A DOT1B/Ribonuclease H2 Protein Complex Is Involved in R-Loop Processing, Genomic Integrity, and Antigenic Variation in Trypanosoma brucei

ABSTRACT The parasite Trypanosoma brucei periodically changes the expression of protective variant surface glycoproteins (VSGs) to evade its host’s immune system in a process known as antigenic variation. One route to change VSG expression is the transcriptional activation of a previously silent VSG expression site (ES), a subtelomeric region containing the VSG genes. Homologous recombination of a different VSG from a large reservoir into the active ES represents another route. The conserved histone methyltransferase DOT1B is involved in transcriptional silencing of inactive ES and influences ES switching kinetics. The molecular machinery that enables DOT1B to execute these regulatory functions remains elusive, however. To better understand DOT1B-mediated regulatory processes, we purified DOT1B-associated proteins using complementary biochemical approaches. We identified several novel DOT1B interactors. One of these was the RNase H2 complex, previously shown to resolve RNA-DNA hybrids, maintain genome integrity, and play a role in antigenic variation. Our study revealed that DOT1B depletion results in an increase in RNA-DNA hybrids, accumulation of DNA damage, and ES switching events. Surprisingly, a similar pattern of VSG deregulation was observed in RNase H2 mutants. We propose that both proteins act together in resolving R-loops to ensure genome integrity and contribute to the tightly regulated process of antigenic variation.

activation of a previously silent ES, a so-called in situ switch, or by recombination of another VSG gene into the active ES.
DNA recombination is the major route of VSG switching and ensures that the full repertoire of VSGs from subtelomeric genome loci can be used (5). DNA accessibility and genome architecture influenced by histone variants H3.V and H4.V impact recombination events in the ES (6). Furthermore, the interplay of several proteins important for homologous recombination is required for this process (7), and it has become clear that DNA lesions trigger recombination events (8,9). Different possibilities for a source of recombination-initiating DNA breaks have been proposed, and a recently published model has suggested the participation of RNA-DNA hybrids (so-called R-loops) (10). Rloops in T. brucei are generated by transcription (11) and usually removed by the two types of RNase H (RH1 and RH2) (12)(13)(14). DNA lesions and altered VSG expression were observed if R-loops were not processed properly (12,13) or if their formation at the ES was not accurately coordinated by the telomere-associated protein RAP1 (15).
The monoallelic expression of VSG from one ES and the coordinated switching to another ES are tightly controlled at multiple levels (16). Chromatin and chromatin-associated factors play ubiquitous roles in VSG expression control (17). The chromatin structure of an active ES is distinct from that of a silent one: the active ES is depleted of nucleosomes (18,19) and is instead enriched for the high-mobility group box protein TDP1 (20). The active ES is transcribed from an unusual extranucleolar locus called the expression site body (ESB) by RNA polymerase I (RNA Pol I) (21). Proteins associated with this ESB, such as the class I transcription factor A (CITFA) complex or the VSGexclusion (VEX) complex, are required for RNA Pol I recruitment and monoallelic transcription, respectively (22,23). Interestingly, R-loops also seem to have an impact on transcriptional control; after depletion of monomeric RH1 or the catalytic subunit of trimeric RH2, trypanosomes derepress silent ESs and express different VSGs on their surface (12,13). Furthermore, transcriptional control is also regulated by nucleosome assembly and chromatin remodeling at the promoter region of the ES, mediated by the ISWI (24), FACT (25), CAF-1b, and ASF1A protein complexes (26). Not only the positioning of nucleosomes but also their posttranslational modifications (PTMs) play a role in ES regulation. For example, the conserved histone methyltransferase DOT1B impacts ES expression on multiple levels (27). DOT1 and its mono-, di-, and trimethylation of histone H3 lysine 79 (H3K79me1/2/3) function in several nuclear processes in eukaryotes, including transcription, the DNA damage response, and telomeric silencing (28). Knowledge of DOT1 interactions with other H3K79 readers has provided valuable insights into these functions. For instance, competition between H3K79 methylation by Saccharomyces cerevisiae Dot1p and the binding of silent information regulator (Sir) proteins to the same chromatin target are involved in the formation of heterochromatin at telomeres (29,30). Additionally, H3K79 methylation is associated with actively transcribed genes (31). The interaction between human DOT1L and the phosphorylated C-terminal domain of RNA Pol II could provide a recruiting mechanism of DOT1L to such active genes (32). DOT1L also interacts directly with members of the AF10 and ENL protein families (33)(34)(35), which are usually found in RNA Pol II transcriptional elongation complexes (36)(37)(38)(39). Incorrect DOT1L recruitment as a result of chimeric gene fusions between DOT1L-interacting proteins AF10/ENL and the histone H4 methyltransferase mixed lineage leukemia (MLL) lead to aberrant H3K79 methylation patterns and increased transcription of oncogenes that cause leukemia (35,40,41), further supporting a role of DOT1B in transcription. DOT1 also helps to maintain genome integrity in mitotic cells by acting through a variety of repair pathways after DNA damage has occurred (28). For instance, binding of the checkpoint adaptor Rad9 to methylated H3K79 in yeast regulates the resectioning step necessary for repair of DNA double-strand breaks (DSBs) by homologous recombination (42). The binding of the Rad9 homolog 53BP1 to methylated H3K79 is also important for detection and repair of DSBs in humans (43) and is further promoted through an interaction of DOT1L with the HLA-B-associated transcript Bat3 (44). Additionally, Dot1p inhibits the error-prone polymerases of the translesion synthesis pathway in response to DNA damage caused by alkylating agents (45). Dot1p is further required for the response to meiotic DSBs and modulates the meiotic checkpoint response (46,47). DOT1 requires a nucleosomal context for its catalytic activity (48) with multivalent interactions to DNA, H2A, H2B, H3 (49,50), and the H4 tail (51). The methylation activity of DOT1 is stimulated by ubiquitination of H2B (52,53). This ubiquitin mark reduces the sampling space of DOT1L on the nucleosome by conformational restriction, which enables DOT1L to catalyze higher methylation states (50). Further work is needed to better understand the exact mechanisms and relative contributions of different regulators and interactors of DOT1.
In contrast to yeast and mammals, trypanosomes possess two DOT1 paralogs, DOT1A and DOT1B, with different enzymatic activities. The mono-and dimethylation of histone H3 on lysine 76 (H3K76) is mediated by DOT1A and is essential for replication initiation (54). In contrast, trimethylation of H3K76 is catalyzed by DOT1B (55). DOT1B knockout (KO) causes a defect in the differentiation from the mammalian-infective stage to the insect-infective stage (56), which is accompanied by severe karyokinesis defects (57). In addition, DOT1B plays an important role in the regulation of VSG transcription: KO cells show derepression of transcriptionally silent ES VSGs and extremely slow ES in situ switch kinetics (27). Furthermore, the attenuation of the active ES in response to inducible expression of an additional VSG gene is not observed in DOT1B-negative parasites (58). The molecular machinery which enables DOT1B to execute regulatory functions at the ES is still elusive, however.
To fully understand how DOT1B can influence these different nuclear processes in T. brucei, we purified DOT1B protein complexes by using complementary biochemical approaches. Surprisingly, one of the most abundant DOT1B-associated protein complexes was RH2, which is important for the maintenance of genome integrity by resolving R-loops (13). Since a contribution of RH2 to antigenic variation was shown recently (13), we predicted that DOT1B could support this function. Consistent with this hypothesis, we found that DOT1B depletion caused an increased R-loop abundance, accumulation of DNA damage, and aberrant VSG expression throughout the VSG repertoire, which suggests additional recombination events at the ESs.

RESULTS
Identification of novel DOT1B-interacting proteins in trypanosomes. To understand the different DOT1B-dependent processes in trypanosomes, it is essential to identify the molecular components that are involved in these distinct functions. Attempts to purify DOT1B under endogenous expression levels using conventional epitope-based methods were consistently unsuccessful (our unpublished data). We therefore employed a highly efficient tandem affinity purification (TAP) approach with the improved PTP (ProtC-TEV-ProtA) tag in order to find DOT1B-interacting proteins (59). The PTP tag was fused at the 59 end of one allele of DOT1B in the mammalian-infective bloodstream form (BSF) stage of the parasite. The second allele of DOT1B was knocked out using homologous recombination (see Fig. S1A in the supplemental material). Correct integration of the targeting and KO constructs was confirmed by PCR analysis of genomic DNA (Fig. S1B). The tagging did not significantly affect the enzymatic activity of DOT1B, because the trimethylation level of H3K76 was comparable to that of wild-type (WT) cells (Fig. S1C). Affinity purification was performed with protein extracts of parasites that expressed the PTP-DOT1B fusion protein.
Extracts of WT cells served as a control. The purified proteins, along with fractions obtained during the procedure, were analyzed by Western blotting (WB) using anti-DOT1B antibodies. The WBs confirmed the successful enrichment of DOT1B (Fig. S2A).
To elucidate the composition of a potential DOT1B complex, we compared the purified samples of PTP::DOT1B cells with those from WT cells using label-free quantitative proteomics. As expected, DOT1B was significantly enriched (P , 0.05) in the eluate fraction of the PTP-DOT1B BSF pulldown, as well as 22 other proteins (Fig. 1A). Three proteins were significantly enriched in the control lysates. The whole data set is summarized in Table S1, tab S1A. As DOT1B has a nuclear localization (60), and acts as a histone methyltransferase on chromatin (56), we further focused on proteins predicted to be nuclear. Four proteins besides DOT1B were categorized as such based on nuclear proteome data (61), TrypTag (60), and relevant literature (as indicated in Table S1, tab S1A).
Unexpectedly, we copurified all three subunits of the RH2 complex (RH2A, RH2B, and RH2C) (62), recently described in trypanosomes by Briggs et al. (13). There are two types of RH in eukaryotes (RH1 and RH2) with overlapping functions in genome stability mediated by removal of R-loops (63). In contrast to RH1, RH2 is also involved in recognizing and cleaving single ribonucleotides falsely incorporated in DNA in a process called ribonucleotide excision repair (RER) (64). Both trypanosomal RH enzymes participate in antigenic variation by processing R-loops (12,13). Furthermore, RH2 has a role in RNA Pol II transcription in T. brucei (13).
To determine whether these interactions are specific for the mammalian-infective stage, we also carried out affinity purification of DOT1B in the vector-specific procyclic form (PCF) and analyzed the results by mass spectrometry (MS). In this instance, we used an improved control cell line, which expressed the PTP tag alone (Fig. S2B). Eleven proteins were significantly enriched together with DOT1B in the PCF PTP:: DOT1B TAP, and one was significantly enriched in the control TAP ( Fig. 1B; Table S1, tab S1B). Remarkably, we again purified two members of the RH2 complex.
In summary, we identified for the first time candidate DOT1B-interacting proteins. Moreover, the results indicate that the association of the RH2 complex with DOT1B is not stage specific. Interestingly, all of the DOT1B copurified nuclear proteins are potential novel interactors, as to our knowledge there are no homologs to these candidates in the list of DOT1 interactors from other organisms. We decided to focus further analysis on the most prominent candidate, the RH2 complex.
Verification of the DOT1B-RH2 interaction. In order to validate the interaction between DOT1B and RH2, we performed reciprocal coimmunoprecipitations (co-IPs) with hemagglutinin (HA)-tagged RH2 subunits. Both alleles of RH2A were endogenously tagged at the 39 end using homologous recombination (Fig. S3A). Integration of the targeting constructs at the endogenous loci was confirmed by PCR analysis of genomic DNA (Fig. S3B). Cells expressing the two tagged alleles grew at a rate comparable to that of controls, demonstrating that the RH2A::HA alleles were functional (Fig. S3C). The cell line expressing RH2A-HA and control WT cells were used in a co-IP experiment with preimmobilized HA antibody-Sepharose. The successful enrichment of RH2A-HA was monitored by WB analysis using anti-HA antibodies (Fig. S3D). MS analysis of eluates identified 17 proteins significantly enriched with RH2A, including both other RH2 subunits as well as DOT1B, confirming the interaction of DOT1B with DOT1B lysates compared to control cell lysates, and the y axis shows the P value (Welch t test) of the four biological replicates. One of the most prominent novel interactors identified in both life cycle stages was the RH2 complex (subunits RH2A, RH2B, and RH2C). Complete data sets are given in Table S1, tabs S1A and S1B, in the supplemental material. the RH2 complex ( Fig. 2A; Table S1, tab S1B). The same IP-MS approach was applied to the other RH2 subunits. While endogenous HA tagging of the RH2B subunit alleles could be confirmed by diagnostic PCR, we could neither detect nor purify this fusion protein biochemically (data not shown). Tagging of the RH2C alleles was similarly confirmed by diagnostic PCR (Fig. S4A and B). In this instance, however, the RH2C-HA protein could be readily immunoprecipitated (Fig. S4C). Mass spectrometry analysis of coimmunoprecipitating proteins yielded an enrichment of 77 candidates, including the other subunits, H2A and H2B, and again DOT1B ( Fig. 2B; Table S1, tab S1D). Apart from the RH2 subunits and DOT1B, the only other overlapping nuclear candidate enriched in both co-IPs was DOT1B's paralog, DOT1A (56).
We combined the candidates obtained from the subunit A and subunit C purifications and categorized them according to their location based on curated gene ontology (GO) terms (TriTrypDB) (Fig. 2C). Since it has been shown that RH2A is a core protein of a nuclear complex (13), we focused on the 26 nuclear proteins annotated (Table S1, tab S1E). We further categorized the nuclear proteins into the different predicted or curated biological processes ( Fig. 2C; Table S1, tab S1E). These proteins are mainly involved in RNA processing and DNA repair, but there were, besides DOT1B, other chromatin-associated proteins such as DOT1A (56) or the ISWI complex protein RCCP (24). Additionally, a potential Flap endonuclease 1 (FEN-1) was enriched. FEN-1 is involved in several mechanisms of RNA processing like RER, Okazaki fragment processing, or stalled replication fork rescue in other eukaryotes (65). In Trypanosoma cruzi, FEN-1 participates in DNA replication and repair (66). FEN-1 was also described in an iPOND (isolation of proteins on nascent DNA) MS screen to be associated with replication in T. brucei (67). Furthermore, the spliceosome component Sm-G was enriched with RH2 (68). A short list of these more promising candidates was compiled (Fig. 2D).
DOT1B-mediated proximity labeling identifies 152 near neighbors. To further validate potential DOT1B-interacting proteins identified by TAP and screen for candidate DOT1B binding partners under native conditions, we next carried out proximity-dependent biotin labeling (BioID). This technique is based on the fusion of a protein of interest to the modified bacterial BirA* biotin ligase and has the advantage that the bait protein can function under native conditions throughout the cell cycle while biotinylating all nearby proteins in a proximity-dependent fashion. Biotinylated proteins can then be purified with high stringency using streptavidin beads (69). We induced the expression of a DOT1B-Myc-BirA* fusion protein for 24 h and incubated PCF cells with 50 mM biotin for another 24 h. Microscopy analysis of DOT1B-Myc-BirA*expressing cells using fluorescent avidin revealed a strong biotinylation signal in the nucleus, which was absent from controls ( Fig. S5A). Biotinylated proteins and associated interaction partners were purified in quadruplicate using streptavidin-coated beads, and uninduced cells served as a control. WB analysis of eluate fractions of DOT1B-Myc-BirA*-expressing cells using streptavidin revealed additional bands compared to the uninduced control (Fig. S5B). Eluted fractions were also analyzed by MS. In total, we identified 152 potential near neighbors, including 11 directly biotinylated proteins ( Fig. 3A; Table S1, tab S1F). All RH2 subunits were significantly enriched, notably by direct biotinylation of subunit A. 79% of the identified proteins were classified as nuclear (based on their curated GO components published on TriTrypDB), including proteins predicted or known to be involved in replication, RNA processing, and transcription (Fig. 3B).
Even though we could not confirm any proteins previously identified by TAP to be DOT1B associated other than the RH2 subunits, we have expanded the list of potential DOT1B interaction partners with other candidates. In addition, we have confirmed an association between DOT1B and the RH2 complex under native conditions, because all RH2 subunits could be identified. Interestingly, we could detect biotinylation only of the A subunit, which suggests that this subunit was closest to DOT1B and provides the interface between DOT1B and the RH2 complex. Yeast two-hybrid data supported the assumption that subunit A of RH2 interacts directly with DOT1B (Fig. S6).
These results, along with all the other experiments mentioned above, support a direct interaction between DOT1B and the RH2 complex, most likely with the RH2A subunit forming the interface of this novel complex.
Contribution of DOT1B to R-loop resolution. We next wanted to understand the function of the DOT1B-RH2 interaction. Depletion of subunit A of the RH2 complex leads to cell cycle stalling with accumulation of R-loops and DNA damage (13). Both phenomena were mapped to transcription initiation sites of RNA Pol II, as well as to the active and inactive ESs. As a consequence, the expression of several genes (including VSGs) was altered, which led to changes in the surface coat composition. Since VSGs throughout the repertoire were derepressed, recombination-based switching events were suggested. In addition, monoallelic expression was impaired, which resulted in the simultaneous expression of two VSGs on the surface of the mutant parasites (13).
In contrast, DOT1B is not essential for parasitic viability (56), indicating that the RH2 protein complex very likely also possesses DOT1B-independent functions. Although transcription-associated proteins were enriched in the DOT1B BioID, no previous experimental data have suggested an influence of DOT1B on RNA Pol II expression to date. Derepression of silent ES-associated VSGs and slow ES switching kinetics have been observed in DOT1B-depleted cells, however (27). Alterations in recombination-based switching have not been investigated. An overlapping contribution of both proteins to antigenic variation could be via regulation of genome integrity. We therefore asked whether DOT1B might support RH2A in minimizing DNA damage caused by R-loop accumulation, which might be a novel route to initiate recombination-based VSG switching (12,13,15). To test this hypothesis, we analyzed R-loop formation, DNA damage accumulation, and VSG switching in DOT1B-depleted cells.
First, we compared genome-wide R-loop formation in DOT1B-depleted cells with that of WT cells in a quantitative dot blot analysis (Fig. 4A). Genomic DNA was analyzed using the RNA-DNA hybrid-specific S9.6 antibody (70). DOT1B RNA interference (RNAi) cells were analyzed instead of DOT1B KO cells in order to capture early events after protein depletion. DOT1B depletion upon induction of RNAi was confirmed by monitoring the loss of DOT1B-specific H3K76me3 methylation and the concomitant increase of H3K76me2 methylation by WB (Fig. S7A). To have a suitable endogenous positive control for this assay, we included DNA from RH2A-depleted cells. RNAi against RH2A resulted in a growth defect (Fig. S7B), as previously described (13). As an additional positive control, synthetic RNA-DNA hybrids were generated by a firststrand synthesis reaction using reverse transcriptase and T. brucei whole-cell RNA extracts as a template. S9.6 antibody specificity was also verified by digestion of half of the samples with recombinant Escherichia coli RNase H (EcRH), which specifically cleaves RNA in RNA-DNA hybrids (71). Strikingly the quantification of the signals clearly demonstrated a 2.5-fold enrichment of R-loop structures in 72-h DOT1B-depleted cells compared to WT cells, suggesting that DOT1B is involved in R-loop processing (Fig. 4B). The signal enrichment was observed in both 48-h RH2A-depleted and cDNA samples. EcRH sensitivity confirmed the specificity of the assay (Fig. 4A, lower blot).
We next wanted to examine whether there is also an accumulation of DNA damage in DOT1B-depleted cells. To test this, we monitored DNA damage after RNAi-mediated depletion of DOT1B in quantitative WB analysis using an antibody specific for phosphorylated histone H2A (gH2A), which is a marker for DNA damage (Fig. 4C) (72). Whole-cell lysates from cultures at 0 h, 24 h, 48 h, and 72 h postinduction of DOT1B RNAi were compared to whole-cell lysates of WT cells. An increased gH2A signal has previously been observed in RH2A-deficient cells (13), so whole-cell lysates at 0 h, 24 h, 48 h, and 72 h after RH2A RNAi induction were used as positive controls. The gH2A signal was normalized to paraflagellar rod (PFR1,2) protein levels. The quantification revealed an accumulation of DNA damage in DOT1B-depleted cells (Fig. 4D). This damage accumulated over time and resulted in a 3-fold increase after 72 h of RNAi induction in comparison to uninduced and WT samples. Interestingly, the accumulation of damaged chromatin appears to be less dramatic than the 6-fold increase of DNA damage after 72 h of RH2A subunit depletion.
We had previously observed a massive accumulation of DNA damage in DOT1B-deficient cells during differentiation from BSF into PCF trypanosomes but not in DOT1B KO BSF cell culture (57). The assumption was that damage accumulation in DOT1B mutants was differentiation dependent. To assess whether these observations were based on overlooked compensation/adaptation processes after prolonged culture of DOT1B KO cells, we reanalyzed DNA damage in a freshly generated KO cell line. A new KO cell line was generated in BSF trypanosomes by replacement of both alleles with drug selection markers in two consecutive transfections. Amplification of the DOT1B locus by PCR and analysis of H3K76me3 and H3K76me2 signals in WB confirmed the deletion of DOT1B ( Fig. S8A and B). The new DOT1B KO cells were cultured for a period of 36 days posttransfection of the cassette that replaced the second allele of DOT1B. Whole-cell lysates of WT and DOT1B KO cells after 14 days, 21 days, 28 days, and 35 days posttransfection were prepared and analyzed in a quantitative gH2A WB. PFR1,2-normalized gH2A signals were 2.8-fold increased at the 14-day time point in the DOT1B KO samples compared to the WT samples. The gH2A signal then declined in subsequent time points, from the initial 2.8-fold increase after 14 days to 1.7-fold after 35 days (Fig. S8C). This supports the hypothesis that trypanosomes can partially compensate for the loss of DOT1B by an unknown adaptation process.
The initial increase of DNA damage in DOT1B KO cells (Fig. S8C), and the comparable amount (3-fold) of DNA damage after depletion of DOT1B by RNAi (Fig. 4C), supports the hypothesis that DOT1B influences genome integrity. Individual DSBs in subtelomeric regions of the active or inactive ES, as well as at chromosome-internal loci, are differentially tolerated by trypanosomes (8,17,(73)(74)(75)(76). Cells show divergent growth outcomes depending on the life cycle stage as well as the extent and location of the damage, which can be explained in part by different cell cycle responses or survival rates. Given that DOT1B is important for genome integrity, we asked whether DOT1B-deficient cells also show growth perturbations. We compared the growth of DOT1B RNAi cells after tetracycline induction to that of uninduced and WT cells. DOT1B-depleted cells showed a slight delay as the depletion time progressed (Fig. S7C). In contrast, DOT1B null mutants showed a severe growth defect after KO generation that gradually reverted back to WT rates over a 5-week culturing period, similar to the amount of DNA damage (Fig. S8D).
DOT1B depletion causes altered VSG expression pattern. Since depletion of DOT1B, like RH2A depletion, affects R-loop levels and is accompanied by reduced genome integrity, we wanted to investigate DOT1B's contribution to the regulation of VSG switching. First, the expression of cell surface VSGs was analyzed in DOT1B RNAi cells by flow cytometry with a monoclonal antibody specific for the predominant VSG-2 isoform. As a control, VSG-13-expressing PN13 cells were included (27). WT cells and uninduced DOT1B RNAi cells were VSG-2 positive, whereas the PN13 cells were VSG-2 negative (Fig. 5A). After 72 h of DOT1B depletion, 7% of the parasites no longer expressed the original VSG-2, suggesting that these cells switched on another VSG gene (Fig. 5A, fourth panel).
To examine which VSGs are expressed in the VSG-2-negative parasite population, we released soluble VSGs from the cell surface of WT cells and DOT1B RNAi cells at 0 h, 48 h, and 72 h after RNAi induction and analyzed them by MS. The protein level of 17 VSGs was more than 2-fold upregulated compared to that of WT cells after 72 h of DOT1B RNAi induction ( Fig. 5B; Fig. S9). Almost half of those VSGs were associated with previously silent BSF ESs (4). In addition, six were detected which are normally only expressed from metacyclic ESs (MES) (77). Strikingly, we also detected several VSGs that are usually located in the subtelomeric arrays of the megabase chromosomes, such as VSG-636. We further analyzed the VSG expression pattern after different culture time points of the newly generated DOT1B KO cells relative to WT cells by MS analysis of whole-cell lysates. Almost the same set of deregulated VSGs was observed at early time points after DOT1B deletion as in the DOT1B-depleted RNAi cells (Fig. S8E Table S1, tab S1G). At late time points after recovery from the DNA damage phenotype, only ES-associated VSGs could be detected.
In addition, we compared DOT1B RNAi with RH2A RNAi cells to assess whether the proportions of the expressed VSG repertoires are similar, further suggesting a shared regulatory function of both proteins in antigenic variation. The abundance of 15 surface VSGs was 2-fold upregulated 72 h after depletion of RH2A compared to the expressed VSGs of WT cells. Similar to DOT1B-depleted cells, VSGs associated with BES and MES but also VSGs from subtelomeric regions were detected after RH2A depletion ( Fig. S9 and Table S1, tab S1H). These VSG genes are most likely activated by recombination into an active BES. Strikingly, in the two cell lines, 80% of the expressed VSGs were identical 72 h postinduction of RNAi, although the percentage of new VSGs in DOT1B-depleted parasites was higher than in RH2A-depleted parasites.
Taken together, our data revealed a novel interaction between DOT1B and RH2 that exists in the mammalian-infective BSF and in the insect PCF of trypanosomes. Since DOT1B-deficient cells showed increased R-loop formation, compromised genome integrity, and a similar deregulated VSG pattern as observed after RH2A depletion, we propose a common function for this complex in regulating antigenic variation by modulation of R-loop structures.

DISCUSSION
DOT1 and the histone modifications it mediates are highly conserved across many species. In yeast and humans, knowledge about DOT1 regulation as well as its contribution to transcription, meiotic checkpoint control, and the DNA DSB response was increased substantially by the discovery of DOT1-binding factors. In contrast, nothing is known about DOT1-associated proteins in T. brucei. DOT1B and other factors involved in chromatin modifications, transcriptional regulation, or telomere structure maintenance were shown to influence antigenic variation in trypanosomes, but our understanding of the compositions and interplay of these molecular machineries remains incomplete. The identification of a DOT1B interactome would improve our understanding of the tightly regulated process of antigenic variation.
In this study, we identified DOT1B-associated proteins using a combination of complementary methods in T. brucei. TAP detected 23 proteins to be significantly enriched in BSF and 12 in PCF, whereas 152 proteins were enriched due to proximity labeling with DOT1B in PCF. The list contains proteins involved in RNA processing, including splicing, as well as replication and transcription. One of the most abundant interactors and the only overlap between all individual data sets was the RH2 complex with its three subunits. The minor overlap of the two complementary approaches may indicate many weak or transient interactions of candidates making them prone to dissociate during the affinity purification step. Alternatively, the substantially larger number of candidates detected in the BioID screen might reflect a ubiquitous presence of DOT1B at chromatin, since DOT1B is likely to be responsible for the entire trimethylation of H3 during the cell cycle and therefore probably in close proximity to many other chromatin-related proteins, such as those involved in replication and transcription. Although we tried to rule out nonspecific binding of proteins to the PTP tag by ectopic expression of the tag in our control cell line, often-reported contaminants such as ribosomal subunits were present in our enriched fractions. Hence, further experiments are needed to validate the list of remaining candidates. Reciprocal IPs and yeast two-hybrid analysis were both used to validate and define the mode of interaction between DOT1B and trimeric RH2 subunits in this study.
Our data did not include previously described DOT1 nuclear interactors found in other species, such as members of the YEATS domain family that are among the beststudied interactors of mammalian DOT1 (33,34,38,78,79). Searching the T. brucei proteome with human ENL or AF9 by BLAST revealed a potential YEATS protein homolog, but this protein was not significantly enriched in our assays, suggesting that this interaction is likely not conserved in trypanosomes. However, this specific regulatory process seems unlikely in trypanosomes, because the generation of di-and trimethylation of H3K76 in trypanosomes is ensured by the division of this task between two enzymes, DOT1A and DOT1B (55,56). Furthermore, a long C-terminal extension outside of the conserved HMT domain of DOT1L is essential for the interaction with DNA and other proteins, including the interaction with ENL/AF9 (33,34). Trypanosomal DOT1B comprises only the conserved HMT domain and lacks such long N-or C-terminal extensions present in yeast and human, indicating that interaction mechanisms might differ across species or might be indeed species specific.
In this study, we identified a novel interaction between DOT1B and RH2 in BSF and PCF parasites, most likely mediated by a DOT1B/RH2 subunit A interface. To date, there are very few studies describing the mode of interactions of the RH2 complex in other organisms. In addition to a structural role to support the activity of RH2A's catalytic subunit, it was speculated that the other subunits might be involved in interactions with other proteins (80). Indeed, the RH2B subunit was shown to act as a platform to interact with PCNA via the conserved PCNA-interacting protein motif (so called PIPbox) (80). The only additional interactions with recruiting functions are mediated by an association of RH2A with BRCA2 (81) in humans and with the telomeric protein RIF2 in yeast (82). Our IPs of subunit A and C in trypanosomes revealed 26 nuclear proteins. We could not find conservation of the above-mentioned interactions, but the detection of FEN1, another important player in the RER pathway downstream of RH2 (83), indicates a conserved role for RH2 in RER, a pathway that has not yet been investigated in trypanosomes. The IP of RH2B was not successful. However, it is possible that an interaction with PCNA also happens in trypanosomes. Sequence analysis in TriTrypDB suggests two putative trypanosomal RH2B isoforms and revealed that a PIP-box motif is also present at the N-terminal end of the longer RH2B isoform.
Subsequent work in this study investigated the purpose of this novel interaction with RH2 in trypanosomes. In contrast to DOT1B, RH2A is essential (13,56); hence, it is likely that RH2 has DOT1B-independent functions. As RH2 contributes to R-loop removal in trypanosomes, and increased R-loop levels are accompanied by elevated DNA damage and VSG switching (12,13), we speculated that the influence on antigenic regulation might be the overlapping function of the two enzymes. Indeed, we showed that DOT1B is an additional player affecting R-loops, as R-loop formation increased in DOT1B-depleted cells. We found additional proteins described to bind to R-loops (84) and prevent their formation (85) in the BioID data set. These include members of the spliceosome (86,87), the exosome (88), topoisomerase 1 (89), and an associated bromodomain-containing protein BRD2 (90), transcription elongation factor TFIIS (91), or FACT (92). These proteins are involved in the resolution of genome maintenance conflicts caused by R-loop formation in mammals. Interestingly, TAP of human DOT1L also identified an R-loop-interacting protein: the helicase DDX21 (93,94). Further studies are needed to evaluate whether those proteins also contribute to R-loop modulation in trypanosomes and whether this is dependent on an interplay with DOT1B.
Depletion of DOT1B results in altered VSG expression. Interestingly, we detected almost the same changes in VSG expression pattern in DOT1B-and RH2A-deficient BSF cells. Activation of new VSGs does not occur randomly (95,96). VSG recombination events first affect BES VSGs, followed by subtelomeric VSG arrays (95). VSG recombination events between BES can be initiated by a break in the 70-bp repeats of the active ES and are most likely accompanied by chromatin conformational changes (6,8,76). We observed similar patterns of VSG expression as reported after break induction in the 70-bp repeats of the active ES (VSG-3, VSG-17, VSG-11) (76) or after H3.V and H4.V deletion (VSG-8, VSG-11) (6). Moreover, VSGs from subtelomeric regions (VSG-775, VSG-336) were expressed. Both indicate that these VSGs may have been transferred into the active ES by homologous recombination. We propose that VSG switching may happen via inefficiently processed R-loops, which lead to DSBs, and in turn can lead to a switching event by homologous recombination as described recently (12,13). Unfortunately, chromatin immunoprecipitation (ChIP) experiments with an antibody specific for gH2A did not show any statistically significant enrichment of DNA damage at ES, most likely due to a very low yield of precipitated chromatin (data not shown).
One possibility for how DOT1B could be involved in R-loop-driven antigenic variation might be a recruiting function of RH2A by DOT1B to R-loop sites in the genome, a task known for RH2 interactors in yeast and humans (81,82). DOT1B is responsible for fast trimethylation of nearly all newly incorporated histones during G 1 phase (our unpublished data) and therefore localizes to every domain of the entire chromatin once per cell cycle, potentially with an R-loop-processing RH2 complex as a backpack. The trimethylation of H3K76 at R-loop-free regions could generate an R-loop-processed memory in this scenario. H3K76me3 distribution assessed by ChIP in comparison to H3 occupancy and sites of R-loop accumulation (14) could be informative in this regard. Another possibility could be that DOT1B modulates RH2 activity. Interestingly, yeast RH2 was shown to perform its functions in G 2 phase of the cell cycle (97). It would be interesting to figure out if this is also true for trypanosomal RH2 and if DOT1B contributes via a cell cycle-dependent interaction. Yeast and human subunit B and subunit C form stable subcomplexes in vitro (80,98), but RH2 is active only after formation of the trimeric complex with subunit A. Therefore, it might be possible that DOT1B plays a role in modulating the activity by recruiting the individual subunits to form an active complex.
Here, we provide the first evidence that DOT1B directly interacts with the RH2 protein complex. This novel physical interaction has further extended our knowledge about DOT1B's function in antigenic variation: we have observed strikingly similar phenotypes in R-loop formation, genome integrity, and deregulation of VSGs after depletion of DOT1B or RH2A. Hence, we propose that both enzymes work together in a molecular complex to influence recombinational switching in African trypanosomes.
For Ddot1b cells, a PCR-based gene deletion approach was used to sequentially replace both alleles of DOT1B with the open reading frames (ORFs) of the resistance markers HYG and BSD in BSF SM cells. HYG was amplified from pHD309 and BSD from pPOTv4 with primers containing 70-bp homology flanks of the DOT1B untranslated regions (UTRs) for homologous recombination. Correct replacement of the ORFs was verified by diagnostic PCR.
For PTP::DOT1B/Ddot1b cells, the first allele of DOT1B (Tb427.1.570) was tagged endogenously at the 59 end with PTP as previously published (104). Briefly, a fragment of the DOT1B ORF (positions 1 to 783) was amplified from genomic DNA and cloned into the p2678_PUR_PTP plasmid. NsiI was used to linearize the vector within the DOT1B fragment prior to transfection of BSF and PCF cells. A PCR-based gene deletion approach was used to replace the second DOT1B allele with the hygromycin phosphotranspherase (HYG) ORF in BSFs or with the blasticidin-S-deaminase (BSD) ORF in PCFs. HYG ORF was amplified from the pHD309 plasmid, and BSD ORF was amplified from the pPOTv4 plasmid with primers containing 60-bp homology flanks to the DOT1B UTRs for recombination. Correct integration of the constructs was verified by PCR using primers annealing in the UTRs of DOT1B and ORFs of the resistance cassettes.
For PTP Ti cells, to express the PTP tag ectopically, its ORF was amplified from p2678 with forward primers containing a HindIII restriction site and reverse primers containing a stop codon and BamHI site. The PTP internal BamHI site was destroyed by site-directed mutagenesis, and the construct was cloned into the pLEW100v5bd plasmid. The plasmid was linearized with NotI prior to transfection of 29-13 PCF cells. Expression was induced with 50 ng/ml tetracycline for 24 h.
For RH2A::HA, RH2B::HA, and RH2C::HA cells, both alleles of RH2A (Tb427.10.5070), RH2B (Tb427.1.4220), and RH2C (Tb427.1.4730) were in situ tagged with a 3ÂHA epitope sequence at the 39 end using a PCRbased method and the pMOTag2H and pMOTag4H plasmids as previously described (105). Correct integration of constructs was tested by using primers annealing in the UTRs of the genes of interest and the ORFs of resistance cassettes.
For DOT1B-Myc-BirATi cells, for inducible expression of the DOT1B-Myc-BirA* fusion protein, the DOT1B ORF (without stop codon) was amplified from genomic DNA and cloned into the pLEW100_Myc_BirA*  (107). 2T1 cells were transfected with SgsI-linearized plasmids, and RNAi of clones with correctly integrated constructs was induced with 1 mg/ml tetracycline.
BioID. BioID was conducted as described previously (106) with minor changes. Expression of DOT1B-Myc-BirA* was induced by addition of 5 ng/ml tetracycline for 24 h, and cells were further incubated for 24 h with 50 mM D-biotin (Invitrogen) and 5 ng/ml tetracycline. Quadruplets of 1 Â 10 9 DOT1B-Myc-BirA*-expressing cells and of the uninduced control cells were harvested by centrifugation (1,800 Â g, 5 min, 4°C). After washing cells three times with PBS, the cells were incubated in 1 ml IP buffer (20ômin, on ice) and then sonicated (6 cycles, 30-s on, 30-s off, settings high). Lysates were cleared by centrifugation (15,000 Â g, 10 min, 4°C) and further stored on ice. Pierce streptavidin agarose beads (Thermo Fisher) were washed twice with 1 ml ice-cold binding buffer (50 mM Na 2 HPO 4 , 50 mM NaH 2 PO 4 , 150 mM NaCl) and equilibrated with 1 ml ice-cold IP buffer (500 Â g, 1 min, 4°C). Fifty-microliter beads were added to each cleared lysate, and the reaction mixtures were incubated under mild agitation (4 h, 4°C). The unbound material was separated from the beads by centrifugation (500 Â g, 5 min, 4°C) and was followed by two washing steps with 1 ml IP buffer (5 min, on ice). Bound proteins were eluted as described above.
Mass spectrometry and data analysis. Samples were run on a Novex Bis-Tris 4%-to-12% gradient gel (Thermo Fisher) with MOPS (morpholinepropanesulfonic acid) buffer (Thermo Fisher) for 10 min at 180 V. The gel was stained with Coomassie blue G250 dye (Biozym). Each lane was cut into pieces, minced, and destained in water/50% ethyl alcohol (EtOH). The gel pieces were dehydrated with pure ACN, reduced with 10 mM DTT (Sigma-Aldrich), and subsequently alkylated with 55 mM iodoacetamide in the dark. The again-dried gel pieces were digested with 1 mmol trypsin at 37°C overnight. The digested peptides were desalted and stored on StageTips (108) until measurement.
Peptides were separated along a 240-min gradient (90 min for VSG analysis) on a EasyLC 1000 ultrahigh-performance liquid chromatography (uHPLC) system using a C 18 reverse-phase column, packed inhouse with Reprosil C 18 (Dr. Maisch GmbH). The column was enclosed in a column oven (Sonation), and the peptides were sprayed into a Q Exactive Plus mass spectrometer (Thermo Fisher). The mass spectrometer was operated in a data-dependent acquisition mode using a top10 method. Spray voltage was set at ca. 2.4 kV.
The aquired raw files were analyzed with MaxQuant (version 1.5.8.2) (109) using the Trypanosoma brucei protein database downloaded from TriTrypDB. Prior to bioinformatics analysis, contaminants, reverse hits, protein groups identified only by site, and protein groups with less than 2 peptides (one of them unique) were removed.
R-loop dot blot. Genomic DNA was isolated from 2.5 Â 10 7 cells using the High Pure PCR template preparation kit (Roche). A 1.1-mg amount of DNA was then either treated with 10 U recombinant E. coli RNase H (NEB) or double-distilled water (ddH 2 O) in 1Â RNase H buffer (2 h, 37°C). All samples were further incubated with 10 mg RNase A (Thermo Fisher) (1 h, 37°C) in a final concentration of 500 mM NaCl. The DNA samples were spotted on a Hybond N1 membrane (GE Healthcare) using a dot blot apparatus in a 2-fold serial reduction and cross-linked with UV (0.12 J). To quantify R-loop formation, the membrane was blocked in 5% milk/PBS (overnight, 4°C). A 1-mg/ml concentration of S9.6 antibody (Merck Millipore) in 0.1% Tween/PBS was then incubated with the membrane (1 h, RT). After three washes with 50 ml 0.2% Tween/PBS, the membrane was incubated with horseradish peroxidase (HRP)-conjugated goat anti-mouse antibody (Dianova) (1:20,000) (1 h, RT). After three washes with 50 ml 0.2% Tween/PBS, the HRP signal was developed with Western Lightning Plus-ECL solutions (PerkinElmer) and imaged using an iBright imaging system (Thermo Fisher). iBright Analysis software was used for quantification of signals.
Isolation of soluble VSGs for MS analysis. Soluble VSGs were isolated in quadruplicate from 4 Â 10 7 cells according to previously published protocols (110). Briefly, cells were precooled (10 min, on ice) prior to harvesting by centrifugation (1,500 Â g, 10 min, 4°C). After a washing step with 10 ml ice-cold TDB (1,500 Â g, 10 min, 4°C), cells were transferred to a microcentrifuge tube in 1 ml ice-cold TDB and pelleted again (1,500 Â g, 10 min, 4°C). Subsequently, the pellet was resuspended in 45 ml 10 mM sodium phosphate buffer supplemented with EDTA-free protease inhibitor cocktail (Roche) and incubated (5 min, 37°C). After the cells were cooled down (2 min, on ice), they were pelleted by centrifugation (14,000 Â g, 5 min, 4°C) and the supernatant containing soluble VSGs was transferred into a new reaction tube. After the addition of 24.5 ml NuPAGE LDS sample buffer (Thermo Fisher) supplemented with 100 mM DTT, the samples were boiled (10 min, 70°C).
Yeast two-hybrid assay. The DOT1B and RH2A ORFs (excluding their stop codon) were amplified from genomic DNA and cloned into the Gal4-binding domain pGBKT7 vector or Gal4 activation domain pGADT7 vector between the NdeI and BamHI sites, respectively. pGADT7-T-AG and pGBKT7-p53 plasmids were used as a positive control, and pGADT7-T-AG with pGBKT7-lamC was used as a negative control (Clontech).
Yeast strain AH109 was cotransformed with binary combinations of the corresponding plasmids using the lithium acetate method. Briefly, 40 ml of culture (optical density at 600 nm [OD 600 ], 1.5) grown in YPDA (20 g/liter Bacto peptone, 10 g/liter yeast extract, 0.2% adenine hemisulfate, pH 5.5) was harvested (700 Â g, 5 min, RT) and resuspended in 2 ml One-Step buffer (0.2 M lithium acetate, 40% PEG 3350, 100 mM DTT). Binary combinations of 1 mg bait and prey plasmids were added to 100 ml of yeast cells and incubated (30 min, 45°C). Cells were plated on tryptophan (trp)-and leucine (leu)-depleted SD plates to select for cotransformed plasmids (4 to 5 days, 30°C). Four to six positive clones were dissolved by rocking at 700 rpm in 100 ml PBS (15 min, RT). Ten microliters of each 1:10, 1:100, and 1:1,000 serial dilution was dropped on SD selection plates (2trp, 2leu) also lacking histidine (his) to select for proteinprotein interactions. Stringency was increased by supplementation of SD selective plates (2trp, 2leu, 2his) with 2 mM 3-aminothiazol (3AT). The plates were incubated for 3 to 4 days at 30°C. An interaction was defined as positive when growth could be observed in at least 75% of the plated clones.
Data availability. Data are available via ProteomeXchange with identifier PXD028588.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.