Divergence in DNA Specificity among Paralogous Transcription Factors Contributes to Their Differential In Vivo Binding

Widespread variation in molecular interactions and regulatory properties among transcription factor isoforms

Most human Transcription factors (TFs) genes encode multiple protein isoforms differing in DNA binding domains, effector domains, or other protein regions. The global extent to which this results in functional differences between isoforms remains unknown. Here, we systematically compared 693 isoforms of 246 TF genes, assessing DNA binding, protein binding, transcriptional activation, subcellular localization, and condensate formation. Relative to reference isoforms, two-thirds of alternative TF isoforms exhibit differences in one or more molecular activities, which often could not be predicted from sequence. We observed two primary categories of alternative TF isoforms: "rewirers" and "negative regulators", both of which were associated with differentiation and cancer. Our results support a model wherein the relative expression levels of, and interactions involving, TF isoforms add an understudied layer of complexity to gene regulatory networks, demonstrating the importance of isoform-aware characterization of TF functions and providing a rich resource for further studies.

High UV damage and low repair, but not cytosine deamination, stimulate mutation hotspots at ETS binding sites in melanoma

Noncoding mutation hotspots have been identified in melanoma and many of them occur at the binding sites of E26 transformation-specific (ETS) proteins; however, their formation mechanism and functional impacts are not fully understood. Here, we used UV (Ultraviolet) damage sequencing data and analyzed cyclobutane pyrimidine dimer (CPD) formation, DNA repair, and CPD deamination in human cells at single-nucleotide resolution. Our data show prominent CPD hotspots immediately after UV irradiation at ETS binding sites, particularly at sites with a conserved TTCCGG motif, which correlate with mutation hotspots identified in cutaneous melanoma. Additionally, CPDs are repaired slower at ETS binding sites than in flanking DNA. Cytosine deamination in CPDs to uracil is suggested as an important step for UV mutagenesis. However, we found that CPD deamination is significantly suppressed at ETS binding sites, particularly for the CPD hotspot on the 5' side of the ETS motif, arguing against a role for CPD deamination in promoting ETS-associated UV mutations. Finally, we analyzed a subset of frequently mutated promoters, including the ribosomal protein genes RPL13A and RPS20, and found that mutations in the ETS motif can significantly reduce the promoter activity. Thus, our data identify high UV damage and low repair, but not CPD deamination, as the main mechanism for ETS-associated mutations in melanoma and uncover important roles of often-overlooked mutation hotspots in perturbing gene transcription.

High-throughput data and modeling reveal insights into the mechanisms of cooperative DNA-binding by transcription factor proteins

Cooperative DNA-binding by transcription factor (TF) proteins is critical for eukaryotic gene regulation. In the human genome, many regulatory regions contain TF-binding sites in close proximity to each other, which can facilitate cooperative interactions. However, binding site proximity does not necessarily imply cooperative binding, as TFs can also bind independently to each of their neighboring target sites. Currently, the rules that drive cooperative TF binding are not well understood. In addition, it is oftentimes difficult to infer direct TF-TF cooperativity from existing DNA-binding data. Here, we show that in vitro binding assays using DNA libraries of a few thousand genomic sequences with putative cooperative TF-binding events can be used to develop accurate models of cooperativity and to gain insights into cooperative binding mechanisms. Using factors ETS1 and RUNX1 as our case study, we show that the distance and orientation between ETS1 sites are critical determinants of cooperative ETS1-ETS1 binding, while cooperative ETS1-RUNX1 interactions show more flexibility in distance and orientation and can be accurately predicted based on the affinity and sequence/shape features of the binding sites. The approach described here, combining custom experimental design with machine-learning modeling, can be easily applied to study the cooperative DNA-binding patterns of any TFs.

Short tandem repeats bind transcription factors to tune eukaryotic gene expression

Short tandem repeats (STRs) are enriched in eukaryotic cis-regulatory elements and alter gene expression, yet how they regulate transcription remains unknown. We found that STRs modulate transcription factor (TF)-DNA affinities and apparent on-rates by about 70-fold by directly binding TF DNA-binding domains, with energetic impacts exceeding many consensus motif mutations. STRs maximize the number of weakly preferred microstates near target sites, thereby increasing TF density, with impacts well predicted by statistical mechanics. Confirming that STRs also affect TF binding in cells, neural networks trained only on in vivo occupancies predicted effects identical to those observed in vitro. Approximately 90% of TFs preferentially bound STRs that need not resemble known motifs, providing a cis-regulatory mechanism to target TFs to genomic sites.

De novo distillation of thermodynamic affinity from deep learning regulatory sequence models of in vivo protein-DNA binding

Transcription factors (TF) are proteins that bind DNA in a sequence-specific manner to regulate gene transcription. Despite their unique intrinsic sequence preferences, in vivo genomic occupancy profiles of TFs differ across cellular contexts. Hence, deciphering the sequence determinants of TF binding, both intrinsic and context-specific, is essential to understand gene regulation and the impact of regulatory, non-coding genetic variation. Biophysical models trained on in vitro TF binding assays can estimate intrinsic affinity landscapes and predict occupancy based on TF concentration and affinity. However, these models cannot adequately explain context-specific, in vivo binding profiles. Conversely, deep learning models, trained on in vivo TF binding assays, effectively predict and explain genomic occupancy profiles as a function of complex regulatory sequence syntax, albeit without a clear biophysical interpretation. To reconcile these complementary models of in vitro and in vivo TF binding, we developed Affinity Distillation (AD), a method that extracts thermodynamic affinities de-novo from deep learning models of TF chromatin immunoprecipitation (ChIP) experiments by marginalizing away the influence of genomic sequence context. Applied to neural networks modeling diverse classes of yeast and mammalian TFs, AD predicts energetic impacts of sequence variation within and surrounding motifs on TF binding as measured by diverse in vitro assays with superior dynamic range and accuracy compared to motif-based methods. Furthermore, AD can accurately discern affinities of TF paralogs. Our results highlight thermodynamic affinity as a key determinant of in vivo binding, suggest that deep learning models of in vivo binding implicitly learn high-resolution affinity landscapes, and show that these affinities can be successfully distilled using AD. This new biophysical interpretation of deep learning models enables high-throughput in silico experiments to explore the influence of sequence context and variation on both intrinsic affinity and in vivo occupancy.

Double DAP-seq uncovered synergistic DNA binding of interacting bZIP transcription factors

Many eukaryotic transcription factors (TF) form homodimer or heterodimer complexes to regulate gene expression. Dimerization of BASIC LEUCINE ZIPPER (bZIP) TFs are critical for their functions, but the molecular mechanism underlying the DNA binding and functional specificity of homo- versus heterodimers remains elusive. To address this gap, we present the double DNA Affinity Purification-sequencing (dDAP-seq) technique that maps heterodimer binding sites on endogenous genomic DNA. Using dDAP-seq we profile twenty pairs of C/S1 bZIP heterodimers and S1 homodimers in Arabidopsis and show that heterodimerization significantly expands the DNA binding preferences of these TFs. Analysis of dDAP-seq binding sites reveals the function of bZIP9 in abscisic acid response and the role of bZIP53 heterodimer-specific binding in seed maturation. The C/S1 heterodimers show distinct preferences for the ACGT elements recognized by plant bZIPs and motifs resembling the yeast GCN4 cis-elements. This study demonstrates the potential of dDAP-seq in deciphering the DNA binding specificities of interacting TFs that are key for combinatorial gene regulation.

Flexibility of flanking DNA is a key determinant of transcription factor affinity for the core motif

Selective gene regulation is mediated by recognition of specific DNA sequences by transcription factors (TFs). The extremely challenging task of searching out specific cognate DNA binding sites among several million putative sites within the eukaryotic genome is achieved by complex molecular recognition mechanisms. Elements of this recognition code include the core binding sequence, the flanking sequence context, and the shape and conformational flexibility of the composite binding site. To unravel the extent to which DNA flexibility modulates TF binding, in this study, we employed experimentally guided molecular dynamics simulations of ternary complex of closely related Hox heterodimers Exd-Ubx and Exd-Scr with DNA. Results demonstrate that flexibility signatures embedded in the flanking sequences impact TF binding at the cognate binding site. A DNA sequence has intrinsic shape and flexibility features. While shape features are localized, our analyses reveal that flexibility features of the flanking sequences percolate several basepairs and allosterically modulate TF binding at the core. We also show that lack of flexibility in the motif context can render the cognate site resistant to protein-induced shape changes and subsequently lower TF binding affinity. Overall, this study suggests that flexibility-guided DNA shape, and not merely the static shape, is a key unexplored component of the complex DNA-TF recognition code.

Context-dependent CpG methylation directs cell-specific binding of transcription factor ZBTB38

DNA methylation on CpGs regulates transcription in mammals, both by decreasing the binding of methylation-repelled factors and by increasing the binding of methylation-attracted factors. Among the latter, zinc finger proteins have the potential to bind methylated CpGs in a sequence-specific context. The protein ZBTB38 is unique in that it has two independent sets of zinc fingers, which recognize two different methylated consensus sequences in vitro. Here, we identify the binding sites of ZBTB38 in a human cell line, and show that they contain the two methylated consensus sequences identified in vitro. In addition, we show that the distribution of ZBTB38 sites is highly unusual: while 10% of the ZBTB38 sites are also bound by CTCF, the other 90% of sites reside in closed chromatin and are not bound by any of the other factors mapped in our model cell line. Finally, a third of ZBTB38 sites are found upstream of long and active CpG islands. Our work therefore validates ZBTB38 as a methyl-DNA binder in vivo and identifies its unique distribution in the genome.

Elucidating the biology of transcription factor-DNA interaction for accurate identification of cis-regulatory elements

Transcription factors (TFs) play a critical role in determining cell fate decisions by integrating developmental and environmental signals through binding to specific cis-regulatory modules and regulating spatio-temporal specificity of gene expression patterns. Precise identification of functional TF binding sites in time and space not only will revolutionize our understanding of regulatory networks governing cell fate decisions but is also instrumental to uncover how genetic variations cause morphological diversity or disease. In this review, we discuss recent advances in mapping TF binding sites and characterizing the various parameters underlying the complexity of binding site recognition by TFs.

Evolution of binding preferences among whole-genome duplicated transcription factors

Throughout evolution, new transcription factors (TFs) emerge by gene duplication, promoting growth and rewiring of transcriptional networks. How TF duplicates diverge was studied in a few cases only. To provide a genome-scale view, we considered the set of budding yeast TFs classified as whole-genome duplication (WGD)-retained paralogs (~35% of all specific TFs). Using high-resolution profiling, we find that ~60% of paralogs evolved differential binding preferences. We show that this divergence results primarily from variations outside the DNA-binding domains (DBDs), while DBD preferences remain largely conserved. Analysis of non-WGD orthologs revealed uneven splitting of ancestral preferences between duplicates, and the preferential acquiring of new targets by the least conserved paralog (biased neo/sub-functionalization). Interactions between paralogs were rare, and, when present, occurred through weak competition for DNA-binding or dependency between dimer-forming paralogs. We discuss the implications of our findings for the evolutionary design of transcriptional networks.

Cis-regulatory sequences in plants: Their importance, discovery, and future challenges

The identification and characterization of cis-regulatory DNA sequences and how they function to coordinate responses to developmental and environmental cues is of paramount importance to plant biology. Key to these regulatory processes are cis-regulatory modules (CRMs), which include enhancers and silencers. Despite the extraordinary advances in high-quality sequence assemblies and genome annotations, the identification and understanding of CRMs, and how they regulate gene expression, lag significantly behind. This is especially true for their distinguishing characteristics and activity states. Here, we review the current knowledge on CRMs and breakthrough technologies enabling identification, characterization, and validation of CRMs; we compare the genomic distributions of CRMs with respect to their target genes between different plant species, and discuss the role of transposable elements harboring CRMs in the evolution of gene expression. This is an exciting time to study cis-regulomes in plants; however, significant existing challenges need to be overcome to fully understand and appreciate the role of CRMs in plant biology and in crop improvement.

Competition for DNA binding between paralogous transcription factors determines their genomic occupancy and regulatory functions

Most eukaryotic transcription factors (TFs) are part of large protein families, with members of the same family (i.e., paralogous TFs) recognizing similar DNA-binding motifs but performing different regulatory functions. Many TF paralogs are coexpressed in the cell and thus can compete for target sites across the genome. However, this competition is rarely taken into account when studying the in vivo binding patterns of eukaryotic TFs. Here, we show that direct competition for DNA binding between TF paralogs is a major determinant of their genomic binding patterns. Using yeast proteins Cbf1 and Pho4 as our model system, we designed a high-throughput quantitative assay to capture the genomic binding profiles of competing TFs in a cell-free system. Our data show that Cbf1 and Pho4 greatly influence each other's occupancy by competing for their common putative genomic binding sites. The competition is different at different genomic sites, as dictated by the TFs' expression levels and their divergence in DNA-binding specificity and affinity. Analyses of ChIP-seq data show that the biophysical rules that dictate the competitive TF binding patterns in vitro are also followed in vivo, in the complex cellular environment. Furthermore, the Cbf1-Pho4 competition for genomic sites, as characterized in vitro using our new assay, plays a critical role in the specific activation of their target genes in the cell. Overall, our study highlights the importance of direct TF-TF competition for genomic binding and gene regulation by TF paralogs, and proposes an approach for studying this competition in a quantitative and high-throughput manner.

Androgen and glucocorticoid receptor direct distinct transcriptional programs by receptor-specific and shared DNA binding sites

The glucocorticoid (GR) and androgen (AR) receptors execute unique functions in vivo, yet have nearly identical DNA binding specificities. To identify mechanisms that facilitate functional diversification among these transcription factor paralogs, we studied them in an equivalent cellular context. Analysis of chromatin and sequence suggest that divergent binding, and corresponding gene regulation, are driven by different abilities of AR and GR to interact with relatively inaccessible chromatin. Divergent genomic binding patterns can also be the result of subtle differences in DNA binding preference between AR and GR. Furthermore, the sequence composition of large regions (>10 kb) surrounding selectively occupied binding sites differs significantly, indicating a role for the sequence environment in guiding AR and GR to distinct binding sites. The comparison of binding sites that are shared shows that the specificity paradox can also be resolved by differences in the events that occur downstream of receptor binding. Specifically, shared binding sites display receptor-specific enhancer activity, cofactor recruitment and changes in histone modifications. Genomic deletion of shared binding sites demonstrates their contribution to directing receptor-specific gene regulation. Together, these data suggest that differences in genomic occupancy as well as divergence in the events that occur downstream of receptor binding direct functional diversification among transcription factor paralogs.

AP-1 subunits converge promiscuously at enhancers to potentiate transcription

The AP-1 transcription factor (TF) dimer contributes to many biological processes and environmental responses. AP-1 can be composed of many interchangeable subunits. Unambiguously determining the binding locations of these subunits in the human genome is challenging because of variable antibody specificity and affinity. Here, we definitively establish the genome-wide binding patterns of five AP-1 subunits by using CRISPR to introduce a common antibody tag on each subunit. We find limited evidence for strong dimerization preferences between subunits at steady state and find that, under a stimulus, dimerization patterns reflect changes in the transcriptome. Further, our analysis suggests that canonical AP-1 motifs indiscriminately recruit all AP-1 subunits to genomic sites, which we term AP-1 hotspots. We find that AP-1 hotspots are predictive of cell type–specific gene expression and of genomic responses to glucocorticoid signaling (more so than super-enhancers) and are significantly enriched in disease-associated genetic variants. Together, these results support a model where promiscuous binding of many AP-1 subunits to the same genomic location play a key role in regulating cell type–specific gene expression and environmental responses.

Speed-Specificity Trade-Offs in the Transcription Factors Search for Their Genomic Binding Sites

Transcription factors (TFs) regulate gene expression by binding DNA sequences recognized by their DNA-binding domains (DBDs). DBD-recognized motifs are short and highly abundant in genomes. The ability of TFs to bind a specific subset of motif-containing sites, and to do so rapidly upon activation, is fundamental for gene expression in all eukaryotes. Despite extensive interest, our understanding of the TF-target search process is fragmented; although binding specificity and detection speed are two facets of this same process, trade-offs between them are rarely addressed. In this opinion article, we discuss potential speed-specificity trade-offs in the context of existing models. We further discuss the recently described 'distributed specificity' paradigm, suggesting that intrinsically disordered regions (IDRs) promote specificity while reducing the TF-target search time.

DNA mismatches reveal conformational penalties in protein-DNA recognition

Transcription factors recognize specific genomic sequences to regulate complex gene-expression programs. Although it is well-established that transcription factors bind to specific DNA sequences using a combination of base readout and shape recognition, some fundamental aspects of protein-DNA binding remain poorly understood1,2. Many DNA-binding proteins induce changes in the structure of the DNA outside the intrinsic B-DNA envelope. However, how the energetic cost that is associated with distorting the DNA contributes to recognition has proven difficult to study, because the distorted DNA exists in low abundance in the unbound ensemble3-9. Here we use a high-throughput assay that we term SaMBA (saturation mismatch-binding assay) to investigate the role of DNA conformational penalties in transcription factor-DNA recognition. In SaMBA, mismatched base pairs are introduced to pre-induce structural distortions in the DNA that are much larger than those induced by changes in the Watson-Crick sequence. Notably, approximately 10% of mismatches increased transcription factor binding, and for each of the 22 transcription factors that were examined, at least one mismatch was found that increased the binding affinity. Mismatches also converted non-specific sites into high-affinity sites, and high-affinity sites into 'super sites' that exhibit stronger affinity than any known canonical binding site. Determination of high-resolution X-ray structures, combined with nuclear magnetic resonance measurements and structural analyses, showed that many of the DNA mismatches that increase binding induce distortions that are similar to those induced by protein binding-thus prepaying some of the energetic cost incurred from deforming the DNA. Our work indicates that conformational penalties are a major determinant of protein-DNA recognition, and reveals mechanisms by which mismatches can recruit transcription factors and thus modulate replication and repair activities in the cell10,11.

Chromatin Environment and Cellular Context Specify Compensatory Activity of Paralogous MEF2 Transcription Factors

Compensation among paralogous transcription factors (TFs) confers genetic robustness of cellular processes, but how TFs dynamically respond to paralog depletion on a genome-wide scale in vivo remains incompletely understood. Using single and double conditional knockout of myocyte enhancer factor 2 (MEF2) family TFs in granule neurons of the mouse cerebellum, we find that MEF2A and MEF2D play functionally redundant roles in cerebellar-dependent motor learning. Although both TFs are highly expressed in granule neurons, transcriptomic analyses show MEF2D is the predominant genomic regulator of gene expression in vivo. Strikingly, genome-wide occupancy analyses reveal upon depletion of MEF2D, MEF2A occupancy robustly increases at a subset of sites normally bound to MEF2D. Importantly, sites experiencing compensatory MEF2A occupancy are concentrated within open chromatin and undergo functional compensation for genomic activation and gene expression. Finally, motor activity induces a switch from non-compensatory to compensatory MEF2-dependent gene regulation. These studies uncover genome-wide functional interdependency between paralogous TFs in the brain.

Majidi et al. study how transcription factors respond to paralog depletion by conditionally depleting MEF2A and MEF2D in mouse cerebellum. Depletion of MEF2D induces functionally compensatory genomic occupancy by MEF2A. Compensation occurs within accessible chromatin in a context-dependent manner. This study explores the interdependency between paralogous transcription factors.

Quantitative analysis of transcription factor binding and expression using calling cards reporter arrays

We report a tool, Calling Cards Reporter Arrays (CCRA), that measures transcription factor (TF) binding and the consequences on gene expression for hundreds of synthetic promoters in yeast. Using Cbf1p and MAX, we demonstrate that the CCRA method is able to detect small changes in binding free energy with a sensitivity comparable to in vitro methods, enabling the measurement of energy landscapes in vivo. We then demonstrate the quantitative analysis of cooperative interactions by measuring Cbf1p binding at synthetic promoters with multiple sites. We find that the cooperativity between Cbf1p dimers varies sinusoidally with a period of 10.65 bp and energetic cost of 1.37 K_BT for sites that are positioned ‘out of phase’. Finally, we characterize the binding and expression of a group of TFs, Tye7p, Gcr1p and Gcr2p, that act together as a ‘TF collective’, an important but poorly characterized model of TF cooperativity. We demonstrate that Tye7p often binds promoters without its recognition site because it is recruited by other collective members, whereas these other members require their recognition sites, suggesting a hierarchy where these factors recruit Tye7p but not vice versa. Our experiments establish CCRA as a useful tool for quantitative investigations into TF binding and function.

Sequence and chromatin determinants of transcription factor binding and the establishment of cell type-specific binding patterns

Transcription factors (TFs) selectively bind distinct sets of sites in different cell types. Such cell type-specific binding specificity is expected to result from interplay between the TF's intrinsic sequence preferences, cooperative interactions with other regulatory proteins, and cell type-specific chromatin landscapes. Cell type-specific TF binding events are highly correlated with patterns of chromatin accessibility and active histone modifications in the same cell type. However, since concurrent chromatin may itself be a consequence of TF binding, chromatin landscapes measured prior to TF activation provide more useful insights into how cell type-specific TF binding events became established in the first place. Here, we review the various sequence and chromatin determinants of cell type-specific TF binding specificity. We identify the current challenges and opportunities associated with computational approaches to characterizing, imputing, and predicting cell type-specific TF binding patterns. We further focus on studies that characterize TF binding in dynamic regulatory settings, and we discuss how these studies are leading to a more complex and nuanced understanding of dynamic protein-DNA binding activities. We propose that TF binding activities at individual sites can be viewed along a two-dimensional continuum of local sequence and chromatin context. Under this view, cell type-specific TF binding activities may result from either strongly favorable sequence features or strongly favorable chromatin context.

QBiC-Pred: quantitative predictions of transcription factor binding changes due to sequence variants

Non-coding genetic variants/mutations can play functional roles in the cell by disrupting regulatory interactions between transcription factors (TFs) and their genomic target sites. For most human TFs, a myriad of DNA-binding models are available and could be used to predict the effects of DNA mutations on TF binding. However, information on the quality of these models is scarce, making it hard to evaluate the statistical significance of predicted binding changes. Here, we present QBiC-Pred, a web server for predicting quantitative TF binding changes due to nucleotide variants. QBiC-Pred uses regression models of TF binding specificity trained on high-throughput in vitro data. The training is done using ordinary least squares (OLS), and we leverage distributional results associated with OLS estimation to compute, for each predicted change in TF binding, a P-value reflecting our confidence in the predicted effect. We show that OLS models are accurate in predicting the effects of mutations on TF binding in vitro and in vivo, outperforming widely-used PWM models as well as recently developed deep learning models of specificity. QBiC-Pred takes as input mutation datasets in several formats, and it allows post-processing of the results through a user-friendly web interface. QBiC-Pred is freely available at http://qbic.genome.duke.edu.

The MYCL and MXD1 transcription factors regulate the fitness of murine dendritic cells

We previously found that MYCL is required by a Batf3-dependent classical dendritic cell subset (cDC1) for optimal CD8 T cell priming, but the underlying mechanism has remained unclear. The MAX-binding proteins encompass a family of transcription factors with overlapping DNA-binding specificities, conferred by a C-terminal basic helix-loop-helix domain, which mediates heterodimerization. Thus, regulation of transcription by these factors is dependent on divergent N-terminal domains. The MYC family, including MYCL, has actions that are reciprocal to the MXD family, which is mediated through the recruitment of higher-order activator and repressor complexes, respectively. As potent proto-oncogenes, models of MYC family function have been largely derived from their activity at supraphysiological levels in tumor cell lines. MYC and MYCN have been studied extensively, but empirical analysis of MYCL function had been limited due to highly restricted, lineage-specific expression in vivo. Here we observed that Mycl is expressed in immature cDC1s but repressed on maturation, concomitant with Mxd1 induction in mature cDC1s. We hypothesized that MYCL and MXD1 regulate a shared, but reciprocal, transcriptional program during cDC1 maturation. In agreement, immature cDC1s in Mycl ^-/- -deficient mice exhibited reduced expression of genes that regulate core biosynthetic processes. Mature cDC1s from Mxd1 ^-/- mice exhibited impaired ability to inhibit the transcriptional signature otherwise supported by MYCL. The present study reveals LMYC and MXD1 as regulators of a transcriptional program that is modulated during the maturation of Batf3-dependent cDC1s.

HCF-2 inhibits cell proliferation and activates differentiation-gene expression programs

HCF-2 is a member of the host-cell-factor protein family, which arose in early vertebrate evolution as a result of gene duplication. Whereas its paralog, HCF-1, is known to act as a versatile chromatin-associated protein required for cell proliferation and differentiation, much less is known about HCF-2. Here, we show that HCF-2 is broadly present in human and mouse cells, and possesses activities distinct from HCF-1. Unlike HCF-1, which is excluded from nucleoli, HCF-2 is nucleolar—an activity conferred by one and a half C-terminal Fibronectin type 3 repeats and inhibited by the HCF-1 nuclear localization signal. Elevated HCF-2 synthesis in HEK-293 cells results in phenotypes reminiscent of HCF-1-depleted cells, including inhibition of cell proliferation and mitotic defects. Furthermore, increased HCF-2 levels in HEK-293 cells lead to inhibition of cell proliferation and metabolism gene-expression programs with parallel activation of differentiation and morphogenesis gene-expression programs. Thus, the HCF ancestor appears to have evolved into a small two-member protein family possessing contrasting nuclear versus nucleolar localization, and cell proliferation and differentiation functions.

BindSpace decodes transcription factor binding signals by large-scale sequence embedding

Decoding transcription factor (TF) binding signals in genomic DNA is a fundamental problem. Here we present a prediction model called BindSpace that learns to embed DNA sequences and TF class/family labels into the same space. By training on binding data for hundreds of TFs and embedding over 1M DNA sequences, BindSpace achieves state-of-the-art multiclass binding prediction performance, in vitro and in vivo, and can distinguish signals of closely related TFs.

Homotypic cooperativity and collective binding are determinants of bHLH specificity and function

Eukaryotic cells express transcription factor (TF) paralogues that bind to nearly identical DNA sequences in vitro but bind at different genomic loci and perform different functions in vivo. Predicting how 2 paralogous TFs bind in vivo using DNA sequence alone is an important open problem. Here, we analyzed 2 yeast bHLH TFs, Cbf1p and Tye7p, which have highly similar binding preferences in vitro, yet bind at almost completely nonoverlapping target loci in vivo. We dissected the determinants of specificity for these 2 proteins by making a number of chimeric TFs in which we swapped different domains of Cbf1p and Tye7p and determined the effects on in vivo binding and cellular function. From these experiments, we learned that the Cbf1p dimer achieves its specificity by binding cooperatively with other Cbf1p dimers bound nearby. In contrast, we found that Tye7p achieves its specificity by binding cooperatively with 3 other DNA-binding proteins, Gcr1p, Gcr2p, and Rap1p. Remarkably, most promoters (63%) that are bound by Tye7p do not contain a consensus Tye7p binding site. Using this information, we were able to build simple models to accurately discriminate bound and unbound genomic loci for both Cbf1p and Tye7p. We then successfully reprogrammed the human bHLH NPAS2 to bind Cbf1p in vivo targets and a Tye7p target intergenic region to be bound by Cbf1p. These results demonstrate that the genome-wide binding targets of paralogous TFs can be discriminated using sequence information, and provide lessons about TF specificity that can be applied across the phylogenetic tree.

Low-Affinity Binding Sites and the Transcription Factor Specificity Paradox in Eukaryotes

Eukaryotic transcription factors (TFs) from the same structural family tend to bind similar DNA sequences, despite the ability of these TFs to execute distinct functions in vivo. The cell partly resolves this specificity paradox through combinatorial strategies and the use of low-affinity binding sites, which are better able to distinguish between similar TFs. However, because these sites have low affinity, it is challenging to understand how TFs recognize them in vivo. Here, we summarize recent findings and technological advancements that allow for the quantification and mechanistic interpretation of TF recognition across a wide range of affinities. We propose a model that integrates insights from the fields of genetics and cell biology to provide further conceptual understanding of TF binding specificity. We argue that in eukaryotes, target specificity is driven by an inhomogeneous 3D nuclear distribution of TFs and by variation in DNA binding affinity such that locally elevated TF concentration allows low-affinity binding sites to be functional.