The 9aaTAD Is Exclusive Activation Domain in Gal4

The Gal4 protein is a well-known prototypic acidic activator that has multiple activation domains. We have previously identified a new activation domain called the nine amino acid transactivation domain (9aaTAD) in Gal4 protein. The family of the 9aaTAD activators currently comprises over 40 members including p53, MLL, E2A and other members of the Gal4 family; Oaf1, Pip2, Pdr1 and Pdr3. In this study, we revised function of all reported Gal4 activation domains. Surprisingly, we found that beside of the activation domain 9aaTAD none of the previously reported activation domains had considerable transactivation potential and were not involved in the activation of transcription. Our results demonstrated that the 9aaTAD domain is the only decisive activation domain in the Gal4 protein. We found that the artificial peptides included in the original Gal4 constructs were results of an unintended consequence of cloning that were responsible for the artificial transcriptional activity. Importantly, the activation domain 9aaTAD, which is the exclusive activation domain in Gal4, is also the central part of a conserved sequence recognized by the inhibitory protein Gal80. We propose a revision of the Gal4 regulation, in which the activation domain 9aaTAD is directly linked to both activation function and Gal80 mediated inhibition.

There are four activation domains reported for Gal4 protein. Early Gal4 studies reported one weak N-terminal activation domain AD-I (148-196 aa; 13-fold less efficient than the intact Gal4) and one strong C-terminal activation domain AD-II (763-823 aa; as efficient as the intact Gal4) [24]. Similarly to the Gcn4 acidic activator, authors correlated the presence of a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 multiple acidic amino acids with activation function. Initially, a deletion of the 33 amino acid long C-terminus of the Gal4 protein had only a minor effect on the Gal4 ability to activate transcription (as 85% activity was detected in the construct pMA230, Gal4 region 1-848 aa) [24]. In the next study, the same authors showed that their mutant was fully independent on galactose stimulation [25].
However, the results from Melcher and Johnston studies [26,27] clearly showed that the Gal4 C-terminal 34 aa long end represented another activation domain AD-III, which provided the interaction with coactivators of transcription and also with inhibitory protein Gal80. The Gal4 loss-of-function mutant (1-852 aa, deletion of the Gal4 C-terminal 29 aa long end) called gal4-64 supported these findings [28][29][30].
In the following Gal4 study from Ptashne, a shorter activation domain AD-III (840-857 aa) was declared as the major activation domain in Gal4 protein [31]. However, the shorter activation domain AD-III did not include the region recognized by inhibitory protein Gal80. Furthermore, the authors reported previously that the Gal4 constructs without the activation domain AD-III were strong activators of transcription (AD-II construct CD15XX, 1-823 aa; AD-II construct CDCD13, 1-844 aa) [24].
We have reported a new activation domain for the Gal4 transcription factor, the nine amino acids transactivation domain, 9aaTAD [32]. The activation domains 9aaTAD are created by a tandem of two hydrophobic clusters that are interspersed by a hydrophilic region. As highlighted previously [32], the hydrophobic clusters of the activation domain 9aaTAD are consistent with a broad evidence of hydrophobic patterns reported for Gcn4, p53 and other transcription factors [12][13][14][15][16][17]22]. The 9aaTAD pattern was reported and the online prediction could be found on www.piskacek.org ( [32,33]. Occasionally, the 9aaTAD function could be significantly enhanced by the adjacent amino acids (one aa at the C-terminus and up to 4 aa at the N-terminus, which included usually one or more hydrophobic amino acids e.g. as shown for the Msn2 orthologs [32,34]). Therefore the activation domains 9aaTAD could be up to 14 amino acid long, when the contribution of the adjacent amino acids is discernible.
The position of the Gal4 9aa TAD activation domain did not correspond to any of the previously reported activation domains AD-I, AD-II or AD-III from Ptashne lab. Rather, the position of the activation domain 9aaTAD correlated with the Gal80 binding region and with the region missing in the Gal4 loss-of-function mutant gal4-64 (1-852 aa) [26,27].
In this study, we revised function of all reported Gal4 activation domains. Surprisingly, we found that beside of the activation domain 9aaTAD none of the previously reported activation domains was functional. Our results showed that the 9aaTAD domain is the only decisive activation domain in the Gal4 protein.

Materials and Methods Constructs
The construct pBTM116-HA was generated by an insertion of the HA cassette into the EcoRI site of the vector pBTM116 (HA cassette nucleotide sequence: TGG CTG-GAATTA-GCC ACC ATG GCT TAC CCA TAC GAT GTT CCA GAT TAC GCT GTC GAG ATA-GAATTC, which render in amino acids sequence: W L-E L-A T M A Y P Y D V P D Y A V E I-E F). The constructs G1-G45 and H1-H45 were generated by PCR and subcloned into pBTM116 EcoRI and BamHI sites. All construct have a spacer of three amino acids inserted into the EcoRI site; peptide -NNN-(NNN cassette: GAATTC-AATAATAAT, which render in peptide: EF-NNN). All constructs were sequenced by Eurofins Genomics. Further detailed information about constructs, primer sequences are available on the request.

Assessment of enzyme activities
The β-galactosidase activity was determined in the yeast strain L40 [35,36]. The strain L40 has integrated the lacZ reporter driven by the lexA operator. In all hybrid assays, we used 2μ vector pBTM116 for generation of the LexA hybrids. The yeast strain L40, the Saccharomyces cerevisiae Genotype: MATa ade2 his3 leu2 trp1 LYS::lexA-HIS3 URA3::lexA-LacZ, is deposited at ATCC (#MYA-3332). For β-galactosidase assays, overnight cultures propagated in YPD medium (1% yeast extract, 2% bactopeptone, 2% glucose) were diluted to an A 600 of 0.3 and further cultivated for two hours and collected by centrifugation. The crude extracts were prepared by vortexing with glass beads for 3 minutes. The assay was done with 10 ul crude extract in 1ml of 100 mM phosphate buffer pH7 with 10 mM KCl, 1 mM MgSO4 and 0.2% 2-Mercaptoethanol; reaction was started by 200 ul 0.4% ONPG and stopped by 500 ul 1 M CaCO3. The average value of the β-galactosidase activities from three independent experiments is presented as a percentage of the reference with the standard deviation (means and plusmn; SD; n = 3). We standardized all results to previously reported Gal4 construct HaY including merely the activation domain 9aaTAD with the activity set to 100% [34].

Western blot analysis
The crude cell extracts were prepared in a buffer containing 200 mM Tris-HCl, pH 8.0, 1 mM EDTA, 10% glycerol (v/v), separated by SDS-PAGE, and blotted to nitrocellulose. The immuno-detection of proteins was carried out using mouse anti-HA antibody (#26183, Ther-moFisher Sci) or mouse anti-LexA (#306-719, EMD Millipore Corp). The secondary antibodies used were anti-mouse IgG antibodies conjugated with horseradish peroxidase (#A9044, Sigma Aldrich). The proteins were visualized using Pierce ECL (#32106, ThermoFisher Sci) according to the manufacturer's instructions.

Functional and non-functional activation domains in Gal4 activator
To we revised function of all reported Gal4 activation domains, we had generated a set of the Gal4 constructs (Figs 1, 2 and 3). The activities of all Gal4 constructs in this study were tested in LexA hybrid assay (the LexA is E.coli DNA binding domain and the Gal4 DBD is S.cerevisiae Gal4 DNA binding domain, both generally used for the generation of fusion hybrids; 2μ vector pBTM116 was used for all constructs; strain L40 with integrated lacZ reporter driven by lexA operator was used to assay the activation of transcription). The expression of the LexA-Gal4 protein hybrids were proved by western blotting (S1 Fig).
The constructs G577, G81, and G49, which all included the activation domain 9aaTAD, have the capacity to activate transcription (Fig 1). Surprisingly, the constructs G10, G42, G405 and G50, with activation domains AD-I, AD-II or AD-III, had no substantial activity.

Artificial peptides in the original Gal4 constructs activated transcription
From the data above, it was not clear why the Gal4 activation domains AD-I, AD-II and AD-III were inactive in our constructs. By the inspection of the previously reported data, we identified artificial peptides in the original constructs as a result of an unintended consequence of cloning [24]. Therefore we have generated a set of the Gal4 constructs with and without the artificial peptides (H41, H43, H45, H47 versus H42, H44, H46, and H48). In addition, we generated G42, G44, G46, and G48 constructs without the HA tags (the standard version of LexA hybrid without any modification of the original BTM116 vector). To avoid any cloning artefacts, we have inserted a peptide spacer of three Asparagines (NNN) between the LexA DNA binding domain and the Gal4 fragments.
We have observed that the three constructs with the artificial peptides including the H41 (Gal4 sequence 766-848 aa with RVWN HYRDV peptide), H43 (Gal4 sequence 766-844 aa with GIPD HYRDV peptide) and H47 (Gal4 sequence 766-823 with PEFR RVWN HYRDV peptide) strongly activated transcription (Fig 2). The activities of the corresponding constructs without the artificial peptides (constructs H42, H44 and H48) were dramatically reduced or completely diminished. This observation was even more pronounced in the constructs without the HA-tags (constructs G42, G44, G46, and G48). The weak activities were observed in two out of four constructs with the HA-tags (for the constructs H42, H48 but not for H44 and H46 constructs). All constructs without the HA-tags had none substantial activity, therefore the weak activity of the construct H42 (and even less in H48 construct) was considered as an elevated basal level, which extended to 15% of the Gal4 C-terminal construct G49 and 10% of the 9aaTAD construct H577. The LexA-HA construct Hdd with the HA-tag was inactive and therefore the HA-tag could not activate transcription (Fig 3).
The artificial peptides in H41, H43 and H47 did not fit the 9aaTAD pattern and therefore are not activation domains of the 9aaTAD family.

The activation domain AD-III did not activate transcription
In order to fully justify the 9aaTAD domain function, we have generated the Gal4 constructs G49 (776-871 aa) and G50 (776-863 aa) that ended before and after the activation domain 9aaTAD. Furthermore, we have generated two different constructs including merely the activation domain 9aaTAD with two different fusion regions (boundaries) to the LexA DNA binding domain-constructs H577 (857-871 aa) and HaY (860-871 aa) (Fig 3). To avoid artefacts, we have inserted the peptide linker of three Asparagines (NNN) between LexA and Gal4 in all our constructs (LexA-HA tag-NNN-Gal4 region).
We have previously reported that fusion of the Gal4 DNA binding domain could occasionally generate strong artificial activation domains [34]. The Gal4 region (92-100 aa) might serve as a half of the artificial activation domain 9aaTAD. We generated construct U39 with a half of the artificial activation domain 9aaTAD originated from the DNA binding domain of the Gal4 (Gal4 region 92-100, peptide LLTGLFVQD) together with four amino acids of the activation domain AD-III, which represented the second half of the artificial activation domain 9aaTAD (Gal4 region 840-843, peptide WTDQ).
As shown in
In this study, we found that none of the previously reported Gal4 activation domains AD-I, AD-II or AD-III was able to activate transcription. Therefore, the domain 9aaTAD is the exclusive activation domain in the Gal4 protein, which has the decisive competence to activate transcription. Thus the term acidic activation domain reported for Gal4 protein [2,25,31 . Nevertheless, we showed in this study that the same Gal4 region of 18 amino acids was unable to activate transcription (AD-III construct G405). We suspected that an artificial activation domain has been generated in the original AD-III construct pRJR200 by fusion of the Gal4 DNA binding domain with otherwise inactive the Gal4 region AD-III (840-857 aa). The latter fusion (construct pRJR200#), the Gal4 DNA binding domain and four amino acids of the Gal4 activation domain AD-III were fused together without any suitable peptide spacer (Gal4 region 92-100 and Gal4 region 840-843, peptide LLTGLFVQD-WTDQ). This fusion was a powerful activator of transcription, which was two-fold stronger than the Gal4 Similar artificial transactivation domain was generated previously by fusion of the Gal4 DNA binding domain (96-100 aa) and P201 activator [44]. The Gal4-P201 hybrid (peptide LFVQD-YLLPTCIP) was strong activator of transcription [44].
The position of the activation domain 9aaTAD (857-871 aa) correlates with the Gal4 region 854-875, which is necessary and sufficient for the interaction with inhibitory protein Gal80. The 21 aa long peptide derived from the Gal4 region 854-875 tightly binds to the inhibitory protein Gal80 (peptide: GMFNTTTM DDVYNYLFD DEDT; the Gal4 activation domain 9aaaTAD was included inside of the peptide: DDVYNYLFD; structural data available at PDB accession code 3E1K) [45]. Similarly, related peptide derived from Gal4's ortholog K.l.Gal9 (peptide: TQQLFNTTTM DDVYNYIFD NDE; structural data available at PDB accession code 3BTS) [11] (Fig 4). Furthermore, recent report has suggested a functional link between the artificial peptides binding to the Gal80 protein and their ability to activate transcription; "Peptides selected to bind the Gal80 repressor are potent transcriptional activation domains in yeast" [46].
Collectively, we inferred the Gal4 regulation, which involves the activation domain 9aaTAD, the Gal4 mediators of transcription and the inhibitory protein Gal80. We propose a new model of the Gal4 regulation (Fig 4) to the previously report [39].