PEC-PRO: A new prognostic score from a series of 87 patients with localized perivascular epithelioid cell neoplasms (PEComas) treated with curative intent.

BACKGROUND: Perivascular epithelioid cell neoplasms (PEComas) encompass a heterogeneous family of mesenchymal tumors. Previously described clinicopathologic features aimed at distinguishing benign from malignant variants but lacked prognostic value. METHODS: This retrospective analysis examined clinicopathologic data from patients who had localized PEComa across French Sarcoma Network centers. The authors analyzed 12 clinicopathologic features in a Cox proportional hazard framework to derive a multivariate prognostic risk model for event-free survival (EFS). They built the PEComa prognostic score (PEC-PRO), in which scores ranged from 0 to 5, based on the coefficients of the multivariate model. Three groups were identified: low risk (score = 0), intermediate risk (score = 1), and high risk (score ≥ 2). RESULTS: Analyzing 87 patients who had a median 46-month follow-up (interquartile range, 20-74 months), the median EFS was 96.5 months (95% confidence interval [CI], 47.1 months to not applicable), with 2-year and 5-year EFS rates of 64.7% and 58%, respectively. The median overall survival was unreached, with 2-year and 5-year overall survival rates of 82.3% and 69.3%, respectively. The simplified Folpe classification did not correlate with EFS. Multivariate analysis identified three factors affecting EFS: positive surgical margins (hazard ratio [HR], 5.17; 95% CI, 1.65-16.24; p = .008), necrosis (HR, 3.94; 95% CI, 1.16-13.43; p = .030), and male sex (HR, 3.13; 95% CI, 1.19-8.27; p = 0.023). Four variables were retained in the prognostic model. Patients with low-risk PEC-PRO scores had a 2-year EFS rate of 93.7% (95% CI, 83.8%-100.0%), those with intermediate-risk PEC-PRO scores had a 2-year EFS rate of 67.4% (95% CI, 53.9%-80.9%), and those with high-risk PEC-PRO scores had a 2-year EFS rate of 2.3% (95% CI, 0.0%-18.3%). CONCLUSIONS: The PEC-PRO score reliably predicts the risk of postoperative recurrence in patients with localized PEComa. It has the potential to improve follow-up strategies but requires validation in a prospective trial.

Decoding complexity in biomolecular recognition of DNA i-motifs with microarrays.

DNA i-motifs (iMs) are non-canonical C-rich secondary structures implicated in numerous cellular processes. Though iMs exist throughout the genome, our understanding of iM recognition by proteins or small molecules is limited to a few examples. We designed a DNA microarray containing 10976 genomic iM sequences to examine the binding profiles of four iM-binding proteins, mitoxantrone and the iMab antibody. iMab microarray screens demonstrated that pH 6.5, 5% BSA buffer was optimal, and fluorescence was correlated with iM C-tract length. hnRNP K broadly recognizes diverse iM sequences, favoring 3-5 cytosine repeats flanked by thymine-rich loops of 1-3 nucleotides. Array binding mirrored public ChIP-Seq datasets, in which 35% of well-bound array iMs are enriched in hnRNP K peaks. In contrast, other reported iM-binding proteins had weaker binding or preferred G-quadruplex (G4) sequences instead. Mitoxantrone broadly binds both shorter iMs and G4s, consistent with an intercalation mechanism. These results suggest that hnRNP K may play a role in iM-mediated regulation of gene expression in vivo, whereas hnRNP A1 and ASF/SF2 are possibly more selective in their binding preferences. This powerful approach represents the most comprehensive investigation of how biomolecules selectively recognize genomic iMs to date.

DECODING COMPLEXITY IN BIOMOLECULAR RECOGNITION OF DNA I-MOTIFS.

DNA i-motifs (iMs) are non-canonical C-rich secondary structures implicated in numerous cellular processes. Though iMs exist throughout the genome, our understanding of iM recognition by proteins or small molecules is limited to a few examples. We designed a DNA microarray containing 10,976 genomic iM sequences to examine the binding profiles of four iM-binding proteins, mitoxantrone, and the iMab antibody. iMab microarray screens demonstrated that pH 6.5, 5% BSA buffer was optimal, and fluorescence was correlated with iM C-tract length. hnRNP K broadly recognizes diverse iM sequences, favoring 3-5 cytosine repeats flanked by thymine-rich loops of 1-3 nucleotides. Array binding mirrored public ChIP-Seq datasets, in which 35% of well-bound array iMs are enriched in hnRNP K peaks. In contrast, other reported iM-binding proteins had weaker binding or preferred G-quadruplex (G4) sequences instead. Mitoxantrone broadly binds both shorter iMs and G4s, consistent with an intercalation mechanism. These results suggest that hnRNP K may play a role in iM-mediated regulation of gene expression in vivo, whereas hnRNP A1 and ASF/SF2 are possibly more selective in their binding preferences. This powerful approach represents the most comprehensive investigation of how biomolecules selectively recognize genomic iMs to date.

NuRD complex recruitment to Thpok mediates CD4+ T cell lineage differentiation.

Although BTB-zinc finger (BTB-ZF) transcription factors control the differentiation of multiple hematopoietic and immune lineages, how they function is poorly understood. The BTB-ZF factor Thpok controls intrathymic CD4+ T cell development and the expression of most CD4+ and CD8+ lineage genes. Here, we identify the nucleosome remodeling and deacetylase (NuRD) complex as a critical Thpok cofactor. Using mass spectrometry and coimmunoprecipitation in primary T cells, we show that Thpok binds NuRD components independently of DNA association. We locate three amino acid residues within the Thpok BTB domain that are required for both NuRD binding and Thpok functions. Conversely, a chimeric protein merging the NuRD component Mta2 to a BTB-less version of Thpok supports CD4+ T cell development, indicating that NuRD recruitment recapitulates the functions of the Thpok BTB domain. We found that NuRD mediates Thpok repression of CD8+ lineage genes, including the transcription factor Runx3, but is dispensable for Cd4 expression. We show that these functions cannot be performed by the BTB domain of the Thpok-related factor Bcl6, which fails to bind NuRD. Thus, cofactor binding critically contributes to the functional specificity of BTB-ZF factors, which control the differentiation of most hematopoietic subsets.

Altering the Double-Stranded DNA Specificity of the bZIP Domain of Zta with Site-Directed Mutagenesis at N182.

Zta, the Epstein-Barr virus bZIP transcription factor (TF), binds both unmethylated and methylated double-stranded DNA (dsDNA) in a sequence-specific manner. We studied the contribution of a conserved asparagine (N182) to sequence-specific dsDNA binding to four types of dsDNA: (i) dsDNA with cytosine in both strands ((DNA(C|C)), (ii, iii) dsDNA with 5-methylcytosine (5mC, M) or 5-hydroxymethylcytosine (5hmC, H) in one strand and cytosine in the second strand ((DNA(5mC|C) and DNA(5hmC|C)), and (iv) dsDNA with methylated cytosine in both strands in all CG dinucleotides ((DNA(5mCG)). We replaced asparagine with five similarly sized amino acids (glutamine (Q), serine (S), threonine (T), isoleucine (I), or valine (V)) and used protein binding microarrays to evaluate sequence-specific dsDNA binding. Zta preferentially binds the pseudo-palindrome TRE (AP1) motif (T-4G-3A-2G/C 0T2C3A4 ). Zta (N182Q) changes binding to A3 in only one half-site. Zta(N182S) changes binding to G3 in one or both halves of the motif. Zta(N182S) and Zta(N182Q) have 34- and 17-fold weaker median dsDNA binding, respectively. Zta(N182V) and Zta(N182I) have increased binding to dsDNA(5mC|C). Molecular dynamics simulations rationalize some of these results, identifying hydrogen bonds between glutamine and A3 , but do not reveal why serine preferentially binds G3 , suggesting that entropic interactions may mediate this new binding specificity.

Regulation and function of AP-1 in insulinoma cells and pancreatic ß-cells.

Cav1.2 L-type voltage-gated Ca2+ channels play a central role in pancreatic ß-cells by integrating extracellular signals with intracellular signaling events leading to insulin secretion and altered gene transcription. Here, we investigated the intracellular signaling pathway following stimulation of Cav1.2 Ca2+ channels and addressed the function of the transcription factor activator protein-1 (AP-1) in pancreatic ß-cells of transgenic mice. Stimulation of Cav1.2 Ca2+ channels activates AP-1 in insulinoma cells. Pharmacological and genetic experiments identified c-Jun N-terminal protein kinase as a signal transducer connecting Cav1.2 Ca2+ channel activation with gene transcription. Moreover, the basic region-leucine zipper proteins ATF2 and c-Jun or c-Jun-related proteins were involved in stimulus-transcription coupling. We addressed the functions of AP-1 in pancreatic ß-cells analyzing a newly generated transgenic mouse model. These transgenic mice expressed A-Fos, a mutant of c-Fos that attenuates DNA binding of c-Fos dimerization partners. In insulinoma cells, A-Fos completely blocked AP-1 activation following stimulation of Cav1.2 Ca2+ channels. The analysis of transgenic A-Fos-expressing mice revealed that the animals displayed impaired glucose tolerance. Thus, we show here for the first time that AP-1 controls an important function of pancreatic ß-cells in vivo, the regulation of glucose homeostasis.

Structural basis for glucocorticoid receptor recognition of both unmodified and methylated binding sites, precursors of a modern recognition element.

The most common form of DNA methylation involves the addition of a methyl group to a cytosine base in the context of a cytosine-phosphate-guanine (CpG) dinucleotide. Genomes from more primitive organisms are more abundant in CpG sites that, through the process of methylation, deamination and subsequent mutation to thymine-phosphate-guanine (TpG) sites, can produce new transcription factor binding sites. Here, we examined the evolutionary history of the over 36 000 glucocorticoid receptor (GR) consensus binding motifs in the human genome and identified a subset of them in regulatory regions that arose via a deamination and subsequent mutation event. GR can bind to both unmodified and methylated pre-GR binding sequences (GBSs) that contain a CpG site. Our structural analyses show that CpG methylation in a pre-GBS generates a favorable interaction with Arg447 mimicking that made with a TpG in a GBS. This methyl-specific recognition arose 420 million years ago and was conserved during the evolution of GR and likely helps fix the methylation on the relevant cytosines. Our study provides the first genetic, biochemical and structural evidence of high-affinity binding for the likely evolutionary precursor of extant TpG-containing GBS.

REL Domain of NFATc2 Binding to Five Types of DNA Using Protein Binding Microarrays.

NFATc2 is a DNA binding protein in the Rel family transcription factors, which binds a CGGAA motif better when both cytosines in the CG dinucleotide are methylated. Using protein binding microarrays (PBMs), we examined the DNA binding of NFATc2 to three additional types of DNA: single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA) with either 5-methylcytosine (5mC, M) or 5-hydroxymethylcytosine (5hmC, H) in one strand and a cytosine in the second strand. ATTTCCAC, the complement of the core GGAA motif, is better bound as ssDNA compared to dsDNA. dsDNA containing the 5-mer CGGAA with either 5mC or 5hmC in one DNA strand is bound stronger than CGGAA. In contrast, the reverse complement TTCCG is bound weaker when it contains 5mC. Analysis of the available NFATc2:dsDNA complexes rationalizes these PBM data.

bZIP Dimers CREB1, ATF2, Zta, ATF3|cJun, and cFos|cJun Prefer to Bind to Some Double-Stranded DNA Sequences Containing 5-Formylcytosine and 5-Carboxylcytosine.

In mammalian cells, 5-methylcytosine (5mC) occurs in genomic double-stranded DNA (dsDNA) and is enzymatically oxidized to 5-hydroxymethylcytosine (5hmC), then to 5-formylcytosine (5fC), and finally to 5-carboxylcytosine (5caC). These cytosine modifications are enriched in regulatory regions of the genome. The effect of these oxidative products on five bZIP dimers (CREB1, ATF2, Zta, ATF3|cJun, and cFos|cJun) binding to five types of dsDNA was measured using protein binding microarrays. The five dsDNAs contain either cytosine in both DNA strands or cytosine in one strand and either 5mC, 5hmC, 5fC, or 5caC in the second strand. Some sequences containing the CEBP half-site GCAA are bound more strongly by all five bZIP domains when dsDNA contains 5mC, 5hmC, or 5fC. dsDNA containing 5caC in some TRE (AP-1)-like sequences, e.g., TGACTAA, is better bound by Zta, ATF3|cJun, and cFos|cJun.

Custom G4 Microarrays Reveal Selective G-Quadruplex Recognition of Small Molecule BMVC: A Large-Scale Assessment of Ligand Binding Selectivity.

G-quadruplexes (G4) are considered new drug targets for human diseases such as cancer. More than 10,000 G4s have been discovered in human chromatin, posing challenges for assessing the selectivity of a G4-interactive ligand. 3,6-bis(1-Methyl-4-vinylpyridinium) carbazole diiodide (BMVC) is the first fluorescent small molecule for G4 detection in vivo. Our previous structural study shows that BMVC binds to the MYC promoter G4 (MycG4) with high specificity. Here, we utilize high-throughput, large-scale custom DNA G4 microarrays to analyze the G4-binding selectivity of BMVC. BMVC preferentially binds to the parallel MycG4 and selectively recognizes flanking sequences of parallel G4s, especially the 3'-flanking thymine. Importantly, the microarray results are confirmed by orthogonal NMR and fluorescence binding analyses. Our study demonstrates the potential of custom G4 microarrays as a platform to broadly and unbiasedly assess the binding selectivity of G4-interactive ligands, and to help understand the properties that govern molecular recognition.

Custom DNA Microarrays Reveal Diverse Binding Preferences of Proteins and Small Molecules to Thousands of G-Quadruplexes.

Single-stranded DNA (ssDNA) containing four guanine repeats can form G-quadruplex (G4) structures. While cellular proteins and small molecules can bind G4s, it has been difficult to broadly assess their DNA-binding specificity. Here, we use custom DNA microarrays to examine the binding specificities of proteins, small molecules, and antibodies across ∼15,000 potential G4 structures. Molecules used include fluorescently labeled pyridostatin (Cy5-PDS, a small molecule), BG4 (Cy5-BG4, a G4-specific antibody), and eight proteins (GST-tagged nucleolin, IGF2, CNBP, FANCJ, PIF1, BLM, DHX36, and WRN). Cy5-PDS and Cy5-BG4 selectively bind sequences known to form G4s, confirming their formation on the microarrays. Cy5-PDS binding decreased when G4 formation was inhibited using lithium or when ssDNA features on the microarray were made double-stranded. Similar conditions inhibited the binding of all other molecules except for CNBP and PIF1. We report that proteins have different G4-binding preferences suggesting unique cellular functions. Finally, competition experiments are used to assess the binding specificity of an unlabeled small molecule, revealing the structural features in the G4 required to achieve selectivity. These data demonstrate that the microarray platform can be used to assess the binding preferences of molecules to G4s on a broad scale, helping to understand the properties that govern molecular recognition.

Transient Delivery of A-C/EBP Protein Perturbs Differentiation of 3T3-L1 Cells and Induces Preadipocyte Marker Genes.

Transformation of committed 3T3-L1 preadipocytes to lipid-laden adipocytes involves the timely appearance of numerous transcription factors (TFs); foremost among them, C/EBPß is expressed during the early phases of differentiation. Here, we describe liposome-mediated protein transfection approach to rapidly downregulate C/EBPß by A-C/EBP protein inhibitor. Signals from EGFP-tagged A-C/EBP protein were observed in 3T3-L1 cells within 2 h of transfections, whereas for A-C/EBP gene transfections, equivalent signals appeared in 48 h. Following transient transfections, the expression profiles of 24 marker genes belonging to pro- and anti-adipogenic, cell cycle, and preadipocyte pathways were analyzed. Expectedly, the mRNA and protein expression profiles of adipocyte marker genes showed lower expression in both A-C/EBP protein- and gene-transfected samples. Interestingly, for preadipocytes and cell fate determinant genes, striking differences were observed between A-C/EBP protein- and A-C/EBP gene-transfected samples. Preadipocyte differentiation factors Stat5a and Creb were downregulated in A-C/EBP protein samples. Five preadipocyte markers, namely, Pdgfrα, Pdgfrß, Ly6A, CD34, and Itgb1, showed high expression in A-C/EBP protein samples, whereas only Ly6A and CD34 were expressed in A-C/EBP gene-transfected samples. Pdgfrα and Pdgfrß, two known cell fate markers, were expressed in A-C/EBP protein-transfected samples, suggesting a possible reversal of differentiation. Our study provides evidences for the immediate and efficient knockdown of C/EBPß protein to understand time-dependent preadipocytes differentiation.

The bZIP mutant CEBPB (V285A) has sequence specific DNA binding propensities similar to CREB1.

The bZIP homodimers CEBPB and CREB1 bind DNA containing methylated cytosines differently. CREB1 binds stronger to the C/EBP half-site GCAA when the cytosine is methylated. For CEBPB, methylation of the same cytosine does not affect DNA binding. The X-ray structure of CREB1 binding the half site GTCA identifies an alanine in the DNA binding region interacting with the methyl group of T, structurally analogous to the methyl group of methylated C. This alanine is replaced with a valine in CEBPB. To explore the contribution of this amino acid to binding with methylated cytosine of the GCAA half-site, we made the reciprocal mutants CEBPB(V285A) and CREB1(A297V) and used protein binding microarrays (PBM) to examine binding to four types of double-stranded DNA (dsDNA): 1) DNA with cytosine in both strands (DNA(C|C)), 2) DNA with 5-methylcytosine (M) in one strand and cytosine in the second strand (DNA(M|C)), 3) DNA with 5-hydroxymethylcytosine (H) in one strand and cytosine in the second strand (DNA(H|C)), and 4) DNA with both cytosines in all CG dinucleotides containing 5-methylcytosine (DNA(5mCG)). When binding to DNA(C|C), CEBPB (V285A) preferentially binds the CRE consensus motif (TGACGTCA), similar to CREB1. The reciprocal mutant, CREB1(A297V) binds DNA with some similarity to CEBPB, with strongest binding to the methylated PAR site 8-mer TTACGTAA. These data demonstrate that V285 residue inhibits CEBPB binding to methylated cytosine of the GCAA half-site.

GABPα and CREB1 Binding to Double Nucleotide Polymorphisms of Their Consensus Motifs and Cooperative Binding to the Composite ETS ⇔ CRE Motif (ACCGGAAGTGACGTCA).

Previously, cooperative binding of the bZIP domain of CREB1 and the ETS domain of GABPα was observed for the composite DNA ETS ⇔ CRE motif (A 0 C 1 C 2 G 3 G 4 A 5 A 6 G 7 T 8 G 9 A 10 C 11 G 12 T 13 C 14 A 15 ). Single nucleotide polymorphisms (SNPs) at the beginning and end of the ETS motif (ACCGGAAGT) increased cooperative binding. Here, we use an Agilent microarray of 60-mers containing all double nucleotide polymorphisms (DNPs) of the ETS ⇔ CRE motif to explore GABPα and CREB1 binding to their individual motifs and their cooperative binding. For GABPα, all DNPs were bound as if each SNP acted independently. In contrast, CREB1 binding to some DNPs was stronger or weaker than expected, depending on the locations of each SNP. CREB1 binding to DNPs where both SNPs were in the same half site, T 8 G 9 A 10 or T 13 C 14 A 15 , was greater than expected, indicating that an additional SNP cannot destroy binding as much as expected, suggesting that an individual SNP is enough to abolish sequence-specific DNA binding of a single bZIP monomer. If a DNP contains SNPs in each half site, binding is weaker than expected. Similar results were observed for additional ETS and bZIP family members. Cooperative binding between GABPα and CREB1 to the ETS ⇔ CRE motif was weaker than expected except for DNPs containing A 7 and SNPs at the beginning of the ETS motif.

Replacing C189 in the bZIP domain of Zta with S, T, V, or A changes DNA binding specificity to four types of double-stranded DNA.

Zta is a bZIP transcription factor (TF) in the Epstein-Barr virus that binds unmethylated and methylated DNA sequences. Substitution of cysteine 189 of Zta to serine (Zta(C189S)) results in a virus that is unable to execute the lytic cycle, which was attributed to a change in binding to methylated DNA sequences. To learn more about the role of this position in defining sequence-specific DNA binding, we mutated cysteine 189 to four other amino acids, producing Zta(C189S), Zta(C189T), Zta(C189A), and Zta(C189V) mutants. Zta and mutants were used in protein binding microarray (PBM) experiments to evaluate sequence-specific DNA binding to four types of double-stranded DNA (dsDNA): 1) with cytosine in both strands (DNA(C|C)), 2) with 5-methylcytosine (5mC) in one strand and cytosine in the second strand (DNA(5mC|C)), 3) with 5-hydroxymethylcytosine (5hmC) in one strand and cytosine in the second strand (DNA(5hmC|C)), and 4) with both cytosines in all CG dinucleotides containing 5mC (DNA(5mCG)). Zta(C189S) and Zta(C189T) bound the TRE (AP-1) motif (TGAG/CTCA) more strongly than wild-type Zta, while binding to other sequences, including the C/EBP half site GCAA was reduced. Binding of Zta(C189S) and Zta(C189T) to DNA containing modified cytosines (DNA(5mC|C), DNA(5hmC|C), and DNA(5mCG)) was reduced compared to Zta. Zta(C189A) and Zta(C189V) had higher non-specific binding to all four types of DNA. Our data suggests that position C189 in Zta impacts sequence-specific binding to DNA containing modified and unmodified cytosine.

Delayed Hair Follicle Morphogenesis and Hair Follicle Dystrophy in a Lipoatrophy Mouse Model of Pparg Total Deletion.

PPARγ regulates multiple aspects of skin physiology, including sebocyte differentiation, keratinocyte proliferation, epithelial stem cell survival, adipocyte biology, and inflammatory skin responses. However, the effects of its global deletion, namely of nonredundant key functions of PPARγ signaling in mammalian skin, are yet unknown because of embryonic lethality. Here, we describe the skin and hair phenotype of a whole-body PPARγ-null mouse (PpargΔ/Δ), obtained by preserving PPARγ expression in the placenta. PpargΔ/Δ mice exhibited total lipoatrophy and complete absence of sebaceous glands. Right after birth, hair follicle (HF) morphogenesis was transiently delayed, along with reduced expression of HF differentiation markers and of transcriptional regulators necessary for HF development. Later, adult PpargΔ/Δ mice developed scarring alopecia and severe perifollicular inflammation. Skin analyses in other models of lipodystrophy, AZIPtg/+ and Adipoq-Cretg/+Ppargfl/fl mice, coupled with skin graft experiments, showed that the early defects observed in hair morphogenesis were caused by the absence of adipose tissue. In contrast, the late alteration of HF cycle and appearance of inflammation were observed only in PpargΔ/Δ mice and likely were due to the lack sebaceous glands. Our findings underscore the increasing appreciation for the importance of adipose tissue-mediated signals in HF development and function.

A-ZIP53, a dominant negative reveals the molecular mechanism of heterodimerization between bZIP53, bZIP10 and bZIP25 involved in Arabidopsis seed maturation.

In Arabidopsis, maturation phase, an intricate process in seed formation is tightly regulated by the DNA binding activity of protagonist basic leucine zipper 53 (bZIP53) transcription factor and its heterodimerizing partners, bZIP10 and bZIP25. Structural determinants responsible for heterodimerization specificity of bZIP53 are poorly understood. Analysis of amino acid sequences of three bZIPs does not identify interactions that may favor heterodimerization. Here, we describe a designed dominant negative termed A-ZIP53 that has a glutamic acid-rich amphipathic peptide sequence attached to N-terminal of bZIP53 leucine zipper. Circular dichroism (CD) and mass spectrometry studies with equimolar mixture of three bZIP proteins in pairs showed no heterodimer formation whereas A-ZIP53 interacted and formed stable heterodimers with bZIP53, bZIP10, and bZIP25. A-ZIP53 electrostatically mimics DNA and can overcome repulsion between basic DNA binding regions of three bZIP proteins. Gel shift experiments showed that A-ZIP53 can inhibit the DNA binding of three proteins. CD studies demonstrated the specificity of A-ZIP53 as it did not interact with bZIP39 and bZIP72. Transient co-transfections in Arabidopsis protoplasts showed that A-ZIP53 inhibited three bZIPs and their putative heterodimers-mediated transactivation of GUS reporter gene. Furthermore, four newly designed acidic extensions were evaluated for their ability to interact with three bZIPs.

The Epstein-Barr Virus B-ZIP Protein Zta Recognizes Specific DNA Sequences Containing 5-Methylcytosine and 5-Hydroxymethylcytosine.

The Epstein-Barr virus (EBV) B-ZIP transcription factor Zta binds to many DNA sequences containing methylated CG dinucleotides. Using protein binding microarrays (PBMs), we analyzed the sequence specific DNA binding of Zta to four kinds of double-stranded DNA (dsDNA): (1) DNA containing cytosine in both strands, (2) DNA with 5-methylcytosine (5mC) in one strand and cytosine in the second strand, (3) DNA with 5-hydroxymethylcytosine (5hmC) in one strand and cytosine in the second strand, and (4) DNA in which both cytosines in all CG dinucleotides contain 5mC. We compared these data to PBM data for three additional B-ZIP proteins (CREB1 and CEBPB homodimers and cJun|cFos heterodimers). With cytosine, Zta binds the TRE motif TGAC/GTCA as previously reported. With CG dinucleotides containing 5mC on both strands, many TRE motif variants containing a methylated CG dinucleotide at two positions in the motif, such as MGAGTCA and TGAGMGA (where M = 5mC), were preferentially bound. 5mC inhibits binding of Zta to both TRE motif half-sites GTCA and CTCA. Like the CREB1 homodimer, the Zta homodimer and the cJun|cFos heterodimer more strongly bind the C/EBP half-site tetranucleotide GCAA when it contains 5mC. Zta also binds dsDNA sequences containing 5hmC in one strand, although the effect is less dramatic than that observed for 5mC. Our results identify new DNA sequences that are well-bound by the viral B-ZIP protein Zta only when they contain 5mC or 5hmC, uncovering the potential for discovery of new viral and host regulatory programs controlled by EBV.

Impact of cytosine methylation on DNA binding specificities of human transcription factors.

The majority of CpG dinucleotides in the human genome are methylated at cytosine bases. However, active gene regulatory elements are generally hypomethylated relative to their flanking regions, and the binding of some transcription factors (TFs) is diminished by methylation of their target sequences. By analysis of 542 human TFs with methylation-sensitive SELEX (systematic evolution of ligands by exponential enrichment), we found that there are also many TFs that prefer CpG-methylated sequences. Most of these are in the extended homeodomain family. Structural analysis showed that homeodomain specificity for methylcytosine depends on direct hydrophobic interactions with the methylcytosine 5-methyl group. This study provides a systematic examination of the effect of an epigenetic DNA modification on human TF binding specificity and reveals that many developmentally important proteins display preference for mCpG-containing sequences.

Distal CpG islands can serve as alternative promoters to transcribe genes with silenced proximal promoters.

DNA methylation at the promoter of a gene is presumed to render it silent, yet a sizable fraction of genes with methylated proximal promoters exhibit elevated expression. Here, we show, through extensive analysis of the methylome and transcriptome in 34 tissues, that in many such cases, transcription is initiated by a distal upstream CpG island (CGI) located several kilobases away that functions as an alternative promoter. Specifically, such genes are expressed precisely when the neighboring CGI is unmethylated but remain silenced otherwise. Based on CAGE and Pol II localization data, we found strong evidence of transcription initiation at the upstream CGI and a lack thereof at the methylated proximal promoter itself. Consistent with their alternative promoter activity, CGI-initiated transcripts are associated with signals of stable elongation and splicing that extend into the gene body, as evidenced by tissue-specific RNA-seq and other DNA-encoded splice signals. Furthermore, based on both inter- and intra-species analyses, such CGIs were found to be under greater purifying selection relative to CGIs upstream of silenced genes. Overall, our study describes a hitherto unreported conserved mechanism of transcription of genes with methylated proximal promoters in a tissue-specific fashion. Importantly, this phenomenon explains the aberrant expression patterns of some cancer driver genes, potentially due to aberrant hypomethylation of distal CGIs, despite methylation at proximal promoters.