Interpretable deep learning reveals the role of an E-box motif in suppressing somatic hypermutation of AGCT motifs within human immunoglobulin variable regions.

Introduction: Somatic hypermutation (SHM) of immunoglobulin variable (V) regions by activation induced deaminase (AID) is essential for robust, long-term humoral immunity against pathogen and vaccine antigens. AID mutates cytosines preferentially within WRCH motifs (where W=A or T, R=A or G and H=A, C or T). However, it has been consistently observed that the mutability of WRCH motifs varies substantially, with large variations in mutation frequency even between multiple occurrences of the same motif within a single V region. This has led to the notion that the immediate sequence context of WRCH motifs contributes to mutability. Recent studies have highlighted the potential role of local DNA sequence features in promoting mutagenesis of AGCT, a commonly mutated WRCH motif. Intriguingly, AGCT motifs closer to 5' ends of V regions, within the framework 1 (FW1) sub-region1, mutate less frequently, suggesting an SHM-suppressing sequence context. Methods: Here, we systematically examined the basis of AGCT positional biases in human SHM datasets with DeepSHM, a machine-learning model designed to predict SHM patterns. This was combined with integrated gradients, an interpretability method, to interrogate the basis of DeepSHM predictions. Results: DeepSHM predicted the observed positional differences in mutation frequencies at AGCT motifs with high accuracy. For the conserved, lowly mutating AGCT motifs in FW1, integrated gradients predicted a large negative contribution of 5'C and 3'G flanking residues, suggesting that a CAGCTG context in this location was suppressive for SHM. CAGCTG is the recognition motif for E-box transcription factors, including E2A, which has been implicated in SHM. Indeed, we found a strong, inverse relationship between E-box motif fidelity and mutation frequency. Moreover, E2A was found to associate with the V region locale in two human B cell lines. Finally, analysis of human SHM datasets revealed that naturally occurring mutations in the 3'G flanking residues, which effectively ablate the E-box motif, were associated with a significantly increased rate of AGCT mutation. Discussion: Our results suggest an antagonistic relationship between mutation frequency and the binding of E-box factors like E2A at specific AGCT motif contexts and, therefore, highlight a new, suppressive mechanism regulating local SHM patterns in human V regions.

RIF1 regulates early replication timing in murine B cells.

The mammalian DNA replication timing (RT) program is crucial for the proper functioning and integrity of the genome. The best-known mechanism for controlling RT is the suppression of late origins of replication in heterochromatin by RIF1. Here, we report that in antigen-activated, hypermutating murine B lymphocytes, RIF1 binds predominantly to early-replicating active chromatin and promotes early replication, but plays a minor role in regulating replication origin activity, gene expression and genome organization in B cells. Furthermore, we find that RIF1 functions in a complementary and non-epistatic manner with minichromosome maintenance (MCM) proteins to establish early RT signatures genome-wide and, specifically, to ensure the early replication of highly transcribed genes. These findings reveal additional layers of regulation within the B cell RT program, driven by the coordinated activity of RIF1 and MCM proteins.

A de novo transcription-dependent TAD boundary underpins critical multiway interactions during antibody class switch recombination.

Interactions between transcription and cohesin-mediated loop extrusion can influence 3D chromatin architecture. However, their relevance in biology is unclear. Here, we report a direct role for such interactions in the mechanism of antibody class switch recombination (CSR) at the murine immunoglobulin heavy chain locus (Igh). Using Tri-C to measure higher-order multiway interactions on single alleles, we find that the juxtaposition (synapsis) of transcriptionally active donor and acceptor Igh switch (S) sequences, an essential step in CSR, occurs via the interaction of loop extrusion complexes with a de novo topologically associating domain (TAD) boundary formed via transcriptional activity across S regions. Surprisingly, synapsis occurs predominantly in proximity to the 3' CTCF-binding element (3'CBE) rather than the Igh super-enhancer, suggesting a two-step mechanism whereby transcription of S regions is not topologically coupled to synapsis, as has been previously proposed. Altogether, these insights advance our understanding of how 3D chromatin architecture regulates CSR.

DNA replication timing directly regulates the frequency of oncogenic chromosomal translocations.

Chromosomal translocations result from the joining of DNA double-strand breaks (DSBs) and frequently cause cancer. However, the steps linking DSB formation to DSB ligation remain undeciphered. We report that DNA replication timing (RT) directly regulates lymphomagenic Myc translocations during antibody maturation in B cells downstream of DSBs and independently of DSB frequency. Depletion of minichromosome maintenance complexes alters replication origin activity, decreases translocations, and deregulates global RT. Ablating a single origin at Myc causes an early-to-late RT switch, loss of translocations, and reduced proximity with the immunoglobulin heavy chain (Igh) gene, its major translocation partner. These phenotypes were reversed by restoring early RT. Disruption of early RT also reduced tumorigenic translocations in human leukemic cells. Thus, RT constitutes a general mechanism in translocation biogenesis linking DSB formation to DSB ligation.

PHF3 regulates neuronal gene expression through the Pol II CTD reader domain SPOC.

The C-terminal domain (CTD) of the largest subunit of RNA polymerase II (Pol II) is a regulatory hub for transcription and RNA processing. Here, we identify PHD-finger protein 3 (PHF3) as a regulator of transcription and mRNA stability that docks onto Pol II CTD through its SPOC domain. We characterize SPOC as a CTD reader domain that preferentially binds two phosphorylated Serine-2 marks in adjacent CTD repeats. PHF3 drives liquid-liquid phase separation of phosphorylated Pol II, colocalizes with Pol II clusters and tracks with Pol II across the length of genes. PHF3 knock-out or SPOC deletion in human cells results in increased Pol II stalling, reduced elongation rate and an increase in mRNA stability, with marked derepression of neuronal genes. Key neuronal genes are aberrantly expressed in Phf3 knock-out mouse embryonic stem cells, resulting in impaired neuronal differentiation. Our data suggest that PHF3 acts as a prominent effector of neuronal gene regulation by bridging transcription with mRNA decay.

Spt5-mediated enhancer transcription directly couples enhancer activation with physical promoter interaction.

Active enhancers are frequently transcribed, yet the regulatory role of enhancer transcription remains debated. Here, we depleted the RNA polymerase II pausing and elongation factor Spt5 in activated mouse B cells and found that approximately 50% of enhancer-gene pairs showed co-regulated transcription, consistent with a potential functional requirement for enhancer transcription. In particular, Spt5 depletion led to loss of super-enhancer-promoter physical interaction and gene expression at the immunoglobulin heavy-chain locus (Igh), abrogating antibody class switch recombination. This defect correlated strictly with loss of enhancer transcription but did not affect acetylation of histone H3 at lysine 27, chromatin accessibility and occupancy of Mediator and cohesin at the enhancer. Strikingly, CRISPRa-mediated rescue of enhancer transcription in Spt5-depleted cells restored Igh gene expression. Our work suggests that Spt5-mediated enhancer transcription underlies the physical and functional interaction between a subset of active enhancers and their target promoters.

Regulation of RNA polymerase II processivity by Spt5 is restricted to a narrow window during elongation.

Spt5 is a highly conserved RNA polymerase II (Pol II)-associated pausing and elongation factor. However, its impact on global elongation and Pol II processivity in mammalian cells has not been clarified. Here, we show that depleting Spt5 in mouse embryonic fibroblasts (MEFs) does not cause global elongation defects or decreased elongation rates. Instead, in Spt5-depleted cells, a fraction of Pol II molecules are dislodged during elongation, thus decreasing the number of Pol II complexes that complete the transcription cycle. Most strikingly, this decrease is restricted to a narrow window between 15 and 20 kb from the promoter, a distance which coincides with the stage where accelerating Pol II attains maximum elongation speed. Consequently, long genes show a greater dependency on Spt5 for optimal elongation efficiency and overall gene expression than short genes. We propose that an important role of Spt5 in mammalian elongation is to promote the processivity of those Pol II complexes that are transitioning toward maximum elongation speed 15-20 kb from the promoter.

R Loops in the Regulation of Antibody Gene Diversification.

For nearly three decades, R loops have been closely linked with class switch recombination (CSR), the process that generates antibody isotypes and that occurs via a complex cascade initiated by transcription-coupled mutagenesis in switch recombination sequences. R loops form during transcription of switch recombination sequences in vitro and in vivo, and there is solid evidence that R loops are required for efficient class switching. The classical model of R loops posits that they boost mutation rates by generating stable and long tracts of single-stranded DNA that serve as the substrate for activation induced deaminase (AID), the enzyme that initiates the CSR reaction cascade by co-transcriptionally mutating ssDNA in switch recombination sequences. Though logical and compelling, this model has not been supported by in vivo evidence. Indeed, several reports suggest that R loops may not be involved in recruiting AID activity to switch regions, meaning that R loops probably serve other unanticipated roles in CSR. Here, I review the key findings in this field to date and propose hypotheses that could help towards elucidating the precise function of R loops in CSR.

DNA Replication Origins in Immunoglobulin Switch Regions Regulate Class Switch Recombination in an R-Loop-Dependent Manner.

Class switch recombination (CSR) at the immunoglobulin heavy chain (IgH) locus generates antibody isotypes. CSR depends on double-strand breaks (DSBs) induced by activation-induced cytidine deaminase (AID). Although DSB formation and repair machineries are active in G1 phase, efficient CSR is dependent on cell proliferation and S phase entry; however, the underlying mechanisms are obscure. Here, we show that efficient CSR requires the replicative helicase, the Mcm complex. Mcm proteins are enriched at IgH switch regions during CSR, leading to assembly of facultative replication origins that require Mcm helicase function for productive CSR. Assembly of CSR-associated origins is facilitated by R loops and promotes the physical proximity (synapsis) of recombining switch regions, which is reduced by R loop inhibition or Mcm complex depletion. Thus, R loops contribute to replication origin specification that promotes DSB resolution in CSR. This suggests a mechanism for the dependence of CSR on S phase and cell division.

Epigenetic targeting of activation-induced cytidine deaminase.

Activation-induced cytidine deaminase (AID) initiates class switch recombination (CSR) and somatic hypermutation (SHM) by deaminating cytosine residues in immunoglobulin genes (Igh, Igκ, and Igλ). At a lower frequency, AID also causes collateral DNA damage at non-Ig loci, including genes that are rearranged or mutated in B-cell lymphoma. Precisely how AID is recruited to these off-target sites is not entirely understood. To gain further insight into how AID selects its targets, we compared AID-mediated translocations in two different cell types, B cells and mouse embryonic fibroblasts (MEFs). AID targets a distinct set of hotspots in the two cell types. In both cases, hotspots are concentrated in highly transcribed but stalled genes. However, transcription alone is insufficient to recruit AID activity. Comparison of genes similarly transcribed in B cells and MEFs but targeted in only one of the two cell types reveals a common set of epigenetic features associated with AID recruitment in both cells. AID target genes are enriched in chromatin modifications associated with active enhancers (such as H3K27Ac) and marks of active transcription (such as H3K36me3) in both fibroblasts and B cells, indicating that these features are universal mediators of AID recruitment.

B cell super-enhancers and regulatory clusters recruit AID tumorigenic activity.

The antibody gene mutator activation-induced cytidine deaminase (AID) promiscuously damages oncogenes, leading to chromosomal translocations and tumorigenesis. Why nonimmunoglobulin loci are susceptible to AID activity is unknown. Here, we study AID-mediated lesions in the context of nuclear architecture and the B cell regulome. We show that AID targets are not randomly distributed across the genome but are predominantly grouped within super-enhancers and regulatory clusters. Unexpectedly, in these domains, AID deaminates active promoters and eRNA(+) enhancers interconnected in some instances over megabases of linear chromatin. Using genome editing, we demonstrate that 3D-linked targets cooperate to recruit AID-mediated breaks. Furthermore, a comparison of hypermutation in mouse B cells, AID-induced kataegis in human lymphomas, and translocations in MEFs reveals that AID damages different genes in different cell types. Yet, in all cases, the targets are predominantly associated with topological complex, highly transcribed super-enhancers, demonstrating that these compartments are key mediators of AID recruitment.

AID targeting in antibody diversity.

Antibody maturation requires class switch recombination (CSR) and somatic hypermutation (SHM), both of which are initiated by activation-induced cytidine deaminase (AID). AID deaminates cytosine residues resulting in mismatches that are differentially processed to produce double-strand breaks in Ig switch (S) regions that lead to CSR, or to point mutations in variable (V) exons resulting in SHM. Although AID was first thought to be Ig-specific, recent work indicates that it also targets a diverse group of non-Ig loci, including genes such as Bcl6 and c-myc, whose modification by AID results in lymphoma-associated mutations and translocations. Here, we review the recent literature on AID targeting and the role for transcriptional stalling in recruitment of this enzyme to Ig and non-Ig loci. We propose a model for AID recruitment based on transcriptional stalling, which reconciles several of the key features of SHM, CSR, and lymphoma-associated translocation.

Activation-induced cytidine deaminase targets DNA at sites of RNA polymerase II stalling by interaction with Spt5.

Activation-induced cytidine deaminase (AID) initiates antibody gene diversification by creating U:G mismatches. However, AID is not specific for antibody genes; Off-target lesions can activate oncogenes or cause chromosome translocations. Despite its importance in these transactions little is known about how AID finds its targets. We performed an shRNA screen to identify factors required for class switch recombination (CSR) of antibody loci. We found that Spt5, a factor associated with stalled RNA polymerase II (Pol II) and single stranded DNA (ssDNA), is required for CSR. Spt5 interacts with AID, it facilitates association between AID and Pol II, and AID recruitment to its Ig and non-Ig targets. ChIP-seq experiments reveal that Spt5 colocalizes with AID and stalled Pol II. Further, Spt5 accumulation at sites of Pol II stalling is predictive of AID-induced mutation. We propose that AID is targeted to sites of Pol II stalling in part via its association with Spt5.

Histone H2B monoubiquitination functions cooperatively with FACT to regulate elongation by RNA polymerase II.

Over the past years, a large number of histone posttranslational modifications have been described, some of which function to attain a repressed chromatin structure, while others facilitate activation by allowing access of regulators to DNA. Histone H2B monoubiquitination is a mark associated with transcriptional activity. Using a highly reconstituted chromatin-transcription system incorporating the inducible RARbeta2 promoter, we find that the establishment of H2B monoubiquitination by RNF20/40 and UbcH6 is dependent on the transcription elongation regulator complex PAF, the histone chaperone FACT, and transcription. H2B monoubiquitination facilitates FACT function, thereby stimulating transcript elongation and the generation of longer transcripts. These in vitro analyses and corroborating in vivo experiments demonstrate that elongation by RNA polymerase II through the nucleosomal barrier is minimally dependent upon (1) FACT and (2) the recruitment of PAF and the H2B monoubiquitination machinery.

PARP-1 determines specificity in a retinoid signaling pathway via direct modulation of mediator.

We show that PARP-1 is indispensable to retinoic acid receptor (RAR)-mediated transcription from the RARbeta2 promoter in a highly purified, reconstituted transcription system and that RA-inducible expression of all RARbeta isoforms is abrogated in PARP-1(-/-) cells in vivo. Importantly, PARP-1 activity was independent of its catalytic domain. PARP-1 directly interacts with RAR and Mediator. Chromatin immunoprecipitation experiments confirmed the presence of PARP-1 and Mediator on RAR-responsive promoters in vivo. Importantly, Mediator was inactive (Cdk8+) under basal conditions but was activated (Cdk8-) upon induction. However, in PARP-1(-/-) cells, Mediator was retained in its inactive state (Cdk8+) upon induction consistent with the absence of gene expression. PARP-1 became dispensable for ligand-dependent transcription in a chromatin reconstituted transcription assay when Mediator was devoid of the Cdk8 module (CRSP). PARP-1 appears to function as a specificity factor regulating the RA-induced switch of Mediator from the inactive (Cdk8+) to the active (Cdk8-) state in RAR-dependent transcription.