Glucokinase (GCK) in diabetes: from molecular mechanisms to disease pathogenesis.

Glucokinase (GCK), a key enzyme in glucose metabolism, plays a central role in glucose sensing and insulin secretion in pancreatic ß-cells, as well as glycogen synthesis in the liver. Mutations in the GCK gene have been associated with various monogenic diabetes (MD) disorders, including permanent neonatal diabetes mellitus (PNDM) and maturity-onset diabetes of the young (MODY), highlighting its importance in maintaining glucose homeostasis. Additionally, GCK gain-of-function mutations lead to a rare congenital form of hyperinsulinism known as hyperinsulinemic hypoglycemia (HH), characterized by increased enzymatic activity and increased glucose sensitivity in pancreatic ß-cells. This review offers a comprehensive exploration of the critical role played by the GCK gene in diabetes development, shedding light on its expression patterns, regulatory mechanisms, and diverse forms of associated monogenic disorders. Structural and mechanistic insights into GCK's involvement in glucose metabolism are discussed, emphasizing its significance in insulin secretion and glycogen synthesis. Animal models have provided valuable insights into the physiological consequences of GCK mutations, although challenges remain in accurately recapitulating human disease phenotypes. In addition, the potential of human pluripotent stem cell (hPSC) technology in overcoming current model limitations is discussed, offering a promising avenue for studying GCK-related diseases at the molecular level. Ultimately, a deeper understanding of GCK's multifaceted role in glucose metabolism and its dysregulation in disease states holds implications for developing targeted therapeutic interventions for diabetes and related disorders.

Long non-coding RNA and RNA-binding protein interactions in cancer: Experimental and machine learning approaches.

Understanding the complex and specific roles played by non-coding RNAs (ncRNAs), which comprise the bulk of the genome, is important for understanding virtually every hallmark of cancer. This large group of molecules plays pivotal roles in key regulatory mechanisms in various cellular processes. Regulatory mechanisms, mediated by long non-coding RNA (lncRNA) and RNA-binding protein (RBP) interactions, are well documented in several types of cancer. Their effects are enabled through networks affecting lncRNA and RBP stability, RNA metabolism including N6-methyladenosine (m6A) and alternative splicing, subcellular localization, and numerous other mechanisms involved in cancer. In this review, we discuss the reciprocal interplay between lncRNAs and RBPs and their involvement in epigenetic regulation via histone modifications, as well as their key role in resistance to cancer therapy. Other aspects of RBPs including their structural domains, provide a deeper knowledge on how lncRNAs and RBPs interact and exert their biological functions. In addition, current state-of-the-art knowledge, facilitated by machine and deep learning approaches, unravels such interactions in better details to further enhance our understanding of the field, and the potential to harness RNA-based therapeutics as an alternative treatment modality for cancer are discussed.

A modeling framework for embedding-based predictions for compound-viral protein activity.

MOTIVATION: A global effort is underway to identify compounds for the treatment of COVID-19. Since de novo compound design is an extremely long, time-consuming and expensive process, efforts are underway to discover existing compounds that can be repurposed for COVID-19 and new viral diseases.We propose a machine learning representation framework that uses deep learning induced vector embeddings of compounds and viral proteins as features to predict compound-viral protein activity. The prediction model in-turn uses a consensus framework to rank approved compounds against viral proteins of interest. RESULTS: Our consensus framework achieves a high mean Pearson correlation of 0.916, mean R2 of 0.840 and a low mean squared error of 0.313 for the task of compound-viral protein activity prediction on an independent test set. As a use case, we identify a ranked list of 47 compounds common to three main proteins of SARS-COV-2 virus (PL-PRO, 3CL-PRO and Spike protein) as potential targets including 21 antivirals, 15 anticancer, 5 antibiotics and 6 other investigational human compounds. We perform additional molecular docking simulations to demonstrate that majority of these compounds have low binding energies and thus high binding affinity with the potential to be effective against the SARS-COV-2 virus. AVAILABILITY AND IMPLEMENTATION: All the source code and data is available at: https://github.com/raghvendra5688/Drug-Repurposing and https://dx.doi.org/10.17632/8rrwnbcgmx.3. We also implemented a web-server at: https://machinelearning-protein.qcri.org/index.html. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Flavin Adenine Dinucleotide (FAD) and Pyridoxal 5'-Phosphate (PLP) Bind to Sox9 and Alter the Expression of Key Pancreatic Progenitor Transcription Factors.

Cofactor flavin adenine dinucleotide (FAD), a compound with flavin moiety and a derivative of riboflavin (vitamin B2), is shown to bind to Sox9 (a key transcription factor in early pancreatic development) and, subsequently, induce a large increase in markers of pancreatic development, including Ngn3 and PTF1a. Pyridoxal 5'-phosphate (PLP), the active form of vitamin B6, also binds to Sox9 and results in a similar increase in pancreatic development markers. Sox9 is known to be specifically important for pancreatic progenitors. Previously, there was no known link between FAD, PLP, or other co-factors and Sox9 for function. Thus, our findings show the mechanism by which FAD and PLP interact with Sox9 and result in the altered expression of pancreatic progenitor transcription factors involved in the pancreas development.

Molecular Analysis and Conformational Dynamics of Human MC4R Disease-Causing Mutations.

Obesity is a chronic disease with increasing cases among children and adolescents. Melanocortin 4 receptor (MC4R) is a G protein-coupled transporter involved in solute transport, enabling it to maintain cellular homeostasis. MC4R mutations are associated with early-onset severe obesity, and the identification of potential pathological variants is crucial for the clinical management of patients with obesity. A number of mutations have been reported in MC4R that are responsible for causing obesity and related complications. Delineating these mutations and analyzing their effect on MC4R's structure will help in the clinical intervention of the disease condition as well as designing potential drugs against it. Sequence-based pathogenicity and structure-based protein stability analyses were conducted on naturally occurring variants. We used computational tools to analyze the conservation of these mutations on MC4R's structure to map the structural variations. Detailed structural analyses were carried out for the active site mutations (i.e., D122N, D126Y, and S188L) and their influence on the binding of calcium and the agonist or antagonist. We performed molecular dynamics (MD) simulations of the wild-type and selected mutations to delineate the conformational changes, which provided us with possible reasons for MC4R's instability in these mutations. This study provides insight into the potential direction toward understanding the molecular basis of MC4R dysfunction in disease progression and obesity.

BCrystal: an interpretable sequence-based protein crystallization predictor.

MOTIVATION: X-ray crystallography has facilitated the majority of protein structures determined to date. Sequence-based predictors that can accurately estimate protein crystallization propensities would be highly beneficial to overcome the high expenditure, large attrition rate, and to reduce the trial-and-error settings required for crystallization. RESULTS: In this study, we present a novel model, BCrystal, which uses an optimized gradient boosting machine (XGBoost) on sequence, structural and physio-chemical features extracted from the proteins of interest. BCrystal also provides explanations, highlighting the most important features for the predicted crystallization propensity of an individual protein using the SHAP algorithm. On three independent test sets, BCrystal outperforms state-of-the-art sequence-based methods by more than 12.5% in accuracy, 18% in recall and 0.253 in Matthew's correlation coefficient, with an average accuracy of 93.7%, recall of 96.63% and Matthew's correlation coefficient of 0.868. For relative solvent accessibility of exposed residues, we observed higher values to associate positively with protein crystallizability and the number of disordered regions, fraction of coils and tripeptide stretches that contain multiple histidines associate negatively with crystallizability. The higher accuracy of BCrystal enables it to accurately screen for sequence variants with enhanced crystallizability. AVAILABILITY AND IMPLEMENTATION: Our BCrystal webserver is at https://machinelearning-protein.qcri.org/ and source code is available at https://github.com/raghvendra5688/BCrystal. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

DeepCrystal: a deep learning framework for sequence-based protein crystallization prediction.

MOTIVATION: Protein structure determination has primarily been performed using X-ray crystallography. To overcome the expensive cost, high attrition rate and series of trial-and-error settings, many in-silico methods have been developed to predict crystallization propensities of proteins based on their sequences. However, the majority of these methods build their predictors by extracting features from protein sequences, which is computationally expensive and can explode the feature space. We propose DeepCrystal, a deep learning framework for sequence-based protein crystallization prediction. It uses deep learning to identify proteins which can produce diffraction-quality crystals without the need to manually engineer additional biochemical and structural features from sequence. Our model is based on convolutional neural networks, which can exploit frequently occurring k-mers and sets of k-mers from the protein sequences to distinguish proteins that will result in diffraction-quality crystals from those that will not. RESULTS: Our model surpasses previous sequence-based protein crystallization predictors in terms of recall, F-score, accuracy and Matthew's correlation coefficient (MCC) on three independent test sets. DeepCrystal achieves an average improvement of 1.4, 12.1% in recall, when compared to its closest competitors, Crysalis II and Crysf, respectively. In addition, DeepCrystal attains an average improvement of 2.1, 6.0% for F-score, 1.9, 3.9% for accuracy and 3.8, 7.0% for MCC w.r.t. Crysalis II and Crysf on independent test sets. AVAILABILITY AND IMPLEMENTATION: The standalone source code and models are available at https://github.com/elbasir/DeepCrystal and a web-server is also available at https://deeplearning-protein.qcri.org. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Identification of a Wells-Dawson polyoxometalate-based AP-2γ inhibitor with pro-apoptotic activity.

AP-2 gamma (AP-2γ) is a transcription factor that plays pivotal roles in breast cancer biology. To search for small molecule inhibitors of AP-2γ, we performed a high-throughput fluorescence anisotropy screen and identified a polyoxometalate compound with Wells-Dawson structure K6[P2Mo18O62] (Dawson-POM) that blocks the DNA-binding activity of AP-2γ. We showed that this blocking activity is due to the direct binding of Dawson-POM to AP-2γ. We also provided evidence to show that Dawson-POM decreases AP-2γ-dependent transcription similar to silencing the gene. Finally, we demonstrated that Dawson-POM contains anti-proliferative and pro-apoptotic effects in breast cancer cells. In summary, we identified the first small molecule inhibitor of AP-2γ and showed Dawson-POM-mediated inhibition of AP-2γ as a potential avenue for cancer therapy.

Structure and decoy-mediated inhibition of the SOX18/Prox1-DNA interaction.

The transcription factor (TF) SOX18 drives lymphatic vessel development in both embryogenesis and tumour-induced neo-lymphangiogenesis. Genetic disruption of Sox18 in a mouse model protects from tumour metastasis and established the SOX18 protein as a molecular target. Here, we report the crystal structure of the SOX18 DNA binding high-mobility group (HMG) box bound to a DNA element regulating Prox1 transcription. The crystals diffracted to 1.75Å presenting the highest resolution structure of a SOX/DNA complex presently available revealing water structure, structural adjustments at the DNA contact interface and non-canonical conformations of the DNA backbone. To explore alternatives to challenging small molecule approaches for targeting the DNA-binding activity of SOX18, we designed a set of five decoys based on modified Prox1-DNA. Four decoys potently inhibited DNA binding of SOX18 in vitro and did not interact with non-SOX TFs. Serum stability, nuclease resistance and thermal denaturation assays demonstrated that a decoy circularized with a hexaethylene glycol linker and terminal phosphorothioate modifications is most stable. This SOX decoy also interfered with the expression of a luciferase reporter under control of a SOX18-dependent VCAM1 promoter in COS7 cells. Collectively, we propose SOX decoys as potential strategy for inhibiting SOX18 activity to disrupt tumour-induced neo-lymphangiogenesis.

Oct4 switches partnering from Sox2 to Sox17 to reinterpret the enhancer code and specify endoderm.

How regulatory information is encoded in the genome is poorly understood and poses a challenge when studying biological processes. We demonstrate here that genomic redistribution of Oct4 by alternative partnering with Sox2 and Sox17 is a fundamental regulatory event of endodermal specification. We show that Sox17 partners with Oct4 and binds to a unique 'compressed' Sox/Oct motif that earmarks endodermal genes. This is in contrast to the pluripotent state where Oct4 selectively partners with Sox2 at 'canonical' binding sites. The distinct selection of binding sites by alternative Sox/Oct partnering is underscored by our demonstration that rationally point-mutated Sox17 partners with Oct4 on pluripotency genes earmarked by the canonical Sox/Oct motif. In an endodermal differentiation assay, we demonstrate that the compressed motif is required for proper expression of endodermal genes. Evidently, Oct4 drives alternative developmental programs by switching Sox partners that affects enhancer selection, leading to either an endodermal or pluripotent cell fate. This work provides insights in understanding cell fate transcriptional regulation by highlighting the direct link between the DNA sequence of an enhancer and a developmental outcome.

Regulation of expression of venom toxins: silencing of prothrombin activator trocarin D by AG-rich motifs.

Trocarin D (TroD), a venom prothrombin activator from Tropidechis carinatus, shares similar structure and function with blood coagulation factor Xa [Tropidechis carinatus FX (TrFX) a]. Their distinct physiologic roles are due to their distinct expression patterns. The genes of TroD and TrFX are highly similar, except for promoter and intron 1, indicating that TroD has probably evolved by duplication of FX, the plasma counterpart. The promoter insertion in TroD accounts for the elevated but not venom gland-specific expression. Here we examined the roles of 3 insertions and 2 deletions in intron 1 of TroD in the regulation of expression using luciferase as a reporter. By systematic deletions, we showed that a 209 bp region within the second insertion silences expression in mammalian and unmilked venom gland cells. Through bioinformatics analysis, we identified 5 AG-rich motifs in this region. All except the 5th motif are important for silencing function. YY1, Sp3 and HMGB2 were identified to bind these AG-rich motifs and silence gene expression in mammalian cells. Similar AG-rich motif clusters are also found in other toxin genes but not in their physiologic counterparts. Thus, AG-rich motifs contribute to regulation of expression of TroD, and probably other toxin genes.-Han, S. X., Kwong, S., Ge, R., Kolatkar, P. R., Woods, A. E., Blanchet, G., Kini, R. M. Regulation of expression of venom toxins: silencing of prothrombin activator trocarin D by AG-rich motifs.

TherMos: Estimating protein-DNA binding energies from in vivo binding profiles.

Accurately characterizing transcription factor (TF)-DNA affinity is a central goal of regulatory genomics. Although thermodynamics provides the most natural language for describing the continuous range of TF-DNA affinity, traditional motif discovery algorithms focus instead on classification paradigms that aim to discriminate 'bound' and 'unbound' sequences. Moreover, these algorithms do not directly model the distribution of tags in ChIP-seq data. Here, we present a new algorithm named Thermodynamic Modeling of ChIP-seq (TherMos), which directly estimates a position-specific binding energy matrix (PSEM) from ChIP-seq/exo tag profiles. In cross-validation tests on seven genome-wide TF-DNA binding profiles, one of which we generated via ChIP-seq on a complex developing tissue, TherMos predicted quantitative TF-DNA binding with greater accuracy than five well-known algorithms. We experimentally validated TherMos binding energy models for Klf4 and Esrrb, using a novel protocol to measure PSEMs in vitro. Strikingly, our measurements revealed strong non-additivity at multiple positions within the two PSEMs. Among the algorithms tested, only TherMos was able to model the entire binding energy landscape of Klf4 and Esrrb. Our study reveals new insights into the energetics of TF-DNA binding in vivo and provides an accurate first-principles approach to binding energy inference from ChIP-seq and ChIP-exo data.

Structure of liganded T-state haemoglobin from cat (Felis silvestris catus), a low oxygen-affinity species, in two different crystal forms.

Haemoglobin (Hb) is an iron-containing metalloprotein which plays a major role in the transportation of oxygen from the lungs to tissues and of carbon dioxide back to the lungs. Hb is in equilibrium between low-affinity tense (T) and high-affinity relaxed (R) states associated with its unliganded and liganded forms, respectively. Mammalian species can be classified into two groups on the basis of whether they express `high' or `low' oxygen-affinity Hbs. Although Hbs from the former group have been studied extensively, a more limited number of structural studies have been performed for low oxygen-affinity Hbs. Here, the crystal structure of low oxygen-affinity cat methaemoglobin (metHb) has been solved at 2.0 and 2.4 Šresolution in two different crystal forms. Even though both structures are fully liganded, they unusually adopt a T-state-like quaternary conformation but with several localized R-like tertiary-structural and quaternary-structural features. The study provides atomic-level insights into the ligand-binding properties of this Hb, including its low cooperativity, blunt response to allosteric effectors and low affinity for oxygen, as well as further contributing to the mechanism underlying Hb allostery.

Sox transcription factors require selective interactions with Oct4 and specific transactivation functions to mediate reprogramming.

The unique ability of Sox2 to cooperate with Oct4 at selective binding sites in the genome is critical for reprogramming somatic cells into induced pluripotent stem cells (iPSCs). We have recently demonstrated that Sox17 can be converted into a reprogramming factor by alteration of a single amino acid (Sox17EK) within its DNA binding HMG domain. Here we expanded this study by introducing analogous mutations to 10 other Sox proteins and interrogated the role of N-and C-termini on the reprogramming efficiency. We found that point-mutated Sox7 and Sox17 can convert human and mouse fibroblasts into iPSCs, but Sox4, Sox5, Sox6, Sox8, Sox9, Sox11, Sox12, Sox13, and Sox18 cannot. Next we studied regions outside the HMG domain and found that the C-terminal transactivation domain of Sox17 and Sox7 enhances the potency of Sox2 in iPSC assays and confers weak reprogramming potential to the otherwise inactive Sox4EK and Sox18EK proteins. These results suggest that the glutamate (E) to lysine (K) mutation in the HMG domain is necessary but insufficient to swap the function of Sox factors. Moreover, the HMG domain alone fused to the VP16 transactivation domain is able to induce reprogramming, albeit at low efficiency. By molecular dissection of the C-terminus of Sox17, we found that the ß-catenin interaction region contributes to the enhanced reprogramming efficiency of Sox17EK. To mechanistically understand the enhanced reprogramming potential of Sox17EK, we analyzed ChIP-sequencing and expression data and identified a subset of candidate genes specifically regulated by Sox17EK and not by Sox2.

Deciphering the Sox-Oct partner code by quantitative cooperativity measurements.

Several Sox-Oct transcription factor (TF) combinations have been shown to cooperate on diverse enhancers to determine cell fates. Here, we developed a method to quantify biochemically the Sox-Oct cooperation and assessed the pairing of the high-mobility group (HMG) domains of 11 Sox TFs with Oct4 on a series of composite DNA elements. This way, we clustered Sox proteins according to their dimerization preferences illustrating that Sox HMG domains evolved different propensities to cooperate with Oct4. Sox2, Sox14, Sox21 and Sox15 strongly cooperate on the canonical element but compete with Oct4 on a recently discovered compressed element. Sry also cooperates on the canonical element but binds additively to the compressed element. In contrast, Sox17 and Sox4 cooperate more strongly on the compressed than on the canonical element. Sox5 and Sox18 show some cooperation on both elements, whereas Sox8 and Sox9 compete on both elements. Testing rationally mutated Sox proteins combined with structural modeling highlights critical amino acids for differential Sox-Oct4 partnerships and demonstrates that the cooperativity correlates with the efficiency in producing induced pluripotent stem cells. Our results suggest selective Sox-Oct partnerships in genome regulation and provide a toolset to study protein cooperation on DNA.

Structural analysis and dimerization profile of the SCAN domain of the pluripotency factor Zfp206.

Zfp206 (also named as Zscan10) belongs to the subfamily of C(2)H(2) zinc finger transcription factors, which is characterized by the N-terminal SCAN domain. The SCAN domain mediates self-association and association between the members of SCAN family transcription factors, but the structural basis and selectivity determinants for complex formation is unknown. Zfp206 is important for maintaining the pluripotency of embryonic stem cells presumably by combinatorial assembly of itself or other SCAN family members on enhancer regions. To gain insights into the folding topology and selectivity determinants for SCAN dimerization, we solved the 1.85 Å crystal structure of the SCAN domain of Zfp206. In vitro binding studies using a panel of 20 SCAN proteins indicate that the SCAN domain Zfp206 can selectively associate with other members of SCAN family transcription factors. Deletion mutations showed that the N-terminal helix 1 is critical for heterodimerization. Double mutations and multiple mutations based on the Zfp206SCAN-Zfp110SCAN model suggested that domain swapped topology is a possible preference for Zfp206SCAN-Zfp110SCAN heterodimer. Together, we demonstrate that the Zfp206SCAN constitutes a protein module that enables C(2)H(2) transcription factor dimerization in a highly selective manner using a domain-swapped interface architecture and identify novel partners for Zfp206 during embryonal development.

BRET-based biosensors for SARS-CoV-2 oligonucleotide detection.

The need for the early detection of emerging pathogenic viruses and their newer variants has driven the urgent demand for developing point-of-care diagnostic tools. Although nucleic acid-based methods such as reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and loop-mediated isothermal amplification (LAMP) have been developed, a more facile and robust platform is still required. To address this need, as a proof-of-principle study, we engineered a prototype-the versatile, sensitive, rapid, and cost-effective bioluminescence resonance energy transfer (BRET)-based biosensor for oligonucleotide detection (BioOD). Specifically, we designed BioODs against the SARS-CoV-2 parental (Wuhan strain) and B.1.617.2 Delta variant through the conjugation of specific, fluorescently modified molecular beacons (sensor module) through a complementary oligonucleotide handle DNA functionalized with the NanoLuc (NLuc) luciferase protein such that the dissolution of the molecular beacon loop upon the binding of the viral oligonucleotide will result in a decrease in BRET efficiency and, thus, a change in the bioluminescence spectra. Following the assembly of the BioODs, we determined their kinetics response, affinity for variant-specific oligonucleotides, and specificity, and found them to be rapid and highly specific. Furthermore, the decrease in BRET efficiency of the BioODs in the presence of viral oligonucleotides can be detected as a change in color in cell phone camera images. We envisage that the BioODs developed here will find application in detecting viral infections with variant specificity in a point-of-care-testing format, thus aiding in large-scale viral infection surveillance.

Structural insights into the unique recognition module between α-synuclein peptide and nanobody.

Nanobodies are single-domain fragments of antibodies with comparable specificity and affinity to antibodies. They are emerging as versatile tools in biology due to their relatively small size. Here, we report the crystal structure of a specific nanobody Nbα-syn01, bound to a 14 amino acid long peptide of α-synuclein (αSyn), a 140-residue protein whose aggregation is associated with Parkinson's disease. The complex structure exhibits a unique binding pattern where the αSyn peptide replaces the N-terminal region of nanobody. Recognition is mediated principally by extended main chain interaction of the αSyn peptide and specificity of the interaction lies in the central 48-52 region of αSyn peptide. Structure-guided truncation of Nbα-syn01 shows tighter binding to αSyn peptide and improved inhibition of α-synuclein aggregation. The structure of the truncated complex was subsequently determined and was indistinguishable to full length complex as the full-length form had no visible electron density for the N-terminal end. These findings reveal the molecular basis for a previously unobserved binding mode for nanobody recognition of α-synuclein, providing an explanation for the enhanced binding, and potential for an alternate framework for structure-based protein engineering of nanobodies to develop better diagnostic and therapeutic tools.

The crystal structure of the Sox4 HMG domain-DNA complex suggests a mechanism for positional interdependence in DNA recognition.

It has recently been proposed that the sequence preferences of DNA-binding TFs (transcription factors) can be well described by models that include the positional interdependence of the nucleotides of the target sites. Such binding models allow for multiple motifs to be invoked, such as principal and secondary motifs differing at two or more nucleotide positions. However, the structural mechanisms underlying the accommodation of such variant motifs by TFs remain elusive. In the present study we examine the crystal structure of the HMG (high-mobility group) domain of Sox4 [Sry (sex-determining region on the Y chromosome)-related HMG box 4] bound to DNA. By comparing this structure with previously solved structures of Sox17 and Sox2, we observed subtle conformational differences at the DNA-binding interface. Furthermore, using quantitative electrophoretic mobility-shift assays we validated the positional interdependence of two nucleotides and the presence of a secondary Sox motif in the affinity landscape of Sox4. These results suggest that a concerted rearrangement of two interface amino acids enables Sox4 to accommodate primary and secondary motifs. The structural adaptations lead to altered dinucleotide preferences that mutually reinforce each other. These analyses underline the complexity of the DNA recognition by TFs and provide an experimental validation for the conceptual framework of positional interdependence and secondary binding motifs.

Structural basis for the cooperative DNA recognition by Smad4 MH1 dimers.

Smad proteins form multimeric complexes consisting of the 'common partner' Smad4 and receptor regulated R-Smads on clustered DNA binding sites. Deciphering how pathway specific Smad complexes multimerize on DNA to regulate gene expression is critical for a better understanding of the cis-regulatory logic of TGF-ß and BMP signaling. To this end, we solved the crystal structure of the dimeric Smad4 MH1 domain bound to a palindromic Smad binding element. Surprisingly, the Smad4 MH1 forms a constitutive dimer on the SBE DNA without exhibiting any direct protein-protein interactions suggesting a DNA mediated indirect readout mechanism. However, the R-Smads Smad1, Smad2 and Smad3 homodimerize with substantially decreased efficiency despite pronounced structural similarities to Smad4. Therefore, intricate variations in the DNA structure induced by different Smads and/or variant energetic profiles likely contribute to their propensity to dimerize on DNA. Indeed, competitive binding assays revealed that the Smad4/R-Smad heterodimers predominate under equilibrium conditions while R-Smad homodimers are least favored. Together, we present the structural basis for DNA recognition by Smad4 and demonstrate that Smad4 constitutively homo- and heterodimerizes on DNA in contrast to its R-Smad partner proteins by a mechanism independent of direct protein contacts.