNA is cleaved with Hindlll and then used as a template for in vitro transcription, run-off RNAs of 2.7 kb (from the plX promoter) and 0.19 kb (from the late promoter) are produced (see Figure 6). If the DNA is addition ally cleaved with Bell, then the length of the plX run-off RNA synthesized is reduced to approximately 0.47 kb. Figure 6 shows the results obtained when run-off RNAs synthesized from pAd6 digested with Hindlll or Hindlll and Bell were incubated in HeLa whole-cell lysate under conditions optimal for polyadenylation. In addition, one further plp4-SVA run-off RNA (obtained by Taql digestion of the DNA template; see Figure 1) is analyzed. (The size of this RNA [0.47 kb] is nearly identical to the run-off RNA obtained from the plX promoter when pAd6 is digested with Bell.) Lanes l-4 show that the p(P4SVA Taql-generated run-off RNA is not detectably polyadenylated. This is as expected if the model mentioned above is correct. Likewise, it appears that the 190nucleotide run-off RNA synthesized from the adenovirus late promoter is not a substrate for polyadenylation (lanes 5-8 and 9-12). The data in lanes 5-8 suggest that, in contrast, the 0.47-kb transcript synthesized from the plX promoter is efficiently polyadenylated in vitro. Following incubation in the whole-cell lysate, the 0.47 kb run-off RNA (arrowhead) can no longer be detected (compare lanes 5 and 6). However, unlike with the p(o4SVA BamHI-generated run-off RNA (see above), no band corresponding to a new, higher molecular weight RNA is apparent. When the processed RNA is separated into poly(A)- (lane 7) and poly(A)+ (lane 8) fractions, a smear of RNA can be detected in the poly(A)+ RNA. The bulk of this RNA (indicated by the bracket) is approximately 150-300 nucleotides larger than the run-off RNA. While it would be difficult to conclude from this experiment alone that this smear results from polyadenylation of the plX run-off RNA, such a conclusion is strongly supported by Sl quantitation experiments described below. The 2.7-kb plX run-off RNA generated by Hindlll cleavage also functions as a substrate for polyadenylation (lanes 9-12). However, the fraction of RNA converted to a poly-

lnOytro Polyadenylation

12

34

56

76

M

Figure 6. Analysis of Additronal Run-off RNAs followrng Incubation in HeLa Whole-Ceil Lysate Transcripts synthesized either from p94SVA that had been digested with Taql (lanes 1-4) or from pAd6 digested with Hindlll plus Bell (lanes 5-8) or Hindlll alone (lanes 9-12) were analyzed before and after RNA processing. Lanes 1. 5, and 9, precursor RNA; lanes 2, 6, and 10, processed RNA; lanes 3, 7, and 11, processed RNA selected on oligo(dT)-cellulose, poly(A)- fraction; lanes 4, 8, and 12. processed RNA selected on oligo(dT)cellulose, poly(A)+ fraction; lane M, RNA size markers Numbers indrcate RNA srze in krlobases. The arrowhead adjacent to lane 5 indicates the band corresponding to the unprocessed run-off RNA synthesrzed from the plX promoter, whrle the arrow Indicates the RNA synthesized from the late pro moter. The line drawing IS a simplified transcnptronal map of the adenoviral DNA insert in pAd6. H represents Hindlll and B IS Bell. Glyoxalated RNAs were analyzed by electrophoresis in a 1 6% agarose gel.

.47

H

A

123

456

B

B

Figure 7. Quantrtatron clease Sl Mapping

123

- 48 .67 - .1g

adenylated form is low (approximately 20%) and the apparent size of the poly(A) added is relatively small. The transcript thus behaves quite analogously to the p(P4SVA Sall-generated RNA (see Figure 5). Quantitation of Polyadenylation by Sl Nuclease Mapping To both confirm the quantitations described above and also obtain additional evidence that the plX Bell-generated run-off RNA is indeed a substrate for efficient polyadenylation, Sl nuclease protection experiments were carried out. Equal aliquots of total processed RNA, poly(A)- and poly(A)+ fractions were hybridized to the appropriate unlabeled DNA probes and treated with Sl nuclease, and the resistant hybrids were denatured with glyoxal and the RNA was analyzed by gel electrophoresis.

of Polyadenylatron

by Nu-

Processed RNAs were analyzed before and after selectron on olrgo(dT)-cellulose. Equal akquots of total RNA, poly(A)- RNA, and poly(A)+ RNA that had been selected twrce on oligo(dT)-cellulose were hybndlzed, treated with Sl nuclease, purrfled. glyoxalated, and electrophoresed in a 1.6% agarose gel. A, RNAs obtarned from p474SVA that had been digested wrth BamHl (lanes 1-3) or Sall (lanes 4-6) were hybridized to the SV40 DNA Pst B fragment. Lanes 1 and 4. total processed RNA, lanes 2 and 5. poly(A)- RNA; lanes 3 and 6. poly(A)+ RNA. 6. RNAs obtained from pAd6 that had been drgested wrth Hrndlll and Bell were hybndtzed to the adenovirus Hindlll fragment purified from pAd6 at 59°C (lanes l-3) or 56°C (lanes 46). Lanes 1 and 4. total processed RNA; lanes 2 and 5, poly(A)- RNA: lanes 3 and 6, poly(A)+ RNA. Numbers Indicate srzes of RNA in kilobases.

In Figure 7A, the results obtained with RNA synthesized from p(04-SVA that had been previously digested with either BamHl (lanes l-3) or Sall (lanes 4-6) are shown. The DNA probe used was the SV40 Pst B fragment (see Figure 36). The intensities of the bands obtained confirm the quantitations presented above: approximately 70% of the BamHI-generated RNA is polyadenylated after incubation in the whole-cell lysate (lanes l-3) while only 20-25% of the Sall RNA is found in the poly(A)+ fraction (lanes 46). (In the experiments shown in Figure 7, the poly(A)+ fractions represent RNA that was bound to and eluted from oligo(dT)-cellulose columns two times.) In Figure 78 the processed transcripts synthesized from pAd6 that had been cleaved with Hindlll plus Bell are shown following nuclease Sl analysis. Approximately 75% of the 470nucleotide plX run-off RNA is found in the

Cell 602

poly(A)+ fraction (compare lanes 1 and 3). This high percentage may even represent an underestimate of the ability of this RNA to function as a substrate for polyadenylation, because a band corresponding in size to the full length run-off RNA is not apparent in the poly(A)- fraction (lane 2). Rather, the protected RNA in this fraction is slightly smaller than the intact run-off RNA, indicating that it may have been slightly degraded during one of the incubations or purification steps, and that this degradation inhibits subsequent polyadenylation. Since the DNA probe used was the entire Ad2 Hindlll fragment from pAd6, it was surprising that the 0.19-kb run-off RNA was not detected. This appears to be related to the temperature at which the hybridization reaction is performed, because when the temperature was reduced from 59°C (lanes l-3) to 56’C (lanes 4-6) this short run-off RNA is now readily visualized. As expected, this transcript is not detectably polyadenylated. Note that, in this case, the percentage of the plX run-off RNA found in the poly(A)+ fraction is lower than in the previous experiment. This again appears to be due to degradation, since the size of the plX-specific RNA in the poly(A)- fraction is smaller than the full length run-off RNA. These results show that the Bell-generated run-off transcript of the plX gene functions at least as efficiently as a substrate for polyadenylation as does the p’P4SVA BamHIgenerated RNA.

Animal cell poly(A) polymerases have been studied for well over 20 years (e.g. Edmonds and Abrams, 1960). Interest in this class of enzymes has been intense since the discoveries that the vast majority of eucaryotic mRNAs are polyadenylated at their 3’ ends, and that these sequences are added post-transcriptionally (see for review Brawerman, 1976). Studies with partially purified poly(A) polymerases, although providing many insights into the nature of the enzyme (reviewed by Edmonds and Winters, 1976) have been somewhat disappointing in several regards. For example, the rate of elongation by the enzyme is very low (2 residues/min), no specificity is observed in the utilization of RNA primers (tRNA and other similarly sized small RNAs function with the highest efficiencies), and the size of the poly(A) segment added is variable (from 2 to over 1000 nucleotides), depending primarily on the duration of the in vitro incubation (i.e. the length to which the poly(A) product can grow is not controlled in vitro as it must be in vivo). This situation is strikingly analogous to what is found when purified eucaryotic RNA polymerases are used to transcribe double-stranded DNA (e.g. Roeder, 1976); transcription initiation is very inefficient, and no signs of specificity are detected. In order to obtain specific and accurate transcription initiation in vitro, it has been necessary to use less pure systems. Thus, the first demonstrations of correct transcription initiation were obtained in studies with isolated nuclei, which subsequently led to the development of the soluble, DNA-dependent transcription systems now in

common use (see Manley, 1982, for review). These studies have established that factors in addition to RNA polymerase are required in order to obtain specific transcription initiation in vitro. Studies on the mechanism of polyadenylation are proceeding on a nearly identical path. Isolated nuclei have thus been shown to be able to bring about the accurate, specific, and efficient polyadenylation of mRNA precursors (Jelinek, 1974; Kieras et al., 1978; Manley et al., 1979, 1982; Yang and Flint, 1979). As with transcription initiation, these studies have paved the way for the development of soluble systems with which polyadenylation can be studied. The only soluble system capable of specific and accurate polyadenylation described previously is one that utilizes adenovirus transcription complexes isolated from infected cells (Chen-Kiang et al., 1982). Adenovirus-specific RNA transcribed in vitro from these complexes is also accurately and specifically polyadenylated. Isolation of transcription complexes affords a substantial initial purification of the components required for RNA processing. However, before it is possible to exploit this fractionation, it will be necessary to determine whether or not conditions can be found that allow first for the uncoupling of transcription and polyadenylation, and then for the reconstitution of the polyadenylation activity. The results presented here establish that an in vitrosynthesized precursor RNA can be accurately and efficiently polyadenylated when incubated in a HeLa wholecell lysate. The requirements for obtaining polyadenylation in vitro are as precise as those for obtaining specific transcription initiation. When an appropriate precursor is incubated under optimal conditions, 70% or more of the RNA becomes polyadenylated, the apparent rate of elongation is high, and the reaction appears to be processive (unpublished results). Also, the size to which the poly(A) segment grows is controlled, resulting in the addition of between 150 and 300 nucleotides to the precursor (e.g. Figure 58 and unpublished results). This size agrees well with an estimate of the size of poly(A) segments on steadystate nuclear RNA, which is approximately 230 nucleotides (Sawicki et al., 1977). However, the length of the poly(A) added in vitro is probably more heterogeneous than is found on nuclear mRNA precursors. Two explanations, not mutually exclusive, for this discrepency are apparent. First, although the length of poly(A) added is controlled in the in vitro reaction, it may be that in vivo a second, “fine-tuning” control is also operative. Second, it may be that the immediate product of poly(A) addition in vitro is as heterogeneous as observed in vitro, but that this heterogeneity is quickly reduced in vivo, perhaps by the action of specific nucleases or a second addition of AMP residues. In fact, evidence for such “trimming” and “filling” reactions has been presented (Brawerman and Diez, 1975; Sawicki et al., 1977). An interesting response is observed when polyadenylation reactions are carried out at suboptimal conditions:

In Vitro Polyadenylation 603

not only is the fraction of precursor that becomes polyadenylated reduced, but also the length of the poly(A) segment added is shortened. It is particularly noteworthy that both of these parameters are influenced by the sequence (or structure) at the 3’ end of the RNA precursor: those precursors with which a high percentage of the molecules become polyadenylated also produce the longest poly(A) segments, While it is easy to imagine that a sequence (or structure) at the 3’ end of a precursor might influence how frequently that RNA is used as a substrate by the poly(A) polymerase, it is somewhat surprising that this feature also affects the lengths of the resultant poly(A) chains added to the RNA. The above discussion should illustrate that, as is the case with animal cell RNA polymerases, the poly(A) polymerase almost certainly requires one (or more) factors to bring about accurate and specific polyadenylation. Perhaps, for example, a specific factor is required for selection of a specific substrate, while another factor might help “set” the length of poly(A) to be added. Whatever their nature, it is now feasible to identify and characterize them by fractionation of HeLa whole-cell lysates. The poly(A) segments appear to be added directly onto the 3’ ends of the RNA precursors generated by run-off transcription (see Figure 38); i.e. the specific endonuclease that has been implicated in 3’ end formation in isolated nuclei and in vivo (see Manley, 1983, for review) is not active in the in vitro assay described here. Previous experiments have suggested that the poly(A) polymerase and endonuclease function in a coupled manner, most likely as an enzyme complex (Manley et al., 1982). A simple notion was that the endonuclease recognizes a specific sequence in pre-mRNA to generate a 3’-OH, which subsequently functions as primer for synthesis of poly(A) by the poly(A) polymerase. It was thus somewhat surprising when a marked specificity in utilization of RNAs as precursors for polyadenylation was observed (Figures 4-7). In particular, of the 10 different run-off transcripts tested, only 2 could be efficiently (greater than 70%) polyadenylated. These two transcripts share one feature in common, which also serves to distinguish them from the other eight: the 3’ ends of these two run-off RNAs are only slightly different than the 3’ ends of the corresponding in vivo mRNAs. These results strongly suggest that the poly(A) polymerase and/or an associated factor recognize(s) a specific sequence (or structure) in pre-mRNA. A strong candidate for this signal is the sequence 5’-AAUAAA-3’. This highly conserved sequence (Proudfoot and Brownlee, 1976) or a variant of it, is found within IO to 30 nucleotides upstream of 3’ ends of all mammalian polyadenylated mRNAs analyzed to date. Genetic studies have shown that this sequence is in fact required for the formation of SV40 late (Fitzgerald and Shenk, 1981) and adenovirus early region IA (A. J. Berk, personal communication) mRNA 3’ ends. In vitro, if a 3’ end is located within a short distance downstream of such a sequence, a poly(A) segment will be polymerized onto the RNA. An attractive model is that

in vivo a similar recognition occurs between the poly(A) polymerase complex and RNA. However, in this case, the associated endonuclease cleaves the RNA a set distance downstream, and this 3’ end is then used as a primer for the poly(A) polymerase. An alternative hypothesis that cannot at the moment be excluded is that the specificity for recognition does indeed reside with the endonuclease. However, it may be that a cofactor is required for RNA cleavage (although not for specific binding), and that this cofactor is lacking in in vitro reaction mixtures. Note that both these models require that the poly(A) polymerase and endonuclease are intimately associated with each other. Although only two of the run-off transcripts tested are efficient substrates for polyadenylation, three others are utilized at low (lo-25%) but reproducible efficiencies. It is not clear what sequence (or structure) serves as a recognition signal for the poly(A) polymerase in these cases; however, several features merit discussion. The pcP4SVA Sall-generated run-off is 275 nucleotides longer than the efficiently utilized BamHl run-off (see Figure 1). It is conceivable that polyadenylation of both of these precursors is brought about by the same signal. Two disparate observations support this possibility. First, when pAd6 cleaved with Hindlll is used as a template for in vitro transcription, a 2.7-kb transcript (from the plX promoter) and a 0.19-kb RNA (from the late promoter) are produced (see Figure 6). These transcripts are co-terminal, sharing the same 190nucleotide sequence at their 3’ ends. When these RNAs are added to an in vitro polyadenylation reaction, only the 2.7-kb transcript is detectably polyadenylated (Figures 6 and 7). This result suggests that sequences more than 190 nucleotides upstream of the RNA 3’ end are required, either directly or indirectly, for polyadenylation. (An intriguing possibility is that the sequence 5’-TATAAA-3’, which constitutes the “TATA box” required for transcription initiation from the late promoter [Corden et al., 1980; Hu and Manley, 19811, can serve as a polyadenylation signal when it is present in RNA). A second observation that is perhaps relevant is that RNA polymerase II, in some instances, is able to initiate transcription at sites far downstream from the promoter site. For example, when the SV40 early region TATA box is deleted, transcription initiation, both in vivo and in vitro, occurs at multiple sites, some located as far downstream as 300 base pairs from the promoter region (Benoist and Chambon, 1981; Mathis and Chambon, 1981). The role that poly(A) sequences play in the control of gene expression is not yet clear, although they have been implicated in nuclear to cytoplasmic transport and/or mRNA stability (see Brawerman, 1981, and Littauer and Soreq, 1982, for reviews). The results presented here show that, in vitro, both the fraction of pre-mRNA that becomes polyadenylated and also the length of the poly(A) segment synthesized are modulated by a variety of factors. If this same situation obtains in vivo, then these findings may reflect an important mechanism by which gene expression can be regulated.

Cell 604

Experimental

Procedures

Pettersson, U. (1980). The gene for polypeptide unspliced messenger RNA. Cell 19, 671-681.

IX of adenovirus

Cells and DNA HeLa cells were maintained in minimal essential medium plus 5% horse serum (Gibco). Recombinant plasmids were grown in E. coli HBIOI, except when it was required that the DNA be digested with the restriction enzyme Bell. In this case, plasmids were grown tn E. coli GM33 (dam-). Plasmid DNA was extracted by established procedures (Davis et al., 1980) digested with restriction endonucleases (New England Biolabs), and purified for transcription as described (Manley et al., 1980).

Benoist, C., and Chambon, P. (1981). In vivo sequence the SV40 early promoter region. Nature 290, 304-310.

In Vitro Transcription HeLa whole-cell lysates were prepared as described (Manley et al., 1980, 1983) except that the final dialysis buffer was 40 mM Tris-HCI, pH 7.9, 100 mM KCI, 10 mM MgCh. 2 mM DTT, 0.1 mM EDTA, 15% glycerol. In vitro transcription reactions were carried out essentially as described. Reaction mrxtures contained 60% (vol/vol) lysate, 50 pg/ml DNA, 4 mM creatine phosphate, disodrum salt, and 50 pM each of the four ribonucleotide triphosphates. Standard preparative reaction mixtures were 100 pl in total volume, and contained 25 pCi o-“P-GTP (New England Nuclear). RNA was extracted from reaction mixtures as described.

Brawerman, G. (1981). The role of the poly(A) mRNA. CRC Crit. Rev. B&hem. 70, l-38.

In Vitro Polyadenylation Standard processing reaction mixtures (25 ~1) contained 5 pl lysate, 500 pM ATP, 500 pM EDTA, 4 mM creatine phosphate, and 5% of the RNA obtained from a standard preparative transcription reactron. Reaction mixtures were incubated at 30°C for 60 min. RNA was extracted by the same method used for purifying RNA from transcription reactions (Manley et al., 1980). RNA Analysis RNA was fractionated by oligo(dT)-cellulose chromatography as described (Manley et al., 1979). If the RNA was to be analyzed by the nuclease Sl method, it was first treated with RNase-free DNase (Boehringer Mannheim) by Incubation rn a 100.pl solution containrng 5 Ag DNase, 10 mM Tris-HCI, pH 7.5, 100 mM NaCI, and 10 mM MgCl2 for IO min at 37°C. RNA was then purified by phenol-chloroform extraction and ethanol precipitation. Nuclease Si analysis was performed essentially as described by Berk and Sharp (1977). as modified for “P-labeled RNA (Manley et al., 1979). The DNA probes were obtained from recombinant plasmids indrcated in the text by restnction endonuclease digestion followed by agarose gel electrophoresrs. DNA was eluted from gels by the freeze-thaw method (Smith, 1980). Roughly 50 ng were used per hybridization reaction, Nuclease Sl was from Boehringer Mannheim or Miles. In all cases, size analysis of RNA was performed by glyoxalation and agarose gel electrophoresrs (McMasters and Carmichael, 1977). Routinely, 25-50% of each RNA sample was analyzed. “P-RNA size markers were obtained from an rn vitro transcription reaction that had contarned the recombinant plasmid p’P4 digested with different restrictron endonucleases (Hu and Manley, 1981). Following electrophoresrs, RNA was deglyoxalated and endogeneous HeLa RNA was vrsualrzed by staining with ethidium bromrde, as described. Gels were then dried and exposed to Kodak X-AR5 film (wrthout intensifying screens). Exposures were routinely for l-4 days. When required, autoradrograms were scanned wrth a Gilford model 250 spectrophotometer. Acknowledgments I thank D. Lewis for constructing and supplying the recombinant plasmid pq4SVA, A. Allen for technical assrstance, and C. Prives for helpful discussions. This work was supported by U. S. Public Health Service Grant GM 28983. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received

February

18. 1983; revrsed April 4. 1983

Alastrom,

P Akusjarvr,

G., Perricaudet,

M., Mathews,

M., Klessig,

D., and

requirements

of

Berk, A. J.. and Sharp, P. A. (1977). Srzing and mapping of early adenovirus mRNAs by gel electrophoresis of Si endonuclease-digested hybrids. Cell 12. 721-732. Brawerman, G. (1976). Characteristics and significance of the polyadenylate sequence in mammalian mRNA. Prog. Nucl. Acids Res. Mol. Biol. 77, 117148. sequence

in mammalran

Brawerman. G., and Diez, J. (1975). Metabolism of the polyadenylate sequence of nuclear RNA and messenger RNA in mammalian cells. Cell 5, 271-280. Buchman, A., Burnett, L., and Berg, P. (1981). The SV40 nucleotide sequence. In DNA Tumor Viruses, 2nd ed., revised, J. Tooze, ed. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory), pp, 799-841. Cepko, C. L., Hansen, transcription-translation Cell Biol. 1, 919-926.

U., Handa, H., and Sharp, P. A. (1981). Sequential of SV40 by using mammalian cell extracts. Mol.

Chen-Kiang. S.. Wolgenmuth. D. J., Hsu, M.-T., and Darnell, J. E., Jr. (1982). Transcription and accurate polyadenylation in vitro of RNA from the major late adenovirus 2 transcription unit. Cell 28, 575-584. Corden, J., Wasylyk. B.. Buchwalder, A., Sassone-Corsi, P., Kedinger, C., and Chambon, P. (1980). Promoter sequences of eukaryotic protein-coding genes. Science 209, 1406-l 414. Davis, R., Roth, J., and Botstein, D. (1980). Advanced Bacterial Cold Spnng Harbor, NY: Cold Spring Harbor Laboratory.

Genetics.

Dynan, W. S., and Tjian, R. (1981). Characterization of factors that Impart selectivity to RNA polymerase II in a reconstituted system. ICN-UCLA Symp. 23, 401-414. Edmonds, M., and Abrams, R. (1960). Polynucleotrde biosynthesrs: tron of a sequence of adenylate units from adenosine triphosphate enzyme from thymus nuclei. J. BIoI. Chem. 235, 1142-l 149. Edmonds, M., and Wtnters, M. A. (1976). Polyadenylate Nucl. Acrds Res. Mol. Biol. 17, 149-179.

polymerases.

formaby an Prog.

Frtzgerald, M., and Shenk, T. (1981). The sequence 5’.AAUAAA-3’ forms part of the recognitron site for polyadenylation of SV40 late mRNAs. Cell 24, 251-260. Flrnt, S. J., and Broker, T. R. (1981). Lytic infection by adenoviruses. In DNA Tumor Vtruses, 2nd ed.. revised. J. Tooze, ed. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory), pp. 443546. Ford, J. P., and Hsu, M.-T. (1978). Transcription late SV40 RNA: equrmolar transcription beyond Vrrol. 28, 795-801.

pattern of in vivo-labelled the mRNA 3’ termrnus. J.

Goldenberg, C. J., and Raskas, H. J. (1981). In vitro splicing of purified precursor RNAs specified by early region 2 of the adenovirus 2 genome. Proc. Nat. Acad. Sci. USA 78, 5430-5434 Handa, H.. Kaufmann, R., Manley, J. L., Gefter, M. L., and Sharp, P. A. (1981) Accurate initration of transcription of SV40 early and late genes in whole-cell extracts. J. Biol. Chem. 256, 478-482. Heintz, N., and Roeder, R. G. (1982). Transcription of eukaryotic genes in soluble cell-free systems. In Genetic Engineering, Vol. 4. J. Setlow and A. Hollaender, eds. (New York: Plenum Press), pp. 57-90. Hu, S.-L., and Manley, J. L. (1981). DNA sequence transcription in vitro from the major late promoter Nat. Acad. Sci. USA 78, 820-824.

required for initiation of of adenovrrus 2. Proc.

Jelinek, W. R. (1974). Polyadenylic acid synthesis in vitro in isolated HeLa cell nuclei and whole cell homogenates. Cell 2, 197-204. Kieras, R. M., Amendinger, R. J., and Edmonds. M. (1978). Characterization of the polyadenylic acid sequences in RNA synthesized in vitro by mouse myeloma nuclerc. Brochemistry 17, 3221-3228. Kole, R., and Weissman, S. M. (1982). Accurate @-globin mRNA. Nucl. Acids Res. 70, 54295445.

References

2 and its

Lai, C-J.,

Dhar, R , and Khoury

G. (1978).

in vitro splicing of human Mapping

the spliced

and

In Vitro Polyadenylation 605

unsplrced

late lytrc SV40 RNAs. Cell 74. 971-982.

Lrttauer. U. Z.. and Soreq, H. (1982). The regulatory function of poly(A) and adjacent 3’ sequences In translated RNA. Prog. Nucl. Acids Res. Mol. Biol. 27. 53-83 Manley, J. L. (1982). Transcnptron of mammalian genes in vitro. In Genetic Engrneering, Vol. 4. J. Setlow and A. Hollaender, eds (New York: Plenum Press), pp. 37-56. Manley, J. L (1983). Analysts of the expressron of genes encoding animal mRNA by /n wtro techniques. Prog. Nucl. Acids Res. Mol. Biol. 30, in press. Manley, J. L., and Colozzo, M. T. (1982). Syntheses in vitro of an exceptionally long RNA transcript promoted by an A/u I sequence. Nature 300. 376379. Manley, J. L., Frre, A., Cano. A.. Sharp, P. A., and Gefter, M. L. (1980). DNA-dependent transcription of adenovirus genes in a soluble whole-cell extract. Proc. Nat. Acad. Sci. USA 77, 3855-3859 Manley, J. L., Fire, A., Samuels, M., and Sharp, P. A. (1983). Transcnption of mammalran genes rn a whole cell extract. Meth. Enzymol., in press. Manley. J. L., Sharp, P. A., and Gefter, M. L. (1979). RNA syntheses in Isolated nuclei: rdentifrcation and characterrzatron of adenovrrus 2 encoded transcrrpts synthesized in wtro and in viva. J. Mol. BIoI. 135, 171-197. Manley, J. L., Sharp, P. A., and Gefter, M. L. (1982). RNA synthesis in isolated nuclerc processrng of the adenovirus 2 major late mRNA precursor in wtro. J. Mol. Brol. 759, 581-600 Mathrs. D., and Chambon, P. (1981). The SV40 early region TATA box IS required for accurate in wtro initlatron of transcription. Nature 290, 310315 Matsur, T., Segall. J., Well, P. A., and Roeder, R. G. (1980). Mulitple factors requrred for accurate inrtratron of transcnption by purified RNA polymerase II. J. Brol. Chem. 255, 11992-I 1996 McMasters, G., and Carmrchael, G. (1977). Analysis of srngle- and doublestranded nuclerc acrds on polycrylamide and agarose gels by using glyoxal and acndrne orange. Proc. Nat. Acad. Sci. USA 79, 4835-4838. Nevins, J. R and Darnell, J. E., Jr. (1978). Steps In the processing of Ad2mRNA. poly(A)’ nuclear sequences are conserved and poly(A) addition precedes splicrng. Cell 75, 1477-1487. Proudfoot, sequences Reddy, (1979). termrnr Infected

N. J., and In eukaryotic

Brownlee. G. G. (1976). 3’ non-coding messenger RNA. Nature 263, 211-214.

regron

V B., Ghosh. P. K.. Lebowrtz, P.. Piatak. M., and Weissman, S. M., Simian vrrus 40 early mRNAs. I, Genomrc localizatron of 3’ and 5’ and two malor splrces in mRNA from transformed and lytically cells. J. Vrrol. 30, 279-296.

Roeder, R. G. (1976). Eukaryotrc nuclear RNA polymerases. In RNA Polymerase, R. Losrck, and M. Chamberlin, eds. (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory), pp 285-330. Samuels, M., Fire, A., and Sharp, P. A. (1982). Separatron and characterization of factors medratrng accurate transcrrptron by RNA polymerase II J Biol. Chem 257, 14419-14427. Sawrckr, S., Jelinek, W., and Darnell. J. E., Jr. (1977). 3’ termrnal addition to HeLa cell nuclear and cytoplasmlc poly(A). J. Mol. Biol. 113, 219-235. Smrth. H 0. (1980). 380.

Recovery

of DNA from gels. Meth. Enzymol.

65. 371-

Tsai, S. Y., Tsar, M. J., Kops, L. E , Minghettr, P. P., and O’Malley, 8. W. (1981). TranscrIptron factors from oviduct and HeLa cells are similar. J. 6101.Chem. 256, 13055-13059. Well, P. A., Luse, D S., Segall, J., and Roeder, R. G. (1979). Selective and accurate rnrtiatron of transcriptron at the Ad2 major late promoter In a solbule system dependent on RNA polymerase II and DNA. Cell 18. 469-484 Werngartner, B., and Keller, W. (1981). Transcriptron and processrng of adenoviral RNA by extracts from HeLa cells Proc. Nat. Acad. Sci. USA 78, 4092-4096. Yang, V. W., and Flrnt, S. J. (1979). Synthesis RNA in Isolated nucler. J Vrrol. 32, 394-407.

and processrng

of adenovrral