Transcription factor clusters enable target search but do not contribute to target gene activation

Abstract

Many transcription factors (TFs) localize in nuclear clusters of locally increased concentrations, but how TF clustering is regulated and how it influences gene expression is not well understood. Here, we use quantitative microscopy in living cells to study the regulation and function of clustering of the budding yeast TF Gal4 in its endogenous context. Our results show that Gal4 forms clusters that overlap with the GAL loci. Cluster number, density and size are regulated in different growth conditions by the Gal4-inhibitor Gal80 and Gal4 concentration. Gal4 truncation mutants reveal that Gal4 clustering is facilitated by, but does not completely depend on DNA binding and intrinsically disordered regions. Moreover, we discover that clustering acts as a double-edged sword: self-interactions aid TF recruitment to target genes, but recruited Gal4 molecules that are not DNA-bound do not contribute to, and may even inhibit, transcription activation. We propose that cells need to balance the different effects of TF clustering on target search and transcription activation to facilitate proper gene expression.

Short tandem repeats bind transcription factors to tune eukaryotic gene expression

“…Our measurements can be explained by a simple four-state model showing that STRs enhance affinities by increasing the rate of DNA association. This is consistent with prior work suggesting that degenerate recognition sites may serve as “DNA antennae” to attract TFs to a particular regulatory site ( 85 , 112 – 115 ). This four-state model likely underestimates the true impacts of STRs on target search because it does not explicitly consider whether TFs can move from flanking STRs to a central motif through one-dimensional sliding, hopping, and intersegmental transfer ( 116 – 119 ), rather than dissociating, diffusing, and rebinding.…”

Transcription factor exchange enables prolonged transcriptional bursts

“…Conversely, even in this dense regime, Gal4 labeling efficiency was likely much less than 100% because of export of the JF646 dye from yeast cells. 25 The Gal4-HaloTag images only showed a few spots in the nucleus ( Figure 3 A) and did not resemble the images from Gal4-EGFP where all molecules are visible, 26 suggesting that there was additional DNA binding of non-labeled Gal4 molecules at the locus.

Figure 3 Simultaneous imaging of Gal4 binding and GAL10 transcription using focus feedback (A) Cropped example images of a GAL10 TS, tracked and imaged every 5 s with focus feedback (top), while simultaneously imaging Gal4 DNA binding at the locus (middle), shown every 100 s. Green and magenta circles indicate the found or interpolated positions of the spot.

…”

“…The diffraction limit and the physical distance between the PP7-PCP-GFPEnvy label and the TF binding sites made it impossible to resolve whether detected Gal4 molecules in the vicinity of GAL10 were actually bound to the TF binding site in the GAL1-GAL10 promoter or represented background binding, for example, to the downstream GAL7 promoter or another nearby gene, or perhaps residing in a Gal4 cluster close to TS. 26 For simplicity, we refer to all Gal4 molecules within 300 nm of the TS as bound to one of the four Gal4 binding sites in the GAL10 promoter. We also observed periods without detectable Gal4 binding, which may represent either periods without Gal4 binding or periods during which unlabeled Gal4 molecules were bound.…”

“…In a model without TF exchange (model 1), the burst duration would be independent of Gal4 concentration. In a previous study, 26 we reduced the Gal4 concentration by 50% by deleting one of the 2 GAL4 alleles in diploid yeast. Although this resulted in a slight but significant reduction of the burst duration, 26 the local Gal4 concentration at the GAL10 gene was likely not reduced to same degree as the global concentration because of concentration buffering by Gal4 clustering.…”

“…In a previous study, 26 we reduced the Gal4 concentration by 50% by deleting one of the 2 GAL4 alleles in diploid yeast. Although this resulted in a slight but significant reduction of the burst duration, 26 the local Gal4 concentration at the GAL10 gene was likely not reduced to same degree as the global concentration because of concentration buffering by Gal4 clustering. To further corroborate this connection, we reduced the Gal4 concentration even further by mutating the remaining GAL4 promoter 28 ( gal4Δ/pGAL4-mut ).…”