RNA expression from a site-specific non-LTR retrotransposon microinjected into Xenopus oocytes

Genetica 104: 67–76, 1998. © 1998 Kluwer Academic Publishers. Printed in the Netherlands.

67

RNA expression from a site-specific non-LTR retrotransposon microinjected into Xenopus oocytes Genevi`eve Pont-Kingdon1, Enxi Chi2 , Shawn Christensen & Dana Carroll Department of Biochemistry University of Utah School of Medicine Salt Lake City, Utah 84132, USA; 1 Present address: Huntsman Cancer Institute 15N, 2030E, room 3260, Salt Lake City, UT 81112-5330, USA: 2 Present address: University of Texas Southwestern Medical Center. Dallas, TX 75235, USA Accepted 22 June 1998

Key words: non-LTR retrotransposon, 3! RNA ends, polyadenylation Abstract Tx1L is a site-specific non-LTR retrotransposon (NLR) that has been identified in the genome of Xenopus laevis. Using microinjection into Xenopus oocytes, several aspects of RNA expression by these elements were investigated. With constructs carrying various parts of the element we saw no evidence of promoter activity, unlike what has been shown for several other elements of this class. Tx1L transcription was induced by linking a whole element to a promoter that is active in oocytes. Among the RNAs produced, about half had 3! ends located near the end of the element, suggesting that instructions for 3! end formation are encoded in the element or its target. Deletion of the 3! UTR of Tx1L and of surrounding target sequences indicated that these regions are not required for termination or processing of the RNA. PolyA or very A-rich sequences were added at these 3! ends, despite the absence of canonical polyA addition signals. A significant proportion of non-A residues was found in the 3! untemplated tails, and this is reminiscent of non-templated insertions often found at the 3! junction of new genomic copies of some NLRs.

Introduction Retroelements, transposable elements that move via RNA intermediates, are ubiquitous and repetitive components of essentially all genomes. The subclass of non-LTR retrotransposons (NLRs) is composed of mobile elements that are dispersed broadly in the genomes of most eukaryotes (for review see Hutchison et al., 1989; Eickbush, 1992). Although the sequences of NLR elements may vary considerably, they are united in carrying coding sequences for a protein homologous to reverse transcriptase (RT) and in lacking the long terminal repeats (LTRs) that are found in the retrovirus-like retrotransposons. Phylogenetic analysis of NLR RT domains confirms their classification as a distinct group of related sequences (Xiong & Eickbush, 1990; McClure, 1991). Studies of examples from many different species have confirmed that NLR elements move via an RNA intermediate (Evans & Palmiter, 1991; Jensen & Hei-

dmann, 1991; Pelisson, Finnegan & Bucheton, 1991; Segal-Bendirdjian & Heidmann, 1991; Kinsey, 1993) and have elucidated various aspects of the transposition process that differ substantially from that of the LTR-containing retrotransposons. NLRs typically contain two long open reading frames (ORFs), although these are fused into a single ORF in some elements. The product of the first ORF (ORF1p) codes for an RNA binding protein (Dawson et al., 1997; Kolosha & Martin, 1997; Pont-Kingdon et al., 1997) found in the cytoplasm associated with element RNA (Martin, 1991; Hohjoh & Singer, 1996). The protein encoded by the second ORF (ORF2p) has RT and endonuclease activities (Xiong & Eickbush, 1988; Gabriel & Boeke, 1991; Ivanov et al., 1991; Mathias et al., 1991; Luan et al., 1993; Feng et al., 1996). Very interestingly, the RT encoded by non-LTR retrotransposons is able to prime RNA-dependent DNA synthesis from a nick in the target site created by the endonuclease activity of the protein (Xiong & Eick-

MENNEN/SCHRIKS:DISK/CP: Pips Nr.:184080; Ordernr.:235591-mc (genekap:bio2fam) v.1.1 gene464.tex; 3/11/1998; 10:11; p.1

68

Figure 1. A Elements of the Tx1 family. Tx1D is an apparent DNA-based transposon, with short terminal inverted repeats (open triangles), two types of internal tandem repeats – PTR-1 (white boxes) and PTR-2 (gray boxes) – and unique sequences flanking the repeats (hatched boxes). Composite Tx1C elements consist of Tx1D with a Tx1L element inserted into one PTR-1 repeat. The dark triangles represent the duplication of 23 bp of PTR-1 target sequence flanking the insertion. Tx1L (Garrett, Knutzon & Carroll, 1989) is a non-LTR retrotransposon that carries two open reading frames: ORF1 is loosely related to retroviral gag genes, while ORF2 is homologous to retroviral pol genes in the reverse transcriptase domain. The offset between the boxes denoting ORF1 and ORF2 indicates that they are in different reading frames. The 5! and 3! untranslated regions (UTRs) of the Tx1L element are also indicated. B. Diagrams of expression constructs. All of the constructs are in the vector pMT3 (Swick et al., 1992), which carries an adenovirus promoter (light hatched box). The bottom four constructs also contain the tripartite leader and a hybrid intron (dark hatched box) from the vector, but these were deleted in pMTx1L. Downstream of the insertions is a DHFR gene followed by an SV40 polyadenylation signal (gray boxes). The transcription initiation site is indicated by a bent arrow. Segments of Tx1L and its target are designated as in part A. The white box in pM2 indicates the hemagglutinin epitope introduced into ORF2. The star in pM2-5mt, !3! UTR, and !ORF2p indicates the position of the myc tags. In pMTx1L, the downstream target sequence extends to the SnaB I (S) site of PTR-1, situated 253 bp from the target duplication; in pM2, pM2-5mt, and !ORF2p, it extends to the Hind III (H) site, 214 bp from the target duplication.

bush, 1988; Luan et al., 1993; Feng et al., 1996). Similarly, the RT encoded by a yeast group II intron nicks its target and initiates reverse transcription from the nick (Zimmerly et al., 1995). These results support a target DNA-primed mechanism of transposition in which reverse transcription initiates at the 3! end of the NLR RNA template.

Full-length element RNAs have been detected in many cases (Mizrokhi, Georgieva & Ilyin, 1988; Skowronski, Fanning & Singer, 1988; Chaboissier et al., 1990; Packer, Manova & Bachvarova, 1993; Schumann et al., 1994; Vaury et al., 1994), and the RNA of the Drosophila I factor is found only in cells which support transposition (i.e., germ line cells of dysgenic females) (Chaboissier et al., 1990). These RNAs are presumably generated from functional internal promoters for RNA polymerase II that have been identified in the 5! untranslated region (5! UTR) of several NLRs (Mizrokhi, Georgieva & Ilyin, 1988; Nur, Pascale & Furano, 1988; Swergold, 1990; Minchiotti & Di Nocera, 1991; Minakami et al., 1992; McLean, Bucheton & Finnegan, 1993; Schumann et al., 1994). Studies of the structure of the 3! termini of natural transcripts are less numerous. Element RNAs are often found in the polyA+ fraction, but many NLRs lack canonical polyadenylation signals, and it is not clear whether transcripts are subject to post-transcriptional polyA addition like cellular mRNAs. In the frog, Xenopus laevis, Tx1L has been identified as a family of repeated sequences with homology to NLRs (Garrett & Carroll, 1986; Carroll, Knutzon & Garrett, 1989; Garrett, Knutzon & Carroll, 1989) (Figure 1A). Tx1L is found in the genome only in association with another family of repeated sequences, Tx1D, which has structural characteristics of a DNAbased, cut-and-paste transposable element (Garrett & Carroll, 1986). Each Tx1L element is flanked by duplicated copies of a 23-bp target sequence, which is found within tandem 400-bp repeats (called PTR-1) of Tx1D (Figure 1A). The composite element (i.e., Tx1L within a Tx1D sequence) is called Tx1C. Because of the similarities of Tx1L to other NLRs, we assume that it is an independent element with a high level of target specificity, although it is still possible that Tx1C is the canonical transposon (Garrett, Knutzon & Carroll, 1989). Each of the above features is shared by a second family of X. laevis elements, called Tx2 (Garrett, Knutzon & Carroll, 1989; D. Carroll et al., unpublished results). Our long-term goal is to investigate in detail the Tx1L transposition mechanism by inducing new transposition events in a convenient laboratory setting. In this paper we address questions regarding Tx1L RNA expression by microinjecting various constructs into Xenopus oocytes. The element does not carry a promoter that is active in oocytes. When transcription is forced, RNA processing creates A-rich 3! ends some of which may be appropriately located to support

gene464.tex; 3/11/1998; 10:11; p.2

69 transposition. The 3! untemplated tails often contain non-A residues that may reflect a novel mechanism of 3! end formation (Furano et al., 1986).

Materials and methods Constructs For construction of pMTx1L (Figure 1B), a fully sequenced Tx1L element (Garrett, Knutzon & Carroll, 1989) was excised from clone λB10 by cleavage with SnaB I, which cuts in the flanking PTR-1 repeats. The resulting 7.3 kb fragment was cloned at the Cla I site of the expression vector pMT3 (Swick et al., 1992). Sequences situated between the adenovirus promoter contained in the vector and the first nucleotide of Tx1L were deleted using a modified recombinant PCR method (Pont-Kingdon, 1994). All oligonucleotides were synthesized at the University of Utah DNA/Peptide Core Facility. Some mapping of the 3! ends of Tx1L RNA was performed on RNA expressed from the plasmid pM2, which was constructed as follows. First, two consecutive Hind III fragments that contain the complete ORF2 of Tx1L, the 3! untranslated region (3! UTR), and 200 bp of the PTR-1 target site were subcloned at the Hind III site of pBluescript KS+ (Stratagene). Serial deletions from the 5! end with exoIII and S1 nuclease brought the first ATG of ORF2 close to the 5! end of the insert, creating an Nco I site between the plasmid and the insert. This construct is called pBORF2. A double-stranded oligonucleotide encoding the hemagglutinin epitope (YPYDVPDYA) (Niman et al., 1983) was inserted into the Xma I site of ORF2 in order to follow expression of the protein in other studies. The tagged ORF2 was excised from pBluescript using Sac II and Cla I and, after blunting the Sac II site, introduced between the Sma I and Cla I sites of pMT3. The construct pM2-5mt was obtained in several steps, beginning with pBORF2. The S1-blunted Sac II-Nco I fragment of this clone was ligated with the S1-blunted Sal I-Nco I fragment of pKSMT6 (Roth, Zahler & Stolk, 1991); this linked ORF2 in frame with the first five of the six myc tags contained in pKSMT6. The Cla I fragment of this clone was cloned at the Cla I site of pMT3. PCR was used to create the deletion of the 3! UTR and PTR-1 sequences in the construct !3! UTR; one primer was homologous to Tx1L from nucleotide 6657 to 6678 (Garrett, Knutzon

& Carroll, 1989); the second primer was the chimera 5! GGGTCGACGGTATTTAAAGTGCACTTAAA 3! containing the new junction between plasmid sequences (5! half) and 16 nucleotides of Tx1L ending at the TAA ORF2 stop codon (underlined). Using pM2-5mt as a template, a 367-bp PCR product was obtained. This fragment was digested with Xma I (position 6717 of Tx1L) and Sal I and used to replace the original Xma I-Sal I fragment from pM2-5mt. The construct !ORF2p was obtained by S1 treatment and ligation of a partial Xho I digest of pM2-5mt. In the resulting clone, ORF2 is disrupted at the Xho I site 156 bp (52 amino acids) from the initiation codon. The putative protein is 70 amino acids long; it contains a short portion of the endonuclease domain, but none of the region of homology with reverse transcriptase. Oocyte injection Xenopus laevis females used in these studies were obtained from Xenopus I (Ann Arbor, MI). Preparation of stage V and VI oocytes and nuclear microinjection of DNA constructs were performed largely as described elsewhere (Maryon & Carroll, 1989). Oocytes were incubated at 19 ◦ C, in OR-2 medium (Wallace, 1973), for 3 days. RNA procedures Total RNA was extracted after homogenization of the oocytes in a buffer containing proteinase K and SDS (Sambrook, Fritsch & Maniatis, 1989); the ethanol precipitation preceding the LiCl precipitation was omitted. PolyA+ RNA was selected on an oligodTcellulose column following instructions from the manufacturer (Collaborative Biomedical Products). RNAs were analyzed in 1% agarose gels containing 220 mM formaldehyde (Tsang et al., 1993). To permit rapid detection, ethidium bromide was usually added directly to the RNA samples at a final concentration of 50 µg/ml (Ogretman et al., 1993). Northern blots were obtained by capillary transfer of the RNA onto positively charged nylon membranes (Boehringer Mannheim) in 20 x SSC (Sambrook, Fritsch & Maniatis, 1989). Hybridization with DNA labeled with 32 P by random primed synthesis (Feinberg & Vogelstein, 1983) was performed at 42 ◦ C in 20 mM PIPES, pH 6.5, 800 mM NaCl, 50% formamide, 1% SDS, and 100 µg/ml denatured salmon sperm DNA. Oligo-directed RNase H cleavage experiments followed a published protocol (Wyatt, Sontheimer & Steitz, 1992). In brief, RNA extracted from 2 oocytes

gene464.tex; 3/11/1998; 10:11; p.3

70 was annealed at 30 ◦ C with 2 µg of oligonucleotide in presence of 1 unit of RNase H (GibcoBRL). The oligonucleotide used was complementary to sequences 1.36 kb upstream of the 3! end of Tx1L. The RNA was then analyzed by electrophoresis in a 1% agarose gel, followed by Northern blot-hybridization, as described above. The RACE protocol was used to amplify 3! ends of Tx1L RNAs for sequencing (Frohman, 1990). Briefly, first strand cDNA was synthesized by AMV reverse transcriptase (Boehringer Mannheim) from a primer containing restriction enzyme recognition sites for Bgl II, BamH I and Pst I followed by 16 T residues. A PCR product was then obtained using the cDNA as template, with either of two ‘inside primers’ that corresponded, respectively, to sequences 128 bp and 449 bp upstream from the target duplication site of Tx1L and an ‘outside primer’ corresponding to the Bgl II, BamH I and Pst I restriction sites. The final PCR products were cloned into pBluescript KS+ using restriction sites within the fragments or in the primers. Recombinant plasmids were tested for the presence of an insert by colony hybridization with a probe containing the 3! UTR of Tx1L and PTR-1 target sequences. The selected colonies were also negative for hybridization with a DHFR probe, but in fact there were very few DHFR+ colonies. Seven DHFR+ colonies from the !3! UTR experiment were also recovered. DNA extracted from individual colonies was subjected to double-strand sequencing using M13RP1 and M13-21 fluorescent primers on a ABI 373 automated sequencer at the Health Sciences Sequencing Core Facility (Salt Lake City, UT). Results Search for a Tx1L promoter Many NLR elements carry functional RNA polymerase II internal promoters near their 5! ends (Mizrokhi, Georgieva & Ilyin, 1988; Nur, Pascale & Furano, 1988; Swergold, 1990; Minchiotti & Di Nocera, 1991; Minakami et al., 1992; McLean, Bucheton & Finnegan, 1993; Schumann et al., 1994). In the case of the site-specific NLRs, it is also conceivable that the element might take advantage of a promoter located (consistently) in the target sequence (Eickbush, 1992). We tested whether an active promoter, internal or external, was present at the 5! end of Tx1L by injection into oocytes of circular plasmid DNAs carrying the 5! target-element junction. Plasmids carrying

205 bp of PTR-1 target and 521 bp of the 5! -UTR of Tx1L were isolated from three different λTx1C clones to allow for the possibility that only a subset of extant elements are still active. None of these showed evidence of specific transcripts in Northern blot or primer extension assays (not shown). The same was true of a single subclone from λB10 (Garrett, Knutzon & Carroll, 1989) with a larger insert representing 293 bp of PTR-1 and 1646 bp from the 5! end of Tx1L. To allow for contributions of sequences anywhere in the element or target, we injected DNAs of two different λ clones containing complete Tx1C elements, again with negative results. Northern analysis of RNA extracted from uninjected oocytes representing different stages of oogenesis (stage I to stage VI. Dumont, 1972) did not detect natural Tx1L transcripts, and it is not known at what developmental stage or in what type of cells Tx1L expression naturally occurs. The lack of detectable promoter activity in oocytes might be due to tissue specificity, rather than the complete absence of a potentially functional promoter. Expression of a full-length Tx1L RNA in oocytes To force the expression of Tx1L RNA in oocytes, we placed a full-length element under the control of the adenovirus major late promoter contained in the vector pMT3. This promoter is active in Xenopus oocytes (Swick et al., 1992). The construct, pMTx1L (Figure 1B), was designed so that transcription would start precisely at the first nucleotide of Tx1L (PontKingdon, 1994). Primer extension analysis of RNA recovered from oocytes injected with pMTx1L showed that the 5! end of the majority of transcripts mapped at the expected nucleotide (data not shown). Northern analysis showed that three days after nuclear injection of 2 ng of pMTx1L DNA, individual oocytes had accumulated approximately 2 ng of Tx1L transcript. When Northern blots were hybridized with Tx1L DNA probes, the RNA resolved into two bands of about equal intensity (Figure 2A, lane 1), in addition to a smear of presumably partially degraded molecules. When the same blots were hybridized with a probe limited to the DHFR gene found in plasmid sequences downstream of Tx1L (see Figure 1B), only the slower of the two bands was revealed (Figure 2A, lane 2). This result is consistent with the presence of two RNA species: larger ones that terminate at the SV40 polyadenylation signal of the vector, and smaller ones that terminate near the end of the Tx1L sequence.

gene464.tex; 3/11/1998; 10:11; p.4

71

Figure 2. A Northern analysis of RNA from pMTx1L. The same blot carrying total RNA extracted from one oocyte injected with pMTx1L was successively probed with an ORF2 probe (lane 1) and a DHFR probe (lane 2). Positions of size markers (in kb) are shown at the left. B Northern analysis of RNA from four oocytes injected with pM2. Probes were derived from ORF2 (lanes 1, 2, 3) or from the DHFR gene (lanes 4, 5). RNA samples are: total RNA (lane 1), polyA+ RNA (lanes 2, 4) and polyA− RNA (lanes 3, 5). C. Oligo-directed RNase H cleavage. RNA from pM2-injected oocytes was recovered and polyA-selected. Samples were incubated with a complementary oligonucleotide and treated (lanes 2, 4) or not (lanes 1, 3) with RNase H. After electrophoresis and blotting, the filter was probed for ORF2 (lanes 1, 2) or DHFR (lanes 3, 4) sequences. The two major RNA bands in lane 2 are indicated with arrowheads. Each lane contained RNA recovered from four oocytes.

To facilitate mapping of the 3! ends, we analyzed the RNAs produced from a shorter element sequence in the plasmid pM2 (Figure 1B). In this construct, the 5! -UTR and ORF1 are absent, and transcription initiates in vector sequences, 450 bp upstream of the beginning of ORF2. The transcript contains a 240-nt intron from the pMT3 vector, which may be spliced out in the oocytes, enhancing ORF2 protein expression. The 3! ends of the transcription units in pM2 and pMTx1L are very similar, although the latter contains 39 bp more of PTR-1 target sequence (see Figure 1B). Northern blots of RNA expressed in oocytes after injection of pM2 are shown in Figure 2B. As was the case for pMTx1L RNA, two main bands hybridized with a Tx1L probe (lane 1). Passage over an oligo dTcellulose column removed much of the smear of heterogeneous RNAs, which we presume are breakdown products (lanes 2, 3). Both discrete RNAs appeared to be polyadenylated (lane 2), although a portion of the faster band was also found in the polyA− fraction (lane 3). A DHFR probe hybridized only with the slower band (lanes 4 & 5). Furthermore, the sizes of the observed RNAs, 5.9 kb for the slower band and 4.8 kb for the faster, were consistent with termination at two discrete sites. The predicted size of an RNA ending at the SV40 polyadenylation site is 5.6 kb, plus the size of the polyA tail. The distance from the tran-

scription start to the end of Tx1L sequences is 4.6 kb, and a polyA tail may be added. To confirm the identification of the two RNA species, oligonucleotide-directed RNase H mapping was employed. RNA isolated from oocytes injected with pM2 was annealed with a DNA oligonucleotide complementary to a sequence within ORF2. The RNA/DNA hybrids were digested with RNase H and analyzed by Northern blotting, using a probe derived from ORF2 downstream of the oligo. The two observed bands (Figure 2C, lane 2) reflect the distances from the oligo site to the RNA 3! ends. As seen earlier, the DHFR probe hybridized only to the slower band (Figure 2C, lane 4), which had a size of 2.6 kb, in agreement with the 2.48 kb + polyA expected for RNA ending at the SV40 polyadenylation signal. The smaller 1.5-kb RNA, which hybridized only with the ORF2 probe, is consistent with molecules having 3! termini that map to the element end. To test if sequences present in the 3! UTR of Tx1L or in the PTR-1 target were responsible for the formation of the smaller RNA, we examined the RNA transcribed from the construct !3! UTR where these sequences were missing (Figure 1B). Northern analysis of RNA transcribed in oocytes from !3! UTR revealed two bands equivalent to the species identified for pM2 (data not shown). This indicates that the 3! UTR and the PTR-1 sequences are not necessary

gene464.tex; 3/11/1998; 10:11; p.5

72

Figure 3. Mapping of Tx1L RNAs recovered by RACE analysis. RNAs isolated from oocytes injected with pMTx1L or pM2 were subjected to RT-PCR between oligo-dT and the primer indicated by the black arrowhead with the bent extension. The internal primer used for RNA obtained after injection of pM2-5mt, !ORF2p or !3! UTR is indicated by the gray arrow. The locations of these primers are indicated on an expanded map of the region around the 3! end of Tx1L, with symbols as in Figure 1. Products were cloned and sequenced as described in the text. Vertical arrowheads show the points at which the sequences of the clones diverge from that of the template.

for the formation of the RNAs that terminate near the element end. Location of RNA 3! ends The RACE protocol was used to amplify RNA 3! termini located near the end of Tx1L (see Materials and Methods). Clones were obtained for sequencing from five RNA samples: polyA+ -selected RNA from pM2injected oocytes, and total RNA from oocytes injected with pMTx1L, pM2-5mt, !ORF2p, or !3! UTR. pM2-5mt (Figure 1B) is very similar to pM2, but has 5 myc tags at the 5! end of ORF2 and lacks the internal HA tag. The myc tags allowed us to demonstrate expression of the ORF2 protein (data not shown). !ORF2p carries a frameshift mutation near the 5! end of the ORF2 coding sequence that abolishes expression of ORF2 protein. It was included to determine whether activity of that protein, which probably binds sequences near the 3! end of element RNA (Luan & Eickbush, 1995), was required for 3! end formation. In all cases, the primer used for reverse transcription carried 16 T residues at its 3! end and had a 5! extension with restriction sites to aid subsequent cloning; use of this primer inherently selects for RNA ends rich in A. Both this step and the polyA selection may exclude some of the ends of the shorter RNA species, since not

all of these molecules bound to oligodT-cellulose (see Figure 2B). Two different internal primers were used, corresponding to sites 339 bp or 660 bp upstream of the end of the PTR-1 sequence (Figure 3). Termini of 16 of the 18 pMTx1L and pM2 clones, mapped with the primer nearer the element 3! end, were clustered in a region 133 bp long that includes the junction between element and target sequences (Figure 3). A somewhat more dispersed distribution of the 3! ends was found when a larger 3! end fragment was tested, but the majority were still within about 100 bp of the element end. The results with !ORF2p show that no element-encoded protein is required for 3! end formation. The distribution of 3! ends from !3! UTR was very similar to the other samples (Figure 3), indicating that no specific sites in the deleted region determine the location of the termini. Sequences of RNA 3! ends Sequences of the 3! ends are presented in Figure 4. Forty-four clones representing ends in or near Tx1L from five independent samples were analysed. All the clones showed untemplated A-rich 3! extensions, which were typically quite short (average of 26 ± 14 untemplated residues). Only 25 (57%) of the extensions were pure polyA, while the remaining 19 (43%) were interrupted by one to ten non-A residues scattered along the A tail. In none of the cases was a plausible template for the non-A residues found in the injected DNA. The end of the templated sequence was different for each clone, but none of them was preceded by a canonical AATAAA polyadenylation signal. It seems unlikely that these clones resulted from internal PCR priming within Tx1L sequences because in only three cases do the termini occur in a stretch of three or more templated A’s that might have annealed with the primer (Figure 4). As with the location of the 3! ends, the presence of non-A residues was not affected by abrogating expression of ORF2p (!ORF2p) or by deleting sequences at the 3! end of Tx1L (!3! UTR). For comparison, we obtained the sequence of seven clones representing 3! ends of RNAs that terminated at the SV40 polyadenylation signal in the vector (DHFR+) (Figure 4). These have polyA tails that begin 16 or 17 nucleotides downstream of the AATAAA, and all but one are pure polyA; the exception contains one C in a tail of 107 residues. These tails are also longer than those located in Tx1L (average of 63 ± 30 untemplated residues). Because these clones came

gene464.tex; 3/11/1998; 10:11; p.6

73

Figure 4. Sequences of the 3! untemplated extensions. All of the nucleotides added at the points shown in Figure 3 are given for each of the cases. Lower case indicates template sequences and capital letters the non templated nucleotide additions. Non-A residues are highlighted in bold. For the DHFR+ sequences, the polyA addition signal, aataaa, is underlined.

from a sample (!3! UTR) that also yielded 3! termini in Tx1L, they provide a good control for PCR artifacts. In the Tx1L 3! end extensions there were 43 non-A residues among 1161 untemplated nucleotides (3.7%) in 44 different clones. In the seven DHFR+ tails, there was a single non-A residue among 444 nucleotides (0.2%). Statistical analysis using the Student T test showed that these two populations are significantly different at the 98% confidence level (p