• Loading metrics

Loss of atrx cooperates with p53-deficiency to promote the development of sarcomas and other malignancies

  • Felix Oppel ,

    Contributed equally to this work with: Felix Oppel, Ting Tao

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Validation, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States of America

  • Ting Tao ,

    Contributed equally to this work with: Felix Oppel, Ting Tao

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Validation, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States of America

  • Hui Shi,

    Roles Data curation, Formal analysis, Visualization, Writing – review & editing

    Affiliation Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States of America

  • Kenneth N. Ross,

    Roles Data curation, Formal analysis, Software, Validation, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States of America

  • Mark W. Zimmerman,

    Roles Conceptualization, Formal analysis, Investigation, Methodology, Writing – review & editing

    Affiliation Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States of America

  • Shuning He,

    Roles Formal analysis, Methodology, Project administration, Resources, Validation, Writing – review & editing

    Affiliation Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States of America

  • Guangxiang Tong,

    Roles Data curation, Formal analysis, Validation, Visualization, Writing – review & editing

    Affiliation Heilongjiang River Fisheries Research Institute of Chinese Academy of Fishery Sciences, Harbin, China

  • Jon C. Aster,

    Roles Data curation, Formal analysis, Investigation, Validation, Writing – review & editing

    Affiliation Department of Pathology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States of America

  • A. Thomas Look

    Roles Conceptualization, Formal analysis, Funding acquisition, Investigation, Project administration, Resources, Supervision, Writing – review & editing

    Affiliation Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States of America

Loss of atrx cooperates with p53-deficiency to promote the development of sarcomas and other malignancies

  • Felix Oppel, 
  • Ting Tao, 
  • Hui Shi, 
  • Kenneth N. Ross, 
  • Mark W. Zimmerman, 
  • Shuning He, 
  • Guangxiang Tong, 
  • Jon C. Aster, 
  • A. Thomas Look


The SWI/SNF-family chromatin remodeling protein ATRX is a tumor suppressor in sarcomas, gliomas and other malignancies. Its loss of function facilitates the alternative lengthening of telomeres (ALT) pathway in tumor cells, while it also affects Polycomb repressive complex 2 (PRC2) silencing of its target genes. To further define the role of inactivating ATRX mutations in carcinogenesis, we knocked out atrx in our previously reported p53/nf1-deficient zebrafish line that develops malignant peripheral nerve sheath tumors and gliomas. Complete inactivation of atrx using CRISPR/Cas9 was lethal in developing fish and resulted in an alpha-thalassemia-like phenotype including reduced alpha-globin expression. In p53/nf1-deficient zebrafish neither peripheral nerve sheath tumors nor gliomas showed accelerated onset in atrx+/- fish, but these fish developed various tumors that were not observed in their atrx+/+ siblings, including epithelioid sarcoma, angiosarcoma, undifferentiated pleomorphic sarcoma and rare types of carcinoma. These cancer types are included in the AACR Genie database of human tumors associated with mutant ATRX, indicating that our zebrafish model reliably mimics a role for ATRX-loss in the early pathogenesis of these human cancer types. RNA-seq of p53/nf1- and p53/nf1/atrx-deficient tumors revealed that down-regulation of telomerase accompanied ALT-mediated lengthening of the telomeres in atrx-mutant samples. Moreover, inactivating mutations in atrx disturbed PRC2-target gene silencing, indicating a connection between ATRX loss and PRC2 dysfunction in cancer development.

Author summary

Somatic mutations in genes coding for epigenetic regulators such as ATRX are found across a diverse group of cancer types, suggesting their broad relevance in tumor induction and progression. However, tumors that have been linked to these chromatin remodelers can arise in many different molecular and cellular contexts, requiring studies with new experimental models to understand the extent and mechanisms of tumor development mediated by these regulatory proteins. Thus, we analyzed the tumor suppressive role of atrx in zebrafish that already harbored inactivating mutations of p53 and nf1. Homozygous deletion of atrx was lethal in developing fish, whereas the partial loss of this gene (atrx+/-) within the p53/nf1-deficient background led to a diverse spectrum of tumors not observed in animals that were wildtype for atrx, including epithelioid sarcoma, angiosarcoma, and rare carcinomas. Most of the cancer types we identified correspond to human tumors in the ATRX-mutant tumor sample cohort within the AACR Genie database, attesting to the relevance of our findings to human cancer. Further analysis revealed downregulation of telomerase during the lengthening of the telomeres through the ALT pathway, and disturbed function of the polycomb repressive complex 2 as key mechanistic components underlying atrx-linked tumorigenesis. These results demonstrate how a p53/nf1 compromised genetic background combined with ATRX haploinsufficiency leads to a broad spectrum of sarcomas and carcinomas associated with loss of this chromatin modulator.


The alpha thalassemia/mental retardation syndrome X-linked (ATRX) protein is involved in the epigenetic regulation of gene expression. It is classified as a SWI/SNF-family chromatin remodeling factor due to its ATP-dependent helicase domain. In humans, germline loss of ATRX function causes mental retardation and alpha thalassemia that is associated with reduced alpha globin expression levels, lower blood-oxygen levels and hypochromia, anisocytosis, and poikilocytosis of red blood cells [15]. Because the ATRX gene is located on the X-chromosome in humans, females can carry a mutant allele heterozygously, without developing symptoms. Loss of ATRX leads to reduced levels of histone 3.3 (H3.3) incorporation, telomere destabilization and increased homologous recombination facilitating the development of ALT. ATRX binds to the death domain-associated protein 6 (DAXX) and recognizes H3K9me3 marks with its cysteine-rich domain termed ADD (ATRX-DNMT3-DNMT3L) [6]. The ATRX/DAXX-mediated deposition of H3.3 maintains the condensed heterochromatic state [7,8].

ATRX also guides the Polycomb repressive complex 2 (PRC2) to its targets for gene silencing by tri-methylation of histone 3 lysine 27 (H3K27me3) [9], a repressive epigenetic mark established by PRC2 [10]. This process is crucial for X-chromosome inactivation, mediated by the noncoding RNA XIST [11], which is expressed only from the to-be-silenced X-chromosome (Xi) and spreads along the Xi in cis [12]. ATRX binding to XIST is essential for PRC2 recruitment following XIST in cis along the Xi and also functions as an adaptor that affects PRC2 function beyond Xi inactivation. Upon ATRX knockdown in a human fibroblast cell line, PRC2 is unable to silence its target genes by deposition of the H3K27me3 mark at specific sites within the gene body. Instead, H3K27me3 is established at ectopic sites in the intergenic space and at non-canonical sites in the target genes, demonstrating the importance of ATRX for normal gene silencing by PRC2 [9].

Over the past 5 years, it has become apparent that mutations in epigenetic regulator genes are involved in the onset and progression of a large number of malignancies. The loss of ATRX in gliomas [13], neuroendocrine tumors [14] and various sarcoma types [1519] facilitates alternative lengthening of the telomeres (ALT) and thereby stabilizes the genome of cancer cells during cancer development, a crucial step in the immortalization of cancer cells in general and a requirement for the formation of malignant tumors in humans [20]. There are two mutually exclusive mechanisms to elongate telomeres in tumor cells, i) telomerase (TERT) re-expression and ii) ALT activation. Individual tumor types differ in the frequency with which these mechanisms are activated [21]. For example, gliomas more frequently re-express TERT, but about 80% of ALT-positive pediatric high-grade gliomas are ATRX-deficient [13]. Comparable results demonstrating a strong association between ATRX deficiency and ALT were obtained in pancreatic neuroendocrine tumors (PanNETs) [22] and sarcomas [1518]. ALT-positive cancer cell lines were recently found to preferentially have ATRX inactivation [23,24].

The mechanisms underlying ALT activation upon ATRX loss are not fully understood. It has been proposed that the heterochromatic state of the telomeres is disrupted when H3.3 can no longer be loaded by ATRX/DAXX, so that chromatin opens up and telomeres become more accessible [25,26]. Others have shown that ALT depends on telomere elongation by homologous recombination-mediated DNA replication (HR) [27]. It was has also been shown in a genetic mouse model of glioma that ATRX deficiency results in reduced activity of the non-homologous end joining pathway, which competes with HR in DNA repair [28]. This suggests that there may be increased HR-activity in ATRX-depleted cells, which further facilitates ALT.

Loss-of-function (lof) mutations in ATRX clearly promote the activity of ALT, which is an important step in establishing cellular immortality. However, the changes in tumor biology (other than activation of ALT) caused by ATRX loss are poorly understood. As ATRX loss is found in a variety of cancer types, we aimed to assess ATRX loss in combination with dysregulation of two of the most prevalent types of oncogenic pathways—the Ras pathway and the p53 pathway—to include a large spectrum of human cancers. Thus, we knocked out ATRX in the germline of the previously published p53- and nf1-deficient zebrafish model, in which mutants develop malignant peripheral nerve sheath tumors (MPNSTs) and high-grade gliomas [29]. Here we demonstrate that homozygous in activation of atrx leads to lethal alpha-thalassemia in zebrafish, while heterozygous loss of atrx in the context of combined p53/nf1-deficiency induces the onset of multiple histologic tumor types that are otherwise not observed in nf1/p53–deficient fish.

Results and discussion

CRISPR/Cas9-mediated knockout of atrx

To create loss-of-function (lof) mutations in atrx in zebrafish germline, we induced frameshift-mutations in exon 4 of atrx using CRISPR/Cas9 (S1 Fig and S1 Table). This resulted in a truncated Atrx protein lacking both ADD and ATPase/helicase domains (Fig 1A) which represents a total loss of Atrx function. Mutant alleles were generated in both wildtype (strain AB) zebrafish and in the previously published nf1/p53-deficient zebrafish line that expresses the green fluorescent protein (GFP) marker under the control of the zebrafish sox10 promoter (sox10:GFP) [29].

Fig 1. CRISPR/Cas9-mediated knockout of atrx in zebrafish germline.

(A) The wildtype (WT) Atrx protein consists of 2013 amino acids (aa). Targeting of the atrx coding sequence with CRISPR/Cas9 resulted in truncation of Atrx before its functional domains and confers a loss of function; numbers indicate the underlying genetic alterations, e.g., +4 = insertion of 4 bases. (B) atrx-/- embryos show developmental abnormalities and die within 18 days post fertilization (dpf); sibling includes both atrx+/- and atrx+/+ embryos.

After injection of gRNAs and Cas9 mRNA into one-cell-embryos, fish were raised to fertility and out-crossed with non-injected siblings. All 10 analyzed injected F0 fish of the p53/nf1-deficient line transmitted mutations into the F1 generation. When 25 F1-embryos derived from these F0 fish were examined for their genomic alterations at the target locus, we observed a frameshift-mutation rate of 40%, with 30% harboring deletions and 70% harboring insertions (S1 Table). Moreover, 48% of zebrafish exhibited the in-frame deletion of a specific triplet, two had other in-frame mutations (8%) and the single remaining fish (4%) had a 3 bp substitution. The most recurrent in-frame mutation deleted codon 98 residing 4–6 bases upstream of the PAM-sequence. It is known that certain gRNAs recurrently induce specific genomic alterations in knockout-approaches using CRISPR/Cas9. Thus, genome editing was highly efficient and the resulting genomic alterations are consistent with previous studies [30,31]. We chose two zebrafish lines with frameshift mutations in the p53/nf1-deficient background and one line in the wildtype background to conduct all experiments described in this manuscript (S1 Fig and S1 Table). All our here presented knockout lines carry frameshift mutations in atrx in germline and thus model a total loss of ATRX, as it is frequently observed in human tumors [1319,32]. As ATRX is a tumor suppressor, our knockout recapitulates most closely the situation in broad carcinogenesis. To our knowledge, specific point mutations disturbing certain functions of ATRX or gain-of-function mutations are not described so far.

F1 fish of all three lines reached fertility at around three months of age, but never produced viable adult offspring with the atrx-/- genotype. Genotyping of developing embryos revealed that atrx-/- embryos of all lines died between 10 and 18 days post fertilization (dpf) (Fig 1B). Each of these embryos displayed a body curvature phenotype and a lack of the swim bladder. The embryonic lethality observed with the atrx-/- genotype is also observed in mice [3335]. In zebrafish, this is also true for nf1 which is duplicated in fish, so that at least 1 out of 4 alleles of the two nf1-genes (either nf1a or nf1b) has to remain wildtype to enable normal development of the embryo [29].

Knockout of atrx reduces globin expression and affects definitive erythropoiesis in zebrafish

Previous reports from human patients have shown that ATRX mutations are associated with reduced α-globin expression and lead to α-thalassemia myelodysplasia syndrome (ATMDS) [15]. To investigate the globin expression in atrx mutant zebrafish, we performed whole-mount in situ hybridization (WISH) using α-e1 and β-e1 globin probes. The results showed that the expression levels of α-e1 globin, but not β-e1 globin, were significantly reduced in atrx-/- homozygous mutants compared to atrx+/- heterozygous and wildtype siblings. This was observed during definitive erythropoiesis at 5 dpf, but not during primitive erythropoiesis at 22 hours post fertilization (hpf) (Fig 2A–2C), analogous to findings in humans [1]. Normal globin gene expression during primitive erythropoiesis likely reflects presence of maternal RNA or protein [36]. The expression levels of c-myb at both 36 hpf and 5 dpf did not show significant differences between atrx-/- homozygous mutants and wildtype embryos (S2A–S2C Fig), indicating that the development of hematopoietic stem/progenitor cells was not affected in atrx-/- homozygous mutants.

Fig 2. Homozygous loss of atrx results in an α-thalassemia-like phenotype in zebrafish development.

(A) Whole-mount in situ hybridization for α-e1 and β-e1 globins in wildtype (WT), atrx+/- heterozygous fish and atrx-/- homozygous mutants at 22 hpf and 5 dpf as indicated. Boxes outline the CHT region at 5 dpf, and are magnified in the right panels. Globin signal intensities of α-e1 (B) and β-e1 (C) in fish with different atrx backgrounds were calculated at 5dpf. Horizontal bars indicate the means ± SEM, which were compared with the two-tailed unpaired t-test; ns = not significant; ***p<0.001. (D) Erythroid progenitors development visualized by GFP in the Tg(gata1:GFP) transgenic line with wildtype (WT) or atrx-/- background at 7 dpf. CHT = caudal hematopoietic tissue; H = heart; KM = kidney marrow; dpf = days post fertilization. (E) Analysis of peripheral blood smears by MGG staining in wildtype fish (WT), atrx-/- homozygous mutant and atrx-/- homozygous mutant injected with zebrafish atrx mRNA at 7 dpf; scale bar: 10 μm. (F) Circularity index for the erythroid cells from fish with different atrx backgrounds (E) was calculated by ImageJ. Horizontal bars indicate the means ± SEM, which were compared with the two-tailed unpaired t-test; *p<0.05; **p<0.01.

To study the effects of reduced globin expression on erythropoiesis, we bred the zebrafish atrx+/- mutant with the Tg(gata1:GFP) transgenic line [37]. In this line, GFP expression is driven by the zebrafish gata1 promoter, providing a useful marker to identify erythroid cells in the Tg(gata1:GFP) transgenic line. At both 7 dpf and 12 dpf, atrx-/- homozygous mutants showed a remarkable increase in GFP-expressing cells in the heart, kidney marrow and caudal hematopoietic tissue (CHT) compared to wildtype fish (Figs 2D and S2D), indicating the accumulation of erythroid progenitors in these regions. Furthermore, May-Grunwald-Giemsa (MGG) staining of peripheral blood smears from atrx-/- homozygous mutants showed that erythrocytes had an aberrant rounded shape compared with the characteristic flattened elliptical morphology observed in the wildtype fish at 7 dpf. Zebrafish atrx mRNA injection rescued the mutant phenotype (Fig 2E and 2F). The rounded erythrocytes indicate the presence of circulating erythroid progenitors in the atrx-/- homozygous mutants, reflecting a block in erythroid cell differentiation resulting from the lack of α-globin expression. Taken together, these data indicate that the zebrafish atrx knockout model closely resembles the phenotype of human thalassemia patients with ATRX mutation, showing reduced globin expression and the accumulation of erythroid progenitors [1].

Additional loss of atrx in p53/nf1-deficient zebrafish promotes initiation of specific tumor types

Because ATRX is a tumor suppressor in many types of human cancers [1319], we examined the oncogenic effects of haploinsufficiency for artx alone and in cooperation with the p53/nf1-deficient genetic background. Both, p53 and NF1 are known tumor suppressors in humans and are inactivated in various cancer types. Loss of NF1 removes a major source of GTPase-activation affecting RAS and thus prolongs and strengthens RAS-MAPK signaling, thus enhancing the proliferation and survival of tumor cells [38,39]. The loss of nf1 has previously been shown to synergize with p53 mutation in a zebrafish model of MPNSTs and high-grade gliomas [29]. In zebrafish, the nf1 gene is duplicated (nf1a and nf1b) resulting in four functional nf1 alleles. Since a complete loss of nf1 is lethal, we bred p53-/-;nf1b-/-;nf1a+/- fish, resulting in offspring with the p53-/-;nf1b-/-;nf1a+/- or p53-/-;nf1b-/-;nf1a+/+ genotypes. Zebrafish with both genotypes are prone to develop MPNSTs, but with much faster onset and increased penetrance in the nf1a+/- genotype fish [29]. Moreover, we have previously observed high grade glioma tumorigenesis in p53-/-;nf1b-/-;nf1a+/- fish arising with low penetrance. Using CRISPR/Cas9, we created the atrx+/- genotype in the p53-/-;nf1b-/-;nf1a+/- background. Since a total loss of both nf1 and atrx is lethal in development, in-cross of this line resulted in viable offspring with 4 genotypes, 1) p53-/-;nf1b-/-;nf1a+/-;atrx+/-, 2) p53-/-;nf1b-/-;nf1a+/+;atrx+/-, 3) p53-/-;nf1b-/-;nf1a+/-;atrx+/+, and 4) p53-/-;nf1b-/-;nf1a+/+;atrx+/+.

Fish of all these genotypes were carefully monitored for tumor onset. Both, atrx+/+ and artx+/- fish of this line developed visually identical tumors located in the eye, gill, head, tail and predominantly in the abdomen (Fig 3A). Surprisingly, tumor watch experiments revealed no differences in time of tumor onset and penetrance associated with the atrx+/- genotype (Fig 3B). However, histopathology analysis confirmed that 100% of the tumor bearing p53-/-;nf1b-/-;nf1a+/-;atrx+/+ and p53-/-;nf1b-/-;nf1a+/+;atrx+/+ control fish had MPNSTs (n = 21 and n = 14 respectively; Tables 1 and S2). In the atrx+/- siblings of both nf1a+/- and nf1a+/+ populations, MPNSTs were identified in between 83.3% and 97.1% of all tumor-bearing fish and were indistinguishable in histology from their atrx+/+ control counterparts (n = 34 and n = 12, respectively; S3A Fig and Tables 1 and S2). Because ATRX is known to influence PRC2-mediated gene silencing, we examined lysine 27 tri-methylation status of histone 3 (H3K27me3), which is an epigenetic modification associated with genes silenced by PRC2, and predicts a worse prognosis for patients when lost in MPNST tissue [40]. Immunofluorescence staining using an H3K27me3-specific antibody revealed that this epigenetic mark was clearly present in atrx+/- and atrx+/+ tumors in the p53/nf1-deficient background (S3B Fig). Thus, partial atrx loss did not appear to have an inhibitory effect on total H3K27me3 deposition.

Fig 3. Addition of frameshift-mutations in atrx to the p53/nf1-deficient background induces specific changes in tumor biology.

(A) The morphologic features of p53/nf1/atrx- and p53/nf1-deficient tumors were indistinguishable by visual inspection. (B) When all tumor types were considered, tumor onset was not significantly altered upon heterozygous loss of atrx; a representative tumor watch experiment using line 1 is shown; the tumor watch has been reproduced in lines 1 and 2 with similar results; ns = not significant.

Table 1. Tumor spectrum of p53/nf1/atrx-deficient and control fish.

Individual fish have tumors of multiple organs with different histology, so that the number of tumors can exceed the number of fish.

Histopathologic analysis further revealed that 5 out of 34 (14.7%) analyzed tumor-bearing fish of the p53-/-;nf1b-/-;nf1a+/-;atrx+/- population had various tumor types other than or in addition to MPNSTs, including epithelioid sarcoma, biliary cancer, angiosarcoma, and an undifferentiated tumor of the eye with small round blue cell morphology (Table 1 and Fig 4). In the p53-/-;nf1b-/-;nf1a+/+;atrx+/- population this proportion was higher, with 3 out of 12 (25%) of zebrafish developing neoplasms consistent in histology with epithelioid sarcoma, undifferentiated pleomorphic sarcoma, or serous carcinoma of the ovary. These tumor types were never observed in atrx-wildtype siblings. It is noteworthy that many individual fish harbored more than one tumor site or type at the time of sacrifice. Epithelioid sarcomas were identified by the histologic appearance and positive cytokeratin (CK) staining of sections from paraffin embedded tumor tissue using the pan-CK antibody AE1/AE3 which is known to positively stain the vast majority of epithelioid sarcomas [41]. CK expression was also found in papillary serous carcinoma, undifferentiated pleomorphic sarcoma and biliary cancer (Fig 5). Faint positive staining for CK was detected in angiosarcoma tissue. All non-MPNST tumors in atrx+/- fish also stained strongly positive for the H3K27me3 mark (Fig 5). All detected sarcoma types other than MPNSTs are summarized below as soft tissue sarcomas.

Fig 4. Histology of tumor types detected in atrx+/- zebrafish with p53/nf1-deficient background.

This figure shows HE-stained sections of all tumor types detected in p53-/-, nf1b-/-, nf1a+/-, atrx+/- or p53-/-, nf1b-/-, nf1a+/+, atrx+/- fish that were not observed in atrx-wildtype siblings; scale bars: 50μm.

Fig 5. Immunofluorescence staining of tumor types detected in atrx+/- zebrafish with p53/nf1-deficient background.

This figure shows all tumor types detected in p53-/-, nf1b-/-, nf1a+/-, atrx+/- or p53-/-, nf1b-/-, nf1a+/+, atrx+/- fish that were not observed in atrx-wildtype siblings. For each tumor a section stained by indirect immunofluorescence for H3K27me3 (green) and pan-cytokeratin (red) is shown. The DNA is visualized with Hoechst (blue); scale bars: 10μm.

In previous studies, ATRX was identified as a tumor suppressor gene involved in the pathogenesis of gliomas, sarcomas and neuroendocrine tumors [1319]. Moreover, in the AACR Genie database, ATRX mutations are annotated in 49 cancer type categories (S3 Table), representing at least 45 distinct malignancies [32]. Within these, uterine sarcoma and glioma have the highest proportion of ATRX-mutant samples (19.60% and 16.74% respectively), whereas soft tissue sarcomas are ranked 15th with 6.42%. Moreover, the ATRX-mutation frequency in ovarian cancer is 2.65%, and 1.12% in hepatobiliary carcinoma. Thus, the tumor types observed in our model of combined p53/nf1/atrx-deficiency are faithfully reflected in the human ATRX-mutant tumor spectrum.

Soft tissue sarcomas and the other tumor types depicted in Fig 4 were not observed in the atrx-wildtype, p53/nf1-deficient controls or in the atrx+/-, p53/nf1-wildtype line. Thus, the atrx+/- genotype cooperated with the p53/nf1-deficient background in the development of these malignancies. Interestingly, p53/nf1/atrx-mutant fish developed various other tumor types regardless of whether two or three nf1 alleles were mutated (Tables 1 and S2), indicating atrx-deficiency synergizes with the combined loss of p53 rather than nf1. This cooperation between lof in p53 and atrx in our model is supported by previous data on the combined lof mutations in ATRX and p53 in high-grade gliomas [13] and leiomyosarcomas [19]. In the AACR Genie database, the overall p53 mutation rate considering all tumor samples analyzed is 38.64%, whereas among the ATRX-mutant samples 55.04% are co-mutants for p53 (Fig 6A) [32]. Gliomas, the tumors with the second highest prevalence of ATRX-mutation in the database (S3 Table) carry about twice as often a mutation in p53 if they have a mutation in ATRX (Fig 6A). In ATRX-mutant soft tissue sarcomas p53 mutations are observed almost 2.5 times as often as in the overall cohort. Together, these relationships indicate a selection advantage during the development of these particular tumor types for malignant clones with combined deficiencies in ATRX and p53. This does not extend as clearly to the loss of NF1 in soft tissue sarcomas, gliomas, and serous ovarian cancer (Fig 6B). However, when again all tumor samples were considered, similar results as for p53 were observed with 6.4% NF1-mutation rate in all tumor samples and 16.28% in the ATRX-mutated cohort. This indicates that both, loss of p53 and NF1, might cooperate with ATRX-deficiency in the right cancer type and genetic background. In hepatobiliary cancer in particular, there is a threefold higher incidence of nf1 mutations in atrx-mutant samples (Fig 6B). Thus loss of nf1b, and therefore RAS-MAPK pathway activation, appears to synergize with atrx deficiency in this cancer type. In tissues with predominate expression of nf1b compared to nf1a, the difference between loss of one allele or retention of both alleles could be very significant in terms of nf1 activity.

Fig 6. AACR Genie data on the co-occurrence of mutations in ATRX and p53 or NF1 in selected tumor types.

Data on mutation frequencies in p53 and NF1 were extracted from the AACR Genie Database [32].

Despite the shift to more heterogeneous tumor types, we did not see an accelerated overall tumor onset in p53/nf1/atrx-deficient fish compared to p53/nf1-knockout control fish, and the proportion of MPNST-bearing fish decreased in the atrx+/- cohort. From this result, we conclude that atrx loss does not promote the pathogenesis of MPNSTs in our model. In the AACR Genie database ATRX mutations are annotated for 1 of 31 analyzed MPNSTs (3.23%), while 2 of 29 samples that were profiled for copy number alterations (CNAs) show a deletion of atrx (6.9%). However, CNAs can occur due to general genomic instability and do not have to be specific for a gene contained in a larger deletion. Thus, we cannot exclude the possibility that ATRX loss can contribute to MPNST malignancy in the right context (e.g. by supporting ALT), although it clearly it does not accelerate MPNST onset in the genetic background of our model. The observation of an undiminished H3K27me3 signal in MPNSTs upon heterozygous atrx knockout in our model fits well to the fact that patients with MPNSTs retaining the H3K27me3 mark have a better prognosis [42].

Partial loss of atrx deregulates tert expression and PRC2-mediated gene silencing

The soft tissue sarcomas and other tumor types detected in p53/nf1/atrx-deficient fish were never seen in atrx wildtype siblings and have not been described in previous genome-editing studies in the p53/nf1-deficient line [29,43]. Thus, we conclude that the onset of these tumors was caused by decreased ATRX activity associated with the atrx+/- genotype. To examine the effect of heterozygous loss of atrx on global gene expression we used mRNA isolated from zebrafish tumor tissue to compare the gene expression profiles of p53-/-;nf1b-/-;nf1a+/-;atrx+/- and p53-/-;nf1b-/-;nf1a+/-;atrx+/+ MPNSTs by RNA-seq with three biological replicates for each group. Strikingly, the telomerase encoding gene tert was significantly downregulated in atrx+/- tumor tissue (Figs 7A and S4 and S4 Table). Since ATRX loss in humans is associated with ALT, we used fluorescence in situ hybridization with a telomere-specific probe (TelC-FISH) to visualize the telomeres of atrx+/- and atrx+/+ tumors in our line. In this experiment, we detected an increased telomere signal consistent with previously described phenotypes of ALT in human tumors (Fig 7B) [13]. This indicates that longer telomeres associated with atrx depletion-mediated ALT triggers a downregulation of tert. Indeed, a negative feedback mechanism that regulates TERT expression in a telomere length-dependent manner has been previously described in human cancer cells [44], and is supported by our observations to occur in our model system. Quantification of the telomere FISH-signal (n = 6) relative to the DNA-signal revealed a significantly larger total telomere area in TelC-FISH images derived from atrx+/- tumors compared to images derived from atrx+/+ controls (Fig 7C and S5 Table; p = 0.0151). Moreover, the total number of detectable telomere spots also increased significantly (Fig 7D and S6 Table; p = 0.042). This effect was even stronger, when only larger telomere signals (>1.5μm2 area) were considered (Fig 7E and S6 Table; p = 0.0013). This shows that heterozygous loss of atrx resulted in a measurable increase in telomere size which indicated atrx-deficiency-related ALT.

Fig 7. Loss of atrx in p53/nf1-depleted zebrafish tumors results in tert downregulation, ALT and disturbed PRC2 function.

(A) FPKM values show a significant downregulation of tert in p53/nf1/atrx-deficient fish compared to p53/nf1-kockout control fish; values were compared with the two-tailed unpaired t-test; n = 3, p = 0.0387. (B) Fluorescence in situ hybridization (FISH) using a Cy3-conjugated telomere-specific probe (TelC-Cy3, red) reveals elongated telomeres visible as large irregularly shaped spots in tumors with a knockout of p53, nf1 and atrx compared to atrx-wildtype control tumors, consistent with alternative lengthening of the telomeres; DNA stained with DAPI (blue); scale bars: 10μm; n = 6. (C) Quantification of relative telomere area in FISH images, normalized to DAPI signal area; values were compared with the two-tailed unpaired t-test; *p = 0.0151; n = 6. (D) Quantification of the relative number (#) of all detectable telomere spots in FISH images, normalized to DAPI signal area; values were compared with the two-tailed unpaired t-test; *p = 0.042; n = 6. (E) Quantification of the relative number (#) of telomere spots larger than ~1.5μm2 in FISH images, normalized to DAPI signal area; values were compared with the two-tailed unpaired t-test; *p = 0.0013; n = 6. (F) RNA-seq analysis demonstrates significantly increased expression of PRC2-target gene sets in atrx-deficient tumors; group 1: p53-/-, nf1b-/-, nf1a+/-, atrx+/+ (control); group 2: p53-/-, nf1b-/-, nf1a+/-, atrx+/-; n = 3.

In adult zebrafish tissues, the Tert protein regulates aging in a process similar to its role in human cells [4547]. In a previous study, tert-/- zebrafish aged faster and developed earlier spontaneous age-related tumors [46]. Thus, in zebrafish, shortening of the telomeres promotes tumor onset, which can be explained by the need in early tumorigenesis to increase the mutational burden by genomic instability in order to develop a malignant cancer. In patients, similar findings were reported for the transition of adenomas to metastatic colon cancer, which is among the most well described cancer progression systems. In particular, cells within adenomas that give rise to metastatic cancer have critically short telomeres [48]. In this context, induction of ALT due to the loss of atrx would not be expected to accelerate the onset of tumors, but rather facilitate later phases of their progression. This may explain why the tumor onset was not uniformly accelerated in our model. Thus, ALT cannot alone be responsible for the atrx lof-mediated onset of various tumor types including soft tissue sarcomas detected in our model.

To identify factors that may have contributed to this carcinogenesis we analyzed mRNA-seq data for gene set enrichment that might contribute to tumor formation. In this survey, we found the enrichment of notch1-targets (one gene set) and jak-stat-signaling (one gene set), epithelial differentiation (three gene sets) and PRC2-function (11 gene sets) in p53/nf1/atrx-deficient tumors compared to the atrx wildtype controls (Figs 7F and S5, S7 and S8 Tables). Interestingly, 10 of 11 PRC2-related gene sets showed a significant up-regulation of PRC2-targets, SUZ12-targets or H3K27me3 marked genes from various stem and progenitor cell types. However, there was one EZH2-target gene set that was significantly downregulated in p53/nf1/atrx-depleted tumors. This indicates that the loss of atrx predominantly led to a re-expression of PRC2 target genes silenced by H3K27me3 deposition, even though this might not affect all PRC2-targets. Interestingly, this was accompanied by an undiminished nuclear H3K27me3 signal in p53/nf1/atrx-deficient tumors detected by immunofluorescence. This observation is consistent with a previous study in human cells where loss of ATRX was found to displace PCR2-deposited H3K27me3 silencing marks away from the target gene promoters to the intergenic space leading to their re-expression [9]. Thus, it is likely that ATRX loss does not abolish H3K27me3 marks in the genome, but rather deregulates their positioning throughout the genome.

In this study, we show the consequences of inactivation of atrx in the zebrafish germline, which resulted in the first zebrafish model atrx deficiency causing alpha-thalassemia and the contribution of loss of atrx to carcinogenesis in the background of p53 loss and RAS-MAPK pathway activation. Faithful models of both of these consequences of atrx loss will be valuable for future preclinical studies. We provide evidence that atrx and p53 cooperate in the carcinogenesis of soft tissue sarcomas and other malignancies. Our data are consistent with a negative feedback loop downregulating telomerase upon loss of atrx, causing alternative lengthening of the telomeres, and indicate that the role of atrx-deficiency in tumor initiation may also be linked to disturbed PRC2-mediated gene silencing.

Materials and methods

Ethics statement

All zebrafish studies and maintenance were done in accord with Dana-Farber Cancer Institute IACUC-approved protocol (#02–107).

Zebrafish lines and maintenance

Zebrafish were raised and maintained according to standard procedures. They were derived from the AB background strain, the Tg(gata1:GFP) line [37] or the p53/nf1-deficient background [29,49]. All p53-/- fish were homozygous for the p53-M214K mutation described previously [49]. The p53/nf1-deficient fish carried a frameshift mutation in exon 26 of nf1a and in exon 17 of nf1b which truncate the Nf1 protein before its functional GRD domain, as published in a previous study [29].

The zebrafish atrx mutant lines were generated by the CRISPR/Cas genome editing system using the previous described protocol [30]. The plasmid constructs pDR274 (#42250) and pMLM3613 (#42251) were purchased from Addgene. Oligonucleotides 5’-TAG GTC CTG AGT TCC GTA ACA A-3’ and 5’-AAA CTT GTT ACG GAA CTC AGG A-3’ were annealed and cloned into the pDR274 vector to generate single guide RNAs (gRNAs) targeting atrx exon 4. Each embryo was injected with 1 nl of solution containing 25 ng/μl gRNA and 600 ng/μl Cas9 mRNA at the 1 cell stage. Mosaic F0 fish with germline mutation were identified, and stable mutant lines were established by outcrossing. To genotype the atrx mutant line, we first amplified a DNA fragment containing the mutated site, using the primer pair atrx_E3I3Fw: 5’-GCA AGC TGC CAC AAG GTT AGT CC-3’ and atrx_I4Rv: 5’-GTC ACA AAC ACG TCA CCA CTT A-3’ with genomic DNA extracted from fin clips serving as template. The DNA product was either sent for sequencing with the primer atrx_seqI3Fw: 5’-TGT TCC GAT CAG TCT TCC TTA GC-3’ or digested by BslI. The wildtype DNA fragment was then digested into 2 fragments of 171 (bp) and 282 bp while the mutant one was resistant to digestion, revealing a band of 455 to 475 bp (please note that the mutant PCR product can be larger due to insertions). Alternatively, if PCR products were not sequenced, primers atrx_shortFw: 5’-GCAGGCACAGTAGTGGTGAAGCCA-3’ and atrx_shortRv: 5’- CACCAGGACGTTTCCGCGCACCCT-3’ were used. The wildtype DNA fragment was then digested into 79 and 45bp fragments. The mutant amplicon remained undigested at a size of 126 to 146 bp (please note that the mutant PCR product can be larger due to insertions).

Injection in 1-cell-stage embryos

All gRNA was transcribed in vitro from DraI (NEB) linearized pDR274-gRNA plasmid DNA using the MAXIscript T7 Kit (Ambion Inc.). Cas9 mRNA was transcribed in vitro from NotI (NEB, Ipswich, MA, United States) linearized pCS2-nCas9n plasmid DNA using the mMessage mMachine SP6 Kit (Ambion Inc., Foster City, CA, United States). Oligonucleotides were mixed between 1:5 and 1:1 with a 0.5% phenol red solution (Sigma-Aldrich, Burlington, MA, United States) to a final concentration of 25ng/μl gRNA and 600ng/μl Cas9 mRNA. To induce mutations in the genome of zebrafish, we injected one-cell-stage embryos with the oligonucleotide/phenol red mix described above within 30min after fertilization using a glass capillary mounted into an air pressure injector (Harvard Apparatus, Cambridge, MA, United States). The injection volume was 1nl oligonucleotide/phenol red mix per embryo. Dead embryos were removed 3 to 6 hours after injection using a Leica M420 microscope (Wetzlar, Germany). For zebrafish atrx rescue experiment, atrx was cloned into pCS2+ vector by PCR from cDNA of zebrafish embryos, using the primer pair atrx_Fw: 5’-CGG CTC GAG ATG GCA ACC AAT GAC GTA AAT ATT-3’ and atrx_Rv: 5’-CGG TAC GTA TTA CAG ACC CTT AGA TGG GCC TGG-3’. Zebrafish atrx mRNA was synthesized in vitro from NotI (NEB, Ipswich, MA, United States) linearized atrx-pCS2+ plasmid DNA using the mMessage mMachine SP6 Kit (Ambion Inc., Foster City, CA, United States). 1 nl atrx mRNA (500 ng/μl) containing phenol red dye was injected into zebrafish embryos at the one-cell stage.

Tumor watch in tumor-prone zebrafish lines

The sox10:GFP zebrafish were genotyped at 2–3 months of age and separated into distinct tanks according to genotype. At least every 2 weeks, all fish were examined for tumor onset using a Nikon C-DSD115 fluorescence microscope (Tokyo, Japan). The time point of tumor onset was defined at the first observation of a GFP+ growth that did not regress within 2 weeks. Fish that died from tumor-unrelated causes were removed from the analysis. A survival analysis with Graph pad Prism 7 software was performed to visualize the tumor onset rate and tumor penetrance.

Zebrafish histopathology

Tumor-bearing fish were sacrificed and fixed in 4% paraformaldehyde (PFA) in PBS at 4°C for 1–3 days. Subsequently, the fish were washed in PBS or 70% ethanol and embedded in paraffin. Paraffin sectioning (3-micron) and hematoxylin/eosin staining were performed at the DF/HCC Research Pathology Core according to standard protocols.

Indirect immunofluorescence of paraffin-embedded tumor tissue

The indirect immunofluorescence protocol was adapted from previous studies [50,51]. Zebrafish tumor sections were incubated twice for 10 min each in xylol to remove paraffin. For rehydration the slides were washed for 5min each in 100% ethanol, 96% ethanol, 70% ethanol twice for 5min each in H2Odd. Epitopes were unmasked by heating for 2min and 30 sec in 10 mmol citric acid/NaCitrate-buffer (pH 6; 18% citric acid, 82% NaCitrate) followed by 5min incubation after each cooking step. Next, slides were blocked for 15min in PBS + 0.1% BSA at room temperature. The DNA was stained for 10 min with Hoechst 33342 (1mg/ml) and diluted 1:250 diluted in PBS + 0.1% BSA. Primary antibodies were diluted in PBS + 0.1% BSA and 100μl antibody-mix was incubated for at least 1h on fat surrounded area around the tissue (wet chamber in the dark). Secondary antibodies were diluted in PBS + 0.1% BSA and 100μl antibody-mix was incubated for 1h on fat surrounded area around the tissue (wet chamber in the dark). Slides were washed for 5-10min in PBS + 0.1% BSA after each staining step. After secondary antibody incubation the slides were washed 1x in PBS + 0.1% BSA, 1 x in PBS and 1x in H2Odd, dried with an absorbent paper and mounted in 30μl PromoFluor Antifade Reagent mounting medium (Promokine, Heidelberg, Germany) using a #1 cover glass. As primary antibodies were used: pan-cytokeratin AE1/AE3 (Novus Biologicals, Littleton, CO, USA), H3K27me3 C36B11 (Cell Signaling Technology, Danvers, MA, USA). Secondary antibodies were conjugated with Alexa 488, 568 (Thermo Fisher Scientific, Waltham, MA, USA). Slides were imaged by a Leica SP5X scanning confocal microscope (Wetzlar, Germany) at the Confocal and Light Microscopy core facility at Dana-Farber Cancer Institute.

Telomere PNA-fluorescence in situ hybridization (FISH)

Telomere PNA-FISH was performed as previously described [47]. Briefly, zebrafish paraffin sections were deparaffinized, followed by hybridization for 2 h at room temperature in the dark. Slides were washed, mounted and imaged by a Leica SP5X scanning confocal microscope at the Confocal and Light Microscopy core facility at Dana-Farber Cancer Institute. Cy3 labeled PNA TelC probes (CCCTAA repeats) were purchased from PNA Bio INC (Thousand Oaks, CA, USA). DNA was counterstained with DAPI. Individual images of the Cy3 and DAPI channels were analyzed in ImageJ to quantify the total area covered by Cy3 stained telomeres or DNA and the number of telomere signals (particles in Cy3 channel). Moreover, the average number of large telomere spots was quantified to determine changes in the abundance of large telomeres reminiscent of alternative lengthening of the telomeres. Large was defined as >0,01 inch2 size in digital image exported from LasX software (Leica) as single channel image which corresponds to ~1,556μm2 real size. For each image these values were normalized to the DNA content of the respective image to quantify the relative telomere signal within the nuclear compartment. For this the area percentage value of the DAPI signal in each image was determined using ImageJ. From this a “DNA content factor” was calculated by the formula 100/% DNA content. Raw values were subsequently multiplied with this DNA content factor. For each analysis the threshold was set to include the specific signal and exclude the background. The threshold was kept constant for all images compared. Images with background signals that could not be eliminated were excluded from the analysis. A two-tailed unpaired t-test type 2 was performed using GraphPad Prism software.

Whole-mount RNA in situ hybridization (WISH)

Riboprobe labeling and WISH were performed as described previously [52,53]. The digoxigenin (DIG)-labeled RNA probes to detect α-e1 and β-e1 globins [54], and c-myb [55] were as previously described.

Peripheral blood smears and May-Grunwald-Giemsa (MGG) staining

Peripheral blood smears were prepared on glass slides as described previously [55]. The slides were fixed and stained with May-Grunwald–Giemsa (MGG) solution (Sigma-Aldrich, St. Louis, MO) according to the manufacturer’s instructions and visualized with a Zeiss AXIO microscope (Zeiss, Oberkochen, Germany).

RNA-Seq data analysis

RNA was isolated from one half of p53/nf1/atrx-deficient MPNSTs and p53/nf1-deficient control MPNSTs using the AllPrep DNA/RNA Mini Kit from Qiagen (Hilden, Germany) and the other half was analyzed by histopathology as described above. Library preparation, quality control, and next-generation sequencing were conducted by the Molecular Biology Core Facility of the Dana-Farber Cancer Institute following standard protocols.

Gene expression values were derived from paired end RNA-Seq data that compared samples from p53/nf1/atrx-deficient tumors to samples from atrx-wildtype controls (3 vs. 3 samples). FastQC was used to evaluate read quality on raw RNA-Seq reads and trimmed reads. Trimming of low quality reads and clipping of sequencing adapters was done using the program Trimmomatic [56] and all reads shorter than 36bp after trimming were dropped. Reads were aligned to the GRCz10 version of the zebrafish reference genome with TopHat [57] version 2.1.0. Bam sorting and indexing was done with SamTools [58] and duplicate reads were removed using Picard-tools ( Gene level counts were obtained with htseq-count version 0.9.1 [59]. Differential gene expression was evaluated with the R Bioconductor package DESeq2 [60] and normalized expression values for individual samples were obtained from DESeq2 using the variance stabilizing transform on the raw counts. The variance stabilizing transformed data were used for GSEA.

Gene set enrichment analysis

Gene Set Enrichment Analysis (GSEA) [61,62] was used to evaluate the association of genes with compared p53/nf1/atrx-deficient tumors to the atrx-wildtype controls. GSEA was run with 2500 permutations of the phenotype using signal-to-noise to rank genes. GSEA was performed with signatures from version 6.0 of the molecular signature database (MolSigDB) ( the c2 curated gene sets from various sources such as online pathway databases, the biomedical literature, and knowledge of domain experts, the c3 motif gene sets, c5 gene ontology (GO) MF, CC, and BP ontologies, and the c7 immunologic signatures.

AACR Genie database search

Data of the AACR Genie database (version 3.0 public release) were extracted online using the cBioPortal application ( and further processed in Microsoft Excel.

Supporting information

S1 Fig. CRISPR/Cas9-induced mutation of the atrx coding sequence in zebrafish germline.

An exon 4 targeting gRNA was used to induce frameshift mutations in atrx coding sequence. Mutations were induced in either wildtype (WT) or p53/nf1-deficient genetic background.


S2 Fig. Homozygous loss of atrx does not affect hematopoietic stem/progenitor cell development.

(A) Whole-mount in situ hybridization for c-myb at 36 hpf and 5 dpf in wildtype (WT), atrx+/- heterozygous fish and atrx-/- homozygous mutants as indicated. Boxes outline the AGM region at 36 hpf and the CHT region at 5 dpf, and are magnified in the right panels. c-myb signal intensities at 36 hpf (B) and 5 dpf (C) in fish with different atrx backgrounds were calculated. Horizontal bars indicate the means ± SEM, which were compared with the two-tailed unpaired t-test; ns = not significant. (D) Erythroid progenitors development visualized by GFP in the Tg(gata1:GFP) transgenic line with wildtype (WT) or atrx-/- background at 12 dpf. AGM = aorta-gonad-mesonephros; CHT = caudal hematopoietic tissue; H = heart; KM = kidney marrow; hpf = hours post fertilization; dpf = days post fertilization.


S3 Fig. MPNST biology in atrx+/+ and atrx+/- siblings in p53/nf1-deficient background.

(A) HE-staining of p53/nf1 atrx+/+ and atrx+/- MPNSTs reveals no differences in histology; scale bars: 50μm. (B) Indirect immunofluorescence staining of two MPNSTs each of the p53/nf1 atrx+/+ and atrx+/- cohorts both show detectable tri-methylation of histone 3, lysine 27 (H3K27me3, green); scale bars: 10μm.


S4 Fig. Head map of RNA-Seq results.

Significantly up- or downregulated genes (p<0.025) between p53-/-, nf1b-/-, nf1a+/-, atrx+/+ and p53-/-, nf1b-/-, nf1a+/-, atrx+/- samples (n = 3). The heat map is row normalized with blue representing minimum expression and red representing maximum expression.


S5 Fig. Gene set enrichment analysis.

RNA-Seq analysis revealed significantly increased expression of PRC2-related gene sets, NOTCH1 and JAK/STAT signaling targets and markers of epithelial differentiation; group 1: p53-/-, nf1b-/-, nf1a+/-, atrx+/+ (control); group 2: p53-/-, nf1b-/-, nf1a+/-, atrx+/-.


S1 Table. Germline atrx mutations in F1 in p53/nf1-deficient background.


S2 Table. Histopathology of tumor-bearing fish by age and genotype.


S3 Table. Analysis of ATRX-mutation frequency in human malignancies in AACR Genie Database.


S4 Table. FPKM and log2 FPKM values from cufflinks for atrx +/+ versus atrx +/- samples in p53/nf1-deficient background.


S5 Table. Relative quantification of telomere signal spots in FISH images derived from atrx+/+ and atrx+/- tumors in p53/nf1-deficient background.


S6 Table. Relative quantification of large telomere signal spots in FISH images derived from atrx+/+ and atrx+/- tumors in p53/nf1-deficient background.


S7 Table. Gene sets enriched in p53-/-, nf1b-/-, nf1a+/-, atrx+/- MPNSTs.


S8 Table. Gene sets enriched in p53-/-, nf1b-/-, nf1a+/- control MPNSTs.



The authors would like to acknowledge the American Association for Cancer Research and its financial and material support in the development of the AACR Project GENIE registry, as well as members of the consortium for their commitment to data sharing. Interpretations are the responsibility of study authors.

F.O. acknowledges the German Cancer Aid for their generous funding within the Mildred-Scheel-Postdoctoral program of the Mildred Scheel Foundation.

The authors thank the Childhood Brain Tumor Foundation for their generous support.

The authors thank Grace Thurston and Daniel Debiasi for their excellent care of the zebrafish.

The authors thank John R. Gilbert for editorial assistance and critical comments and Myzafere Ormenaj for help with the manuscript.


  1. 1. Gibbons RJ, Pellagatti A, Garrick D, Wood WG, Malik N, et al. (2003) Identification of acquired somatic mutations in the gene encoding chromatin-remodeling factor ATRX in the alpha-thalassemia myelodysplasia syndrome (ATMDS). Nat Genet 34: 446–449. pmid:12858175
  2. 2. Steensma DP, Gibbons RJ, Higgs DR (2005) Acquired alpha-thalassemia in association with myelodysplastic syndrome and other hematologic malignancies. Blood 105: 443–452. pmid:15358626
  3. 3. Steensma DP, Higgs DR, Fisher CA, Gibbons RJ (2004) Acquired somatic ATRX mutations in myelodysplastic syndrome associated with alpha thalassemia (ATMDS) convey a more severe hematologic phenotype than germline ATRX mutations. Blood 103: 2019–2026. pmid:14592816
  4. 4. Steensma DP, Viprakasit V, Hendrick A, Goff DK, Leach J, et al. (2004) Deletion of the alpha-globin gene cluster as a cause of acquired alpha-thalassemia in myelodysplastic syndrome. Blood 103: 1518–1520. pmid:14576055
  5. 5. Stevenson RE (1993) Alpha-Thalassemia X-Linked Intellectual Disability Syndrome. In: Adam MP, Ardinger HH, Pagon RA, Wallace SE, Bean LJH et al., editors. GeneReviews((R)). Seattle (WA).
  6. 6. Dhayalan A, Tamas R, Bock I, Tattermusch A, Dimitrova E, et al. (2011) The ATRX-ADD domain binds to H3 tail peptides and reads the combined methylation state of K4 and K9. Hum Mol Genet 20: 2195–2203. pmid:21421568
  7. 7. Voon HP, Hughes JR, Rode C, De La Rosa-Velazquez IA, Jenuwein T, et al. (2015) ATRX Plays a Key Role in Maintaining Silencing at Interstitial Heterochromatic Loci and Imprinted Genes. Cell Rep 11: 405–418. pmid:25865896
  8. 8. Rapkin LM, Ahmed K, Dulev S, Li R, Kimura H, et al. (2015) The histone chaperone DAXX maintains the structural organization of heterochromatin domains. Epigenetics Chromatin 8: 44. pmid:26500702
  9. 9. Sarma K, Cifuentes-Rojas C, Ergun A, Del Rosario A, Jeon Y, et al. (2014) ATRX directs binding of PRC2 to Xist RNA and Polycomb targets. Cell 159: 869–883. pmid:25417162
  10. 10. Muller J, Verrijzer P (2009) Biochemical mechanisms of gene regulation by polycomb group protein complexes. Curr Opin Genet Dev 19: 150–158. pmid:19345089
  11. 11. Cerase A, Pintacuda G, Tattermusch A, Avner P (2015) Xist localization and function: new insights from multiple levels. Genome Biol 16: 166. pmid:26282267
  12. 12. Clemson CM, McNeil JA, Willard HF, Lawrence JB (1996) XIST RNA paints the inactive X chromosome at interphase: evidence for a novel RNA involved in nuclear/chromosome structure. J Cell Biol 132: 259–275. pmid:8636206
  13. 13. Schwartzentruber J, Korshunov A, Liu XY, Jones DT, Pfaff E, et al. (2012) Driver mutations in histone H3.3 and chromatin remodelling genes in paediatric glioblastoma. Nature 482: 226–231. pmid:22286061
  14. 14. Jiao Y, Shi C, Edil BH, de Wilde RF, Klimstra DS, et al. (2011) DAXX/ATRX, MEN1, and mTOR pathway genes are frequently altered in pancreatic neuroendocrine tumors. Science 331: 1199–1203. pmid:21252315
  15. 15. Koelsche C, Renner M, Johann P, Leiss I, Sahm F, et al. (2016) Differential nuclear ATRX expression in sarcomas. Histopathology 68: 738–745. pmid:26291601
  16. 16. Liau JY, Lee JC, Tsai JH, Yang CY, Liu TL, et al. (2015) Comprehensive screening of alternative lengthening of telomeres phenotype and loss of ATRX expression in sarcomas. Mod Pathol 28: 1545–1554. pmid:26428317
  17. 17. Liau JY, Tsai JH, Jeng YM, Lee JC, Hsu HH, et al. (2015) Leiomyosarcoma with alternative lengthening of telomeres is associated with aggressive histologic features, loss of ATRX expression, and poor clinical outcome. Am J Surg Pathol 39: 236–244. pmid:25229770
  18. 18. Liau JY, Tsai JH, Yang CY, Lee JC, Liang CW, et al. (2015) Alternative lengthening of telomeres phenotype in malignant vascular tumors is highly associated with loss of ATRX expression and is frequently observed in hepatic angiosarcomas. Hum Pathol 46: 1360–1366. pmid:26190196
  19. 19. Yang CY, Liau JY, Huang WJ, Chang YT, Chang MC, et al. (2015) Targeted next-generation sequencing of cancer genes identified frequent TP53 and ATRX mutations in leiomyosarcoma. Am J Transl Res 7: 2072–2081. pmid:26692951
  20. 20. Maser RS, DePinho RA (2002) Connecting chromosomes, crisis, and cancer. Science 297: 565–569. pmid:12142527
  21. 21. Heaphy CM, Subhawong AP, Hong SM, Goggins MG, Montgomery EA, et al. (2011) Prevalence of the alternative lengthening of telomeres telomere maintenance mechanism in human cancer subtypes. Am J Pathol 179: 1608–1615. pmid:21888887
  22. 22. Singhi AD, Liu TC, Roncaioli JL, Cao D, Zeh HJ, et al. (2017) Alternative Lengthening of Telomeres and Loss of DAXX/ATRX Expression Predicts Metastatic Disease and Poor Survival in Patients with Pancreatic Neuroendocrine Tumors. Clin Cancer Res 23: 600–609. pmid:27407094
  23. 23. Lovejoy CA, Li W, Reisenweber S, Thongthip S, Bruno J, et al. (2012) Loss of ATRX, genome instability, and an altered DNA damage response are hallmarks of the alternative lengthening of telomeres pathway. PLoS Genet 8: e1002772. pmid:22829774
  24. 24. Bower K, Napier CE, Cole SL, Dagg RA, Lau LM, et al. (2012) Loss of wild-type ATRX expression in somatic cell hybrids segregates with activation of Alternative Lengthening of Telomeres. PLoS One 7: e50062. pmid:23185534
  25. 25. Voon HPJ, Collas P, Wong LH (2016) Compromised Telomeric Heterochromatin Promotes ALTernative Lengthening of Telomeres. Trends Cancer 2: 114–116. pmid:28741530
  26. 26. Amorim JP, Santos G, Vinagre J, Soares P (2016) The Role of ATRX in the Alternative Lengthening of Telomeres (ALT) Phenotype. Genes (Basel) 7.
  27. 27. Dunham MA, Neumann AA, Fasching CL, Reddel RR (2000) Telomere maintenance by recombination in human cells. Nat Genet 26: 447–450. pmid:11101843
  28. 28. Koschmann C, Calinescu AA, Nunez FJ, Mackay A, Fazal-Salom J, et al. (2016) ATRX loss promotes tumor growth and impairs nonhomologous end joining DNA repair in glioma. Sci Transl Med 8: 328ra328.
  29. 29. Shin J, Padmanabhan A, de Groh ED, Lee JS, Haidar S, et al. (2012) Zebrafish neurofibromatosis type 1 genes have redundant functions in tumorigenesis and embryonic development. Dis Model Mech 5: 881–894. pmid:22773753
  30. 30. Hwang WY, Fu Y, Reyon D, Maeder ML, Tsai SQ, et al. (2013) Efficient genome editing in zebrafish using a CRISPR-Cas system. Nat Biotechnol 31: 227–229. pmid:23360964
  31. 31. Jao LE, Wente SR, Chen W (2013) Efficient multiplex biallelic zebrafish genome editing using a CRISPR nuclease system. Proc Natl Acad Sci U S A 110: 13904–13909. pmid:23918387
  32. 32. The AACR Project GENIE Consortium (2017) AACR Project GENIE: Powering Precision Medicine through an International Consortium. Cancer Discov 7: 818–831. pmid:28572459
  33. 33. De Raedt T, Beert E, Pasmant E, Luscan A, Brems H, et al. (2014) PRC2 loss amplifies Ras-driven transcription and confers sensitivity to BRD4-based therapies. Nature 514: 247–251. pmid:25119042
  34. 34. Garrick D, Sharpe JA, Arkell R, Dobbie L, Smith AJ, et al. (2006) Loss of Atrx affects trophoblast development and the pattern of X-inactivation in extraembryonic tissues. PLoS Genet 2: e58. pmid:16628246
  35. 35. Roberts CW, Galusha SA, McMenamin ME, Fletcher CD, Orkin SH (2000) Haploinsufficiency of Snf5 (integrase interactor 1) predisposes to malignant rhabdoid tumors in mice. Proc Natl Acad Sci U S A 97: 13796–13800. pmid:11095756
  36. 36. Baumann C, Viveiros MM, De La Fuente R (2010) Loss of maternal ATRX results in centromere instability and aneuploidy in the mammalian oocyte and pre-implantation embryo. PLoS Genet 6: e1001137. pmid:20885787
  37. 37. Long Q, Meng A, Wang H, Jessen JR, Farrell MJ, et al. (1997) GATA-1 expression pattern can be recapitulated in living transgenic zebrafish using GFP reporter gene. Development 124: 4105–4111. pmid:9374406
  38. 38. Ballester R, Marchuk D, Boguski M, Saulino A, Letcher R, et al. (1990) The NF1 locus encodes a protein functionally related to mammalian GAP and yeast IRA proteins. Cell 63: 851–859. pmid:2121371
  39. 39. McCormick F (1995) Ras signaling and NF1. Curr Opin Genet Dev 5: 51–55. pmid:7749326
  40. 40. Cleven AH, Al Sannaa GA, Briaire-de Bruijn I, Ingram DR, van de Rijn M, et al. (2016) Loss of H3K27 tri-methylation is a diagnostic marker for malignant peripheral nerve sheath tumors and an indicator for an inferior survival. Mod Pathol 29: 1113. pmid:27573709
  41. 41. Hasegawa T, Matsuno Y, Shimoda T, Umeda T, Yokoyama R, et al. (2001) Proximal-type epithelioid sarcoma: a clinicopathologic study of 20 cases. Mod Pathol 14: 655–663. pmid:11454997
  42. 42. Dieter SM, Ball CR, Hoffmann CM, Nowrouzi A, Herbst F, et al. (2011) Distinct types of tumor-initiating cells form human colon cancer tumors and metastases. Cell Stem Cell 9: 357–365. pmid:21982235
  43. 43. Ki DH, He S, Rodig S, Look AT (2017) Overexpression of PDGFRA cooperates with loss of NF1 and p53 to accelerate the molecular pathogenesis of malignant peripheral nerve sheath tumors. Oncogene 36: 1058–1068. pmid:27477693
  44. 44. Kim W, Ludlow AT, Min J, Robin JD, Stadler G, et al. (2016) Regulation of the Human Telomerase Gene TERT by Telomere Position Effect-Over Long Distances (TPE-OLD): Implications for Aging and Cancer. PLoS Biol 14: e2000016. pmid:27977688
  45. 45. Carneiro MC, de Castro IP, Ferreira MG (2016) Telomeres in aging and disease: lessons from zebrafish. Dis Model Mech 9: 737–748. pmid:27482813
  46. 46. Carneiro MC, Henriques CM, Nabais J, Ferreira T, Carvalho T, et al. (2016) Short Telomeres in Key Tissues Initiate Local and Systemic Aging in Zebrafish. PLoS Genet 12: e1005798. pmid:26789415
  47. 47. Henriques CM, Carneiro MC, Tenente IM, Jacinto A, Ferreira MG (2013) Telomerase is required for zebrafish lifespan. PLoS Genet 9: e1003214. pmid:23349637
  48. 48. Plentz RR, Wiemann SU, Flemming P, Meier PN, Kubicka S, et al. (2003) Telomere shortening of epithelial cells characterises the adenoma-carcinoma transition of human colorectal cancer. Gut 52: 1304–1307. pmid:12912862
  49. 49. Berghmans S, Murphey RD, Wienholds E, Neuberg D, Kutok JL, et al. (2005) tp53 mutant zebrafish develop malignant peripheral nerve sheath tumors. Proc Natl Acad Sci U S A 102: 407–412. pmid:15630097
  50. 50. Ball CR, Oppel F, Ehrenberg KR, Dubash TD, Dieter SM, et al. (2017) Succession of transiently active tumor-initiating cell clones in human pancreatic cancer xenografts. EMBO Mol Med 9: 918–932. pmid:28526679
  51. 51. Oppel F, Muller N, Schackert G, Hendruschk S, Martin D, et al. (2011) SOX2-RNAi attenuates S-phase entry and induces RhoA-dependent switch to protease-independent amoeboid migration in human glioma cells. Mol Cancer 10: 137. pmid:22070920
  52. 52. Chen J, Ruan H, Ng SM, Gao C, Soo HM, et al. (2005) Loss of function of def selectively up-regulates Delta113p53 expression to arrest expansion growth of digestive organs in zebrafish. Genes Dev 19: 2900–2911. pmid:16322560
  53. 53. Tao T, Shi H, Guan Y, Huang D, Chen Y, et al. (2013) Def defines a conserved nucleolar pathway that leads p53 to proteasome-independent degradation. Cell Res 23: 620–634. pmid:23357851
  54. 54. Ganis JJ, Hsia N, Trompouki E, de Jong JL, DiBiase A, et al. (2012) Zebrafish globin switching occurs in two developmental stages and is controlled by the LCR. Dev Biol 366: 185–194. pmid:22537494
  55. 55. Gjini E, Mansour MR, Sander JD, Moritz N, Nguyen AT, et al. (2015) A zebrafish model of myelodysplastic syndrome produced through tet2 genomic editing. Mol Cell Biol 35: 789–804. pmid:25512612
  56. 56. Bolger AM, Lohse M, Usadel B (2014) Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30: 2114–2120. pmid:24695404
  57. 57. Langmead B, Trapnell C, Pop M, Salzberg SL (2009) Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol 10: R25. pmid:19261174
  58. 58. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, et al. (2009) The Sequence Alignment/Map format and SAMtools. Bioinformatics 25: 2078–2079. pmid:19505943
  59. 59. Anders S, Pyl PT, Huber W (2015) HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics 31: 166–169. pmid:25260700
  60. 60. Anders S, Huber W (2010) Differential expression analysis for sequence count data. Genome Biol 11: R106. pmid:20979621
  61. 61. Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, et al. (2003) PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat Genet 34: 267–273. pmid:12808457
  62. 62. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, et al. (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102: 15545–15550. pmid:16199517