PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of molcellbPermissionsJournals.ASM.orgJournalMCB ArticleJournal InfoAuthorsReviewers
 
Mol Cell Biol. May 2010; 30(10): 2376–2390.
Published online Mar 22, 2010. doi:  10.1128/MCB.01046-09
PMCID: PMC2863704
Mechanism of Mediator Recruitment by Tandem Gcn4 Activation Domains and Three Gal11 Activator-Binding Domains [down-pointing small open triangle]
Eric Herbig,1,2 Linda Warfield,1 Lisa Fish,1 James Fishburn,1 Bruce A. Knutson,1 Beth Moorefield,1 Derek Pacheco,1 and Steven Hahn1*
Fred Hutchinson Cancer Research Center, P.O. Box 19024, 1100 Fairview Ave. N., Mailstop A1-162, Seattle, Washington 98109,1 Molecular and Cellular Biology Graduate Program, Fred Hutchinson Cancer Research Center and University of Washington, Seattle, Washington 981092
*Corresponding author. Mailing address: Fred Hutchinson Cancer Research Center, P.O. Box 19024, 1100 Fairview Ave. N., Mailstop A1-162, Seattle, WA 98109. Phone: (206) 667-5261. Fax: (206) 667-6497. E-mail: shahn/at/fhcrc.org
Present address: The Rockefeller University, 1230 York Avenue, New York, NY 10021.
Received August 6, 2009; Revised September 2, 2009; Accepted March 4, 2010.
Targets of the tandem Gcn4 acidic activation domains in transcription preinitiation complexes were identified by site-specific cross-linking. The individual Gcn4 activation domains cross-link to three common targets, Gal11/Med15, Taf12, and Tra1, which are subunits of four conserved coactivator complexes, Mediator, SAGA, TFIID, and NuA4. The Gcn4 N-terminal activation domain also cross-links to the Mediator subunit Sin4/Med16. The contribution of the two Gcn4 activation domains to transcription was gene specific and varied from synergistic to less than additive. Gcn4-dependent genes had a requirement for Gal11 ranging from 10-fold dependence to complete Gal11 independence, while the Gcn4-Taf12 interaction did not significantly contribute to the expression of any gene studied. Complementary methods identified three conserved Gal11 activator-binding domains that bind each Gcn4 activation domain with micromolar affinity. These Gal11 activator-binding domains contribute additively to transcription activation and Mediator recruitment at Gcn4- and Gal11-dependent genes. Although we found that the conserved Gal11 KIX domain contributes to Gal11 function, we found no evidence of specific Gcn4-KIX interaction and conclude that the Gal11 KIX domain does not function by specific interaction with Gcn4. Our combined results show gene-specific coactivator requirements, a surprising redundancy in activator-target interactions, and an activator-coactivator interaction mediated by multiple low-affinity protein-protein interactions.
Activation of transcription, a key regulatory step in gene control, is the endpoint of many signal transduction pathways controlling cell growth, development, and the response to stress. Sequence-specific binding of transcription activators to gene regulatory regions initiates a cascade of events ultimately leading to the assembly of a functional transcription preinitiation complex (PIC) (52). This recruitment pathway involves the cooperative action of coactivator complexes and the transcription machinery (3, 37, 42, 43, 53). A subset of these coactivators (e.g., ATP-dependent remodelers and histone acetyltransferases) act to modify and remodel chromatin, allowing access of additional gene-specific factors and the transcription machinery to promoters, while other coactivators (e.g., SAGA, Mediator, and TFIID) directly interact with PolII and the general transcription factors to promote PIC assembly.
Most of the activator-target interactions characterized to date involve activator-coactivator interactions rather than direct interactions with the general transcription factors (16, 23, 41, 51, 59, 60, 64, 66), and individual activators are generally found to interact with multiple factors. However, for many of these activator-target contacts, it is not clear if the activator-target contacts are promoter specific and why the requirement for specific coactivators varies at different promoters (12, 43).
The acidic transcription activators are an important and universal class of transcription factors that activate transcription in all of the eukaryotes tested (52). Originally recognized in yeast Gal4 and Gcn4 (27, 39), the acidic activators encompass most of the well-characterized yeast activation domains, as well as important mammalian activators such as p53, c-Myc, and E2F and the strong viral activator VP16. The acidic activation “domains” of p53 and VP16 are disordered in the absence of a binding partner and adopt a helical conformation when bound to their targets, interacting through hydrophobic, charged, and polar interactions (16, 33, 35, 36, 63). p53 binding is further regulated by phosphorylation, resulting in an increased affinity of p53 for the Taz2 domain of p300 while decreasing its affinity for the Mdm2 repressor (18, 32). Several acidic activators, including p53, VP16, and Gcn4, have tandem activation domains, although the functional significance of multiple activation domains in the same activator is not well understood (17, 30, 32, 57). The two p53 activation domains have some specificity for distinct targets, while the two VP16 activation domains have similar affinities for at least one factor, the Tfb1 subunit of the general transcription factor TFIIH (4, 11, 18, 65).
Yeast Gcn4 contains tandem acidic activation domains and directly regulates >70 genes involved in diverse processes such as amino acid metabolism, energy homeostasis, purine synthesis, and transcriptional control (17, 25, 44). Gcn4 synthesis is regulated primarily at the level of translation, and Gcn4 levels are elevated under stress conditions, leading to the activation of multiple genes (26). The two Gcn4 activation domains were defined by deletions and mutations in short clusters of hydrophobic residues that reduced the transcription of several Gcn4-regulated genes (17, 27, 30). The Gcn4 N-terminal activation domain spans residues 1 to 100 and central activation domain residues 101 to 134 (17). These two activation domains show no obvious sequence similarity, apart from a preponderance of acidic residues. From earlier studies, the mechanism of how the tandem activation domains cooperate to regulate Gcn4 target genes is not fully understood.
A previous study mapped targets of the Gcn4 central activation domain. Studies of site-specific cross-linking in PICs showed that the central activation domain cross-links to three coactivator subunits: Taf12, a subunit of TFIID and SAGA, Tra1, a subunit of SAGA and NuA4, and the Mediator subunit Gal11/Med15 (19). In vitro studies showed that contacts of this Gcn4 domain with Gal11 and Tra1 are important for stimulation of transcription (19). Other studies utilizing cross-linking, in vivo fluorescence resonance energy transfer, and chromatin immunoprecipitation (ChIP) showed that Mediator, SAGA, and the chromatin remodeler Swi/Snf functionally interact with full-length Gcn4 (3, 7, 22, 45, 51, 67). Pulldown experiments also suggested that Gcn4 interacts with Mediator subunits other than Gal11, although the in vivo importance of these interactions has not yet been resolved (24, 49, 67).
In this work, we investigated the role and mechanism of the tandem Gcn4 activation domains by examining the cooperativity between the two activation domains, identifying the activator-coactivator contacts in PICs, and determining the functional significance of these contacts. We also found an unexpected mechanism for the interaction of the Gcn4 activation domains with one Mediator subunit involving multiple low-affinity interactions that contribute additively to Mediator recruitment and transcription activation.
Protein purification.
All proteins except Gcn4 derivatives used for PEAS cross-linking were expressed in Escherichia coli as N-terminally His6-SUMO-tagged proteins (Invitrogen). Cells containing the SUMO fusions were lysed in 50 mM HEPES (pH 7.0)-500 mM NaCl-40 mM imidazole-10% glycerol-0.1 mM phenylmethylsulfonyl fluoride (PMSF)-3 mM dithiothreitol (DTT) and purified using Ni-Sepharose High Performance resin (GE Healthcare) or GE HisTrap HP columns. Purified proteins were dialyzed versus 50 mM HEPES (pH 7.0)-100 to 500 mM NaCl-10% glycerol-0.1 mM PMSF-1 mM DTT and digested with SUMO protease for 3 to 5 h at 4°C using a 1:500 ratio of protease to protein. The cleaved His6-SUMO tag was removed using Ni-Sepharose, except for Gal11 430-680, which binds Ni-Sepharose. Gal11 and Taf12 polypeptides were further purified using HiTrap Heparin (GE Healthcare) in 20 mM HEPES (pH 7)-1 mM DTT-0.5 mM PMSF, eluting with an 80 to 350 mM NaCl gradient. All Gcn4 activation domain derivatives lacking the DNA-binding domain, except 1-100, were further purified by chromatography on Source 15Q (GE Healthcare) using a 50 to 350 mM NaCl gradient. Gcn4 1-100 was purified using Source 15Q and HiTrap Phenyl Sepharose HP (GE Healthcare). Protein was loaded onto Source 15Q at 120 mM NaCl and flowed though the column. This unbound fraction was adjusted to 0.8 M (NH4)2SO4, bound to phenyl Sepharose, and eluted with a 750 to 150 mM (NH4)2SO4 gradient. All of the proteins used in fluorescence polarization (FP) were further purified using size exclusion chromatography on Superdex 75 or Superdex 200 HR 10/30 (GE Healthcare) in 20 mM HEPES (pH 7.5)-150 mM KCl-1 mM DTT-1 mM PMSF. All proteins eluted in a single peak. Gcn4 derivatives containing the DNA-binding domain were further purified on Source 15Q using a 100 to 300 mM KCl gradient. Gcn4 derivatives used for PEAS cross-linking were expressed as N-terminally His-tagged proteins and purified by methods used for Gcn4 derivatives containing the central activation and DNA-binding domains (19).
Gcn4 cross-linking, Gcn4-FeBABE cleavage, and in vitro transcription.
Gcn4 derivatives containing single cysteine residues were linked to PEAS as described previously (19). These Gcn4-PEAS derivatives were used to stimulate PIC formation with yeast nuclear extract and the HIS4 promoter immobilized to Dynabeads. PICs were washed and cross-linked using 2 × 0.75 J/cm−1 (~5 min total) in a Spectrolinker XL-1500 UV cross-linker (Spectronics Corp.) (19). Cross-linked samples were processed and analyzed as described previously (19). Immunoprecipitation of cross-linked complexes was done as described previously (19). FeBABE attachment to Gcn4, activation of hydroxyl radical production, and analysis of cleavage products were done as described previously (8). Gcn4 derivatives with single or multiple activation domains were assayed for activity at the HIS4 promoter in multiround transcription assays (19).
To identify the cross-linked p47 polypeptide, extracts containing the following Flag-tagged polypeptides were tested in the cross-linking assay: Ccl1, Ssl1, Med2, Sfh1, Caf1, Npl6, Taf4, Eaf3, Eaf7, and Tfa2. In a second approach, immunoprecipitation of TFIID, SAGA, PolII, NuA4, and the 19S proteasome complex after cross-linking was used to show that p47 was not a component of these complexes (19).
Chromatin template assembly.
Nucleosomes were reconstituted from recombinant yeast histones expressed individually in E. coli strain BL21-Codon+RIL (Stratagene) and purified under denaturing conditions as described previously (21). Histones were mixed at equal molar concentrations in guanidine-containing buffer, and multimers were allowed to form during dialysis at 4°C in refolding buffer (38). Histone octamers were isolated by gel filtration on a Superdex S200 column (GE Healthcare), and their concentration was determined by UV absorbance. A chromatin assembly template was generated by replacing the adenovirus E4 promoter of p5SG5E4 (28) with a 303-bp SphI-StyI fragment bearing the yeast HIS4 promoter with a single Gcn4 binding site (p5SHIS4). A 2.3-kb fragment containing the HIS4 promoter flanked by five repeats of 5S rRNA nucleosome positioning sequences was PCR amplified using a biotinylated 5′ primer to permit immobilization on streptavidin beads. These biotinylated templates were gel purified and mixed with a 1:1 molar ratio of histone octamers in a Nap1-mediated chromatin assembly reaction mixture as described previously (21). The assembly of nucleosomal arrays was confirmed by micrococcal nuclease digestion (50), and their protein composition was monitored by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and silver staining to confirm Nap1 elimination. Chromatin-assembled templates were coupled to streptavidin according to Ranish et al. (55), except that the wash buffer contained 0.6 M KCl to enhance the removal of NAP1.
FP assay.
Proteins were labeled with the Oregon Green 488 dye 5-isomer based on the manufacturer's suggested protocols (Invitrogen). For N-terminal labeling, the succinimidyl ester derivative of Oregon Green was resuspended in anhydrous dimethyl sulfoxide to a concentration of 10 mg/ml. For cysteine conjugation, the maleimide derivative of Oregon Green was resuspended to 10 mg/ml in dimethylformamide. For N-terminal labeling, proteins were dialyzed into 20 mM HEPES (pH 7.5)-150 mM KCl. For conjugation to the maleimide derivative of Oregon Green, the proteins were dialyzed into 20 mM HEPES (pH 7.2)-150 mM KCl-10 mM TCEP [Tris(2-carboxyethyl)phosphine]. All proteins were at a minimum concentration of 2 mg/ml. Dye was added dropwise to protein at a 15:1 dye-to-protein molar ratio final concentration. The N-terminal labeling reactions were carried out with gentle mixing at 4°C for 4 h. Cysteine conjugation was performed with gentle mixing at 25°C for 3 h. Following the conjugation reactions, free dye was separated from protein using a Nap5 buffer exchange column (GE Healthcare) that had been equilibrated using 3 column volumes of 20 mM HEPES (pH 7.5)-150 mM KCl and eluted with 1 ml of the same buffer. The degree of labeling was determined using absorbance measured at 280 nM and 496 nM and calculated based on the manufacturer's instructions. Following labeling, protein concentrations were determined using the Pierce BCA assay.
All FP measurements were conducted using a Beacon 2000 instrument (Invitrogen) and based on the manufacturer's protocols. The instrument settings were static, batch blank, blank delay 1 s, sample delay 1 s, number of average read cycles 3, temperature 22°C, and autorange. One-hundred-microliter samples of either Gal11 or Taf12 were assembled in triplicate using approximately 15 concentrations typically spanning 0 to 200 μM. These reaction mixtures were used to blank the Beacon 2000, after which Oregon Green-labeled Gcn4 was added to a final concentration of 50 to 75 nM. Control experiments confirmed that KD measurements were not impacted by increasing the concentration of Gcn4 by severalfold and determined that the binding reaction reached equilibrium in less then 10 s.
Data from the experiments were analyzed using Prism 4 (GraphPad Software, Inc.) to perform nonlinear regression analysis. First, the background signal (the signal from a sample containing only the relevant labeled Gcn4 protein) was subtracted from all measurements. Prism 4 was then used to determine the KD and 95% confidence interval by fitting the data to a one-site binding model using the equation Y = Bmax × [X/(KD + X)], where Y equals arbitrary polarization units and X equals the protein concentration.
Gcn4 pulldown assays.
C-terminally His6-tagged Gcn4 or an inactive Gcn4 derivative composed of amino acids 101 to 281 and containing alanine mutations at residues 107, 110, 113, 120, 123, and 124 (17) was expressed in E. coli. Cells were suspended in buffer containing 50 mM HEPES, 150 mM NaCl, 10% glycerol, 0.1% Tween, 1 mM PMSF, and 5 mM β-mercaptoethanol. Cells were lysed by three passes through a microfluidizer (Microfluidics Corp.) and spun at 15,000 rpm for 30 min. Gcn4 in the extracts was quantitated by Western blotting. A total of 350 μg of Gcn4 was combined with 210 μl of a 50% slurry of Talon metal affinity beads (Clontech), and the volume was brought to 900 μl. The extract and beads were incubated at 4°C for 45 min with gentle mixing. The samples were then applied to a disposable Poly-Prep chromatography column (Bio-Rad) and washed with 3 ml of cell lysis buffer. The columns were capped, and the Gcn4-conjugated beads were resuspended in 210 μl of cell lysis buffer. Concentrations of Gcn4 conjugated to the beads were determined by SDS-PAGE analysis, followed by Coomassie blue staining and comparison to bovine serum albumin standards.
In vitro-translated fragments of Taf12 and Gal11 were generated by cloning open reading frame fragments into the pCITE4b vector (Novagen) and translated using the TNT Quick Coupled transcription/translation system (Promega). Depending on the in vitro translation efficiency, between 0.5 and 2 μl of in vitro-translated fragments was combined with 15 μg of the bead-bound Gcn4 constructs. The reaction mixtures were incubated and gently mixed for 3 h at 4°C, placed into Wizard minicolumns (Promega), and then washed four times with 175 μl of buffer. The samples were eluted with 25 μl of 2× lithium dodecyl sulfate (LDS) buffer containing 500 mM imidazole. The eluted samples, along with 25% of the starting in vitro-translated input, were analyzed by SDS-PAGE, and input Gcn4 levels were compared using Coomassie blue staining. Gels were soaked for 30 min in Amplify solution (GE Healthcare), dried, and visualized by PhosphorImager and/or autoradiography.
Yeast strains and plasmids.
For the yeast strains and plasmids used in this study, see Tables S1 and S2 in the supplemental material. The strains are all derivatives of BY4705 (6). The GAL11 plasmids (see Fig. Fig.55 and and7)7) are derivatives of pEH30, and all lack an epitope tag. All internal deletions, except Δ418-696, Δ116-255, and Δ116-630, contain the sequence GSGSGS at the internal deletion junctions. For these experiments, strain SHY822 was transformed with the indicated GAL11 plasmid and pSH828 containing GCN4. Selected deletions were transferred to plasmid pEH284 containing a six-hemagglutinin (HA) tag at the C terminus of Gal11 and transformed to strain EHY49 for the ChIP, Gal11 stability, and pulldown experiments (see Fig. Fig.66 and and88).
FIG. 5.
FIG. 5.
Multiple conserved elements contribute additively to Gal11 coactivator activity. (A) Gray boxes 1 to 5 indicate conserved motifs within Gal11. Elements 2 to 4 specifically bind Gcn4. Brackets indicate the positions of Glu-Asn- or Glu-Ala-rich regions (more ...)
FIG. 7.
FIG. 7.
Multiple conserved Gal11 elements contribute to Gcn4 activation at genes with modest dependence on Gal11 coactivator function. (A to C) mRNA levels from the indicated Gcn4-dependent genes were determined using RNA from the experiment described in Fig. (more ...)
FIG. 6.
FIG. 6.
Characterization of Gal11 and Taf12 activator-binding domain mutants. (A) Western blotting of whole-cell yeast extracts from strains containing the indicated Gal11-HA derivatives (lanes 1 to 13) and Srb2-Flag. Cells were induced with SM as in other experiments. (more ...)
FIG. 8.
FIG. 8.
Multiple conserved Gal11 elements contribute additively to Mediator recruitment. Strains containing wild-type (WT) Gcn4, epitope-tagged Srb2, and the indicated Gal11 derivatives were grown as described for Fig. Fig.55 to to7,7, induced (more ...)
Yeast whole-cell extracts.
Two hundred milliliters of yeast cells was grown in glucose minimal medium to an optical density at 600 nm (OD600) of 0.8 to 1.0. Cells were washed with water and with whole-cell extraction buffer (100 mM Tris [pH 7.9], 260 mM ammonium sulfate, 1 mM EDTA, 10% glycerol, 0.5 mM DTT, protease inhibitors). Cells were resuspended in 1 ml of buffer and put into a 2-ml Cryovial about three-fourths filled with zirconia/silica beads (Biospec Products, Inc.). Cells were lysed by bead beating (Mini-Beadbeater 96; Biospec) three times for 3 min. The lysed cells and beads were pelleted, and the lysate was collected. The whole-cell extracts were used for Western blotting and immunoprecipitation as described previously (19).
A subset of the Gal11 mutants was HA tagged at the C terminus, and whole-cell extracts were prepared as described above to assess protein expression and stability by Western blotting using anti-HA antibody (SC-7392; Santa Cruz Biotech). Immunoprecipitated complexes were analyzed by Western blotting using an alternative anti-HA antibody (PBR-101; Covance), an anti-Flag antibody (F3165; Sigma), and a rabbit polyclonal anti-Pgd1 antibody.
ChIP assays.
ChIP assays were conducted as described previously (46, 47). Briefly, cell cultures were grown in triplicate to an OD600 of 0.5 to 0.8 in 2% dextrose synthetic complete medium lacking Ile, Val, and Trp at 30°C. Cells were induced with 0.5 to 1.0 μg/ml sulfometuron-methyl (SM) for 1 h and cross-linked with 1% formaldehyde (Sigma) for 15 min at room temperature. After cross-linking, cells were harvested and washed twice with cold phosphate-buffered saline buffer. Cells were resuspended in ChIP buffer (150 mM NaCl, 50 mM Tris-HCl [pH 7.5], 5 mM EDTA, 0.5% NP-40, 1.0% Triton X-100) supplemented with 0.5 mM DTT, 0.5 mM PMSF, and EDTA-free complete protease inhibitor cocktail (Roche). Cells were lysed with 0.5-mm zirconia beads (Biospec) using the Mini-Beadbeater 96 (Biospec) at 3-min intervals until greater than 95% cell breakage was achieved. Lysates were sonicated three times for 10 min each time on the high setting with a Bioruptor UCD-200 (Diagenode). Sonicated lysates were cleared by centrifugation, and protein concentrations were determined by Bradford assay. For immunoprecipitations, 1 mg of sheared chromatin was incubated overnight at 4°C with either 2 μg of anti-HA antibody (sc-7392; Santa Cruz Biotech), 2 μg of anti-Flag M2 antibody (F1804; Sigma), or 4 μg of anti-Gcn4 antibody. Antibody-protein complexes were recovered with 15 μl of protein G Dynabeads (Dynal), and the beads were washed four times in ChIP buffer containing 500 mM NaCl and once with standard ChIP buffer. Next, 100 μl of 10% Chelex-100 (Bio-Rad) was added to the beads and they were boiled for 10 min at 90°C. The suspension was centrifuged, and supernatant containing the eluted DNA was collected. The beads were washed with 100 μl of water and centrifuged, and the supernatants were combined. Eluted DNA was used directly in a PCR. To determine the percentage immunoprecipitated, values were calculated as the ratio of the percentage precipitated at a specific locus to the percentage precipitated at the POL1 reference locus. All values are expressed relative to that of the wild type, which was set at 1.0. Experiments were performed in at least biological duplicate.
Total RNA isolation.
Cell cultures were grown in triplicate to an OD600 of 0.5 to 0.8 in 2% dextrose synthetic complete medium lacking Ile, Val, Ura, and Leu at 30°C. Cells were induced with 0.5 μg/ml SM for 1 h (except for Fig. Fig.3A,3A, where cells were induced for 30 min), and approximately 10 ml of cells was pelleted by centrifugation and washed in cold water. Cells were incubated at 65°C for 1 h in equal volumes of TES (10 mM Tris [pH 7.5], 10 mM EDTA, 0.5% SDS) and acid phenol (Ambion). Cells were extracted twice with acid phenol and once with chloroform (Sigma), and RNA was isolated by ethanol precipitation. Fifteen micrograms of RNA was treated with the Turbo DNase kit (Ambion), and 1 μg of DNA-free RNA was used to generate cDNA using Transcriptor (Roche), anchored oligo(dT)18 primer, and the manufacturer's instructions. cDNA was diluted 1:50 for quantitative PCR (qPCR).
FIG. 3.
FIG. 3.
Activity of the individual and combined Gcn4 activation domains and coactivator dependence of Gcn4-regulated genes. (A) Strain SHY823 containing wild-type (WT) Gcn4 or Gcn4 with a single functional N-terminal (NTAD) or central (CAD) activation domain (more ...)
qPCR.
Gene-specific qPCR was preformed in triplicate using primers near the 3′ end of the gene. Promoter-specific qPCR was also performed in triplicate using primers spanning the upstream activating sequence (Gal11 and Gcn4 ChIPs) or the ATG codon (Srb2 ChIPs). For the sequences of the primers used, see Table S3 in the supplemental material. Primers were designed by either PrimerQuest (IDT) or Primer Express 3 (ABI) software using default parameters. qPCRs were assembled in 5-μl reaction mixtures in a 384-well plate format using an ABI 7900HT Sequence Detection System for real-time PCR (ABI) and Power SYBR green master mix (ABI). Relative amounts of DNA were calculated using a standard curve generated from 10-fold serial dilutions of purified genomic DNA ranging from 10 ng to 0.001 ng. Relative amounts were normalized to ACT1 for RNA expression analysis and to POL1 for ChIP analysis. All values are expressed relative to that of the wild type, which was set at 1.0.
Targets of the Gcn4 N-terminal activation domain.
To understand the mechanism of transcription activation by Gcn4, we first identified targets of the Gcn4 N-terminal activation domain in PICs. By monitoring factors that cross-link to the activator in the context of the PIC, we identified activator-interacting surfaces that are exposed and close to the activator when the transcription complex is fully assembled on DNA. To identify these activator targets, a Gcn4 derivative containing a single functional activation domain (Gcn4 Δ119-208, defective in central activation domain function) was constructed and used for cross-linking studies. The minimal N-terminal Gcn4 polypeptide that retains full transcription activity is composed of residues 1 to 100 in addition to the DNA-binding domain (residues 209 to 281) (17; see below; data not shown). Cysteine substitutions were introduced at six individual positions within or directly adjacent to the N-terminal activation domain. These six activator derivatives were conjugated to the radiolabeled photo-cross-linker N-[(2-pyridyldithio)ethyl]-4-azidosalicylamide (PEAS), which cross-links proteins within 14 Å of the cysteine α carbon (10). When combined with yeast nuclear extract, all Gcn4-PEAS derivatives were active in stimulating in vitro transcription of the yeast HIS4 gene (Fig. (Fig.1A),1A), a TATA-containing, SAGA- and Mediator-dependent promoter that is TFIID independent. The HIS4 promoter derivative used here contains one Gcn4 binding site that binds a single Gcn4 dimer (54).
FIG. 1.
FIG. 1.
The Gcn4 N-terminal activation domain cross-links to Taf12, Gal11, Tra1, and Sin4 in PICs. (A) In vitro transcription from the HIS4 promoter comparing basal transcription (lane 1) to transcription activated by the Gcn4 N-terminal activation domain (Gcn4 (more ...)
To probe for proteins in close contact with the activation domain, the Gcn4-PEAS derivatives were used to stimulate PIC formation on an immobilized HIS4 promoter and these PICs were purified and UV cross-linked. The cross-linked complexes were treated with DTT to reduce the disulfide bond linking PEAS to Gcn4, thereby transferring radiolabeled PEAS to the cross-linked polypeptide. Analysis of these reaction products by SDS-PAGE and autoradiography showed that the Gcn4 N-terminal activation domain cross-linked to six polypeptides ranging in size from 47 to >200 kDa, in addition to cross-linking subunits of the Gcn4 dimer (Fig. (Fig.1B).1B). Cross-linking was strongest when PEAS was inserted at positions 92 and 101 at the C-terminal end of the activation domain. Cross-linking was barely detectable at the N terminus of the activation domain, even though this region is required for optimal activity in vitro, as in vitro transcription of HIS4 driven by the Gcn4 N-terminal activation domain decreases 1.7-fold upon the deletion of residues 2 to 18 (data not shown).
We hypothesized that other Gcn4 targets, such as the chromatin remodeling factor Swi/Snf, may require chromatin to stably associate with the PIC under these conditions. To test if additional cross-linking targets could be observed using a chromatin template, Gcn4-PEAS was used to stimulate PIC formation on HIS4 assembled into chromatin reconstituted from recombinant yeast nucleosomes and immobilized to magnetic beads. This chromatin template contains five 5S nucleosome positioning sequences on either side of HIS4. In this transcription system, transcription is completely dependent on the activator and on the addition of acetyl coenzyme A (acetyl-CoA) (data not shown). Under these conditions, Gcn4 cross-linked to the same set of polypeptides, with the exception that a cross-link to a presumed breakdown product of Tra1 (Tra1*; see below) was not detected (Fig. (Fig.1C1C).
To identify the Gcn4 cross-linking targets, cross-linking reactions were repeated using a series of nuclear extracts with individual PIC subunits triple Flag tagged, altering the mobility of the tagged protein by ~5 kDa. Taf12, a shared subunit of TFIID and SAGA, and the Mediator subunit Gal11/Med15 were identified as the 64- and 100-kDa cross-linking targets, respectively (Fig. (Fig.2A,2A, lanes 1 to 3). Cross-linking to other factors occurred independently of Gal11, since Gcn4 still cross-linked to other targets in a PIC formed from a Δgal11 mutant strain (Fig. (Fig.2A,2A, lane 4). Since Tra1, a subunit of both SAGA and NuA4, is too large to show an SDS-PAGE mobility shift upon Flag tagging, we cross-linked PICs, disrupted the cross-linked complexes with high-salt buffer, and immunoprecipitated SAGA or NuA4 using Flag-tagged subunits unique to each complex (19). Consistent with Tra1 being a direct target of the Gcn4 N-terminal activation domain, a radiolabeled polypeptide the size of Tra1 was coimmunoprecipitated with both Ada1 and Esa1, subunits of SAGA and NuA4, respectively (Fig. (Fig.2B).2B). We believe that the ~150-kDa cross-linking product is a degradation product of Tra1 since it is in both SAGA and NuA4 immunoprecipitates and is larger than the other known subunits of these complexes. The ~85-kDa subunit was shown to be Sin4/Med16 by Flag tag-induced mobility shift and by immunoprecipitation of the Mediator complex using either Flag-tagged Sin4 or Flag-tagged Gal11 (Fig. 2C and D, lanes 2 to 4). Cross-linking to Sin4 increased upon the elimination of Gal11, consistent with the fact that both of these subunits are located in the Mediator tail domain (Fig. (Fig.2A,2A, lane 4). The interaction of Gcn4 with Sin4 agrees with a previous binding study which demonstrated that deletion of Gal11 did not eliminate coprecipitation of Gcn4 and Mediator, while deletion of Sin4, which eliminates both Sin4 and Gal11 from Mediator, abolished coprecipitation (67).
FIG. 2.
FIG. 2.
Identification of Gcn4 N-terminal activation domain cross-linking products. (A) Cross-linking reaction in PICs using nuclear extracts from the wild type (WT) (lane 1), Taf12-Flag (lane 2), Gal11-Flag (lane 3), or a Gal11 deletion strain visualized by (more ...)
Flag-tagged subunits of all known PIC components in the range of 47 kDa were tested in an attempt to identify the cross-linked factor (see Materials and Methods), but none of the candidates tested corresponded to the 47-kDa protein. We therefore speculate that the 47-kDa protein is not a general transcription factor or known coactivator subunit but may instead correspond to a gene-specific transcription factor.
It is striking that both Gcn4 activation domains and the activation domain from Gal4 all cross-link to Tra1, Taf12, and Gal11 since these activation domains have very different sequences (19, 56). These observations suggest that the activator-interacting domains of these coactivator subunits possess common features that facilitate interaction with a variety of acidic activators. In contrast, the Mediator subunit Sin4 cross-links only to the Gcn4 N-terminal activation domain.
The above results (Fig. (Fig.2B)2B) reveal that Tra1 cross-links to Gcn4 in the context of both the SAGA and NuA4 HAT-containing coactivators. However, it was previously found that NuA4 is not required for Gcn4-activated transcription on nonchromatin templates (19). To test the relevance of the Gcn4-NuA4 interaction, we immunodepleted nuclear extract of NuA4 using an HA tag on the NuA4 HAT subunit Esa1 and then tested Gcn4-activated transcription of HIS4 using either chromatin or nonchromatin templates (Fig. (Fig.1D).1D). We found that activated transcription on chromatin templates was highly dependent on the NuA4 complex, while depletion of NuA4 had no effect on transcription from a nonchromatin promoter. This shows that NuA4 is a functional Gcn4 target and a chromatin-specific coactivator.
Relative contributions of the two Gcn4 activation domains are gene specific.
To determine the functional significance of the tandem Gcn4 activation domains, we compared the activated transcription levels of several Gcn4-dependent genes in strains containing wild-type Gcn4 or Gcn4 containing only the N-terminal (Δ101-124) or central (Δ2-100) activation domain. Cells were treated with SM to induce Gcn4 expression, and mRNA levels were measured by reverse transcription (RT)-qPCR. The induced protein levels of both Gcn4 mutants were ≥4-fold higher than that of wild-type Gcn4 (data not shown), consistent with findings that Gcn4 is targeted via the activation domain for degradation under transcription activation conditions (26, 29).
Surprisingly, we observed that the relative contributions of the two Gcn4 activation domains are gene specific (Fig. (Fig.3A).3A). The two activation domains acted synergistically at ARG3, where removal of either activation domain reduced expression to 10 to 25% of the wild-type level. In contrast, at STR3, the two activation domains were less than additive, as either activation domain alone activated transcription to 50 to 75% of the level observed with wild-type Gcn4. However, at HIS4, the two activation domains show additive activity. Subtraction of basal HIS4 activity shows that the activity of wild-type Gcn4 at HIS4 is approximately equal to the sum of the activities of the individual activation domains. These results show that the relative role of the two activation domains is promoter dependent and may be due in part to different coactivator requirements at different Gcn4-regulated genes and/or other regulatory factors acting at these natural promoters.
We also examined cooperation of the Gcn4 activation domains at HIS4 in vitro under conditions similar to those used in the Gcn4-PEAS cross-linking experiments. HIS4 transcription in the presence of Gcn4 containing either the N-terminal or central activation domain was activated about 10-fold and about 30-fold when both activation domains were combined, showing modest synergy between the two activation domains (data not shown).
Role of the coactivator subunits Gal11, Taf12, and Sin4 at Gcn4-dependent genes.
We next examined the requirement of Gcn4-dependent genes for the coactivator subunits Gal11, Sin4, and Taf12. Strains with or without Gal11 were induced with SM, and mRNA levels were measured from 10 Gcn4-dependent genes (Fig. (Fig.3B).3B). Surprisingly, the requirement for Gal11 varies from ~10-fold dependence at ARG3 to complete Gal11 independence at SNO1 and ILV6. One possibility is that the Gcn4-Sin4 interaction is redundant with the Gcn4-Gal11 interaction at some genes. To test this model, we measured mRNA levels in strains with Gal11 and/or Sin4 deleted (Fig. (Fig.3C).3C). These results showed that mutation of the gene for Gal11, the gene for Sin4, or the both genes together had similar effects and that the Gal11-independent genes are also Sin4 independent. As shown below, Mediator recruitment is dependent on the Gcn4-Gal11 interaction at a strongly Gal11-dependent gene. Our combined results suggest that at some Gcn4-dependent genes, the Gcn4-Gal11 interaction may be redundant or partially redundant. This may reflect direct recruitment of Mediator by other transcription factors that interact with other Mediator subunits and/or by cooperative association of Mediator with other coactivators associated with these genes, making direct Gcn4-Mediator contacts unnecessary. It is possible that Gcn4-Gal11 contact occurs only at Gcn4- and Gal11-dependent genes, although we think this less likely than the model in which Gcn4-Gal11 contacts occur at all Gcn4-regulated genes but are redundant with other pathways of Mediator recruitment at some genes.
To test the functional significance of Gcn4-Taf12 interaction, strains with wild-type Taf12 or with Taf12 containing a deletion of sequences within the activator-binding domain (Taf12 Δ138-236; see below) were induced with SM and mRNA levels measured from several Gcn4-dependent genes (Fig. (Fig.3D).3D). In contrast to the results with Gal11, where a number of genes showed strong-to-modest dependence, mutation of the Taf12 activator-binding domain did not strongly alter transcription from any of the eight genes examined. Similar results were observed when the entire Taf12 activator-binding domain was deleted (not shown). Microarray studies examining genome-wide effects of the Taf12 mutation Δ138-236 in SM-induced cells also showed no significant changes compared to the wild type for the expression of any cellular genes. We therefore conclude that the Taf12 activator-binding domain is either redundant with other activator-coactivator interactions or not important for transcription activation in vivo.
Activator-binding domains in Gal11 and Taf12.
To begin to understand how multiple activation domains with unrelated sequences can interact with common coactivator targets, we first mapped the activator-binding domains of Gal11. We initially mapped activator-binding domains using pulldown experiments where His-tagged Gcn4 was immobilized to beads and tested for binding to in vitro-translated segments of Gal11 (Fig. (Fig.4A).4A). To determine if Gcn4-Gal11 binding was specific for the activation function of Gcn4, we compared the binding of full-length Gcn4 to the binding of the Gcn4 central activation domain containing six alanine substitutions (Gcn4 6 × Ala) that eliminate transcription activation in vivo (17). Gal11 contains an N-terminal alpha-helical KIX domain that functionally interacts with the activators Pdr1 and Oaf1 (61, 62). Although segments containing the KIX domain bound Gcn4, binding did not require a functional Gcn4 activation domain as these Gal11 segments bound equally well to the inactive Gcn4 6 × Ala derivative. In contrast, two Gal11 segments encompassing residues 158 to 238 and 430 to 680 showed strong activator-specific binding. In addition, a Gal11 polypeptide containing residues 277 to 368 showed very weak activator-specific binding. These three segments directly and specifically bound the Gcn4 activation domains in other assays shown below. Our results agree with those of Park et al. (49), who observed Gcn4 binding to Gal11 residues 116 to 255, 256 to 410, and 411 to 864. Recently, Jedidi et al. (31) found that at least three Gal11 segments bind the combined Gcn4 activation domains (6 to 90 [KIX domain], 87 to 289, and 233 to 499), two of which correspond to the activator-binding domains shown above. However, as demonstrated in Fig. Fig.4A4A and in two other experiments shown below, we found no evidence of specific binding of the Gcn4 activation domains to the Gal11 KIX domain.
FIG. 4.
FIG. 4.
Mapping of the Gcn4 activator-binding domains within Gal11. (A) Pulldown assay for initial localization of Gal11 activator-binding domains. Brackets above Gal11 indicate the three regions shown in this work to bind Gcn4 activation domains and the KIX (more ...)
To map the regions of Gal11 that interact with Gcn4 in the context of the PIC, we attached the Fe-EDTA probe FeBABE (13) to the N-terminal or central activation domain. When activated by peroxide, FeBABE generates hydroxyl-radical ions that cleave polypeptides within ~30 Å of the FeEDTA reaction center, mapping protein-protein interactions in the context of large complexes (9). FeBABE was linked to Gcn4 at cysteine residues substituted at positions within or near the N-terminal activation domain (residues 92 and 101) or at position 126 within the central activation domain. All Gcn4 derivatives were active in stimulating HIS4 transcription in vitro (data not shown). Gcn4-FeBABE derivatives were used to stimulate PIC formation using nuclear extracts made from cells with a triple Flag tag at the C terminus of Gal11. After washing isolated PICs, FeBABE was activated with H2O2 to produce hydroxyl radicals and the resulting protein cleavage products were visualized on Western blots probed with anti-Flag antibody (Fig. (Fig.4B).4B). Black asterisks mark the cleavage products generated with both activation domains, while blue and red asterisks mark activation domain-specific cleavage sites. The sizes of the cleavage products were calculated relative to a standard curve of in vitro-translated fragments of Gal11-Flag (9), and the positions of the Gal11 cleavage sites are summarized in Fig. Fig.4C4C.
We observed Gal11 cleavage in four regions, three of which correspond to Gcn4 activator-binding domains (158 to 238, 277 to 368, and 430 to 680). In addition, we observed strong cleavage at Gal11 residues 744 to 834 generated almost exclusively by the central activation domain. However, this region does not directly bind Gcn4, as shown above. Experiments shown below to test the functional significance of Gal11 residues 744 to 834 reveal that this region is dispensable for Gal11 function at Gcn4-dependent genes. From these results, we conclude that Gal11 residues 744 to 834 are near the Gcn4 activation domains in the PIC but do not directly bind the activator. Finally, we did not observe cleavage within the Gal11 KIX domain, consistent with our other assays showing no specific binding of Gcn4 to this domain.
Similar pulldown and FeBABE mapping experiments were conducted with Taf12 to map the activator-binding domain. Pulldown studies localized the activator-binding domain to residues 1 to 277 in the nonessential N-terminal region (data not shown). Gcn4-FeBABE mapping showed that most of the cleavage sites fell within or close to this Taf12 segment, with the strongest cleavage centered near residue 140. FP studies shown below confirmed direct binding of the Gcn4 activation domains to Taf12 residues 1 to 277. This Taf12 region is conserved among closely related fungi and also binds the activation domain of the transcription factor Rap1, a factor that functions in part to regulate expression of the TATA-less ribosomal protein genes (20). However, since we found no major changes in genome-wide expression upon mutation of this activator-binding domain, it is unlikely that this segment is essential for activation of ribosomal protein gene transcription.
Functional importance of the conserved Gal11 domains for transcription activation by Gcn4.
We next mutated the three Gal11 activator-binding domains and other conserved regions of Gal11 to determine their importance in Gcn4 transcription activation. Three regions of Gal11/Med15 have been widely conserved in a broad spectrum of eukaryotes (5) corresponding to Saccharomyces cerevisiae residues 17 to 78 (KIX domain), 538 to 606 (one activator-binding domain), and 874 to 895 (a region required for Gal11 association with the Mediator complex) (2, 31, 48). The two other Gcn4 activator-binding domains described above (residues 168 to 233 and 291 to 507) are also conserved in yeasts, with the 168-to-233 region showing the most conservation (see Fig. S1 in the supplemental material). These conserved regions were deleted both individually and in almost all possible combinations to test their function in Gcn4-dependent expression. Figure Figure5A5A shows a schematic of Gal11 with the conserved regions labeled 1 to 5. Conserved regions 1 to 4 are separated by long glutamine-asparagine- or glutamine-alanine-rich stretches. Indicated below Gal11 are the segments deleted in the 21 Gal11 derivatives tested for effects on gene expression.
To examine in vivo stability and association with Mediator, a representative set of the Gal11 derivatives lacking conserved regions 1 to 4 in different combinations were epitope tagged with a C-terminal HA tag. We found that Gal11 derivatives lacking individual conserved domains or mutations removing combinations of two, three, or four conserved domains were all stably expressed in yeast (Fig. (Fig.6A).6A). Removal of conserved domains 1 to 4 or selected combinations of these domains also did not affect the immunoprecipitation of Gal11 with Mediator, as revealed by coprecipitation with Srb2/Med20-Flag (Mediator head module subunit) or Pgd1/Med3 (Mediator tail subunit) (Fig. (Fig.6B).6B). Therefore, these mutations do not affect the stability of Gal11 or its association with the Mediator complex.
Strains containing these Gal11 derivatives lacking an epitope tag were induced with SM, and the expression of four Gcn4-dependent genes was examined. The Gal11 mutations did not alter Gcn4 binding to these gene regulatory regions, as shown by ChIP in a representative set of Gal11 deletion strains (data not shown). We first examined the expression of ARG3, which shows a 10-fold dependence on Gal11 (Fig. (Fig.5B).5B). These measurements strikingly revealed that the conserved Gal11 domains contribute in a nearly additive fashion to Gcn4 activation at ARG3. Deletion of any one conserved domain decreased expression to about 75% of the wild-type level, deletion of any two elements decreased it to about 40% of the wild-type level, deletion of any combination of three decreased it to about 20% of the wild-type level, and deletion of all four elements decreased it to a level near, but slightly above, that observed upon complete GAL11 deletion. The Q/N-rich region between the KIX and activator-binding domains (residues 116 to 157) also contributed to Gal11 function, as deletions removing this region (Δ3, Δ4) were more severe than a deletion that disrupted only the activator-binding domain (Δ2) (also see Fig. Fig.7).7). However, this segment adjacent to conserved region 2 is not required for stable folding of the activator-binding domain (E. Herbig, data now shown) and is not necessary for specific binding of Gcn4 to conserved region 2 (see below). Likewise, the KIX domain contributes to Gal11 function but apparently not by specific interaction with the Gcn4 activation domains. Finally, as described above, deletion of Gal11 residues 631 to 768 produces no defect in gene expression and deletion of residues important for Gal11-Mediator association (Δ20, Δ21) produces defects almost as severe as those caused by a complete GAL11 deletion.
The same general trend of additive effects of the Gal11 conserved regions was observed at three other Gcn4-dependent genes with 2- to 5-fold Gal11 dependence (Fig. (Fig.7),7), although there were a few exceptions. For example, individual deletion of conserved region 1 or 2 had no measurable effect unless they were mutated in combination with other conserved domains. This is in contrast to a recent report that deletion of the KIX domain alone has a significant effect on expression from the ARG1 promoter (31). For these three genes, deletion of three or four conserved regions was required to reduce expression to levels near those observed in the absence of Gal11. Because resistance of cells to SM is a measure of Gcn4 function, cells containing the Gal11 mutations were examined for growth on SM plates (see Table S4 in the supplemental material). Consistent with the mRNA measurements, we found that deletion of multiple conserved Gal11 regions 1 to 4 was required to inhibit growth on SM medium.
The conserved Gal11 domains function additively in Mediator recruitment.
We next examined the function of the conserved Gal11 domains in Mediator recruitment at ARG3 (Fig. (Fig.8).8). Strains containing HA-tagged Gal11 derivatives and chromosomal Srb2-Flag were induced with SM, cross-linked with formaldehyde, and assayed by ChIP for Gall11 and Srb2 association with ARG3. These ChIP results closely parallel the observed effects on mRNA expression where deletion of single conserved domains produces small defects and combinations of two, three, or four domain mutations produce progressively larger defects in Mediator recruitment. From our combined results examining the defects in the Gal11 mutations, we conclude that these conserved domains function additively at Gcn4- and Gal11-dependent genes in both Mediator recruitment and gene activation.
Gcn4 has micromolar affinity for three Gal11 activator-binding domains.
To investigate the direct interaction of Gcn4 with Gal11 and Taf12 in more detail, we developed a quantitative binding assay using FP and highly purified recombinant polypeptides. In this assay, binding of the fluorescently labeled activator to the coactivator target results in an increase in emitted polarized light. The activation domains were labeled with the fluorescent dye Oregon Green either at the N-terminal amine group or on a cysteine residue positioned at the activator C terminus. Gcn4 derivatives labeled at either position gave very similar results for binding to Taf12. The affinity of the Gcn4 N-terminal (residues 1 to 100), central (residues 101 to 134), and combined activation domains (residues 1 to 134) was measured for each of the conserved Gal11 domains, as well as for two Gal11 polypeptides containing two or three activator-binding domains (Fig. (Fig.9;9; Table Table1).1). Gal11 conserved regions 2, 3, and 4 all specifically bound Gcn4 with micromolar affinity. Activator binding was specific since the binding of Gcn4 101-134 6 × Ala to each of these domains was much weaker than the binding of a functional activation domain. Conserved regions 2 and 4 bound the individual Gcn4 activation domains with similar affinities and with observed dissociation constants of 2 to 5 μM. In contrast, activator-binding domain 277-368 bound the Gcn4 N-terminal activation domain about 6-fold more weakly than conserved region 2 or 4. Gal11 residues 277 to 368 favored binding to the Gcn4 N-terminal activation domain by 5-fold compared to the central activation domain.
FIG. 9.
FIG. 9.
Measurement of activator-Gal11 binding by FP. Assay for binding of conserved Gal11 domains to the indicated Oregon Green-labeled Gcn4 activation domains or the 6 × Ala Gcn4 derivative lacking transcription activation activity. Panels: A, Gal11 (more ...)
TABLE 1.
TABLE 1.
Affinity of Gcn4 activation domains for Gal11 and Taf12 activator-binding domainsa
Gcn4 1-134 containing the tandem activation domains bound to conserved Gal11 regions 2 and 3 with similar affinity compared to each individual activation domain. In contrast, Gcn4 1-134 bound 5-fold more tightly to conserved region 4 than did the individual activation domains, suggesting that the two activation domains interact simultaneously with this region. We also observed that the highest-affinity Gcn4-Gal11 binding was to either of two Gal11 polypeptides containing two or three activator-binding domains (Gal11 153-238 and 496-651 or Gal11 143-680). Binding of Gcn4 to Gal11 polypeptides containing multiple activator-binding domains showed 2- to 7-fold higher apparent affinity than Gcn4 binding to the individual activator-binding domains (Table (Table11).
It was recently proposed that Gcn4 directly and specifically binds the Gal11 KIX domain and that this direct interaction is important for transcription activation (31). In contrast, we found that both the individual and combined Gcn4 activation domains bound very weakly and nonspecifically to the KIX domain (Fig. (Fig.9D).9D). Gcn4 binding was not saturated at Gal11 concentrations of up to 200 μM, and we observed that interaction of Gcn4 101-134 and the Gcn4 101-134 6 × Ala mutant was equivalent. Characterization of the recombinant KIX domain by circular dichroism showed that it was alpha helical, as expected, and had a sharp melting point with a melting temperature of 42°C, suggesting a stably folded domain (data not shown). Based on this finding, as well as the pulldown and FeBABE assays, we conclude that the Gcn4 activation domains do not specifically bind the isolated Gal11 KIX domain.
We also measured the binding of the Gcn4 activation domains to Taf12 residues 1 to 277 and observed binding with affinities ranging from 3 to 12 μM (Table (Table1).1). This binding was specific for the activation function of Gcn4, as Gcn4 101-134 6 × Ala bound with nearly 20-fold lower affinity. The polypeptide containing both Gcn4 activation domains (1-134) bound with 3-fold higher affinity than the individual activation domains.
Transcription activation is an important step in gene regulation, and much remains to be understood about the mechanism of activation. Here, we used the yeast acidic activator Gcn4 to investigate the targets of its tandem activation domains and to determine the functional consequences of these activator-target interactions. Our results show a surprising overlap in activation domain function where both Gcn4 activation domains interact with a common set of coactivator targets. Another unexpected finding is the mechanism of Gcn4-Gal11 interaction where both Gcn4 activation domains contact the same three conserved Gal11 activator-binding domains with relatively low affinity. These three activator-binding domains additively contribute to Gal11 function and Mediator recruitment in vivo (Fig. (Fig.1010).
FIG. 10.
FIG. 10.
Summary of Gcn4 activation domain (AD) and Gal11-Gcn4 interactions. (A) Cross-linking targets of the tandem Gcn4 activation domains in PICs. (B) Three Gal11 activator-binding domains bind both Gcn4 activation domains. Gray boxes 1 to 5 indicate conserved (more ...)
We used site-specific cross-linking in PICs to identify Taf12, Gal11, Tra1, and Sin4 as targets of the Gcn4 N-terminal activation domain in both nonchromatin and chromatin-assembled promoters. With the exception of Sin4, the coactivator subunits Taf12, Gal11, and Tra1 are also cross-linking targets of two other acidic activation domains: the Gcn4 central activation domain and the Gal4 activation domain (19, 56). Thus, unlike p53, in which the two individual activation domains may target distinct factors, the tandem Gcn4 activation domains primarily target the same three coactivator subunits, which may be general targets of acidic activators.
We performed a series of experiments to test the functional importance of these Gcn4-coactivator interactions. Gcn4 cross-links to Tra1 in the context of two distinct coactivator complexes, SAGA and NuA4, and it has previously been shown that Gcn4-Tra1 interaction is important for transcription activation via interaction with SAGA (7, 19). Here, we showed that the NuA4 complex, containing the histone acetyltransferase subunit Esa1, is an important chromatin-specific coactivator for Gcn4 function. However, we were surprised to find that the activator-binding domain of Taf12 was not required for Gcn4 activity and was not obviously important for the expression of any tested yeast gene under starvation conditions. It is possible that activator-Taf12 interaction is redundant with other activator-target interactions or that it is simply not functionally important. In contrast to these two extremes, Gcn4-regulated genes show a broad spectrum of dependence on Gal11, ranging from 10-fold dependence to complete Gal11 independence. We speculate that modest or absent Gal11 dependence at some genes is, at least in part, due to other transcription factors that bind to these gene regulatory regions. For example, at the Gal11-independent genes, Gcn4 may work in concert with other factors that make contacts with alternative Mediator subunits, and these other contacts likely compensate for the loss of Gcn4-Gal11 contact. Alternatively, Mediator may be recruited to these Gal11-independent genes by cooperative interactions with other coactivators that are recruited by different pathways.
We found that the two Gcn4 activation domains cooperate in a gene-specific fashion, which was surprising since the two activation domains contact three identical coactivator subunits. At three Gcn4-dependent genes examined, we found that the two activation domains can synergize or function additively or less than additively to promote transcription. Although the basis for this behavior is unknown, it may be due to variable coactivator requirements at these three genes. A better understanding of the reasons for this variable cooperation will help explain how tandem activation domains function in a wide variety of eukaryotic transcription factors.
Three independent experimental approaches mapped the Gal11 domains that specifically bind the two Gcn4 activation domains. These three Gal11 activator-binding domains bound each Gcn4 activation domain with low micromolar affinity. Two of these domains bound either activator with approximately equal affinity, while Gal11 residues 277 to 368 bound with 6-fold higher affinity to the Gcn4 N-terminal activation domain. At least two of these activator-binding domains have previously been suggested to bind the combined Gcn4 activation domains (31, 49). Also, the Gal11 activator-binding domain corresponding to conserved region 2 has been reported to bind the activation domain of glucocorticoid receptor Tau, which normally interacts with a glutamine-rich domain in mammalian steroid receptor coactivator 1 (34).
Recently, Jedidi et al. (31) suggested a key functional interaction between Gcn4 and the Gal11 KIX domain based on pulldown assays, in vivo functional tests, and a nuclear magnetic resonance binding experiment. However, our results from several lines of experiments do not support specific binding of the KIX domain to the Gcn4 activation domains. Particularly revealing are FP experiments showing very weak interaction of Gcn4 with the Gal11 KIX domain that did not saturate at concentrations of KIX polypeptide near 200 μM. This weak interaction is not specific for the activation function of Gcn4 since it is not reduced by a Gcn4 mutation lacking transcription activation function.
Measurement of transcription and Mediator recruitment in vivo showed the functional importance of the activator-binding domains and the KIX domain. Although the KIX domain contributes to Gal11 function, we believe it does not function via direct interaction with Gcn4. Perhaps it interacts with other gene-specific regulators at the promoters tested here, with another Gal11 domain, or with other coactivators or general transcription factors, contributing to the stability of PIC and/or Mediator recruitment.
The three Gcn4 activator-binding domains we identified are unrelated to each other, and sequence analysis did not detect similarity to any known protein fold, although all are predicted to be alpha helical (40, 58). However, one common feature of these activator-binding domains, as well as those found in the Swi/Snf subunits Swi1 and Snf5 (51), is that they are located in close proximity to nearby polyglutamine or polyglutamine/asparagine repeats. Since these glutamine-rich repeats are expected to be flexible (1, 14, 15), they may function to enhance accessibility or exposure of the activator-binding domains in the context of large coactivator complexes.
It was surprising that the Gal11 activator-binding domains acted additively to promote transcription by Gcn4 since we found that a polypeptide containing two or three activator-binding domains bound with significantly higher affinity to Gcn4 than did the single activator-binding domains. This shows that activator-target affinity does not strictly correlate with function and that multiple weak low-micromolar interactions are sufficient to recruit Gal11 and Mediator to gene regulatory regions. Perhaps the relatively weak affinity sufficient for activation is due to the ability of the activation domains to interact with multiple coactivators, which in turn cooperatively bind to the promoter. In other words, recruitment of multiple factors by protein-protein interactions of modest affinity is an effective strategy for assembly of the PIC since many of these recruited factors cooperatively interact with each other and with DNA. The cooperative interactions among coactivators, general transcription factors, and DNA likely circumvent the need for high-affinity, high-specificity protein-protein interactions that are important for the specific interaction of individual molecules and small protein complexes.
In summary, although activator-target contacts may be conserved at many promoters, the relative importance of these contacts varies in a promoter-specific manner. It will be important for future studies to examine specific classes of genes to determine the basis for these mechanistic differences. These differences likely play an important role in regulating the response of different genes to specific activators and signaling pathways and provide a mechanism for a single activator to regulate transcription in a gene-specific fashion.
Supplementary Material
[Supplemental material]
Acknowledgments
We thank Barry Stoddard and Rachel Klevit for encouragement and much valuable advice on protein-protein interactions and protein structure, Leonid Kisselev for assistance with protein purification and FP assays, and members of the Hahn lab and Ted Young for comments throughout the course of this work. We also thank Roland Strong and Vlad Vigdorovich for advice on protein binding assays and Jeff Delrow for assistance with microarray analysis.
This work was supported by training grants T32 CA80416 to E.H. and T32 CA09657 to B.A.K. and grant 5RO1 GM075114 to S.H.
Footnotes
[down-pointing small open triangle]Published ahead of print on 22 March 2010.
Supplemental material for this article may be found at http://mcb.asm.org/.
1. Altschuler, E. L., N. V. Hud, J. A. Mazrimas, and B. Rupp. 2000. Structure of polyglutamine. FEBS Lett. 472:166-168. [PubMed]
2. Barberis, A., J. Pearlberg, N. Simkovich, S. Farrell, P. Reinagel, C. Bamdad, G. Sigal, and M. Ptashne. 1995. Contact with a component of the polymerase II holoenzyme suffices for gene activation. Cell 81:359-368. [PubMed]
3. Bhaumik, S. R., T. Raha, D. P. Aiello, and M. R. Green. 2004. In vivo target of a transcriptional activator revealed by fluorescence resonance energy transfer. Genes Dev. 18:333-343. [PubMed]
4. Bochkareva, E., L. Kaustov, A. Ayed, G. S. Yi, Y. Lu, A. Pineda-Lucena, J. C. Liao, A. L. Okorokov, J. Milner, C. H. Arrowsmith, and A. Bochkarev. 2005. Single-stranded DNA mimicry in the p53 transactivation domain interaction with replication protein A. Proc. Natl. Acad. Sci. U. S. A. 102:15412-15417. [PubMed]
5. Bourbon, H. M. 2008. Comparative genomics supports a deep evolutionary origin for the large, four-module transcriptional mediator complex. Nucleic Acids Res. 36:3993-4008. [PMC free article] [PubMed]
6. Brachmann, C. B., A. Davies, G. J. Cost, E. Caputo, J. Li, P. Hieter, and J. D. Boeke. 1998. Designer deletion strains derived from Saccharomyces cerevisiae S288C: a useful set of strains and plasmids for PCR-mediated gene disruption and other applications. Yeast 14:115-132. [PubMed]
7. Brown, C. E., L. Howe, K. Sousa, S. C. Alley, M. J. Carrozza, S. Tan, and J. L. Workman. 2001. Recruitment of HAT complexes by direct activator interactions with the ATM-related Tra1 subunit. Science 292:2333-2337. [PubMed]
8. Chen, H.-T., and S. Hahn. 2004. Mapping the location of TFIIB within the RNA polymerase II transcription preinitiation complex: a model for the structure of the PIC. Cell 119:169-180. [PubMed]
9. Chen, H. T., and S. Hahn. 2003. Binding of TFIIB to RNA polymerase II: mapping the binding site for the TFIIB zinc ribbon domain within the preinitiation complex. Mol. Cell 12:437-447. [PubMed]
10. Chen, Y., Y. W. Ebright, and R. H. Ebright. 1994. Identification of the target of a transcription activator protein by protein-protein photocrosslinking. Science 265:90-92. [PubMed]
11. Chi, S. W., S. H. Lee, D. H. Kim, M. J. Ahn, J. S. Kim, J. Y. Woo, T. Torizawa, M. Kainosho, and K. H. Han. 2005. Structural details on mdm2-p53 interaction. J. Biol. Chem. 280:38795-38802. [PubMed]
12. Cosma, M. P. 2002. Ordered recruitment: gene-specific mechanism of transcription activation. Mol. Cell 10:227-236. [PubMed]
13. Datwyler, S. A., and C. F. Meares. 2000. Protein-protein interactions mapped by artificial proteases: where sigma factors bind to RNA polymerase. Trends Biochem. Sci. 25:408-414. [PubMed]
14. Davies, P., K. Watt, S. Kelly, C. Clark, N. Price, and I. McEwan. 2008. Consequences of poly-glutamine repeat length for the conformation and folding of the androgen receptor amino-terminal domain. J. Mol. Endocrinol. 41:301-314. [PubMed]
15. Demarest, S. J., M. Martinez-Yamout, J. Chung, H. Chen, W. Xu, H. J. Dyson, R. M. Evans, and P. E. Wright. 2002. Mutual synergistic folding in recruitment of CBP/p300 by p160 nuclear receptor coactivators. Nature 415:549-553. [PubMed]
16. Di Lello, P., L. M. Jenkins, T. N. Jones, B. D. Nguyen, T. Hara, H. Yamaguchi, J. D. Dikeakos, E. Appella, P. Legault, and J. G. Omichinski. 2006. Structure of the Tfb1/p53 complex: insights into the interaction between the p62/Tfb1 subunit of TFIIH and the activation domain of p53. Mol. Cell 22:731-740. [PubMed]
17. Drysdale, C. M., E. Duenas, B. M. Jackson, U. Reusser, G. H. Braus, and A. G. Hinnebusch. 1995. The transcriptional activator GCN4 contains multiple activation domains that are critically dependent on hydrophobic amino acids. Mol. Cell. Biol. 15:1220-1233. [PMC free article] [PubMed]
18. Feng, H., L. M. Jenkins, S. R. Durell, R. Hayashi, S. J. Mazur, S. Cherry, J. E. Tropea, M. Miller, A. Wlodawer, E. Appella, and Y. Bai. 2009. Structural basis for p300 Taz2-p53 TAD1 binding and modulation by phosphorylation. Structure 17:202-210. [PMC free article] [PubMed]
19. Fishburn, J., N. Mohibullah, and S. Hahn. 2005. Function of a eukaryotic transcription activator during the transcription cycle. Mol. Cell 18:369-378. [PubMed]
20. Garbett, K. A., M. K. Tripathi, B. Cencki, J. H. Layer, and P. A. Weil. 2007. Yeast TFIID serves as a coactivator for Rap1p by direct protein-protein interaction. Mol. Cell. Biol. 27:297-311. [PMC free article] [PubMed]
21. Gelbart, M. E., T. Rechsteiner, T. J. Richmond, and T. Tsukiyama. 2001. Interactions of Isw2 chromatin remodeling complex with nucleosomal arrays: analyses using recombinant yeast histones and immobilized templates. Mol. Cell. Biol. 21:2098-2106. [PMC free article] [PubMed]
22. Govind, C. K., S. Yoon, H. Qiu, S. Govind, and A. G. Hinnebusch. 2005. Simultaneous recruitment of coactivators by Gcn4p stimulates multiple steps of transcription in vivo. Mol. Cell. Biol. 25:5626-5638. [PMC free article] [PubMed]
23. Green, M. R. 2005. Eukaryotic transcription activation: right on target. Mol. Cell 18:399-402. [PubMed]
24. Han, S. J., Y. C. Lee, B. S. Gim, G. H. Ryu, S. J. Park, W. S. Lane, and Y. J. Kim. 1999. Activator-specific requirement of yeast mediator proteins for RNA polymerase II transcriptional activation. Mol. Cell. Biol. 19:979-988. [PMC free article] [PubMed]
25. Harbison, C. T., D. B. Gordon, T. I. Lee, N. J. Rinaldi, K. D. Macisaac, T. W. Danford, N. M. Hannett, J. B. Tagne, D. B. Reynolds, J. Yoo, E. G. Jennings, J. Zeitlinger, D. K. Pokholok, M. Kellis, P. A. Rolfe, K. T. Takusagawa, E. S. Lander, D. K. Gifford, E. Fraenkel, and R. A. Young. 2004. Transcriptional regulatory code of a eukaryotic genome. Nature 431:99-104. [PMC free article] [PubMed]
26. Hinnebusch, A. G. 2005. Translational regulation of GCN4 and the general amino acid control of yeast. Annu. Rev. Microbiol. 59:407-450. [PubMed]
27. Hope, I. A., S. Mahadevan, and K. Struhl. 1988. Structural and functional characterization of the short acidic transcriptional activation region of yeast GCN4 protein. Nature 333:635-640. [PubMed]
28. Ikeda, K., D. J. Steger, A. Eberharter, and J. L. Workman. 1999. Activation domain-specific and general transcription stimulation by native histone acetyltransferase complexes. Mol. Cell. Biol. 19:855-863. [PMC free article] [PubMed]
29. Irniger, S., and G. H. Braus. 2003. Controlling transcription by destruction: the regulation of yeast Gcn4p stability. Curr. Genet. 44:8-18. [PubMed]
30. Jackson, B. M., C. M. Drysdale, K. Natarajan, and A. G. Hinnebusch. 1996. Identification of seven hydrophobic clusters in GCN4 making redundant contributions to transcriptional activation. Mol. Cell. Biol. 16:5557-5571. [PMC free article] [PubMed]
31. Jedidi, I., F. Zhang, H. Qiu, S. J. Stahl, I. Palmer, J. D. Kaufman, P. S. Nadaud, S. Mukherjee, P. T. Wingfield, C. P. Jaroniec, and A. G. Hinnebusch. 2010. Activator Gcn4 employs multiple segments of Med15/Gal11, including the KIX domain, to recruit mediator to target genes in vivo. J. Biol. Chem. 285:2438-2455. [PubMed]
32. Jenkins, L. M., H. Yamaguchi, R. Hayashi, S. Cherry, J. E. Tropea, M. Miller, A. Wlodawer, E. Appella, and S. J. Mazur. 2009. Two distinct motifs within the p53 transactivation domain bind to the Taz2 domain of p300 and are differentially affected by phosphorylation. Biochemistry 48:1244-1255. [PMC free article] [PubMed]
33. Kaustov, L., G. S. Yi, A. Ayed, E. Bochkareva, A. Bochkarev, and C. H. Arrowsmith. 2006. p53 transcriptional activation domain: a molecular chameleon? Cell Cycle 5:489-494. [PubMed]
34. Kim, D. H., G. S. Kim, C. H. Yun, and Y. C. Lee. 2008. Functional conservation of the glutamine-rich domains of yeast Gal11 and human SRC-1 in the transactivation of glucocorticoid receptor Tau 1 in Saccharomyces cerevisiae. Mol. Cell. Biol. 28:913-925. [PMC free article] [PubMed]
35. Kussie, P., S. Gorina, V. Marechal, B. Elenbaas, J. Moreau, A. J. Levine, and N. P. Pavletich. 1996. Structure of the MDM2 oncoprotein bound to the p53 tumor suppressor transactivation domain. Science 274:948-953. [PubMed]
36. Langlois, C., C. Mas, P. Di Lello, L. M. Jenkins, P. Legault, and J. G. Omichinski. 2008. NMR structure of the complex between the Tfb1 subunit of TFIIH and the activation domain of VP16: structural similarities between VP16 and p53. J. Am. Chem. Soc. 130:10596-10604. [PubMed]
37. Liu, X., M. Vorontchikhina, Y. L. Wang, F. Faiola, and E. Martinez. 2008. STAGA recruits Mediator to the MYC oncoprotein to stimulate transcription and cell proliferation. Mol. Cell. Biol. 28:108-121. [PMC free article] [PubMed]
38. Luger, K., T. J. Rechsteiner, and T. J. Richmond. 1999. Preparation of nucleosome core particle from recombinant histones. Methods Enzymol. 304:3-19. [PubMed]
39. Ma, J., and M. Ptashne. 1987. Deletion analysis of GAL4 defines two transcriptional activating segments. Cell 48:847-853. [PubMed]
40. McGuffin, L. J., K. Bryson, and D. T. Jones. 2000. The PSIPRED protein structure prediction server. Bioinformatics 16:404-405. [PubMed]
41. Mittler, G., T. Stuhler, L. Santolin, T. Uhlmann, E. Kremmer, F. Lottspeich, L. Berti, and M. Meisterernst. 2003. A novel docking site on Mediator is critical for activation by VP16 in mammalian cells. EMBO J. 22:6494-6504. [PubMed]
42. Näär, A. M., B. D. Lemon, and R. Tjian. 2001. Transcriptional coactivator complexes. Annu. Rev. Biochem. 70:475-501. [PubMed]
43. Narlikar, G. J., H. Y. Fan, and R. E. Kingston. 2002. Cooperation between complexes that regulate chromatin structure and transcription. Cell 108:475-487. [PubMed]
44. Natarajan, K., M. R. Meyer, B. M. Jackson, D. Slade, C. Roberts, A. G. Hinnebusch, and M. J. Marton. 2001. Transcriptional profiling shows that Gcn4p is a master regulator of gene expression during amino acid starvation in yeast. Mol. Cell. Biol. 21:4347-4368. [PMC free article] [PubMed]
45. Neely, K. E., A. H. Hassan, C. E. Brown, L. Howe, and J. L. Workman. 2002. Transcription activator interactions with multiple SWI/SNF subunits. Mol. Cell. Biol. 22:1615-1625. [PMC free article] [PubMed]
46. Nelson, J., O. Denisenko, and K. Bomsztyk. 2009. The fast chromatin immunoprecipitation method. Methods Mol. Biol. 567:45-57. [PubMed]
47. Nelson, J. D., O. Denisenko, and K. Bomsztyk. 2006. Protocol for the fast chromatin immunoprecipitation (ChIP) method. Nat. Protoc. 1:179-185. [PubMed]
48. Nishizawa, M. 2001. Negative regulation of transcription by the yeast global transcription factors, Gal11 and Sin4. Yeast 18:1099-1110. [PubMed]
49. Park, J. M., H. S. Kim, S. J. Han, M. S. Hwang, Y. C. Lee, and Y. J. Kim. 2000. In vivo requirement of activator-specific binding targets of mediator. Mol. Cell. Biol. 20:8709-8719. [PMC free article] [PubMed]
50. Pazin, M. J., and J. T. Kadonaga. 1998. Transcriptional and structural analysis of chromatin assembled in vitro, p. 173-194. In H. Gould (ed.), Chromatin: a practical approach. Oxford University Press, Oxford, United Kingdom.
51. Prochasson, P., K. E. Neely, A. H. Hassan, B. Li, and J. L. Workman. 2003. Targeting activity is required for SWI/SNF function in vivo and is accomplished through two partially redundant activator-interaction domains. Mol. Cell 12:983-990. [PubMed]
52. Ptashne, M., and A. Gann. 2002. Genes and signals. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
53. Qiu, H., C. Hu, F. Zhang, G. J. Hwang, M. J. Swanson, C. Boonchird, and A. G. Hinnebusch. 2005. Interdependent recruitment of SAGA and Srb mediator by transcriptional activator Gcn4p. Mol. Cell. Biol. 25:3461-3474. [PMC free article] [PubMed]
54. Ranish, J. A. 1999. Mechanisms of transcription by RNA polymerase II. Ph.D. thesis. University of Washington, Seattle, WA.
55. Ranish, J. A., N. Yudkovsky, and S. Hahn. 1999. Intermediates in formation and activity of the RNA polymerase II preinitiation complex: holoenzyme recruitment and a postrecruitment role for the TATA box and TFIIB. Genes Dev. 13:49-63. [PubMed]
56. Reeves, W. M., and S. Hahn. 2005. Targets of the Gal4 transcription activator in functional transcription complexes. Mol. Cell. Biol. 25:9092-9102. [PMC free article] [PubMed]
57. Regier, J. L., F. Shen, and S. J. Triezenberg. 1993. Pattern of aromatic and hydrophobic amino acids critical for one of two subdomains of the VP16 transcriptional activator. Proc. Natl. Acad. Sci. U. S. A. 90:883-887. [PubMed]
58. Rost, B., G. Yachdav, and J. Liu. 2004. The PredictProtein server. Nucleic Acids Res. 32:W321-W326. [PMC free article] [PubMed]
59. Stevens, J. L., G. T. Cantin, G. Wang, A. Shevchenko, and A. J. Berk. 2002. Transcription control by E1A and MAP kinase pathway via Sur2 mediator subunit. Science 296:755-758. [PubMed]
60. Teufel, D. P., S. M. Freund, M. Bycroft, and A. R. Fersht. 2007. Four domains of p300 each bind tightly to a sequence spanning both transactivation subdomains of p53. Proc. Natl. Acad. Sci. U. S. A. 104:7009-7014. [PubMed]
61. Thakur, J. K., H. Arthanari, F. Yang, K. H. Chau, G. Wagner, and A. M. Näär. 2009. Mediator subunit Gal11p/MED15 is required for fatty acid-dependent gene activation by yeast transcription factor Oaf1p. J. Biol. Chem. 284:4422-4428. [PubMed]
62. Thakur, J. K., H. Arthanari, F. Yang, S. J. Pan, X. Fan, J. Breger, D. P. Frueh, K. Gulshan, D. K. Li, E. Mylonakis, K. Struhl, W. S. Moye-Rowley, B. P. Cormack, G. Wagner, and A. M. Näär. 2008. A nuclear receptor-like pathway regulating multidrug resistance in fungi. Nature 452:604-609. [PubMed]
63. Uesugi, M., O. Nyanguile, H. Lu, A. J. Levine, and G. L. Verdine. 1997. Induced alpha helix in the VP16 activation domain upon binding to a human TAF. Science 277:1310-1313. [PubMed]
64. Waters, L., B. Yue, V. Veverka, P. Renshaw, J. Bramham, S. Matsuda, T. Frenkiel, G. Kelly, F. Muskett, M. Carr, and D. M. Heery. 2006. Structural diversity in p160/CREB-binding protein coactivator complexes. J. Biol. Chem. 281:14787-14795. [PubMed]
65. Xiao, H., A. Pearson, B. Coulombe, R. Truant, S. Zhang, J. L. Regier, S. J. Triezenberg, D. Reinberg, O. Flores, C. J. Ingles, and J. Greenblatt. 1994. Binding of basal transcription factor TFIIH to the acidic activation domains of VP16 and p53. Mol. Cell. Biol. 14:7013-7024. [PMC free article] [PubMed]
66. Yang, F., R. DeBeaumont, S. Zhou, and A. M. Näär. 2004. The activator-recruited cofactor/Mediator coactivator subunit ARC92 is a functionally important target of the VP16 transcriptional activator. Proc. Natl. Acad. Sci. U. S. A. 101:2339-2344. [PubMed]
67. Zhang, F., L. Sumibcay, A. G. Hinnebusch, and M. J. Swanson. 2004. A triad of subunits from the Gal11/tail domain of Srb mediator is an in vivo target of transcriptional activator Gcn4p. Mol. Cell. Biol. 24:6871-6886. [PMC free article] [PubMed]
Articles from Molecular and Cellular Biology are provided here courtesy of
American Society for Microbiology (ASM)