Requirement of RNA polymerase III transcription factors for in vitro position-specific integration of a retroviruslike element.

The yeast retroviruslike element Ty3 inserts at the transcription initiation sites of genes transcribed by RNA polymerase III (Pol 111). An in vitro integration assay was developed with the use of Ty3 viruslike particles and a modified SUP2 tyrosine transfer RNA (tRNATYr) gene target. Integration was position-specific and required Ty3 integrase, Pol Ill transcription extract, and a transcriptionally competent tRNA gene. Use of individual transcription factor (TF) IIIB-, TFIIIC-, and Pol Ill-containing fractions showed that TFIIIB and TFIIIC, together, were sufficient for position-specific Ty3 integration, but not for transcription. This report demonstrates that in vitro integration of a retroelement can be targeted by cellular proteins. RNA encoding the B subunit of the a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) subtype of ionotropic glutamate receptor (GluR-B) undergoes a posttranscriptional modification in which a genomically encoded adenosine is represented as a guanosine in the GluR-B complementary DNA. In vitro editing of GluR-B RNA transcripts with HeLa cell nuclear extracts was found to result from

. To permit detection of the newly phosphorylated IKB-ox in this experiment, we included phosphatase inhibitors during cell extraction (2,5). To enhance detection, we trapped the phosphorylated intermediate by blocking proteolysis with calpain inhibitor 1 (2).
Of note, mutation of the protein kinase C consensus site (residues Ser262 and Thr263) (15) did not prevent activation of NF-KB (12). 14. Proteolytic mapping experiments confirmed that induced phosphorylation of IKB-ox occurred at a site or sites that are NH2-terminal to the ankyrin domain (12). Thus together with the observation that mutations of all other potential phosphorylation sites within the NH2-terminal domain of IKB-a did not block signaling indicates that the possibility that the Ser32 and Ser36 mutations act directly at one or both of these sites to block phosphorylation. 15 These experiments also showed that the murine IKB-aL was phosphorylated in response to signals (shown by a shift in charge toward negative pH in the first dimension) and that the induced phosphorylation of this species resulted in only a minor shift in apparent size (12). Analyses of the AN and mt36 mutants yielded results similar to those obtained with mt32 (12).
Phosphatase treatment resulted in a single spot (12).
Endogenous IKB-CL from human cells was indistinguishable from exogenously expressed human IKB-CL on these gels. 19 integrated with a preference for the 5' portions of the three RNA Pol II-transcribed genes examined (4). In addition, unselected transposition of Tyl into chromosome III revealed an insertion bias for regions near tRNA genes and outside of open reading frames; this pattern was not observed in vitro with a naked DNA substrate (5). These findings have implicated host DNA-associated proteins in integration site selection. DNA structure influences retroviral integration site selection. This was suggested by a study that identified in vivo integration sites in DNA fragments containing bends (6). In addition, in vitro data indicate that integration occurs preferentially into particular DNA structures. Integration mediated by human immunodeficiency virus (HIV) pre-integration complexes or purified integrase (IN) occurred nonrandomly into unmodified, naked DNA and preferentially into runs of CpG modified by 5-methylation of cytosine (7). With nucleoprotein com-plexes of murine leukemia virus (MLV), integration into naked DNA also occurred nonrandomly and more frequently into the exposed major groove of DNA assembled into nucleosomes (8). Moreover, DNA positions predicted to be distorted within the nucleosome were used preferentially for integration mediated by HIV IN (9). Preferred integration into static or induced bends that were not associated with nucleosomes occurred with both MLV and HIV IN (10).
Ty3, a retroviruslike element in Saccharomyces cerevisiae, integrates solely upstream of genes transcribed by Pol III. The gene-proximal joint in the host DNA occurs within one to three nucleotides of the transcription initiation site (11,12). On the basis of in vivo experiments, it has been suggested that integration is mediated by a protein-protein interaction between the Ty3 integration complex and the Pol III transcription complex. To identify cellular factors required for Ty3 integration, we developed an in vitro assay. Ty3 viruslike particles (VLPs), analogous to retroviral core particles, were isolated as a complex fraction containing Ty3 RNA, replicated DNA, and the mature Ty3 proteins, including IN (13). The target plasmid, pDLC374 (Fig. 1A), was used previously as a target for Ty3 transposition in vivo monitored by genetic selection (12) or by a polymerase chain reaction (PCR) assay (14). This plasmid contained the yeast HIS3 gene, a 2-pm plasmid origin of replication, and a modified version of the SUP2 tRNATYr gene cloned into pIBI20. To reconstitute the natural genomic targets of Ty3, we prepared a Pol III transcription extract from yeast cells (15). The BioRex7O 500 mM NaCl fraction (BR500) from this preparation was used together with Ty3 VLPs and the tRNA gene target plasmid for in vitro integration reactions (16). DNA was isolated from each reaction and amplified by PCR with primers U REPORTS complementary to the HIS3 gene (primer set 284-285) to quantitate the amount of pDLC374, or with primers complementary to Ty3 and the tDNA (primer set 278-279) to detect position-specific integration (   with each primer set and either pDLC374 or pTM42, a pDLC374 derivative which has a Ty3 insertion in divergent orientation immediately upstream of the transcription initiation site of the tRNA gene (14). A 474base pair (bp) product from the 278-279 PCR was detected for pTM42 but not for pDLC374. The 240-bp product from the HIS3 primers (284-285 PCR) was detected for both plasmid templates.
In vitro integration reactions were performed with the standard conditions as described (16) or with variations in order to identify essential reaction components (Fig.  1B). When the standard integration reaction conditions were used, the 474-bp 278-279 PCR product was observed, which indicated that integration occurred upstream of the tRNA gene. Integration required target DNA, VLPs, and a divalent cation. No integration occurred if the incubation time was 0 min or if the VLPs were boiled for 10 min before being added. There was no detectable integration in the absence of BR500 transcription extract, suggesting that a cellular protein or proteins were required for the position-specific integration. Differences in the amount of the 474-bp product were not the result of variation in the amount of DNA template for the PCR as demonstrated by equivalent amounts of the control 240-bp HIS3 product. The 278-279 PCR products of several integration samples were compared with the 278-279 PCR products of a serial dilution of pTM42 DNA in pDLC374 DNA. On the basis of this comparison, integration was estimated to occur into less than 5% of the target DNA.
To  (18) and in vivo (19). Mutations engineered in Ty3 IN by site-directed mutagenesis changed the second Asp residue of this motif (amino acid position 225) (20) to Glu, and the Glu (amino acid position 261) to Asp. The mutations were introduced separately and in combination. The IN mutants produced wild-type amounts of mature Ty3 proteins and replicated DNA, but they did not transpose in vivo (21). When used in standard integration reactions in parallel with wildtype VLPs, the mutant VLPs were unable to catalyze integration (Fig. 2). Thus, in vitro, as in vivo, position-specific integration required Ty3 IN activity.
The dependence of the integration reaction on the presence of the BR500 extract suggested that integration in vitro depended on the ability of the tDNA to form a tran-SCIENCE * VOL. 267 * 10 MARCH 1995 1489 scription complex, as had been observed in vivo (12). To test this directly, we introduced a mutation into the tDNA promoter. Two internal promoter elements (box A and box B) in tRNA genes mediate the binding of the Pol III transcription factors TFIIIC and TFIIIB. TFIIIC binds first to the box A and box B sequences, then mediates the binding of initiation factor TFIIIB (22). We introduced a mutation by site-directed mutagenesis at the position in the modified SUP2 tRNATyr gene corresponding to position 56 in the wild-type tRNATyr molecule. This mutation (G56) altered the first absolutely conserved cytosine in the box B promoter element, replacing it with guanine. The G56 mutation causes a decrease in the binding of TFIIIC to the SUP4 tRNATyr gene in vitro to approximately 0.3% of wildtype amounts (23), which results in a decrease in transcription of SUP4 tRNATyr in vivo to 5% of wild-type amounts (24). This mutation inactivates the SUP2 tRNATyr gene as a target for Ty3 integration in vivo (12). In vitro, the modified SUP2 tRNATyr gene with a wild-type box B was transcribed by the BR500 extract, but the G56 mutant was not (Fig. 3A). The G56 mutant was not used as a target for position-specific integration (Fig. 3B, lanes 7 to 9). These data showed that in vitro, as in vivo, integration required an intact box B. This result is consistent with the dependence of Ty3 integration on a Pol III transcription factor (or factors), Pol III, or both. Integration into the modified SUP2 tRNATyr gene with a wild-type or G56 mutant box B was also examined by performing the PCR with primers that did not anneal to the Ty3-target junction (21). With this integration position-independent primer set, a product of the predicted size for Ty3 integration immediately upstream of the tRNA gene was formed when the target was wild-type, but not when it was the G56 mutant. It is unlikely that our system can detect single insertions; thus, we cannot address the possibility that some insertions occur nonspecifically. However, the major site of in vitro integration detected with these primers was immediately upstream of the transcription initiation site.
To determine which Pol III transcription components were required for integration, we separated the BR500 extract on a DEAE Sephadex column into TFIIIB-, TFIIIC-, and Pol III-containing fractions (15). The fractions were tested in transcription reactions with pDLC374 (Fig. 4A). In vitro transcription activity was dependent on the presence of all three fractions. Transcription was not observed with the G56 mutant template or in the absence of template. The DEAE fractions were added to integration reactions in all combinations (Fig. 4B). No single fraction was sufficient for integration. Integration was observed when all three fractions were present, as expected, but also when only TFIIIC and TFIIIB were added. When Pol III was included, the amount of integration was lower. This result was obtained in duplicate experiments for three preparations of extract. Because TFIIIC is required to load TFIIIB onto the tDNA, these data do not address whether the role of TFIIIC in integration is direct or indirect, that is, as a loading factor.
The simplest model to explain the position-specific integration of Ty3 is that a protein-protein interaction occurs between some component of TFIIIB, either alone or together with TFIIIC, and the Ty3 integration complex. A tRNA gene that had reduced transcriptional activity as a result of A + | t+ + + TFIII B + 1 + +I++ bd.+ + VLPs mutations at the initiation site was competitive as a target for integration in vivo (12). That result suggested that transcription is not rate-limiting for transposition and is consistent with the in vitro result which showed that Pol III is not required for integration. The decrease in integration in the presence of Pol III may be the result of a competition between Pol III and the integration complex for contact with the TFIII(C+B)-tDNA complex.
In addition to a protein-protein interaction, there may be a DNA structural component to Ty3 integration. TFIIIC and TFIIIB bend the tDNA upon binding to the promoter elements and 5' flanking region (25). The data are consistent with the center of the bend being positioned near the transcription start site. In addition, the initiation site is DNAase I-hypersensitive (15). DNA distortion in this region may facilitate access or activity of the Ty3 integration complex. Kinked and unwound regions of DNA appear to be used preferentially for retroviral integration, and distortion of DNA is an inherent feature in the activity of other endonucleases, for example Eco RI (26), and recombinases, for example FLP (27).
Ty3 is the most specific among wellstudied retroelements in its pattem of integration, yet characterization of its mechanism of integration may elucidate the basis of more subtle integration patterns of retroviruses and other retroviruslike elements. The results presented here show the dependence of Ty3 position-specific integration on Pol III transcription factors. Retroviral and Tyl integrations display some preference for transcribed regions and transcription initiation regions, respectively, and thus could be facilitated by interaction with transcription factors. HIV IN interacts with a human gene product (InIl1) which has a region similar to one in the yeast transcription factor Snf5p and stimulates the joining activity of IN in vitro (28). Targeting of Ty3 preintegration complexes by transcription factors is compatible with a retroviral mechanism of integration and suggests a basis for the observed insertion of retroviruses into preferred positions in host genomes.  scriptional modification in which a genomically encoded adenosine is represented as a guanosine in the GluR-B complementary DNA. In vitro editing of GluR-B RNA transcripts with HeLa cell nuclear extracts was found to result from an activity that converts adenosine to inosine in regions of double-stranded RNA by enzymatic base modification. This activity is consistent with that of a double-stranded RNA-specific adenosine deaminase previously described in Xenopus oocytes and widely distributed in mammalian tissues.
The editing of mRNA transcripts is an important mechanism for augmenting the flexibility of eukaryotic gene expression (1). The A-*G conversion seen in mRNAs encoding glutamatergic ion channel subunits exemplifies the functional relevance of this posttranscriptional modification (2). Glutamate receptors of the AMPA subtype are generated by the assembly of GluR-A, -B, -C, and -D subunits into homo-and heteromeric channels (