The function of Spt3, a subunit of the SAGA complex, in PGK1 transcription is restored only partially when reintroduced by plasmid into taf1 spt3 double mutant yeast strains

In Saccharomyces cerevisiae , class II gene promoters contain two classes of TATA elements: the TATA box and the TATA-like element. Functional loss of TFIID and SAGA transcription complexes selectively impacts steady-state mRNA levels expressed from TATA-like element-containing (i.e., TATA-less) and TATA box-containing promoters, respectively. While nascent RNA analysis has revealed that TFIID and SAGA are required for both types of promoters, the division of their roles remains unclear. We show here that transcription from the PGK1 promoter decreased in some cases by more than half after disruption of the TATA box or SPT3 ( spt3 Δ ), whereas spt3 Δ did not affect transcription from the TATA-less promoter, consistent with the prevailing view that Spt3 functions speciﬁcally in a TATA box-dependent manner. Transcription from this promoter was abolished in the spt3 Δ taf1-N568 Δ strain but unaffected in the taf1-N568 Δ strain, regardless of TATA box presence, suggesting that TFIID was functionally dispensable for PGK1 transcription at least in the SPT3 strain. Furthermore, transcription from the TATA box-containing PGK1 promoter was slightly reduced in the taf1 strain lacking TAND ( taf1- Δ TAND ) upon temperature shift from 25 to 37 ° C, with resto-ration to normal levels within 2 h, in an Spt3-dependent manner. Interestingly, when SPT3 was reintroduced into the spt3 Δ TAF1 , spt3 Δ taf1-N568 Δ or spt3 Δ taf1-Δ TAND strain, TATA box-dependent transcription from this promoter was largely restored, but TFIID independence in transcription was not restored, especially from the TATA-less promoter, and transient TAND/Spt3-dependent ﬂuctuations of transcription after the temperature shift were also not recapitulated. Collectively, these observations suggest that Spt3 has multiple functions in PGK1 transcription, some of which may be intimately connected to Taf1 function and may therefore become unrestorable when the TFIID and SAGA functions are simultaneously compromised.

In Saccharomyces cerevisiae, class II gene promoters contain two classes of TATA elements: the TATA box and the TATA-like element. Functional loss of TFIID and SAGA transcription complexes selectively impacts steady-state mRNA levels expressed from TATA-like element-containing (i.e., TATA-less) and TATA box-containing promoters, respectively. While nascent RNA analysis has revealed that TFIID and SAGA are required for both types of promoters, the division of their roles remains unclear. We show here that transcription from the PGK1 promoter decreased in some cases by more than half after disruption of the TATA box or SPT3 (spt3Δ), whereas spt3Δ did not affect transcription from the TATA-less promoter, consistent with the prevailing view that Spt3 functions specifically in a TATA box-dependent manner. Transcription from this promoter was abolished in the spt3Δ taf1-N568Δ strain but unaffected in the taf1-N568Δ strain, regardless of TATA box presence, suggesting that TFIID was functionally dispensable for PGK1 transcription at least in the SPT3 strain. Furthermore, transcription from the TATA box-containing PGK1 promoter was slightly reduced in the taf1 strain lacking TAND (taf1-ΔTAND) upon temperature shift from 25 to 37 °C, with restoration to normal levels within 2 h, in an Spt3-dependent manner. Interestingly, when SPT3 was reintroduced into the spt3Δ TAF1, spt3Δ taf1-N568Δ or spt3Δ taf1-ΔTAND strain, TATA box-dependent transcription from this promoter was largely restored, but TFIID independence in transcription was not restored, especially from the TATA-less promoter, and transient TAND/Spt3-dependent fluctuations of transcription after the temperature shift were also not recapitulated. Collectively, these observations suggest that Spt3 has multiple functions in PGK1 transcription, some of which may be intimately connected to Taf1 function and may therefore become unrestorable when the TFIID and SAGA functions are simultaneously compromised.

INTRODUCTION
During eukaryotic transcription of class II genes, a set of general transcription factors (GTFs; TFIIA, B, D, E, F and H) assemble at the core promoter region alongside Mediator and RNA polymerase II (Pol II) to form a pre-initiation complex (PIC) that directs accurate transcriptional initiation (Thomas and Chiang, 2006;Hahn and Young, 2011;Sainsbury et al., 2015). There are two types of core promoter in budding yeast. These contain either the TATA box or a TATA-like element (a 1-or 2-bp-mismatched version of the TATA box), the latter of which has been termed the "TATA-less" core promoter (Rhee and Pugh, 2012). TFIID is a large multiprotein complex composed of TATA-binding protein (TBP) and 14 TBP-associated factors (Tafs) (Thomas and Chiang, 2006). TFIID binds to the TATA box or TATA-like element as a first step during PIC assembly and then recruits other factors to the core promoter (Sainsbury et al., 2015;Gupta et al., 2016;Nogales et al., 2017;Bhuiyan and Timmers, 2019;Fischer et al., 2019). In previous studies, it was proposed that the Taf1 N-terminal domain (TAND) of TFIID plays an important role in TBP binding to the TATA box or TATA-like element (Kotani et al., 2000;Cheng et al., 2002b;Chitikila et al., 2002;Huisinga and Pugh, 2007;Patel et al., 2018;Kasahara et al., 2019), but the detailed mechanisms remained unelucidated (Gupta et al., 2017;Bhuiyan and Timmers, 2019;Patel et al., 2019;Wei et al., 2019).
In general, transcription of stress-inducible genes is from TATA box-containing promoters and is dominated by SAGA, while transcription of housekeeping genes is from TATA-less promoters and is dominated by TFIID (Huisinga and Pugh, 2004). This suggests that TFIID and SAGA play distinct roles in transcription according to the presence of a TATA box in the core promoter (Huisinga and Pugh, 2004;Bhaumik, 2011;Rhee and Pugh, 2012;de Jonge et al., 2017). However, more recent studies have demonstrated that TFIID and SAGA are both involved in the transcription of nearly all class II genes Warfield et al., 2017;Fischer et al., 2019), although the effects of their deficiencies on steady-state mRNA levels are obscured by transcript buffering phenomena (Timmers and Tora, 2018).
A recent in vitro study showed that while transcription from both TATA box-and TATA-like element-containing (i.e., TATA-less) promoters was impaired by degronmediated depletion of Taf1 from cell extracts, only the former could be restored by addition of recombinant TBP (Donczew and Hahn, 2017). Furthermore, transcription from TATA-less promoters could not be restored by addition of TBP, even when their TATA-like elements were substituted with TATA boxes (Donczew and Hahn, 2017), indicating that the mechanisms of transcription from these two types of promoters are fundamentally different. These observations also suggest that TATAless promoters may contain as-yet-uncharacterized core promoter sequence(s) that specifically require TFIID function to promote transcription and/or prevent the promotive function of free TBP or SAGA in the absence of Taf1/ TFIID. In higher eukaryotes, there are several wellcharacterized core promoter sequences other than the TATA box, such as Inr (Initiator), DPE (downstream core promoter element), MTE (motif ten element), and DCE (downstream core element), all of which are recognized by TFIID (Zabidi and Stark, 2016;Nogales et al., 2017;Vo Ngoc et al., 2017Patel et al., 2019). Although similar sequences recognizable by TFIID remain unidentified in budding yeast (Hahn and Young, 2011;Lubliner et al., 2015), it was recently shown that yeast TFIID could bind to downstream DNA upon transcriptional initiation, especially on TFIID-dominated promoters, to facilitate re-initiation in an activator-independent manner (Joo et al., 2017). Such interaction of TFIID with downstream DNA, which presumably cannot be mediated by free TBP or SAGA, may play a more important role in transcription from TATA-less promoters than in transcription from TATA-containing promoters.
Consistent with recent observations that SAGA acts as a general cofactor for both TATA box-containing and TATA-less promoters Fischer et al., 2019), we previously showed that SAGA could facilitate transcription from the TATA-containing AGP1 promoter in a Taf1-independent manner, even when its TATA box was substituted with TATA-like elements (Watanabe and Kokubo, 2017). Furthermore, chimeric promoter mapping analyses using the Taf1-dependent CYC1 and Taf1-independent AGP1 promoters demonstrated that while Taf1 dependence or independence was conferred by the upstream activating sequence (UAS), Spt3/SAGA dependence of the latter promoter was conferred by certain specific core promoter sequence(s) other than the TATA box or TATA-like element (Watanabe and Kokubo, 2017). These observations suggested that, like TFIID, SAGA could mediate transcription in a manner dependent on or independent of the TATA box and/or TATA-like element.
Previous research showed that addition of a TATA box abrogated or much reduced the Taf1 dependence of transcription from some promoters (Tsukihashi et al., 2000(Tsukihashi et al., , 2001Cheng et al., 2002a). Such observations clearly indicate that the mechanisms underlying TFIID-mediated transcription from TATA box-containing or TATAless promoters are qualitatively different, at least in the requirement for a particular Taf1 function. To date, no SAGA mutants have been identified that affect transcription differently according to the presence or absence of the TATA box. Thus, it remains unknown whether SAGA can mediate transcription from both types of pro-moter via the same or different mechanisms.
In this study, we found that introduction of plasmidborne SPT3 into spt3Δ or spt3Δ taf1 mutant strains largely rescued TATA box-dependent transcription at the PGK1 promoter, but did not restore TFIID independence in transcription, particularly from a TATA-less PGK1 promoter. These observations indicate that Spt3/ SAGA has a range of distinct functions, only some of which can be restored by reintroduction of SPT3, and that, like TFIID, Spt3/SAGA mediates transcription from TATA box-containing and TATA-less promoters by different mechanisms. Furthermore, we also found that TFIID independence of PGK1 transcription could not be altered simply by changing the core promoter sequences, even though the promoter had the potential to be transcribed in a TFIID-dependent manner. Our observations lead us to propose that some functions of Spt3/SAGA are so closely connected with those of Taf1/TFIID that functional rescue is not possible when the two factors are compromised simultaneously.

MATERIALS AND METHODS
Yeast strains Standard techniques were used for yeast growth and transformation (Amberg et al., 2005). Yeast strains used in this study are listed in Supplementary  Table S1. Oligonucleotide sequences used for strain construction are listed in Supplementary Table S2. All strains used in this study were generated from BY4741 or BY4742, as described below.
YTK11406 and YTK11411, both of which carry a deletion of the chromosomal TAF1 coding region and the wild type TAF1 gene in a URA3-based low copy vector (pYN1), were described previously (Ohyama et al., 2010).
Northern analysis Northern analysis was performed as described previously (Tsukihashi et al., 2000). Briefly, total RNA (10-20 μg) was isolated from the indicated strains grown in synthetic complete (SC) medium to logphase at 25 °C or after continued incubation at 37 °C for the indicated time periods. Total RNA was subjected to electrophoresis, blotted onto a nylon membrane, and hybridized with gene-specific probes.
For detection of PGK1, VTC1 and SCR1, DNA fragments were amplified by PCR from genomic DNA of BY4741, purified, and 32 P-labeled by random priming with the Klenow fragment. The PCR primer pairs used were as follows: PGK1, TK1224 and TK1225; VTC1, TK9030 and TK9013; and SCR1, TK9507 and TK10081.
Primer extension analysis Primer extension analysis was performed as described previously (Watanabe et al., 2015). The primer used was TK8251 ( + 40 to + 60 of VTC1). The cDNA products were analyzed on a 6% polyacrylamide DNA sequencing gel. Gels were exposed to imaging plates for visualization (Typhoon FLA 7000, GE Healthcare Life Sciences), scanning, and quantification of electrophoretic images (ImageQuant TL software version 8.1, GE Healthcare Life Sciences).

RESULTS
TFIID dependence of PGK1 transcription is reliant upon the chromosomal or plasmid location of SPT3 The PGK1 promoter is one of the most extensively characterized promoters in S. cerevisiae (Chambers et al., 1989;Henry et al., 1994;Ogden et al., 1986;Rathjen and Mellor, 1990). Although the PGK1 promoter contains a TATA box (TATATATAAA) located at -153 to -144, and a TATA-like element (TACATATT) located at -116 to -109, their roles in transcription remain unclear (Ogden et al., 1986;Rathjen and Mellor, 1990). Previously, PGK1 transcription was decreased by inactivation of SAGA but not TFIID (Li et al., 2000;Huisinga and Pugh, 2004;Layer et al., 2010). Nevertheless, TFIID was detected at the PGK1 promoter at 25 °C (Venters et al., 2011), and occupancy increased after heat shock (Kuras et al., 2000), suggesting that TFIID might yet be involved in PGK1 transcription. On this basis, we reasoned that the kinetics of PGK1 transcription after heat shock might be affected by loss of TFIID function.
During the preliminary stage of this work, we observed that the effects of taf1 mutations on PGK1 transcription differed between two distinct types of taf1 strain: those carrying SPT3 at the native locus and those with chromosomal spt3Δ complemented with plasmid-borne SPT3. We therefore sought to examine the detailed kinetics of PGK1 transcription after temperature shift from 25 °C to 37 °C, a procedure which inactivates TFIID function in the temperature-sensitive taf1-ΔTAND and taf1-N568Δ strains carrying SPT3 either on the chromosome or on the plasmid. In addition, a VTC1 reporter system carried by the chromosome (Watanabe et al., 2015) was used to test the importance of the TATA box for PGK1 transcription (Supplementary Fig. S1).
Levels of VTC1 mRNA gradually decreased after the temperature shift even in the wild-type strain ([m] in Fig. 1A, 1C), indicating that this reporter was more heat-sensitive than the endogenous system, for unknown reasons. Accordingly, the transient increase of PGK1 transcription in the taf1-N568Δ strain became unobservable ([u] in Fig. 1A, 1C), but its transient decrease in the taf1-ΔTAND strain was more apparent ([q] in Fig. 1A,  1C). Disruption of the TATA box by substitution with the GAGA sequence (see the Fig. 3A Fig. 1A, 1C), indicating that Spt3 functions in a TATA box-dependent manner. By contrast, while TATA disruption did not directly affect the fluctuating PGK1 transcription profile (transient Spt3-dependent decrease and subsequent recovery) after the temperature shift in the taf1-ΔTAND strain ([q, s] in Fig. 1A, 1C), substantial further decreases in mRNA levels were observed in the spt3Δ taf1-ΔTAND strain (compare [r] and [t] in Fig. 1A, 1C), suggesting that the taf1-ΔTAND mutation rendered the Spt3 function less dependent on the TATA box.
Next, similar experiments were conducted using strains carrying spt3Δ on the chromosome and SPT3 on a highcopy plasmid (spt3Δ + SPT3), or those containing an empty vector (spt3Δ + empty) (Fig. 1B, 1D). Transient fluctuations in mRNA level after the temperature shift were not observed with PGK1 ([e, g] in Fig. 1B, 1D), and were much diminished with VTC1 ([q, s] in Fig. 1B, 1D) in the taf1-ΔTAND strain, indicating that Spt3 expressed from the plasmid is not fully functional. Consistent with this, PGK1 mRNA levels transiently increased but then decreased more rapidly during the later phase of the 2-h period following the temperature shift in the taf1-N568Δ strain ([i, k] in Fig. 1B, 1D), suggesting that Spt3 expressed from the plasmid could support initial increases in PGK1 transcription but could not support transcription during the later phase in this particular taf1 strain. This was supported by transcription results for VTC1 mRNA, particularly when the TATA box of the PGK1 promoter was disrupted ([w] in Fig. 1B, 1D). By contrast, a stimulatory function of Spt3 for TATA box-dependent transcription was observed even when expressed from the plasmid ([m, n] in Fig. 1B, 1D). Notably, spt3Δ strains carrying SPT3 on a low-copy plasmid yielded similar results to spt3Δ strains carrying SPT3 on a high-copy plasmid (Supplementary Fig. S2), suggesting that such partial loss of Spt3 function was not due to differences in Spt3 expression levels, which varied over a wide range ( Supplementary  Fig. S3). Taken together, these results indicate that Spt3 performs multiple functions in the mediation of PGK1 transcription, only some of which could be restored when Spt3 was reintroduced into spt3Δ taf1 strains by plasmid expression.
The effects of plasmid-borne SPT3 on the growth of spt3Δ taf1-ΔTAND and spt3Δ taf1-N568Δ strains In a previous study, taf1-ts2 and spt3Δ mutants were found to display mild synthetic growth defects when cultured at    (m, q, u, n, r, v, o, s, w, p, t, x in the middle panels), PGK1 (a, e, i, b, f, j, c, g, k, d, h, l in the top panels) or SCR1 (control; lanes 1-28 in the bottom panels) in 12 strains carrying the VTC1 reporter driven by the PGK1 promoter in which the TATA box (TATA) was intact or substituted with the GAGA sequence (GAGA) (see the Fig. 3A legend), as indicated in (C). In these strains, chromosomal TAF1 was deleted and then substituted with TAF1 (m, n, o, p), taf1-ΔTAND (q, r, s, t) or taf1-N568Δ (u, v, w, x) on a plasmid, as indicated in (C). Chromosomal SPT3 was intact (m, o, q, s, u, w) or deleted (n, p, r, t, v, x), as indicated in (C). These strains were grown at 25 °C (0 min) and then further incubated at 37 °C for the time period indicated above the blot image (15, 30, 45, 60, 90, 120 min) in synthetic medium containing 2% glucose. (B) Northern analysis of expression of VTC1 (m, q, u, n, r, v, o, s, w, p, t, x in the middle panels), PGK1 (a, e, i, b, f, j, c, g, k, d, h, l in the top panels), or SCR1 (control; lanes 1-28 in the bottom panels) in 12 strains carrying the VTC1 reporter driven by the PGK1 promoter in which the TATA box (TATA) was intact or substituted with the GAGA sequence (GAGA), as indicated in (D). In these strains, chromosomal TAF1 was deleted and then substituted with TAF1 (m, n, o, p), taf1-ΔTAND (q, r, s, t) or taf1-N568Δ (u, v, w, x) on the plasmid, as indicated in (D). Chromosomal SPT3 was deleted and strains were transformed with either SPT3-containing (m, o, q, s, u, w) or empty (n, p, r, t, v, x) plasmid, as indicated in (D). Transformations were performed in strains in which chromosomal TAF1 was substituted with TAF1 or taf1 on the plasmid. These strains were incubated as described in ( 33 °C on SC medium (van Oevelen et al., 2005). Here, we examined whether taf1-ΔTAND and taf1-N568Δ mutants would display similar growth defects in the spt3Δ background. Strains were cultured for 3 days at 25, 37 and 38 °C on SC medium ( Supplementary Fig.  S4). Alone, the spt3Δ mutation resulted in a mild slow growth defect (compare lanes 1 and 4 in Supplementary  Fig. S4). Severe growth defects were observed for spt3Δ in combination with taf1-N568Δ (compare lanes 3 and 6 at 25 °C in Supplementary Fig. S4) but not with taf1-ΔTAND (compare lanes 2 and 5 in Supplementary Fig.  S4). Consistent with the results shown in Fig. 1, the introduction of plasmid-borne SPT3 into the spt3Δ taf1-N568Δ strain only partially restored the synthetic growth defects (compare lanes 3, 6 and 9 in Supplementary Fig.  S4). In addition, introduction of plasmid-borne SPT3 into the spt3Δ taf1-ΔTAND strain was able to suppress, albeit weakly, the temperature-sensitive growth defects associated with taf1-ΔTAND (compare lanes 2, 5 and 8 at 37 or 38 °C in Supplementary Fig. S4). This observation suggested that ectopically expressed Spt3 was able to substitute, at least in part, for the loss of TAND at higher temperatures, even though the two molecules did not appear to be functionally redundant at 25 °C. Therefore, the suppressive effect of Spt3 on growth could explain the aforementioned reduced PGK1 transcription fluctuations observed after the temperature shift in the taf1-ΔTAND (spt3Δ + SPT3) strain ([e, g, q, s] in Fig. 1B, 1D). Taken together, these results indicate that plasmid-borne SPT3 could only partially compensate for the lack of Spt3, at least in the spt3Δ taf1-N568Δ strain, with respect to both PGK1 transcription (Fig. 1) and cell growth (Supplementary Fig. S4).
Transcription start sites of the PGK1 promoter are not altered by disruption of the TATA box or plasmid-borne SPT3 Previous research showed that PGK1 mRNA levels were not affected by deletion of either or both of the TATA box (TATATATAAA, labeled as #1 in Fig. 3A) and the TATA-like element (TACATATT, labeled as #3 in Fig. 3A) (Ogden et al., 1986;Rathjen and Mellor, 1990). Furthermore, one of these previous studies showed that deletion of both these elements caused a downstream shift in the transcription start sites (TSSs) of the PGK1 promoter (Rathjen and Mellor, 1990). By contrast, we observed that VTC1 mRNA levels were significantly reduced by disruption of the TATA box ([m, o] in Fig. 1A-1D). The reasons for this discrepancy are unclear, but one possibility is the difference in copy number of the reporter gene, which was placed on a multi-copy plasmid in earlier research (Ogden et al., 1986;Rathjen and Mellor, 1990) but integrated into the chromosome in this study. We next conducted primer extension analyses to explore the possibility that disruption of the TATA box or introduction of plasmid-borne SPT3 causes a TSS shift at the  PGK1 promoter ( Fig. 2A, 2B). The TSS profiles observed for the SPT3 and [spt3Δ + SPT3] strains were similar, irrespective of the presence of the TATA box, with the exception that mRNA levels were lower with the taf1-N568Δ mutation at 37 °C in the latter strain (compare lanes 9, 12, 21 and 24 in Fig. 2A, 2B). This suggested that neither the function of the TATA box nor those specifically lost for plasmid-borne SPT3 (Fig. 1) were required for accurate transcription initiation at TSSs of this promoter.
CATAAATT is a functional TATA-like element of the PGK1 promoter Disruption of the TATA box decreased the levels of PGK1 and/or VTC1 mRNA, by more than half in some cases (Fig. 1A-1D and Fig. 2A, 2B), but a substantial proportion of transcription was unaffected. This suggested that a core promoter element(s) (CE) other than the TATA box was present and supported transcription from this promoter. Previously, multiple AT-rich sequences were shown to function as a CE in the RPS5 promoter (Sugihara et al., 2011), and we therefore sought to determine whether two TATA-like elements, CATAAATT (#2 in Fig. 3A) and TACATATT (#3 in Fig.  3A) (Ogden et al., 1986;Rathjen and Mellor, 1990), could function as CEs in this promoter. Northern analysis was used to compare the levels of VTC1 mRNA derived from PGK1 promoters bearing combinations of the intact or mutated #1-3 sequences (Fig. 3B, 3C). The results showed that the two TATA-like elements (#2 and #3) were not functional in the presence of the intact TATA box (#1) (WT and mut 2-4 in Fig. 3C). However, in the presence of a mutated TATA box (#1), the CATAAATT (#2) sequence became functional (compare mut 1/lane 2 and mut 6/lane 7 in Fig. 3C), whereas the TACATATT (#3) sequence remained nonfunctional (compare mut 1/ lane 2 and mut 5/lane 6 in Fig. 3C). Furthermore, when both the TATA box (#1) and CATAAATT (#2) sequence were mutated, the TACATATT (#3) sequence became . Sequences #1 and #3 were described previously as CE (Ogden et al., 1986;Rathjen and Mellor, 1990). The TATA box (#1) was substituted with a GAGAGAGAAA sequence denoted as "GAGA" in  functional (compare mut 6/lane 7 and mut 7/lane 8 in Fig.  3C). Therefore, we concluded that the three sequences (#1-3) functioned as CEs in the PGK1 promoter only when located at the most upstream position among the potential CE sequences.
TFIID dependence or independence of PGK1 transcription cannot be altered by changing the combinations of CE sequences PGK1 transcription became TFIID-dependent when SPT3 was introduced into the spt3Δ taf1-N568Δ strain on a plasmid (e.g., compare [w] in Fig. 1C, 1D), suggesting that this promoter had the potential to be transcribed in a TFIID-dependent manner. We suspected that TFIID dependence could be conferred on PGK1 transcription by changing the combinations of CE sequences. To explore this possibility, the promoter activities of eight constructs, WT and mut 1-7 (Fig. 3B Fig. 4A, 4B). Furthermore, the TATA box-specific stimulatory function of Spt3 was not affected by the absence of the CATAAATT (#2) and/or TACATATT (#3) sequences (WT and mut 2-4 in Fig. 4B). Together, these observations indicate that CE sequences cannot by themselves govern the TFIID dependence or independence of this promoter.
Transient fluctuation of PGK1 transcription in the taf1-ΔTAND strain occurs independently of the TATA box and two TATA-like elements Transient fluctuation of VTC1 mRNA levels was observed after the temperature shift in the taf1-ΔTAND strain, irrespective of the presence or absence of the TATA box (#1) ([q, s] in Fig. 1C). To explore whether the fluctuation was enabled by the presence of the two TATA-like elements (#2 and #3) in the promoter, Northern analysis was used to measure the accumulation kinetics of VTC1 mRNA expressed from the "mut 4" promoter ( Fig. 3B), which contained the TATA box (#1) but lacked the two additional TATA-like elements (#2 and #3), in the TAF1, taf1-ΔTAND, taf1-N568Δ, spt3Δ, spt3Δ taf1-ΔTAND and spt3Δ taf1-N568Δ strains (Supplementary Fig. S5A, S5B). Similar transient fluctuation was observed for the "mut 4"-derived VTC1 mRNA specifically in the taf1-ΔTAND strain ([i] in Supplementary  Fig. S5B), indicating that the fluctuation occurred independently of the two TATA-like elements (#2 and #3), as well as of the TATA box (#1). As mentioned previously, the VTC1 reporter system was more heat-sensitive ([m], [u] in Fig. 1C) than the endogenous system ([a], [i] in Fig.  1C). However, this appeared to be alleviated when the VTC1 mRNA was expressed from the "mut 4" promoter ([g], [k] in Supplementary Fig. S5B). This observation suggests that the two TATA-like elements (#2 and #3) affect the stability of VTC1 mRNA at 37 °C by unknown post-transcriptional mechanisms.

DISCUSSION
In this study, we showed that plasmid-borne SPT3 was functionally different from that SPT3 carried on the chromosome. First, plasmid SPT3 could not fully recapitulate the transient fluctuations of PGK1 and/or VTC1 mRNA levels observed during the initial 2-h phase, following the temperature shift from 25 °C to 37 °C, that was observed in the taf1-ΔTAND strain with chromosomal SPT3 ([e, g, q, s] in Fig. 1B, 1D). However, only plasmid SPT3 was able to suppress temperature-sensitive growth of the taf1-ΔTAND strain (lanes 2, 5 and 8 in Supplementary Fig.  S4), and the differential effects of the two alleles on PGK1 transcription might therefore be a simple consequence of suppression effects caused by copy number variation (the copy number of plasmid SPT3 was approximately 50-fold higher than that of chromosomal SPT3 when quantified by qPCR). Second, plasmid SPT3 could not recapitulate the TFIID independence of PGK1 transcription observed in the taf1-N568Δ strain with chromosomal SPT3, particularly when the promoter lacked the TATA box ([w] in Fig. 1B, 1D). As plasmid SPT3 was unable to restore the growth defect of the spt3Δ taf1-N568Δ strain to a similar extent as chromosomal SPT3 (lanes 3, 6 and 9 in Supplementary Fig. S4), the differential effect on transcription in the taf1-N568Δ strain may be due to intrinsically different properties of the two alleles, rather than to a simple suppression effect.
As stated above, introduction of plasmid-borne SPT3 into the spt3Δ taf1-N568Δ strain failed to restore TFIID independence in transcription from the TATA-less PGK1 promoter ([w] in Fig. 1B, 1D), whereas significant restoration of transcription from the TATA box-containing PGK1 promoter occurred ([i, k, u] in Fig. 1B, 1D). Furthermore, the TATA box-dependent stimulatory function of SPT3 for PGK1 transcription was observed at similar levels both when SPT3 was on the chromosome ([m, n, o, p] in Fig. 1A, 1C) and when provided ectopically via plasmid in the spt3Δ TAF1 strain ([m, n, o, p] in Fig. 1B, 1D). These observations suggest two functions for Spt3: conferring TFIID independence on PGK1 transcription, in particular from the TATA-less promoter when Taf1 function is impaired at 37 °C in the taf1-N568Δ strain ([w] in Fig.  1C, 1D), and stimulating PGK1 transcription in a TATA box-dependent manner regardless of Taf1 functionality ([m-x] in Fig. 1C, 1D). Importantly, plasmid-expressed SPT3 lacked the first of these two activities, thus indi-cating that the two Spt3 functions were mediated by two different mechanisms.
An alternative interpretation of the different results observed with plasmid and chromosomal SPT3 is that the alleles are not functionally different but that some unknown epigenetic mechanism hampers the restoration of TFIID independence of PGK1 transcription when both Spt3 and Taf1 are compromised simultaneously in the spt3Δ taf1-N568Δ strain. This model is consistent with the observation that SPT3 provided by the plasmid appears to be hypomorphic to the chromosomal counterpart, not only in PGK1 transcription ( Fig. 1) but also in growth ( Supplementary Fig. S4), at least when introduced into the spt3Δ taf1-N568Δ strain. Further analysis is needed to discriminate between the two hypotheses, for example by introducing an SPT3 expression plasmid prior to disruption of chromosomal SPT3.
The implications of our findings are two-fold. First, Spt3 has dual functions as described above. Second, plasmid SPT3 was unable to support transcription from the TATA-less promoter in the taf1-N568Δ strain, indicating that Spt3/SAGA mediates PGK1 transcription by different mechanisms depending on the presence or absence of the TATA box in the promoter. To our knowledge, this is the first demonstration that SAGA can mediate transcription by different mechanisms at a single promoter according to the presence of a TATA box or TATA-like element, indicating the requirement for particular subunitspecific functions.
Finally, although the PGK1 promoter had the potential to be transcribed in a TFIID-dependent manner (Fig. 1), TFIID dependence of PGK1 transcription could not be induced by changing the combinations of CE sequences (Fig. 4). This observation is consistent with our previous two-step model for the action of TFIID and SAGA, in which the UAS first specifies either TFIID or SAGA as the predominant factor for a specific promoter, and then the core promoter structure guides that factor to conduct transcription in an appropriate manner (Watanabe and Kokubo, 2017). If this model holds true for the PGK1 promoter, TFIID dependence should not emerge unless the UAS is altered. However, TFIID dependence was evident, particularly at the TATA-less PGK1 promoter, when plasmid SPT3 was introduced into the spt3Δ taf1-N568Δ strain ([w] in Fig. 1B, 1D). Therefore, we propose that the initial "predominant factor-specifying" step mediated by the UAS is affected differently by the chromosomal and plasmid SPT3 alleles, and/or that epigenetic events occur upon simultaneous functional loss of Spt3 and Taf1 as discussed above. We further suggest that the second step guided by core promoter structures is also affected differently by the two SPT3 alleles, depending on the presence of the TATA box. This would suggest that involvement of the TATA box is in contrast to results obtained for the AGP1 promoter, where CEs other than the TATA box or TATA-like element guided SAGA to conduct transcription in an appropriate manner (Watanabe and Kokubo, 2017). Further analyses are needed to identify the molecular mechanisms underpinning the different effects of the chromosomal and plasmid SPT3 alleles, and to further elucidate the TFIID dependence of PGK1 transcription.