Emerging molecular subtypes and therapeutic targets in B-cell precursor acute lymphoblastic leukemia

B-cell precursor acute lymphoblastic leukemia (BCP-ALL) is characterized by genetic alterations with high heterogeneity. Precise subtypes with distinct genomic and/or gene expression patterns have been recently revealed using high-throughput sequencing technology. Most of these profiles are associated with recurrent non-overlapping rearrangements or hotspot point mutations that are analogous to the established subtypes, such as DUX4 rearrangements, MEF2D rearrangements, ZNF384/ZNF362 rearrangements, NUTM1 rearrangements, BCL2/MYC and/or BCL6 rearrangements, ETV6-RUNX1-like gene expression, PAX5alt (diverse PAX5 alterations, including rearrangements, intragenic amplifications, or mutations), and hotspot mutations PAX5 (p.Pro80Arg) with biallelic PAX5 alterations, IKZF1 (p.Asn159Tyr), and ZEB2 (p.His1038Arg). These molecular subtypes could be classified by gene expression patterns with RNA-seq technology. Refined molecular classification greatly improved the treatment strategy. Multiagent therapy regimens, including target inhibitors (e.g., imatinib), immunomodulators, monoclonal antibodies, and chimeric antigen receptor T-cell (CAR-T) therapy, are transforming the clinical practice from chemotherapy drugs to personalized medicine in the field of risk-directed disease management. We provide an update on our knowledge of emerging molecular subtypes and therapeutic targets in BCP-ALL.

The well-established molecular subtypes of BCP-ALL have been mostly integrated into the guidance of therapy options (e.g., dose and times of chemotherapy and types of targeted drugs) [13]. Their integration has significantly improved the long-term survival of both adult and children patients. However, subtypes with poor/intermediate prognosis, including BCR-ABL1, KMT2A fusions, TCF3-PBX1, HLF fusions, and low hypodiploid, are needed in intensive chemotherapy [5]. BCR-ABL1 and KMT2A fusions not only occur in BCP-ALL but also in other types of acute or chronic leukemia, including acute myeloid leukemia (AML), mixed phenotype acute leukemia (MPAL), and chronic myeloid leukemia [61][62][63]. A potential supposition is that the fusion of genes may occur at different stages of hematopoietic stem cell development.
DUX4, located within a D4Z4 repeat array in the subtelomeric region of chromosome 4, is a key transcription factor regulating embryonic development and is not expressed in normal B-cells. DUX4 can bind a large percentage of activated genes in early developing embryos and improve the accessibility of genes as early as the 2-to 4-cell stages [20,21,69,70]. Intragenic deletions of ERG were previously reported in approximately 5% of children with BCP-ALLs, but they were designated as the biological feature in DUX4-r cases [5,21]. Functional studies show that DUX4 rearrangements, as an early initiating event in leukemogenesis, can bind to an intragenic region of ERG and cause the overexpression of ERGalt with a noncanonical first exon and transcript. Aberrant ERGalt preserves the DNA binding and transactivating domains and encodes a truncated C-terminal ERG protein. The transcriptional activity of wild-type ERG is inhibited by the ERGalt protein-transforming leukemic blast cells [5,20,21,71]. Interestingly, despite the presence of approximately 40%-50% of genetic alterations in IKZF1 deletions related to poor outcomes in other subtypes with BCP-ALL [72], DUX4-r BCP-ALL can achieve excellent outcome [5,12,17,18,21]. Recent studies have reported that IGH-DUX4 translocation occurs on the silenced IGH allele, thereby reducing the oncogenic stress of DUX4's high-level expression. The ERG deletions have a positive impact on the prognosis (ERG deletion positive/ negative: five-year EFS 93%/68%, P = 0.022; five-year OS 97%/75%, P = 0.029) [73,74].
MEF2D is a transcription factor that can specifically bind to the myocyte-specific enhancer factor 2 (MEF2) element 5′-YTA[AT](4)TAR-3′. It belongs to the MEF2 gene family that contributes to the differentiation of muscle and neural cells, cardiac morphogenesis, formation of blood vessels, growth factor responsiveness, survival of neuronal cells, and acute leukemia [75,76]. As a member of the MEF2 gene family, MEF2C is an activated oncogene in early T-cell precursor ALL, a subtype of high risk Tlineage ALL [12,[77][78][79][80][81][82]. In 2005, MEF2D-DAZAP1 was reported in the TS-2 cell line, as established from a three-year-old girl with ALL; it was found to contain t(1;19)(q23;p13.3) but was lacking the TCF3-PBX1 fusion [83]. In 2016, multiple research groups who used RNA-seq recognized simultaneously the MEF2D fusion in a subgroup of patients with BCP-ALL [16,18,20,64]. These studies determined that MEF2D fusions can retain the MADS box domain required to mediate DNA binding with enhanced MEF2D transcriptional activity. The heterogeneous nuclear ribonucleoproteins, including HNRNPUL1, HNRNPH1, and HNRNPM, are involved in MEF2D-r ALL [8,9,15]. These proteins can bind RNA and are associated with pre-mRNA processing in the nucleus. Histone deacetylase 9 (HDAC9), a target gene of MEF2D, is significantly upregulated in MEF2D-r patients, providing an optional therapeutic strategy by using histone deacetylase inhibitors, such as panobinostat [5,16,18]. Staurosporine and venetoclax have recently been reported to be effective in inducing the caspase-dependent proteolysis of MEF2D-fusion proteins and apoptosis in MEF2Dfusion + ALL cells [84].
ZNF384, also called CIZ or NMP4, can encode a C2H2type zinc finger protein, whose function remains largely elusive, although it may function as a transcription factor. ZNF384 fusions are often diagnosed as BCP-ALLs with an expression of cell surface markers of myeloid lineage (CD13 and CD33) or as B/myeloid (B/M) MPAL comprising approximately 50% of B/M MPAL [63]. In 2002, seven cases with ZNF384-r acute leukemia and recurrent EWSR1-ZNF384 and TAF15-ZNF384 were reported [85]. On the basis of large-cohort RNA-seq data analysis, ZNF384-positive patients with BCP-ALL were identified as a prognostic subtype with distinct gene expression features [8,9,[16][17][18]20,22,65,66,86,87]. All ZNF384 fusions keep their entire coding region. Cell apoptotic response, MAPK signaling, and JAK-STAT signaling pathways are significantly upregulated in this subtype [8,22]. Cardiotrophin-like cytokine factor 1 (CLCF1) is one of the most upregulated genes in ZNF384-r BCP-ALL that can bind to CRLF1 to form a compound cytokine, thereby ultimately activating the JAK-STAT signaling pathway and B-cell proliferation in vivo [22,88]. In addition, up to 60% of patients with ZNF384 fusions also show alterations in their signaling molecules, such as NRAS and FLT3, and 40% have epigenetic mutations, particularly SETD1B, CREBBP, and EZH2.
"Like" or "phenocopy" subtypes in BCP-ALL Somatic mutations in the coding region and gene expression profile can help us recognize "like" or "phenocopy" subtypes that share similar gene expression features but lack consistent biomarkers. In recent years, several multi-omics and large-cohort studies have identified such subtypes, including Ph-like, ETV6-RUNX1-like, KMT2A-like, and ZNF384-like [8,9]. Integrated datasets showed that Ph-like and ETV6-RUNX1-like account for approximately 6%-15% and 2%-3% of children patients and 20%-25% and < 1% of adult patients with BCP-ALL. However, both KMT2A-like and ZNF384-like are rare (< 1%). Ph-like and ETV6-RUNX1-like are associated with poor and intermediate outcomes. The prognoses of KMT2A-like and ZNF384-like remain unclear because of limitation of sample size [8,9,12].
The Ph-like subtype can successfully exemplify significant improvements in prognosis by seeking the phenocopied events. However, extra effort is required to identify other possible phenocopied subtypes (e.g., HLF fusion-like and hypodiploid-like subtypes) [8]. Moreover, the natural history and cell-of-origin of these "like" or "phenocopy" subtypes are still largely elusive. These gaps may prevent researchers from understanding the potential prognostic factors. A comprehensive comparison of the subtypes sharing similar gene expression profiles in many other aspects, such as cancer cell population, inherited or de novo variants within a noncoding region, aberrant splicing, bacterial and viral infections, chromatin accessibility, and/or epigenetics alterations, may provide more hints [27,87,[94][95][96][97][98][99][100][101][102][103][104][105].
Point mutations of PAX5 P80R in the DNA binding can affect the ability of PAX5 to bind DNAs and regulate expressions. These mutations represent the first molecular subtype defined on the basis of homogeneous hotspot mutations and gene expression profiles [8,9,65,86,108,109]. Most patients with PAX5 P80R presented distinct gene expression profiles and uniform genetic alterations. PAX5 P80R patients can promote the biallelic alteration of PAX5 in vivo, and this alteration frequently coexists with the hemizygous loss of PAX5. This scenario results in a higher mutation allele frequency compared with that of other PAX5 point mutations [9]. Hemizygous PAX5 P80R without the deletion of the wildtype PAX5 allele may present gene expression features in other subtypes, such as the Ph-like and PAX5alt. Activating mutations in signaling (Ras and JAK/STAT pathways), including PTPN11 and IL7R, and inactivating mutations in epigenetic factor SETD2 are the most common genetic alterations coexisting with PAX5 P80R [8,9,65,86].
PAX5-driven subtypes showed high heterogeneity in genetics, such as diverse patterning genes involved in PAX5 fusions, and sequence mutations scattered on the coding region of PAX5. This characteristic may further require diverse therapy agents, including a combination of chemotherapy and multi-inhibitors, and possibly immunotherapy. Additionally, different prognoses between adult and childhood PAX5alt indicates the presence of additional age-dependent factors that may affect the vitality, aggressiveness, and drug responsiveness of leukemic cells in this BCP-ALL subtype [9].

Other gene fusions and point mutations
Rearrangements of BCL2/MYC and/or BCL6 Translocations t(14;18)(q32;q21), t(8;14)(q24;q32), and t(3;14)(q27;q32) that result in BCL2-IGH, MYC-IGH, and/or BCL6-IGH fusions occur in approximately 2% of adults (mostly > 30 years old) with BCP-ALL and are less common in children. These translocations are associated with consistently poor responses in early treatment [9]. Low levels of circulating t(14;18)-positive cells can be found in approximately 50%-70% of healthy individuals but never develop into a disease [118][119][120][121]. Chromosomal alterations in BCL2, MYC, and/or BCL6 have been frequently reported in chronic lymphocytic leukemia and double-or triple-hit lymphoma, which are rare in ALL with a B-cell-precursor immunophenotype [16,[121][122][123][124][125][126][127][128][129][130][131][132][133]. The promoter regions of BCL2, MYC, and BCL6 are usually affected by IGH translocations, causing an overexpression of the rearranged allele compared with the germline allele. This characteristic can dramatically activate the proliferation of leukemogenic blast cells [123]. At present, patients with the abovementioned alterations cannot be well cured because of the development of chemotherapy resistance [9,121]. This limitation poses a great challenge to patient treatment. The inhibition of cyclin-dependent kinase 7 (CDK7) has been reported as an optional target for reducing the resistance of BCL-2 in B-cell lymphoma models [133]. The developed therapeutic agents in B-cell lymphoma may also be effective in BCP-ALL. Additional functional assays and clinical tests are urgently needed to improve the survival rate in the BCL2/ MYC and/or BCL6-r BCP-ALL.

ZEB2 p.His1038Arg (H1038R) and IGH-CEBPE gene fusion
Zinc finger E-box binding homeobox 2 (ZEB2), a member of the Zfh1 family of 2-handed zinc finger/homeodomain proteins, is a nuclear protein that can bind DNA and repress its transcriptional activity; this protein interacts with activated SMAD, a DNA binding protein, and recognizes an 8-bp palindromic sequence (GTCTAGAC) called the Smad binding element [134]. CCAAT enhancer binding protein epsilon (CEPBE) is a bZIP transcription factor that binds to certain DNA regulatory regions by means of homodimer formation. In a recent study, a hotspot mutation of ZEB2 p.His1038Arg (H1038R) and translocation t(14;14) (q11;q32), which results in IGH-CEBPE with a truncation of the 3′ UTR, is defined as a rare subtype ( < 1%) sharing similar gene expression features. The rs2239630 G > A at the promoter of CEBPE is associated with the IGH-CEBPE translocated ALL with increased CEBPE expression [135]. More than 50% of patients in this subtype also have NRAS sequence mutations. The leukemic oncogene LMO1 is significantly upregulated in this subtype and is an important component of a transcriptional complex that includes TAL1, TCF12/ HEB, TCF3/E2A, MYB, RUNX1, GATA3, and LDB1. This transcription complex can form a positive interconnected autoregulatory circuit that impacts the transformation in approximately 60% of patients with T-cell ALL (T-ALL) [77,136]. In addition, SMAD1 and BMP2 are significantly downregulated in this subtype; these genes can regulate the signals of the bone morphogenetic proteins and are involved in a range of biological activities, including morphogenesis, cell growth, apoptosis, development, and immune responses [137][138][139][140]. Preliminarily, ZEB2 mutation is associated with poor event-free survival and high relapse in patients with BCP-ALL [141]. The clinical implications of ZEB2 p.His1038Arg (H1038R) and IGH-CEBPE gene fusion subtype are still largely unclear. Much larger sample sizes and functional tests are needed for this BCP-ALL subtype. IKZF1 p.Asn159Tyr (N159Y) IKZF1, also known as IKAROS, is a critical transcription factor related to the differentiation and maturation of the Bcell precursor. Somatic alterations of IKZF1 are a hallmark of high-risk BCP-ALL with poor response to therapy [72,109,142]. Point mutations of IKZF1 p.Asn159Tyr (N159Y) were recently recognized as a rare subtype (< 1%) in BCP-ALL. In light of the limitation of sample size, the outcomes of IKZF1 N159Y are still undermined. IKZF1 N159Y, which is located in the DNA binding domain, may impact its capability for DNA binding and gene transcription regulation. Sequence mutations of KRAS are recurrent in patients with IKZF1 N159Y but lack extra copy number alterations. The gene expression data showed that the transcriptional coactivator YAP1 is significantly upregulated, which can drive KRAS-induced transformation through rescued cell viability in KRASdependent cells [143]. The chromatin remodeling SALL1 and the signaling factor ARHGEF28 are also significantly upregulated in IKZF1 N159Y-positive patients. Meanwhile, the B-cell receptor signaling and JAK-STAT signaling pathways (e.g., FLT3, FLT4, and STAT5) are down-regulated in patients with IKZF1 N159Y. These two pathways are commonly activated in other BCP-ALLs [8,9,[144][145][146]. At present, the public data show that IKZF1 N159Y-positive cases are still below 10. One case of IKZF1 N159T was reported in the chronic myelomonocytic leukemia cohort [147]. Surprisingly, de novo germline mutations, including IKZF1 N159S (n = 6) and IKZF1 N159T (n = 1), could exist at the same IKZF1 amino acid N159, which then would cause T-, B-, and myeloid cell-combined immunodeficiency, and a patient would develop T-ALL [148]. Pneumocystis jirovecii pneumonia is positive in all of patients with N159S or N159T mutations. A patient died at 2 years old without leukemia phenotype although the child received hematopoietic stem cell transplants [148]. Multi-center collaboration may accelerate the collection of patient data in this subtype for the further evaluation of its clinical significance (Table 1; Fig. 1A and 1B).

Diagnosis and molecular classification of BCP-ALL based on RNA-seq
Genetic variations with clinical significance in BCP-ALL are structurally heterogeneous [18]. Multiple highthroughput sequencing approaches are highly recommended, as they can be utilized to accurately recognize the prognostic factors (e.g., fusion genes, gene expressiondependent subtypes, small sequence variants, and genomic duplications and deletions) and determine the strategic therapy [149]. RNA-seq is a single and comprehensive platform for BCP-ALL diagnosis and genomic classification in the laboratory and clinical settings [9,43,[150][151][152].

New therapeutic targets and agents in BCP-ALL
Intensive chemotherapy and allogeneic hematopoietic cell transplantation were the core options of BCP-ALL treatment in the past [27]. Treatment toxicity with relapsed events induced by chemotherapy drugs has been one of the most critical concerns awaiting further resolution in BCP-ALL [13,27]. The advent of emerging inhibitors/antagonists and immunotherapeutic has launched a new era of target therapy in several molecular subtypes or unselected BCP-ALL patients ( Table 2). This ongoing transformation may continuously reduce the use of chemotherapy drugs and consequently achieve less treatment-induced resistance events [160]. Molecular target therapy and cellular immunotherapy in BCP-ALL are mainly dependent on the specific genetics and gene expression markers of the patients' leukemic cells in various molecular subtypes (e.g., BCR-ABL1, Ph-like, and KMT2A fusions) or the shared or unselected cell surface marker (e.g., CD19 and CD22) (Fig. 2). Several molecular subtypes in BCP-ALL, including R/R subtypes, have benefited from the new therapeutic targets and agents, although a large number of biomarkers are still rarely used as therapy targets.
Apart from the inhibitors/antagonists, immunotherapeutic agents, such as CAR-T and monoclonal antibodies, are another promising strategy for targeting specific cell surface markers overexpressed in leukemic cells. Immunotherapeutic agents can be used in the therapy of most children and adults with R/R BCP-ALL. The cell membrane antigens CD19, CD20, and CD22 are the three most promising targets of immunotherapeutic agents for BCP-ALL [167][168][169][170][171][172][173][174]. Other bispecific agents (e.g., Combobox for CD19/CD22) are becoming the nextgeneration CAR-T treatment options. As one of the hottest fields in BCP-ALL, a series of clinical trials of CAR-T for R/R BCP-ALL is ongoing (e.g., NCT03330691, NCT03937544, NCT00450944, NCT04012879, and NCT04094766). The immunotherapeutic agents inotuzumab and blinatumomab were approved by the United States Food and Drug Authority for the therapy of adults with R/R BCP-ALL, which can help to improve in the future the overall survival of both children and adults with BCP-ALL. In addition, the combination of the TKIs imatinib or dasatinib with multiagent chemotherapy is currently used in multiple clinical trials of R/R BCP-ALL. This treatment has markedly improved the outcome in BCP-ALL subset, as the five-year overall survival of patients has increased to 75% from less than 50% [12,175]. Controlling early death, reducing therapy-induced resistance mutations, and comprehensive clinical management of adult patients are the potential key issues of medical precision in BCP-ALL treatment (Table 2) [5,8,9,12,13,25,40,60,61,160,167,169,[176][177][178][179][180][181][182].

Conclusions
Through decades of collaboration, more than 95% of patients with BCP-ALL have been classified and labeled using detectable genetic alterations, including distinct translocations of chromosomes, aneuploidy, DNA copy number alterations, sequence mutations, and gene expression patterns [8][9][10][11][12]183]. Refined risk stratification and the approval of new drugs and agents, including target inhibitors, monoclonal antibodies, and immunomodulators, and CAR-T, have resulted in an excellent five-year event-free survival and a five-year overall survival in children with BCP-ALL. However, the development of resistance and early death during treatments continue to pose daunting challenges for some subtypes of BCP-ALL, including hypodiploid (< 44 chromosomes), HLFarranged, BCL2/MYC-r, BCR-ABL1/Ph-like, KMT2A-r, and MEF2D-r BCP-ALL. Additionally, single-cell-based traces of cancer cell populations, chromatin accessibility, epigenetic alterations, genome-wide germline mutations, functional non-coding and synonymous mutations, and other abnormalities at the non-genomic levels, such as protein and metabolic levels, are still poorly understood. Information on these aspects can help in drug discovery and improvement in the prognoses of leukemia patients, including BCP-ALLs [103][104][105][183][184][185][186][187][188]. Through the comprehensive identification of prognostic biomarkers and the development of new techniques in diagnosis and target treatment, we can further improve the survival time and life quality of patients with BCP-ALL.

Compliance with ethical guidelines
Jianfeng Li, Yuting Dai, Liang Wu, Ming Zhang, Wen Ouyang, Jinyan Huang, and Saijuan Chen declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. This manuscript is a review article and does not involve a research protocol requiring approval by the relevant institutional review board or ethics committee.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as 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 changes were made.
The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.