Oncogenic fusions are major drivers in leukemogenesis and may serve as potent targets for treatment. DUX4/IGHs have been shown to trigger the abnormal expression of ERG through binding to DUX4-Responsive-Element (DRE), which alt leads to B-cell differentiation arrest and a full-ﬂedged B-ALL. Here, we determined the crystal structures of Apo- and DNA -bound DUX4 and revealed a clamp-like transactivation mechanism via the double homeobox domain. DRE HD2 Biophysical characterization showed that mutations in the interacting interfaces signiﬁcantly impaired the DNA binding afﬁnity of DUX4 homeobox. These mutations, when introduced into DUX4/IGH, abrogated its transactivation activity in Reh cells. More importantly, the structure-based mutants signiﬁcantly impaired the inhibitory effects of DUX4/IGH upon B- cell differentiation in mouse progenitor cells. All these results help to deﬁne a key DUX4/IGH-DRE recognition/step in B- ALL. Introduction the insertion of chromosome fragments containing DUX4 gene into the IGH locus, has been reported in ~7% of B- Oncogenic fusions are important causes/targets in leukemia ALL patients [3, 6, 7]. and related treatment [1, 2]. Acute lymphoblastic leukemia Double homeobox 4 (DUX4) is the core component of (ALL) is the most common childhood malignancy. B-ALL, human subtelomeric macrostalelite D4Z4 . It has been which stems from the clonal proliferation of B lineage shown that 11 to >150 D4Z4 repeats can be identiﬁed in progenitors [3, 4], constitutes 80% of ALL cases . Very chromosome 4 and 10. In each of the D4Z4 repeat (~3.3 recently, using next-generation sequencing technologies, a kb), it constitutes a single copy of the intronless DUX4 novel B-ALL oncogenic driver DUX4/IGH, derived from gene. At physiological condition, DUX4 is usually silenced . In pathogenic condition known as facioscapulohumeral muscular dystrophy (FSHD) , the D4Z4 repeats are These authors contributed equally: Xue Dong, Weina Zhang, Haiyan contracted to 1–10. The change of chromatin packing is Wu, Jinyan Huang. thought to be a critical factor that leads to the abnormal expression of DUX4. Consistently, in B-cell ALL, the Electronic supplementary material The online version of this article (https://doi.org/10.1038/s41375-018-0093-1) contains supplementary abnormal expression of DUX4/IGHs derived from chro- material, which is available to authorized users. mosome translocation is also demonstrated to be a driver factor of leukemogenesis . Furthermore, the deregulation * Sai-Juan Chen of DUX4 is frequently associated with the biogenesis of email@example.com ERG . Very brieﬂy, ERG is an ETS-family transcrip- * Guoyu Meng alt firstname.lastname@example.org tion factor playing important role in normal hematopoietic regulation  and is involved in the pathogenesis of T- State Key Laboratory of Medical Genomics, Shanghai Institute of ALL and acute myeloid leukemia [12–14]. ERG , derived alt Hematology, Rui-Jin Hospital, Shanghai JiaoTong University from an abnormal transcription starting from the intron 6 of School of Medicine and School of Life Sciences and −/− ERG, could induce lymphoid leukemia in Arf mice . Biotechnology, Shanghai JiaoTong University, 197 Ruijin Er Therefore, the ERG biogenesis is recognized as a reporter alt Road, Shanghai 200025, China and a potent secondary hit in DUX4-driven leukemogen- Key Laboratory of Systems Biomedicine, Shanghai Center for esis. However, due to the lack of structural information, it is Systems Biomedicine, Shanghai JiaoTong University, 800 Dong Chuan Road, Shanghai 200240, China not clear how DUX4/IGH recognizes, binds, and activates 1234567890();,: 1234567890();,: Structural basis of DUX4/IGH-driven transactivation 1467 the transcription of its target genes via the double homeo- terminus of the protein (Fig. 1b). Consistent with previous box (HD1–HD2) domain. observations [3, 4], human leukemia cell line Reh cells In this report, we determined the crystal structures of harboring the newly identiﬁed DUX4/IGHs showed the Apo-DUX4 and DNA-bound complex. The atomic expression of ERG (Fig. 1c). All the DUX4 truncated HD2 alt structures, together with functional characterization, reveal mutants derived from Chinese patients harbor an intact N- a two-step clamp-like binding mechanism between double terminal HD1–HD2 domain. This is also consistently homeobox and TAATCTAAT DNA site, the latter of which observed in other B-ALL patients reported by the Japanese is frequently observed in DUX4 target genes, and hence and American investigators [3, 4], suggesting that the termed DUX4-Responsive-Element (DRE). Consistent with double homeobox HD1–HD2 is critical for DUX4/IGH previous observation that a single HD domain is not sufﬁ- activity and B-ALL leukemogenesis. Supportively, it has cient for productive protein–DNA interaction, our data been reported that wild-type (WT) DUX4 can selectively showed that both HD1 and HD2 are essential in DRE bind to DNA [4, 15–17]. Using published ChIP-seq data recognition. Single mutation in the major-groove-binding , we have investigated a number of DUX4/IGH target interfaces of HD1 and HD2 can signiﬁcantly impair DUX4- genes including ERG site TAATCTCAT (Supplementary alt DRE binding and DUX4-driven transactivation. When Figure 1a,b). This has led to the discovery of a DUX4/IGH investigated in the context of DUX4/IGH and B-cell dif- consensus binding site TAATCTAAT, and hence termed ferentiation, structure-based mutations also abolish DUX4/ DRE (Supplementary Figure 1c). IGH’s transcriptional activities and their aggressive/inhibitory roles upon B-cell differentiation. All these results have led to Crystal structures of Apo- and DNA -bound DRE the novel proposal that the DUX4-DRE recognition and two- DUX4 HD2 step clamp-like binding mechanism are critical for DUX4/ IGH-driven transactivation and B-ALL leukemogenesis. In order to investigate how DUX4/DUX4 fusions recognize DNA, we have determined the crystal structures of Apo-DUX4 and DNA -bound complex at the resolu- HD2 DRE Materials and methods tions of 1.5 and 2.6 Å, respectively (Fig. 2). The Apo structure is a classic homeobox fold with three α-helices, Information concerning bacterial and eukaryotic cells, termed α1, α2, α3, respectively (Fig. 2a). The N-terminal DNAs, and other materials used in this study is described in α1, α2 helices are anti-parallel. The α3 helix is perpendi- Supplementary Data. The experimental details of structure cular to the axis of α1/α2 helices. Compatible with its DNA determination, functional/genetic characterizations includ- binding activity, the electrostatic surface of Apo-DUX4 HD2 ing biolayer interferometry assay, transactivation assay, reveals a highly positively charged pocket delineated by luciferase assay, B-cell differentiation assay, RNA-seq residues Arg137, Trp141, Arg145, Arg148, and His149 in analysis etc. are also described in Supplementary Data. the α3 helix (Fig. 2b). The statistics of the data collection and structure reﬁnement The HD2-DNA complex structure reveals a clamp-like are shown in Supplementary Table 1. Coordinates of DNA-binding mechanism, in which the N-terminal Apo-DUX4 and DUX4 -DNA have been depos- R RKR loop and the α3 helix interact with the TAAT HD2 HD2 DRE 95 98 ited into the Protein Database Bank. The PDB codes are site in the minor and major grooves, respectively (Fig. 2c, 5Z2S and 5Z2T, respectively. d). In the minor groove, the N-terminal R RKR loop 95 98 adopts an extended conﬁguration making loosen interaction with the DNA backbone (Fig. 2e). In the asymmetry unit Results (ASU), two DUX4 molecules were observed. The N- HD2 terminal R RKR loop is highly ﬂexible and might adopt 95 98 DUX4/IGH-driven transactivation in B-ALL different conﬁgurations for the binding of DNA minor groove. The two HD2 molecules in ASU are superimposed We have previously reported the genomic proﬁling of adult in the context of DNA binding. Upon DNA binding, Arg96/ and pediatric B-cell ALL based on a cohort of Chinese Lys97 or Arg95/Lys97 are utilized to make direct contract patients . Using RNA-seq technology, we observed that with the phosphate groups of DNA backbone in the minor the abnormal expression of DUX4/IGHs often led to the groove (Supplementary Figure 2a). In compatible with the overexpression of a subset of DUX4 target genes including negatively charged nature of DNA molecules, the highly PCDH17, STAP1, AGAP1, ERG , etc. (Fig. 1a). The positively charged N-terminal RRKR loop could serve as alt expressions of DUX4 fusions were often associated with the important recruitment platform to initiate DNA binding via genetic events of NRAS mutations (Supplementary Table 2). electrostatic interaction (see structure-based characterization The breakpoints in DUX4 fusions were all located in the C- below). 1468 X. Dong et al. b c Vec S1 S2 S3 S4 S5 JP3 DUX4 HD1 HD2 1 CTD 19 78 94 153 424 S1 IGH HA-DUX4/IGH 255 289 S2 IGH 234 242 S3 IGH ERG 330 336 IGH S4 312 321 ERG alt S5 IGH 1 412 453 IGH JP3 ACTIN 399 431 Fig. 1 DUX4/IGH-driven transactivation in Chinese B-ALL patients. with bracket symbol on the left. b Domain arrangement of WT DUX4 a Heatmap of abnormal gene expression in DUX4-ALL patients (blue) and its oncogenic fusions obtained from Chinese patients S1 to S5. and other B-ALL subgroups including MEF2D fusion (black), TCF3- DUX4/IGH was previously demonstrated to induce leukemia in JP3 PBX1 (red), ETV6-RUNX1 (green), ZNF384 (cyan) fusion, hyperdi- mice . c Transactivation activity of DUX4/IGHs. Western blot ploidy (purple), MLL fusion (brown), and BCR-ABL1 (gray). The analysis of ERG in Reh cells overexpressing HA-tagged DUX4/IGH alt DUX4-driven transactivation of genes in patient samples is highlighted fusions. The bracket is used to highlight the expression of DUX4/IGHs In the major groove, the α3 helix forms direct hydro- deep into the major groove reading the “barcode” of gen bonds with DNA (Fig. 2f). The residues Arg137, T A A T . Gln143 is located 4.6 Å away from the DRE 1 2 3 4 Trp141, Arg145, and His149 are interacting with the nucleotides. Structural superimposition of Apo- and DNA backbone (Supplementary Figure 2b). Compared DNA-bound DUX4 reveals interesting sidechain HD2 with the Apo-DUX4 structure, the guanidinium head reshufﬂes of Asn144 and Arg148 upon the recognition of HD2 group of Arg137 in the DNA-bound complex shifts 2.9 Å DRE. The carboxamide head group of Asn144 rotates 30° upward to form two direct hydrogen bonds with the anti-clockwise to make two direct hydrogen bonds with phosphate groups in the DNA backbone. In parallel, the A nucleotide. In a concerted fashion, the guanidi- His149, located in the opposite end of the α3 helix, nium head group of Arg148 swings 5.3 Å into the major moves 1.7 Å upward, ﬂanking Arg145 via a side-chain π– groove to form hydrogen bonds with the A site. The π interaction. As a result, the side chains of Trp141, minor- and major-groove binding residues including Arg145, and His149 all adopt extended conﬁgurations, RRRR –RRKR ,QNR –QNR motifs are HD1 HD2 HD1 HD2 with the alkyl side-chain in parallel to each other and the highly conserved in DUX protein family (Supplementary charged head groups pointing towards DNA. This has Figure 2c), among which HD1 shares 54% sequence given rise to a pedestal-like base, on which the DNA identity with HD2. In particular, Asn144 is strictly con- backbone can rest. Inside the DNA major groove lie the served in homeobox superfamily (Supplementary Fig- residues Gln143, Asn144, and Arg148, termed QNR ure 2d), highlighting the signiﬁcance of DNA binding motif (Fig. 2f). The side-chains of these residues insert ability in DUX4/IGH-driven transactivation. Structural basis of DUX4/IGH-driven transactivation 1469 Fig. 2 Crystal structures of Apo- a b and DNA-bound DUX4 . a HD2 RRKR motif Apo-form of HD2 domain. The homeobox subdomain is shown in cartoon representation. The N/ R137 C-termini and α1/α2/α3 are labeled. b Electrostatic surface W141 of Apo-DUX4 . The highly HD2 3 R145 positively charged patch in Apo- R148 H149 HD2 is delineated by residues Arg137, Trp141, Arg145, Arg148, and His149. The ﬂexible N-terminal poly-Arg motif is shown with a dashed c d line. c Crystal structure of DUX4 -DNA complex. 2 HD2 DRE The HD2 molecules and TAATCTAAT sequences Major groove (DNA ) are shown in cartoon DRE 3 3 representation and colored in cyan, green, and orange, respectively. d The electrostatic surfaces of the N-terminal 2 loop and C-terminal α3 3 RRKR helix that engage with DNA Minor groove RRKR motif binding. e The interface between the N-terminal RRKR motif and DNA minor groove. The DRE poly-Arg residues are shown in RRKR motif stick representation. f The interface between the α3 helix and DNA major groove. DRE e f Gln143 (yellow), Asn144 (red), and Arg148 (red) are shown in K97 stick representation. The C R98 hydrogen bonds are highlighted RRKR motif R96 with dashed lines. The TAAT R148 nucleotides are labeled R95 N144 Q143 DUX4-DRE is important for DNA binding, As previously reported , the ERG biogenesis could alt transactivation, and B-cell differentiation be used to monitor the transactivation activities of DUX4/ IGH and mutants. The WT DUX4/IGH transactivated the To characterize the crystal structure of DUX4 -DNA , expression of ERG (Fig. 4a and Supplementary Figure 4). HD2 DRE alt the biolayer interferometry (BLI) was used. The recombi- Contrarily, the perturbation of DNA binding residues nant DUX4 and DUX4 were shown to bind to a abrogated the biogenesis of ERG . The poly-Ala mutations HD2 HD1–HD2 alt DRE sequence at K values of 7.2 and 1.4 μM, respectively of RRRR –RRKR and QNR –QNR motifs in D HD1 HD2 HD1 HD2 (Fig. 3a and Supplementary Figure 3a). By contrast, the either HD1 or HD2 completely abolished the transactivation mutations of R95A, R96A, K97A, and R98A in the poly- activity of DUX4/IGH, suggesting that HD1 and HD2 Arg motif signiﬁcantly abolished the DNA binding activity domains are both required for DRE recognition. of DUX4 (Fig. 3b). Consistently, in the α3 helix region, In order to investigate which DNA-binding motif HD2 Q143E, N144E, and R148E resulted in 4- to 9-fold weaker (RRRR –RRKR or QNR –QNR ) could be more HD1 HD2 HD1 HD2 binding. The measured K values of these mutants were 31, important, a series of single mutations were designed. The 55, and 67 μM, respectively (Fig. 3c). Supportively, similar DUX4/IGH-driven transactivation activities (i.e., the structure-based mutations in DUX4 completely abnormal biogenesis of ERG ) of wild type and mutants HD1–HD2 alt abrogated DNA binding activity (Supplementary Figure 3b). were monitored in Reh cells. The single mutation in QNR 1470 X. Dong et al. Fig. 3 The BLI characterization of DNA binding by DUX4 and μM recombinant DUX4 was used. Under these experiment condi- DRE HD2 HD2 mutants. a The interaction between DUX4 and DNA . a WT tions, the DNA was immobilized on the probe. Left, the association HD2 DRE DRE DUX4 . b DUX4 mutations in the minor groove binding resi- and dissociation curves. Right, the binding curves and K values HD2 HD2 D dues, i.e., poly-Arg motif. c DUX4 mutations in the major groove derived from Scatchard plots  HD2 binding residues, i.e., QNR motif. A gradient concentration of 0.5–32 motifs (i.e., N69A, R73A in HD1 and N144A, R148A in contrast, the double Q68/143A mutation dramatically HD2) was sufﬁcient to inhibit the DNA recognition and reduced ERG biogenesis, suggestive of rescue effect DRE alt abolish subsequent ERG transactivation (Fig. 4a). Inter- between HD1 and HD2 (Fig. 4a). Consistently, the com- alt estingly, the single mutations, Q68A and Q143A, failed to plementary binding mode was also observed in the single inhibit DUX4/IGH-driven transactivation. In marked mutations of RRRR –RRKR . Although single HD1 HD2 Structural basis of DUX4/IGH-driven transactivation 1471 R RKR Q N R 95 98 143 144 148 ERG ERG alt HA ACTIN R RKR Q N R 95 98 143 144 148 R RRR Q N R 68 69 73 20 23 ERG ERG alt HA ACTIN R RRR Q N R 20 23 68 69 73 R RRR Q N R 20 23 68 69 73 R RKR Q N R 95 98 143 144 148 ERG ERG alt HA ACTIN R RRR Q N R 20 23 68 69 73 R RKR Q N R 95 98 143 144 148 b Vector DUX4-IGH R20A&R95A 85.6% 12.1% 66.1% R21A&R96A R23A&R98A R22A&K97A ** *** 81.9% 44.1% 80.6% 20 *** Q68A&Q143A R73A&R148A N69A&N144A 27.4% 84.3% 90.6% CD19 mutation in RRKR could preclude the formation of disruption of DUX4-driven transactivation (Fig. 4a), HD2 DUX4 -DNA (Fig. 3b), most single mutations in implying the DNA binding between DUX4/IGH and DNA HD2 DRE RRKR were less destructive, resulting in partial might be rescued by the equivalent poly-Arg motif in HD1. HD2 Mac-1 Expression of ERG Expression of ERG Expression of ERG alt alt alt CD19 (%) 1472 X. Dong et al. Fig. 4 DRE recognition is important for DUX4/IGH-driven transac- DUX4 -DNA , but also suggest DUX4 -DRE HD2 DRE HD1–HD2 tivation and B-cell differentiation. a Transactivation activity of DUX4/ interacting motif as a potential therapeutic target for future IGH and mutants in Reh cells. Left column, the biogenesis of ERG in alt treatment against B-ALL harboring DUX4/IGH fusion. Reh cells that contain WT and mutant DUX4/IGHs was detected by Supportively, the structure-based mutations also inhibit the western blot. Right column, the real-time PCR technique was used to conﬁrm the DUX4-driven transactivation at mRNA level. Top row, apoptotic function of WT DUX4 (Supplementary Fig- mutations in HD2. Middle row, mutations in HD1. Bottom row, ure 5a), and its transactivation ability on target gene as double mutations in HD1 and HD2. All experiments have been done at monitored by luciferase assay (Supplementary Figure 5b). least with three independent replicates. b In vitro B cell differentiation Altogether, these have led to hypothesis that WT DUX4 and assay of DUX4/IGH and mutants. FACS technique was used to ana- lyze the effect of DUX4/IGH and mutants on the differentiating ability DUX4/IGH might share a similar DNA binding mechanism − Low of mouse progenitor cells (Lin /c-Kit ) into CD19 positive B cells. (see below). All experiments have been done at least with three independent However, until now, it is not clear how the C-terminal replicates. Values are means ± SE domain (CTD) of DUX4, whose molecular boundary is not yet clearly deﬁned, might contribute to the transactivation activities of WT DUX4 and DUX4/IGHs. WT DUX4, Indeed, when an extra Arg/Lys to Ala mutation was engi- overexpression of which often triggers apoptosis, is the neered in the equivalent position (such as R20/95A, R21/ main reason of FSHD disease . In marked contrast, the 96A, R22/K97A, R23/98A), the accumulative impacts were loss/break-down of CTD has enabled DUX4/IGH to cause very clear in spite of the fact that the QNR –QNR B-cell arrest and proliferation, and hence serves as an HD1 HD2 motif remained intact in these mutants (Fig. 4a). Suppor- oncogenic driver in leukemia [3, 4]. In order to understand tively, similar results can be obtained using NALM6 cells how CTD might inﬂuence the transactivation activity of (Supplementary Figure 4). All these results support a con- DUX/IGH, a series of CTD truncated mutants were certed binding clamp-like binding between HD1 and HD2. designed (Fig. 5a). Interestingly, when the CTD was Furthermore, as shown in Fig. 4a, the single mutations in restricted to residue 200, the DUX4/IGH activity was sig- HD1 appear to be more destructive than those of HD2, niﬁcantly impaired (Fig. 5b, c). Consistently, this correlates prompting the hypothesis that two HDs might have different with the observation in B-ALL patients. The shortest ver- roles in DNA binding, in which HD1 might be more afﬁ- sion of DUX4/IGH derived from Patient S2 also displayed nitive subunit to engage DRE site. less ERG expression (Fig. 1c), suggestive of CTD con- alt Next, we want to know whether the DRE recognition via tribution in DUX4/IGH and B-ALL. However, the size of HD domains could play a key role in disease mechanism. CTD is not the only structural factor in DUX4/IGH’s As demonstrated elsewhere , the leukemogenic activity transactivation activity. DUX4 /IGH from Patient S5 1-412 of DUX4/IGH is constantly associated with the arrest of B- displayed less ERG expression than those shorter fusion alt cell differentiation. We therefore addressed the phenotypic derivatives from Patients S3 and S4 (Fig. 1b, c). Similar effect of distinct fusion proteins with WT or mutated DRE- results could be observed in CTD truncated mutants binding motifs. In control experiment, ~85% primary (Fig. 5b, c), suggesting that the overall structure/fold of murine progenitors transfected with vehicle (empty vector) CTD, which remains elusive, might play critical role in could undergo normal B-cell differentiation as assessed by DUX4/IGH-driven transactivation. ﬂow cytometry using antibodies against mouse CD19 (Fig. 4b). DUX4/IGH-transfected cells, by contrast, showed Abnormal transactivation controlled by much less B-cell differentiation (~12%), in concordance DUX4 -DNA HD1–HD2 DRE with its transforming ability and leukemogenic effects . The perturbations in the RRRR –RRKR and The RNA-seq data mining of our 172 patients showed that HD1 HD2 QNR –QNR , which hampered the DNA binding and the expression proﬁle of DUX4/IGH positive ALL was HD1 HD2 transactivation, also abrogated the inhibitory activity of signiﬁcantly different from other B-ALL subtypes (Fig. 1a). DUX4/IGH on B-cell differentiation. As shown in Fig. 4b, Notably, a group of genes such as AGAP1, CHST2, most mutations rescued the lymphoid lineage differentiation CLEC12A, PCDH17, PTPRM, STAP1, etc., which all at levels similar to that of the cells transfected with vehicle contain DREs, were found upregulated in patients harboring (i.e., 66 – 90% differentiation for R20/95A, R21/96A, R23/ DUX4 fusions. This is further supported by the observation 98A, N69/144A, R73/148A). Of note, cells transfected with that knock-down (KD) of DUX4/IGH through shRNA in R22/K97A and Q68/143A still bore certain levels of ERG NALM6 cells , which contains an endogenous DUX4/ alt expression (Fig. 4a) and showed only partially restored IGH fusion, downregulated these target genes (Supple- differentiation compartments in FACS analysis (44 and mentary Figure 6a). We next investigated the function of 27%, respectively) (Fig. 4b). These results taken together these DUX4/IGH-targeted genes by FACS proﬁling and not only support our observation in the crystal structure of apoptotic assay. Firstly, FACS proﬁling showed that the Structural basis of DUX4/IGH-driven transactivation 1473 ab c *** ERG DUX4 HD1 HD2 IGH 80 1 431 399 ERG alt 1 350 1 300 HA 1 250 ACTIN 1 200 Fig. 5 The effect of C-terminal domain in DUX4/IGH-driven transactivation. a Scheme of CTD truncated mutants. The biogenesis of ERG in alt Reh cells by WT DUX4/IGH and C-terminal truncated mutants were monitored by western blot (b) and RT-PCR (c) over-expression of AGAP1, but none of other target genes, In addition, the tandem arrangement of double homeobox is led to a partial arrest of B-cell differentiation in murine in striking similarity with the domain architecture observed progenitor cells (Supplementary Figure 6b). Secondly, the in PRD proteins. The PAI sub-domain is a homeobox, and genetic KD of AGAP1 in NALM6 cells exerted a signiﬁcant the RED sub-domain can fold into a HD-like structure with impact on apoptosis, whereas KD of other DUX4/IGH three α-helices [20, 22]. Hence PAI-RED appears to be a target genes (except for CHST2) did not show obvious double-homeobox-like structure. However, it has been apoptotic effects (Supplementary Figure 6c). More impor- shown that the RED sub-domain is not really a HD domain tantly, as monitored by luciferase assays, the structure- in term of DNA binding and recognition. As demonstrated based mutations signiﬁcantly abolished the DUX4/IGH- in the structural and functional study of PAX -DNA PAIRED driven transactivation of AGAP1 (Supplementary Fig- complex, the RED sub-domain is not essential in DNA ure 6d). Supportively, abnormal expression of AGAP1 is binding . In line with this observation, many studies frequently observed in B-ALL patients with DUX4 fusions have shown that a single homeobox domain is not sufﬁcient reported elsewhere [4, 18]. These results highlight a critical to mediate DNA binding/recognition for productive trans- DUX4 -DNA recognition in B-ALL pathogenesis activation. Thus far, it has been reported that homeobox HD1–HD2 DRE (Fig. 6). However, although AGAP1 displays potential in B- proteins often require: (1) the protein–protein interaction cell differentiation, further investigation such as in vivo involving a second transcription factor; (2) inter-molecular characterization involving the use of knock-out mice model dimerization; and (3) an extra intra-molecule DNA binding is required to establish the leukemogenic role of DUX4/IGH domain to bind DNA [24–27]. target gene. In DUX4, the presence of double homeobox appears to be sufﬁcient for DNA recognition and transactivation. The gain of an extra HD domain in DUX4, possibly derived Discussion from gene reshufﬂe and infusion , gives rise to a powerful transcription factor for TAATCTAAT (DRE) site Protein diversity in DUX4 (Supplementary Figure 1). In B-ALL patients, nearly all the DUX4/IGHs contain a structural and functional intact DUX4, a member of homeobox protein family, is exclu- double homeobox, leading to the hypothesis that the DUX4 sively found in placental mammals [19, 20]. Compared to double homeobox might play the central role in DNA other homeobox proteins, DUX4 contains two adjacent binding and transactivation. Indeed, when the Reh cells HD1 and HD2 domain, connected by a short linker of 15 were transfected with DUX4 fusion with different IGH tails, residues. Thus far, no structural information is available for ERG biogenesis were consistently observed (Fig. 1b, c). alt the double homeobox. In this section, we will focus our When the endogenous expression of DUX4/IGH was sup- discussion on the protein diversity in DUX4 double pressed by shRNA, the target genes containing DRE sites homeobox. were consistently downregulated (Supplementary Fig- It has been proposed that DUX4 belongs to the PAIRED ure 6a). More importantly, when the DNA binding interface (PRD) homeobox subclass because their HD domains share in either DUX4 or DUX4 was mutated, the binding HD1 HD2 the most sequence identity (Supplementary Figure 2d) . afﬁnity and transactivation function were severely impaired Expression of ERG alt 1474 X. Dong et al. Loop RRKR HD2 DNA DRE recognition HD1 HD2 Loop RRRR HD1 DUX4-IGH driven transactivation CTD-IGH ERG alt Other target genes? Normal cells Leukemia cells Fig. 6 Oncogenic network controlled by DUX4 -DNA Based on 54% sequence identity, the HD1 model in this ﬁgure is built HD1–HD2 DRE recognition in B-ALL pathogenesis. A two-step cooperative binding/ by homology modeling using DUX4 coordinates. The covalent HD2 transactivation mechanism via HD1 and HD2 domains. In the ﬁrst step linkage between HD1 and HD2 is shown in dashed line. CTD is (top-left panel), poly-Arg motifs lead the initial DNA binding via a indicated with octagon. The symbols “?” are used to highlight/prompt non-speciﬁc eletrostatic attraction. This in turn primes the second step further investigations to understand (1) what role CTD might play in (top-right panel) for the intimate/speciﬁc clamp-like association DUX4- and DUX4/IGH-driven transactivations; (2) whether DUX4/ between DUX4-DNA via QNR –QNR motifs. The two-step IGH target genes such as AGAP1 could accelerate/or is required in DRE HD1 HD2 DUX4/IGH is a critical step in B-ALL leukemogenesis (low panel). DUX4/IGH-mediated leukemogenesis (Figs. 3 and 4). Altogether, these results suggest that the The two-step binding mode is further supported by evolutionary tinkering of double homeobox design is an structure-based mutagenesis. The single mutation of important mean to endow DUX4 with transactivation QNR –QNR (i.e., N69A, N144A, R73A, R148A) HD1 HD2 activity. could ~100% inhibit transactivation, suggesting both HD1 and HD2 are essential. In marked contrast, single mutations Two-step clamp-like binding mechanism of RRRR –RRKR were less dramatic, implying that HD1 HD2 (1) poly-Arg motif is less critical when compared to the In this report, the crystal structure of DUX4 -DRE QNR motif; (2) the existence of cooperative binding HD2 reveals a novel two-step clamp-like DNA binding between HD1 and HD2. This is further supported by the mechanism for double homeobox. In the asymmetric unit observation that double mutations of RRRR –RRKR HD1 HD2 (ASU), it contains two HD2 molecules and a DNA duplex (i.e., R20/95A, R21/96A, R22/K97A, R23/98A) resulted a of TAATCTAAT (DRE) sequence. Based on homology complete loss of transactivation even in the presence of modeling, a HD1 molecule can be mapped onto DRE site. QNR –QNR motifs (Fig. 4). This is also the case in HD1 HD2 Supportively, the C-termini of HD1 and the N-termini of BLI characterization, the poly-Glu mutation of HD2 are both correctly positioned, giving ample space to RRRR–RRKR resulted in no binding between double envisage the missing linker in between. More importantly, homeobox and DNA (Supplementary Figure 3b). These the QNR motif is predicted to insert into the DNA major data appear to suggest that the DNA binding of poly-Arg HD1 groove engaging the direct hydrogen bonds with TAAT. In motifs might precede the interaction of TAAT-QNR. Based parallel, the RRRR motif of HD1 wraps around the DNA on these observations, we proposed a two-step clamp-like minor groove (Fig. 6). All these data suggest that: (1) HD1 DRE recognition mechanism by DUX4 double homeobox can bind TAAT sequence in the same way as HD2 and (2) (Fig. 6). In the ﬁrst step, the highly positively charged the presence of an extra HD domain might enhance its RRRR –RRKR is important to direct DUX4 towards HD1 HD2 overall binding. Indeed, the BLI characterization showed its DNA substrate (via electrostatic interaction). In this step, that the double homeobox binds to a DRE sequence at a K the HD1 subdomain might take the leading role to make the value of 1.4 μM, >5 fold stronger than HD2 alone (Fig. 3a ﬁrst kiss with target DNA. The non-speciﬁc association and Supplementary Figure 3a). between RRRR –RRKR with DNA in turn enables HD1 HD2 Structural basis of DUX4/IGH-driven transactivation 1475 the double homeobox to recognize the DRE site via personnel of beamlines BL17U/18U1/19U1 (SSRF/NFPS, Shanghai, China) for help during data collection. QNR –QNR . The double kiss between TAATs and HD1 HD2 QNR motifs might be critical to “clamp” DUX4-DRE into Author contributions Conceived and designed the experiments: ZC, an intimate association that enables productive transactiva- S-JC, GM. Performed the experiments: XD, WZ, HW, JH, MZ, PW, HZ. Analyzed the data: XD, WZ, HW, JH, MZ, PW, HZ, ZC, S-JC, tion of its target genes (Fig. 6). Similar DNA binding mode GM. Preparation of the ﬁgures: XD, WZ, HW, JH, ZC, S-JC, GM. should also present in WT DUX4 (Supplementary Fig- Wrote the paper: GM. ure 5). In addition, the CTD domain, which clear con- tributes to DUX4/IGH-driven transactivation (Fig. 5), Compliance with ethical standards should remain an interesting subject for future investigation in the context of WT DUX4 and its oncogenic derivatives. Conﬂict of interest The authors declare that they have no conﬂict of interest. Oncogenic potential of DUX4-DRE Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, It has been reported that homeobox-containing protein adaptation, distribution and reproduction in any medium or format, as including PRD-type subclass is involved in hematopoiesis long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if and leukemogenesis . The abnormal expression of changes were made. The images or other third party material in this PAX5, a member of the PRD-type subclass, derived from article are included in the article’s Creative Commons license, unless chromosome translocation with the IGH locus can dereg- indicated otherwise in a credit line to the material. If material is not ulate the lymphoid cell gene expression program and con- included in the article’s Creative Commons license and your intended tribute to B and T-lineage neoplasms [29–31]. Recent use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright studies showed that, the homeobox-containing DUX4 can holder. To view a copy of this license, visit http://creativecommons. also translocate into the IGH locus, and the aberrant org/licenses/by/4.0/. expression of DUX4 can cause B-cell leukemia in immu- nocompromised mice . A follow-up study by Zhang and References co-workers showed that the aberrant expression of DUX4/ IGHs were constantly associated with the production of a 1. de The H, Chen Z. Acute promyelocytic leukaemia: novel insights novel ERG isoform that can also contribute to leukemo- into the mechanisms of cure. Nat Rev Cancer. 2010;10:775–83. alt 2. Hehlmann R, Hochhaus A, Baccarani M, European LeukemiaNet. genesis in vivo . These observations have led to the Chronic myeloid leukaemia. Lancet. 2007;370:342–50. proposal of DUX4/IGH-ERG axis in B-ALL develop- alt 3. Yasuda T, et al. Recurrent DUX4 fusions in B cell acute lym- ment. However, not all patients in DUX4/IGH subgroup phoblastic leukemia of adolescents and young adults. Nat Genet. were ERG positive (Supplementary Table 2), suggestive 2016;48:569–74. alt 4. Zhang J, et al. Deregulation of DUX4 and ERG in acute lym- of other genetic lesion in B-ALL. The RNA-seq data phoblastic leukemia. Nat Genet. 2016;48:1481–1489. mining of our 172 patients showed that the expression 5. Hunger SP, Mullighan CG. Acute lymphoblastic leukemia in proﬁle of DUX4/IGH positive ALL was signiﬁcantly dif- children. N Engl J Med. 2015;373:1541–52. ferent from other B-ALL subtypes (Fig. 1 and Supple- 6. Liu YF, et al. Genomic proﬁling of adult and pediatric B-cell acute lymphoblastic leukemia. EBioMedicine. 2016;8:173–83. mentary Figure 1). Using FACS and apoptotic assays, we 7. Lilljebjorn H, et al. Identiﬁcation of ETV6-RUNX1-like and have found that AGAP1, like DUX4/IGH and ERG , could alt DUX4-rearranged subtypes in paediatric B-cell precursor acute contribute to B cells proliferation and differentiation. lymphoblastic leukaemia. Nat Commun. 2016;7:11790. However, it should be noted, cytoplasm-bound AGAP1 is 8. Gatica LV, Rosa AL. A complex interplay of genetic and epige- netic events leads to abnormal expression of the DUX4 gene in not a transcription factor like ERG or ERG , and its alt facioscapulohumeral muscular dystrophy. Neuromuscul Disord. oncogenic pathway to B-cell arrest is not yet clear. Further 2016;26:844–52. investigations such as in vivo study in the background of 9. Lemmers RJ, et al. Digenic inheritance of an SMCHD1 mutation AGAP1 knock-out mice should be carried out to char- and an FSHD-permissive D4Z4 allele causes facioscapulohumeral muscular dystrophy type 2. Nat Genet. 2012;44:1370–4. acterize its leukemogenic cross-talk with DUX4/IGH and 10. Geng LN, et al. DUX4 activates germline genes, retroelements, ERG . alt and immune mediators: implications for facioscapulohumeral dystrophy. Dev Cell. 2012;22:38–51. Acknowledgements This work was supported by research grants 11. Loughran SJ, et al. The transcription factor Erg is essential for 81770142, 81370620, 81570120, 31070645 from National Natural deﬁnitive hematopoiesis and the function of adult hematopoietic Scientiﬁc Foundation of China (to GM), a research grant 20152504 stem cells. Nat Immunol. 2008;9:810–9. from “Shanghai Municipal Education Commission—Gaofeng Clinical 12. Tursky ML, et al. Overexpression of ERG in cord blood pro- Medicine Grant Support” (to GM), “The Program for Professor of genitors promotes expansion and recapitulates molecular sig- Special Appointment (Eastern Scholar) at Shanghai Institute of Higher natures of high ERG leukemias. Leukemia. 2015;29:819–27. Learning” (to GM), a research grant 11JC1407200 from SMSTC (to 13. Tsuzuki S, Taguchi O, Seto M. Promotion and maintenance of GM), a research grant 12ZZ109 from SME (to GM), We thank the leukemia by ERG. Blood. 2011;117:3858–68. 1476 X. Dong et al. 14. Thoms JA, et al. ERG promotes T-acute lymphoblastic leukemia 24. Verrijzer CP, et al. The DNA binding speciﬁcity of the bipartite and is transcriptionally regulated in leukemic cells by a stem cell POU domain and its subdomains. EMBO J. 1992;11:4993–5003. enhancer. Blood. 2011;117:7079–89. 25. Grueneberg DA, Natesan S, Alexandre C, Gilman MZ. Human 15. Krom YD, et al. Intrinsic epigenetic regulation of the D4Z4 and Drosophila homeodomain proteins that enhance the DNA- macrosatellite repeat in a transgenic mouse model for FSHD. binding activity of serum response factor. Science. PLoS Genet. 2013;9:e1003415. 1992;257:1089–95. 16. De Iaco A, et al. DUX-family transcription factors regulate 26. Wilson DS, Guenther B, Desplan C, Kuriyan J. High resolution zygotic genome activation in placental mammals. Nat Genet. crystal structure of a paired (Pax) class cooperative homeodomain 2017;49:941–5. dimer on DNA. Cell. 1995;82:709–19. 17. Hendrickson PG, et al. Conserved roles of mouse DUX and 27. Wilson D, Sheng G, Lecuit T, Dostatni N, Desplan C. Cooperative human DUX4 in activating cleavage-stage genes and MERVL/ dimerization of paired class homeo domains on DNA. Genes Dev. HERVL retrotransposons. Nat Genet. 2017;49:925–34. 1993;7:2120–34. 18. Harvey RC, et al. Identiﬁcation of novel cluster groups in pediatric 28. Mullighan CG, et al. Genome-wide analysis of genetic alterations high-risk B-precursor acute lymphoblastic leukemia with gene in acute lymphoblastic leukaemia. Nature. 2007;446:758–64. expression proﬁling: correlation with genome-wide DNA copy 29. Iida S, et al. The t(9;14)(p13; q32) chromosomal translocation number alterations, clinical characteristics, and outcome. Blood. associated with lymphoplasmacytoid lymphoma involves the 2010;116:4874–84. PAX-5 gene. Blood. 1996;88:4110–7. 19. Leidenroth A, Hewitt JE. A family history of DUX4: phylogenetic 30. Busslinger M, Klix N, Pfeffer P, Graninger PG, Kozmik Z. analysis of DUXA, B, C and Duxbl reveals the ancestral DUX Deregulation of PAX-5 by translocation of the Emu enhancer of gene. BMC Evol Biol. 2010;10:364. the IgH locus adjacent to two alternative PAX-5 promoters in a 20. Burglin TR, Affolter M. Homeodomain proteins: an update. diffuse large-cell lymphoma. Proc Natl Acad Sci USA. Chromosoma. 2016;125:497–521. 1996;93:6129–34. 21. Holland PW. Evolution of homeobox genes. Wiley Interdiscip 31. Souabni A, Jochum W, Busslinger M. Oncogenic role of Pax5 in Rev Dev Biol. 2013;2:31–45. the T-lymphoid lineage upon ectopic expression from the immu- 22. Gehring WJ, Affolter M, Burglin T. Homeodomain proteins. noglobulin heavy-chain locus. Blood. 2007;109:281–9. Annu Rev Biochem. 1994;63:487–526. 32. Abdiche Y, Malashock D, Pinkerton A, Pons J. Determining 23. Xu W, Rould MA, Jun S, Desplan C, Pabo CO. Crystal structure of kinetics and afﬁnities of protein interactions using a parallel real- a paired domain-DNA complex at 2.5 A resolution reveals struc- time label-free biosensor, the Octet. Anal Biochem. tural basis for Pax developmental mutations. Cell. 1995;80:639–50. 2008;377:209–17.
– Springer Journals
Published: Mar 15, 2018