RNA polymerase II transcription complexes may become arrested if the nascent RNA is shortened to less than 50 nucleotides

A significant fraction of RNA polymerase II transcription complexes become arrested when halted within a particular initially transcribed region after the synthesis of 23-32-nucleotide RNAs. If polymerases are halted within the same sequence at a promoter-distal location, they remain elongation-competent. However, when the RNAs within these promoter-distal complexes are truncated to between 21 and 48 nucleotides, many of the polymerases become arrested. The degree of the arrest correlates very well with the length of the RNA in both the promoter-proximal and -distal complexes. This effect is also observed when comparing promoter-proximal and promoter-distal complexes halted over a completely different sequence. The unusual propensity of many promoter-proximal RNA polymerase II complexes to arrest may therefore be recreated in promoter-distal complexes simply by shortening the nascent RNA. Thus, the transition to full elongation competence by RNA polymerase II is dependent on the synthesis of about 50 nt of RNA, and this transition is reversible. We also found that arrest is facilitated in promoter-distal complexes by the hybridization of oligonucleotides to the transcript between 30 and 45 bases upstream of the 3'-end.


6
DNA elements, since initially transcribed sequences within which upstream translocation takes place present no barrier to polymerase translocation at promoterdistal locations. Complexes with transcripts longer than about 45 nt no longer showed upstream translocation (26), so transcript length seemed to be an important determinant of lateral stability on the template. However, the earlier study could not separate effects based simply on transcript length from those which also depended upon proximity to promoter sequences.
We have now investigated directly the role of transcript length on the elongation competence of transcription complexes, independent of the presence of the promoter.
Through the use of carefully controlled transcript cleavage procedures, we have prepared matched sets of transcription complexes with identical transcript length and nearly identical transcript sequence at promoter proximal and promoter distal locations.
The corresponding complexes display very similar levels of transcriptional arrest. Thus, the nascent RNA itself, well upstream (i.e., 20-50 nt) of the 3' end, facilitates continued transcript elongation by the ternary complex.

9
nonlabled UTP, or UTP and GTP, or UTP, GTP and ATP, for 5 min at 37°C (see Fig. 1 for the sequences).
Ternary complexes (G17) stalled at position +17 on the pML20-23like3 template were synthesized by incubating preinitiation complex with 1 mM ApC, 20 FM ATP, 50 FM dATP and 1 FM "-32 P-GTP for 5 min at 30°C, followed by incubation for 5 min at 30°C with nonlabeled 20 FM GTP and sarkosyl rinsing. Similarly, C23 was made from preinitiation complex with 1 mM ApC, 20 FM each of ATP and GTP, 50 FM dATP and 1 FM "-32 P-CTP for 5 min at 30°C followed by the addition of 20 FM nonlabeled CTP, 5 min of further incubation at 30°C, and sarkosyl rinsing. G17 complexes were elongated to C18 and C23 complexes to U26, G29 or A32 by the addition of the appropriate NTPs at 50 FM and incubation for 5 min at 25°C.
The experiment in Fig. 8 used bead-attached templates generated by PCR from the pML20-55M template. Primers for generating the PCR fragments, preparation of the bead attached templates and assembly of preinitiation complexes on these templates were as described (16). Complexes were advanced to the desired positions as given in the Fig. 8 legend.
the chimera (28,29). We believe that the faint bands which were occasionally seen above the main products (see for example lanes 2, 6, 10 of Fig. 2) are not due to variations in the location of RNAse H digestion but rather to low levels of read-through RNA in the original transcription reaction (note for example the presence of 152-mers in the C151 preparation in ref. 10). This explanation is consistent with our results with U154 complexes, which were generated from C151 with low levels of UTP and which do not appear to have any read-through products. In that case, RNAse H cleavage with the chimeric oligonucleotides gave essentially a single band (see Fig. 3). Transcript truncation for the experiment in Fig. 8 was performed using RNase T1 digestion as described in the Fig. 8 legend.
Complexes treated with RNase H were tested for transcriptional competence by the addition of 50 FM NTPs at 37°C as described in detail in the figure legends. No chimeras or RNase H were added to the complexes synthesized on pML20-23like3 but these were also incubated at 37°C for 10 min to mimic conditions of the RNaseH cleavage experiment. These complexes were then immediately chased by further incubation with 50 FM NTPs at 37°C for 5 min. When indicated a final concentration of 18 Fg/ml of SII (or 29 Fg/ml, for the experiment in Fig. 8) was included in the chase reactions.
It should be noted that C17 and G18 complexes showed significant transcript truncation, in the absence of added SII, during a 5 min incubation at 25E C (compare lanes 2 and 3 with lane 1 in Fig. 4). Truncation was more pronounced during mockchase incubations at 37E C (compare lanes 2 and 3 with lanes 4 and 7). This transcript cleavage involved RNA polymerase, since it was completely inhibited by "-amanitin (data not shown) and was essentially confined to the C17 and G18 complexes. We did not observe breakdown with early elongation complexes walked to position +23 or later ( Fig. 4), nor did we see this effect with promoter distal complexes whose transcripts had been trimmed with RNase H to 18 nt or any other length (Figs 2 and 3). We believe that the source of the cleavage is a very low level of SII from the nuclear extract that was not removed by the sakosyl rinsing, since the addition of sarkosyl to the C17 or G18 complexes prevented RNA breakdown without inhibiting the elongation of the transcripts upon chase (data not shown). However, we cannot exclude the possibility that very early elongation complexes differ from later complexes in a way that renders their transcripts exceptionally labile to cleavage. If residual SII did remain in the promoter proximal C17 and G18 complexes (Fig. 4), this cannot explain their very low level of arrest relative to, for example, C23 promoter proximal complexes since promoter distal C18 complexes generated by RNase H cleavage, which show no RNA breakdown upon incubation, are also fully elongation compentent (Fig. 2).
It is important to note that after the sarkosyl washing step, transcript initiation factors are removed (see 26,30). Thus, once transcript cleavage was performed with oligos and RNase H, there was no need to remove these components during subsequent chase reactions since the transcript segments targeted by the oligos could not be regenerated.
All transcription reactions were phenol/chloroform extracted and (except for the  (10) reported on the exonuclease III footprints of several RNA polymerase II complexes stalled far downstream from the adenovirus major late promoter. That study employed a template, called pML20-23, which was constructed in such a way that RNA polymerase could be advanced in only two steps to position +151.

Samkurashvili and Luse
The sequence TTTGGGAAACCC on the nontemplate strand immediately downstream of position +151 allowed walking of the polymerase from C151 to U154, G157, or A160 by incubating with a subset of NTPs (see Fig. 1). All of the complexes stalled from +151 to +160 were found to be fully active in chase reactions (only about 5-10 % failed to restart transcription) and their exonuclease III footprints advanced synchronously along the template with the polymerase active site.
To study the same DNA sequence in a promoter-proximal context the region downstream of position +130 on plasmid pML20-23 was moved to the beginning of the initially transcribed region of the adenovirus major late promoter, creating the pML20-23like series of plasmids (16). In this case, the TTTGGGAAACCC sequence begins only 24 bases downstream of +1 (see Fig. 1). Surprisingly, the RNA polymerase II complexes corresponding to the +151 to +160 complexes on pML20-23, namely C23, U26, G29 and A32, were all found to be severely arrested on this template (16; also see Fig. 4). In order to determine the basis for the difference in elongation competence of these complexes, we cleaved the RNAs in the promoter-distal complexes so that they matched, in length and sequence, the transcripts in the promoter-proximal complexes.
Transcript Truncation in Promoter Distal Complexes-In our initial pilot experiments we used RNase A to cut the transcripts in C151 complexes made on the pML20-23 plasmid. After the RNase A digestion a prominent 21 nt long RNA remained in the ternary complexes. We found that these 21-mer complexes were predominantly arrested (data not shown). Encouraged by this observation, we employed a procedure  were separately annealed to body labeled RNA in the C151 complex (uncut RNA is shown in lane 1). The conditions of the RNase H digestion were optimized so that more than 90% of the input RNA was cut in 10 min at 37°C. We also confirmed that this incubation without the addition of RNase H did not deactivate the C151 complexes (see as compared to C18 in lane 2. We consider complex C18 fully active, since about the same fraction (14 %) of its parent complex, C151, cannot be chased after a 10 min incubation at 37°C without RNase H (Fig. 6A). In contrast to the C18 complex, complexes C23 and C45 were found to be arrested (73% and 53%, respectively) in this experiment. If the elongation factor SII was included in the chase reactions nearly all (> 95%) of the complexes could be elongated (data not shown; see Fig. 4).
The same experiment was repeated with complex U154 (Fig. 3). In this case the RNA was 3' end labeled and therefore only the 3' segment is detectable after the cleavage reaction. Note the disappearance of the 5' cleavage products from the top of the gel. We observed that the U21, U26 and U48 complexes were 56, 89 and 35% arrested, respectively. These results show that the promoter distal complexes that are fully elongation competent can be converted to complexes that are easily arrested, like their promoter proximal counterparts.
During these experiments we noticed that the extent of arrest for a given complex To sample more complexes containing different lengths of RNA after the RNase H cleavage reaction we repeated the above experiments on complexes G157 and A160. The sequences of these RNAs are shown in Fig. 1 and results are summarized in Fig. 5. A clear trend could be discerned from these tests. If the RNA in the ternary complex after cleavage was longer than about 20 nt and shorter than about 50 nt the complex was prone to arrest. However, complexes with shorter or longer RNAs were as elongation competent as noncleaved controls.
Promoter-proximal complexes-As noted above, other work in our laboratory has shown that RNA polymerase II complexes stalled from 20 to 32 nt downstream of transcription start on the pML20-23like template are arrested to a significant extent (16).
In order to quantitatively compare the elongation competence of promoter proximal and promoter distal complexes in the pML20-23 sequence context, it was necessary to subject the promoter proximal complexes to the same extensive 37EC incubations used in RNase H digestion of the promoter distal complexes. Figure 4 shows a representative experiment of this type with the pML20-23like3 template. The first position on this template where the polymerase complex can be stalled by NTP limitation is +17. Complex G17 was synthesized and body labeled with radioactive GTP. After sarkosyl rinsing this complex could be advanced to C18 by the addition of CTP. The G17 and C18 complexes were further incubated at 37°C for 10 min (in the absence of RNase H) to treat them identically to the RNase H cleaved samples (lanes [4][5][6][7][8][9]. A subsequent chase with all four NTPs was carried out in the presence or absence of the transcription elongation factor SII. We found that both G17 and C18 were fully active under these conditions, since only 2% of the complexes could not be chased (see lanes 5 and 8). However, we obtained a very different result when we assayed complexes with longer nascent RNAs. Complex C23 was synthesized with radioactive CTP, so that all of the label was incorporated in the 3'-most 6 nt of this RNA (see Fig. 1). After sarkosyl rinsing C23 was extended to U26, G29 or A32 by the addition of the corresponding nonlabeled NTPs at 25°C for 5 min. All complexes were incubated at 37°C for 10 min and then chased with all four NTPs at 37°C for 5 min. All of the complexes with 23 to 32 nt RNAs were found to be severely arrested in this assay (lanes 11, 14, 17 and 20). Inclusion of SII in the chase reaction resulted in the disappearance of all of the 23-32mer RNAs and the appearance of labeled cleavage products of about 10-20 nt (lanes 12, 15, 18 and 21). Almost no labeled run-off RNA was made in these SII-containing chase reactions since all of the label in the initial ternary complex RNAs was released by transcript cleavage and subsequent elongation took place with nonlabeled NTPs. We did confirm with uniformly labeled C23-A32 RNA that the transcript which remained in ternary complex was fully extended to run-off during chase in the presence of SII (data not shown).
The data on the elongation competence of halted pML20-23like3 complexes are summarized in Figure 5. There was a very close correspondence in arrest levels for the promoter proximal and promoter distal complexes with identical transcript lengths.
Those complexes with RNAs shorter than about 20 nt RNA were active, while complexes with 20 to 32 nt RNAs were predominantly arrested. We did not attempt to study promoter proximal complexes containing RNAs longer than 32 nt with the techniques used here, since all possible precursor complexes, from C23 to A32, are highly arrested. We have repeated this experiment several times with both complexes C151 and U154. The results are summarized in Figure 6B. The percent arrest values were normalized to the background that is measured in the absence of the chimeras. Arrest above background was seen with oligos hybridizing from 24 to 45 nt upstream of the 3' end. The greatest effect was observed with chimera 1 and the U154 complex, where the oligo hybrid extended from 27 to 36 nt upstream. Hybridization to locations more than 46 nt upstream did not significantly diminish elongation competence. This is consistent with our observation (Fig. 5) that complete elongation competence requires a transcript at least 51 nt long.
We hypothesized that arrest from oligo hybridization results from interfering with a mechanism that would otherwise block upstream translocation by the RNA polymerase. It was therefore important to demonstrate directly that the arrest seen in Fig. 6 did result from upstream translocation. In order to do this we produced U154 complexes labeled only in the last 3 nt and incubated these complexes with chimeric oligonucleotides 1, 2, 3 or 5 (or with no oligo, as control). The complexes were then treated with SII (Fig. 7). Mock-hybridized complexes (lanes 3 and 4), which were mostly elongation competent (see Fig. 6), gave predominantly short SII cleavage products as expected (32,33). The longer (about 6-13 nt) SII cleavage products in lane 4 presumably arose from the low level of arrested complexes. Hybridization with chimera 2, which did not reduce transcriptional competence in the experiment in Fig. 6, did not substantially change the partitioning of SII cleavage fragments as compared to the mock hybridization control. However, significant increases in the proportion of large SII cleavage products and coordinate diminution of short cleavage products was seen after hybridization with chimera 5 and particularly with chimera 1; the latter oligo was the most effective in inducing arrest in the Fig. 6 assay. Thus, the induction of arrest through oligonucleotide hybridization is accompanied by upstream translocation by the RNA polymerase.
Transcript truncation causes reduced elongation competence in another, completely different sequence context. In order to eliminate the possibility that the effects we have observed are somehow specific to the purine-rich initially transcribed sequence of pML20-23like3, we have also repeated our basic experiment using a template with a completely different initially transcribed sequence. In our earlier work we studied adenovirus major late promoter-based templates, such as pML20-42, which have pyrimidine-rich initially transcribed regions. RNA polymerases halted in promoterproximal locations on these templates generally show much lower levels of arrest than we observed for polymerases stalled in the promoter proximal region of the pML20-23like plasmids (16). A partial exception to this rule was provided by complexes stalled at +23 on pML20-42; about 40% of these polymerases could not extend their RNA chains after a 5 min chase under the conditions of our original study (16). We therefore constructed a new template, designated pML20-55M, which is diagramed in Fig. 8A.
On this template RNA synthesis will proceed to +23 when preinitiation complexes are incubated with CpA, CTP, UTP and ATP. After sarkosyl rinsing the complexes may be advanced to +54 in an ATP-less reaction, rinsed under native conditions and then walked to +77 with CTP, ATP and UTP. Note that treatment of complexes stalled at +77 with RNase T1 will result in truncation of the RNAs in the complex to 24-mers, which are identical to the RNAs present in CpA-primed, promoter proximal complexes stalled at +23 (underlined sequences in Fig. 8A). When the initial, CpA-primed transcription reaction on pML20-55M included radiolabled CTP, we could observe the resulting A24 complexes (lanes 1-4 of Fig. 8B). Exposure of these complexes to mock T1 digestion (incubation at 37EC for 5 min) before chase resulted in a somewhat greater level of arrest than we had seen in our earlier study (16). In the example shown in Fig.   8B, 67% of the A24 complexes became arrested (compare lanes 1 and 2). When polymerases were walked to +77 on the pML20-55M template, 32

DISCUSSION
RNA polymerase II complexes paused in promoter-proximal locations have a strong tendency to translocate upstream (16,26), an event which in many cases results in transcriptional arrest. This is an unexpected finding because the promoter proximal sequences in question do not cause arrest when traversed by the polymerase far downstream of transcription start. In this paper we show directly that this unusual property of promoter-proximal RNA polymerase II complexes may be recreated simply by shortening the nascent RNA within complexes that had transcribed to a promoterdistal location. We conclude that the transition to full elongation competence by RNA polymerase II is dependent on the synthesis of about 50 nt of RNA. Interestingly, this transition appears to be reversible. A major finding of the current work is that this hybrid-plus-sliding clamp model does not predict the behavior of RNA polymerase II complexes containing RNAs of less than 50 nt. It is important to stress that the sequence context we chose for most of our study (all experiments but those in Fig. 8) had been shown to provide no barriers to transcription in a promoter distal location. RNA polymerases stalled within this region remained transcriptionally active and did not show upstream translocated footprints (10). One would therefore predict that polymerases halted in this local sequence context have neither unfavorable interactions with downstream DNA nor particularly weak RNA-DNA hybrids. In reference to the latter point, it is worth noting that many of the complexes we studied have transcripts with GC-rich 3' ends. For example, the C23 complex (Fig. 4) and the corresponding C151 complex trimmed back to C23 (Fig. 2) were both severely arrested, and yet both complexes should contain RNA-DNA hybrids in which 6 of the 8 base pairs, including the four base pairs at the 3' end, are CG or GC (see Fig. 1).

Current Models for Transcript Elongation Complexes Do Not Predict Our Results-
Our results indicate that in addition to the RNA-DNA hybrid and the sliding clamp, The Transition by RNA Polymerase II from Initiation to Elongation-We had previously suggested (16,25) that the acquisition of lateral stability on the template, between roughly positions +20 and +50, is the last step in the commitment of RNA polymerase II to the transcript elongation phase of RNA synthesis. Among the changes which such a transition could reflect are the complete closure of the clamp structure which constrains the template within the ternary complex (7) and the formation of a transcription bubble with the dimensions characteristic of the elongation-committed polymerase (45). The net effect of these events presumably locks the polymerase in the stable elongation configuration. While such changes may indeed be characteristic of the initiation-elongation transition, the somewhat surprising result from the present study is that these changes are reversible. The response of the transcription complex to a particular length and sequence of transcript is essentially the same, regardless of whether that complex had previously passed through the initiation-elongation transition (Fig. 5).
A major unanswered question from our work concerns the molecular basis for the strong tendency of RNA polymerase II complexes with between 20 and 50 nt of nascent RNA to translocate upstream. What destabilizing effect makes the stabilizing influence of upstream RNA necessary? We speculate that once the nascent RNA reaches a critical length (which would be about 20 nt in the sequence context of the pML20- It is important to stress that the role of the transcript in preserving elongation competence must extend beyond the simple formation of secondary structure. This point can be most easily made by considering successive complexes in early elongation which do or do not display upstream translocation when stalled. For example, the A32 complex (Fig. 4) is predominantly arrested and upstream translocated (16); however, the C35 complex on the same template is neither upstream translocated nor arrested.
When RNA polymerase reaches position +35, the sequence of the RNA which has emerged from the polymerase should be 5'ACAGGAAGAGGAAGAAGC (assuming that 17 nt upstream of the 3' end remain protected by the polymerase-see 41). There is no apparent opportunity for secondary structure to form within this RNA. Similarly, while the G25 complex on the pML20-42 template is upstream translocated, the C27 complex on the same template occupies the normal template position for elongation competent complexes (26). In the case of the C27 complex, the RNA external to the polymerase should have the sequence 5'ACUCUCUUCC; again, one would not predict any secondary structure for this RNA.
Since the nascent transcript is probably not acting to stabilize early elongation complexes by forming secondary structure, a potential alternative mechanism would involve interaction of the RNA with some protein component of the transcription The significance of the distinctive properties of RNA polymerase II complexes with 20-50 nt nascent RNAs-Finally, it is important to recall that in those genes in which the transition to elongation is regulated, stalled RNA polymerases are typically found 20-50 nt downstream of transcription start; for example, between +21 and +35 on the Drosophila hsp70 gene (reviewed in 19) and at about +45 on the human hsp70 gene (23). We propose that the cell exploits the tendency of promoter proximal RNA polymerases to arrest in order to provide an opportunity for negatively acting transcription factors to interact with the enzyme. The concerted action of negatively (e.g. DSIF, NELF) and positively (e.g. P-TEFb, FACT) acting transcription factors could then regulate the transition into productive transcript elongation (48)(49)(50)(51). complexes formed on bead-attached, PCR-generated pML20-55M linear templates were incubated with CpA (which pairs with the template at positions -1/+1), α-32 P-CTP, UTP and ATP to obtain complexes stalled at +23. These promoter-proximal A24 complexes were rinsed with sarkosyl prior to testing for the elongation competence (lanes 1-4). A78 complexes (the precursors for promoter distal A24 complexes) were generated by incubating preinitiation complexes with the same set of NTPs used for generating the promoter proximal A24 complexes except that the CTP was not radiolabeled. After rinsing with sarkosyl, the nonlabled A24 complexes were walked to +54 with CTP, GTP and UTP, rinsed with transcription buffer to remove free NTPs, and then advanced to +77 with α-32 P-CTP, UTP and ATP. The promoter distal A24 complexes were obtained by digesting A78 complexes with RNAse T1 (10U of T1 in a 10 µl reaction for 5 min at 37EC) to liberate the upstream segment of the transcript (the 3'-most T1 cleavage site is indicated by the arrow in panel A). The T1-treated