JAK2 Alterations in Acute Lymphoblastic Leukemia: Molecular Insights for Superior Precision Medicine Strategies

Abstract

Acute lymphoblastic leukemia (ALL) is the most common pediatric cancer, arising from immature lymphocytes that show uncontrolled proliferation and arrested differentiation. Genomic alterations affecting Janus kinase 2 (JAK2) correlate with some of the poorest outcomes within the Philadelphia-like subtype of ALL. Given the success of kinase inhibitors in the treatment of chronic myeloid leukemia, the discovery of activating JAK2 point mutations and JAK2 fusion genes in ALL, was a breakthrough for potential targeted therapies. However, the molecular mechanisms by which these alterations activate JAK2 and promote downstream signaling is poorly understood. Furthermore, as clinical data regarding the limitations of approved JAK inhibitors in myeloproliferative disorders matures, there is a growing awareness of the need for alternative precision medicine approaches for specific JAK2 lesions. This review focuses on the molecular mechanisms behind ALL-associated JAK2 mutations and JAK2 fusion genes, known and potential causes of JAK-inhibitor resistance, and how JAK2 alterations could be targeted using alternative and novel rationally designed therapies to guide precision medicine approaches for these high-risk subtypes of ALL.

Keywords: JAK2; Janus kinases; acute lymphoblastic leukemia; kinase inhibitor; leukemia; targeted therapy.

FIGURE 1

The CRLF2r/JAK-mutant and JAK2/EPORr subtypes of Ph-like ALL are associated with poor outcomes. Outcome analyses for different genomic subtypes of Ph-like ALL for all ages combined, probabilities of 5-years event-free survival (EFS) and overall survival (OS) are shown. There are significant differences in the 5-years EFS and OS of CRLF2r/JAK-mutant and JAK2/EPORr cases compared with other Ph-like ALL subtypes. Figure adapted from Roberts et al. (2014a).

FIGURE 2

WT JAK2 structure and function (A) Schematic representation of the JAK2 domain structure (NCBI reference sequence: NP_004963.1) encoded by the seven JAK homology (JH) domains. The FERM (4.1 protein, ezrin, radixin, moesin), SH2 (Src homology 2)-like (SH2L), pseudokinase (JH2) and kinase (JH1) domains are represented by purple, red, light blue, and dark blue respectively. Key residues for phosphorylation for positive (green) or negative (red) regulation are shown. Mutations commonly associated with ALL (black lines) and JAK2 fusion breakpoints (black arrows) are indicated. Adapted from Silvennoinen and Hubbard (2015b) (Silvennoinen and Hubbard, 2015a). (B) Schematic representation of JAK/STAT signaling pathway activation through JAK2. The JAK2 FERM and SH2L domains associate with the cytoplasmic juxtamembrane motifs of a cytokine receptor (grey) to recruit JAK2 to the cell membrane. The four domains of JAK2 are presented: FERM (green), SH2-like (orange), pseudokinase (JH2, purple), and kinase (JH1, blue). JAK2 is shown bound to ATP (black). The proposed model of JAK2 activation suggests that JAK2 exists in an equilibrium between inactive and partially active conformations. In the inactive conformation (left), the JAK2 kinase domain is inhibited by a JH2-mediated autoinhibitory interaction. In the partially active conformation (right), the JAK2 kinase domain is released from the JH2-mediated auto-inhibition and is available for some limited transphosphorylation. Cytokine (cyan) binding to their cognate receptor promotes receptor dimerisation, which facilitates JAK2 activation by transphosphorylation (arrows). JAK2 then auto-phosphorylates the cytoplasmic region of the receptor creating recruitment sites for cytoplasmic STATs (red). JAK2-mediated STAT phosphorylation facilitates STAT dimerisation. These STAT dimers are then translocated to the nucleus where they regulate gene expression by binding to promoters with STAT-binding sites. Adapted from Hubbard (2018) and “Cytokine Signaling through the JAK-STAT Pathway” (BioRender.com, 2021).

FIGURE 3

Active and inactive conformations of JAK2. (A) Co-crystal structure of the type-I JAK inhibitor, ruxolitinib, bound to the JAK2 kinase domain in the active conformation (PDB: 6VGL). Ruxolitinib (yellow) is presented in ball-and-stick representation with nitrogen atoms in blue. The JAK2 kinase domain is presented in ribbon representation with amino acid side chains shown for essential phosphotyrosine residues, JAK2 p. Y1007 and p. Y1008. The N-terminal lobe (residues 840–931), shown in cyan, comprises a 5-stranded antiparallel β-sheet and one α-helix (αC). The C-terminal lobe (residues 932–1,132), shown in green, comprises 8 α-helices, 3 3/10 helices, and 3 pairs of antiparallel β-strands. The glycine loop is colored in pink, the hinge region between the 2 lobes in peach, the catalytic loop in orange, the activation loop in red, DFG-motif in blue, and the insertion loop in grey. (B) Superimposition of the active (blue) and inactive (pink) conformations of the JAK2 kinase domain ATP-binding site within co-crystal structures bound to JAK inhibitors. Ruxolitinib (dark blue) and the type-II JAK inhibitor, BBT594 (dark pink), are bound to the active (PDB: 6VGL) and inactive (PDB: 3UGC) conformations respectively. JAK inhibitors are presented in ball-and-stick representations with oxygen atoms in red and nitrogen atoms in blue. The JAK2 activation loop is disordered in the inactive conformation. Structures were visualized with PyMOL 2.0 (Schrödinger, LCC). (C) 2D chemical structures of ruxolitinib (ChemSpider, CSID: 25027389) and BBT594 (ChemSpider, CSID: 34980928)

FIGURE 4

JAK2 mutations in hematological malignancies. (A) Model of JAK2 JH2-JH1 interface showing the positions of known activating JAK2 mutations. The JAK2 JH2-JH1 model was generated by Shan et al. (2014) using molecular dynamics simulations and annotated using ChimeraX-1.2.5 (University of California). The JH2 (pseudokinase domain) N-terminal (residues 536–629) and C-terminal (residues 630–839) lobe are colored in light blue and green respectively. The JH1 (kinase domain) N-terminal (residues 840–931) and C-terminal (932–1,132) lobes are colored in pink and purple respectively. Residues that when mutated are known to be activating are shown as red spheres (α carbon). Two critical inhibitory phosphorylation sites, pS523 and pY570, are encircled. Other key residues predicted to be involved in activating JAK2 mechanisms are colored in red and presented with amino acid side chains shown. Proposed interactions are represented by dashed lines. Figure adapted from Shan et al. (2014), Hammarén et al. (2019a), Leroy et al. (2016) and Lupardus et al. (2014). (B) Schematic representation of JAK/STAT signaling pathway activation through mutant-JAK2. CRLF2 (dark grey) heterodimerizes with IL-7Rα (light grey) to form the cytokine receptor for TSLP. The JAK2 FERM and SH2-like domains associate with the cytoplasmic juxtamembrane motifs of the receptor to recruit JAK2 to the cell membrane. The four domains of JAK2 are presented: FERM (green), SH2-like (orange), pseudokinase (JH2, purple), and kinase (JH1, blue). JAK2 is shown bound to ATP (black). The proposed model of mutant-JAK2 activation suggests that mutations such as JAK2 p. R683G (represented by a yellow sphere) disrupt the JH2-mediated autoinhibitory interaction with the kinase domain. This shifts the equilibrium of JAK2 from the inactive, auto-inhibited state towards the partially active state, supporting mutant-JAK2 dimerisation. Although mutant-JAK2 alone remains dependent on cytokine binding, additional mechanisms such as receptor overexpression may promote malignant transformation. Adapted from “Cytokine Signaling through the JAK-STAT Pathway” (BioRender.com, 2021).

FIGURE 5

JAK2 fusion proteins in ALL. (A) Schematic representation of a genomic rearrangement between JAK2 exon 15 and BCR exon 1 that produces the BCR-JAK2 fusion gene. BCR isoform 1 (encoded by BCR variant 1) contains the following domains: BCR coiled-coil (CC), serine/threonine kinase (S/T kinase), DH (Dbl homology), PH (pleckstrin homology), Cal-B (calcium-dependent lipid-binding) and Rac-GAP (Rac GTPase-activating protein) domains. The BCR DH and PH domains form the Rho-GEF domain (Rho guanine nucleotide exchange factor). JAK2 isoform A (encoded by JAK2 variant 1) contains FERM (4.1 protein, ezrin, radixin, moesin), SH2-like (SH2L, Src homology 2), pseudokinase (JH2) and kinase (JH1) domains. The BCR-JAK2 fusion protein retains the BCR CC and S/T kinase domains, three exons of the JAK2 pseudokinase domain and the full-length JAK2 kinase domain. BCR-JAK2 is predicted to homodimerise via its retained BCR CC motif. Domains encoded by the BCR, JAK2 and BCR-JAK2 transcripts were annotated using InterPro (EMBL-EBI, 2021) (Jones et al., 2014; Blum et al., 2021) and Maru amd Witte (1991). (B) Schematic representation of JAK/STAT signaling pathway activation through JAK2 fusions. All JAK2 fusions comprise of an N-terminal fusion partner (orange) and the full-length JAK2 kinase domain (JH1, blue). The full-length or truncated JAK2 pseudokinase domain (JH2, purple) may also be present or absent in different JAK2 fusions. The absence of the JAK2 FERM and SH2-like domains prevent JAK2 fusions from associating with the cytoplasmic juxtamembrane motifs of cytokine receptors (dark grey). JAK2 fusions are shown bound to ATP (black). The proposed model of JAK2 fusion activation suggests that oligomerization domains within the fusion partner may facilitate JAK2 fusion dimerisation and subsequent trans-phosphorylation, promoting malignant transformation. Adapted from “Cytokine Signaling through the JAK-STAT Pathway” (BioRender.com, 2021).