Saccharomyces cerevisiae strains used in this study are BY4741 (MATa, his3Δ1, leu2Δ0, met15Δ0, ura3Δ0) and congenic strains with a single non-essential ORF replaced by a kanMX cassette (Brachmann et al., 1998; Giaever and Nislow, 2014). Six additional strains, each containing a kanMX-tagged DAmP allele of an essential gene (Breslow et al., 2008), were obtained from Thermo Scientific. Strain JC6464 is a derivative of BY4741 constructed by retrotransposition of Ty1kanMXAI into the host genome, as described previously. Strain JC6474 is the med15∆::URA3 derivative of JC6464 (Salinero et al., 2018).

Yeast media used included Yeast Peptone Dextrose (YPD: 1% bactoyeast extract, 2% bactopeptone extract, 2% glucose) broth and agar. Mutant strains harboring deletion and hypomorphic alleles marked with kanMX were taken from frozen stocks in 15% glycerol and grown as single colonies on YPD agar containing 200 μg/ml Geneticin (G418) prior to transformation with plasmid DNA. A standard lithium acetate transformation protocol was used to introduce plasmid DNA into yeast strains (Gietz and Schiestl, 2007). Strains containing a plasmid were grown in Synthetic Complete (SC) broth [0.67% bactoyeast nitrogen base without amino acids (6.7 g/L), 0.08% SC dropout powder (0.8 g/L)] with 2% glucose as a carbon source unless otherwise indicated. SC dropout powder lacking leucine (SC-Leu), histidine (SC-His), leucine and histidine (SC-Leu-His), uracil (SC-Ura) or uracil and histidine (SC-Ura-His) were obtained from Sunrise Science Products.

Plasmid-borne Ty1 elements used in this study carry the his3AI-[∆1] retrotransposition indicator gene (Scholes et al., 2001) cloned into the BglII site between the pol ORF and the 3′ LTR of Ty1-H3. DNA recombination between the his3AI-[∆1] allele and the his3∆1 allele in strain BY4741 does not result in a functional HIS3 allele.

Plasmid pBJC1250 consist of the LEU2-CEN vector pRS415 carrying the element P _TEF1 -Ty1his3AI-[∆1], in which the U3 region of the 5′ LTR in Ty1-H3 is replaced with a TEF1 promoter (P _TEF1 ) (Salinero et al., 2018). P _TEF1 was amplified with flanking ApaI and XhoI sites by PCR using plasmid pUG6 (Guldener et al., 1996) as template DNA and primers PTEFforward and PTEFreverse (Table 1). The PCR product was digested with ApaI and XhoI, and used to replace the ApaI-XhoI fragment containing the GAL1 promoter on plasmid pBJC838 (Stamenova et al., 2009).

Plasmid pBDG633 consists of the URA3-CEN vector YCp50 carrying a functional, his3AI-[∆1]-tagged hybrid Ty1-912/Ty1-H3 element (Lee et al., 1998) in which the U3 region of the 5′ LTR of Ty1-912 is fused to the transcriptional start site of Ty1-H3, referred to as U3 (912)-Ty1his3AI. There are five nucleotide differences between the U3 domain of Ty1-912 and that of Ty1-H3, but none of these occur in a known regulatory motif (Clare and Farabaugh, 1985; Boeke et al., 1988). Plasmid pBDG633 was generously provided by Dr. David Garfinkel.

Plasmid pBJC998 consists of the LEU2-CEN vector pRS415 carrying a U3 promoter-driven Ty1his3AI-[∆1] element, herein referred to as U3(H3)-Ty1his3AI. The U3 region of the 3′ LTR of Ty1-H3 in plasmid pGTy1-Cla (Garfinkel et al., 1988) was amplified with flanking ApaI and XhoI sites using primers PJ762 and PJ763. The U3(H3) PCR fragment was digested with ApaI and XhoI and used to replace the ApaI-XhoI fragment on plasmid pBJC838 (Stamenova et al., 2009).

Plasmid pBJC1270 is a derivative of plasmid pLTR_p:Gag_1-401:GFP:ADH1 _TER (Doh et al., 2014) in which the first AUG codon of the GAG ORF has been changed to CUG. Two PCR reactions were performed using pBJC998 DNA as a template and primers PJ795 and PJ797 or primers PJ796 and PJ749. The resulting PCR products were purified and then combined as a template for overlap extension PCR performed with primers PJ795 and PJ749. This final PCR product, containing the U5 region of the LTR and beginning of the GAG ORF, was digested with XhoI and HpaI, and then used to replace the XhoI-HpaI fragment in plasmid pLTR_p:Gag_1-401:GFP:ADH1 _TER. The XhoI-HpaI fragment was sequenced to ensure that it contained only the intended nucleotide substitution.

Plasmid pBJC1279 consists of the LEU2-CEN vector pRS415 carrying a PSP2 promoter-driven Ty1his3AI-[∆1] element, referred to as P _PSP2 -Ty1his3AI. The plasmid was constructed by amplifying the PSP2 promoter (Chr XIII, nucleotide 235889–236513) from strain BY4741 genomic DNA using primers PJ1480 and PJ1481. The promoter was cloned into plasmid BJC838 digested with ApaI and XhoI using New England Biolabs NEBuilder^® HiFi DNA Assembly Master Mix.

As a screen for retromobility of P _TEF1 -Ty1his3AI, U3 (912)-Ty1his3AI and P _PSP2-Ty1his3AI in different mutants, a semi-quantitative retromobility patch assay that measures the relative level of His⁺ papillation in cells in a ∼1 cm² patch was performed. Patches of cells from four to seven independent transformants of plasmid pBJC1250, pBDG633 or pBJC1279 in strain BY4741 or a mutant derivative were spread on SC-Leu or SC-Ura agar, and plates were incubated at 30°C until patches of the mutant strains had grown to a similar density as patches of the wild-type strain BY4741 (2–4 days). The SC-Leu and SC-Ura plates were photographed, and then replicated to YPD agar and grown at 20°C for three to 6 days. After photographing the YPD plates, they were replicated to SC-His-Leu or SC-His-Ura agar, incubated for 4 days at 30°C., and photographed.

Single colonies of strain BY4741 and mutant derivatives harboring plasmid pBJC838 (Stamenova et al., 2009), which carries the P _GAL1 -Ty1his3AI element, were grown in SC-Leu broth at 30°C overnight. Cells were diluted 1:100 in SC-Leu + 2% raffinose + 2% sucrose + 2% galactose and grown for 3 days at 20°C. A 1 or 2 µl aliquot was plated on SC-Leu agar and an aliquot of 10–1,000 µl was plated on SC-Leu-His agar. Plated cells grew at 30°C for three to 5 days. Colonies were counted, and the retromobility frequency of each culture was calculated as the number of His⁺ Leu⁺ colonies divided by the total number of Leu⁺ colonies in 1 ml of culture. The mean retromobility frequency (MRF) of seven independent transformants of the same genotype was determined. The relative retromobility frequency of each strain is the mean retromobility frequency (MRF) of seven independent transformants of the mutant compared to the mean retromobility frequency (MRF) of seven independent transformants of the wild-type strain measured in parallel at the same time.

Independent transformants of strain JC6464 and JC6474 harboring plasmid pBJC1279 were grown overnight in SC-Leu broth at 30°C. Cultures were diluted 1:1,000 in YPD broth and grown at 20°C for 3 days. A 1:1,000 dilution of each culture was plated on YPD agar to determine the number of cells (colony forming units) in each culture. Aliquots of each culture were plated on YPD agar with 200 μg/ml Geneticin to determine the number of G418^R colony forming units per culture, and on SC-His agar to determine the number of His⁺ colony forming units per culture. Plates were grown for 3 days at 30°C before counting colonies. The Ty1kanMXAI retromobility frequency was calculated for each transformant as the number of G418^R colonies divided by the total number of cells in an equal volume of culture, and the Ty1kanMXAI mean retromobility frequency (MRF) was the average from six independent transformants. The P _PSP2 -Ty1his3AI retromobility frequency was calculated for each transformant as the number of His⁺ colonies divided by the total number of cells in an equal volume of culture, and the P _PSP2 -Ty1his3AI mean retromobility frequency (MRF) was the average from the same six independent transformants.

Gene ontology enrichment analysis was performed using the Gene Ontology Term Finder at the Saccharomyces genome database (SGD) (Cherry et al., 2012). Relative growth rates for 1,312 mutant strains was downloaded from (O'Duibhir et al., 2014). p-values using the Mann-Whitney U-test were calculated using the online calculator at https://astatsa.com/WilcoxonTest/. Previously published ChIP-seq data used in Supplementary Presentation 1 is archived at the NCBI BioProject database (https://www.ncbi.nlm.nih.gov/bioproject/) under accession number SRP047524 (Paul et al., 2015) and PRJNA657372 (Sarkar et al., 2022).

Yeast was grown in YPD broth at 30°C overnight. Cultures were diluted to an OD₆₀₀ of 0.1 in 50 ml YPD broth and incubated at 20°C until an OD₆₀₀ of 0.6–0.8 was reached. Cells were pelleted, and pellets were washed in water before being frozen on dry ice and stored at −80°C. Total cellular RNA was isolated using a hot phenol extraction method (Collart and Oliviero, 2001). Between 250 µg and 1 mg of total cellular RNA was used to purify polyA⁺ RNA with the Magnetic mRNA Isolation Kit (New England Biolabs) following the manufacturer’s protocol. Purified PolyA⁺ RNA (6 µg) was incubated in glyoxal loading dye (Ambion) for 30 min at 50°C. and fractionated on a 0.8% SeaKemGTG agarose gel. RNA was transferred to a HybondXL membrane (Amersham) utilizing a SSC gradient transfer system between 6x SSC and 10× SSC overnight at room temperature. Following transfer, membranes were UV crosslinked using the Optimal Crosslinking setting on the Spectrolinker XL-1500 (Spectronics Corporation). Membranes with bound RNA were stored at 4°C in 5× SSC.

Riboprobes labeled with α-³²P-CTP were synthesized using an SP6 polymerase in vitro transcription reaction (Promega). The template for the antisense riboprobe that hybridizes to Ty1 gRNA and Ty1i RNA was plasmid pGEM-TyA1 digested with XhoI (Curcio et al., 1990); for the antisense riboprobe that hybridizes to PYK1 RNA, plasmid pGEM-PYK1 digested with ClaI (Curcio and Garfinkel, 1992). Riboprobes purified on a NucAway spin column (Thermo Fisher) were added to membranes in 25 ml NorthernMax Prehybridization buffer (Ambion) and allowed to hybridize to membranes overnight at 65°C. Membranes were washed in 2× SSC, 0.1% SDS for 45 min at 65°C, 0.1× SSC, 0.1% SDS for 45 min at 65°C and twice in 0.1% SDS for 10 min at 65°C. Bands were imaged on a Typhoon 9400 phosphoimager and relative quantifications were obtained using ImageQuant software (Molecular Dynamics). Membranes probed with a Ty1 riboprobe were stripped by washing twice in boiling 0.1% SDS incubated at 70°C for 10 min. Stripped membranes were then re-probed to detect PYK1 mRNA, and the signal quantitated as above.

Strain BY4741 and selected mutant derivatives were transformed with the vector, pRS415 (Sikorski and Hieter, 1989) or plasmid pBJC1270. Three independent transformants of pRS415 and another three of pBJC1270 in each strain were grown for 18–24 h in SC-Leu broth at 30°C. The OD₆₀₀ of each culture was determined, and cells were diluted in SC-Leu broth to an OD₆₀₀ of 0.2. Diluted cultures were grown at 20°C for 3 hours before measuring fluorescence with a FACSCalibur (Becton Dickinson) flow cytometer. The geometric mean of fluorescence activity in 10,000 cells was measured for each pBJC1270 transformant and each pRS415 transformant. The geometric means of three pRS415 transformants of a specific strain were averaged to obtain a value for autofluorescence of that strain. This value was subtracted from the geometric mean of each of three pBJC1270 transformants in that strain, and the three resulting values were averaged to obtain the p22-Gag:GFP activity for each strain.

To identify RTT and RHF genes that regulate Ty1 retromobility via the U3 promoter, mutant alleles were tested for their effects on retromobility of a Ty1his3AI element in which the U3 promoter was replaced with the TEF1 promoter (P _TEF1 ; Figure 1A). The TATA-less, TFIID-dominated TEF1 promoter was chosen because it has a different architecture than the TATA-containing, SAGA-dependent U3 promoter and because P _TEF1 is a strong promoter (Schirmaier and Philippsen, 1984; Lee et al., 2015), whereas U3 is relatively weak (Morillon et al., 2002) (Supplementary Presentation 1). The P _TEF1 promoter was fused to the Ty1-H3 element at the transcription start site such that only U3 and no transcribed sequences are deleted in P _TEF1 -Ty1his3AI. The retrotransposition indicator gene, his3AI allows retromobility of a Ty1 element to be measured in a simple, quantitative genetic assay; it consists of an artificial intron (AI) within the coding region of the HIS3 gene in an antisense orientation. The his3AI gene is inserted in the 3′ untranslated region of the Ty1-H3 element in the opposite transcriptional orientation (Figure 1A); therefore, the AI cannot be spliced from the his3AI transcript but is in the correct orientation to be spliced from Ty1his3AI RNA. The use of a spliced Ty1his3 transcript as a template for reverse transcription generates cDNA in which the functional HIS3 allele is restored. Integration of this Ty1HIS3 cDNA into the host genome, either by integration or by recombination of the cDNA with preexisting Ty1 sequences in the genome, allows for expression of HIS3 and renders the cell phenotypically His⁺. The frequency of His⁺ prototroph formation is a direct measure of Ty1his3AI retrotransposition or cDNA recombination, together known as retromobility (Curcio and Garfinkel, 1991).

A CEN-based, low copy plasmid harboring the P _TEF1 -Ty1his3AI element was introduced into strains bearing mutant alleles of 188 RTT and RHF genes (Supplementary Table 1). Most of the rtt and rhf mutations analyzed were deletions of non-essential genes identified in one or more of four large-scale screens that used a chromosomal or integrating plasmid-based Ty1his3AI element as a retromobility reporter (Scholes et al., 2001; Nyswaner et al., 2008; Dakshinamurthy et al., 2010; Risler et al., 2012). Retromobility host factors that were identified in a screen employing a GAL1 promoter-driven Ty1 element were excluded from our study, since by definition, these host factors do not require the Ty1 promoter to control retromobility (Griffith et al., 2003). Like Med3 and Med15, many of the RHFs and RTTs identified in the four Ty1his3AI screens impact Ty1 cDNA levels without substantially altering levels of Ty1 RNA or its primary translation product, Gag (Lee et al., 1998; Rattray et al., 2000; Scholes et al., 2003; Sundararajan et al., 2003; Curcio et al., 2007; Dutko et al., 2010; Risler et al., 2012; Salinero et al., 2018). To increase the likelihood of identifying U3-dependent regulators, we excluded rtt and rhf mutants that were reported to have no change in cDNA levels, since these mutations are more likely to affect Ty1 cDNA utilization than a potentially promoter-dependent step such as transcription or RNA fate. Two additional non-essential genes analyzed were newly identified RHF genes, CAF130 and NOT3, which encode subunits of the CCR4-NOT complex, a master regulator of gene expression from transcription to mRNA degradation (Collart, 2016; Buschauer et al., 2020). Hypomorphic DAmP alleles of six essential RHF genes were also analyzed. DAmP alleles harbor the kanMX gene inserted into the 3′ untranslated region of the essential gene, thereby destabilizing and reducing the steady-state level of mRNA produced (Breslow et al., 2008). The six essential RHF genes include SRP68 (Doh et al., 2014) and five others that we identified in a small screen of essential genes related to known regulators of retromobility (CDC36/NOT2, CDC39/NOT1, MSL5, NDC1, RPB7). As controls, mutants harboring a deletion of CHK1, which has no effect on Ty1 retrotransposition (Curcio et al., 2007), or SPT3, which encodes a SAGA complex component that activates transcription of Ty1 and represses transcription of Ty1i (Winston et al., 1984; Grant et al., 1997; Saha et al., 2015), were included in our analysis.

A semi-quantitative patch assay was used to measure retromobility in four to six transformants of the CEN- P _TEF1 -Ty1his3AI plasmid in wild-type and mutant strains (Figure 1B). Transformants, grown initially as small patches of cells at 30°C on SC-Leu agar to maintain the plasmid, were replicated to YEPD agar and grown at 20°C, as these are optimal conditions for retromobility. Growth at 20°C continued until patches of mutant cells and wild-type cells reached equivalent densities. Patches were then replicated to SC-Leu-His agar and incubated at 30°C to allow growth of cells that had retained the plasmid and sustained a retromobility event. The level of His⁺ Leu⁺ papillation in transformants of each mutant was compared to that in the wild-type strain grown on the same plate (Figure 1B). Transformants of the spt3∆ mutant were included on each plate as an additional control. Images of mutants are shown in Supplementary Data Sheet 1. Classification of mutants and test plate number for each mutant is provided in Supplementary Table 1.

The spt3∆ mutant had few or no His⁺ Leu⁺ papillae on all P _TEF1 -Ty1his3AI screen plates (Figure 1B; Supplementary Data Sheet 1); the difference in His⁺ papillae numbers between the wild-type strain and the spt3∆ mutant provides a partial range of the assay for each plate. The chk1∆ mutant, which has no retromobility defect (Curcio et al., 2007), had wild-type levels of P _TEF1 -Ty1his3AI retromobility, as expected (Plate 42, Supplementary Data Sheet 1). Based on prior studies, both the dhh1∆ mutant and the srp68-DAmP mutant were expected to have reduced P _TEF1 -Ty1his3AI retromobility, since Dhh1 is required for P _GAL1 -driven retromobility of Ty1his3AI (Dutko et al., 2010), and Srp68 is required co-translationally for Ty1his3AI retromobility (Doh et al., 2014). Indeed, both genes were required for efficient retromobility of P _TEF1 -Ty1his3AI (Plates 55 and 63, respectively; Supplementary Data Sheet 1).

The P _TEF1 -Ty1his3AI screen identified 82 of the 188 rhf and rtt mutants as having a level of retromobility that was either higher or lower than that of the wild-type strain, using a threshold of ∼2-fold increase or decrease in His⁺ Leu⁺ papillae (Figure 1C; Supplementary Table 1). The corresponding factors regulate retromobility regardless of whether Ty1his3AI RNA is driven by P _TEF1 or U3, and thus were classified as U3-Independent Regulators, or U3IRs. 76 mutants (including dhh1∆ and srp68-DAmP) were previously reported regulators of Ty1his3AI retromobility. Mutants harboring DAmP alleles of all five newly identified, essential RHF genes, CDC36/NOT2, CDC39/NOT1, MSL5, NDC1 and RPB7, are also included in the cohort of 82. The final one is not3∆, another newly identified rhf mutant. Greater than 80% of these U3IRs are retromobility co-factors.

As an additional test of the idea that U3IRs regulate retromobility regardless of the strength or architecture of the promoter driving Ty1his3AI expression, retromobility was measured quantitatively in eight rhf-u3ir mutants using a CEN-plasmid-based Ty1his3AI element driven from the inducible GAL1 promoter (P _GAL1 ; Figure 1D). P _GAL1 is a strong promoter like P _TEF1 , but it contains a TATA box, like U3. Retromobility of P _GAL1- Ty1his3AI in the dhh1∆ mutant was 2% of that in a wild-type strain, consistent with previous results (Dutko et al., 2010), and less than 10% of wild-type in the hcr1∆, hmo1∆, rpl27a∆, rpl31a∆ or rpl43a∆ mutants. Retromobility in the loc1∆ and rpl19a∆ mutant was 26 and 78% of that in the congenic wild-type strain, respectively; these decreases were small but significant (p < 0.00001 and p < 0.05, respectively; Student’s t-test). Together with the results of the P _TEF1 -Ty1his3AI assay, these findings suggest that U3IRs act in a manner that is at least partially independent of the expression level of Ty1 RNA and the architecture of the U3 promoter.

A subset of 106 rtt and rhf mutants had no detectable defect in P _TEF1 -Ty1his3AI retromobility relative to the wild-type strain (Figure 1C). To identify the members of this set whose phenotype reflects the substitution of P _TEF1 for U3 and not the Ty1his3AI element being located on an extrachromosomal plasmid, we tested retromobility of a U3-driven Ty1his3AI element on a CEN-based plasmid in 105 of the 106 rtt and rhf mutants and the chk1∆ negative control (Supplementary Table 2). One mutant, erg6∆, could not be transformed with the U3 (912)-Ty1his3AI plasmid in multiple attempts. Transformants were grown as patches on SC-URA agar at 30°C, replicated to YPD agar and grown at 20°C, and then replicated to SC-Ura-His agar to select for His⁺ papillae in cells that retained the plasmid. A detailed summary of the results of the U3-Ty1his3AI screen and plate reference numbers can be found in Supplementary Table 2. Images of each plate showing growth and papillation for multiple transformants of each strain are provided in Supplementary Data Sheet 2.

Fifty-five mutants had wild-type levels of retromobility of the U3-Ty1his3AI element (Figure 1C). The 55 corresponding genes do not affect retromobility of a P _TEF1 - or U3-driven Ty1his3AI element on a CEN-plasmid in this strain background. These results do not necessarily contradict the results of earlier studies in which the genes were identified as regulators of Ty1his3AI retromobility; instead, they may indicate that these genes act in a context-dependent fashion. The critical context may not have been reproduced in our screen because the Ty1his3AI element is in a different chromatin environment or because of host strain differences. These host factors were classified as conditional regulators (CR) of retromobility.

A second major class of mutants identified by this screen contains 51 deletion mutants with a Ty1his3AI retromobility phenotype that differed from that of the wild-type strain (Figure 1C). The 51 corresponding genes regulate retromobility of a CEN-based, U3-driven but not P _TEF1 -driven Ty1his3AI element. This set of U3-Dependent Regulators, or U3DRs, contains 23 RTTs and 28 RHFs including Snf2, a component of the SWI/SNF complex that is required for Ty1 RNA transcription (Ciriacy et al., 1991; Happel et al., 1991). Rtt101, which restricts the mobility of a U3-driven but not P _GAL1 -driven Ty1his3AI element (Curcio et al., 2007), was also found in this class, as predicted. These findings validate the screening method used to identify U3DRs. Both components of the nuclear exosome that were in the cohort of 188 Ty1 regulators, Rrp6 and Lrp1, were indentified as U3DRs.

While all 51 u3dr mutants were previously found to affect Ty1his3AI retromobility, six of the 51 (12%) had the opposite effect on CEN-Ty1his3AI retromobility as that reported previously (Sundararajan et al., 2003; Nyswaner et al., 2008; Dakshinamurthy et al., 2010). Deletion of ARG4, CKB2, RAD27 or YLR282C resulted in hypermobility of a chromosomal Ty1his3AI element in previous studies, but had decreased retromobility of the CEN-Ty1his3AI element; similarly, pap2∆ and ssk1∆ had reduced levels of chromosomal Ty1his3AI retromobility in a previous study but had substantially increased retromobility of the CEN- Ty1his3AI element (Supplementary Table 2; Supplementary Data Sheet 2). This unexpected finding underscores the influence of the context of the Ty1his3AI element on the retromobility phenotype of certain mutants.

To determine whether the subset of 82 U3IR genes differs functionally from the subset of 51 U3DR genes, we performed Gene Ontology (GO) analysis. We further divided the genes in these two categories, and in the starting cohort of 188 genes, into RTT and RHF genes for this analysis. The complete set of enriched GO categories having p < 0.01 after correcting for multiple category testing is contained in Supplementary Table 3. Many categories contain highly overlapping gene sets; representative enrichments are depicted in Figure 2A, where we focused on categories having p < 10⁻⁵ for at least one gene set, and including fewer than 1,000 genes. As expected, distinct GO enrichments were observed for RTT and RHF genes. RTT genes were enriched for genes in the related categories of DNA repair and recombination, telomere maintenance, and response to stress (Figure 2A; Supplementary Table 3). Enrichment in these categories was also seen for both U3DR and U3IR cohorts, suggesting that distinct genes within these categories can affect transposition in both promoter-dependent and independent manners. Four of the eight genes in the GO category of Negative Regulators of Transposition (GO:0010529) were among the 22 U3DR-RTT genes. The enriched category of histone ubiquitylation seen for U3IR-RTT genes includes two members of the CDC73-PAF1 complex, CDC73 and LEO1. Other members of this complex were not tested. The 29 U3DR-RHF genes showed no enriched categories using our criteria, while the 68 U3IR-RHF genes were enriched for genes involved in mRNA catabolism, gene expression, and transcriptional elongation, the latter including four members of the CCR4-Not complex out of five that were tested (Figure 2; Supplementary Table 3).

A previous study identified a cohort of 1312 S. cerevisiae genes contributing to gene expression changes in a large number of transcriptome studies; deletion strains corresponding to these genes share the property of slow growth and have a common gene expression signature that was primarily attributed to a redistribution of cells over the cell cycle, with a greater fraction of cells in G1 (O'Duibhir et al., 2014). RHF and U3IR genes (which are mostly RHF genes) are enriched for this slow growth gene set and show substantial effects on growth when mutated, while RTT and U3DR genes exhibit only modest effects (Figure 2B). Ty1 retrotransposition is thought to be cell-cycle dependent (Xu and Boeke, 1991; Curcio et al., 2007; Curcio et al., 2015); this result suggests that many of the mutations identified as causing decreased Ty1 mobility do so in part by extending the G1 phase of the cell cycle. However, it should be emphasized that this is almost certainly only a partial effect, as many mutants that result in longer doubling times do not alter Ty1 mobility.

Host factors that require the U3 promoter for their effects on retromobility might be expected to regulate Ty1 or Ty1i RNA levels. In mutants lacking one of the U3-dependent RHF genes MED3 or MED15, Ty1 gRNA levels were similar to that in the wild-type strain, but Ty1i RNA levels were increased, resulting in elevated levels of the p22-Gag restriction factor (Salinero et al., 2018). We screened other u3dr-rhf mutants identified here for a substantially lower level of Ty1 gRNA or detectable level of Ty1i RNA that could result in a hypomobility defect. A northern blot of oligo-dT purified RNA from 21 u3dr-rhf∆, two u3ir-rhf∆ and four CR/rhf∆ mutants was performed using a riboprobe that hybridizes to both Ty1 RNA and Ty1i RNA, and the levels of Ty1 gRNA relative to a control transcript, PYK1 mRNA, were measured. (Supplementary Presentation 2). Compared to the relative Ty1 gRNA level in the wild-type strain BY4741, none of these 27 rhf∆ mutants had Ty1 gRNA levels that were decreased ≥2-fold except for the CR/rhf∆ mutant, vps34∆, which had a 20-fold decrease in Ty1 gRNA. Thus, a substantial decrease in Ty1 gRNA is not likely to be the cause of the hypomobility phenotype of the 21 u3dr-rhf∆ mutants that were screened. For the most part, we also failed to detect a conspicuous Ty1i RNA band in these mutants, except in the rpl7a∆ mutant, which has previously been reported to have an elevated level of Ty1i RNA (Ahn et al., 2017). It is possible that diffuse signals below the Ty1 gRNA bands in the rrp6∆, cst1∆ and scj1∆ mutants are also indicative of increases in Ty1i RNA, but a more thorough analysis is necessary to determine whether this signal is due to smearing of the Ty1 gRNA band or increased Ty1i RNA. In summary, most of the U3-dependent and U3-independent rhf mutants tested did not have a conspicuous increase in Ty1i RNA that accompanied their decreased Ty1 retromobility.

Oligo-dT purified RNA from 16 u3dr-rtt∆ mutants, four u3ir-rtt∆ mutants and six CR/rtt∆ mutants was also compared to that of the wild-type strain to determine whether increased Ty1 gRNA underlies the elevated retromobility in these mutants. Only the u3dr-rtt mutants rtt109∆ and mms1∆ had an increase of ≥2-fold (2.7-fold and 2-fold, respectively; Supplementary Presentation 2). A previous report saw no increase in total Ty1 gRNA in the rtt109∆ mutant (Scholes et al., 2001); potentially, the difference between our results reflects the use of different strains or the use of total RNA earlier and polyA⁺ RNA here. The same study observed a 2-fold increase in steady-state Ty1 gRNA levels in the mms1∆ (rtt108∆) mutant, in agreement with our results. We conclude that higher Ty1 RNA expression in the mms1∆ and rtt109∆ mutants may be an underlying cause of higher retromobility levels. The other 24 rtt mutants tested had only minor changes in Ty1 gRNA compared to the wild-type strain. Together, these findings indicate that most of the rhf and rtt mutants tested do not have substantially altered levels of Ty1 gRNA that are consistent with their retromobility defect; thus, some U3 promoter-dependent host factors may act at a post-transcriptional step in retromobility or affect expression of Ty1i RNA. While major increases in Ty1i RNA were not observed in this single northern, resolving Ty1i RNA presents a challenge, and it may be that subtle changes in Ty1i RNA levels can have more pronounced effects on retromobility. Furthermore, Ty1i RNA was not detected in the wild-type strain, so northern blotting could not be used to determine whether a decrease in Ty1i RNA accompanies the increase in retromobility in u3dr-rtt mutants. These challenges prompted us to develop a sensitive assay to measure the relative level of Ty1i RNA product, p22-Gag, in u3dr-rtt and other mutants.

To compare levels of the self-encoded Ty1 restriction factor among hypermobile rtt and hypomobile rhf mutants, we designed a CEN-plasmid reporter construct that expresses p22-Gag:GFP from the Ty1i promoter (Figure 3). Since the Ty1i promoter has not been completely delineated and may overlap regions of the Ty1 promoter, the plasmid included the entire 5′ LTR and most of the GAG ORF, up to the C-terminal protease cleavage site, where GFP was fused (Doh et al., 2014). A single A to C nucleotide substitution was introduced in the most upstream start codon within GAG, such that p49-Gag:GFP can no longer be translated from Ty1 gRNA, and protein expression from a downstream in-frame AUG at nucleotide 321 is obstructed by four out-of-frame AUG codons between CUG₁ and AUG₃₂₁ (Dvir et al., 2013). However, p22-Gag is not expressed from Ty1 gRNA, and the nucleotide substitution is not present in Ty1i RNA, so p22-Gag:GFP expression should not be affected by this nucleotide substitution (Figure 3). The p22-Gag:GFP expression vector was introduced into strain BY4741 and mutant derivatives, and the relative level of GFP activity in 10,000 individual cells of each of three biological replicates was measured by flow cytometry.

To determine whether this assay yields results that are comparable to those in earlier studies, we compared p22-Gag:GFP expression in the wild-type strain to that of the spt3∆ mutant, which has little Ty1 gRNA but increased levels of Ty1i RNA and p22-Gag (Saha et al., 2015). In two individual experiments, deletion of SPT3 resulted in a 7- or 10-fold increase in the p22-Gag:GFP level (p < 0.001 to p < 0.000001, Student’s t-test; Figure 3). Retromobility mutants, including control strains lacking Mediator subunits Med2 and Med20, were then compared to the wild-type strain assayed in parallel in one of six separate experiments. In the hypomobile med2∆ mutant, p22-Gag:GFP was increased 1.57-fold (p < 0.05; Student’s t-test), in agreement with previous findings of elevated Ty1i RNA and p22-Gag (Salinero et al., 2018). In contrast, p22-Gag:GFP activity in the hypermobile med20∆ strain decreased to 71% of that in the wild-type strain (p < 0.05, Student’s t-test; note that because GFP activity of mutants was compared to that of wild type assayed in parallel and not all mutants were assayed together, significance is not necessarily apparent from visual inspection of expression level and standard deviation). This reduction in p22-Gag:GFP activity is consistent with the hypermobility of a med20∆ strain but decreased expression of Ty1i RNA or p22-Gag was not detectable by northern or western blot analysis (Salinero et al., 2018). Together, these findings support the use of this assay to monitor p22-Gag expression, and suggest that the assay provides greater sensitivity in detecting decreased Ty1i or p22-Gag expression than northern or western blot analysis.

The p22-Gag:GFP activity in three u3dr-rtt mutants was measured. The elg1∆ mutant had reduced GFP activity (p < 0.05, Student’s t-test), but rtt109∆ and rad54∆ did not. Of seven u3dr-rhf mutants tested, only rpl7a∆ and snf2∆ had significant increases in p22-Gag:GFP levels consistent with their hypomobility phenotypes (p < 0.01 and p < 0.00001, respectively; Student’s t-test). An elevated level of p22-Gag in the rpl7a∆ mutant parallels the elevated level of Ty1i RNA observed by northern blot analysis (Supplementary Presentation 2). The other five u3dr-rhf mutants, rrp6∆, rps19b∆, rps25a∆, itc1∆ and cst6∆ did not have significantly elevated levels of GFP activity (Figure 3), consistent with the absence of a discrete Ty1i RNA band in northern blot analysis (Supplementary Presentation 2). Together these findings reveal that a host factor’s dependence on the U3 promoter is not necessarily predictive of it modulating the level of Ty1i RNA or p22-Gag.

Other classes of mutants were examined for comparison. Three U3-independent rhf mutations, upf3∆, loc1∆ and hcr1∆ had little effect on p22-Gag:GFP levels. Notably, two mutants that had a wild-type Ty1his3AI-CEN retromobility phenotype did show significant changes in p22-Gag:GFP activity. The gcr2∆ mutant, which has been reported to have a hypermobility phenotype (Nyswaner et al., 2008), had a reduced amount of p22-Gag:GFP activity (p < 0.05), while opi3∆, shown previously to have hypomobile phenotype (Dakshinamurthy et al., 2010), had more GFP activity than the wild-type strain (p < 0.05) (Figure 3). These changes in the amount of p22-Gag:GFP were detected in the absence of any detectable change in Ty1i RNA (Supplementary Presentation 2). Since Gcr2 and Opi3 modulate Ty1 retromobility under certain conditions, they may do so by controlling p22-Gag expression. Overall though, our findings suggest that some U3-dependent regulators, notably Elg1, Rpl7a and Snf2, can modulate the level of Ty1i RNA or p22-Gag in a manner consistent with their retromobility phenotype, while others may regulate retromobility by mechanisms that do not directly impact expression from the Ty1 or Ty1i promoter. The latter possibility is explored below.

The U3-dependence of some modulators of Ty1 retromobility could be related to U3 being a weak promoter. A co-factor could improve the translational efficiency of Ty1 gRNA translation when U3 drives Ty1 expression but be dispensable when higher levels of Ty1 RNA are expressed from a stronger promoter. Another possibility is that host factors exhibit dependence on U3 because of its architecture. U3DR host factors could be directed via an interaction with the U3 promoter to associate with Ty1 RNA and control the modification, export, localization or translation of Ty1 RNA or Ty1i RNA (Zid and O'Shea, 2014; Espinar et al., 2018). We asked whether U3DR genes fall into one or both of these functional categories by testing retromobility of a Ty1 element driven by the P _PSP2 promoter in the corresponding mutants (Figure 1A). P _PSP2 is a weak promoter, like U3 (Supplementary Presentation 1), but is TATA-less and Taf1-enriched (Rhee and Pugh, 2012), and is insensitive to rapid depletion of SAGA (Donczew et al., 2020), indicating that it is more similar in architecture to P _TEF1 than to the SAGA-dominated U3 promoter. A u3dr mutation that does not affect P _PSP2 -driven Ty1his3AI retromobility by definition also does not affect P _TEF1 -driven Ty1his3AI retromobility and therefore is likely to depend primarily on the architecture of U3 rather than the level of expression. On the other hand, any u3dr mutation that also affects P _PSP2 -driven Ty1his3AI retromobility may exert its effect in a way that depends on promoter strength and not architecture, or may reflect an architectural effect that is suppressed when Ty1his3AI is driven by a strong promoter that is not affected by the specific u3dr mutation.

The relative strength of U3, P _TEF1 , and P _PSP2 in driving retromobility in the wild-type strain was determined using the identical CEN-vector and Ty1his3AI sequence. P _PSP2 had the weakest activity, followed by U3 and then P _TEF1 (Figure 4A). Thus, if the dependence of retromobility on a host factor requires the weak strength of the U3 promoter but is independent of its architecture, that factor should also regulate P _PSP2 -Ty1his3AI. Our labs have previously shown that med15∆ abolishes retromobility of a chromosomal Ty1kanMXAI element but has only a minor effect on retromobility of a plasmid-based P _TEF1 -Ty1his3AI in the same strain (Salinero et al., 2018). In a parallel experiment, we compared the effect of deleting MED15 on the same chromosomal U3-driven Ty1kanMXAI element and the plasmid-based P _PSP2 -Ty1his3AI element transformed into the same strain (Figure 4B). In the absence of MED15, retromobility of both elements was abolished. These findings indicate that suppression of the reduced Ty1 retromobility of the med15∆ mutant seen when Ty1his3AI is driven by the P _TEF1 promoter requires the strong promoter activity of the TEF1 promoter in addition to any possible contribution of promoter architecture.

P _PSP2 -Ty1his3AI-CEN was introduced into the 51 u3dr mutants, and the retromobility patch assay was performed as it was for P _TEF1 -Ty1his3AI (Figure 1B). The relative number of His⁺ Leu⁺ prototrophs formed in patches of multiple transformants of each mutant was compared to that of the congenic wild-type strain, grown on the same plate (Figure 4C; Supplementary Data Sheet 3). Mutants with an elevated or reduced number of His⁺ Leu⁺ papillae per patch such as ygl214w∆ and yel008w∆ (Figure 4C), respectively, control Ty1his3AI retromobility when transcription was driven from the PSP2 promoter. Like Med15, these host co-factors and restriction factors may depend primarily on Ty1 gRNA being driven from a weak promoter for their effect on retromobility, and thus are designated as U3 low activity-dependent regulators (U3laDRs). Nineteen U3laDR genes were identified, 12 (63%) of which encode Ty1 repressors (Figure 4D, red box, RTT genes in bold). Another 31 mutants, including yhr130c∆ (Figure 4C), had a similar level of His⁺ papillation as the wild-type strain (Figure 4D, blue box), suggesting that the architecture of U3 is more critical than its low activity for their function. The phenotype of one mutant, caf130∆, could not be determined due to a growth defect (Plate 10, Supplementary Data Sheet 3). Genes that didn’t modulate P _PSP2 -driven retromobility were classified as U3 architecture-dependent regulators (U3arDRs). Most U3arDRs (21/31) were Ty1 retromobility co-factors. Snf2, a component of the SWI/SNF complex that is required for U3 but not Ty1i promoter activity, was identified as a U3 architecture-dependent regulator, as predicted. Other U3arDR transcription-related proteins include the SAGA and TFIID component, Taf14; SAGA component, Sgf29; histone gene transcription factors, Hir2 and Hir3; and the putative transcription factor Cth1.