Host Genes That Affect the Target-Site Distribution of the Yeast Retrotransposon Ty1

Corresponding author: Susan W. Liebman, Room 4068, MBRB (m/c567), Laboratory for Molecular Biology, University of Illinois, 900 S. Ashland Ave., Chicago, IL 60607., suel{at}uic.edu (E-mail)

Communicating editor: F. WINSTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We report here a simple genetic system for investigating factors affecting Ty1 target-site preference within an RNAP II transcribed gene. The target in this system is a functional fusion of the regulatable MET3 promoter with the URA3 gene. We found that the simultaneous inactivation of Hir3 (a histone transcription regulator) and Cac3 (a subunit of the chromatin assembly factor I), which was previously shown by us to increase the Ty1 transposition rate, eliminated the normally observed bias for Ty1 elements to insert into the 5' vs. 3' regions of the MET3-URA3 and CAN1 genes. The double cac3 hir3 mutation also caused the production of a short transcript from the MET3-URA3 fusion under both repressed and derepressed conditions. In a hir3 single-mutant strain, the Ty1 target-site distribution into MET3-URA3 was altered only when transposition occurred while the MET3-URA3 fusion was actively transcribed. In contrast, transcription of the MET3-URA3 fusion did not alter the Ty1 target-site distribution in wild-type or other mutant strains. Deletion of RAD6 was shown to alter the Ty1 target-site preference in the MET3-URA3 fusion and the LYS2 gene. These data, together with previous studies of Ty1 integration positions at CAN1 and SUP4, indicate that the rad6 effect on Ty1 target-site selection is not gene specific.

RETROTRANSPOSONS are a class of transposable elements that resemble eukaryotic retroviruses structurally and functionally. Insertions of retrotransposons and retroviruses can alter the regulation of nearby genes and cause disease in mammals. The choice of insertion sites for most retroelements is not random, although little sequence specificity is displayed (for review see SANDMEYER et al. 1990 ). In several studies, AT-rich, transcriptionally active, or nuclease-sensitive chromatin regions were found to be preferred proviral insertion sites (SHIMOTOHNO and TEMIN 1980 ; FREUND and MESELSON 1984 ; ROBINSON and GAGNON 1986 ; VIJAYA et al. 1986 ; ROHDEWOHLD et al. 1987 ; MOOSLEHNER et al. 1990 ; SCHERDIN et al. 1990 ).

Common laboratory yeast strains contain five types of retrotransposons, called Ty elements, in which long terminal direct repeats (LTRs) flank central regions of DNA. The central region encodes structural and enzymatic proteins (for reviews see BOEKE and SANDMEYER 1991 ; FARABAUGH 1995 ). Ty1 and Ty2 elements are closely related and share the same LTR sequences (called delta elements). They are present in ~30–35 and 5–15 copies per genome, respectively. The more distantly related retrotransposons, Ty3, Ty4, and Ty5, with different sets of LTRs, are present in lower copy number and have not been observed to be insertional mutagens. Ty1, Ty2, Ty4, and Ty5 elements belong to the copia class of retrotransposons, while Ty3 is a member of the gypsy class.

The complete yeast genomic sequence and the mapping of many Ty1 elements in several yeast strains indicate a nonrandom distribution of Ty1 elements and the other yeast Ty elements. Ty1 elements are found preferentially in AT-rich regions (OYEN and GABRIELSEN 1983 ; NATSOULIS et al. 1989 ; WILKE et al. 1989 ) and are frequently associated with tRNA genes (GAFNER et al. 1983 ; OLIVER et al. 1992 ). In experimental systems with induced Ty1 elements, integration was also shown to be clustered in hot spots upstream of RNAP III promoters (e.g., tRNA; JI et al. 1993 ; DEVINE and BOEKE 1996 ). The high level of Ty1 integration near RNAP III promoters was shown to be linked to the potential for transcriptional activity (DEVINE and BOEKE 1996 ). Ty3 elements show an even more pronounced specificity for the upstream region of tRNA genes. This specificity appears to depend upon the ability of the Ty3 integration machinery to compete with RNAP III for interaction with the TFIIIB and TFIIIC transcription factors (KIRCHNER et al. 1995 ; CONNOLLY and SANDMEYER 1997 ). Ty5 elements appear to be targeted for integration by protein complexes assembled at silenced regions (ZOU and VOYTAS 1997 ).

Analyses of Ty1 insertions into several RNAP II transcribed genes indicate a preference for the promoter or 5' end of the coding sequence. In some instances, this specificity is caused by the type of selection used. For example, insertions in the vicinity of HIS3 (BOEKE et al. 1985 , BOEKE et al. 1986 ) and ADH2 (WILLIAMSON et al. 1983 ) were selected on the basis of their activation of the target gene and, thus, would be expected to occur in the promoter. However, preferential integration of Ty1 elements into the 5' regions of LYS2, URA3, and CAN1 was also observed when Ty1 insertions were selected on the basis of gene inactivation (EIBEL and PHILIPPSEN 1984 ; SIMCHEN et al. 1984 ; NATSOULIS et al. 1989 ; WILKE et al. 1989 ; LIEBMAN and NEWNAM 1993 ).

Although RNAP III promoters are the preferred endogenous sites for Ty1, the rare insertions into RNAP II transcribed genes can have dramatic effects on the host cells. There have been recent efforts to use retroelements as carriers for gene therapy (for reviews see GUNZBURG and SALMONS 1995 ; TURCHETTO et al. 1997 ; NIELSEN and MANEVAL 1998 ). It is thus of practical importance to understand the factors that influence retroelement target-site specificity, so that integration can be directed into innocuous sites.

Mutations in the DNA repair gene RAD6(UBC2) have been shown to alter the target-site specificity of Ty1 insertions in the CAN1 (LIEBMAN and NEWNAM 1993 ) and SUP4 (KANG et al. 1992 ) genes, and to increase the Ty1 transposition rate into CAN1 (PICOLOGLOU et al. 1990 ) without causing an increase in the level of total Ty1 RNA (LIEBMAN and NEWNAM 1993 ; BRYK et al. 1997 ). RAD6 encodes a ubiquitin-conjugating enzyme and mediates N-end rule-dependent protein degradation (DOHMEN et al. 1991 ; SUNG et al. 1991 ). Deletion of RAD6 has been shown to reduce silencing at telomeres, HM loci (HUANG et al. 1997 ), and rDNA (BRYK et al. 1997 ), which indicates that RAD6 may play some role in the assembly or maintenance of chromatin structures. Rad6 is also involved in DNA repair, recombination, induced mutagenesis, and sporulation (PRAKASH et al. 1993 ). Rad6 can ubiquitinate histones H2A, H2B, and H3 in vitro, although no ubiquitinated histones have been detected in vivo in Saccharomyces cerevisiae (SUNG et al. 1988 ; HAAS et al. 1990 ). Reduced levels of histone proteins (H2A and H2B) have been shown to disrupt the asymmetric orientation bias of Ty1 elements inserted into the CAN1 promoter region (RINCKEL and GARFINKEL 1996 ).

DNA blot hybridization, PCR, and sequencing have been used to identify the distribution of Ty1 elements inserted into different target loci. Here we report a simple genetic system to study the promoter preference of Ty1 integration. Using this system, we found that the simultaneous deletion of CAC3 and HIR3, which was previously shown by us to increase the Ty1 transposition rate (QIAN et al. 1998 ), eliminated the promoter bias of Ty1 integration. CAC3 encodes a subunit of chromatin assembly factor I (CAF-I, KAUFMAN et al. 1997 ), and HIR3 is a histone transcription regulation gene (OSLEY and LYCAN 1987 ; MORAN et al. 1990 ; RECHT et al. 1996 ). We also found that transcription of the target gene altered the target-site preference in a hir3 strain but did not affect the Ty1 distribution in the other strains tested.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plasmids:
The hir3 disruption was made using the gamma deletion method (SIKORSKI and HIETER 1989 ) with SmaI-digested pHH23 (hir3::HIS3). To make this plasmid, two fragments cloned from pHIR3 (QIAN et al. 1998 ), the 1.35-kb EcoRI fragment from the 3' region of HIR3 and the 631-bp BamHI-BglII fragment from the 5' region of HIR3, were inserted into the EcoRI and BamHI sites, respectively, in the polylinker of pRS403 (SIKORSKI and HIETER 1989 ). The orientation of the insertions was engineered such that upon digestion with the unique SmaI site in the polylinker between the two insertions, the linear DNA would promote the desired disruption.

Plasmid pSH2 (WILKE et al. 1989 ), which carries the CAN1 gene, was used as a hybridization probe. Two plasmids, pJEF1114 (2µ, TRP1) and pJEF724 (2µ, URA3), kindly provided by J. Boeke, which contain Ty1 elements under the control of the GAL1 promoter (GAL1-Ty1), were used to induce Ty1 transposition (BOEKE et al. 1985 ; NATSOULIS et al. 1989 ). The LEU2-bearing integrative plasmid pHAM18 was kindly provided by H. Mountain, and it contains a fusion of the MET3 promoter and its 5' coding region with the URA3 gene via a small fragment from HIS4 (Figure 1). The MET3-URA3 fusion is transcriptionally active in the absence of methionine and is repressed by methionine (MOUNTAIN et al. 1991 , MOUNTAIN et al. 1993 ).

View larger version (20K): In this window In a new window Download PPT slide   Figure 1. The Ty1 target-site assay system. The pHAM18 plasmid containing the MET3-URA3 fusion and LEU2 (MOUNTAIN et al. 1991 , MOUNTAIN et al. 1993 ) is integrated into the genomic MET3 locus. The in-frame fusion is composed of the MET3 promoter (crosshatched box), part of its 5' coding sequence (aminoacids 1–182, hatched box), a small HIS4 coding region (amino acids 439–531, solid box), and the URA3 coding region (amino acids 6–267, open box). The LEU2 marker and the extra copy of the MET3 gene are shown. Ty1 elements inserted into the MET3 region of the fusion are frequently lost, while those inserted in the URA3 region are not (see text). The primers used for PCR to map the positions of the Ty1 inserts are indicated by arrows. Note that the HH9 and U5in primers are complementary to both the 3' and 5' LTRs, but they are shown only in the positions in which they could give rise to amplicons in the current experiments.

Strains and media:
Standard media were used to grow yeast (SHERMAN et al. 1986 ) and bacteria (SAMBROOK et al. 1989 ). Nutritional markers were scored by growth on synthetic glucose media lacking a specific component (e.g., -Ura). To detect resistance to 5-fluorootic acid (FOA) caused by Ty1 insertions or other mutations in the MET3-URA3 fusion gene, 1 g/liter FOA and 12 mg/liter uracil were added to synthetic glucose medium lacking methionine, leucine, and uracil (+FOA-Met-Leu), since FOA poisons Ura⁺ cells (BOEKE et al. 1984 ). Galactose (Gal) instead of glucose was used in media for inducing Ty1 transposition in the presence of the GAL1-Ty1 plasmids, pJEF1114, and pJEF724. Canavanine-resistant colonies were selected on -Arg media with 60 mg/liter canavanine. S. cerevisiae were routinely grown at 30°, but were grown at 20° when we wished to enhance the Ty1 transposition rate (PAQUIN and WILLIAMSON 1984 , PAQUIN and WILLIAMSON 1986 ).

S. cerevisiae was transformed using the lithium acetate procedure (ITO et al. 1983 ). The yeast strains used are listed in Table 1. Strain L1528 was made by transforming JB282 with HindIII-BamHI-cut pJJ105 (rad6::LEU2), kindly provided by L. Prakash (MORRISON et al. 1988 ). A series of isogenic strains was made in YJL170 (bearing a complete deletion of URA3) transformed with pJEF1114 (GAL1-Ty1). In each case, a disruption was constructed in the gene to be tested. If the disruption used the hisG-URA3-hisG marker, the URA3 marker was looped out on +FOA medium. Next, each disruption strain was transformed to Leu⁺ with XhoI-cut pHAM18 to integrate the MET3-URA3 fusion at the MET3 locus. Strain L1706 was made with the rad6 deletion generating plasmid pR671 (rad6::hisG-URA3-hisG), kindly supplied by L. Prakash (MORRISON et al. 1988 ), and cut with BamHI. Strain L1707 was made with the rad18 deletion generating plasmid pJJ239 (rad18::hisG-URA3-hisG), kindly supplied by L. Prakash, and cut with EcoRI (BAILLY et al. 1994 ). Both rad6 and rad18 deletions were identified by UV sensitivity and genetic complementation. L1712 was made with the fus3 deletion generating plasmid BJC396 (fus3::hisG-URA3-hisG), kindly supplied by J. Curcio, and cut with BamHI and EcoRI (CONTE et al. 1998 ). PCR analysis was used to identify these deletions, which were sterile in our strain. The primers used to confirm fus3 were FUS3-1/FUS3-2 (Table 2). For the wild-type strain, the two primers give rise to a 632-bp fragment, while the fus3::hisG deletions give rise to an ~1.4-kb fragment. Strain L1714 was made with SmaI-cut pHH23 (hir3::HIS3). The primer pair used for PCR to identify hir3::HIS3 was 140CF/MCN02 (Table 2). An ~2-kb fragment was expected for the hir3::HIS3 deletion as opposed to no fragment for the wild-type strain. Strains L1713(cac3::hisG) and L1715(hir3::HIS3, cac3::hisG) were made with BamHI-SalI-cut pPK104 (cac3::hisG-URA3-kanR-hisG), kindly supplied by P. Kaufman (KAUFMAN et al. 1997 ). The primer pair used for PCR to identify cac3::hisG was CAC3-3'C/hisG (Table 2). An ~1.2-kb fragment is expected for the cac3::hisG deletion vs. no fragment for the wild-type strain. Strain L1708 was made with EcoRI-BamHI-cut pJR675 (sir1::HIS3; RINE and HERSKOWITZ 1987 ). Strain L1709 was made with SphI-EcoRV-cut pJR531(sir2::HIS3; RINE and HERSKOWITZ 1987 ). Strain L1710 was made with EcoRI-cut pSIR3HIS3 (sir3::HIS3, APARICIO et al. 1991 ), kindly supplied by D. Gottschling. Strain L1711 was made with PvuII-cut pRS4.2 (sir4::HIS3; RINE and HERSKOWITZ 1987 ). DNA blot hybridization was used to identify these deletions.

Ty1 transposition induction and target-site assay:
The integrated MET3-URA3 fusion plasmid, pHAM18 (Figure 1), was maintained by growth on -Leu media. Strains bearing the fusion grew on -Ura-Leu-Met medium, but they failed to grow on +FOA-Leu-Met medium because the fusion was derepressed as a result of the absence of methionine. These strains normally did not grow on -Leu-Ura+Met because the fusion was repressed as a result of the presence of methionine (MOUNTAIN et al. 1993 ). Strains bearing integrated pHAM18 (MET3-URA3) and pJEF1114 (GAL1-Ty1) were grown at 30° on -Met-Leu-Trp-Ura plates. To induce Ty1 transposition, cells were then transferred to plates containing Gal-Leu-Trp medium without or with methionine (200 mg/liter) and were incubated at 20° for 1 wk. Cells were then replica plated to +FOA-Leu-Met plates to select for mutations in the fusion that caused a Ura^- mutant phenotype. Independent Ura^- mutants were streaked for single colonies, patched on YPD (complete medium), and spotted on -Met-Leu-Ura to score for revertibility to Ura⁺ (Figure 2).

View larger version (53K): In this window In a new window Download PPT slide   Figure 2. Scheme for selection and rough genetic mapping of mutations in the MET3-URA3 fusion under conditions where the fusion gene is expressed (-Met) or repressed (+Met). The majority of the mutations obtained from this scheme result from the insertions of Ty1 elements. Strains bearing an integrated LEU2 plasmid carrying the MET3-URA3 fusion and a TRP1 2µ GAL1-Ty1 plasmid were streaked for single colonies on -Met-Leu-Trp-Ura. (A) The colonies were then patched on plates containing Gal-Leu-Trp with or without methionine and were incubated at 20° for 1 wk. Galactose-induced transposition of the GAL1-Ty1, together with growth at low temperature, increased the Ty1 transposition rate. (B) The plates were replicated to -Met-Leu+FOA to select for Ura- mutants. (C) One Ura- mutant was picked from each patch, streaked for single colonies, and patched on YPD master plates. (D) The master plates were spotted to -Met-Leu-Ura to score for revertibility of the Ura- mutants. The Ura- mutant pictured in the middle of the bottom row was mapped to the URA3 region of the fusion because it failed to revert. The other five Ura- mutants pictured were mapped to the 5' region of the fusion because they were revertible (see text for further explanation).

Ty1 insertions in the LYS2 gene were obtained in strains JB282 and L1528 carrying pJEF724 (GAL1-Ty1). Patches made from individual colonies grown on -Ura at 30° were replica plated to Gal-Ura and incubated at 20° for 5 days to allow high levels of Ty1 transposition. The plates were then replica plated to medium containing 6% -aminoadipate, where lys2 and lys5 mutants can grow but wild-type cells cannot (CHATTOO et al. 1979 ). One -aminoadipate-resistant mutant from each patch was picked and streaked for single colonies and used for PCR analysis.

Independent colonies of both L1717 and L1675 were grown for 1 wk on YPD medium at 20° to induce Ty1 transposition, and they were then spread on -Arg+Can medium and incubated at 30°. Canavanine-resistant mutants were analyzed via DNA blot hybridization. Because most of the can1 mutations in L1717 result from point mutations, several canavanine-resistant mutants were analyzed from each original colony. To identify independent Ty1 insertions in the CAN1 gene, however, only a single Ty1 mutation was included from each original colony. Total genomic DNA was isolated, completely digested with EcoRI, and probed with the 1.8-kb BamHI-SmaI CAN1 fragment from pSH2. There is an EcoRI site 673 bp downstream from the ATG of CAN1, which together with the EcoRI sites in the 5' and 3' flanking regions, respectively, gives rise to 6.0- and 8.7-kb fragments. The presence of both wild-type bands indicates a point mutation in CAN1. The disappearance of either of the two wild-type fragments, as well as the appearance of other bands consistent with a 6-kb insertion into the CAN1 gene with the Ty1-characteristic EcoRI map, indicated the presence of a Ty1 element within the CAN1 region in the missing fragment. Except for an EcoRI restriction site polymorphism in the CAN1 5' flanking region, the analysis used to identify and localize Ty1 elements is identical to that described previously (LIEBMAN and NEWNAM 1993 ). An increase in fragment size of ~300 bp was taken as an indication of a solo delta element insertion.

Mapping Ty1 inserts by PCR:
Four primers, MU-A, MU-B, MU-C, and HAM-1, were designed for the MET3-URA3 construct (Figure 1, Table 2). Three primers for Ty1 were used: U5in, HH9, and HH10 (Figure 1, Table 2). To map Ty1 insertions in the Ura^- mutants, the primer pairs MU-A/MU-C and MU-B/HAM-1 were used to check if the Ura^- mutants were caused by point mutations. Point mutations were expected to yield wild-type fragments of 1.7 and 0.9 kb, respectively. The absence of either band suggested the presence of a Ty1 insertion, since the fragment amplified would be too large for the PCR protocol used. To confirm the presence of a Ty1 element and to map it, primer pairs MU-A/U5in, HAM-1/U5in, and sometimes MU-B/U5in and MU-C/U5in were used. Ty1 primers HH9 and HH10 were also used. The positions of the Ty1 elements were determined within 50–100 bp by estimating the size of amplicons, usually in the 500- to 1500-bp range.

The LYS2 locus was divided into three regions by three sets of primer pairs (Table 2; see Figure 6): LYS2-0/LYS2-1C, LYS2-1/LYS2-3C, and LYS2-2/LYS2-4C. The inability to amplify the expected fragment suggested the presence of a Ty element. To confirm this and to map the position of the Ty1 insertion, primers U5in and/or HH9, specific for Ty1, were used in combination with the different LYS2 primers. Each of the Ty1 inserts was unambiguously localized to one of five regions of LYS2 that were used in an earlier study (NATSOULIS et al. 1989 ) and that are defined by restriction sites (PstI-EcoRV, EcoRV-BglII, BglII-NcoI, NcoI-BamHI, and BamHI-NcoI). These assignments were based on the sizes of the PCR amplicons. In borderline cases, numerous primers were used and restriction sites were mapped in the PCR amplicons.

View larger version (86K): In this window In a new window Download PPT slide   Figure 3. Transcriptional regulation of the MET3-URA3 fusion was altered in cac3 hir3 mutants. Growth on Gal-Leu-Ura and RNA hybridization were used to assay the expression of the MET3-URA3 fusion in L1716 (WT), L1713 (cac3), L1714 (hir3), L1715 (cac3 hir3), L1706 (rad6), L1712 (fus3), and L1710 (sir3) grown under repressed and derepressed conditions (+ or -Met). Plates containing Gal-Leu (+ or -Met) were used as growth controls. Total RNA was isolated from cells grown in Gal-Leu (+ or -Met) at 20°. The RNA blot was sequentially probed with DNA complementary to URA3, RAD6, and the loading control TEF1.

View larger version (21K): In this window In a new window Download PPT slide   Figure 4. Effects of hir3, cac3 hir3, and rad6 on the distribution of Ty1 elements that transposed into the MET3-URA3 fusion when the target was under repressed or derepressed conditions. Ty1 insertions were obtained in the isogenic strains L1716 (WT), L1714 (hir3), L1715 (cac3 hir3), and L1706 (rad6), carrying the MET3-URA3 fusion and pJEF1114 (GAL1-Ty1). PCR analyses were used to map the Ty1 insertions. A schematic diagram of the MET3-URA3 fusion is shown at the bottom. The location of primer MU-A at position -769 is used as the zero point of the x-axis. The hatched box corresponds to the 5' coding region of MET3. The solid box is the HIS4 bridge sequence. The open box is the URA3 coding region. Ty1 inserts within every 200 bp starting at position -769 were grouped, and the number of Ty1 insertions found within that region is presented. Open bars indicate Ty1 insertions in the forward direction (same transcriptional orientation as MET3-URA3); shaded bars are Ty1 insertions in the reverse direction.

View larger version (10K): In this window In a new window Download PPT slide   Figure 5. Restriction map of the CAN1 gene showing Ty1 insertions. The BamHI-SalI fragment is the probe for DNA hybridization used to locate Ty1 insertions. The solid box indicates the CAN1 coding region. Two Ty1 elements are shown, one inserted 5' and the other 3' of the EcoRI site in CAN1. R, EcoRI; B, BamHI; S, SalI.

View larger version (23K): In this window In a new window Download PPT slide   Figure 6. Deletion of RAD6 altered the Ty1 target-site preference in LYS2. The location and direction of primers used to map Ty1 inserts in LYS2 are shown on the top. Ty1 insertions were obtained from isogenic RAD6 and rad6 strains carrying pJEF724 (GAL1-Ty1). Ty1 insertions were grouped into five regions defined by the indicated restriction enzymes (P, PstI; R, EcoRV; G, BglII; N, NcoI; B, BamHI), and the number of insertions per kilobase is presented. The graph of the RAD6 (WT) results shows our data pooled with previously published data obtained in the same strain under the identical conditions (NATSOULIS et al. 1989 ).

In addition to the DNA blot analyses, the locations of some of the Ty1 and delta elements in CAN1 were verified by PCR analysis using primers (Table 2) U5in/CAN1(+1941X)' (LIEBMAN and NEWNAM 1993 ; RINCKEL and GARFINKEL 1996 ).

Northern blot analysis:
The transcription status of the MET3-URA3 fusion in L1716 (wild type), L1713 (cac3), L1714 (hir3), L1715 (cac3, hir3), L1706 (rad6), L1712 (fus3), and L1710 (sir3) were examined under repressed (+Met) or derepressed (-Met) conditions. Strains were grown to log phase in +GAL-Leu-Met and +Gal-Leu+200 mg/liter methionine at 20°. Total RNA was isolated, separated, and probed sequentially (SAMBROOK et al. 1989 ) with URA3 (amplified by primer MU-B and HAM-1), RAD6 (EcoRI fragment of pHH1; HUANG et al. 1997 ), and TEF1 (EcoRI-HindIII of pSP36; PICOLOGLOU et al. 1990 ) DNA fragments.

Statistical analysis:
A contingency chi-square test was used to determine whether the Ty1 distributions and numbers of revertible vs. nonrevertible Ura^- mutants obtained were significantly different from the wild type (HERSKOWITZ 1977 ). The Yates correction was used where applicable. For the purpose of the contingency chi-square test, the Ty1 insertions in MET3-URA3 were grouped into five regions on the basis of their locations (-369 to -168, -169 to +31, +32 to +631, +632 to +1231, and +1232 to +1632). Likewise, for the contingency chi-square test, the Ty1 insertions in LYS2 were grouped into four regions (PstI-EcoRV, EcoRV-BglII, BglII-NcoI, and NcoI-NcoI).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A system to study Ty1 target-site selection:
A simple genetic system was devised that can distinguish Ty1 insertions and mutations located in the 5' region of a target gene from those in the 3' region. The target gene we used contains the promoter and 5' region of MET3 fused in-frame to the URA3 coding sequence via a small HIS4 bridge region (Figure 1). When the fusion is derepressed by the absence of methionine, it can complement a deletion of URA3. Since the genetic system described below cannot distinguish between point mutations and Ty1 insertions at the URA3 locus, we increased the rate of Ty1 transposition by transforming cells with pJEF1114 (GAL-Ty1) and inducing Ty1 transcription on galactose medium (BOEKE et al. 1985 ). Ura^- mutants were then selected on +FOA-Leu-Met medium. Using this procedure, we ensured that almost all the Ura^- mutants isolated had resulted from Ty1 insertions. Indeed, by using PCR analysis, 93 of 100 independent Ura^- mutants isolated in strain L1716 were shown to contain a Ty1 insertion within the MET3-URA3 fusion (see below).

We found that Ura^- mutants caused by Ty1 insertions into the MET3 promoter or its coding region (5' region of the fusion) frequently reverted to Ura⁺ (Figure 2), presumably via gene conversion, because a complete copy of the MET3 gene was present downstream of the fusion (Figure 1). The frequent reversion of such Ura^- mutants was detected as growth on -Ura-Leu-Met medium. In contrast, Ura^- mutants caused by Ty1 insertions in the URA3 coding region (3' region of the fusion) were not revertible, presumably because, as a result of the absence of the normal URA3 gene in the strain used, they cannot be converted.

Transcriptional regulation of the MET3-URA3 fusion:
It was previously shown that methionine can repress the expression of the MET3 gene (MOUNTAIN et al. 1991 , MOUNTAIN et al. 1993 ). Likewise, we found that in the presence of 200 mg/liter methionine, L1716 (wild type), which contains the MET3-URA3 fusion, was unable to grow in the absence of uracil and contained a very reduced level of fusion message or none at all (Figure 3). In the absence of methionine, cells grew without uracil, and the ~2-kb MET3-URA3 mRNA was easily detected. Similar results were obtained for isogenic strains containing a disruption of CAC3 (L-1713), HIR3 (L1714), RAD6 (L1706), FUS3 (L1712), or SIR3 (L1710). In contrast, the double cac3 hir3 mutants grew on medium lacking uracil, even in the presence of methionine. Thus, in cac3 hir3 mutants, the fusion is expressed despite the presence of methionine levels that normally lead to repression (Figure 3). In agreement with this observation, RNA hybridization with a URA3 probe showed that there is an ~1-kb mRNA present in cac3 hir3 mutants regardless of the presence of methionine (Figure 3), although the normal 2-kb MET3-URA3 mRNA was reduced when methionine was added to the culture.

The transcriptional status of the target gene has no effect on the Ty1 distribution in a wild-type strain:
To test if transcription of the MET3-URA3 fusion affects the Ty1 target-site preference, transposition was induced in L1716 on media containing or lacking methionine. PCR analysis was used to determine the number of Ura^- mutants caused by Ty1 insertions and to map the positions of these Ty1 insertions (see MATERIALS AND METHODS; Figure 1). In addition, each of the Ura^- mutants was tested for its ability to revert to Ura⁺.

Of 50 independent Ura^- mutants selected after induction of transposition while the MET3-URA3 fusion was repressed, PCR analysis showed that 46 result from Ty1 insertions (Table 3, Figure 4). Of these, 41 were in the MET3 region of the fusion and all were revertible, and 5 were in the URA3 region of the fusion and none were revertible.

Of 50 independent Ura^- mutants selected after induction of transposition while the MET3-URA3 fusion was derepressed, 47 were shown to be caused by Ty1 insertions via PCR analysis (Table 3, Figure 4). Of 42 Ty1 elements that mapped to the MET3 region of the fusion (including the MET3 promoter and its 5' coding region), 41 were revertible. The single nonrevertible Ura^- mutant in this group mapped closest to the MET3-HIS4 border (within 170 bp). An additional 4 Ty1 elements mapped to the URA3 region of the fusion, and none of these was revertible. One Ty1 element was mapped to the small HIS4 region and was not revertible.

The data show that the revertibility of the Ura^- mutants can generally predict the location of Ty1 elements in the MET3-URA3 fusion. The data further show that the pattern of Ty1 insertions throughout the gene is not significantly different (P = 0.5) whether the MET3-URA3 fusion was repressed or derepressed while Ty1 transposition was induced (Figure 4).

In analogy with results described previously, when the CAN1 gene was the target of Ty1 transposition (WILKE et al. 1989 ; LIEBMAN and NEWNAM 1993 ), we found a dramatic bias for Ty1 elements in the promoter region to insert into the same transcriptional orientation as MET3-URA3, while there was no obvious orientation bias for Ty1 elements inserted into downstream regions of the fusion. The transcriptional status of the MET3-URA3 target gene during transposition did not significantly alter the asymmetric orientation bias (Figure 4, Table 4).

The Ty1 insertion-site promoter bias is completely eliminated in a cac3 hir3 double mutant:
Simultaneous mutations in CAC3 and HIR3, but neither of the two single mutations, dramatically increase the rate of retrotransposition and MMS sensitivity, and they reduce telomeric silencing and growth rate (KAUFMAN et al. 1998 ; QIAN et al. 1998 ). HIR3 belongs to a class of HIR genes that regulate the transcription of histones (OSLEY and LYCAN 1987 ; MORAN et al. 1990 ; RECHT et al. 1996 ), while Cac3 combined with Cac1 and Cac2 make up CAF-I (KAUFMAN et al. 1997 ).

Strains that carry the MET3-URA3 fusion contain a single cac3 (L1713) or hir3 (L1714) mutation or simultaneous cac3 hir3 (L1715) double mutations, and are isogenic to L1716 were constructed. Table 5 shows the comparison of analyses of independent Ura^- mutants isolated in the cac3, hir3, cac3 hir3, and wild-type strains. In each case, Ura^- mutants were obtained after induction of Ty1 transposition under conditions where the MET3-URA3 fusion was either repressed or derepressed. The results for deletion of HIR3 alone are described in the next section.

View this table: In this window In a new window   Table 5. The revertibility of Ura- mutants from cac3, hir3 and cac3 hir3 strains

The Ty1 distribution pattern and orientation bias obtained in the cac3 strain were not dramatically different from the wild-type strain. As described above for the wild-type, transcription of the MET3-URA3 fusion did not affect the Ty1 target-site distribution in the cac3 strain (Table 5). A total of 144 Ura^- mutants were obtained in the cac3 strain. Of these, 124 were revertible and 20 were not. There was no significant alteration in the asymmetric orientation bias of Ty1 elements inserted into the promoter region under either repressed or derepressed conditions (Table 4).

For the cac3 hir3 strain (L1715) of 64 Ura^- mutants obtained under repressed conditions, 26 were revertible and 38 were not. Of 64 Ura^- mutants obtained under derepressed conditions, 22 were revertible and 42 were not. This represents a dramatic alteration compared with the wild-type strain, L1716 (Table 5). PCR was used to identify Ty1 insertions among the Ura^- mutants in the cac3 hir3 strain, L1715, and to map them (Figure 4). Of 64 Ura^- mutants isolated under repressed conditions, 51 were found to contain Ty1 insertions. Of these, 19 were in the MET3 region and all were revertible, while 32 were in the URA3 region and were not revertible. Of 64 Ura^- mutants isolated under derepressed conditions, 55 were found to contain Ty1 insertions. Of these, 16 were in the MET3 region and all were revertible; 4 were in the HIS4 region and none were revertible, and 35 were in the URA3 region and none were revertible. The data clearly show that the distribution of Ty1 insertion sites differs in cac3 hir3 and wild-type strains under both repressed and derepressed conditions (P < 0.0001). Furthermore, while the difference from the wild type is not statistically significant, the data suggest that cac3 hir3 mutations relax the asymmetric orientation bias in the promoter region (Table 4).

We also examined the effect of cac3 hir3 double mutations on the distribution of Ty1 integration positions into the CAN1 gene after growth at 20° in YPD. The CAN1 gene is known to be transcribed under these conditions (HOFFMANN 1985 ; AHMAD and BUSSEY 1986 ). Independent, spontaneous canavanine-resistant mutants were isolated in L1717 (CAC3 HIR3) and in its isogenic derivative, L1675 (cac3 hir3), respectively. DNA hybridization was used to identify those mutants that had EcoRI patterns indicative of the insertion of a Ty1 element. The integration positions of 10 and 28 independent Ty1 insertions in L1717 and L1675, respectively, were unambiguously determined to be either upstream or downstream of the EcoRI site within the CAN1 open reading frame (ORF; Figure 5, Table 6). A similar approach was previously used to map the integration positions of Ty1 elements in RAD6 wild-type and isogenic rad6 mutant strains (WILKE et al. 1989 ; PICOLOGLOU et al. 1990 ; LIEBMAN and NEWNAM 1993 ), and these results are included in Table 6 for comparison. The distribution of Ty1 elements located upstream vs. downstream of the CAN1 EcoRI site was similar in L1717 (wild type) and LP2752-4B (wild type; P = 0.55), exhibiting a strong promoter bias in both cases. In contrast, the distribution in L1675 (cac3 hir3) is dramatically different from that in LP2752-4B (P = 0.0015). The Ty1 promoter preference at CAN1 is clearly eliminated in the cac3 hir3 strain. This indicates that the effect of the double cac3 hir3 mutation on the Ty1 distribution is not specific for the target gene.

Transcription of the target MET3-URA3 fusion gene alters the distribution of Ty1 integration positions in a hir3 strain:
HIR3 is a histone transcription regulator and has been shown to be involved in modulating the balance of histones (OSLEY and LYCAN 1987 ; MORAN et al. 1990 ; RECHT et al. 1996 ). As described above, transcription of the MET3-URA3 fusion did not affect the Ty1 target-site distribution in wild-type, cac3 HIR3, or cac3 hir3 strains. In the hir3 strain, however, transcription of the target caused a dramatic alteration in the proportion of revertible Ura^- mutants, which suggests an alteration in the Ty1 preference for integrating into the promoter region (Table 5).

PCR was used to map the location and orientation of the Ty1 inserts in the first 50 Ura^- mutants obtained under both repressed and derepressed conditions (Figure 4). In the repressed data set, 35 were in the MET3 region and all of these were revertible; 4 were in the URA3 region and were not revertible. In the derepressed data set, 25 were in the MET3 region and 22 of these were revertible; 19 were in the URA3 region and none were revertible. One was mapped into the HIS4 region and was not revertible. The data show that the pattern of Ty1 insertions throughout the gene are significantly different in the repressed and derepressed data sets (P = 0.001). Furthermore, the orientation bias exhibited by Ty1 elements in the promoter regions in the wild-type strain L1716 and its hir3 derivative, L1714, were statistically different (P = 0.005). These data indicate that the hir3 mutation causes loss of the target-site preference for the promoter when the target is transcribed, and that it relaxes the Ty1 asymmetric orientation bias under both repressed and derepressed conditions.

Deletion of RAD6 alters the Ty1 target-site distribution at LYS2 and MET3-URA3:
Deletion of RAD6 was previously shown to alter the Ty1 target-site distribution and transposition rate into the CAN1 gene (PICOLOGLOU et al. 1990 ; LIEBMAN and NEWNAM 1993 ). The pattern of Ty1 insertions into a plasmid-borne SUP4-o target (an ochre suppressor allele of a tyrosyl-tRNA gene) was also altered in a rad6 strain (KANG et al. 1992 ). Rad6 is likely to be involved in the maintenance of chromatin structure since it affects silencing at mating type loci, telomeres (HUANG et al. 1997 ), and rDNA (BRYK et al. 1997 ). Because the CAN1 gene is close to telomere V-L (within 40 kb), it seemed possible that the proximity of the target to a telomere might be necessary for rad6 to have an effect on Ty1 target-site selection. We thus tested if deletion of RAD6 could affect the Ty1 target-site distribution in LYS2, which is >200 kb away from a telomere or centromere, and the MET3-URA3 fusion, which is integrated into the MET3 locus, ~20 kb from centromere X and 300 kb from the telomere.

To study the Ty1 target-site distribution in LYS2, Ty1 transposition was induced in strain JB282, which bears the wild-type RAD6 allele, kindly supplied by J. Boeke (NATSOULIS et al. 1989 ), and its isogenic rad6 derivative, L1528, both of which carry the GAL1-Ty1 plasmid pJEF724 (BOEKE et al. 1985 ), by growth on galactose medium at 20°. Independent mutations at the LYS2 locus were selected on medium containing -aminoadipate. PCR analyses were used to map Ty1 insertions in these mutants to specific regions of LYS2, as described in MATERIALS AND METHODS. The distribution of 17 Ty1 insertions isolated here in JB282 (for details see BURCK 1996 ) was compared with previously published data (NATSOULIS et al. 1989 ) obtained under identical conditions in the same strain. The distributions were found not to be significantly different from each other (P = 0.75), and neither exhibited an orientation asymmetry in the promoter region. We thus combined these two RAD6 data sets and compared them with the distribution obtained in the isogenic rad6 strain, L1528. The data (Figure 6) show that the RAD6 and rad6 distributions are significantly different (P < 0.002).

Table 7 shows the comparison of analyses of independent Ura^- mutants isolated from isogenic RAD6 (L1716) and rad6 (L1706) strains carrying the MET3-URA3 fusion and pJEF1114 (GAL1-Ty1). In each case, half of the Ura^- mutants were obtained after the induction of Ty1 transposition under conditions where transcription of the MET3-URA3 fusion was either repressed or derepressed. Since there was no difference in the results obtained under these conditions, we have pooled the data. Of 220 mutants obtained in L1706, 98 were revertible and 122 were not. In contrast, only 13 of 100 Ura^- mutants obtained from L1716 failed to revert to Ura⁺.

PCR was used to map the location of the Ty1 inserts (Figure 4), and it verified that 83 of the revertible and 82 of the nonrevertible Ura^- mutants isolated in L1706 (summing those isolated under repressed and derepressed conditions) resulted from Ty1 insertions. Of the 165 Ty1 insertions, 89 were in the MET3 region and 81 of these were revertible, 57 were in the URA3 region and none were revertible, and 19 were in the small HIS4 region and 2 of these were revertible. All 8 nonrevertible Ura^- mutants mapping to the MET3 region were in the coding region within 270 bp of the MET3/HIS4 border. Since the HIS4 region is only 276 bp, most Ty1 insertions are too close to one end of the HIS4 region to permit efficient gene conversion. Only insertions that occur into the center of the HIS4 region, allowing for ~138 bp of HIS4 homology on each end, would be expected to gene convert to Ura⁺. A comparison of these data with those obtained in the isogenic RAD6 strain, L1716, shows a dramatic difference (P < 0.0001) in Ty1 distribution patterns (Figure 4). These data indicate that the effect of rad6 on the Ty1 target-site distribution is not gene specific. The rad6 mutation did not significantly alter the asymmetric orientation bias at the MET3-URA3 promoter (Table 4).

Deletions of RAD18, FUS3, and SIR genes do not affect Ty1 distribution:
Rad18 is a key member of the RAD6 DNA repair pathway, and it is believed to bring Rad6 to DNA damage sites because it can bind with both Rad6 and single-stranded DNA (BAILLY et al. 1994 , BAILLY et al. 1997A , BAILLY et al. 1997B ). Of 156 independent Ura^- mutants obtained in the rad18 deletion strain, L1707, 127 of them were revertible. This indicates that the vast majority of Ty1 insertions are in the MET3 region in the rad18 strain, and that the Ty1 target-site distribution is not affected (Table 7). Mutations in the FUS3 gene can increase the Ty1 transposition rate dramatically (CONTE et al. 1998 ). We show in Table 7, however, that deletion of FUS3 had no effect on target-site selection.

Because RAD6, HIR3, and CAC3 have all been shown to be involved in transcriptional silencing, the question arose as to whether silencing factors play any role in Ty1 target-site selection. The data shown in Table 7 indicate that deletions of any one of the four SIR genes did not dramatically alter the Ty1 target-site distribution.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

By selecting for gene inactivation, Ty1 elements were previously found to preferentially insert into the promoter or 5' end of the coding sequence of several genes transcribed by RNAP II (EIBEL and PHILIPPSEN 1984 ; SIMCHEN et al. 1984 ; NATSOULIS et al. 1989 ; WILKE et al. 1989 ; LIEBMAN and NEWNAM 1993 ). In these studies, the locations of the Ty1 elements were mapped using PCR, DNA blot hybridization, and/or DNA sequencing. Here, we report a simple genetic system with which the rough position of Ty1 insertions into a target gene, the MET3-URA3 fusion, can be easily scored. We have used this system to screen candidate mutations for effects on Ty1 target-site preferences. The system also allowed us to examine the effects of transcriptional repression or derepression of the target gene on Ty1 transposition-site preference.

Effects of cac3 hir3 mutations on transcriptional regulation of the MET3-URA3 fusion:
The MET3 promoter is tightly regulated (MOUNTAIN et al. 1991 , MOUNTAIN et al. 1993 ) by the presence or absence of methionine. Thus, wild-type strains containing the MET3-URA3 fusion can grow on -Met-Ura medium but not +Met-Ura medium. Also, the 2-kb fusion mRNA is abundant in the absence of methionine, but cannot easily be detected when methionine is added to the culture. Similar results were obtained when strains carrying single deletions in CAC3 and HIR3 were examined. Cac3 is a component of CAF-I, which transports histones H3 and H4 into nucleosomes (KAUFMAN et al. 1997 ), and Hir3 controls the levels and balance of transcription of the different histones (OSLEY and LYCAN 1987 ; MORAN et al. 1990 ; RECHT et al. 1996 ). Mutations in the HIR genes suppress the his4-912 and lys2-128 alleles, which are mutant because of the presence of an upstream Ty1 delta sequence. In these alleles, the normal transcriptional initiation site is replaced by a site within the delta insert, resulting in a longer, nonfunctional transcript. In the presence of hir mutations, the transcription initiation occurs at the normal HIS4 start site in addition to the delta sequence (SHERWOOD and OSLEY 1991 ). Likewise, alterations in histone gene dosages caused by overexpression or deletion of histone gene loci can suppress the delta insertion mutations (CLARK-ADAMS et al. 1988 ). As shown in Figure 3, however, the hir3 mutation itself appears to have no affect on the transcription initiation site of the MET3-URA3 fusion.

In contrast to these results, strains containing simultaneous deletions of CAC3 and HIR3 contained a shortened mRNA complementary to URA3 (~1 kb) under both repressed and derepressed conditions, and they grew on +Met-Ura media. Growth on media lacking uracil under repressed conditions is presumably caused by the presence of this short mRNA, which must contain the complete URA3 coding region of 786 bp and, therefore, appears to be synthesized from a hidden promoter in the MET3 or HIS4 coding region. The normal 2-kb MET3-URA3 mRNA was reduced in the presence of methionine.

Effects of cac3 and hir3 mutations on the Ty1 integration-site preference and orientation bias at the MET3-URA3 target locus:
As described above, relative to their coding regions, the promoter regions of RNAP II transcribed genes are the preferred integration sites for Ty1 elements. In the background of a wild-type strain, we found that this promoter bias occurs whether or not the MET3-URA3 target gene is actively transcribed when Ty1 transposition is induced. We further showed that the Ty1 target-site preference was always present in cac3 strains and always absent in cac3 hir3 strains whether or not MET3-URA3 was transcribed. In hir3 strains, however, the Ty1 insertion-site promoter preference was affected by transcription of the MET3-URA3 target: the preference was present when transcription was repressed, but it was eliminated under derepressed conditions. In the background of a hir3 strain, transcription of the target gene possibly alters its chromatin structure, and this alteration affects the integration of retroelements.

In wild-type strains, the majority of Ty1 elements that inserted into the promoter region were in the same transcriptional orientation as MET3-URA3. In contrast, there was no obvious orientation bias for Ty1 elements inserted into other regions of the fusion. It has been proposed that a similar Ty1 orientation bias at the CAN1 promoter is caused by properties of the genomic target and is not the result of the selection (RINCKEL and GARFINKEL 1996 ). Our results show that in a wild-type strain, transcription of the target gene does not alter the features of the genome responsible for the orientation bias, while in single hir3 mutants, the orientation bias of Ty1 insertions in the promoter region is reduced whether the target gene is repressed or derepressed. This is similar to the previous finding by RINCKEL and GARFINKEL 1996 that the asymmetric insertion pattern of Ty1 elements at the CAN1 promoter is dramatically relaxed in the presence of a hta1-htb1 mutant. They suggest that the altered stoichiometry of histones caused by the hta1-htb1 causes a change in the feature of the target genome that affects the orientation bias. According to this hypothesis, a hir3 mutation, which also alters the relative histone levels, should have a similar effect on orientation bias.

Synergistic effects of cac and hir mutations may result from alterations in chromatin structure:
Simultaneous mutations of CAC3 and HIR3, but not mutation of either gene alone, dramatically increase the Ty1 transposition rate, cause MMS sensitivity, reduce telomeric and HM silencing, and retard growth rate (KAUFMAN et al. 1998 ; QIAN et al. 1998 ). It has been proposed that the simultaneous inactivation of CAF-I and a HIR gene causes alterations in the chromatin structure that result in these phenotypes. Indeed, the telomeric chromatin of cac2 hir1 mutants has been shown to display an increase in accessibility to dam methylase relative to that of the wild-type strain or mutants lacking either gene alone (KAUFMAN et al. 1998 ). The findings reported here showing that cac3 hir3 double mutations alter the target-site bias of Ty1 elements at two loci, MET3-URA3 and CAN1, support the hypothesis that the chromatin structure of the target DNA has been altered. In addition to its dramatic effect on the Ty1 target-site distribution, the data suggest that the double cac3 hir3 mutation may reduce the promoter orientation bias.

One possibility to explain the proposed synergistic interaction between cac and hir on chromatin structure is that in the absence of CAF-I, an alternate pathway deposits H3-H4 into nucleosomes. If the alternative pathway were less efficient than CAF-I, it would be more sensitive to the relative reduction in H3-H4 levels that may be caused by hir mutations. Indeed, the relative reduction of H3-H4 levels in a cac background caused by deletions of genes encoding histones H3 and H4 or amplification of genes encoding H2A-H2B appears to mimic the effect of a hir mutation on telomeric silencing (KAUFMAN et al. 1998 ). It is also possible that the Hir proteins are directly involved in chromatin assembly and interact with the alternative chromatin assembly pathway. The alteration in chromatin structure caused by double mutations in CAC3 and HIR3 may make the DNA more accessible to Ty1 integration events, explaining the increased rate in Ty1 transposition (QIAN et al. 1998 ) and the effects on Ty1 target-site distribution.

Mutations in RAD6 affect the Ty1 target-site distribution at several loci:
We previously showed that mutations in RAD6 alter the Ty1 target-site distribution (LIEBMAN and NEWNAM 1993 ) in the telomere-proximal gene, CAN1, and reduce telomeric and HM silencing (HUANG et al. 1997 ). Mutations in RAD6 also altered a Ty1 hot spot within the SUP4-o gene on a plasmid (KANG et al. 1992 ). These changes were hypothesized to result from alterations in chromatin structure caused by the rad6 mutation. We now show that deletion of RAD6 can also dramatically alter the Ty1 target-site preference into two other genes, the MET3-URA3 fusion and LYS2. Since neither of these genes is located near a telomere, these results show that RAD6 has a general effect on Ty1 target-site selection. Although the rad6 mutation altered the target-site distribution of Ty1 elements at MET3-URA3, it failed to significantly affect the orientation bias. Ty1 elements in the LYS2 gene do not have an orientation bias, even in the RAD6 wild-type background (NATSOULIS et al. 1989 ; BURCK 1996 ). Since we found that double mutations in CAC3 and HIR3 cause alterations in the target-site distribution of Ty1 elements that are similar to those caused by rad6 (LIEBMAN and NEWNAM 1993 ), we also tested the possibility that cac3 hir3 double mutations may inhibit transcription of RAD6. This was found not to be the case.

RAD18, a member of the RAD6 DNA repair epistasis group, encodes a Rad6-binding protein and is thought to bring Rad6 to DNA damage sites (BAILLY et al. 1994 , BAILLY et al. 1997A , BAILLY et al. 1997B ). Deletion of RAD18 did not affect telomeric silencing (HUANG et al. 1997 ) and did not alter the Ty1 target-site preference in the MET3-URA3 fusion.

Not all loci that affect silencing or Ty1 transposition rates have effects on the Ty1 target-site preference at the MET3-URA3 fusion:
RAD6, CAC3, and HIR3 have roles in both silencing and Ty1 transposition, suggesting a connection between these two processes. There are numerous other genes required for silencing, and some of them have been shown to affect DNA repair and recombination (GOTTLIEB and ESPOSITO 1989 ; PAETKAU et al. 1994 ; JACKSON 1997 ; TSUKAMOTO et al. 1997 ). However, we found that deleting any one of the four SIR genes, which are required for silencing, did not affect Ty1 target-site preference. Thus, while the reduction of silencing in the cac3 hir3 strain is an indication of chromatin structure alteration, the defect in silencing itself is not the reason for the loss of Ty1 promoter preference.

Inactivation of FUS3 has recently been shown to increase the transposition rate of Ty1 elements into RNAP III transcribed genes. Furthermore, the steady-state levels of Ty1-associated proteins and cDNA were shown to be increased in fus3 mutants (CONTE et al. 1998 ). This led to the hypothesis that Fus3, a haploid-specific, mitogen-activated protein kinase (ERREDE et al. 1995 ), represses Ty1 transposition by promoting the turnover of Ty1-associated proteins (CONTE et al. 1998 ). Since the effect of Fus3 on transposition is not proposed to involve alteration of the chromatin structure of the target, our finding that deletion of FUS3 does not affect the Ty1 target-site distribution supports this hypothesis. Similarly, LEE et al. 1998 show that mutations that alter the Ssl2 excision repair/transcription factor stimulate Ty1 retrotransposition without altering Ty1 target site preferences.

Comparison between factors affecting Ty1 transposition into RNAP II and RNAP III transcribed genes:
The major chromosomal hot spots for Ty1/Ty2 elements are upstream of RNAP III transcribed genes (GAFNER et al. 1983 ; CURCIO and GARFINKEL 1992 ; OLIVER et al. 1992 ; DEVINE and BOEKE 1996 ). Active targeting of Ty1/Ty2 and Ty3 to regions upstream of RNAP III transcribed genes is always reduced by mutations that compromise transcription of the target gene. No direct requirement for transcription has been proven, however, and it is possible that targeting depends on the binding of certain transcription factors rather than the actual transcription itself (JI et al. 1993 ; DEVINE and BOEKE 1996 ). Indeed, this appears to be the case for Ty3 targeting in vitro (KIRCHNER et al. 1995 ; CONNOLLY and SANDMEYER 1997 ).

In this article, we examined transposition of Ty1 elements into a minor but important type of target, an RNAP II transcribed gene. We show that transcription of the RNAP II transcribed MET3-URA3 gene is not required for the Ty1 insertional bias for promoter regions. As hypothesized above for the RNAP III transcribed genes, proteins bound to the promoter region of RNAP II transcribed genes may attract Ty1 elements whether transcription is repressed or derepressed. RNAP II and RNAP III initiation complexes are known indeed to be composed of some of the same protein components (for review see GABRIELSEN and SENTENAC 1991 ; STRUHL 1995 ).

Mutations in FUS3 cause an 18- to 56-fold increase in the transposition frequency of Ty1 elements marked with his3AI (CONTE et al. 1998 ), which predominantly integrate into chromosomal hot spots, such as upstream of tRNA and other RNAP III transcribed genes (CURCIO and GARFINKEL 1991 , CURCIO and GARFINKEL 1992 ; DEVINE and BOEKE 1996 ), and a 6-fold increase in the Ty1 transposition rate into the RNAP II transcribed CAN1 gene (CONTE et al. 1998 ). The increased stability of the Ty1 viruslike particles and the increase in Ty1 cDNA caused by fus3 apparently increase the frequency with which Ty1 elements can transpose into both RNAP III and RNAP II transcribed genes.

As we reported previously (QIAN et al. 1998 ), double cac3 hir3 mutations cause an increase in the Ty1 transposition rate of 20- to 60-fold into a promoterless his3 allele and of 10-fold into the CAN1 gene. The transposition frequency of a Ty1 element marked with his3AI was increased ~4-fold in cac3 hir3 mutants (QIAN et al. 1998 ). Since there is a considerable difference between the effects of cac3 hir3 mutations on the transposition rates at two different RNAP II transcribed genes, it not clear if the lower increase in transposition frequency associated with the his3AI system represents a fundamental difference in transposition pathways associated with RNAP II and RNAP III transcribed gene targets. cac3 hir3 mutations possibly cause changes in chromatin structure at Ty1 chromosomal hot spots, but these changes do not significantly increase their availability to Ty1 insertion because of the high basal rate of transposition at the hot spots. Alternatively, cac3 hir3 may not alter the chromatin structure of RNAP III transcri