The Loz1 transcription factor from Schizosaccharomyces pombe binds to Loz1 Response Elements and represses gene expression when zinc is in excess

Summary

In Schizosaccharomyces pombe the expression of the zrt1 zinc uptake gene is tightly regulated by zinc status. When intracellular zinc levels are low, zrt1 is highly expressed. However, when zinc levels are high, transcription of zrt1 is blocked in a manner that is dependent upon the transcription factor Loz1. To gain additional insight into the mechanism by which Loz1 inhibits gene expression in high zinc we used RNA-seq to identify Loz1-regulated genes, and ChIP-seq to analyze the recruitment of Loz1 to target gene promoters. We find that Loz1 is recruited to the promoters of 27 genes that are also repressed in high zinc in a Loz1-dependent manner. We also find that the recruitment of Loz1 to the majority of target gene promoters is dependent upon zinc and the motif 5’-CGN(A/C)GATCNTY-3’, which we have named the Loz1 Response Element (LRE). Using reporter assays we show that LREs are both required and sufficient for Loz1-mediated gene repression, and that the level of gene repression is dependent upon the number and sequence of LREs. Our results elucidate the Loz1 regulon in fission yeast and provide new insight into how eukaryotic cells are able to respond to changes in zinc availability in the environment.

Abbreviated Summary

In fission yeast the transcription factor Loz1 plays a central role in zinc homeostasis by inhibiting gene expression when zinc is in excess. Here we show that Loz1 DNA binding activity is dependent upon cellular zinc status and that Loz1 binds in a site-specific manner to Loz1 Response Elements (LREs) that are located in target genes promoters. These studies provide new mechanistic insight into Loz1 function, and highlight new pathways that are regulated by zinc.

Introduction

A diverse array of proteins require zinc for their function, including transcription factors containing C₂H₂-type zinc finger domains, and many enzymes that are required for basic cellular processes including: DNA replication, transcription, translation, and respiration (Andreini et al., 2009, Coleman, 1992). As zinc ions can also be toxic in excess, cells require homeostasis mechanisms to maintain sufficient levels of zinc for incorporation into newly synthesized zinc proteins and at the same time do not inhibit growth (Bird, 2015). The importance of these zinc homeostasis mechanisms is underscored by the fact that multiple genetic diseases can result from zinc deficiency or imbalances in zinc ion levels including: Acrodermatitis enteropathica; transient neonatal zinc deficiency; early onset agammaglobulinemia; and the spondylocheirodysplastic form of Ehlers-Danlos syndrome (Wang et al., 2004, Jeong et al., 2012, Chowanadisai et al., 2006, Anzilotti et al., 2019).

The budding yeast Saccharomyces cerevisiae has been an important model system for deciphering mechanisms of cellular zinc homeostasis. In this yeast the transcription factor Zap1 plays a key role in the zinc starvation response by controlling the expression of genes required for zinc acquisition (ZRT1, ZRT2, and FET4) and the release of zinc from intracellular stores (ZRT3) (Eide, 2009, Wilson & Bird, 2016). When zinc levels are low, Zap1 binds to zinc responsive elements (ZREs) that are located within the promoters of these genes, increasing gene expression. In contrast, when zinc levels are high, Zap1 loses its ability to bind to DNA and to activate gene expression. Zap1 also controls how cells allocate zinc by regulating the expression of ZRR1 and ZRR2 - the non-coding RNAs that inhibit the expression of genes encoding alcohol dehydrogenase 1 and 3, respectively (Bird et al., 2006). As alcohol dehydrogenases are abundant zinc-requiring proteins, the inhibition of ADH1 and ADH3 gene expression in low zinc helps cells to conserve zinc ions for more important functions.

In addition to having a primary role in zinc homeostasis, Zap1-dependent changes in gene expression alter cell metabolism so that they are able to survive longer periods of zinc starvation. Under these more severe conditions Zap1 induces the expression of additional genes that are required for stress resistance, cell wall function, and secretory pathway function (Wu et al., 2008). Examples of these second-response genes include TSA1, which encodes a peroxiredoxin that has a dual role as a peroxidase and a holdase, and ZRG17, which encodes a CDF family member that is required for zinc transport from the cytoplasm into the endoplasmic reticulum (Wu et al., 2007, Wu et al., 2011). Further analyses of these genes and their protein products have shown that the holdase function of Tsa1 is critical for survival during severe zinc starvation, and that in the absence of the Zap1-dependent increase in ZRG17 expression, cells experience increased protein unfolding in the endoplasmic reticulum (MacDiarmid et al., 2013, Choi et al., 2018). Studies of these Zap1 target genes therefore suggest that cells have programmed responses to counter stresses that arise from zinc starvation.

In the fission yeast Schizosaccharomyces pombe multiple zinc homeostasis genes are regulated at a transcriptional level in response to zinc status, including zrt1 and adh1 (Dainty et al., 2008, Ehrensberger et al., 2013). However, in this yeast this transcriptional control is dependent upon a different transcription factor called Loz1. In contrast to Zap1, which has a primary role in activating gene expression in low zinc, deletion of the loz1 gene leads to the constitutive expression of zrt1 and a regulatory adh1 antisense transcript (adh1AS), suggesting that the primary function of Loz1 is to repress gene expression when zinc is in excess (Corkins et al., 2013). Studies with Loz1 have also revealed different genes and pathways that are regulated by zinc availability. One novel Loz1 target gene is gcd1, which encodes glucose dehydrogenase 1 (Corkins et al., 2017). In fission yeast this enzyme is part of the gluconate shunt which is a secondary route for directing glucose into the pentose phosphate pathway (Tsai et al., 1995, Corkins et al., 2017). As the pentose phosphate pathway plays a primary role in the regeneration of NADPH and produces ribose-5 phosphate sugars which are precursors of many other metabolic intermediates, the regulation of gcd1 by Loz1 suggests that increased levels of one or more of these intermediates may be important for growth or survival in low zinc conditions.

The goals of this study were to further our knowledge of zinc homeostasis by identifying additional genes that are regulated by Loz1 and to test whether Loz1 directly binds to its target gene promoters to repress gene expression. By using ChIP-seq we show that Loz1 binds in a zinc-dependent manner to the sequence CGN(A/C)GATCNTY when zinc is in excess, highlighting a mechanism by which Loz1 is able to specifically repress target gene expression in high zinc. We also find that in high zinc Loz1 represses the expression of 27 genes that encode proteins and ncRNAs required for zinc homeostasis, as well as carbohydrate metabolism, the transport of small molecules, and non-covalent protein folding.

Results

Loz1 binds to target genes promoters in high zinc

In our previous studies we found that Loz1 negatively auto-regulates its own expression which leads to very low levels of the active Loz1 repressor accumulating in high zinc conditions (Corkins et al., 2013). When we attempted to perform ChIP analysis using a functional Loz1GFP fusion protein that was expressed from the loz1 promoter, we were unable to immunoprecipitate Loz1GFP from cells grown in high zinc, potentially due to the low levels of Loz1GFP protein that accumulate under this condition. To overcome this limitation, we generated a new construct to express Loz1GFP from a weaker derivative of the pgk1 promoter that contained a deletion within the canonical TATA box (pgk1^ΔTATA). To compare the strength of the pgk1^ΔTATA promoter to the loz1 promoter (loz1^p), strains expressing Loz1GFP from these promoters were grown overnight in a zinc-limited minimal medium (ZL-EMM) (Choi et al., 2018) supplemented with 0 (-Zn) or 100 μM Zn (+Zn) and crude protein extracts prepped for immunoblot analysis. The immunoblot analysis revealed that the pgk1^ΔTATA-driven Loz1GFP accumulated in both low and high zinc conditions (Fig. 1A). When compared to cells expressing Loz1GFP from the loz1 promoter, expression from the pgk1^ΔTATA promoter resulted in ~ 2-fold and ~ 7-fold higher levels of Loz1GFP in low and high zinc conditions, respectively.

Figure 1.

Generation of a Loz1GFP expression vector for ChIP-seq analysis. A) Immunoblot analysis of crude protein extracts isolated from wild-type cells and loz1Δ cells expressing pgk1^ΔTATA-Loz1GFP or loz1^p-Loz1GFP following growth overnight in ZL-EMM (-Zn) or in ZL-EMM + 100 μM Zn²⁺ (+Zn). Blots were probed with anti-GFP and anti-Act1 (Actin), which served as a loading control. The positions of the molecular weight markers (in kDa) are indicated on the left. Immunoblots were performed in triplicate and a representative blot is shown. The numbers below the anti-GFP panel indicate the mean increase in pgk1^ΔTATA-Loz1GFP relative to the loz1^p-Loz1GFP with +/− standard deviations. N=3

(B) β-galactosidase activity was assayed in loz1Δ cells co-expressing a zrt1-lacZ reporter and the indicated plasmids following growth overnight in ZL-EMM supplemented with 0, 0.5, 1, 10, or 100 μM Zn²⁺. Data represents the mean values with error bars representing standard deviations. N=3

(C) ChIP-qPCR experiments to validate that the pgk1^ΔTATA-Loz1GFP protein could be used for ChIP analysis. loz1Δ cells expressing pgk1^ΔTATA-Loz1GFP were grown overnight in ZL-EMM with 0 (-Zn) or 100 μM Zn²⁺(+Zn), and ChIP-qPCR carried out with primers flanking the zrt1 and adh4 promoters. The fold enrichment in the immunoprecipitations relative to input DNA was calculated as described in the materials and methods. Primers specific to the psi2 promoter were used as a negative control. The numbers are the average fold enrichment with error bars showing standard deviations. N=3

To test whether expression of loz1GFP from the pgk1^ΔTATA promoter would affect Loz1-mediated gene expression, we co-expressed the Loz1GFP alleles described above with a zrt1-lacZ reporter and assayed β-galactosidase activity over a range of zinc levels. In cells expressing pgk1^ΔTATA-Loz1GFP, growth in 100 μM zinc resulted in an ~600-fold reduction in zrt1-lacZ reporter activity, indicating that the activity of this fusion protein was regulated by zinc (Fig. 1B). However, at 0 μM zinc zrt1-lacZ reporter activity was 3-fold lower than that observed in loz1Δ expressing loz1^p-Loz1GFP, suggesting that over-expression of Loz1 also leads to higher levels of gene repression. When the association of pgk1^ΔTATA-Loz1GFP with the zrt1 and adh4 promoters was examined using ChIP analysis, there was a large enrichment of Loz1GFP at both of these promoters in high zinc conditions (Fig. 1C). A similar enrichment was not observed at a promoter of a control gene (pci2) that is not regulated by zinc (Dainty et al., 2008, Ehrensberger et al., 2013). Although over-expression of Loz1 from the pgk1^ΔTATA promoter leads to higher levels of gene repression, these preliminary ChIP analyses suggested that the pgk1^ΔTATA-driven Loz1GFP could be used to examine zinc-dependent changes in Loz1 DNA binding activity.

To determine if the recruitment of Loz1 to other promoter regions was dependent upon zinc, loz1Δ cells expressing pgk1^ΔTATA-Loz1GFP were grown in ZL-EMM supplemented with 0 or 100 μM zinc, and purified DNA associated with Loz1GFP further analyzed by ChIP-seq (see Experimental procedures). 136 peaks were above peak-ID thresholds in at least three independent repeats. 98 of these peaks were located upstream of protein-coding genes or non-coding RNAs (Fig. 2A), whereas the remaining 38 peaks were detected within the coding regions of tRNAs or other small RNAs. Of the 98 peaks that were located within promoter regions, 60 were detected during growth in high zinc, 34 were detected in both high and low zinc conditions, and 4 were detected in low zinc only (Table S1 and S2). The most abundant peaks were detected in the high zinc conditions and included peaks mapping to the promoter regions of the known Loz1 target genes adh4, zrt1, and the adh1AS transcript (Fig. 2B). We also determined which peaks displayed a differential binding pattern in response to cellular zinc status. For these analyses we defined differential binding as a Loz1 peak that displayed a 2-fold increase in one condition as compared to the other condition. 63 Loz1 binding peaks were identified that were enriched in cells grown in high zinc as compared to low zinc, including adh4, zrt1, and adh1AS (Fig. 2C and Table S3). Only a single Loz1-binding peak was identified that was enriched in low zinc as compared to high zinc. This peak mapped to the promoter region of ght8, which encodes a putative hexose transporter.

Figure 2.

Identification of Loz1 binding sites using ChIP-seq (A) ChIP-seq was performed as described in the materials and methods. The input-normalized fold enrichment profiles for peaks enriched in high zinc (orange dots) and low zinc (blue dots) were plotted against chromosome position. Positions of peaks in the promoters of characterized Loz1 target genes adh4, zrt1, and adh1AS are indicated.

(B) ChIP-seq data was visualized with the Integrative Genomics Viewer Software. Snapshots of the peaks identified in the adh4, zrt1, and adh1AS promoters in low and high zinc are shown.

(C) Volcano plot of the confidence scores of Loz1 binding sites plotted against the zinc-dependent fold-change in normalized read counts. Loz1 binding sites where there is a greater than 2-fold increase in binding under one condition (low zinc or high zinc) as compared to the other, have been colored pink.

As the pgk1^ΔTATA-driven Loz1GFP protein accumulates at higher levels inside of cells relative to the endogenous Loz1 protein, it was possible that some of the ChIP peaks identified were false positives arising from over-expression. For example, many of the Loz1-binding peaks were found in the promoters of highly expressed genes (e.g. pgk1), which are commonly found as false positives in ChIP-seq studies (Park et al., 2013). To determine which of the genes identified by ChIP-seq were bona fide Loz1 targets, we used RNA-seq to identify transcripts that accumulated in wild-type and loz1Δ cells following growth overnight in high zinc. As Loz1 is required for transcriptional repression when zinc is in excess, we predicted that Loz1 target genes would be: 1) expressed at a higher level in zinc-replete loz1Δ cells when compared to wild-type cells; 2) would be identified as a target gene in the ChIP-seq analysis; and 3) would contain a binding site for Loz1 in their promoter.

In the RNA-seq analysis 126 unique mRNAs accumulated significantly by ≥ 2-fold in zinc-replete loz1Δ cells. Of these, 24 were also identified as ChIP-seq targets, suggesting that these were the genes that were repressed by Loz1 in high zinc under steady state conditions (Fig. 3A). In order to validate these results, a number of transcripts were also analyzed by RNA blot analysis. In all cases deletion of loz1 led to higher levels of the transcript accumulating in zinc-replete cells, consistent with these genes being regulated by Loz1 (Fig. 3B and 3C). There were also notable differences in the transcript profiles, which could potentially reflect variation in the recruitment of Loz1 to each promoter, or combinatorial effects of other regulatory proteins. As one example, higher levels of dak2 transcripts were detected in loz1Δ cells relative to zinc-limited wild-type cells, suggesting that deletion of loz1 results in increased levels of dak2 transcripts by an alternative mechanism. In a few cases, known Loz1 targets were identified as strong ChIP targets and their fold induction in the RNA-seq analysis did not reach statistical significance (adh1AS), or they were identified as strong RNA-seq targets and their ChIP-seq peaks fell just below the threshold peak ID (loz1 and SPBC1348.06c). As the expression of these genes has previously been shown to be dependent upon Loz1 (Corkins et al., 2013), they were also included as Loz1 target genes in further analyses.

Figure 3.

Identification of genes that are repressed by Loz1 in high zinc conditions A) Venn diagram representing the overlap of genes expressed at a higher level in loz1Δ cells grown in high zinc, and protein coding genes that contain a Loz1 ChIP-seq peak within 1500 bp of the transcriptional start site following growth in zinc-replete conditions. When ChIP-seq peaks were located in the promoters of two divergently transcribed genes (49 of 98 peaks), both genes were included.

B) To validate results from the RNA-seq analysis the indicated strains were grown overnight in ZL-EMM supplemented with 0, 1,10,100 μM Zn. Total RNA was prepared from the indicated strains and subjected to RNA blot hybridization with the indicated probes. A representative blot is shown on the left. Ribosomal RNAs were used as the loading control, and zrt1 representing a previously characterized Loz1 target gene. Due to the differences in dak2 transcript abundance in wild-type and loz1Δ cells, short and long exposures are shown.

C) Quantification of RNA blot analysis. Hybridization signals obtained with the indicated P32 labelled probes were detected by phosphorimaging and were quantitated using Image J. The values were normalized to the 18 S rRNA and represent the average value of 3–4 independent replicates. Error bars represent +/− standard deviations and p values were determined using a student’s t-test. ***p<0.001, **p<0.01, *p<0.05, ns not significant.

To determine if the genes identified in the RNA-seq and/or ChIP-seq were under the control of any common regulatory elements, we next searched the promoter regions and ChIP-seq peaks for regulatory DNA motifs using the de novo computational motif discovery program Multiple Em for Motif Elicitation (MEME) (Bailey et al., 2009). When the search was restricted to ChIP-seq peaks of genes repressed in high zinc in a Loz1-dependent manner, a consensus of CGN(A/C)GATCNTY was present in 16 peaks (Fig. 4A), the core of which resembled the GNNGATC sequence that was previously suggested to be important for Loz1-mediated gene repression (Corkins et al., 2013). Related searches with other programs including HOMER, CONSENSUS, and Weeder revealed shorter motifs that also had a similar core CGNNGATC consensus (Fig. S1). No common consensus was identified when all ChIP-seq peaks were included in the analyses, or in ChIP-seq peaks that were not RNA-seq targets. From here on, the CGN(A/C)GATCNTY element was named the Loz1 Response Element (LRE).

Figure 4.

The Loz1 response element is required for Loz1-mediated gene repression. A) The program MEME was used to identify putative cis-acting DNA elements that were located in the promoters of Loz1 target genes. The shown motif (named the Loz1 response element) was enriched in ChIP-seq peaks of genes that were repressed by Loz1 in high zinc.

B) A schematic diagram of the SPBC1348.06c-lacZ reporter highlighting the positions and sequences of putative LREs. The numbers indicate the distance of the first base of the motif to the putative ATG initiation codon. Nucleotide substitutions that were introduced to disrupt LRE function are shown in bold.

C) Wild-type or loz1Δ cells expressing the indicated SPBC1348.06c-lacZ reporter genes were grown overnight in ZL-EMM supplemented with 0 or 100 μM zinc before cells were harvested and assayed for β-galactosidase activity. The relative positions of the putative LREs in the reporters are indicated by green boxes. The positions of mutated LREs are shown by green boxes with red crosses. Data represents the mean values with error bars representing standard deviations. N=3

The LRE is necessary for Loz1-mediated gene repression

To validate that the LRE was required for Loz1-mediated gene repression, we used site directed mutagenesis to introduce nucleotide substitutions in putative elements in a SPBC1348.06c-lacZ reporter. We used this reporter as SPBC1348.06c was a highly regulated gene in the RNA-seq analysis, and we had previously found that the activity of the SPBC1348.06c-lacZ reporter was tightly regulated by Loz1 and zinc (Corkins et al., 2013). The MEME analysis indicated that the SPBC1348.06c promoter contained a single LRE (5’-CGACGATCATG-3’) at nucleotide positions −147 to −137 relative to the translational ATG (Fig. 4B). In addition to this element, we identified two closely related sequences at positions −380 to −370 (5’-GGAAGATCTAC-3’) and −130 to −140 on the reverse strand (5’-CGTAGATCATG-3’). To assess whether these elements were required for Loz1-mediated gene repression, we generated constructs containing nucleotide substitutions that targeted the highly conserved ‘G’ (at position 5) and ‘A’ (at position 6) in LRE2 and LRE3, and the highly conserved ‘T’ (at position 7) and ‘C’ (at position 8) in LRE1 (Fig. 4B). Consistent with previous studies we found that the activity of the SPBC1348.06c-lacZ reporter was ~50-fold lower in zinc-replete cells, in a manner that was dependent upon Loz1 (Fig. 4C strains 1 and 9). Nucleotide substitutions that targeted individual LREs reduced Loz1-mediated repression by ~10–20%, with substitutions targeting LRE2 having the most significant effect (Fig. 4C strains 3–5). When the mutations disrupting LRE1 and LRE2 were combined, there was no further reduction (Fig. 4C strain 6). However, when nucleotide substitutions were introduced that targeted all three LREs, Loz1-mediated gene repression was eliminated (Fig. 4C strain 7). Loz1-mediated gene repression was also eliminated when nucleotide substitutions were introduced that targeting the G and A in all three LREs (data not shown). The results are consistent with at least one LRE being required for Loz1-mediated gene repression.

To gain additional evidence that LREs were required for Loz1 regulation we tested whether this element was required for the Loz1-mediated repression of a gcd1-lacZ reporter. The gcd1 promoter contains two putative LREs at positions −303 to −293 and −384 to −374, relative to the transcriptional start site (Fig. 5A). To assess the role of the individual elements we generated long and short gcd1-lacZ reporter constructs containing one or both of the elements. Growth in high zinc led to a 12-fold reduction in reporter activity of the longer gcd1-lacZ reporter, in a manner that was dependent upon Loz1 (Fig. 5B strains 2 and 7). The shorter reporter was also regulated by zinc and Loz1. However, this reporter had higher levels of activity in low zinc, and growth in high zinc only resulted in an ~2-fold decrease in β-galactosidase activity (Fig. 5B, strain 3). To test whether LRE1 was required for the 2-fold decrease in the activity of the shorter reporter we substituted the highly conserved ‘C’ at position 8 in the LRE to a ‘G’. While this mutation was expected to interfere with Loz1-mediated gene repression, it did not significantly alter the level of gene repression observed in high zinc conditions (Fig. 5B strain 4).

Figure 5.

Loz1-dependent repression of the gcd1 promoter is dependent upon multiple LREs A) A schematic diagram of the gcd1-lacZ reporter highlighting the positions and sequences of putative LREs (green and blue boxes) and the position of the gcd1 5’UTR (purple boxes and line). The numbers indicate the distance of the first base of each motif relative to the putative transcriptional start site for gcd1. Nucleotide substitutions that were introduced to disrupt LRE function are shown in bold.

B) Wild-type or loz1Δ cells expressing the indicated gcd1-lacZ reporter genes were grown overnight in ZL-EMM supplemented with 0 or 100 μM zinc before cells were harvested and assayed for β-galactosidase activity. The relative positions of the putative LREs in the reporters are indicated by green and blue boxes. The positions of mutated LREs are shown by green and blue boxes with red crosses. Numbers represent the average value with error bars representing standard deviations. N=3

Potential explanations of the above results are that Loz1 binds to additional sites in the gcd1 promoter, or that Loz1 regulates the expression of other proteins or RNAs that affect gcd1 expression. To further probe Loz1 DNA binding specificity, we used electrophoretic mobility shift analysis (EMSA) to screen for nucleotide substitutions that affected Loz1 DNA binding in vitro. The recombinant Loz1 protein used for these analyses consisted of the C-terminal amino acid residues 426–522 containing two C₂H₂-type zinc fingers and an adjacent accessory domain. A DNA-protein complex was detected when purified Loz1^426−522 was incubated with ³²P-labeled double-stranded oligomers containing the ‘wild-type’ motif 5’-CGNMGATCNTY-3 (Fig. S2). The formation of this complex was also inhibited by incubation with competitor unlabeled ‘wild-type’ oligomers indicating that Loz1 binds to this sequence in a site-specific manner. When EMSAs were performed using related oligomers containing systematic nucleotide substitutions to the core LRE sequence, no DNA-protein complexes were detected when nucleotide substitutions were introduced that altered the ‘G’ at position 2, and the ‘A’, ‘T’, or ‘C’, at positions 5 to 7 (Fig. S3 and S4). Substitutions of the G at position 5 with a C and T also led to the loss of complex formation, whereas substitutions altering nucleotides at positions 3 or 4 in the LRE had no effect on DNA-complex formation (Fig. S3 and S4).

The above screen suggested that the G at position 2 in the LRE and the core ATC nucleotides at positions 6–8 were critical for site-specific Loz1 DNA binding activity. They also suggested that a G or A can be present at position 5. When we searched the shorter gcd1 promoter region for motifs with a core sequence of 5’-NGNN(G/A)ATCNNN-3’, we found one additional motif on the reverse strand at position −276 to −286 (5’-TGCAAATCTTC-3’), which we designated gcd1 LRE3 (Fig. 5A LRE3). When mutations were introduced targeting key nucleotides in gcd1 LRE3, mutation of this element led to a reduction in Loz1-mediated gene repression in high zinc, whereas disruption of both LRE2 and LRE3 led to a further loss of Loz1-mediated gene expression (Fig. 5B strains 5 and 6). These results are consistent with Loz1 binding to derivatives of the LREs with an A at position 5. They also suggest that Loz1-mediated gene repression is relatively complex, and that multiple factors including the LRE sequence, the total number of LREs within a promoter, and the position of the LREs, may affect the levels of gene repression observed in high zinc conditions.

The LRE is sufficient for Loz1-mediated gene repression

To determine if the LRE was sufficient for Loz1-dependent gene repression, we developed constructs to express a minimal CYC1-lacZ (^MinCYC1-lacZ) reporter in S. pombe (Fig. 6A). This reporter contains the well characterized minimal region of the CYC1 promoter from S. cerevisiae, which includes the canonical TATA box, but not any UASs (Bachhawat et al., 1995, Zhao et al., 1998). As this minimal system could potentially contain binding sites for transcription factors from S. pombe, we first integrated the ^MinCYC1-lacZ reporter into the leu1 locus of wild-type and loz1Δ cells and measured β-Galactosidase activity following growth in low and high zinc. As shown in Fig. 6B, very low levels of β-Galactosidase activity were observed in cells expressing the ^MinCYC1-lacZ reporter, suggesting that this minimal promoter region does not contain sequences that function as UASs in S. pombe.

Figure 6.

The LRE is sufficient for Loz1-mediated gene repression.A) A schematic diagram of the ^MinCYC1-lacZ reporter containing a thiamine responsive element (TRS) or LRE. The numbers indicate the distance of the first base of the motif to the putative ATG initiation codon.

B) Wild-type or loz1Δ cells expressing the indicated ^MinCYC1-lacZ reporter gene were grown overnight in ZL-EMM supplemented with 0 or 100 μM zinc before cells were harvested and assayed for β-galactosidase activity. Numbers represent the average value with error bars representing standard deviations. N=3

C) A schematic diagram of the ^MinCYC1-lacZ reporter containing the TRS and 2 copies of LRE1 and LRE2 from the SPBC1348.06c-lacZ reporter. The numbers indicate the distance of the first base of the motif to the putative ATG initiation codon.

D) Wild-type or loz1Δ cells expressing the indicated ^MinCYC1-lacZ reporter gene were grown overnight in ZL-EMM supplemented with 0 or 100 μM zinc before cells were harvested and assayed for β-galactosidase activity. Numbers represent the average value with error bars representing standard deviations. N=3.

We next introduced a thiamine-regulatory sequence (TRS) 241 bp upstream of the translational ATG (Fig. 6A). The TRS is a previously mapped cis-acting element that is required for strong activation of the nmt1 promoter (Zurlinden & Schweingruber, 1997). Consistent with this sequence containing a binding site for a transcriptional activator, the insertion of the TRS led to an ~600-fold increase in β-Galactosidase activity (Fig. 6B). This increase in activity was also independent of Loz1 and was not affected by cellular zinc status. When an LRE was inserted into a similar site, it was not able to confer any increase in activity.

To test whether the LRE was sufficient for Loz1-mediated gene repression, we inserted 1 or 2 copies of the overlapping LRE1 and LRE2 from the SPBC1348.06c promoter, upstream of the TRS (Fig. 6C). We introduced multiple copies of the LRE for these studies because it is unclear how Loz1 facilitates gene repression, and whether other factors such as distance of the LRE from UAS’s and other promoter elements are important for Loz1-mediated repression. In this minimal system, the insertion of two LREs upstream of the TRS resulted in ~1.5-fold decrease in β-galactosidase activity in zinc-replete cells, whereas the introduction of 4 LREs led to a ~5-fold decrease (Fig. 6D). In both cases the zinc-dependent decrease in β-galactosidase activity was dependent upon Loz1. These results are consistent with the LRE being sufficient for Loz1-mediated gene repression.

As the LRE sequence was required and sufficient for Loz1-mediated gene repression we searched the promoters of the putative Loz1 target genes for consensus and non-consensus LREs. 25 of the 27 target genes contained at least one NGNNGATCNNN motif within the ChIP-seq. peak or within 100 bp of the peak (Table 1). Although two ChIP-seq peaks did not contain this motif, multiple NGNNAATCNNN sequences were present suggesting that Loz1 may bind to non-consensus elements at these promoters. As these genes are Loz1 ChIP-seq targets and are repressed in high zinc in a manner that is dependent upon Loz1, we propose that these 27 genes form the Loz1 regulon in fission yeast. As discussed below, these genes encode proteins that have known roles in zinc homeostasis, small molecule transport, covalent protein folding, and carbohydrate metabolism.

Table 1. Loz1 target genes that were identified by RNA-seq and ChIP-seq.

Gene	Common Name	FunctionÏ	RNA-seq Δloz1/WT +Zn Fold increase Log2 (average)	Putative LRE Sequence	Position†	Loz1-Dependent regulation‡
Transmembrane transport
SPAC977.17		MIP water channel	7.40	TGCAGATCCCG	−970
				TGCAGATCCAA	−993
SPBC16D10.06	zrt1	Plasma membrane ZIP zinc transmembrane transporter	5.16	AGCAGATCCTT	−735	(Corkins et al., 2013, Ehrensberger et al., 2014)
				TGACGATCGAA	−1515
SPCC548.07c$	ght1	Plasma membrane high−affinity glucose:proton symporter	4.70	GGAAAATCCAA	−1766
				TGGCAATCCCC	−1808
				TGCAAATCGCC	−1817
SPCC1529.01$		Transmembrane transporter	3.48	GGAAAATCCAA	−1281
				TGGCAATCCCC	−1239
				TGCAAATCGCC	−1230
SPCC569.05c		Spermidine family transporter	2.84	CGTAGATCATG	−639	This study
SPBC1683.08	ght4	Plasma membrane hexose:proton symporter	1.92	GGACGATCTCC	−959
				CGGAGATCGTC	−980
				TGAAGATCGAT	−1056
SPAC4H3.01		DNAJ domain protein Caj1/Djp1 type	1.39	CGTCGATCATT	−1321
				TGCCGATCACT	−1339
Carbohydrate metabolic processes
SPAC977.16c	dak2	Dihydroxyacetone kinase	8.61	TGCAGATCCAA	+530	This study
				CGGGATCTGCA	+497
SPCC13B11.02c	adh1AS	Adh1 antisense transcript	7.99	CGTCGATCATT	−295	(Corkins et al., 2013)
				CGGCGATCCAT	−106
				TGCCGATCAGA	−163
				AGAAGATCAAA	−185
				TGATGATCAAA	−207
SPAC5H10.06c	adh4	Alcohol dehydrogenase	6.80	CGACGATCACG	−917	(Corkins et al., 2013, Ehrensberger et al., 2014)
				AGAAGATCTTG	−953
				AGAAGATCCAC	−954
				CGTGGATCTTC	−965
				TGTGGATCACT	−981
SPCC794.01c	gcd1	Glucose dehydrogenase	6.30	GGATGATCGAA	−816	(Corkins et al., 2017)
SPAC4F8.07c	hxk2	Hexokinase 2	1.08	AGTCGATCGAC	−1870
				CGTCGATCGAC	−1859
Gene Regulation and signaling
SPAC25B8.19C	loz1	Zinc−responsive transcription factor	3.01	CGCAGATCATT	−1397	(Corkins et al., 2013)
				CGTCGATCATA	−1386
				CGCCGATCACC	−1075
SPAC57A10.05c	pof1	F−box/WD repeat protein	1.39	CGTAGATCATT	−750
				TGCAGATCAGT	−855
SPAC19D5.01¶	pyp2	Tyrosine phosphatase	1.46	CGAGGATCAGA	−882
SPAC4G8.03c	puf5	Pumilio family RNA−binding protein mRNA metabolic process	1.38	AGTCGATCTTG	−2728
Unknown/pombe specific
SPBC1348.06c§ (SPBPB2B2.15 SPAC977.05c)		Velum formation protein	11.07	CGTAGATCATG	−130	(Corkins et al., 2013)
				CGACGATCATG	−147
				GGAAGATCTAC	−380
SPAC57A10.06	mug15		4.95	CGTAGATCATT	−954
				TGCAGATCAGT	−859
SPBC12C2.14c		dubious	3.76	CGTCGATCTTT	−692
				CGCAGATCTTT	+75
				CGTCGATCCTG	+188
				CGTCGATCTTG	+158
				TGTCGATCACG	+149
				TGGTGATCCAA	+113
SPBC660.05¶			3.06	GGAAGATCATA	−606
SPAC5H10.07			2.51	CGACGATCACG	−599
				AGAAGATCTTG	−563
				AGAAGATCCAC	−562
				CGTGGATCTTC	−551
				TGTGGATCACT	−535
SPAC2E1P3.05c¶		Fungal cellulose binding domain protein	1.82	AGTCGATCCTG	−141	This study
Other
SPAC1F12.10c		NADPH−hemoprotein reductase	2.02	CGTCGATCATT	−229	This study
				TGCCGATCACT	−211
SPCC576.03c	tpx1	thioredoxin peroxidase unfolded protein binding	1.54	CGCCGATCATT	−194	This study
				TGCAGATCCTG	−156
SPCC11E10.01$		Cystathionine beta−lyase	1.37	AGCTAATCATA	−989
SPBC12C2.03c		Methionine synthase reductase	1.18	CGCAGATCTTT	−1648
				CGTCGATCCTG	−1761
				CGTCGATCTTG	−1731
				TGTCGATCACG	−1722
				TGGTGATCCAA	−1696
				CGTCGATCTTT	−882
SPCC757.12		Alpha−amylase homolog	1.04	GGCCGATCGCT	−702

^†Numbers represent the position of the LRE relative to the putative translational ATG codon, with the exception of the adh1AS transcript. For this ncRNA the number represents the distance of the LRE relative to the previously mapped transcriptional start in low zinc conditions (Ehrensberger et al., 2013).

^‡Genes where Loz1-dependent regulation has been confirmed independently by RNA blot or lacZ-reporter gene analysis, in this study or in other reports.

^§These genes were grouped together because SPBC1348.06c shares 100% and ≥99% identity SPBPB2B2.15 and SPAC977.05c at the mRNA level, respectively. The sequence of the LREs and surrounding regions are also identical for these genes.

^¶Genes where the putative LRE sequence was found with the region surrounding the peak detected in the ChIP-seq analysis

^$Genes that did not contain a NGNNGATCNNN sequence in the Loz1 ChIP seq. peak or surrounding region. These ChIP-seq peaks contained NGNNAATCNNN sequences, which are shown in gray text.

^ÏFunctional categories were assigned based on GO term annotations obtained from PomBase GO tool analysis (Lock et al., 2018, The Gene Ontology, 2019).

Discussion

In this study we examined the role of the Loz1 transcription factor in controlling gene expression in response to cellular zinc status. We have found that at the majority of promoters Loz1 binds to LREs in high zinc conditions and that the recruitment of Loz1 to these elements is associated with lower levels of gene expression. Based on these observations we propose that the primary function of Loz1 is gene repression when zinc is in excess.

When expressed from the constitutive pgk1^ΔTATA promoter, ~ 75% of the genes identified as Loz1 target genes had a Loz1-binding peak that was enriched in high zinc growth conditions (Table 1 and S3). As Loz1 is localized to the nucleus under all growth conditions (Corkins et al., 2013), these results are consistent with Loz1 DNA binding activity being regulated by zinc. How might zinc affect Loz1 DNA binding function? Functional dissection of the Loz1 protein has shown that the smallest region that is able to confer zinc-dependent gene repression maps to its last 96 amino acids (Ehrensberger et al., 2014). This same small region contains two C₂H₂ zinc finger domains and an adjacent accessory domain, both of which are necessary for Loz1 DNA binding activity in vitro and Loz1-mediated gene repression in vivo (Ehrensberger et al., 2014). As the minimal zinc responsive domain of Loz1 contains two C₂H₂ zinc finger domains, one potential explanation for the zinc-responsiveness of the Loz1 protein is that one or both of these domains bind zinc with a low affinity. In this model the low affinity zinc finger domain(s) would be in a largely unstructured apo-form in low zinc conditions preventing Loz1 from binding to DNA. On the other hand, when there were adequate levels of intracellular zinc, the zinc finger domain(s) would bind zinc and fold into a conformer that permits DNA binding and gene repression.

Our current results also do not exclude models where both of the zinc finger domains are occupied with zinc under all growth conditions, and therefore an alternative regulatory event controls Loz1 DNA binding activity. For example, in some transcription factors post-translational modifications of key residues in zinc finger domains or their linker regions can interfere with DNA binding (Dovat et al., 2002, Kluska et al., 2018). Other viable models therefore include the possibility that Loz1 is post-translationally modified or interacts with other regulatory factors in a manner that is dependent upon cellular zinc status, and that this regulatory event affects DNA binding function. Studies of other eukaryotic zinc-responsive transcription factors including MTF-1 and Zap1 have shown that they are able to sense changes in intracellular zinc levels using regulatory zinc finger domains (Bird et al., 2003, Chen et al., 1999, Gunther et al., 2012, Wang et al., 2006). We therefore favor a model where Loz1 is the direct sensor of zinc in yeast. Our ongoing experiments are currently testing whether one or both of the Loz1 zinc finger domains bind zinc with a reduced affinity.

By examining the activity of the pgk1^ΔTATA-driven Loz1GFP we found that the over-expression of Loz1GFP results in higher levels of gene repression under low zinc conditions. If Loz1 DNA binding activity is dependent upon zinc, why would overexpression of Loz1 in a zinc-limited cell lead to increased gene repression? One possible explanation is that the levels of Loz1 within cells are not sufficient to occupy all of the Loz1 binding sites within the genome. In this scenario, an increase in the total levels of Loz1 could result in a higher number of active Loz1 proteins under a given condition, which in turn may lead to an increased occupancy of LREs and a higher level of gene repression. An alternative possibility is that a yet to be identified factor(s) maintains Loz1 in its inactive state in low zinc. While we have not yet identified any additional factors required for Loz1-mediated gene repression, there are precedents for this type of regulation. For example, in S. cerevisiae the inactivation of the transcription factor Hap1 in low heme is dependent upon the recruitment of Ssa-type Hsp70 molecular chaperones that mask the Hap1 DNA binding and transactivation domains (Hon et al., 2001). These interactions are critical to maintain Hap1 in an inactive state as reduced expression of Hsp70 chaperones or overexpression of Hap1 both lead to the derepression of Hap1 target genes. In fission yeast a related model could exist where additional proteins maintain Loz1 in its inactive state in low zinc, and over-expression of Loz1 overrides this mechanism.

While the preferentially recruitment of Loz1 to its target gene promoters in high zinc provides a simple explanation of how gene repression occurs in a manner that is dependent upon zinc, approximately 25% of the genes that were identified as Loz1 targets had a Loz1-binding peak that was enriched in high and low zinc conditions (Fig. 2C and Table S3). Given that the Loz1GFP fusion protein used for the ChIP-seq analysis was over-expressed, we currently cannot eliminate the possibility that the ability of Loz1 to bind to these promoters in a zinc-independent manner is an artefact of the higher levels of Loz1 in these cells. However, studies of different zinc-responsive transcription factors have found that these proteins often contain multiple zinc-responsive domains that are independently regulated by zinc (Frey & Eide, 2011, Wilson & Bird, 2016, Gunther et al., 2012). It is therefore possible that Loz1-mediated gene repression at some promoters may be more complex, with Loz1 being bound under all conditions and an alternative regulatory mechanism triggering its inactivation in high zinc.

By analyzing ChIP-seq and RNA-Seq targets we identified 27 genes that are repressed in high zinc in a Loz1-dependent manner. Of these genes, eight have homologs or functional homologs in S. cerevisiae that are regulated by Zap1 (zrt1, adh4, adh1AS, tpx1, pof1, SPAC977.05c, SPBC1348.06c, SPBPB2B2.15). Given the considerable evolutionary distance between budding and fission yeast, the conserved regulation of these genes in both systems suggests that this subset of genes have particularly important roles in zinc homeostasis or for survival when zinc levels are low. Consistent with this hypothesis, Zrt1 is critical for the survival of S. pombe and S. cerevisiae under zinc-deficient conditions (Zhao & Eide, 1996, Boch et al., 2008). The ADH4 and TSA1 genes are also required for growth in low zinc in S. cerevisiae (North et al., 2012).

The identification of tpx1 (TSA1 in S. cerevisiae) and pof1 (MET30 in S. cerevisiae) as Loz1 target genes highlight two new conserved zinc-regulated pathways. In S. cerevisiae the peroxiredoxin Tsa1 is critical for survival in low zinc because it has a molecular holdase function which prevents aggregation of unfolded and misfolded proteins (MacDiarmid et al., 2013, Wu et al., 2007). While it is not yet known if the molecular chaperone function of Tpx1 is critical for survival of fission yeast in low zinc, the conserved regulation of peroxiredoxin gene expression by zinc suggests that increased levels of unfolded and misfolded proteins may be a widespread problem in low zinc conditions. Based on studies of Met30 in S. cerevisiae the regulation of pof1 by Loz1 could be a mechanism to help reduce oxidative stress under conditions of zinc deficiency. In S. cerevisiae, Met30 facilitates the degradation of Met4, a transcription factor required for the expression of sulfate assimilation genes (Wu et al., 2009). Since sulfate assimilation requires high levels of NADPH, an important cofactor for oxidative stress defense, the net result of increased expression of MET30 in low zinc is reduced sulfate assimilation and conservation of NADPH. In fission yeast Pof1 controls the degradation of Zip1, a transcription factor required for the expression of sulfate assimilation genes (Harrison et al., 2005, Guo et al., 2012). The Loz1-dependent regulation of pof1 expression could therefore be a related mechanism to conserve NADPH. It is also noteworthy that gcd1 is a Loz1 target and that derepression of gcd1 expression in low zinc results in glucose being directed into the pentose phosphate pathway via the gluconate shunt (Corkins et al., 2017). As the pentose phosphate pathway is a major route for the regeneration of NADPH, S. pombe may rely on multiple mechanisms to combat higher levels of oxidative stress in low zinc.

In addition to identifying conserved regulatory pathways Loz1 controls the expression of many unique genes. As examples, Loz1 target genes include puf5, which encodes a member of the pumilio family of RNA binding proteins, and pyp2 which encodes a putative tyrosine phosphatase. As 102 additional transcripts were expressed at higher levels in loz1Δ cells that were not Loz1 ChIP-seq targets, it is tempting to speculate that alterations in the expression of these Loz1 target genes may affect the expression of other genes by influencing mRNA decay or by affecting the activity of other transcription factors. Another new Loz1 target gene is dak2, which encodes dihydroxyacetone kinase, an enzyme that is involved in the breakdown of glycerol via a dihydroxyacetone phosphate (DHAP) intermediate. In the DHAP pathway glycerol is oxidized to dihydroxyacetone, which is then phosphorylated to DHAP by Dak1 or Dak2 (Matsuzawa et al., 2010). Why would dak2 expression be regulated by Loz1? One potential explanation is that in low zinc there is a significant reduction in levels of the zinc-requiring enzyme aldolase, which catalyzes the reversible cleavage of fructose 1,6, bisphosphate into DHAP and Glyceraldehyde-3-phosphate (G3P) (Wang et al., 2018). As increased Dak2 activity would increase the generation of DHAP, increased expression of dak2 in low zinc could be a mechanism to compensate for reduced aldolase activity under this condition. While additional experiments are needed to test this hypothesis, they highlight that studies with other Loz1 target genes will likely provide additional insight into the molecular consequences of zinc deficiency.

Experimental Procedures

Strains and growth conditions

The yeast strains used in this study are shown in Table S4. All strains were grown in YES medium supplemented with 3% glucose and 225 mg/l leucine, 225 mg/l uracil and 338 mg/l adenine, or in Zinc-Limited Edinburgh Minimal Medium (ZL-EMM) which is a derivative of EMM which does not contain zinc (Choi et al., 2018). Prior to inoculation in ZL-EMM, cells were pre-grown to exponential phase in Yeast Extract with supplements (YES) medium and washed twice with ZL-EMM with no added zinc.

Plasmid construction

To generate pgk1^ΔTATA-Loz1GFP a cloning strategy was used that first involved generating Pgk1-Loz1-GFP, which contains a fusion of the loz1 and GFP open reading frames under the control of 840 bp of the pgk1 promoter and its 5’UTR. To create Pgk1-Loz1-GFP, the pgk1 promoter and 5’UTR were amplified by PCR using primers containing Kpn1/EcoRI sites. The Kpn1/EcoRI digested PCR product was then inserted into similar sites of the vector JK-GFP (Corkins et al., 2013) to generate JK-pgk1-GFP. To generate an in-frame fusion of Loz1 to GFP, the loz1 open reading frame without its stop codon was released from the vector pSK-loz1 (Corkins et al., 2013) by digestion with EcoRI and BamHI and subcloned into similar sites in the vector JK-pgk1-GFP to generate pgk1-Loz1GFP. To generate pgk1^ΔTATA-Loz1GFP, QuikChange mutagenesis (Agilent) was used to delete the first four nucleotides of the canonical TATA box found at position −85 to −82 in the pgk1 promoter relative to the translational ATG. The construction of the zrt1-lacZ reporter was described previously (Ehrensberger et al., 2014). To create the SPBC1348.06c-lacZ reporter a DNA fragment encompassing positions +3 to −1205 bp of the SPBC1348.06c promoter and ORF was amplified using PCR. This fragment was inserted into the EagI and BamHI sites of the vector JK-lacZ (Ehrensberger et al., 2013). Primers for these reactions were designed so that the translational ATG of SPBC1348.06c was inserted in frame with the lacZ coding region. Derivatives of this vector containing nucleotide substitutions within LREs were generated by cloning overlapping PCR products containing the following nucleotide substitutions shown in bold: LRE1 (5’-CGTAGAGGATG-3’); LRE2 (5’-CGACCCTCATA-3’); and LRE3 (5’-GGAACCTCTAC-3’). Constructs containing substitutions in multiple LREs, were generated using the vector templates that already contained substitutions in alternative LREs. The construction of the reporter gcd1-lacZ, which contains 1452 bp of the gcd1 promoter and 5’UTR was described previously (Corkins et al., 2017). The shorter gcd1 reporter was generated in a similar manner and contains 1108 bp of the gcd1 promoter and 5’UTR fused to the lacZ open reading frame. The minimal CYC1 promoter was generated by using PCR to amplify a 700 bp minimal CYC1 promoter fragment from the vector pNB404 (Bachhawat et al., 1995). Primers were designed to introduce this fragment into the SacII/BamHI sites of the vector JK-lacZ. Cloning of oligonucleotide TRS insert was performed as described by (Zhao et al., 1998). Briefly, complementary oligonucleotides containing suitable overhangs to clone into the SalI restriction sites were phosphorylated using T4 polynucleotide kinase, annealed in the presence of 0.32 × SSC, and were subcloned into SalI digested JK-lacZ. Combinations of LREs from the SPBC1348.06c promoter were introduced upstream of the TRS by two-step overlapping PCR (Ho et al., 1989). All PCR products generated by overlapping PCR were inserted into the SacII/BamHI of JK-lacZ. All constructs were confirmed by sequencing.

ChIP Analysis

For ChIP analysis 75 ml of the strain loz1Δ pgk1^ΔTATA-Loz1GFP were grown to an OD⁶⁰⁰ of ~6.0 in ZL-EMM supplemented with or without 100 μM Zn²⁺ and were treated with 1% formaldehyde for 10 min at room temperature. Crosslinking was quenched by the addition of 250 mM glycine. Cells were then washed, resuspended in the lysis buffer (50mM Hepes-KOH pH 7.5, 140mM NaCl, 1mM EDTA, 1% (v/v) Triton X-100, 0.1% (w/v) sodium deoxycholate, 0.4mM DTT, and protease inhibitors) and lysed by vortexing for 7 × 1 min in the presence of zirconium beads. To obtain cross-linked chromatin, cell lysates were centrifuged at 14,000 rpm for 5 min and the supernatant discarded. After centrifugation, cross-linked chromatin was resuspended in lysis buffer, DNA sheared to an average length of 500 bp by sonication, and cell debris removed by centrifugation at 14000 rpm for 5 min. For immunoprecipitations, chromatin solutions were incubated with anti-GFP antibodies (Abcam Ab290) and protein A magnetic beads (Invitrogen™ Dynabeads™ Protein A) overnight at 4°C. After Immuno-complexes were harvested the crosslinks were reversed by heating at 65°C for 5–7 hrs and DNA purified by Qiagen PCR purification kit. The percent input method was used to calculate the signal of enrichment as described by (Chung et al., 2014).

ChIP-seq Analysis

ChIP seq samples were prepared as described above with the exception that the sonication step was performed using a Covaris Evolution ultrasonicator. The ChIP-seq Methodology performed by the Institute for Genomic Medicine at Nationwide Children’s Hospital. ChIP-seq libraries were constructed from 5 ng of fragmented ChIP DNA using NEB Ultra II FS library prep kit (New England Biolabs, Ipswhich MA). Briefly, DNA fragment ends were 5´ Phosphorylated, dA-tailed, and ligated with a unique, dual UMI indexed adaptor (Integrated DNA Technologies, Iowa). Following purification, using a magnetic-bead based approach (AMPure XP System), adaptor-ligated DNA was amplified by 8–9 PCR cycles. Quality of libraries were determined via Agilent 2200 TapeStation using High Sensitivity D1000 reagents, and quantified using Kappa SYBR®Fast qPCR kit (KAPA Biosystems, Inc, MA). Approximately, 4 million paired-end 150 bp sequence reads were generated per sample using Illumina MiniSeq platform. Sequencing reads were analyzed using the HOMER suite (Heinz et al., 2010). Tag directories were created using HOMER makeTagDirectory with -fragLength parameter manually specified for each sample. The makeUCSCfile program was used to generate bed graph files used for visualization in the Integrative Genomics Viewer (Robinson et al., 2011). The findPeaks program with the -style option set to ‘factor’ was used for peak-calling and findMotifsGenome.pl was used to identify motifs enriched in the ChIP-seq peaks. Differentially binding analysis for ChIP-seq signals between low and high zinc replicates was conducted using the Diffbind package, which uses the TMM normalization method from the edgeR package to identify differential peaks (Afgan et al., 2016, Ross-Innes et al., 2012, Stark & Brown, 2011).

Motif discovery and Enrichment in Loz1-regulated genes

For analysis of conserved promoter elements, the MEME motif finder was accessed at meme-suite.org and run using default parameters (Bailey et al., 2009). Weeder and CONSENSUS were run at default parameter settings using their respective standalone applications for Linux (Pavesi et al., 2004, Hertz & Stormo, 1999).

Protein extraction and Immunoblot analysis

Crude protein extracts were prepared using a Trichloroacetic Acid (TCA) precipitation as described by (Peter et al., 1993). Briefly 5 ml of cells were harvested by centrifugation for 2 min at 3500 rpm. Cell pellets were resuspended in 0.5 ml of ice-cold buffer A (20 mM Tris (pH 8.0), 50 mM NH₄OAc, 0.5 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride) and were immediately mixed with 0.5 ml of ice cold 20% trichloroacetic Acid. Cells were then vortexed in the presence of glass beads for 2 × 0.5 min, with 2 min cooling on ice in-between. The supernatant was the placed into a new tube, proteins pelleted by centrifugation for 10 min at 12,000 rpm, and the pellet resuspended in trichloroacetic acid-sample buffer (3% SDS, 100 mM Tris base pH 11, 3 mM DTT). Extracts were then boiled for 10 min and insoluble cell debris pelleted by centrifugation for 2 min at 12,000 rpm. Proteins were separated by SDS-PAGE polyacrylamide gene electrophoresis followed by immunoblotting to a nitrocellulose membrane. Immunoblots were probed with anti-GFP (G1544, Sigma) and anti-Act1 (ab3280–500) primary antibodies and IRDye800CW-conjugated anti-mouse IgG (LI-COR) and IRDye680-conjugated anti-rabbit IgG (LI-COR) secondary antibodies. An Odyssey infrared image system (LI-COR) was used for visualization and analysis of signal intensities.

β-galactosidase assays

β-galactosidase assays were performed as described by (Guarente, 1983). Briefly in these assays 1 ml of cells are resuspended in lacZ buffer (0.06 M Na₂HPO₄.7H₂O, 0.04M NaH₂PO₄.H₂O, 0.01 M KCl, and 0.001 M MgSO₄) and lysed by vortexing for 10 s in the presence of 0.05 ml of CHCl₃ and 0.05 ml of 0.1% SDS. β-galactosidase activity was determined by monitoring the hydrolysis of the substrate o-Nitrophenyl-β-D-galactosidase (ONPG), and was calculated using the following equation: (ΔA 420 × 1000)/(min × ml of culture × culture absorbance at 600 nm). Error bars represent the standard deviation of values obtained from 3 independent repeats.

RNA-blot hybridizations and RNA-seq analysis

For RNA blot analyses total RNA was purified from the indicated cells using the hot acidic phenol method (Collart & Oliviero, 1993). Depending on transcript abundance, between 5–20 μg of denatured total RNA was separated on formaldehyde gels and transferred to nylon membrane according to the method described by Sambrook et al., (Sambrook & Russell, 2001). Following transfer RNA blots were incubated with single stranded ³²P-labeled RNA probes that were generated from purified PCR products using the Ambion™ MAXIscript™ T7 In Vitro Transcription kit according to manufacturer’s instructions. Signal intensities representing transcript abundance were visualized using GE Typhoon FLA 9500. For RNA seq analysis wild-type and loz1 deletion strains were grown to mid log phase in zinc-limited EMM supplemented with 100 μM Zn. Total RNA was isolated using the RiboPureTM RNA Purification Kit for yeast (ThermoFisher Scientific) according to the manufacturer’s instructions and subsequent library preparation and next generation sequencing performed by Beckman Coulter Genomics. Reads were aligned to the S. pombe reference genome using Hisat2 (Kim et al., 2015), and transcripts assembled and FPKM (fragments per kb for a million reads) values determined using Cufflinks (Trapnell et al., 2012). Following normalization, differential expression of transcripts was assessed with cuffdiff of the cufflinks suite, and significant genes identified based on a false discovery rate of 0.05 (Trapnell et al., 2012, Trapnell et al., 2013).

Data availability

The RNA‐Seq and ChIP‐Seq data generated in this study were submitted to the Gene Expression Omnibus 347. The accession number for both data sets (SuperSeries record) is GSE130846. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE130846