Differing SAGA module requirements for NCR‐sensitive gene transcription in yeast

Nitrogen catabolite repression (NCR) is a means for yeast to adapt its transcriptome to changing nitrogen sources in its environment. In conditions of derepression (under poor nitrogen conditions, upon rapamycin treatment, or when glutamine production is inhibited), two transcriptional activators of the GATA family are recruited to NCR‐sensitive promoters and activate transcription of NCR‐sensitive genes. Earlier observations have involved the Spt‐Ada‐Gcn5 acetyltransferase (SAGA) chromatin remodeling complex in these transcriptional regulations. In this report, we provide an illustration of the varying NCR‐sensitive responses and question whether differing SAGA recruitment could explain this diversity of responses.

Interestingly, the SAGA CORE element Ada1 has been involved in the expression of many NCR-sensitive genes at different degrees according to the nitrogen source available and the gene considered (Soussi-Boudekou & Andre, 1999).
These observations suggest a role for SAGA that could connect the GATA factors to the variations in NCR-sensitive gene expression.In this report, we illustrate the diversity of NCRsensitive responses: an NCR-sensitive gene does not respond similarly to different stimuli, different NCR-sensitive genes do not respond similarly to the same stimuli, and GATA factor recruitment and requirement differ according to genes and stimuli.Here, we also show that SAGA recruitment does not escape the rule: it largely correlated with GATA factor recruitment, consistent with activation, confirming that chromatin remodeling, through SAGA function, plays a role in NCRsensitive gene derepression.

| Media and strain cultivation
Yeast cells were grown at 29°C to mid-log phase (A 660 nm = 0.5) in yeast nitrogen base (YNB) containing 2% glucose and 80 mM ammonium and sampled at the mid-exponential phase.Auxotrophic requirements were covered with 25 µg/mL of histidine, uracil, and/or tryptophan.Where indicated, exponentially growing cells were treated with final concentrations of 200 ng/mL Rap (LC Laboratories; dissolved in 10% Tween 20 + 90% ethanol) or 2 mM L-Msx (Sigma-Aldrich; freshly dissolved in water) for 30 min or transferred after filtration and rinsing to 0.1% (w/v) proline-containing YNB medium for 1 h.

| Strain construction
The strains used in this work are isogenic to S288C and are listed in Table 1.
Strain tagging constructions were carried out using the short flanking homology strategy (Longtine et al., 1998) and genes were deleted by the full replacement of open-reading frames by the kanMX or natMX cassettes constructed using the long flanking homology strategy (Wach, 1996).Primer sequences used for gene tagging and deletions are given in Table 2.

| Quantitative RT-PCR analysis
Quantitative reverse transcription-polymerase chain reaction (RT-PCR) was performed as described previously

Take-away
The expression of nitrogen catabolite repression (NCR)sensitive genes has been analyzed in Spt-Ada-Gcn5 acetyltransferase (SAGA) factor mutants in different nitrogen conditions.Varying NCR-sensitive responses have been noticed and differing SAGA recruitment could explain this diversity of responses.(Georis, Feller, Tate, et al., 2009;Georis, Feller, Vierendeels, et al., 2009).Total RNA was extracted from 4-mL cultures and complementary DNA (cDNA) was generated from 100 to 500 ng of total RNA using a Thermo Scientific™ RevertAid RT Reverse | 209 T A B L E 2 List of oligonucleotides used in this work for strain constructions.

| RESULTS
3.1 | Differences in GATA and Ada1 requirements for NCR-sensitive gene expression Before investigating the impact of SAGA on NCR-sensitive gene expression in chosen conditions, we have analyzed the expression of four deeply characterized NCR-sensitive genes (DAL5, GAP1, GDH2, and MEP2) in ammonia-grown cells and in response to three distinct NCR-derepressing treatments (the addition of Rap, L-Msx, or a transfer to proline as sole nitrogen source), some of them having been presented in our previous publications (Fayyadkazan et al., 2014;Georis et al., 2011).The expression of the four NCR-sensitive genes was analyzed in wild-type cells and in the absence of one or both the GATA transcriptional activators, Gln3 and Gat1 (Figure 1).Wild-type cells derepressed all four NCR-sensitive genes in response to the three stimuli but with varying degrees: GAP1 and MEP2 were induced at similar levels irrespective of the derepressing stimulus, whereas DAL5 was far less sensitive to Msx and GDH2 to Rap.All four genes required at least one of the GATA activators: their expression was fully abolished in the double gln3Δ gat1Δ mutant (Figure 1).Gat1 was largely dispensable for derepression when Gln3 was present, except when DAL5 expression was considered.In the absence of Gln3, however, Gat1 was required for GAP1 and MEP2 derepression.
Finally, responses elicited by an Msx treatment more often relied on Gln3 only (in Msx-treated cells, for GAP1, GDH2, and MEP2 expression), and deleting GAT1 even seemed to increase Msxinduced derepression, suggesting that it could play a negative role in these conditions.
We have characterized DAL5 and GDH2 expression in wildtype cells and cells lacking Ada1 (Figure 1).Single deletion of ADA1 decreased DAL5 expression to levels reached by the single GATA gene deletions in all three inducing conditions (Figure 1a).
GDH2 derepression was strongly impacted by the deletion of ADA1 reaching intensities comparable to those obtained with a GLN3 deletion (Figure 1b).Further, the influence of an ADA1 deleting was much less impacting on GAP1 and MEP2.From these observations, we can conclude that the involvement of Ada1 appears similar to Rap-induced DAL5 and Msx-induced GDH2 in the conditions assayed and is synergistic with the activating GATA factor.
From the abovementioned observations, we have selected two of these four NCR genes with one condition that allowed for elevated and Ada1-responsive gene transcription.Therefore, in the context of this report, yeast cells will be grown in a defined culture medium with ammonia as the sole nitrogen source.First, DAL5 derepression will be analyzed after treating exponentially growing cells with Rap, an inducing condition that requires both Gat1 and Gln3.Second, GDH2 expression will be assayed in response to an Msx treatment, a condition that requires only Gln3, and for which Gat1 could play a negative role.To ascertain the physiological relevance of our observations, proline shift experiments will be provided as supplemental data but, for clarity reasons, would not be described in the results section to allow a smoother reading of the main results of the manuscript.
When ADA1 deletion was assayed in combination with the deletion of GLN3 or GAT1 (Figure 2), any double deletion much more drastically impaired Rap-induced DAL5 expression (Figure 2a), whereas the effect of GDH2 expression upon Msx treatment occurred irrespective of the presence of Gat1 (Figure 2b).

| Differences in SAGA module requirements
To determine the involvement of each of the SAGA modules on NCR-sensitive gene expression, we have analyzed the impact of deleting the genes coding for some of their representative members: ADA1 and SPT7 for the CORE module, GCN5 and SGF79 for the HAT module, UBP8 and SGF73 for the DUB module, and SPT3 and SPT8 for the TBP module.DAL5 and GDH2 expression was monitored after Rap or Msx treatment, respectively, in each of the single mutants (Figure 3).As a control experiment, the expression of BDF2, a typical SAGAdependent gene (Bhaumik & Green, 2002), was analyzed in all mutants.BDF2 required the SAGA core and TBP module for maximal expression but did not require the HAT and DUB modules (Figure 3a,b).
When Rap-induced DAL5 expression was analyzed (Figure 3b), CORE module deletions had the strongest effect (two-to five-fold reduction compared to wild type), DUB and TBP deletions had a reduced impact (less than twofold), and HAT had no influence.
Combining SPT3 and SPT8 deletions had no additive effect on DAL5 expression compared to the respective single mutants.Deleting SAGA modules had no impact on Msx-induced GDH2 expression, apart from CORE module elements (Figure 3c).CORE complex, and hence recruitment of the whole SAGA complex, at NCR-sensitive promoters.Therefore, we assayed the recruitment of Ada1 in cells lacking Spt7 and vice versa (Figure 5).
In cells, lacking SPT7, Ada1 binding was not observed in response to Rap (Figure 5a) or Msx (Figure 5c).Similarly, in cells lacking ADA1, derepressed Spt7 binding was abolished at the DAL5 (Figure 5b) and GDH2 (Figure 5d) promoters.These results indicate a strong interdependence of CORE elements toward one another, suggesting that the SAGA CORE module must be recruited at NCR-sensitive promoters as a whole (similar conclusions for proline-shifted cells; Figure 5e,f).Interestingly, Spt7 binding levels were twice higher than those of Ada1.

| Promoter-specific requirements of CORE integrity for GATA factor recruitment
Further, in our analysis, we aimed at determining if SAGA integrity was required for GATA factor recruitment at NCR-sensitive promoters.Gat1 | 217 and Gln3 bindings were assayed in wild-type cells and in mutants lacking either Ada1 or the other GATA activator (Figure 6).
At the DAL5 promoter, Ada1 was required for Rap-induced Gat1 binding but not for Gln3 binding (Figure 6a,b).Also, as previously shown (Georis et al., 2008), Gat1 and Gln3 binding were strongly interdependent at the DAL5 promoter.At the GDH2 promoter, deleting ADA1 only slightly affected Gln3 and Gat1 binding (Figure 6c,d).Further, Gln3 binding was very robust: it did not require Gat1.On the contrary, Gat1 binding was largely Gln3dependent.In proline-grown cells, binding of both GATA factors at the DAL5 promoter required a functional Ada1 and the other GATA activator (Figure 6e,f), and the same was true for Gat1 binding to the GDH2 promoter (Figure 6g), whereas, again, Gln3 binding to GDH2 was largely immune to the absence of Gat1 and Ada1 (Figure 6h).

| DISCUSSION
This paper provides a thorough illustration of the gene, GATA activator, and inducer specificities among NCR responses.Earlier reports have given some possible explanations for these observations.First, the promoter architecture, the number, and the orientation of GATA sites likely influence the way how an NCR-sensitive gene responds to a particular stimulus (Bysani et al., 1991;Cooper et al., 1989;Rai et al., 1989).It has been recognized that, in some cases, auxiliary promoter sequences were shown to contribute to transcriptional activation (Miller & Magasanik, 1991), although this was never as efficient as additional GATA sites (Minehart & Magasanik, 1992).Second, we cannot neglect the impact of the two other GATA factors on NCR-sensitive transcription, the Dal80 and Gzf3 repressors (Coornaert et al., 1992;Cunningham et al., 1991;Rowen et al., 1997;Soussi-Boudekou et al., 1997).Under limiting nitrogen conditions, DAL80 is highly induced (Cunningham et al., 1991), and Gzf3 has been detected at NCR-sensitive promoters (Georis, Feller, Tate, et al., 2009;Georis, Feller, Vierendeels, et al., 2009).Given the elevated sequence conservation among the four yeast GATA factors, with 100% conservation of the residues involved in contacts with DNA (Ravagnani et al., 1997), it is very likely that the repressors do bind the exact same DNA sequences in vivo and may interfere with the GATA activators, in a gene-specific manner.Third, we have previously shown that Gln3 and Gat1 localizations are controlled by two different regulatory pathways, with Gln3 localization predominantly responding to intracellular nitrogen levels, and Gat1 localization mainly responding to Rap (Tate et al., 2010), suggesting that the different stimuli do not borrow the same signalization cascades.Differences in GATA activator requirements for the transcriptional responses to different activation signals can therefore be anticipated.
This paper also provides a characterization of the role of the SAGA complex controlling DAL5 and GDH2 expression, together with the GATA activators (see Figure 7 and comments therein).In wild-type cells, elevated binding of both GATA activators and SAGA ensure maximal activation of DAL5 upon Rap treatment or proline shift.The HAT module is dispensable here.In the absence of Ada1, DAL5 expression is strongly reduced despite the presence of Gln3 at the promoter, probably because SAGA is not recruited in this CORE mutant.Interestingly, the absence of SAGA prevents Gat1 recruitment at the DAL5 promoter.Low expression of DAL5 is also observed in cells lacking Gat1 or Gln3, in which recruitment of the other GATA activator to the DAL5 promoter is largely impaired, as well as the SAGA complex.DAL5 activation is nevertheless significant in all three single mutants, although this could be attributed to the binding of the GATA activators and/or SAGA subunits being so weak that it escaped our detection.Since DAL5 expression is abolished in the double gln3Δ ada1Δ and gat1Δ ada1Δ mutants, we propose that SAGA could be recruited even though we could hardly detect it in the gln3Δ and gat1Δ single mutants.Since Gln3 binding can occur without SAGA, which is not the case for Gat1 binding, we propose that SAGA could be a better coactivator for Gat1-dependent transcriptional activation.Accordingly, Gln3 could be corecruited with Gat1, leading to SAGA recruitment, and allowing chromatin remodeling, thereby increasing Gat1 binding and subsequent transcription.Interdependence for binding suggests the existence of synergistic interferences between the different partners at the DAL5 promoter upon Rap treatment (Figure 7).
As far as GDH2 is concerned, the conclusions are clearly different.
GDH2 expression is maximal in wild-type cells treated with Msx or shifted to proline, consistent with maximal GATA factor and SAGA subunit binding.Interestingly, this expression does not require Gat1, even though the activator is detected at the GDH2 promoter at elevated levels.
Consistently, Gln3 and SAGA are still bound at the GDH2 promoter in gat1Δ cells, ensuring elevated GDH2 transcription levels.The activation levels were even higher in gat1Δ mutant cells, and this could be due to Gat1 being negatively interfering with Gln3 transcription activation, more potent.Alternatively, increased GDH2 expression in gat1Δ cells could be explained by the decreased presence of the Gat1-dependent Dal80negative regulator.In the absence of Gln3, neither Gat1 nor SAGA could be recruited at the GDH2 promoter, consistent with the lack of activation in gln3Δ mutant cells.Finally, in ada1Δ cells, binding of the GATA factors was only weakly altered, with much lower activation in the mutant cells, F I G U R E 6 SAGA-GATA (Spt-Ada-Gcn5 acetyltransferase-GATA) interdependence for recruitment at nitrogen catabolite repressionsensitive promoters.Cells of FV034 (wt, Gat1-Myc 13 ), FV547 (ada1Δ, Gat1-Myc 13 ), 03740c (gln3Δ, Gat1-Myc 13 ), FV036 (wt, Gln3-Myc 13 ), FV548 (ada1Δ, Gln3-Myc 13 ), and FV041 (gat1Δ, Gln3-Myc 13 ) were grown in YNB media containing ammonium as the sole nitrogen source.Rapamycin (Rap) or L-methionine sulfoximine (Msx) were added, or cells were shifted to a proline-containing medium for 1 h (Pro), where indicated.Chromatin immunoprecipitation (ChIP) of Gat1-Myc 13 (a, c, e, g) and Gln3-Myc 13 (b, d, f, h) was performed as described in the Materials and Methods section.Quantitative polymerase chain reaction of IP and IN fractions was performed with primers specific for the DAL5 (a, b, e, f) and GDH2 (c, d, g, h) promoters.Histograms represent the averages of two immunoprecipitations performed on a minimum of two experiments from independent cultures (n ≥ 2) and error bars delineate a 95% confidence interval.*Statistically significant difference between each mutant and the wild type as determined with the unpaired t-test.*p < 0.05; **p < 0.01.
suggesting that SAGA is not mainly needed for GATA factor recruitment but most strongly for transcription activation.In sum, Pro-or Msxinduced Gln3 binding at the GDH2 promoter can occur on its own, allowing for concomitant Gat1 and SAGA recruitment.Gat1 recruitment is ineffective for GDH2 activation, whereas that of SAGA is crucial (Figure 7).
In this work, we have demonstrated an absolute respective interdependence for Rap-induced GATA activator binding at the DAL5 promoter.This has not always been the case.Our previous reports have shown that Rap-induced Gat1 binding at the DAL5 promoter could occur in the absence of Gln3, in both TB (Georis et al., 2008) and FY (Georis, Feller, Tate, et al., 2009;Georis, Feller, Vierendeels, et al., 2009) genetic backgrounds.In these experiments, however, cells were grown with glutamine as the nitrogen source, whereas they were grown with ammonia in the present work.Such a difference is striking and suggests that much more complexity is to be expected in the context of nitrogen source-specific regulations of NCR-sensitive genes.
Our work has demonstrated that SAGA subunits are recruited to NCR-sensitive promoters upon derepression.The SAGA CORE module is structurally made of a histone octamer-like fold and an adjacent submodule (Taf5, Taf6, and Spt20; Wang et al., 2020).The octamer-like domain is made of three pairs of subunits (Taf6-Taf9, Taf10-Spt7, and Taf12-Ada1), each pair contributing to two histone folds, plus Spt3, which brings two more histone folds on its own (Green, 2020).In our experiments, the strongest phenotypes with respect to NCR-sensitive transcription in SAGA mutants have been Altogether, these observations suggest that regulating the expression of NCR-sensitive genes by coactivators has gene and nitrogen source specificities.We can also hypothesize that each of the catalytic activities of the SAGA complex could play a complementary role in the transcriptional activation of the NCR-sensitive genes, which could explain that only the deletion of the structural subunit of the SAGA complex generates a strong effect on their expression.
Our results illustrate the complexity and interactions taking place at NCR-sensitive promoters, enabling fine-tuned responses to environmental changes.Further work will be required to decipher F I G U R E 7 Model for GATA-and Spt-Ada-Gcn5 acetyltransferase (SAGA)-dependent expression of DAL5 and GDH2.
In ammonia-grown, wild-type cells, no GATA factor is associated with promoters nor SAGA elements.Gene expression is low.Upon rapamycin treatment, both Gat1 and Gln3 plus SAGA are recruited to the DAL5 promoter and this recruitment is interdependent for all three components (arrows).DAL5 expression is high and maximal expression requires both the GATA and SAGA integrity.Upon Msx treatment, both Gat1 and Gln3 plus SAGA are recruited to the GDH2 promoter, but here Gln3 binding can occur independently of the presence of Gat1 and SAGA.On the other hand, SAGA and Gat1 binding relies on Gln3 presence.Gat1 recruitment does not impact transcription, whereas the presence of SAGA, and more importantly Gln3, is needed for full derepression of GDH2.In prolineshifted cells, the situation at the DAL5 promoter is similar to rapamycin-treated cells, except for SAGA binding that does not require Gln3.GDH2 expression in proline-shifted cells has the same factor requirements, although Gat1 recruitment here is more stringent and requires, in addition to Gln3, SAGA as well.

3. 3 |
Similar recruitment of the CORE and TBP modules at NCR-sensitive promotersAiming at determining if the expression consequences of deleting SAGA modules are explained by differences in recruitment at the respective promoters, we assayed the recruitment of Ada1 (CORE) and Spt8 (TBP) in wild type and in cells lacking one of the GATA activators (Figure4).At the DAL5 promoter (Figure 4a,b), the binding of Ada1 and Spt8 was similar: strongly induced by Rap and severely affected in the single GATA mutants.At the GDH2 promoter (Figure 4c,d), Ada1 F I G U R E 1 Differences in GATA and Ada1 requirements and sensitivity to different inducers for nitrogen catabolite repression-sensitive gene expression.Cells of 25T0b (wt), FV022 (gln3Δ), FV023 (gat1Δ), FV531 (ada1Δ), and FV024 (gln3Δ gat1Δ) were grown in yeast nitrogen base (YNB) media containing ammonium as sole nitrogen source.Rapamycin (Rap) or L-methionine sulfoximine (Msx) were added, or cells were shifted to a proline-containing medium (Pro) for 1 h, where indicated.Total RNA was isolated and TBP1-normalized DAL5 (a), GAP1 (b), GDH2 (c), and MEP2 (d) messenger RNA (mRNA) levels were quantified by quantitative reverse transcription-polymerase chain reaction.Values represent the averages of at least two experiments from independent cultures (n ≥ 3; n = 1 for FV024) and error bars delineate a 95% confidence interval.*Statistically significant difference between each mutant and the wild type as determined with the unpaired t-test.*p < 0.05; **p < 0.01; ***p < 0.001.and Spt8 binding was also alike: strongly induced by Msx, but only affected by a GLN3 deletion.The requirement of the GATA activators for Ada1 binding depended on the gene and the nitrogen source considered (Figure 4e,f).It nicely paralleled the expression levels, correlating with the dependence of the expression of DAL5 and GDH2 toward the GATA activators.

3. 4 |
Integrity of the CORE is required to allow the remaining CORE recruitment Given the importance of the impact of deleting one CORE component on NCR-sensitive gene expression, we wanted to determine if deleting one component of the CORE could compromise the integrity of the remaining F I G U R E 2 Differences in Ada1 requirements for catabolite repression (nitrogen catabolite repression-sensitive gene expression).Cells of 25T0b (wt), FV022 (gln3Δ), FV023 (gat1Δ), FV531 (ada1Δ), FV645 (gln3Δ ada1Δ), and FV646 (gat1Δ ada1Δ) were grown in yeast nitrogen base (YNB) media containing ammonium as sole nitrogen source.Rapamycin (Rap) or L-methionine sulfoximine (Msx) were added, where indicated.Total RNA was isolated and TBP1-normalized DAL5 (a) and GDH2 (b) messenger RNA (mRNA) levels were quantified by quantitative reverse transcription-polymerase chain reaction (qRT-PCR).Values represent the averages of at least two experiments from independent cultures (n ≥ 3) and error bars delineate a 95% confidence interval.*Statistically significant difference between each mutant and the wild type as determined with the unpaired t-test.***p < 0.001.GEORIS ET AL. | 213

F
I G U R E 5 CORE integrity is required for its recruitment at nitrogen catabolite repression-sensitive promoters.Cells of FV530 (wt, Ada1-HA 3 ), FV688 (spt7Δ, Ada1-HA 3 ), FV1256 (wt, Spt7-HA 3 ), and FV1257 (ada1Δ, Spt7-HA 3 ) were grown in yeast nitrogen base (YNB) media containing ammonium as the sole nitrogen source.Rapamycin (Rap) or L-methionine sulfoximine (Msx) were added, or cells were shifted to a proline-containing medium for 1 h (Pro), where indicated.Chromatin immunoprecipitation (ChIP) of Ada1-HA 3 (a, c, e, f) and Spt7-HA 3 (b, d) was performed as described in the Materials and methods section.Quantitative polymerase chain reaction of IP and IN fractions was performed with primers specific for the DAL5 (a, b, e) and GDH2 (c, d, f) promoters.Histograms represent the averages of two immunoprecipitations performed on minimum two experiments from independent cultures (n ≥ 2) and error bars delineate a 95% confidence interval.*Statistically significant difference between each mutant and the wild type as determined with the unpaired t-test.*p < 0.05; **p < 0.01; ***p < 0.001.F I G U R E 6 (See caption on next page) GEORIS ET AL.
observed in the ada1Δ and spt7Δ mutant strains.This observation is consistent with the fact that Ada1 and Spt7 belong to the CORE of SAGA and their absence has previously been shown to fully impair SAGA complex integrity(Nuño-Cabanes et al., 2020;Sterner et al., 1999;Wu & Winston, 2002).Consistently, Ada1 and Spt7 recruitment at NCR-sensitive genes were fully interdependent.In addition, our observation that Spt7 ChIP levels were systematically twice higher than those of Ada1 could indicate better accessibility of the former.Indeed, from the published structure of SAGA, Ada1 could be less accessible, interfacing the CORE module with the large Tra1 subunit and the DUB module, whereas Spt7's position appears more peripheral(Green, 2020).Interestingly, the HAT module was fully dispensable for NCRsensitive gene transcription, whereas the DUB module was only mildly required, for Rap-induced DAL5 expression only.This could indicate that SAGA modules fulfill complementary activities at NCR-sensitive promoters and altering only one of these activities may not alter its capacity to mediate transcriptional activation.More surprisingly, the reduced impact, on NCR-sensitive transcription, of the joint deletion of SPT3 and SPT8, supposedly altering TBP recruitment, suggests that other coactivator complexes may help its recruitment, or that GATA activators can perform this function themselves.