|Home | About | Journals | Submit | Contact Us | Français|
Alterations in minisatellite DNA repeat tracts are associated with a variety of human diseases including type 1 diabetes, progressive myoclonus epilepsy, and some types of cancer. However, in spite of their role in human health, the factors required for minisatellite alterations are not well understood. We previously identified a stationary phase specific increase in minisatellite instability caused by mutations in the high affinity zinc transporter ZRT1, using a minisatellite inserted into the ADE2 locus in Saccharomyces cerevisiae. Here, we examined ZRT1-mediated minisatellite instability in yeast strains lacking key recombination genes to determine the mechanisms by which these alterations occur. Our analysis revealed that minisatellite alterations in a Δzrt1 mutant occur by a combination of RAD52-dependent and RAD52-independent mechanisms. In this study, plasmid-based experiments demonstrate that ZRT1-mediated minisatellite alterations occur independently of chromosomal context or adenine auxotrophy, and confirmed the stationary phase timing of the events. To further examine the stationary phase specificity of ZRT1-mediated minisatellite alterations, we deleted ETR1 and POR1, genes that were previously shown to differentially affect the viability of quiescent or nonquiescent cells in stationary phase populations. These experiments revealed that minisatellite alterations in Δzrt1 mutants occur exclusively in quiescent stationary phase cells. Finally, we show that loss of ZRT1 stimulates alterations in a derivative of the human HRAS1 minisatellite. We propose that the mechanism of ZRT1-mediated minisatellite instability during quiescence is relevant to human cells, and thus, human disease.
Minisatellite DNA repeat tracts are found throughout eukaryotic genomes [1, 2]. They consist of repeat units ranging from 16 to 100bp in length. The highly polymorphic nature of minisatellites makes them ideal for use as genetic markers, but they also have important biological functions. Minisatellites have been shown to act as enhancer elements for adjacent genes [2, 3]. The human minisatellite associated with the HRAS1 oncogene has been shown to bind the rel/NF-κB transcription factor and can enhance expression of a reporter gene [4, 5]. Minisatellites also frequently act as fragile sites, possibly due to stalling of DNA polymerases at non-B DNA structures formed by the repeats (reviewed in [1, 6]). It has also been suggested that minisatellites may have a role in chromosome pairing, specifically during male meiosis .
In addition to their important biological functions, these repetitive sequence elements also have a profound influence on human health. Rare altered alleles of human minisatellites have been correlated with increased risk of type 1 diabetes, progressive myoclonus epilepsy, and various cancer subtypes [3, 7–9]. Recently minisatellite alleles have also been associated with disorders as varied as asthma , ulcerative colitis , and attention-deficit hyperactivity disorder [12, 13].
Little is known about how minisatellites alter to give rise to disease-associated minisatellite alleles. Human minisatellites have been shown to change in tract length and repeat composition during meiosis, while remaining relatively stable during mitotic cell cycles . We previously demonstrated that these phenotypes are recapitulated in the budding yeast Saccharomyces cerevisiae [14, 15]. Since the patterns of minisatellite alteration are similar in yeast and human cells, we used the more genetically tractable yeast to identify factors that control minisatellite stability during meiosis. Our work demonstrated that meiotic minisatellite alterations require the meiosis-specific endonuclease Spo11p and the DNA loop repair protein Rad1p, while others have shown the recombination protein Rad50p is also critical [14, 16]. Additional studies have demonstrated that mutation of the key replication genes RAD27, DNA2, POL3, or the PCNA complex can destabilize minisatellite tracts during mitotic cell cycles [17–20]. In addition, the loss of the helicase PIF1 can destabilize minisatellite tracts during mitotic growth .
We previously conducted a screen for mutants that destabilize a minisatellite reporter tract in the ADE2 gene . The reporter, designated ade2-min3, employed three 20 bp direct repeats plus a 5 bp linker inserted into the ADE2 gene (Fig. 1a). This insertion causes a frameshift mutation in ADE2, resulting in a red colony color. Deletion of one repeat unit or addition of two repeat units to the array restores white colony color. If such a repeat tract expansion or contraction occurs during log phase growth of the ade2-min3 colony, then a red-white sectoring phenotype will be observed.
We isolated mutants of the zinc-homeostasis genes ZRT1 and ZAP1 with a novel color segregation phenotype, called blebbing, in the ade2-min3 background (, Fig. 2). In ZRT1 or ZAP1 mutants, red colonies form, then white microcolonies (‘blebs’) develop on the surface of the colony. We confirmed by PCR that cells in the white blebs have minisatellite tracts in which the repeat number has been altered. As we observe no sectoring in these colonies, we concluded that the blebs arise from cells in which the minisatellite tract has altered after colony growth has ceased, and the cells within the colony have entered a post-mitotic, stationary phase state. In support of our interpretation of this colony morphology phenotype, time course experiments and fluctuation analysis demonstrated that blebbing is the result of minisatellite alterations that occur after cells have entered stationary phase. Our work was the first to address the stability of minisatellites during stationary phase – potentially a significant advance since most human somatic cells spend the majority of their lifespan in a quiescent state.
Preliminary genetic analysis revealed a partial decrease in the intensity of blebbing in Δzrt1 Δrad50 mutants . These data, combined with the fact that blebbing is caused by alteration in minisatellite repeat copy number, implied a role for recombination in the minisatellite alterations of ZRT1 mutants. In S. cerevisiae, the majority of mitotic recombination requires RAD52 , but some rare recombination events are RAD52-independent - these require RAD50 or RAD51. It was clear that the mechanism responsible for these events would be of significant interest given the limited data on DNA repair in quiescent cells.
In this study, we examine the requirements for stationary phase minisatellite tract alterations in the Δzrt1 zinc homeostasis mutant. We show that stationary phase minisatellite alterations occur by both RAD52-dependent and RAD52-independent mechanisms. In addition, we demonstrate that ZRT1-dependent minisatellite alterations likely have broad genomic relevance, as they occur independently of chromosomal context, adenine auxotrophy, or marker assay gene. We confirm the stationary phase timing of minisatellite alterations in the ZRT1 mutant and demonstrate that these events are limited to the quiescent population of stationary phase cells. Finally, we establish that loss of ZRT1 can destabilize a tract derived from the human HRAS1-associated minisatellite. Our findings indicate that the mechanism of ZRT1-dependent minisatellite alteration is likely to be relevant to minisatellites found in a broad range of biological contexts.
Media was made as previously described . YPD + G418 media was composed of standard YPD solid media + 200 mg/l of G418 sulfate (geneticin). Sporulation and dissection protocols were used as described .
Plasmid pKK055 was used to integrate the ade2-hras7.5 allele’s 28bp minisatellite repeats into the ADE2 gene (Fig. 1). To construct this plasmid, a PCR product containing the HRAS1 minisatellite repeats from DTK759a  and XbaI-cleavable ends was amplified using primers 35138391 and 42180832. This cassette was digested with XbaI and ligated into the XbaI site of the ADE2 fragment contained in the plasmid pEAS8 . Insertion was verified by sequencing. The insert includes 28bp of unique sequence from the human genome on the 5′ end, one 14bp HRAS1 minisatellite repeat fragment, seven 28bp HRAS1 minisatellite repeats, and 57bp of unique sequence from the human genome on the 3′end. With the 6bp duplication of the XbaI restriction site, the insert is 301bp in length. The sequence of the insert (5′ to 3′ and excluding XbaI sites) is: 5′GAAGCTGGTCACTCGGAGGCTGCTGGGGAGAAGGGGGAGTGTGGCATTCC CTGGACAGAAGGGCAAGTGTGGCGTCCCCTGGAGAGAAGGGCGAGTGTGG CGTCCCCTGGAGAGAAGGGGGAGTGTGGCGTCCCCTGGACAGAAGGGGGA GTGTGGCGTCCCCTGGAGAGAAGGGGGAGTGTGGCGTCCCCTGGAGAGAA GGGCGAGTGTGGCGTCCCCTGGACAGAAGGGGGAGTGTCCCTGCAGCCCT GGCCAGCCAGCGGCATGGCCTACTAGCTCTCCCCCAACCTCAGGG 3′.
S. cerevisiae strains used in this study (Table 1) were derived from EAS28 (MATa his7-2 trp1-289 ura3-52) from Sia et al. 2001. Those strains whose construction is not mentioned here are described in . Strain DTK1188, containing the ade2-h7.5 allele, was made by transforming DTK260  with BglII digested pKK055. Ura− derivatives were then selected on 5-fluoroorotic acid (5-FOA) plates. Integration of the ade2-h7.5 allele was verified by sequencing.
The following strains were constructed by amplifying genomic DNA as template from the Yeast Deletion Consortium Strains. The PCR products are composed of the G418 resistance gene flanked by 3′ and 5′ homology to the target sequence. DTK944 and DTK945, bearing deletions of RAD1, were constructed by amplifying a Δrad1::KAN cassette using oligos 6934207 and 6934208. Strains DTK1074 and DTK1076, bearing deletions of RAD51, and strains DTK1309 and DTK1316, bearing deletions of DNL4, were constructed in a similar manner using oligos 26876476 and 26876477 for Δrad51::KAN and oligos 39724850 and 39724851 for Δdnl4::KAN. Strains DTK1244 and DTK1360, bearing a deletion of ETR1, were constructed by amplifying an Δetr1::KAN cassette using oligos 37616572 and 37616573. DTK1245 and DTK1361, bearing a deletion of POR1, were constructed using Δpor1::KAN cassette DNA with oligos 37616574 and 37616575. Strain DTK1666, bearing a deletion of DNL4, was constructed using genomic DNA from strain DTK1316 using oligos 39724850 and 39724851 for Δdnl4::KAN. DTK1690, bearing a deletion of RAD52, was constructed using genomic DNA amplified from DTK1253 using oligos 34756508 and 34756507 for Δrad52::URA3. Strain DTK1200, ade2-h7.5 with a deletion of ZRT1, was also constructed using a PCR construct generated using oligos 12966370 and 12966371 for Δzrt1::KAN with genomic DNA from strain DTK878 as a template. Parental strains were transformed with PCR product, grown for 4 hours in liquid YPD and then plated on YPD + G418 or SD-ura solid media to select for integration events. Transformants were checked by PCR.
To construct the RAD52 null mutant DTK1191, the Δrad52::URA3 strain DNY101  was crossed to DTK904, sporulated, and dissected as previously described . A spore bearing the ade2-min3 allele, the Δzrt1::LEU2 allele, and the Δrad52::URA3 allele was isolated by color and survival on SD-leu, SD-ura, and YPD+G418 sulfate media. This isolate was backcrossed twice to DTK904 and dissected as above to generate DTK1191. The RAD59 null strain DTK1229 was constructed by crossing DTK1195, a haploid spore isolate from sporulation and dissection of the Δrad59::KAN Yeast Deletion Consortium strain, to DTK904 and sporulating and dissecting. A spore bearing the ade2-min3 allele, the Δzrt1::LEU2 allele, and the Δrad59::KAN allele was isolated by red color, survival on SD-leu, and G418 sulfate resistance. This isolate was backcrossed twice to DTK904 and dissected to generate DTK1229. DTK1230 was constructed in a similar fashion, by crossing DTK1196, a haploid spore isolate of the Δrad59::KAN Yeast Deletion Consortium strain, to DTK271 and sporulating and dissecting as above. A spore bearing ade2-min3 and Δrad59::KAN was isolated by color and G418 sulfate resistance, then backcrossed twice to DTK271. The following strains were constructed by crossing, sporulating, and dissecting: DTK1253 (DTK1191 × DTK284), DTK1254 (DTK1057 × DTK1076), DTK1268 (DTK1253 × DTK1057), DTK1269 (DTK1191 × DTK1076), DTK1289 (DTK1268 × DTK284), DTK1290 (DTK1269 × DTK284), DTK1299 (DTK1254 × DTK284), DTK1320 (DTK1191 × DTK1299), DTK1323 (DTK1191 × DTK1299), DTK1400 (DTK1188 × DTK1056), DTK1401 (DTK1188 × DTK1074), DTK1403 (DTK1200 × DTK1191), DTK1406 (DTK1200 × DTK1056), DTK1407 (DTK1400 × DTK1191), DTK1426 (DTK1400 × DTK1191), DTK1427 (DTK1426 × DTK1188), DTK1428 (DTK1401 × DTK1200), DTK1438 (DTK1401 × DTK1407), DTK1677 (DTK1316 × 1229) and DTK1678 (DTK1316 × DTK1229).
The following primers were used in this study:
Fluctuation analysis was performed as described , with the following modifications: Plasmid pEAS20 , containing the ura3-min3 construct (Fig. 1), was transformed into DTK271 and DTK904 and plated on SD-ura to select for plasmid uptake. Ten Ade+ and ten Ade− transformants for each strain were picked and used to inoculate liquid cultures. SD-trp media, rather than YPD, was used for growing liquid cultures in order to maintain selection for the plasmid, without selecting for a particular conformation of the ura3-min3 repeats. SD-trp liquid cultures of strains bearing pEAS20 were grown at 30°C. Cultures were diluted and plated on YPD and SD-trp 5-FOA plates at 0, 4, 8, and 24 hours after inoculation. The OD600 of the cultures was measured at each timepoint to monitor culture growth. Three independent trials were conducted for each strain and the average frequency of minisatellite alteration and standard error for each strain was calculated. In contrast to previous applications of this technique, we did not extend our calculations to determine rate of minisatellite alteration over generations; as described in Results, we determined that minisatellite alterations occur specifically in quiescent cells in stationary phase, thus making it impossible to calculate a rate measurement.
To quantify the amount of blebbing in our strains we counted the number of blebs on the surface of each colony after 7 days (30°C for 3 days and then room temperature for 4 days). For each strain we surveyed at least 100 colonies over 3 independent experiments, calculating the mean number of blebs per colony and the 95% confidence interval for the mean. The size of the 95% confidence intervals was examined to determine the repeatability of the assay. In order to ensure accurate comparisons, we measured the diameter of each colony and calculated the mean colony diameter and the 95% confidence interval for the mean for each strain. Mean colony diameters for all strains in this analysis ranged from 1.24–1.29mm and 95% confidence intervals for all strains overlapped, indicating that there was not a significant difference in colony size between strains. The timepoint at which we counted blebs was chosen because at that time blebs are readily detected but still small enough to preclude overlap between blebs, which could potentially result in an undercount. We note that the extremely small 95% confidence intervals observed in our measurements demonstrate that our bleb counts in this assay are highly reproducible and are therefore likely to be a true indication of the alteration frequency of the minisatellite tract in the strain.
The ade2-min3 allele allows us to monitor minisatellite stability in S. cerevisiae (Fig. 1a, ). Three minisatellite repeats inserted into the ADE2 gene disrupt the reading frame; cells carrying this allele are Ade− and form red colonies. Loss of a repeat unit from the minisatellite tract restores the correct reading frame; cells carrying these altered alleles are Ade+ and form white colonies.
Using the ade2-min3 reporter, we previously demonstrated that ZRT1 mutants exhibit a novel color segregation phenotype that we called blebbing (Fig. 2). The blebbing phenotype consists of white microcolonies growing on the surface of the red colony. Colony morphology examination, PCR analysis, and timing experiments determined that blebbing in an ade2-min3 strain results from the loss of one minisatellite repeat specifically during stationary phase . Previous studies have shown that stationary phase cultures are composed of quiescent cells that have exited the cell cycle and nonquiescent cells which may still be undergoing some cell division . These studies also demonstrated that loss of ETR1 specifically reduces the capacity of the quiescent population of stationary phase cells to reenter the cell cycle when plated on rich media. The loss of POR1 differentially affects the reproductive capacity of nonquiescent cells in a similar manner. ETR1 encodes a 2-enoyl thioester reductase, which is involved in fatty acid synthesis , while POR1 is a mitochondrial porin, necessary for maintenance of mitochondrial osmotic stability and aerobic respiration .
These distinct genetic requirements for cell cycle reentry in quiescent and nonquiescent stationary phase cells provide a means to distinguish which cell population is responsible for ZRT1-mediated minisatellite alteration. If the reproductive capacity of the cell fraction in which the minisatellite alterations are occurring is reduced, then the cells will be unable to return to growth after minisatellite tract changes restore adenine biosynthesis, preventing blebs from forming. In order to determine whether minisatellite alterations in a Δzrt1 mutant occur in quiescent or nonquiescent stationary phase cells, we deleted ETR1 or POR1 in a Δzrt1 strain background.
The parental Δzrt1 strain has 20.6 blebs/colony (Table 2). The extremely small 95% confidence interval for this strain (20.6 +/− 0.8), as well as all other strains examined, demonstrated that our bleb counts for this assay are highly repeatable. Loss of ETR1 in a Δzrt1 strain reduced blebbing to 0.09 blebs/colony (Table 2, Fig. 2). This indicates that most, if not all, minisatellite alterations in a Δzrt1 mutant occur in quiescent cells. In contrast, loss of POR1 had little effect on blebbing, with 17.9 blebs/colony in a Δzrt1 por1 strain, indicating that minisatellite alterations in a Δzrt1 mutant likely do not occur in nonquiescent cells.
We also examined the effect of ETR1 or POR1 deletion on minisatellite alteration in wild-type ZRT1 cells. A wild-type ade2-min3 strain has 3.7 blebs/colony. Loss of ETR1 completely suppressed bleb formation in this strain background, reducing it to 0.06 blebs/colony (Table 2). Loss of POR1 reduced blebbing to 1.8 blebs/colony. Therefore, while it is likely that background minisatellite alterations occur primarily in quiescent cells, some events may be occurring in nonquiescent stationary phase cells.
Minisatellite stability can be affected by cis-acting sequence elements . In addition, loss of ADE2 function can be highly mutagenic [31, 32]. To determine if local DNA context or the use of the ADE2 gene as the marker in our minisatellite stability assay influences the stability of the minisatellite tract, we used another minisatellite assay system in which the min3 minisatellite tract is incorporated into the URA3 gene on a plasmid, pEAS20 . This TRP1 CEN plasmid contains three tandem repeats, identical to those in the ade2-min3 allele, in-frame with the URA3 coding region, forming the ura3-min3 allele (Fig. 1b). Alterations in the minisatellite repeats will throw the URA3 gene out of frame, making cells carrying the altered plasmid Ura− and resistant to 5-fluoorotic acid (5-FOA).
We employed a modified fluctuation analysis assay  to determine the stability of the minisatellite tract in pEAS20. In our fluctuation assay, we calculated the frequency of 5-FOA resistant colony formation (and thus minisatellite alteration) rather than the rate over generations, as the events being measured do not occur until after cells have ceased dividing and entered stationary phase. Cells bearing pEAS20 were grown in SD-trp liquid media in order to maintain selection for the plasmid without selection for a particular minisatellite conformation. We determined the frequency of minisatellite alteration in cells at 0, 4, 8, and 24 hours after inoculation, while monitoring the growth of the cells in the cultures to determine entry into stationary phase. An increase in minisatellite alteration, as indicated by a dramatic increase in the frequency of 5-FOA-resistant colonies in the Δzrt1 strain, was not detected until the 24 hour timepoint (Table 3), which corresponded with the onset of stationary phase (data not shown). Entry into stationary phase and the increase in minisatellite alteration frequency in ura3-min3 Δzrt1 cells was consistent with prior data from ade2-min3 Δzrt1 cells , although stationary phase onset, as determined by OD600 observation of the liquid cultures, was earlier in the ura3-min3 experiments, presumably due to the use of synthetic drop-out media (required to maintain selection for the ura3-min3 pEAS20 plasmid) rather than rich media. Wild-type pEAS20 cells maintained a level of minisatellite alteration at least 86-fold lower than the corresponding Δzrt1 cells at every time point, demonstrating that loss of ZRT1 significantly increases the frequency of minisatellite alterations independent of chromosomal context (Table 3). Control experiments indicated that the low level of white colonies present at early timepoints in both the wild-type DTK271 and Δzrt1 DTK904 strains were likely derived from cells that entered stationary phase in the colony that was chosen for the initial liquid media inoculation, as their frequency could be dramatically reduced by using colonies that were grown for a much shorter period of time prior to use. Finally, there was no significant difference in the frequency of minisatellite alteration between Ade+ and Ade− Δzrt1 transformants (2.87 × 10−1 events/cell vs. 2.47 × 10−1 events/cell at 24hrs after inoculation), indicating that ZRT1-dependent minisatellite instability is also independent of adenine auxotrophy.
Blebbing in the Δzrt1 mutant is partially suppressed by loss of RAD50 , which encodes a recombination protein . In S. cerevisiae, mitotic recombination can occur by the canonical RAD52-dependent pathway, or one of two RAD52-independent pathways requiring either RAD50 or RAD51 . To determine which recombination pathways are required for minisatellite alteration in the Δzrt1 mutant, we constructed ade2-min3 Δzrt1 strains containing deletions for key recombination genes. We then quantified the average number of blebs per colony for each of our strains (Table 2).
We examined Δzrt1 strains bearing deletions of RAD50, RAD51, and RAD52, along with double and triple mutant combinations of these genes. The parental Δzrt1 strain has 20.6 blebs/colony (Table 2). Minisatellite alterations in the Δzrt1 Δrad50 double mutant are reduced over 50% to 9.0 blebs/colony. Similarly, minisatellite alterations in the Δzrt1 Δrad51 and Δzrt1 Δrad52 double mutants are reduced to 10.0 blebs/colony and 9.4 blebs/colony, respectively. Therefore, minisatellite alterations in a Δzrt1 strain are partially dependent on RAD51 and RAD52, as well as RAD50.
To distinguish between the relative contributions of the RAD52-dependent and the RAD52-independent pathways, we examined mutants with combinations of recombination gene deletions in the Δzrt1 background. The extent of blebbing in a Δzrt1 Δrad50 Δrad52 triple mutant was reduced to 3.9 blebs/colony (Table 2), lower than any of the double mutants, suggesting that ZRT1-dependent minisatellite alterations can occur by a RAD52-dependent mechanism or a RAD52-independent mechanism that requires RAD50. In contrast, a Δzrt1 Δrad51 Δrad52 strain was phenotypically identical to the Δzrt1 Δrad51 and Δzrt1 Δrad52 parents (9.8 blebs/colony vs 10.0 and 9.4 blebs/colony). This result indicates that the RAD52-independent pathway that requires RAD51 does not significantly contribute to minisatellite alterations in a Δzrt1 mutant. A Δzrt1 Δrad50 Δrad51 strain was also phenotypically identical to the corresponding parent strains (9.1 blebs/colony vs. 9.0 and 10.0 blebs/colony). These data indicate that RAD51 does not contribute to ZRT1-dependent minisatellite alterations as long as wild-type RAD50 or RAD52 is present.
In order to explain the residual minisatellite alterations in the Δzrt1 Δrad50 Δrad52 triple mutant, we examined a Δzrt1 Δrad50 Δrad51 Δrad52 quadruple mutant. Blebbing in this strain was reduced to 0.5 blebs per colony (Table 2). Thus, any remaining events in a Δzrt1 Δrad50 Δrad52 mutant are dependent on a RAD52-independent mechanism that requires RAD51, although this pathway does not contribute to minisatellite alterations in the presence of wild-type RAD50 or RAD52.
A significant loss in cell viability could be one possible explanation for the observed decrease in blebbing in the recombination mutants, rather than loss of recombination activity. To address this possibility we examined cell viability, by comparing cell number via hemocytometer counts with viable colony forming units, in colonies that were beginning to exhibit blebbing. We observed no significant difference in cell viability in wild type (DTK271), Δzrt1 (DTK904), Δrad50 Δrad51 Δrad52 (DTK1320), and Δzrt1 Δrad50 Δrad51 Δrad52 (DTK1323) strains – all exhibited viability between 25% and 32%. These data also indicate that it is unlikely that the loss of ZRT1 is causing a mutator phenotype, as significantly increased mutation in these haploid strains would likely lead to increased cell death.
Single-strand annealing (SSA) is a RAD52-dependent recombination pathway that can result in repeat deletions like those seen in ZRT1-dependent minisatellite alteration . SSA is partially dependent on RAD59 and RAD1 [35, 36]. When RAD59 or RAD1 is deleted in a Δzrt1 background, blebbing is reduced to 16.0 blebs/colony and 17.2 blebs/colony, respectively. While this is a smaller reduction than is seen in the Δzrt1 Δrad52 double mutant (9.4 blebs/colony, Table 2), it indicates that at least some of the RAD52-dependent minisatellite alterations in the Δzrt1 mutant may occur by SSA.
As RAD50 is also involved in non-homologous end joining, we wanted to determine if non-homologous end-joining (NHEJ) has a role in ZRT1-dependent minisatellite alterations. We introduced a deletion of DNL4, a critical NHEJ ligase , into the Δzrt1 background. The Δzrt1 Δdnl4 double mutant had an average of 18.4 blebs/colony, which was not substantially different from the Δzrt1 single mutant (Table 2). A Δzrt1 Δdnl4 Δrad59 triple mutant exhibited a similar blebbing frequency. However, a Δzrt1 Δdnl4 Δrad52 mutant had a significantly reduced frequency (2.4 blebs/colony), which is well below the Δzrt1 Δrad52 mutant (9.4 blebs/colony). Thus, while NHEJ does not play a major role in the minisatellite alterations occurring in Δzrt1 strains, nor does it affect events occurring by SSA, it does influence some of the events remaining after the RAD52-dependent pathway has been eliminated.
We also determined if recombination plays a role in the infrequent minisatellite alterations occurring in a strain with a wild-type ZRT1 gene. The parental ade2-min3 strain has 3.7 blebs per colony (Table 2). Loss of RAD50, RAD51, RAD52 or DNL4 in a wild-type background reduced minisatellite alterations to 0.9, 2.1, 1.4, and 1.8 blebs/colony, respectively. These results indicate that the small number of minisatellite tract alterations seen in the wild-type ade2-min3 strain involve DNA double-strand break repair, but require both homologous recombination and non-homologous end joining (NHEJ).
Both the ade2-min3 and the ura3-min3 minisatellite constructs utilize three identical tandem repeats with roughly 50% GC content (Fig. 1a, b). However, human minisatellites are often composed of variable repeats that are enriched for CG content , and therefore may be altered by different mechanisms than our reporter minisatellite constructs. To determine whether the stability of a human minisatellite is affected by loss of ZRT1, we constructed the ade2-h7.5 allele, which is composed of one partial and seven complete 28bp repeats of the human HRAS1 minisatellite plus HRAS1 locus flanking DNA inserted into the ADE2 gene (Fig. 1c). This minisatellite has a CG content of 68%, and variable nucleotides in each repeat at the 14th and 22nd positions.
The insertion of the HRAS1 minisatellite-derived DNA throws the ADE2 gene out of frame, making the cells carrying this allele red and Ade−. Alterations in the minisatellite tract can restore ADE2 to the correct reading frame, making cells in which alterations have taken place white and Ade+. Thus, colony color segregation serves as a reporter for minisatellite alteration in strains with the ade2-h7.5 allele, as with the ade2-min3 allele. The ade2-h7.5 minisatellite allele is very stable, with a blebbing frequency of 0.7 blebs/colony in a WT strain background (Table 2).
To examine the effect of ZRT1 on the stability of the ade2-h7.5 minisatellite repeats, we deleted ZRT1 in the ade2-h7.5 background. The ade2-h7.5 Δzrt1 strain displayed a blebbing phenotype similar to the ade2-min3 Δzrt1 strain, but the frequency of minisatellite alterations in the ade2-h7.5 Δzrt1 strain (4.8 blebs/colony) was lower than the ade2-min3 Δzrt1 strain (20.6 blebs/colony) (Table 2). However, the seven-fold increase seen in the Δzrt1 ade2-h7.5 strain relative to the wild-type parental ade2-h7.5 strain is comparable to the six-fold increase seen in ade2-min3 strains (Table 2). Thus, the ZRT1-dependent pathway of minisatellite alteration in yeast may also be relevant to minisatellite stability in human cells. The lower frequency of minisatellite alterations in the ade2-h7.5 Δzrt1 strain may be due to the fact that the HRAS1 minisatellite repeats are variable at two nucleotides in their sequence (Fig. 1c), which could influence homologous recombination events occurring between repeats. Prior work in our lab demonstrated that single nucleotide heterozygosities in HRAS1 minisatellite repeats act to stabilize the DNA tracts .
We examined the nature of the tract alterations that led to bleb formation in the Δzrt1 ade2-h7.5 strain. Whole-cell PCR was used to examine the ade2-h7.5 allele in 104 independent white blebs. While almost all alleles (99/104 – 95%) were contractions of the repetitive tract, we did observe tract expansions in 5% (5/104) of the blebs examined. All expansion alleles gained two repeat units. Of the 99 tracts exhibiting loss, 91 had lost a single repeat while 8 lost four repeats.
We sequenced the tracts in 27 of the 104 ade2-h7.5 alleles examined by PCR. As shown in Figure 3, both expansions sequenced added repeat units to the middle of the tract. The majority of single repeat losses occurred at the fourth or fifth repeat. The first allele with a four-repeat contraction lost the second through fifth repeats, while the second four-repeat contraction allele lost DNA from the middle of the first repeat (between the variable 14th and 22nd nucleotides) to the middle of the fifth repeat (between the variable nucleotides) to generate a hybrid Type 3 repeat.
Finally, we asked if the ZRT1-dependent alterations in the ade2-h7.5 minisatellite, like alterations in the ade2-min3 minisatellite, occur via RAD52-dependent and – independent pathways. As with ade2-min3 strains, we deleted key recombination factors in the ade2-h7.5 parent and ade2-h7.5 Δzrt1 mutant strains. The ade2-h7.5 Δzrt1 mutant displays an average of 4.8 blebs/colony. Loss of RAD50, RAD51, or RAD52 in the ade2-h7.5 Δzrt1 background reduced blebbing to 1.1 blebs/colony, 1.8 blebs/colony, and 2.5 blebs/colony, respectively (Table 2). An ade2-h7.5 Δzrt1 Δrad50 Δrad52 strain showed a further reduction in blebbing – to 0.9 blebs/colony. Finally, loss of RAD50, RAD51, and RAD52 reduced blebbing in the ade2-h7.5 Δzrt1 background to 0.3 blebs/colony, a level not significantly higher than the ZRT1 wild-type strain. This indicates that HRAS1 minisatellite alterations in the Δzrt1 mutant likely occur via a combination of RAD52-dependent and RAD52-independent mechanisms, as is the case in for ade2-min3 minisatellite alterations.
In this work we examined minisatellite alterations during stationary phase in S. cerevisiae. First, we demonstrated that ZRT1-dependent stationary phase minisatellite alterations occur specifically in quiescent cells, and not in nonquiescent cells within the stationary phase population. We showed that minisatellite alterations in Δzrt1 mutants are not dependent on chromosomal context or adenine auxotrophy. We determined that recombination factors are required for alterations in minisatellite tracts during quiescence. The majority of minisatellite alterations in a Δzrt1 strain occurs by RAD52-dependent homologous recombination or a RAD52-independent pathway that requires RAD50. Finally, we find that mutation of ZRT1 can destabilize a human minisatellite, implying that zinc homeostasis may play a role in minisatellite stability in human cells.
Stationary phase cultures of S. cerevisiae consist of a quiescent cell fraction, in which cells are uniformly arrested and bulk DNA synthesis does not occur, and a nonquiescent cell fraction, in which some budded cells are present and DNA replication may occur . Therefore, we considered that ZRT1-dependent minisatellite alterations during stationary phase might occur as a result of polymerase slippage during whole-genome DNA synthesis in nonquiescent cells. However, we found that loss of ETR1, which specifically reduces the ability of quiescent cells to reenter the cell cycle, completely eliminated blebbing in a Δzrt1 mutant. In contrast, loss of POR1, which specifically reduces the reproductive capacity of nonquiescent cells, has little effect on blebbing in a Δzrt1 mutant. Therefore, our data argue that ZRT1-dependent minisatellite alterations occur as a result of events in quiescent cells that do not require bulk DNA synthesis. Recent work  demonstrates limited DNA synthesis at specific locations in the genome of stationary phase yeast cells; the events that lead to minisatellite tract alterations may utilize this type of limited repair synthesis.
Using the chromosomal ade2-min3 allele (Fig. 1), we determined that loss of ZRT1 triggers an increase in minisatellite alterations exclusively in quiescent cells. We utilized the ura3-min3 allele to show that minisatellite alterations in a Δzrt1 mutant strain occur independently of chromosomal context and adenine auxotrophy. In stationary phase Δzrt1 cells, we see an increase in plasmid-borne ura3-min3 minisatellite alterations, eliminating the possibility that the Δzrt1 minisatellite instability phenotype is an artifact of cis-acting sequences surrounding ADE2 and implying that chromosomal context is not likely to have an influence on these events. Alterations occurred with equal frequency in Ade+ or Ade− strains, indicating that the alterations are not due to the disruption of the adenine biosynthetic pathway. Taken together, these results argue that loss of ZRT1 could potentially destabilize minisatellites at many loci and in many genetic backgrounds.
Our data demonstrate that most minisatellite alterations in a Δzrt1 mutant require homologous recombination. Loss of RAD50, RAD51, or RAD52 in a Δzrt1 mutant strain reduces the minisatellite alterations by approximately 50% compared to the parental Δzrt1 single mutant (Table 2). Similarly, loss of both RAD50 and RAD51 or both RAD51 and RAD52 in a Δzrt1 mutant also reduce minisatellite alterations by ~50%. However, loss of both RAD50 and RAD52 in a Δzrt1 strain reduces minisatellite alterations by ~81%, indicating that both RAD52-dependent recombination and a RAD50-dependent, RAD52-independent mechanism are required for minisatellite alterations in the Δzrt1 mutant. Finally, loss of RAD50, RAD51, and RAD52 in a Δzrt1 mutant reduces minisatellite alterations to a level not significantly different from the parental Δrad50 Δrad51 Δrad52 strain, indicating RAD51 plays a role in ZRT1-dependent minisatellite alterations in the absence of RAD50 and RAD52. Finally, deletion of the non-homologous end-joining (NHEJ) ligase DNL4 in a Δzrt1 mutant strain does not reduce the frequency of minisatellite alterations, but a significant reduction is seen when RAD52 is also deleted, indicating that NHEJ can play a role in stability maintenance if the homologous recombination pathway has been compromised. These results are consistent with a model in which two homologous recombination pathways, one RAD52-dependent and the other RAD52-independent and requiring RAD50, are required for most alterations occurring in the Δzrt1 mutant cells, with other pathways becoming active when homologous recombination has been compromised.
While recombinational mechanisms have previously been implicated in minisatellite instability during yeast mitosis [1, 22], minisatellite alterations in Δzrt1 mutants occur during stationary phase, a stage at which most yeast cells are quiescent . Recombination modulation previously has been linked to genome alterations during stationary phase in prokaryotic E. coli cells (see, for example [40–42]); our data extend these findings to eukaryotic yeast cells.
Stationary phase in our Δzrt1 strain consists of a relatively uniform G1 arrest; visual inspection of the ade2-min3 Δzrt1 strain shows that by the time the culture enters stationary phase, which occurs at ~48hrs after inoculation , approximately 98% of the cells are unbudded (G1) (data not shown). G1 cells have only one copy of each chromosome while G2 cells, which have undergone DNA replication, have two sister chromatids. Thus, minisatellite alterations in a Δzrt1 mutant are not likely to be a result of recombination between sister chromatids. This is significant because all the strains we have used to examine minisatellite stability in this study are haploid and therefore cannot undergo recombination between homologous chromosomes.
What is the molecular mechanism underlying stationary phase minisatellite tract alteration? Since we have shown that ZRT1-dependent minisatellite alterations occur in quiescent, G1-arrested stationary phase haploid cells, the mechanism for these events must not rely on bulk DNA synthesis or involve exchange between sister chromatids or homologous chromosomes. Therefore, we propose that minisatellite alterations in Δzrt1 mutants during stationary phase occur by two mechanisms: single strand annealing (SSA) and intramolecular repair events. Simple misalignment of repeats during SSA could easily result in minisatellite repeat deletion, and the role of SSA in repeat array contraction has been well established . An intramolecular repair event could be initiated by a single-strand gap (Fig. 4). During the limited DNA synthesis needed to repair the gap, polymerase slippage and misalignment could result in repeat units forming a single-stranded loop. If the loop forms on the template DNA strand and is removed, a deletion of repeat units will result. If the loop forms on the newly synthesized DNA strand and is repaired by nicking the template DNA strand opposite the loop, an expansion of the repeat tract will result (Fig. 4). The frequency of this type of event is dependent on the frequency of polymerase slippage and the formation of single-stranded nicks or gaps. It is quite possible that the polymerase slippage frequency is significantly increased at repetitive tracts in stationary phase cells. Nucleotide reserves in quiescent cells are likely to be small relative to the nucleotide pools available in actively-dividing cells, leading to increased polymerase pausing (especially at repetitive DNA tracts) thereby increasing the likelihood of polymerase dissociation and re-association events. Finally, defects in zinc transport have been shown to elevate single-stranded DNA damage in mammalian cells [43, 44].
Our data on the types of minisatellite alterations observed in Δzrt1 mutants are consistent with these models. We previously showed that minisatellite alterations in Δzrt1 mutants consist of deletions of repeat units from the ade2-min3 tract . In this study we observed both repeat deletions and tract expansion in the ade2-h7.5 minisatellite (Fig. 3). Intramolecular repair events and SSA can both lead to deletions in direct repeat tracts when repeats are misaligned . We find that minisatellite alterations in a Δzrt1 mutant during stationary phase occur by both RAD52-dependent and RAD52-independent mechanisms. SSA is a RAD52-dependent process for shorter repeat arrays [45, 46]. RAD52-independent mechanisms of SSA have been demonstrated only for large CUP1 and rDNA repeat tracts . Since our ade2-min3 and ade2-h7.5 reporters are significantly smaller than those tracts both in terms of repeat unit length and number of repeats, it is unlikely that RAD52-independent SSA contributes to our events. RAD59 and RAD1 are involved in SSA [36, 48, 49]. Deletion of either RAD1 or RAD59 in a Δzrt1 strain leads to a ~20% reduction in minisatellite alterations (Table 2, data not shown), clearly implicating SSA in these events. Since some SSA has been demonstrated to occur even without RAD59 , it is difficult to estimate what proportion of ZRT1-dependent minisatellite alterations result from SSA. It may be that all RAD52-dependent minisatellite alterations in the Δzrt1 mutant occur by SSA. However, as we have argued above, RAD52-independent SSA is not likely to contribute to minisatellite alterations in the ade2-min3 minisatellite tract. Therefore, all RAD52-independent minisatellite alterations in the ZRT1 mutant may occur via some form of intramolecular repair. Consistent with this, we find that most RAD52-independent minisatellite alterations require RAD50; prior studies demonstrated a requirement for RAD50 in RAD52-independent intramolecular recombination . Rad50p possesses zinc hooks that are employed in linking DNA ends [50, 51], which could play a role in forming the single stranded DNA loops in our model of intramolecular minisatellite repair. Previous work has demonstrated that single stranded DNA loops over 16 nt can be repaired by loop removal, but the genetic requirements for this process remain elusive, likely because redundant pathways facilitate it . Therefore, it is difficult to provide evidence for this aspect of our model. Finally, our DNL4 data show that NHEJ can be used as a repair mechanism when homologous recombination has been compromised.
Importantly, we demonstrate here that a human HRAS1 minisatellite tract alters in quiescent cells, and that loss of ZRT1 further destabilizes the minisatellite (Table 2). The yeast ZRT1 protein is a member of the ZIP (Zrt-, Irt-like Protein) family of zinc transporters, which are found throughout bacteria and eukaryotes, including humans . Mammals have three orthologs of the yeast ZRT1 gene: ZIP1 (also known as ZIRTL), ZIP2 and ZIP3 [54–59]. The ZIP1 protein in humans (hZip1) is likely to be the major zinc transporter for most of the cells in the body, as it is expressed in most cell types . The human Zip2 protein is expressed in prostate and uterine epithelial cells. Mutations in ZIP family zinc transporters lead to zinc deficiency and associated health problems in humans . For example, mutation of Zip2 or Zip3 has been linked to prostate cancers .
Our data provide a mechanism to link zinc deficiencies with human disease generation. DNA strand breaks, triggered by zinc deficiency, could be the initiating lesions for minisatellite alterations in post-mitotic cells. In agreement with this model, zinc deficiency has been linked to increased DNA strand breakage in mammals [43, 44]. Variations in the HRAS1 minisatellite tract are known to alter HRAS1 transcription [4, 61, 62], modifying the activity of this important oncogene and influencing the onset of particular cancers. These alteration events may be elevated in quiescent, post-mitotic cells, as our data demonstrate that they are particularly sensitive to loss of zinc transporters.
Links that support this hypothesis have been detected between zinc homeostasis and minisatellite stability in human disease. Prostate cancer has been linked to both zinc deficiency and rare HRAS1 minisatellite alleles [8, 43]. Type 1 diabetes has long been associated with rare alleles of the IDDM1 minisatellite, and has recently been correlated with SNPs in the ZIP family zinc transporter SLC30A8 [3, 63]. Thus, the proposal that zinc homeostasis may directly affect minisatellite stability and influence minisatellite-correlated disease in humans clearly merits further investigation.
We previously discovered that a reporter minisatellite DNA tract is destabilized in stationary phase yeast cells with defects in zinc homeostasis . Here we demonstrate that minisatellites are destabilized specifically in quiescent cells. Tract alterations occur regardless of minisatellite context, and can also take place in a human minisatellite inserted into the yeast genome. We have determined that both RAD52-dependent and RAD52-independent recombination pathways are necessary to alter minisatellite tracts in quiescent Δzrt1 cells. Our findings indicate that analogous zinc-mediated minisatellite rearrangements may be occurring in human cells, leading to minisatellite-associated pathologies such as cancer.
We thank Laura Brosnan for technical assistance in constructing the ade2-h7.5 allele. This work was sponsored by a grant from the National Institutes of Health (5RO1-GM072598) and ARRA supplement 3R01-GM072598-05S1.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.