In contrast to the latter 3 subtypes and under the current treatment regimens, t(9,22)(q34;q11), t(4;11)(q21;q23) – the BCR/ABL1 fusion and the MLL/AF4 fusion, respectively- and hypodiploidy (displaying <44 chromosomes) are associated with poor outcome in BCP-ALL (Pui et al., 2008, Pui et al., 2011). Apart from cytogenetic aberrations and immunophenotypic characterization, further prognostic parameters are important for clinical outcome in BCP-ALL cases. For several ALL subtypes, age at diagnosis of the leukemia is associated with prognosis: children between the ages 1-9 have a 5 year event-free survival of 88%, while adolescents between 10 and 15 years of age have a 5 year event-free survival rate of 73% and infants, younger than 12 months, a 44% 5 year event-free survival rate. The worst outcome is observed in infants younger than 6 months of age. To date, the patient’s response to therapy is the best predictive parameter. In this regard, Minimal residual disease (MRD) measurement has become an effective tool to assess treatment response in ALL.
MRD has been incorporated in clinical trials and is used as treatment stratifying parameter. MRD levels can be assessed most effectively by PCR analysis or flow cytometry and provide standardized tools to determine the effectiveness of ALL treatment (van Dongen et al., 1998, Pui and Evans, 2006, Conter et al., 2010, Schrappe et al., 2011).
Treatment of ALL at initial diagnosis and at relapse ALL treatment is characterized by 3 subsequent stages. The first stage is the so called induction therapy. Here, the main goal is to achieve remission of the disease by a very intense treatment aimed at eradicating 90% of the initially present leukemic blast count and re-establishment of normal hematopoiesis. In induction therapy, glucocorticoids (dexamethasone or prednisolone), anthracyclin or asparaginase or 16 both, in addition to vincristine are applied for one month. The current treatment protocols achieve a disease remission, defined by regular blood counts and bone marrow samples that are morphologically normal in approximately 98% of cases. After achieving remission, consolidation therapy is started.
Here, commonly used regimens for ALL include high-dose asparaginase and methotrexate with mercaptopurine over an extended period of time and reinduction therapy. After 4 to 8 months of consolidation treatment, the third and last stage, continuation treatment, also called maintenance treatment, is started. Methotrexate is administered on a weekly basis, while mercaptopurine is given on a daily basis, for total treatment duration of 2 years. If patients have an initial poor response to treatment or are considered especially high-risk patients, allogeneic hematopoietic stem-cell transplantation is considered (Pui and Evans, 2006). With improvement of survival rates, ALL treatment has started to focus on preventing acute and late deleterious side-effects of treatment. Therefore, carcinogenic or major organ-damaging drugs are avoided or dosage is reduced in standard-risk ALL cases. Further efforts to reduce treatment-related toxicity are, for instance, the use of toxicity-counteracting drugs and the generation and application of targeted approaches and advanced molecular diagnostics (Pui and Evans, 2006).
Under the right circumstances, these advances in therapy could increase cure rates in childhood ALL cells, while reducing toxicity. Analysis of gene expression in ALL Profiling gene expression of ALL blasts by microarray or next generation sequencing -based RNA sequencing allows monitoring the entire transcriptome. This has since led to the identification of a new childhood ALL subgroup (Ph-like or BCR-ABL1-like) 17 and significantly contributed to our understanding of leukemia in numerous ways (Andersson et al.
, 2005, Den Boer et al., 2009): expression analysis is an important tool to reveal potential candidate pathways and genes that might be of prognostic or therapeutic value for specific ALL subtypes (Yeoh et al., 2002, Ross et al., 2003, Pui et al., 2004).
Furthermore, expression analysis showed that primary ALLs can be distinguished according to their specific signature upon correlation with established cyto/genetic subgroups. Similarly, leukemic cell lines share a similar gene expression pattern as their primary counterpart, indicating that they represent good models to study the respective leukemia (Fine et al., 2004, Andersson et al., 2005). Analysis of genome-wide copy number aberrations in ALL Single nucleotide polymorphism (SNP) arrays are most effective in detecting loss or gain of genetic material on a genome-wide scale and are powerful means to identify genetic alterations in ALL. This enabled the detection of ALL-specific deletions affecting genes important for lymphoid differentiation (EBF1, PAX5, IKZF1-3, TCF3, LEF1, RAG1-2), lymphoid signaling (BTLA), genetic mismatch repair (MSH6, MSH2), tumor suppression (PTEN, RB1) and drug resistance (NR3C1) (Mullighan et al., 2007, Yang et al., 2012, Mullighan, 2012, Kuster et al.
, 2011). Furthermore, analyzing patterns of genetic losses and gains from matched diagnosis and relapse samples by SNP array revealed the clonal origins of relapses of ALL (Fig. 2).
Thereby, 52% of relapses were shown to have a similar genetic makeup as the diagnostic leukemia, but did arise from clonal evolution of an ancestral prediagnosis clone, while 34% are a direct product of clonal evolution from the dominant diagnosis clone. Only 8% are a reemerged clone identical to the diagnosis clone and 18 6% of all relapse clones evolved completely independently from the clone present at primary diagnosis of ALL (Mullighan et al., 2008, Mullighan and Downing, 2009). Figure 2: Clonal evolution of relapsed ALL. 52% of relapse clones arise either from a common ancestral clone or by acquiring additional genetic aberrations and since have a clear relationship to the ancestral clone. 34% of relapse clones are directly derived from the leukemia clone present at diagnosis of ALL. 8% of relapses are caused by a clone that is genetically identical to the one present at diagnosis.
Only 6% of relapses emerge from a clone that is genetically distinct and evolved independently from the clone present at diagnosis. Adapted from (Mullighan et al., 2008). ETV6/RUNX1-positive B cell precursor ALL The ETV6/RUNX1 (E/R) fusion, t(12;21)(p13q22), defines one of the largest subgroups of BCP-ALL cases and is the most frequently occurring chromosomal rearrangement in pediatric cancer (Pui et al.
, 2011). E/R is nearly exclusively associated with pediatric ALL (Speck and Gilliland, 2002). While associated with an overall favorable prognosis, up to 20% of the affected children experience predominantly late relapses (Pui et al., 2011, Kuster et al., 2011, Conter et al.
, 2010, Borkhardt et al., 1997 Jul 15, Loh et al., 2006) that can pose a significant challenge to treatment, since they are associated with drug resistance and poor outcome (Pui et al., 2011, Kuster et al.
, 2011, Malempati et al., 2007). 19 The ETV6/RUNX1 fusion There are several chromosomal translocations that involve the ETV6 or RUNX1 genes. However, E/R is the only one with RUNX1 as a fusion partner that has been identified in ALL.
The role of E/R is largely defined by the properties of its fusion partners ETV6 and RUNX1. ETV6 (ets variant 6) (also known as TEL) is a ~300 kb gene consisting of 8 exons and is a member of the E-twenty-six (ETS) transcription factor family. ETS factors play key roles in cell proliferation, cell cycle, cell migration, differentiation and apoptosis and are frequently associated with cancer through gene-fusions (Mikhail et al., 2006, Baens et al.
, 1996, Wang et al., 1998). The ETV6 protein is comprised of 2 major domains. The N-terminal helix-loop-helix (HLH) domain is responsible for dimerization. The C-terminal ETS domain allows binding to consensus ETS DNAbinding sites.
Commonly, ETV6 is described as a repressor of target genes and mediates its activity through binding of the nuclear receptor co-repressor (N-Cor), histone deacetylases (HDAC) and mSin3A. While ETV6 expression can be found in wide variety of tissues, genetic abnormalities of ETV6 mainly occur in hematopoietic malignancies of lymphoid and myeloid origin and congenital fibrosarcoma (Zelent et al., 2004). Studies in mice have revealed that Etv6 is important for yolk sac angiogenesis during embryonic development, for the establishment of hematopoietic lineages in the bone marrow and is a key component for HSC survival. Etv6-/- embryos die at E11.
5 due to failure in maintaining the developing vascular system of the yolk sac and additionally, in specific regions of the embryo apoptosis is occurring (Wang et al., 1998). Loss of 20 Etv6 function leads to a depletion of HSCs in mouse adult bone marrow, demonstrating that Etv6 is regulating HSC survival (Hock et al., 2004).
Runt-related transcription factor 1 (RUNX1), also called AML1, is a ~250kb gene, with 12 exons, located at chromosome 21 band q22 and is frequently translocated in hematopoietic malignancies. RUNX1 belongs to the family of core-binding factors (CBFs), which consists of the 3 interchangeable CBF? subunits (RUNX1, RUNX2 and RUNX3) and the CBF? subunit that increases RUNX affinity for DNA-binding (Meyers et al., 1993, Mikhail et al., 2006). All RUNX proteins have a similar 2 domain structure: the Runt homology domain (RHD) for binding to specific DNA motifs and the C-terminally located transactivation domain (TD) for protein-protein interactions (Mikhail et al., 2006).
RUNX1 expression is found in most types of hematopoietic cells, including B lymphocytes (Zelent et al., 2004). RUNX1 has two separate promoters, from where the many alternatively spliced RUNX1 transcripts are generated (Mikhail et al.
, 2006). RUNX1 functions mainly as a transcriptional modulator that recruits cofactors to form a nuclear protein complex (Mikhail et al., 2006). Transactivating RUNX1 activity is mediated by p300 and CREB-binding protein (CBP), two coactivators that boost transcriptional activity through histone acetyl transferase (HAT) activity.
Repression of RUNX1 target genes is, for instance, mediated by mSin3A, which recruits histone deacetylases (HDAC) leading to a compact chromatin structure (Mikhail et al., 2006). 21 Figure 3: Schematic representation of full-length ETV6, RUNX1 and ETV6/RUNX1 (E/R) proteins. For ETV6, the pointed domain (PD), the central repression domain and the (ETS) DNA binding domain are depicted. For RUNX1, the Runt DNA binding domain, the mSin3A interaction domain (SID), the p300 interaction domain (p300ID), the transcriptional activation domain and the terminal VWRPY motif are shown.
Fusion sites for ETV6 and RUNX1 are indicated by arrows. Numbers depict amino acids bordering key functional domains, where the number 1 is the first methionine. Adapted from (Zelent et al., 2004). ETV6/RUNX1 is the product of the t(12,21)(p13q22) translocation (Fig. 3), a result of the in frame fusion of the 5’ region of ETV6 with RUNX1 that retains nearly all RUNX1 functional regions. The breakpoints of the E/R-fusion are located in intron 5 of ETV6 and in intron 1 or, occasionally in intron 2 of RUNX1.
The E/R-fusion is under the transcriptional control of the ETV6 promoter (Fig. 3). Of ETV6, the fusion retains only the central repression domain and the helix-loop-helix (HLH) dimerization domain, which is responsible for activating transcriptional repressors. Of RUNX1 the following domains have been shown to be necessary for the full function of the E/R fusion protein: the mSin3A interaction domain (SID), the p300 interaction domain and the C-terminal VWRPY region of RUNX1 that is responsible for interacting with corepressors (Fig.
3) (Morrow et al., 2007, Zelent et al., 2004). 22 Initial studies described that that the activity of reporter constructs, under the regulatory control of specific hematopoietic genes, is suppressed by expression of the E/R protein and that E/R-expression specifically counteracts RUNX1 driven gene activation (Hiebert et al., 1996, Fenrick et al., 1999). E/R can form a very stable repressor complex by binding N-CoR and mSin3A corepressors through the ETV6 oligomerization moiety. Since E/R binds RUNX1 target sequences, it can constitutively repress RUNX1 target genes in an HDAC dependent fashion (Fig.
4) (Zelent et al., 2004). In addition, the repressive effects mediated by E/R were abrogated by the HDAC inhibitor trichostatin A (Fenrick et al., 1999).
Figure 4: Proposed model for the regulatory function of RUNX1 and ETV6/RUNX1. (a) RUNX1 (AML1) represses target genes by recruitment of mSin3A and HDACs or activates target gene function by recruitment of p300. (b)The ETV6 part of E/R (TEL/AML1) facilitates dimerization, binding of N-Cor and mSin3A.
This allows formation of a stable, constitutive repressor complex that is unresponsive to the regulatory mechanisms shown in (a). Adapted from (Zelent et al., 2004). 23 However, recent studies suggest a more complex mode for the regulatory functions of E/R. One study utilizing expression microarrays in E/R expressing human fibroblast cell lines confirmed the repression of RUNX1 activated genes by E/R, but also demonstrated that E/R induces RUNX1 repressed genes, suggesting an activating function for the E/R fusion protein (Wotton et al., 2008).
Further studies support the notion that E/R is also an activator of gene transcription: one study demonstrated that expression of E/R in BaF3, a B lymphoid murine cell line, increases expression of heat shock protein 90 (HSP90) and survivin (Diakos et al., 2007). In leukemia, expression of the erythropoietin receptor (EPOR) is upregulated by E/R and E/R directly binds to the EPOR promoter region (Inthal et al., 2008, Torrano et al., 2011). Further, Kaindl et al. observed that E/R directly induces MDM2 activity by binding to promoter-inherent RUNX1-motifs (Kaindl et al., 2014).
Modeling E/R-positive ALL in mice has yielded further insights into leukemia development and biology. Evidence has emerged that E/R leads to premalignant activity, when expressed in murine fetal liver hematopoietic stem cells. It has been suggested that E/R blocks further differentiation among the lymphoid line, stopping development before the lymphoid primed multipotent progenitor (LMPP) or CLP stage. Consequently, this leads to an abnormal expansion of HSCs and renders them prone to malignant transformation, hence creating a pool of preleukemic cells (Schindler et al., 2009, Morrow et al.
, 2004, Tsuzuki et al., 2004). These findings were corroborated by a zebrafish study that could only demonstrate the accumulation of immature lymphocytes with low frequency and long latency , when E/R was expressed in HSC , suggesting that for leukemogenesis, E/R has to be expressed in cells more primitive than the CLP stage (Sabaawy et al.
, 2006). 24 While highly valuable, none of the mentioned models could reconstruct the full phenotype of E/R-positive BCP-ALL, suggesting that further models are required that incorporate additional oncogenic hits. For instance, several studies have indicated that forward mutagenesis, using the sleeping beauty transposon system, or loss of genes, frequently associated with E/R-positive ALL, are needed as additional hits to induce leukemia in mice (van der Weyden et al., 2011, Schindler et al.
, 2009, Bernardin et al., 2002). 25 II. Genetic alterations in glucocorticoid signaling pathway components are associated with adverse prognosis in children with relapsed ETV6/RUNX1-positive acute lymphoblastic leukemia Project background E/R-positive leukemia is associated with low risk features and an overall fast response to the currently used treatment regimens (Pui et al., 2011).
Despite the overall favorable prognostic outlook, up to 20% of cases treated with BFM-based protocols suffer a relapse (Seeger et al., 1999). Disease recurrence is more difficult to treat and hence responsible for dismal outcome in a notable proportion of patients (Kuster et al., 2011). A factor implicated in disease recurrence and drug resistance is an impaired glucocorticoid pathway (Kuster et al., 2011). Glucocorticoids and glucocorticoid-mediated signaling Synthetic glucocorticoids (GCs) such as prednisolone (PRED) or dexamethasone (DEX) are essential components in all childhood ALL treatment protocols (Pui and Evans, 2006, Inaba and Pui, 2010).
For instance, in Berlin-Frankfurt-Münster (BFM) based protocols, with the exception of one single dose of intrathecal methotrexate, PRED is the only agent administered for one week prior to induction treatment (Schrappe et al., 2012). Based on a bad response to this one week mono treatment with PRED, patients are designated prednisolone poor responders (PPR) and are 26 therefore allocated to high risk treatment, due to their dismal treatment outcome in previous studies (Schrappe et al., 2000, Moricke et al., 2008). However, in relapse treatment protocols multidrug chemotherapy is immediately applied and therefore it is impossible to single out the effects of PRED. GCs assert their function by binding to the high-affinity glucocorticoid receptor (GR), the gene product of the NR3C1 gene, leading to a combination of non-genomic and genomic effects. Non-genomic effects include the initiation of the apoptotic cascade by mitochondrial translocation upon binding of GCs to membrane-associated GR (Kfir-Erenfeld et al.
, 2010). The more important genomic effects are triggered by translocation of the activated GR dimers and monomers from the cytoplasm to the nucleus. Here, homo-dimers repress or activate expression of genes by binding to glucocorticoid response elements (GRE) on the DNA, while monomers interact with a number of transcription factors (Inaba and Pui, 2010, Schmidt et al., 2006b, Renner et al., 2003). In the nucleus, apoptosis can be induced by regulating typical apoptosis genes, such as Bcl-2 family member genes (Schmidt et al.
, 2004, Ploner et al., 2008). Two pro-apoptotic Bcl-2 family genes have recently been demonstrated to be especially important for GC-induced apoptosis in children with systemic GC treatment (Ploner et al., 2008): induction of BMF (Bcl2-modifying factor) and BIM (Bcl2interacting mediator of cell death, BCL2L11).
While alterations in genes involved in GC signaling, primarily in the form of genetic deletions, are involved in the development of GC resistance in ALL (Schmidt et al., 2004), primary GC-resistant leukemia rarely have shown genetic aberrations of NR3C1 (Mullighan et al., 2007, Irving et al., 2005a, Zhang et al.
, 2011). At the receptor level, resistance to GCs has been linked with a lack of GR autoinduction or defective GR expression (Schmidt et 27 al., 2004, Schwartz et al.
, 2010). Thus, the molecular mechanisms that lead to GC resistance remain poorly understood. Somatic aberrations affecting the glucocorticoid pathway in ETV6/RUNX1