The GIS2 Gene Is Repressed by a Zinc-Regulated Bicistronic RNA in Saccharomyces cerevisiae

Zinc homeostasis is essential for all organisms. The Zap1 transcriptional activator regulates these processes in the yeast Saccharomyces cerevisiae. During zinc deficiency, Zap1 increases expression of zinc transporters and proteins involved in adapting to the stress of zinc deficiency. Transcriptional activation by Zap1 can also repress expression of some genes, e.g., RTC4. In zinc-replete cells, RTC4 mRNA is produced with a short transcript leader that is efficiently translated. During deficiency, Zap1-dependent expression of an RNA with a longer transcript leader represses the RTC4 promoter. This long leader transcript (LLT) is not translated due to the presence of small open reading frames upstream of the RTC4 coding region. In this study, we show that the RTC4 LLT RNA also plays a second function, i.e., repression of the adjacent GIS2 gene. In generating the LLT transcript, RNA polymerase II transcribes RTC4 through the GIS2 promoter. Production of the LLT RNA correlates with the decreased expression of GIS2 mRNA and mutations that prevent synthesis of the LLT RNA or terminate it before the GIS2 promoter renders GIS2 mRNA expression and Gis2 protein accumulation constitutive. Thus, we have discovered an unusual regulatory mechanism that uses a bicistronic RNA to control two genes simultaneously.


Introduction
In prokaryotic organisms, there are many examples where the expression of multiple genes is controlled by a single regulatory region. The classic bacterial operon comprises two or more protein-coding regions (i.e., cistrons) that are regulated by a single upstream promoter. Production of polycistronic mRNA allows for the co-regulated production of multiple, often functionally related proteins because prokaryotic ribosomes are capable of internal translation initiation. While it has been often suggested that mechanisms of multigene regulation also exist in eukaryotes, specific examples are much less common. One mechanism of multigene regulation involves genes that are divergently transcribed and contain a common promoter region between them that controls both genes. Regulation of the GAL1 and GAL10 genes by Gal4 in Saccharomyces cerevisiae is one such example [1]. This and other bidirectional promoters generate a large number of RNA species in this yeast [2] and adjacent positioning of co-regulated genes has been observed widely across eukaryotes [3]. Another mechanism found in eukaryotes that is more similar to bacterial operons is where two or more genes

Plasmid Constructs
To generate pRHGM and pRHGM mZRE , five tandem myc epitope tags were amplified from the pFL44-S-MNR2-MYC plasmid [14] with flanking regions homologous to the 3 end of the GIS2 ORF and its terminator and inserted into pRTC4-HA or pRTC4-HA mZRE [8] by homologous recombination [15]. To generate pRG mZRE , pRHGM mZRE was digested with NheI and PacI restriction enzymes and the corresponding region, lacking the HA and myc epitope tags, was amplified from the chromosome and inserted by homologous recombination. To generate pRHGM Ter , the TEF terminator fragment was amplified from pFA6-KanMX4 [16] with flanking homology to the 3 end of the RTC4 ORF and its terminator and inserted into pRHGM by homologous recombination. To generate pGM, the GIS2 gene was amplified from pRHGM with 500 bp upstream and 1331 bp downstream of DNA flanking the GIS2 ORF and inserted into pRS315 [17] by homologous recombination. pGM retains the HA tags at the 3 end of the truncated RTC4 ORF.

Quantitative RT-PCR Analysis
Quantitative analysis of gene expression by real-time polymerase chain reaction (RT-PCR) was performed as previously described using an ABI Prism 7000 (Thermo Fisher Scientific, Waltham, MA USA) [18].

Northern Blot Analysis
Northern blotting was performed as previously described [18]. To make strand-specific antisense probes, PCR-generated fragments were amplified from template DNA introducing the T7 RNA polymerase promoter into the products in antisense orientation by inclusion in the downstream primer.
Full probe primer sequences (5 -3 ) were GIS2: CGGTCACGTTCAAACGGA and agttaatacgactcactatagggaCTAATTCCGTCCTCTTTCATACAGTC and TAF10: ATGGATTTTGAGGAAGATTAC and agttaatacgactcactatagggaCTAACGATAAA-AGTCTGGGCG. Primers specific to the epitope tags were: HA: GGATCCGCTGGCTCCGCT and agttaatacgactcactatagggaTTAAGCGTAATCTGGAACGTCATATGGA, and myc: taaagctatg GAG-CAAAAGCTC and agttaatacgactcactatagggaatgTCGACAAGGCCTTGAATT. The common T7 promoter portion of each primer pair is indicated by lower case letters.

Mass Spectrometry Analysis
Frozen yeast pellets were resuspended in 100 µL 6 M guanidine-HCl, 100 mM Tris pH 8 and boiled for 5 min at 100 • C. Methanol was added to a concentration of 90% to precipitate protein and the sample was centrifuged for 10 min at 14,000× g. The protein pellet was resuspended in 200 µL lysis buffer (8 M urea, 100 mM Tris, 20 mM TCEP, 80 mM chloroacetamide) and diluted with 1 mL 100 mM Tris pH 8. Protein digestion was performed overnight with trypsin (4 µg) before centrifuging for 5 minutes at 10,000× g. The resulting supernatant was de-salted with Strata C18 solid phase extraction cartridges and quantified using Pierce Quantitative Colorimetric Peptide Assay (Thermo Fisher Scientific, Waltham, MA USA). Peptides were then dried in a vacuum centrifuge before resuspending in 0.2% formic acid.
Samples were analyzed using a liquid chromatography-mass spectrometry (LC/MS) instrument comprising an Orbitrap Fusion Lumos tribrid mass spectrometer (Thermo Fisher Scientific, Waltham, MA USA). Mobile phase A consisted of 0.2% formic acid in water and mobile phase B consisted of 0.2% formic acid in acetonitrile. A 75-min gradient ranging from 0% to 55% B was employed spanning a total runtime of 90 min. Analytes were injected onto a 1.7-micron C18 column packed in-house to a length of 35 cm and heated to 60 • C. Survey scans of peptide precursors were collected from 350-1350 Th with an AGC target of 1,000,000 and a resolution of 240,000 in the orbitrap followed by Higher-energy collisional dissociation (HCD) tandem mass spectrometry (MS/MS) turbo scans taken in the ion trap.
The resulting LC-MS proteomic data were processed using Maxquant software version 1.5.2.8 [20] and searched against a S. cerevisiae database downloaded from Uniprot on 8/10/16 [21]. The digestion enzyme was set to trypsin with up to two missed cleavages, carbamidomethylation of cysteines as a fixed modification, and oxidation of methionines and protein N-terminal acetylation as variable modifications. The match between runs feature was used to decrease missing data values within the data set. Precursor mass tolerance was 20 ppm and product ions were searched at 4.5 ppm tolerances. Peptides were filtered to a 1% false discovery rate (FDR) and combined to protein groups based on the rules of parsimony. The resulting label free quantitation (LFQ) values were log2 transformed and fold changes were calculated for each condition. A Student's t-test was used to calculate p-values, and these were converted to q-values to correct for multiple hypothesis testing. Statistically significant effects were defined as those proteins with 2-fold or greater changes and a q-value of < 0.05 after Benjamini-Hochberg correction.

Competitive Growth Assays
Growth assays were carried out as previously described [18]. Wild-type cells expressing GFP and isogenic untagged gis2∆ mutants were grown separately in liquid culture, mixed in equal numbers, and grown for 15 generations in zinc-replete or deficient media. Alternatively, wild-type cells expressing GFP and transformed with the plasmid vector and untagged wild-type cells transformed with the vector (pRS315) or pRG mZRE were grown separately in liquid culture, mixed in equal numbers, and grown for 15 generations in zinc-replete or deficient media. After analysis by flow cytometry on a FACSCalibur flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA), the final percentages of untagged cells in each culture were determined. A ratio below one indicated a zinc-limited growth defect. Statistical significance was determined by Student's t-test after Benjamini-Hochberg correction.

Opposite Regulation of GIS2 mRNA and Gis2 Protein Levels in Response to Zinc
We recently completed an analysis of the effect of zinc deficiency on zinc-binding proteins of S. cerevisiae [22]. Of the 582 predicted zinc-binding proteins in yeast, 229 were detectable by mass spectrometry in that study. Of those,~30 increased in abundance during the transition to zinc deficiency while~60 decreased. To evaluate the role of transcriptional regulation in these effects, we compared the proteomics results with RNA-sequencing (RNA-seq) data from cells grown under the same zinc-replete and deficient conditions [23]. While the observed changes in the level of many proteins correlated with changes in their mRNA levels, several others did not. One of the outliers in this analysis was the GIS2 gene and its protein product; we observed an~2-fold increase in RNA and a 4-fold decrease in protein abundance during deficiency. GIS2 encodes a ribosome-associated RNA-binding protein that has been proposed to facilitate translation of specific mRNAs that contain multiple repeats of the G-A/U-A/U sequence [24]. Gis2 contains seven consecutive CCHC zinc knuckle domains with the sequence C-X 2 -C-X 4 -H-X 4 -C (C = Cys, H = His, X = any amino acid).
To verify these results, we used a GFP-tagged chromosomal allele of GIS2 and RT-PCR and immunoblotting to assess mRNA and protein levels, respectively. mRNA levels of the known Zap1-regulated gene ZRT3 increased~12-fold in zinc-deficient cells thereby confirming the zinc status of our growth conditions, Figure 1A. compared the proteomics results with RNA-sequencing (RNA-seq) data from cells grown under the same zinc-replete and deficient conditions [23]. While the observed changes in the level of many proteins correlated with changes in their mRNA levels, several others did not. One of the outliers in this analysis was the GIS2 gene and its protein product; we observed an ~2-fold increase in RNA and a 4-fold decrease in protein abundance during deficiency. GIS2 encodes a ribosome-associated RNAbinding protein that has been proposed to facilitate translation of specific mRNAs that contain multiple repeats of the G-A/U-A/U sequence [24]. Gis2 contains seven consecutive CCHC zinc knuckle domains with the sequence C-X2-C-X4-H-X4-C (C = Cys, H = His, X = any amino acid).
To verify these results, we used a GFP-tagged chromosomal allele of GIS2 and RT-PCR and immunoblotting to assess mRNA and protein levels, respectively. mRNA levels of the known Zap1regulated gene ZRT3 increased ~12-fold in zinc-deficient cells thereby confirming the zinc status of our growth conditions, Figure 1A. Similarly, levels of RTC4 mRNA, encoded by the Zap1-regulated gene immediately adjacent to GIS2, increased 15-fold in zinc-deficient cells. Consistent with the RNA-seq results, GIS2 mRNA The data are the means of three biological replicates and the error bars indicate ± 1 SD. All changes in response to zinc status were found to be statistically significant with p-values of 0.004, 0.007, and 0.035 for ZRT3, RTC4, and GIS2, respectively. (B) Immunoblot analysis of wild-type (BY4741) and BY4741 GIS2::green flourecent protein (GFP) cells grown under the same conditions. Anti-GFP and anti-Pgk1 antibodies were used for protein detection. The location of size markers is indicated on the left side in kDa. (C) Transcription start sites mapped by Deep-RACE for CMD1 and GIS2 under zinc-replete (ZnR) and zinc-deficient (ZnD) growth conditions are shown relative to each gene's protein-coding region (shaded in gray). Independent sequencing reads representing mRNA 5 ends are plotted across each region. The x-axis coordinates represent the position on the corresponding chromosomes. Peak values of the number of sequence reads for each gene and growth condition are shown. The transcription start site mapping data are from Wu et al. [25].
Similarly, levels of RTC4 mRNA, encoded by the Zap1-regulated gene immediately adjacent to GIS2, increased 15-fold in zinc-deficient cells. Consistent with the RNA-seq results, GIS2 mRNA levels increased 1.5-fold. In contrast to the increase in mRNA abundance and consistent with the proteomics data, the level of Gis2-GFP protein decreased markedly in deficient cells, Figure 1B. The constitutively expressed Pgk1 protein showed little change in these samples. We had previously mapped transcriptional start sites genome-wide in zinc-replete and deficient cells [25]. While transcription start sites for a control gene, CMD1, were little affected by zinc status, there was a marked decrease in the number of transcript start sites mapped to the GIS2 promoter region despite the observed increase in RNA levels, Figure 1C. These results suggested that there was a major disruption of GIS2 promoter function in zinc-deficient cells and that GIS2-containing RNA were being produced from a site farther upstream of the promoter and ORF.

Co-Regulation of RTC4 and GIS2 by a Bicistronic Regulatory RNA
A potential explanation for these observations came from our recent studies of the RTC4 gene [8]. RTC4 is located immediately upstream of the GIS2 gene, transcribed in the same direction as GIS2, and is regulated by Zap1 by an unusual mechanism, Figure 2. In zinc-replete conditions, RTC4 mRNA is transcribed with a short 5 transcript leader that is efficiently translated. Two forms of RTC4 mRNA were detected, designated SLT1 and SLT2 for "short leader transcripts" one and two. Under zinc-deficient conditions, our studies indicated that Zap1 binds to a ZRE located 432 bp upstream of the RTC4 ORF and activates expression of a more abundant transcript with a longer 5 transcript leader that is poorly translated due to the presence of four small ORFs in the transcript leader. Production of this "long leader transcript" (LLT) inhibits the RTC4 promoter and synthesis of the RTC4 SLT1 and SLT2 mRNA likely by directly interfering with the binding of transcription factors and/or by causing formation of repressive chromatin structure. Using Northern blotting to map the endpoints of the LLT transcript, we found that it extends far beyond the RTC4 ORF passing through the GIS2 promoter and ORF and into the divergently transcribed FOL1 gene [8]. Given these prior observations, we hypothesized that the bicistronic LLT transcript not only regulates RTC4 but also controls the GIS2 promoter by shutting off its activity in zinc-deficient cells. In contrast, because of its low level of expression, the SLT2 bicistronic transcript does not inhibit GIS2 promoter function. proteomics data, the level of Gis2-GFP protein decreased markedly in deficient cells, Figure 1B. The constitutively expressed Pgk1 protein showed little change in these samples. We had previously mapped transcriptional start sites genome-wide in zinc-replete and deficient cells [25]. While transcription start sites for a control gene, CMD1, were little affected by zinc status, there was a marked decrease in the number of transcript start sites mapped to the GIS2 promoter region despite the observed increase in RNA levels, Figure 1C. These results suggested that there was a major disruption of GIS2 promoter function in zinc-deficient cells and that GIS2-containing RNA were being produced from a site farther upstream of the promoter and ORF.

Co-Regulation of RTC4 and GIS2 by a Bicistronic Regulatory RNA
A potential explanation for these observations came from our recent studies of the RTC4 gene [8]. RTC4 is located immediately upstream of the GIS2 gene, transcribed in the same direction as GIS2, and is regulated by Zap1 by an unusual mechanism, Figure 2. In zinc-replete conditions, RTC4 mRNA is transcribed with a short 5′ transcript leader that is efficiently translated. Two forms of RTC4 mRNA were detected, designated SLT1 and SLT2 for "short leader transcripts" one and two. Under zincdeficient conditions, our studies indicated that Zap1 binds to a ZRE located 432 bp upstream of the RTC4 ORF and activates expression of a more abundant transcript with a longer 5′ transcript leader that is poorly translated due to the presence of four small ORFs in the transcript leader. Production of this "long leader transcript" (LLT) inhibits the RTC4 promoter and synthesis of the RTC4 SLT1 and SLT2 mRNA likely by directly interfering with the binding of transcription factors and/or by causing formation of repressive chromatin structure. Using Northern blotting to map the endpoints of the LLT transcript, we found that it extends far beyond the RTC4 ORF passing through the GIS2 promoter and ORF and into the divergently transcribed FOL1 gene [8]. Given these prior observations, we hypothesized that the bicistronic LLT transcript not only regulates RTC4 but also controls the GIS2 promoter by shutting off its activity in zinc-deficient cells. In contrast, because of its low level of expression, the SLT2 bicistronic transcript does not inhibit GIS2 promoter function. . Several small ORFs located in the region between the RTC4 and GIS2 protein-coding regions are not shown. ZnR and ZnD denote zincreplete and zinc-deficient growth conditions, respectively. SLT1/2 and LLT refer to short leader transcripts and long leader transcripts, respectively. The box labeled ZRE indicates the location of a zinc-responsive element.
As a first test of this hypothesis, we assessed whether Rtc4 and Gis2 protein levels were coregulated in response to differences in zinc status. Using a plasmid expressing HA-tagged Rtc4 and myc-tagged Gis2, we found a very close correlation between the regulation of both proteins in response to zinc status, Figure 3. As zinc levels in the culture medium decreased, so were the levels of Rtc4 and Gis2 protein and with very similar dependence on zinc concentration. This correlation strongly suggested a mechanistic link regulating these two proteins.  As a first test of this hypothesis, we assessed whether Rtc4 and Gis2 protein levels were co-regulated in response to differences in zinc status. Using a plasmid expressing HA-tagged Rtc4 and myc-tagged Gis2, we found a very close correlation between the regulation of both proteins in response to zinc status, Figure 3. As zinc levels in the culture medium decreased, so were the levels of Rtc4 and Gis2 protein and with very similar dependence on zinc concentration. This correlation strongly suggested a mechanistic link regulating these two proteins. To test the hypothesis more directly, we used an rtc4∆::kanMX mutant allele in which the entire RTC4 ORF was replaced with the kanMX gene disruption cassette. This cassette contains a strong transcription terminator. Therefore, in addition to disrupting the RTC4 gene, we previously demonstrated that the inserted transcription terminator blocks RNA polymerase from extending the LLT RNA beyond the cassette and into the GIS2 promoter [8]. Thus, if repression of the GIS2 promoter is mediated by the LLT RNA, this allele is expected to result in constitutive GIS2 expression. Using a GIS2-GFP fusion allele, Northern blotting and immunoblotting indicated that the rtc4∆::kanMX allele resulted in constitutive GIS2 mRNA and Gis2 protein expression, Figure 4. This result supported the hypothesis that the LLT transcript represses the GIS2 promoter. A small decrease in Gis2-GFP protein was observed in zinc-deficient rtc4∆::kanMX cells perhaps because of some protein degradation.    To test the hypothesis more directly, we used an rtc4∆::kanMX mutant allele in which the entire RTC4 ORF was replaced with the kanMX gene disruption cassette. This cassette contains a strong transcription terminator. Therefore, in addition to disrupting the RTC4 gene, we previously demonstrated that the inserted transcription terminator blocks RNA polymerase from extending the LLT RNA beyond the cassette and into the GIS2 promoter [8]. Thus, if repression of the GIS2 promoter is mediated by the LLT RNA, this allele is expected to result in constitutive GIS2 expression. Using a GIS2-GFP fusion allele, Northern blotting and immunoblotting indicated that the rtc4∆::kanMX allele resulted in constitutive GIS2 mRNA and Gis2 protein expression, Figure 4. This result supported the hypothesis that the LLT transcript represses the GIS2 promoter. A small decrease in Gis2-GFP protein was observed in zinc-deficient rtc4∆::kanMX cells perhaps because of some protein degradation.  To test the hypothesis more directly, we used an rtc4∆::kanMX mutant allele in which the entire RTC4 ORF was replaced with the kanMX gene disruption cassette. This cassette contains a strong transcription terminator. Therefore, in addition to disrupting the RTC4 gene, we previously demonstrated that the inserted transcription terminator blocks RNA polymerase from extending the LLT RNA beyond the cassette and into the GIS2 promoter [8]. Thus, if repression of the GIS2 promoter is mediated by the LLT RNA, this allele is expected to result in constitutive GIS2 expression. Using a GIS2-GFP fusion allele, Northern blotting and immunoblotting indicated that the rtc4∆::kanMX allele resulted in constitutive GIS2 mRNA and Gis2 protein expression, Figure 4. This result supported the hypothesis that the LLT transcript represses the GIS2 promoter. A small decrease in Gis2-GFP protein was observed in zinc-deficient rtc4∆::kanMX cells perhaps because of some protein degradation.   While the rtc4∆::kanMX allele introduces a transcriptional terminator, it also deletes a large chromosomal region upstream of GIS2 gene. Thus, an alternative hypothesis to explain the results in Figure 4 was that this deletion/insertion disrupts elements of the GIS2 promoter located in or near the RTC4 ORF that are required for decreasing GIS2 expression in zinc-deficient cells. To address this hypothesis, we generated additional mutant constructs in plasmids encoding HA-tagged Rtc4 and myc-tagged Gis2. With the wild-type version of this plasmid (pRHGM), Rtc4-HA and Gis2-myc protein levels were down-regulated in zinc-deficient cells and this coincided with production of the LLT transcript and loss of RTC4 (SLT2) and GIS2 mRNA expression, Figure 5B,C. SLT1 was not detectable in this experiment due to its lower abundance. With a mutant allele in which the ZRE was mutated to a sequence not bound by Zap1 (pRHGM mZRE , Figure 5A), no LLT RNA was produced, RTC4 SLT2 and GIS2 mRNAs were constitutively expressed, and Rtc4 and Gis2 proteins were also constitutively expressed. Insertion of a strong transcriptional terminator just downstream of the RTC4 stop codon (pRHGM Ter ) had no effect on zinc-responsive Rtc4 regulation but Gis2 protein was constitutively produced. This allele produced a truncated LLT transcript detectable with the HA tag-specific NB probe. Finally, a plasmid containing the GIS2 promoter and 245 bp of additional sequence upstream from the RTC4 ORF (pGM) produced no Rtc4 protein and Gis2 was constitutively expressed. These results are all consistent with both RTC4 and GIS2 being regulated by the same bicistronic regulatory RNA. Two new transcripts were observed with the GIS2-only pGM plasmid, Figure 5C, asterisks, and their sizes suggested that they arose from transcription initiation in the plasmid vector upstream of the insert.

Probing the Functional Significance of GIS2 Regulation
The function of Gis2 is not well characterized. However, several studies have suggested a role for Gis2 in controlling the accumulation of specific proteins through mechanisms affecting either translation or transcription. Therefore, we carried out an analysis of the total proteome of yeast with altered expression of Gis2 because this approach would be likely to detect effects in either transcriptional or post-transcriptional mechanisms that alter protein abundance. In one experiment, we compared wild-type vs gis2∆ cells grown in zinc-replete conditions. Under these conditions, we hypothesized that the abundance of proteins that were influenced by Gis2 function either positively or negatively would be altered. Second, we compared wild-type cells transformed with a plasmid vector vs. wild-type cells transformed with a plasmid (pRG mZRE ) containing the RTC4-GIS2 chromosomal region that was mutated in its ZRE thereby resulting in constitutive RTC4 and GIS2 expression. These cells were grown under zinc-deficient conditions where Rtc4 and Gis2 are normally repressed. We hypothesized that the abundance of any other proteins affected positively or negatively by repression of RTC4 and GIS2 would be detected. This study used untagged alleles of RTC4 and GIS2 to avoid any disruption of function that may be caused by epitope tagging. Furthermore, we used a condition of more moderate zinc deficiency (LZM + 6 µM ZnCl 2 ) because under these conditions repression of Rtc4 and Gis2 is normally still strong, Figure 3, yet the level of zinc available for binding to Gis2 to confer function is higher than under more severe deficiency conditions. Deletion of GIS2 in replete cells resulted in decreased abundance of three proteins in addition to Gis2, i.e., Imd2, Smd3, and Fap1, Figure 6A. No proteins were observed to increase significantly in replete gis2∆ mutant cells. Overexpression of RTC4 and GIS2 in zinc-deficient cells caused an increase in only Gis2 and one other protein, Srb7; statistically significant changes were not detected for any other proteins, Figure 6B. Rtc4 was undetectable in these experiments because it is of very low abundance. Moreover, immunoblot analysis was performed with GFP-tagged alleles of four mRNA previously proposed to be direct targets of Gis2 translational control (BDH1, BUD2, FCY1, and THS1) [24]. These four proteins are down-regulated during zinc deficiency [26] but Gis2 overexpression in zinc-deficient cells had no effect on their protein abundance (data not shown). These results indicated that in our strains and under our growth conditions, Gis2 plays only a limited role in zinc-replete cells and loss of GIS2 regulation has little effect on the proteome of zinc-deficient cells. TAF10 was used as a loading control. The locations of the GIS2 and RTC4 SLT2 mRNAs are marked as well as the LLT RNA and truncated LLT RNA produced in the pRHGM Ter transformant. RTC4 SLT1 was not detectable in this experiment because of its low abundance. The asterisks mark two transcripts that arise from transcription initiation in the plasmid vector for pGM. The location of size markers is indicated on the left side of each panel in kDa (panel B) or kb (panel C).

Probing the Functional Significance of GIS2 Regulation
The function of Gis2 is not well characterized. However, several studies have suggested a role for Gis2 in controlling the accumulation of specific proteins through mechanisms affecting either translation or transcription. Therefore, we carried out an analysis of the total proteome of yeast with altered expression of Gis2 because this approach would be likely to detect effects in either transcriptional or post-transcriptional mechanisms that alter protein abundance. In one experiment, we compared wild-type vs gis2∆ cells grown in zinc-replete conditions. Under these conditions, we hypothesized that the abundance of proteins that were influenced by Gis2 function either positively or negatively would be altered. Second, we compared wild-type cells transformed with a plasmid vector vs. wild-type cells transformed with a plasmid (pRG mZRE ) containing the RTC4-GIS2 chromosomal region that was mutated in its ZRE thereby resulting in constitutive RTC4 and GIS2 expression. These cells were grown under zinc-deficient conditions where Rtc4 and Gis2 are normally repressed. We hypothesized that the abundance of any other proteins affected positively or negatively by repression of RTC4 and GIS2 would be detected. This study used untagged alleles of  Finally, using a sensitive co-culturing method, we examined the effect of gis2∆ mutation and constitutive RTC4/GIS2 expression on cell growth. Cells with altered Gis2 expression were mixed in approximately equal numbers with GFP-tagged control cells. The mixed cells were then inoculated into zinc-replete and deficient cultures and grown for 15 generations prior to measuring the abundance of GFP-tagged control cells and untagged cells by Fluorescence-activated cell sorting (FACS). In zinc-replete conditions, we confirmed previously published results [27,28] that mutation of GIS2 caused a small but detectable decrease in fitness, Table 1. In zinc-deficient cells, mutation of GIS2 had no detectable effect on growth rate. Cells mutant for the TSA1 gene were included as a positive control in this experiment; the growth defect of tsa1∆ mutants in zinc-replete conditions that is greatly exacerbated by deficiency observed previously [18] was confirmed here. Constitutive expression of RTC4 and GIS2 had no effect on zinc-replete growth but, surprisingly, conferred a small growth advantage in deficient cells. Thus, Gis2 is of some importance to the growth of cells under zinc-replete conditions but overexpression in zinc-deficient cells when it is normally repressed had no negative effects on growth.
abundance. Moreover, immunoblot analysis was performed with GFP-tagged alleles of four mRNA previously proposed to be direct targets of Gis2 translational control (BDH1, BUD2, FCY1, and THS1) [24]. These four proteins are down-regulated during zinc deficiency [26] but Gis2 overexpression in zinc-deficient cells had no effect on their protein abundance (data not shown). These results indicated that in our strains and under our growth conditions, Gis2 plays only a limited role in zinc-replete cells and loss of GIS2 regulation has little effect on the proteome of zinc-deficient cells. Statistically significant effects were defined as those proteins with 2-fold (log2 = 1) or greater changes and a q-value of < 0.05 after Benjamini-Hochberg correction. The full data from this analysis are provided in Table S1.  . The volcano plots show the data plotted as −log (q-value) vs. log2 fold change. Data are based on four biological replicates for each strain/growth condition. Statistically significant effects were defined as those proteins with 2-fold (log2 = 1) or greater changes and a q-value of < 0.05 after Benjamini-Hochberg correction. The full data from this analysis are provided in Table S1.

Discussion
In a recent report, we described an unusual mechanism in which the transcriptional activator Zap1 mediates repressed translation of RTC4 mRNA [8]. Under zinc-replete conditions, the SLT1 and SLT2 mRNA are produced from the RTC4 gene and are efficiently translated. During zinc deficiency, Zap1 activates expression of the longer LLT RNA. This regulatory RNA shuts off expression from the RTC4 promoter. In addition, the RTC4 ORF contained within the LLT RNA is not efficiently translated because of four small ORFs located upstream in the transcript leader. Thus, Rtc4 protein levels decrease despite a > 10-fold elevation in RNA levels. In this report, we show that the LLT regulatory RNA may have a second function, i.e., to shut off the GIS2 promoter. To our knowledge, this mechanism whereby a single bicistronic regulatory RNA represses two adjacent genes is unprecedented. Despite its novelty, however, similar mechanisms may function to regulate other genes in yeast as well in other organisms. Pelechano et al. discovered that 185 of 2727 (6.7%) adjacent gene pairs in yeast were expressed to at least some level as bicistronic RNAs [29]. Bicistronic RNAs have also been found in several viruses of eukaryotes, C. elegans, Drosophila, and vertebrates [4,30,31]. For many of these examples, the bicistronic RNAs have been either shown or proposed to coordinate protein expression through IRESs. IRESs are specialized RNA elements that allow for the cap-independent translation of internal ORFs in mRNA [32]. In contrast, the bicistronic LLT RNA of the RTC4/GIS2 region serves a repressive regulatory role and does not produce either Rtc4 or Gis2 protein to a detectable level.
Through what mechanism does the bicistronic LLT transcript shut off the GIS2 promoter? Possible mechanisms fall into either of two general categories, i.e., cis-acting and trans-acting mechanisms. In a trans-acting mechanism, the LLT transcript is produced and released from the DNA template after transcription and somehow inhibits the promoter, e.g., by binding to and interfering with transcription factor function. Our results argue against this general type of mechanism. As shown in Figure 5 where expression of the LLT transcript from the plasmid is disrupted (e.g., the mutant ZRE construct), the LLT transcript is still produced by the chromosomal gene yet it exerts no apparent negative impact on GIS2 expression from the plasmid. Possible cis-acting regulatory mechanisms include the physical interference with transcription factor binding to the GIS2 promoter by the movement of LLT-transcribing RNA polymerase II through that region of the DNA. Alternatively, it is well established that RNA polymerase II recruits chromatin-modifying proteins to transcribed regions that repress transcription from initiating within those regions [33]. This mechanism commonly inhibits the activity of cryptic promoters within genes and thereby prevents expression of truncated proteins. Either or both cis-acting mechanisms may be involved in repressing the RTC4 and GIS2 promoters.
Our studies of RTC4 and GIS2 strongly suggest that they are directly regulated by the zinc-responsive Zap1 transcription factor. The LLT transcript is dependent on Zap1, is constitutively produced in cells with a constitutively active Zap1 allele, and the consensus Zap1 binding site upstream of the RTC4 promoter is required for LLT expression [8], Figure 5. What is the purpose of this regulation in zinc-deficient cells? The fact that RTC4 and GIS2 are co-regulated by the LLT RNA suggests that they share a common function [34]. The function of the Rtc4 protein is unclear so it provides few clues to possible roles for Gis2. Rtc4 is found in both the cytosol and nucleus and genetic studies have linked it to telomere maintenance [13,35]. Studies of Gis2, however, have provided more specific hypotheses to consider. The GIS2 gene was first identified as a high copy suppressor of a strain unable to grow on galactose-containing media because of disrupted transcription of galactose metabolism genes [36]. The Gis2 protein contains seven CCHC-type zinc knuckle domains and related proteins bind to single-stranded DNA and to RNA. Gis2 immunoprecipitation experiments combined with microarray analysis indicated that Gis2 binds to stretches of G-A/U-A/U repeats in hundreds of mRNAs [24]. Based on a combined transcriptomics/proteomics analysis, it was concluded that Gis2 reduces the levels of many mRNA to which it binds. Many of these target mRNAs are involved in cytosolic translation. Gis2 was also found in stress-induced P-bodies and stress granules where it was proposed to inhibit translation of specific mRNA during stress conditions [37]. Finally, a recent study of Gis2 orthologs in Cryptococcus neoformans implicated their function in the post-transcriptional control of ribosomal proteins [38].
Gis2 is the yeast ortholog of mammalian CNBP/ZNF9 and these proteins are functionally conserved; expression of CNBP/ZNF9 in yeast gis2∆ mutants complemented some of its phenotypes [24]. Therefore, clues to Gis2's function may also come from the analysis of its mammalian relative. CNBP/ZNF9 is ubiquitously expressed in adult tissues and whole-animal knockout mutations in mice have been shown to be embryonic lethal [39]. In humans, expansion of a CCTG repeat in intron 1 of CNBP/ZNF9 causes myotonic dystrophy 1, a common type of muscular dystrophy [40]. At least some of the symptoms of this disease-causing repeat expansion are due to loss of CNBP/ZNF9 function [41]. Analysis of CNBP/ZNF9 indicated that it also binds to G-rich mRNA sequences that can form G-quadruplex structures [42]. These structures inhibit translation [43] and CNBP/ZNF9 binding prevented their formation and promoted translation of the affected mRNA. In addition to these effects, CNBP/ZNF9 has also been implicated as a DNA-binding transcriptional regulator [44]. Thus, prior studies of Gis2 and CNBP/ZNF9 suggested transcriptional or post-transcriptional roles for Gis2 in regulating the abundance of specific proteins.
In our proteomics analysis, Gis2 was implicated in affecting the expression of only three proteins in replete cells, Imd2, Smd3, and Fap1. Imd2 is involved in GTP synthesis, Smd3 is a spliceosomal subunit, and Fap1 is a transcription factor. This limited effect was surprising given the hypothesized role of Gis2 in regulating the translation of hundreds of mRNAs [24]. In addition, disrupting the mechanism that represses GIS2 expression in zinc-deficient cells had little detectable impact on the proteome in these cells; only Srb7, a component of the Mediator complex, was affected [45]. One caveat is that our proteomics analysis only detected about half of the total yeast proteome. Nonetheless, many of the proteins previously found to be impacted by Gis2 were detectable in our analysis. Disruption of the GIS2 gene in zinc-replete cells and constitutive expression of Gis2 protein in deficient cells had little impact on growth. Thus, we obtained no evidence indicating that repression of GIS2 serves any broad regulatory role under the specific conditions of our experiments. It remains possible that Gis2 regulates the abundance of many of its target proteins under other growth conditions and that regulation of Gis2 activity in response to zinc status more broadly impacts protein expression under those conditions. An alternative purpose of shutting off Gis2 expression in zinc-deficient cells is zinc sparing, i.e., the decreased expression of specific zinc-dependent proteins, as a means of reducing the cellular zinc requirement during deficiency [5]. Gis2 has been estimated to accumulate between 24,000 and 100,000 molecules per cell under replete conditions [46,47]. With seven zinc ions bound per monomer, this translates into 168,000-700,000 zinc ions required by Gis2 per cell or 2-8% of the total zinc requirement of the cell [48]. Thus, repression of Gis2 may be a mechanism that allows zinc-deficient cells to divert a limited supply of the metal to more important functions. The absence of a growth defect in zinc-deficient cells with constitutive Gis2 accumulation may then reflect the cumulative importance of multiple zinc-sparing mechanisms and the limited impact of disrupting any single such mechanism on the total zinc requirement [49].