Transcript cleavage in an RNA polymerase I elongation complex. Evidence for a dissociable activity similar to but distinct from TFIIS.

Stalled Xenopus RNA polymerase I (pol I) elongation complexes bearing a 52-nucleotide RNA were prepared by promoter-initiated transcription in the absence of UTP. When such complexes were isolated and incubated in the presence of Mg2+, the associated RNA was shortened from the 3'-end, and mono- and dinucleotides were released. Shortened transcripts were still associated with the DNA and were quantitatively reelongated upon addition of NTPs. The cleavage activity could be removed from the pol I-ternary complex with buffers containing 0.25% Sarkosyl. These findings indicate that a factor with characteristics similar to elongation factor TFIIS is associated with the pol I elongation complex. However, addition of recombinant Xenopus TFIIS to Sarkosyl-washed pol I elongation complexes had no effect, whereas it showed the expected effects in control reactions with identically prepared pol II elongation complexes. The results thus suggest the existence of a pol I-specific cleavage/elongation factor. I also report the sequence of a novel type of Xenopus TFIIS. The predicted amino acid sequences of the present and previously identified Xenopus TFIIS are less than 65% conserved. Thus, like mammalian species, Xenopus has at least two highly divergent forms of TFIIS.

The process of transcribing a gene can be divided into three main functional steps: initiation, elongation, and termination of transcription. For the eukaryotic ribosomal genes, transcribed by RNA polymerase I (pol I), 1 the cis-acting DNA elements comprising the promoter and the terminator have been identified for several species from yeast to man, and transacting protein factors required for transcription initiation and termination have been purified and in many cases molecularly cloned (1)(2)(3)(4)(5). On the other hand, relatively little is known about the process of transcription elongation by pol I. Due to the considerable length of the ribosomal genes and in light of the finding that at least some pol I even transcribe through the entire intergenic spacer (6), elongation of pol I transcription is likely to be an important process and can be expected to be regulated as well. Indeed, a previous study with Xenopus oocytes provided evidence that there are two types of ribosomal gene transcription, which differ in their elongation characteristics (7). The only pol I elongation factor identified to date is TIF-IC, a mouse pol I transcription factor that is involved in both initiation and elongation (8).
Much more is known about transcription elongation by pol II, and several protein factors affecting different steps of the elongation process have been identified. TFIIF and SIII, for example, stimulate the rate of elongation (9,10), whereas TFIIS (SII) promotes read-through of pol II at arrest or pause sites on the template (11). Others, like TFIIH and TFIIE, are required for promoter escape, an early step in elongation (12). The present study is concerned with TFIIS and TFIIS-like factors. cDNAs encoding TFIIS have been isolated from various species, including Xenopus (Refs. 13 and 14, and this study). TFIIS binds to pol II in the elongation complex and activates an intrinsic RNase activity that shortens the nascent transcript from the 3Ј-end (15,16) (reviewed in Ref. 17). The transcript cleavage is a hydrolytic reaction and liberates mono-and/or short oligonucleotides depending on the conformation and location of the paused elongation complex (18,19). The shortened transcript can be reextended in the presence of NTPs indicating that the 3Ј-ends of the shortened transcripts are still associated with the catalytic site of pol II. It is thought that RNA cleavage is important for the observed activation of readthrough by TFIIS possibly by allowing the arrested elongation complex to make several attempts to overcome an obstacle. Whether these in vitro observations reflect the in vivo role of TFIIS is an unresolved question. In yeast, TFIIS is not essential for viability, but cells lacking TFIIS are sensitive toward 6-azauracil (20,21).
TFIIS is generally considered to be specific for pol II. Thus, TFIIS purified from mouse cells did not affect pol I in nonspecific transcription assays (22), and in a different study TFIIS was found to bind to pol II but not to pol III (23). However, another early paper reported that partially purified yeast TFIIS stimulated both pol I and pol II (24). The question of the pol specificity of TFIIS has not been reinvestigated by looking at the more defined effects of TFIIS on paused elongation complexes. In search of pol I-specific elongation factors, I tested whether evidence for an involvement of TFIIS or a TFIIS-like factor in transcription by pol I could be obtained. I used the cleavage of nascent transcripts in a stalled pol I elongation complex as an assay for such a factor. My results show that there is indeed an activity associated with pol I ternary complexes that has similar functional characteristics as TFIIS but is distinct from TFIIS. initiation by pol I co-eluted from heparin-Sepharose between 0.4 and 0.8 M KCl (H-0.8 fraction). The protocol described by Dignam et al. (26) was used to prepare a nuclear extract from the Xenopus cell line. To remove endogenous nucleotides, the nuclear extract was precipitated twice with ammonium sulfate (0.33 g/ml of extract) and dialyzed against buffer D (0.1 M KCl, 20 mM HEPES, pH 7.9, 20% glycerol, 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride).
Transcription templates were immobilized on Dynabeads (Dynal Corp.) as described (27). For pol I transcription, plasmid pXlr245 (28) was biotinylated at a HindIII site 317 bp downstream of the transcription initiation site, followed by restriction with PvuII 1663 bp upstream of the initiation site in the vector pBR322. To create a heterogeneous population of pol II complexes, plasmid pSP-HSV106 containing a 3.58kilobase BamHI fragment of herpes simplex virus (29) in vector pSP64 was biotinylated at the HindIII site in the multiple cloning site, and a second cut was made with KpnI, resulting in two biotinylated fragments of 1.39 and 5.19 kilobases.
Immobilized DNA (100 -200 ng) was incubated in a 25-l reaction containing 10 l of H-0.8 fraction, 6 mM MgCl 2 , 90 mM KCl, 20 mM HEPES, pH 7.9, 10% glycerol, 0.1 mM EDTA, 2 mM dithiothreitol, and 100 g/ml ␣-amanitin. After 15-30 min at room temperature, 50 M ATP, 50 M GTP, 0.33 M [␣-32 P]CTP (3000 Ci/mmol), and 4 mM phosphocreatine were added to initiate transcription. (Final concentrations of the components in the reaction are given.) After 15 min, the reaction was chased with 0.1 mM each of ATP, GTP, and CTP and 100 g/ml heparin. pol II transcription reactions were similar, except that the MgCl 2 and KCl concentrations were 10 and 60 mM, respectively, and no ␣-amanitin was present. After another 15 min, the template beads were washed four times in 100 l of wash buffer as indicated in the figure legends. Wash buffer consisted of 23 mM HEPES, pH 7.9, 90 mM KCl, 10% glycerol, 0.1 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 100 g/ml bovine serum albumin, and 0.05% Nonidet P-40. Magnesium buffer was wash buffer plus 8 mM MgCl 2 , EDTA buffer was wash buffer plus 1 mM EDTA, and Sarkosyl buffer was EDTA buffer plus 0.25% Sarkosyl. In the reactions in which the isolated complexes were incubated with recombinant Xenopus TFIIS, two washes in Sarkosyl buffer were followed by two washes in EDTA buffer. Washed beads were either directly processed for analysis of the labeled RNA or incubated in a volume of 25 l under the conditions indicated in the figure legends. After treatment with proteinase K/SDS and extraction with phenol/chloroform, the RNA was concentrated by precipitation with ethanol and analyzed on 12% (acrylamide:bisacrylamide, 29:1) polyacrylamide-sequencing gels. For the analysis of the cleavage products, the beads were washed three times in EDTA buffer and one time in a buffer containing 50 mM KCl, 10 mM HEPES, pH 7.9, and 0.5 mM dithiothreitol. They were incubated in 10 l of the same buffer supplemented with 8 mM MgCl 2 , and the reaction was stopped by the addition of 15 l containing 0.25% SDS, 12 mM EDTA, 250 g/ml proteinase K, and 250 g of carrier RNA. After extraction with phenol/ chloroform and chloroform, the sample was dried in a SpeedVac concentrator. RNA samples were analyzed on 25% (25:3) polyacrylamidesequencing gels.
Cloning and Expression of Xenopus TFIIS-To isolate a cDNA clone encoding the Xenopus homologue of TFIIS, a degenerate oligonucleotide encoding QTRSADEP (CAAACTCGT(A/T)(G/C)TGCNGATGA(A/G)CC) and a degenerate oligonucleotide complementary to the coding sequence for CGNRWKF (AACTTCCATC(G/T)(A/G)TTNCC(A/G)CA) were used to amplify from Xenopus genomic DNA a 71-bp fragment using Taq DNA polymerase (see Fig. 6). The 71-bp product was labeled by additional rounds of amplification with 32 P-5Ј-end-labeled primers and used to screen a Xenopus liver cDNA library under high stringency hybridization conditions. A positive clone was isolated, sequenced, and found to contain an open reading frame of 292 amino acids with high similarity to other reported TFIIS sequences.
For expression of recombinant proteins, a BglII site was added upstream of the protein-coding sequence by oligonucleotide-primed synthesis using Pfu DNA polymerase (Stratagene). The fragment from the BglII site to a natural HindIII site at nucleotide position 268 was ligated to a HindIII-EcoRI fragment containing the remainder of the coding sequence plus 164-bp downstream flanking sequences and cloned into the pET28a(ϩ) expression vector (Novagen). The sequence of the in vitro synthesized section was reconfirmed. Deletion mutants were created by using naturally occurring restriction sites. Deletion of a StuI-SspI fragment replaced 59 amino acids from the C terminus with the four residues DLGL (mutant ⌬1), and deletion of a PstI-PstI fragment led to the internal deletion of residues 45-171 (mutant ⌬2). Hexahistidine-tagged proteins were expressed and purified as described (31) except that lysis buffer contained 20 mM imidazole and no DNase I. Likewise, the nickel-nitrilotriacetate-agarose column was equilibrated in lysis buffer containing 20 mM imidazole, and Xenopus TFIIS was eluted with 250 mM imidazole.

Formation of Stalled Xenopus pol I Elongation Complexes-To generate stalled
Xenopus pol I elongation complexes, a minigene construct containing the ribosomal gene promoter and the first 115 bp of the ribosomal precursor-coding region was immobilized on magnetic beads by a streptavidinbiotin linkage ( Fig. 1) and transcribed in a chromatographic fraction that supported specific transcription initiation by pol I. The first U residue at the 5Ј-end of the Xenopus 40 S ribosomal precursor is at position 17, followed by a U triplet at positions 53-55. Transcription in the absence of UTP was therefore expected to yield stalled elongation complexes carrying transcripts of 16 and, due to unavoidable traces of UTP in the reaction, 52 nt in length. These transcripts and the corresponding stalled complexes were termed C16 and G52 according to the type and position of the last incorporated nucleotide (see Fig. 1). Analysis of the template-associated RNA showed that the C16 and G52 RNA were indeed formed ( Because a high concentration of ␣-amanitin was present in all reactions and because in the transcription system used the formation of specific run-off products was dependent on an intact pol I promoter (data not shown, see also Ref. 27), the C16 and G52 transcripts are interpreted to represent stalled pol I elongation complexes. Lanes 3 and 4 show that the G52 complex was still elongationcompetent. If UTP was added to isolated G52 complexes, they elongated for three nucleotides to position U55 (lane 3). If all four NTPs were added, the label was quantitatively chased into the 317-nt run-off (lane 4).
Sarkosyl Removes an Endogenous TFIIS-like Activity-In the reactions analyzed in Fig. 2 (lanes 2-4), the template beads were washed in Sarkosyl buffer; this process stripped the ternary complex of most if not all associated proteins. If the beads were washed in magnesium buffer, the template-associated RNA was found to be shortened by about 10 nt (Fig. 2, lane 5) suggesting that a cleavage activity was active during the washing in magnesium buffer (which takes 5-10 min) and that this activity was removed by the Sarkosyl buffer. This shortened transcript was still in an active elongation complex, since it could be reextended to the G52 RNA upon incubation with ATP, CTP, and GTP (lane 7). The finding that the G52 complex could be regenerated also demonstrated that the G52 RNA was shortened at the 3Ј-end. Reextension reactions in the presence of only two of these three NTPs are shown in lanes 8 -10. Based on these "walking" experiments, the major cleavage product in the magnesium-washed complexes was found to be the G43 complex. Extension of the G43 complex with ATP and GTP resulted in the A44 complex (lane 8), and extension with ATP and CTP yielded the C46 complex (lane 9). Incubation of the G43 complex in the presence of GTP and CTP showed some persisting G43 RNA as expected from the sequence, whereas other transcripts were shortened back to G35 (lane 10). In addition, some extension to what appears to be position C51 was observed, probably due to a small amount of contaminating ATP (lane 10). If all four NTPs were added to the complexes washed in magnesium buffer, all the cleavage products were chased into run-off RNA (lane 11). Prolonged incubation in magnesium buffer without NTPs led to further degradation of the template-associated transcripts (lane 6). None of these shortened, and partially reextended transcripts were released from the template (lanes 13, 14, and 16 -19). Transcript release from the ternary complexes was observed only when chased to the end of the template (lanes 12 and 15).
To test whether the endogenous cleavage activity requires divalent cations, complexes were first washed in EDTA buffer. course identifies the G48/49 and G43 RNA as major metastable intermediates in the cleavage reaction. Under the present labeling conditions the shortest G52-derived cleavage product that could be detected on the autoradiographs was about 20 nt long. Addition of ATP, GTP, and CTP again led to reextension of the shortened transcripts to the G52 RNA (lane 6). The finding that transcripts in a pol I ternary complex can be shortened and reextended in a magnesium-dependent reaction suggests that the catalytic site of pol I is involved in the process. Consistent with this notion is the result that, like pol I transcription elongation, the cleavage reaction was resistant to high concentrations of ␣-amanitin (200 g/ml, lane 12). Interestingly, the amount of RNA in the C16 complex decreased only slightly during incubation in the magnesium buffer, and no cleavage products were detected. Additional data showed that elongation of the C16 complex was also less quantitative; whereas about 70% of the C16 RNA was elongated to U17 RNA upon addition of UTP (data not shown), this was significantly less than the almost 100% of the G52 complexes that reelongated (see Fig. 2, lanes 3 and 4). The basis for the apparent difference between the G52 and C16 complex remains to be investigated.
To determine the Sarkosyl concentration required for dissociation of the cleavage activity, stalled pol I complexes were washed with EDTA buffers containing increasing concentrations of Sarkosyl and then incubated in magnesium buffer for 5 min. The result shows that a slight inhibition of the cleavage was observed after washing in 0.1% Sarkosyl (Fig. 3B, lane 6), whereas transcript shortening was fully abolished after washing in a buffer containing 0.25% Sarkosyl (lane 7).
The Transcript-shortening Reaction Liberates Mono-and Dinucleotides-To learn more about the mechanism of the cleavage reaction, it was important to identify the released products. The reactions were therefore analyzed on high percentage polyacrylamide gels, which had previously been used to separate mono-and dinucleotides (32). My standard transcription protocol employed a chase with ATP, CTP, and GTP after labeling of the initiated transcripts with [␣-32 P]CTP at low NTP concentrations. Therefore, the G52 RNA was expected to be labeled predominantly near its 5Ј-end. Because shortening of such RNA from its 3Ј-end would not yield labeled cleavage products until late in the reaction, I also performed transcription reactions without the chase. Fig. 4A, lane 2, shows that these low nucleotide conditions caused most of the elongation complexes to stall before C16. Based on the template sequence (see Fig. 1), the two major additional stalled complexes were tentatively identified as C10 and C14. Incubation in magnesium buffer led to a decrease in the amount of RNA associated with C10 and C14 and the concomitant appearance of a cluster of bands that migrated slower than the CTP/CMP markers (lanes 2-5). Note that again the C16 complex was relatively unaffected. If "chased" complexes were incubated in magnesium buffer, these same bands also appeared, but as predicted, later in the reaction (lanes 6 -9). The slowest of these cleavage products co-migrated with the pCpG marker (lanes 14 and 16). Note that pGpC and pCpG cannot be distinguished in this gel system. The faster bands in the cluster might represent pCpC and pApC, which are known to migrate ahead of pCpG (32). All of these dinucleotides could be generated from both G52 and C14 RNA. The identification of these cleavage products as 5Ј-phosphorylated dinucleotides was supported by the finding that all bands disappeared upon digestion of the sample with  5, 6, 11, and 12 as indicated. Note that the band comigrating with the CMP marker is increasing in intensity during the endogenous cleavage reaction (lanes 1-4) and that this band is not formed during pyrophosphorolysis (lanes 7-10). The smear in lane 9 is due to incomplete washing of the DNA beads. calf intestinal phosphatase, and a new cluster of bands with much reduced mobility was generated (lanes 17 and 18). This dramatic shift in mobility upon removal of the 5Ј-phosphate from dinucleotide diphosphates has previously been described (32).
Due to the high amount of input [␣-32 P]CTP, some residual CTP as well as inorganic phosphate was present in all lanes even after extensive washing of the beads in EDTA buffer. However, the time course clearly showed that CMP, which runs slightly behind CTP, was absent at time 0 (Fig. 4B, lane 1) and was formed during incubation of the cleavage reactions (Fig.  4B, lanes 2-4). The kinetics of appearance of the CMP band paralleled the appearance of the dinucleotide bands in both the chased and the not chased reactions (see Fig. 4A). These comigration and phosphatase digestion data thus suggest that nucleoside 5Ј-monophosphates and 5Ј-phosphorylated dinucleotides are the main products of the cleavage activity associated with stalled pol I elongation complexes.
Pyrophosphorolysis-The data described in the previous section indicate that pyrophosphorolytic cleavage of the nascent transcript does not make a significant contribution to the observed transcript shortening. Nevertheless, I wondered whether RNA in stalled pol I complexes would undergo pyrophosphorolysis and how this reaction would compare with the shortening by the endogenous Sarkosyl-sensitive activity. Pyrophosphorolysis is known to remove nucleoside 5Ј-monophosphates from the 3Ј-end of nascent RNA in ternary complexes to produce NTPs. Stalled complexes that had been washed in Sarkosyl buffer were incubated in the presence of 2 mM sodium pyrophosphate and Mg 2ϩ . Fig. 3A, lanes 7-9, shows that pyrophosphate resulted in rapid shortening and disappearance of the G52 RNA. With longer incubation times, even the C16 RNA underwent pyrophosphorolytic degradation (Fig. 4A, lanes 19 -22). Pyrophosphorolytic shortening went through different metastable intermediates than the endogenous cleavage reaction (compare lane 8 in Fig. 3A with lanes 3-5). Furthermore, analysis on a high percentage polyacrylamide gel showed that dinucleotides (Fig. 4A, lanes 20 -22) and CMP (Fig. 4B, lanes  8 -10) were not formed. Since Sarkosyl buffer was more efficient in removing the input [␣-32 P]CTP (Fig. 4B, lane 7), this analysis also revealed that, as expected, CTP was regenerated in the presence of pyrophosphate (Fig. 4B, lanes 8 -10). Thus, the combined results strongly indicate that the endogenous cleavage reaction is hydrolysis and not pyrophosphorolysis of the RNA.
TFIIS Does Not Act on the Stalled pol I Complex-The results described so far suggest that an activity that is very similar to the pol II elongation factor TFIIS is associated with the pol I elongation complex. To test whether this activity is identical to TFIIS, I investigated the effect of recombinant Xenopus TFIIS on stalled pol I elongation complexes. As a positive control, I prepared a heterogeneous population of stalled pol II elongation complexes by transcription of an immobilized plasmid in a Xenopus nuclear extract in the absence of UTP, followed by washing of the template beads in Sarkosyl buffer (Fig. 5, lane 1). The RNA associated with these stalled pol II complexes ranged in size from about 10 to 60 nt. If all four NTPs were present during transcription, high molecular weight RNA was formed (lane 2). The synthesis of both stalled and run-off transcripts was sensitive to 5 g/ml ␣-amanitin, indicating that they were the result of transcription by pol II (lanes 3 and 4). The RNA in the stalled pol II complexes was not affected by a 1-h incubation in EDTA or magnesium buffer (lanes 7, 8, and 14). Incubation in magnesium buffer containing Xenopus TFIIS led to a shortening of the template-associated transcripts (lanes 10, 15, and 16), whereas the same amount of Xenopus TFIIS in EDTA buffer had no effect (lane 9). The specificity of the Xenopus TFIIS-induced transcript shortening was demonstrated by testing two deletions of the recombinant protein. Both a deletion of the C-terminal 59 amino acids containing the zinc ribbon (33) as well as a large internal deletion abolished the transcript cleavage activity (lanes 11 and 12). Furthermore, Xenopus TFIIS-induced transcript cleavage was sensitive to 5 g/ml ␣-amanitin (lane 17). These results show that the recombinant Xenopus TFIIS used in the present study was active and had the expected characteristics and effects on Sarkosyl-washed, stalled pol II elongation complexes. If Xenopus TFIIS or one of the mutant proteins was added to identically prepared pol I elongation complexes, no effects were seen (lanes 18 -21). DISCUSSION TFIIS is a transcription factor that was originally identified as an activity that stimulates nonspecific transcription by purified pol II (34). It was later classified as an elongation factor because it suppresses pausing and increases the yield of long  15 and 19) or 100 ng (lanes 9 -12, 16, 17, 20, and 21). Lanes 5 and 13, DNA size marker as in Fig. 2. The sizes of the five shortest fragments are 9 (visible in lane 13 only), 15,26,34, and 67 nt. The gel shown in lanes 1-4 was run longer than the gels shown in lanes 5-12 and 13-21. transcripts in specific transcription reactions (23,35). TFIIS exerts its function in elongation by helping stalled or arrested pol II to read through various transcriptional blockages (11,36,37). In stalled and arrested elongation complexes, TFIIS stimulates the cleavage of the nascent transcripts at the 3Ј-end and a concomitant backward movement of the catalytic site of pol II (15,16). This cleavage reaction is thought to be important or even required for efficient read-through at these arrest sites.
It is reasonable to postulate that the transcription of ribosomal genes also requires an elongation factor. As discussed in the Introduction, the available data appeared insufficient to fully rule out that TFIIS itself would also serve as an elongation factor for pol I. The stimulation of the pol-associated transcript cleavage activity is a more diagnostic assay for TFIIS than the stimulation of nonspecific transcription. I therefore investigated whether TFIIS would stimulate the cleavage of nascent transcripts in a stalled pol I elongation complex. My data show that recombinant Xenopus TFIIS, when added to Sarkosyl-washed pol I elongation complexes that stalled after transcription of a 52-nt RNA, did not stimulate cleavage of this transcript. This negative result was substantiated by control experiments with Sarkosyl-washed pol II elongation complexes, where the expected effects of Xenopus TFIIS were readily detected. Furthermore, Xenopus TFIIS showed all the known characteristics in these pol II controls, like Mg 2ϩ dependence, ␣-amanitin sensitivity, and dependence on the zinc ribbon domain near its C terminus.
Whereas the present results make an involvement of TFIIS in pol I transcription very unlikely, they also show that a different activity that is similar to TFIIS is present in the pol I elongation complex. Like TFIIS, this activity shortens the nascent RNA from the 3Ј-end releasing mono-and dinucleotides. The shortened transcripts are still associated with the catalytic site of pol I, since they are readily reextended upon addition of the appropriate NTPs. Finally, the cleavage activity is removed from the pol I elongation complex with 0.1-0.25% Sarkosyl; the same conditions also remove TFIIS from the pol II elongation complex (38). Unlike the TFIIS-induced RNA cleavage reaction in pol II ternary complexes, the pol I-associated cleavage reaction is resistant to high concentrations of ␣-amanitin, a finding that is consistent with an involvement of the catalytic site of pol I in the transcript shortening.
In stalled or arrested pol II elongation complexes a low level of RNA cleavage is taking place even in the absence of an auxiliary factor like TFIIS, indicating that the nucleolytic activity resides in pol II (15)(16)(17)39). For pol III this type of hydrolytic cleavage is the only type that has been observed so far, and a TFIIS-like, dissociable activity like the one described in this study has not been identified in pol III complexes (30). In the present experiments I did not detect an intrinsic activity of pol I that would be active without an auxiliary factor. It can be expected, however, that pol I by itself would also show some nucleolytic activity if the sensitivity of the assay was increased. In any case, it is clear that for both pol II and pol I, TFIIS or a TFIIS-like factor is required for efficient transcript cleavage.
After submission of the original version of this manuscript, the identification of a yeast activity that shortens nascent pol I transcripts from the 3Ј-end was reported (40). Whereas this yeast factor and the present Xenopus activity clearly seem to be related, the available data are insufficient to decide whether they represent homologous protein factors. It will be interesting to determine the Mg 2ϩ requirement and the Sarkosyl sensitivity of the yeast factor as well as the size of the liberated cleavage products. Apparent differences lie in the chromatographic behavior and in the finding that the yeast activity can be detected even in the presence of nucleotides.
Sequences encoding Xenopus TFIIS were recently reported by Kugawa et al. (13) and Plant et al. (14). The latter paper identified two genes for Xenopus TFIIS, termed xTFIIS.oA and xTFIIS.oB, which were interpreted to represent the two gene copies present in the two homologous genomes of the tetraploid Xenopus laevis. The two predicted proteins are 91% identical. The deduced protein sequence reported by Kugawa et al. (13) is 100% identical to xTFIIS.oA. Interestingly, the sequence of the cDNA isolated for the present study, which I will refer to as xTFIIS.l, is very different from both xTFIIS.oA and xTFIIS.oB (Fig. 6). The predicted xTFIIS.l and xTFIIS.oA/B proteins are only 68 -70% identical in the 82-amino acid long N-terminal domain and 72-74% identical in the 175-amino acid C-terminal domain. The regions between these two domains are not conserved and of different lengths (46 versus 35 amino acids). The similarity between these two Xenopus proteins is thus about the same as between TFIIS from different vertebrate species. Using reverse transcription-polymerase chain reaction, I found that mRNAs encoding xTFIIS.oA/B and xTFIIS.l are both expressed in a Xenopus tissue culture cell line. 2 Xenopus thus has two highly divergent forms of TFIIS, which do not appear to reflect the duplication of its genome during evolution. This finding has a precedent in mammalian species, where in addition to the general form of TFIIS a testis-specific type was identified (41,42 (14), and xTFIIS.l (this work) is shown. Residues shown in the consensus line are conserved among all three sequences; ϩ, indicates a conservative substitution. The sequences shown in italics are not conserved between xTFIIS.oA/B and xTFIIS.l, and their alignment is arbitrary. The underlined residues near the C terminus were used to design the degenerate oligonucleotides described under "Experimental Procedures." It will be interesting to see whether the present TFIIS-like factor belongs to the TFIIS family of proteins. In addition, it will be important to investigate whether and how it changes the properties of the pol I elongation complex. The experiment shown in Fig. 2 shows that in the presence of NTPs a pol I elongation complex is able to elongate to the end of the template regardless of whether it contains this factor (lanes 4 and  11). Furthermore, on the template that was used no intrinsic arrest or pause sites were uncovered during elongation of Sarkosyl-washed complexes. The purification and cloning of this factor will be necessary to investigate its role, if any, in transcription elongation by pol I.