|Home | About | Journals | Submit | Contact Us | Français|
Gal4 is a prototypical eukaryotic transcriptional activator whose recruitment function is inhibited in the absence of galactose by the Gal80 protein through masking of its transcriptional activation domain (AD). A long-standing nondissociation model posits that galactose-activated Gal3 interacts with Gal4-bound Gal80 at the promoter, yielding a tripartite Gal3-Gal80-Gal4 complex with altered Gal80-Gal4 conformation to enable Gal4 AD activity. Some recent data challenge this model, whereas other recent data support the model. To address this controversy, we imaged fluorescent-protein-tagged Gal80, Gal4, and Gal3 in live cells containing a novel GAL gene array. We find that Gal80 rapidly dissociates from Gal4 in response to galactose. Importantly, this dissociation is Gal3 dependent and concurrent with Gal4-activated GAL gene expression. When galactose-triggered dissociation is followed by galactose depletion, preexisting Gal80 reassociates with Gal4, indicating that sequestration of Gal80 by Gal3 contributes to the observed Gal80-Gal4 dissociation. Moreover, the ratio of nuclear Gal80 to cytoplasmic Gal80 decreases in response to Gal80-Gal3 interaction. Taken together, these and other results provide strong support for a GAL gene switch model wherein Gal80 rapidly dissociates from Gal4 through a mechanism that involves sequestration of Gal80 by galactose-activated Gal3.
Direct masking of the activation domain (AD) of a transcriptional activator by an inhibitory protein and relief of such masking in response to signals is typical for several eukaryotic gene regulatory systems. Such is the case for the transcriptional inhibitors RB, MDM4/MDMX, ZFM1, Opi1, and Gal80, all of which target DNA-binding transcription activators to exert their inhibitory effect (5, 6, 16, 33). In the case of RB, it binds to a site embedded in the transactivation domain of the E2F protein. Phosphorylation of RB lowers its binding affinity to E2F and results in gene activation (6). For the Opi1 protein in Saccharomyces cerevisiae, signal-responsive tethering of Opi1 to the ER membrane physically separates Opi1 and its target, Ino2, the transcriptional activator protein involved in the inositol pathway (12). For the yeast Gal80 protein, its ability to mask the AD of the transcriptional activator Gal4 is somehow overcome through interaction with the galactose-activated form of the Gal3 protein. For each of these systems, the molecular activities that cause the unmasking event constitute the overall signal-responsive gene activation mechanism.
In our efforts to elucidate the mechanism of GAL gene activation in yeast, we have focused on how galactose relieves Gal80 masking of the Gal4 AD. Gal4 is a prototypical acidic transcriptional activator that binds to a 17-bp upstream activation sequence within the GAL gene promoters (UASGAL) (3, 7). In the absence of galactose, Gal80 masks the Gal4 AD, preventing recruitment of the transcriptional machinery (13-15). In the presence of galactose and ATP, Gal3 physically interacts with Gal80, and such interaction is required for relieving the Gal80 inhibition of Gal4, resulting in induced GAL gene expression (2, 26). However, how the Gal3-Gal80 interaction leads to Gal4 activation has been unresolved.
There are two contrasting bodies of evidence concerning how the Gal80-Gal3 interaction relieves Gal80 inhibition of Gal4. Historically, two independent studies led to the view that Gal80 remains associated with Gal4 at the promoter in galactose-induced cells (11, 21). The initial evidence came from experiments that used a Gal80-VP16 hybrid protein in which the transcriptional AD of VP16 was fused to Gal80. Gal80-VP16 was found to stimulate transcription of a GAL reporter gene in the presence of galactose, indicating that Gal80 stayed bound to Gal4 at the GAL gene promoter after induction (11). Later, others using a constitutive mutant of Gal3 protein at a 30× excess relative to Gal80 detected a complex of Gal3-Gal80-Gal4 associated with a UASGAL-containing DNA fragment in an electrophoresis mobility shift assay (21). Taken together, these data supported a nondissociation model proposing that galactose-activated Gal3 binds to Gal4-associated Gal80 at the GAL gene promoter in the nucleus and causes the Gal80-Gal4 complex to adopt a conformation that exposes the Gal4 AD (21).
Challenging the nondissociation model is a more recent body of evidence that points to dissociation of Gal80 from Gal4 after induction. Gal3 was detectable by two different methods only in the cytoplasm, and cells in which Gal3 was tethered to membranes outside the nucleus exhibited a magnitude of induction similar to that exhibited by wild-type cells (18). In addition, chromatin immunoprecipitation experiments revealed reduced binding of Gal80 to Gal4 after galactose induction (19). Accordingly, a dissociation model was proposed, in which the binding of Gal80 to cytoplasm-localized Gal3 results in a decrease of Gal80 content in the nucleus, leading to its dissociation from Gal4 (19). However, Gal80 dissociation from Gal4 has in turn been called into question by the report of a fluorescence resonance energy transfer (FRET) between Gal80-enhanced cyan fluorescent protein (ECFP) and Gal4-enhanced yellow fluorescent protein (EYFP) in galactose-grown yeast (1).
Here we present the results of new experiments aimed at resolving the conflicting models for how galactose triggers relief of Gal80 inhibition of Gal4. We demonstrate with the use of a GAL1 promoter-controlled reporter gene array and live-cell imaging that Gal80 rapidly dissociates from Gal4 in response to galactose. Our results further show that such dissociation depends on interaction between Gal3 and Gal80 and is temporally correlated with GAL reporter gene expression. We also find that Gal80 is able to reassociate with Gal4 when galactose is depleted and protein synthesis is blocked, suggesting that reversible binding of Gal80 by Gal3 contributes to the galactose-triggered Gal80-Gal4 dissociation event. We also detect a modest redistribution of Gal80 from the nucleus to the cytoplasm by 15 to 25 min following galactose addition. Finally, we provide here the first evidence that Gal3 is detectable within the nucleus before and after galactose addition. Based on these results, we conclude that the dissociation of Gal80 from Gal4 by Gal80 interaction with Gal3 is the event that initiates the active state of Gal4.
All yeast strains used in this study are haploid and were derived from S. cerevisiae ScTEB652 (a ade1 ile leu2-3,112 ura3-52 trp1-HIII his3-Δ1 MEL1), which is wild type for the GAL regulatory genes GAL4, GAL80, and GAL3 (2). Both ScTEB723 and ScTEB724, containing the wild-type GAL3 gene and a disrupted gal3 allele, respectively, were derived from ScTEB652 (2). ScPX729 was constructed from ScTEB723 by replacement of the chromosomal wild-type GAL80 allele with the noninducible GAL80S-2 allele, which encodes a Gal80 variant that does not interact with Gal3 (30). To measure Gal80-2YFP fluorescence intensity, the isogenic yeast strains ScTEB723 (GAL3), ScTEB724 (gal3Δ), and ScPX729 (GAL80S-2) were used to generate yeast strains ScBF856, ScBF858, and ScBF857, respectively, in which chromosomal GAL80 was tagged with two tandem copies of EmCitrine (2YFP) at its 3′ end. To investigate Gal3 subcellular localization, GAL3 in ScTEB723 was genomically tagged with two tandem copies of the green fluorescent protein gene (GFP) at its 3′ end, yielding the yeast strain ScFJ859. Plasmid pFJ35, expressing the mCherry-tagged histone H2B (H2B-mCherry) fusion protein as a nuclear marker, was constructed by joining the yeast ADH2 promoter, HTB2, and mCherry coding sequences in the CEN TRP1 plasmid pRS414 (from New England Biolabs).
The chromosomal “64× LacO array” was constructed as follows. A LacO-containing sequence, 5′-CCACAAATTGTTATCCGCTCACAATTCCACA-3′, was inserted into YIplac128 (GenBank accession no. X75463) and reiterated 64 times using a cloning strategy described in reference 22 to yield plasmid pFJ58×64. pFJ58×64 was then linearized at the EcoRV site in the LEU2 gene and integrated into the leu2 locus of ScTEB652, yielding ScFJ900. To construct the reporter array, the full-length GAL1 promoter containing four Gal4 binding sites was amplified from genomic DNA isolated from the yeast strain ScTEB652 using primers 5′TAACTCGGATCCGAGCCCCATTATCTTAGC-3′ and 5′-GGGTATAGTTTTTTCTCCTTGACGT-3′. The GAL1 promoter was then inserted 5′ to the glutathione S-transferase (GST) coding sequence followed by the ADH1 terminator sequence in plasmid pFA6a-kanMX6 (28) to yield plasmid pFJ57, containing the PGAL1-GST-TADH1 fusion. The reporter array was then constructed by reiterating the PGAL1-GST-TADH1 fusion eight times to yield plasmid pFJ57×8. pFJ57×8 was linearized and integrated into the chromosome about 2 kb upstream of the 64× LacO array in ScFJ900, resulting in yeast strain ScFJ902. GAL80, GAL3, and GAL4 were genomically tagged with 2YFP at their 3′ ends in ScFJ902 to generate ScFJ914, ScFJ925, and ScFJ927, respectively. ScFJ919 was made from ScFJ902 by replacing chromosomal GAL80 with GAL80S-2-2YFP. ScFJ935 (gal3Δ) was derived from ScFJ914 by disrupting the GAL3 gene with the N-acetyltransferase gene, which confers resistance to the antibiotic nourseothricin. RFP-LacI with a nuclear localization signal (8) was introduced as a point mutation, R197L, in LacI (25) and cloned downstream of the yeast ADH2 promoter in pRS414, resulting in plasmid pME08. In this work, the identities of all strains were verified by PCR and microscopy. The fluorescent-protein-tagged Gal80, Gal4, and Gal3 derivatives were tested for function in vivo using reporter assays, and they were functionally indistinguishable from the untagged proteins (data not shown).
Yeast cells were grown to log phase in selective synthetic media with noninducing carbon sources (2% glycerol, 3% lactic acid, and 1% raffinose). The culture was then split, and to one portion galactose was added to yield a 2% (wt/vol) final concentration. For the experiments with cycloheximide treatment, cycloheximide was added to the culture to a 25-μg/ml final concentration 10 min prior to addition of galactose. The inhibition of protein synthesis by cycloheximide was verified by a lack of GST or Gal80-2YFP expression in Western blot assays using anti-GST or anti-GFP (Roche). At various time points of induction, cells were transferred to agarose-coated slides, and images were acquired using a SlideBook v4.2-controlled Zeiss Axioplan II-based 3I imaging system equipped with a 63× objective lens. SlideBook v4.2 software running the ratio/FRET modules was used to obtain fluorescence intensity measurements on acquired images. The YFP and RFP filters that we used are components of the CFP/YFP-2x2M-A-HZ and GFP/DsRed-2x2M-B sets purchased from Semrock (Rochester, NY). These filters allow no bleed-through between the YFP and RFP channels. Spinning-disc confocal imaging (Perkin-Elmer) of Gal3-2GFP in yeast cells was carried out with a Nikon TE-2000U microscope with a 100× objective lens to determine Gal3 subcellular distribution. A minimum of 200 yeast cells were imaged for each set of fluorescence data presented in Fig. 1B and C (see also Fig. 3A to C, E, and F). Of the total cells, at least 50% showed both red (mRFP-lacI) and green/yellow (Gal80-2YFP or Gal4-2YFP) dots when Gal80 wild-type, Gal80S-2, or gal3Δ cells were imaged under either inducing or noninducing conditions. When Gal80-2YFP and Gal4-2YFP dots were monitored in the time course experiment (Fig. 2A and B), a minimum of 100 yeast cells were imaged at each time point.
For Gal80-2YFP fluorescence intensity measurements, fluorescent images were taken for both induced and noninduced cells of ScBF856, ScBF857, and ScBF858 expressing H2B-mCherry from pFJ35. For the induced cells, the time window for capturing images was 15 to 25 min after the addition of galactose. Each image used for quantitative measurements of fluorescent intensity was deconvoluted with SlideBook software. H2B-mCherry was used as a marker to define the nuclear region (N region), and the remainder of each cell was defined as the R region. Fluorescent intensity was measured in the N and R regions, and the intensity ratio (N/R) was calculated. The percent change in the ratio was defined as (ratio after induction − ratio before induction)/(ratio before induction) × 100%. The quantitative measurements were performed in at least three independent experiments for each yeast strain and were based on about 300 cells per experiment.
Yeast cells were grown in 25 ml of synthetic medium with noninducing carbon sources (2% glycerol, 3% lactic acid, and 1% raffinose). At an optical density at 600 nm (OD600) of 0.5 to ~0.7, 2% galactose was added. At various time points postinduction, cells with an OD600 of 1.5 were taken from the culture, chilled on an ice slurry, and spun down at 4°C. Protein extracts were prepared using the sodium hydroxide method, as described elsewhere (10). Proteins were separated with sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis, and GST was detected by Western blotting using anti-GST (Roche).
Arrays containing LacI or Gal4 binding sites have been successfully used in studying chromosome organization and Gal4-DNA interaction (17, 22). Fluorescent-protein-tagged LacI and Gal4 can bind to such arrays and form visible foci in the nuclei. In this study, we constructed a reporter array containing eight tandem units of PGAL1-GST, a GAL1 promoter-controlled GST reporter gene (Fig. (Fig.1A).1A). This array was inserted within 2 kb of a 64× LacO array located at the leu2 locus. To be visualized with a fluorescence microscope, endogenous GAL4 or GAL80 was tagged with two tandem copies of EmCitrine (2YFP) at its 3′ end, and mRFP-tagged LacI with a 3′ nuclear localization signal (LacI-mRFP) was carried in a centromeric plasmid. In cells expressing Gal4-2YFP and mRFP-LacI, we observed overlapping fluorescent green dots and red dots, confirming that the dots were indeed formed by reporter array-bound Gal4-2YFP and LacO array-bound mRFP-LacI (Fig. (Fig.1B).1B). It is well established that Gal80 associates with Gal4 and that Gal4 associates with DNA, but Gal80 does not itself bind to DNA (13, 21). When the yeast cells with the arrays, Gal80-2YFP and mRFP-LacI, were grown under noninducing conditions, Gal80-2YFP formed visible fluorescent dots on the reporter array, as indicated by overlap with the red mRFP-LacI dots (Fig. (Fig.1C).1C). Therefore, the mRFP-LacI dots serve to verify that dots of Gal80-2YFP and Gal4-2YFP are associated at the reporter array.
The nondissociation model specifies that the Gal3-Gal80 interaction causes Gal80 to shift its position from the Gal4 AD to another site on Gal4. In contrast, the dissociation model emphasizes that Gal80 leaves Gal4 in response to galactose. In order to discriminate the two models, we used our array system to monitor the Gal80 association with Gal4 in live cells before and after galactose induction.
We grew cells containing the arrays to mid-log phase, added galactose, and acquired a timed series of images. In the cells with Gal80-2YFP, the fluorescent intensity of the dots dramatically decreased within 10 min after the addition of galactose. By 15 min, the Gal80-2YFP dots had almost disappeared, while the mRFP-LacI dots persisted and served as a control for verifying that the position of the reporter array was in focus (Fig. (Fig.2A).2A). In contrast to the Gal80-2YFP dots, the Gal4-2YFP dots persisted, with only a slight reduction in intensity due to photo bleaching (Fig. (Fig.2B).2B). To determine whether Gal3 stably associates with Gal4-associated Gal80 at the array, we performed an identical time course experiment (3, 10, and 15 min after galactose addition) using cells expressing Gal3-2YFP. In this case we monitored a total of 287 cells, of which 65 showed the mRFP-LacI red dot. None of the cells showed a Gal3-2YFP dot (data not shown). These results from live cell imaging reveal no evidence of stable galactose-triggered association of Gal3 with Gal80 at the array but, on the contrary, show that Gal80 dissociates from Gal4 in response to galactose. Since the array contains eight copies of the GAL1 promoter-controlled GST gene, the GST expression serves as a sensitive reporter for GAL gene activation. By Western blot analyses, we detected a marked increase in GST protein levels by 15 min after the addition of galactose (Fig. (Fig.2C).2C). This time course of GST reporter expression in the array strain is similar to what has been observed for MEL1 and GAL1 gene expression in cells lacking the arrays (24, 31). These results show that galactose-triggered dissociation of Gal80 from Gal4 temporally correlates with activation of the GAL1 promoter.
Next we asked whether the dissociation of Gal80 from Gal4 on the reporter array depends on a Gal80-Gal3 interaction. In the otherwise wild-type cells bearing Gal80-2YFP and the arrays, we observed that after 30 min of galactose induction, the fluorescent dots completely disappeared (Fig. (Fig.3A).3A). In contrast, the Gal80-2YFP dots persisted after galactose induction in otherwise isogenic cells that either did not express Gal3 (gal3Δ) or expressed a non-Gal3-binding variant of Gal80 (Gal80S-2-2YFP) (30, 32) (Fig. 3B and C). Consistent with no dissociation of Gal80 from Gal4, no GST expression was detected after 1 h of galactose addition to the gal3Δ and GAL80S-2-2YFP cells (Fig. (Fig.3D).3D). These results demonstrate that Gal80 dissociation from Gal4 and GAL gene activation depend on Gal80-Gal3 interaction.
If the consequence of Gal80-Gal3 interaction is to sequester Gal80 from Gal4, we ought to observe reassociation of Gal80-2YFP with Gal4 as re-formation of fluorescent foci following the removal of galactose. Such an experiment requires the use of cycloheximide to prevent the normal galactose-induced increase in synthesis of the Gal80 protein (23). It was previously shown that the addition of cycloheximide to cells prior to galactose does not alter the GAL gene induction kinetics (20). When the cells with wild-type GAL3 were treated with cycloheximide followed by galactose, the Gal80-2YFP dots disappeared (Fig. (Fig.3E).3E). The cells were then removed from galactose medium and transferred to noninducing medium containing cycloheximide. After a 2-hour incubation, the Gal80-2YFP dots reappeared on the reporter array (Fig. (Fig.3F).3F). The somewhat-dimmer dot observed in Fig. Fig.3F3F is most likely due to a decrease in the Gal80 protein in cells held in cycloheximide for 2.5 to 3.0 h, as determined by Western blotting (data not shown). The position of the array corresponded to the mRFP-LacI dot (data not shown). These results demonstrate that throughout a period of approximately 3 h in the presence of cycloheximide, a significant pool of Gal3-sequestered Gal80 persists that can reassociate with Gal4 upon release from Gal3.
The previous finding that Gal3 was detectable only in the cytoplasm led to the expectation that Gal80 nuclear concentration will decrease in response to galactose (19). However, measurements of the nuclear concentration of Gal80 in the absence and presence of galactose have not been reported, and such measurements constitute a critical test of whether a redistribution of Gal80 from the nucleus to the cytoplasm is a feature of the GAL gene switch. To address this question, we monitored the subcellular distribution of Gal80-2YFP before and after galactose addition. To avoid changes in fluorescence due to newly synthesized Gal80-2YFP after galactose induction, control experiments were conducted in the presence of cycloheximide to block protein synthesis. When cells were grown in the absence of galactose, Gal80-2YFP appears to be predominantly localized in the nucleus, as indicated by colocalization with H2B-mCherry (Fig. (Fig.4A).4A). Simple viewing of such images revealed no apparent decrease of Gal80-2YFP nuclear intensity by 15 min after galactose addition. However, by 60 min, Gal80-2YFP appeared to be more uniformly distributed between the nucleus and the cytoplasm (Fig. (Fig.4A).4A). We then quantitatively determined the average fluorescence intensity in the nuclei and in the remainder of the cells by using H2B-mCherry to define the nuclear region. The intensity ratio, N/R, was taken as a measurement of Gal80 distribution between the nucleus and the cytoplasm. A decrease in this ratio in response to galactose is indicative of a relocation of Gal80 from the nucleus to the cytoplasm. We consistently observed a decrease in the ratio by approximately 15% in the wild-type GAL3 after 15 min of galactose induction, and a similar decrease was seen when the cells were treated with cycloheximide (Fig. (Fig.4B).4B). In contrast, isogenic cells expressing no Gal3 protein (gal3Δ) or the noninteracting variant Gal80 protein (GAL80S-2-2YFP) exhibited no such decrease (Fig. (Fig.4B).4B). Taken together, these results provide evidence for redistribution of Gal80 from the nucleus to the cytoplasm in response to a galactose-triggered Gal3-Gal80 interaction.
The Gal80 redistribution detected in response to the Gal3-Gal80 interaction was only approximately 15% by 15 to 25 min after galactose addition, which is the time by which we detected considerable dissociation of Gal80-2YFP from Gal4. This raised doubt as to whether such a small redistribution could be responsible for triggering the rapid dissociation of Gal80 from Gal4 in response to galactose. To address this issue, we tested whether a predominantly nucleus-localized Gal3 would impair the kinetics of induction. For this experiment we utilized NLS-Gal3, in which the simian virus 40 nuclear localization signal peptide was fused to the N terminus of Gal3. It was previously shown that NLS-Gal3 is predominantly nucleus localized and that such localization does not grossly impair induction as determined by cell growth at 4 days after galactose addition (18). To assess the possible effect of NLS-Gal3 on the kinetics of induction during the early phase, we monitored the kinetics of galactose-induced PGAL1-GST reporter expression by Western blotting in the cells with wild-type Gal3 or NLS-Gal3. The kinetics of GST expression in wild-type GAL3 and NLS-GAL3 cells were indistinguishable (Fig. (Fig.55).
We were surprised by the rather modest redistribution of Gal80 that we observed in response to the galactose-triggered Gal80-Gal3 interaction. If Gal3 is exclusively cytoplasmic, as we previously reported (18, 19), one might anticipate a larger subcellular redistribution of Gal80 from nucleus to cytoplasm in response to galactose. To reinvestigate whether some Gal3 resides in the nucleus, endogenous GAL3 was tagged with two copies of GFP to boost sensitivity of detection. To avoid the fluorescence bleeding from the cytoplasm to the nucleus, cells producing Gal3-2GFP were imaged with a spinning-disc confocal microscope. We observed that Gal3-2GFP appeared to be evenly distributed throughout the cells (i.e., not excluded from the nucleus) in the absence of galactose. By 60 min after galactose addition, we observed no change in the cellular distribution of Gal3 (Fig. (Fig.66).
In this study, we have demonstrated by real-time imaging of living cells that dissociation of Gal80 from the transcriptional activator Gal4 is evident beginning very early after galactose addition. Dissociation of Gal80 from Gal4 was dramatically evident in single cells and represented the predominant population behavior. Previously, it was shown by chromatin immunoprecipitation that a decrease in the association of Gal80 with Gal4 at the UASGAL site is evident 20 min following galactose addition (19). In that work, no earlier time points were assayed, and the 20-min time point has since been shown to be much later than the midpoint of recruitment of RNA polymerase by Gal4 (6.4 to 9.8 min) and the onset of GAL1 mRNA (4). Moreover, the earlier work (19) did not address whether or not reduced association of Gal80 and Gal4 was dependent on Gal3-Gal80 interaction, which is known to drive Gal4-mediated GAL gene activation. Importantly, in the work presented here, we monitored the earlier phase, after galactose addition, and observed a time course of galactose-induced Gal80 dissociation from Gal4 that correlates well with the time course of Gal4-mediated activation of GST reporter gene expression. As well, in the present work we showed that dissociation of Gal80 from Gal4 does not occur in live cells in response to galactose if we mutationally disable the Gal3-Gal80 interaction. Thus, dissociation of Gal80 from Gal4 requires the galactose-activated interaction between Gal3 and Gal80 that is known to be required for the relief of Gal80 inhibition of Gal4. The results from our fluorescence intensity measurements of Gal80-2YFP reveal a modest galactose-induced nucleus-to-cytoplasm redistribution of Gal80. As in the case of Gal80-Gal4 dissociation, this subcellular redistribution of Gal80 is not observed if the Gal3-Gal80 interaction is disabled. Also importantly, we show here that in the absence of new protein synthesis, Gal80 can repopulate promoter-associated Gal4 following the depletion of galactose, suggesting that reversible sequestration of Gal80 by Gal3 contributes to its dissociation from Gal4.
Taken together, our results provide direct evidence with live cells that the dissociation of Gal80 from Gal4 through reversible sequestration of Gal80 by Gal3 is an event that initiates the active state of Gal4 in the GAL gene switch. These data, as well as our data showing a reduction in the nuclear Gal80/cytoplasmic Gal80 ratio in response to the Gal3-Gal80 interaction, provide new direct evidence for the previously advanced dissociation model (19). Also in this work we were able with the aid of a double GFP tag and spinning-disc confocal imaging to detect for the first time the occurrence of Gal3 in both nucleus and cytoplasm in the absence of galactose.
Because a fundamental event driving Gal80-Gal4 dissociation, and thereby GAL gene activation, is the Gal80-Gal3 interaction, a major question remaining concerns whether Gal3 interacts with Gal4-bound Gal80. The nondissociation model specifies that Gal4 in complex with Gal80 (11) or a Gal3-Gal80-Gal4 complex (21) represents the persistently active form of Gal4 in galactose-induced cells. To address whether Gal3 does stably associate with Gal4-bound Gal80, we looked for an association of Gal3 with Gal4-bound Gal80 by imaging Gal3-2YFP or NLS-Gal3-GFP in live cells with the arrays. Neither Gal3-2YFP nor NLS-Gal3-GFP formed visible dots at the GAL promoter, even though with the latter construct Gal3 predominantly resides within the nucleus and shows strong nucleus-localized fluorescence (F. Jiang and J. E. Hopper, unpublished observations). While these negative results do not rule out the formation of a Gal3-Gal80-Gal4 complex, our results documenting dissociation of Gal80-2YFP from Gal4 in response to the galactose-dependent Gal3-Gal80 interaction weigh against the involvement of such a complex as a persistent feature of the galactose-induced state. We argue, then, that if the initiating event is the binding of galactose-activated Gal3 to Gal4-bound Gal80 at the promoter, the putative tripartite complex must be short-lived.
Previous attempts in this lab failed to detect Gal3 as Gal3-GFP in the nucleus in either the presence or absence of galactose (19). Recently, others were able to detect Gal3-YFP in the nucleus in the presence of galactose but not in the absence of galactose (29). Apparently, the level of Gal3 present in uninduced cells limits detection. However, in this work, for the first time, with the aid of brighter Gal3-2GFP and high-sensitivity confocal imaging, we did detect Gal3 in the nucleus prior to galactose addition. Thus, we presume that in galactose, Gal80-Gal3 association occurs in both nucleus and cytoplasm. Indeed, by 16 h following galactose addition, FRET between Gal80-CPF and Gal3-YPet is observed within both compartments (29). A nuclear pool of Gal3 in preinduced cells is expected to facilitate a rapid GAL gene response to galactose due to locally available Gal3 to compete with Gal4 for binding to Gal80. Such a scenario could also contribute to why we detect only a modest level of Gal80 redistribution after adding galactose. Given that the nuclear volume is only approximately 8% of the total cellular volume (9), there is within the nucleus compared to within the cytoplasm seemingly less Gal3 with which Gal80 can interact. In this light, we propose that the modest level of nuclear-to-cytoplasmic redistribution of Gal80 that we observe in response to galactose is simply a consequence of the Gal80-Gal3 interaction and their respective subcellular pools. The fact that the induction kinetics observed with NLSGal3 and wild-type Gal3 are similar is consistent with this view. We imagine that the nucleo-cytoplasmic shuttling of Gal80 and Gal80's binding to the larger pool of cytoplasmic Gal3 would likely, as previously proposed (18, 19), set up a linked equilibrium that favors maintenance of the induced state. Binding of Gal3 to Gal80 in both compartments could conceivably compete effectively with Gal4 binding to Gal80 and explain the dissociation of Gal80 from Gal4 that we document here.
This work highlights the use of fluorescence microscopy coupled with a locus-specific array in studying protein-protein interactions at gene promoters in live cells. Gal4 interacts with a wide variety of proteins, besides Gal80, that are implicated with recruitment of RNA polymerase II, including components of TATA-binding protein, SAGA, proteasome, and mediator complexes (27). This array system provides us a useful tool to study the interactions between Gal4 and its interacting partners and the process of recruitment in live cells. This approach can be readily extended to other regulatory systems that involve DNA-binding transcriptional factors.
We thank Andrew Belmont, Andrew Murray, and Cornel Fraefel for providing a 256× LacO array, GFP-LacI constructs, and mRFP-LacI constructs, respectively. We thank Anita Hopper and Gang Peng for critical reading of the manuscript.
This work was supported by grant GM027925 from the National Institutes of Health (to J.E.H.).
Published ahead of print on 3 August 2009.