Structural modeling of JAK1 mutations in T-cell acute lymphoblastic leukemia reveals a second contact site between pseudokinase and kinase domains

With great interest we read the extensive review on JAK kinase targeting in hematologic malignancies by Springuel, Renauld and Knoops.1 It provides a clear overview of the plethora of cytokine receptor complexes and downstream signaling adapter molecules that are mutated in hematologic malignancies. In particular, Janus kinase family members play a clear role in both normal hematopoiesis and malignancies. Four JAK kinases (JAK1, JAK2, JAK3, and TYK2) transmit extracellular signals from ligand-activated cytokine receptors to downstream transcriptional effectors such as STATs to regulate gene transcription. Myeloid development is predominantly driven by various cytokine receptors that recruit JAK2, the likes of which then interacts with the intracellular domains of said receptors. In addition to activating mutations in receptors such as TPO-R and CSF3R, myeloid malignancies frequently have JAK2 mutations. Early B-cell development depends on signaling through the thymic stromal-derived lymphopoietin (TSLP) receptor, a heterodimeric receptor composed of a CRLF2 chain and an IL7Ra chain. The TSLP receptor recruits JAK2 and JAK1 upon ligand binding. In addition to mutations in the CRLF2 and IL7Ra chains and chromosomal rearrangements of JAK2, precursor B-cell acute lymphoblastic leukemia (BCP-ALL) frequently has activating JAK1 and JAK2 mutations.42 Early lymphoid and T-cell development rely on signaling through the IL7 receptor, which is a dimer composed of an IL7Ra chain and a common-γ chain. Upon binding IL7, this receptor recruits and activates both JAK1 and JAK3. Consistent with a pivotal role in normal T-cell development, mutations in IL7Ra, JAK1, and JAK3 (but not JAK2) are commonly found in T-ALL.752

Herein, we expand on the structural consequences of JAK mutations as briefly discussed by Springuel et al.1 The interaction between the N-lobes of the pseudokinase and kinase domains is believed to prevent aberrant (i.e., ligand-independent) kinase activity. This interaction predominantly relies on residues between the hinge region in the pseudokinase domain and the kinase domain loop (Figure 1). Thus, mutations in the N-lobes of either domain may directly affect the interaction between the two domains, thereby triggering a conformational change that induces ligand-independent activation. We screened our cohort of 146 pediatric T-ALL patients and identified five novel JAK1 mutations (plus one previously identified mutation, R724H85); the majority of these mutations are located in N-lobe interaction sites between the pseudokinase and kinase domains (Figure 1A). To better understand the impact of these mutations on the function of JAKs, we superimposed these JAK1 pseudokinase9 and kinase domain10 mutants on the TYK2 crystallographic structure11 (Figure 1B). This model is consistent with the interaction between the hinge region in the pseudokinase domain (residues V666 through K736) and the kinase domain loop (residues R893 through E904); this interaction is supported by strong salt bridges formed between the following four pairs of amino acid residues: E668-R893, D709-H1015, R724-E897, and K736-D899 (Figure 1B–D). Mutations in these residues, or nearby residues, were identified in our T-ALL cohort and include E668Q, R724H (corresponding to the R657Q mutation in JAK3) and T901G (Figure 1C–D). It is likely that these mutations disrupt the pseudokinase-kinase domain interaction, leading to constitutive JAK1 activation.

Figure 1.JAK mutations affect the pseudokinase-kinase domain interaction. (A) Schematic representation of the JAK1 and JAK3 proteins indicating the mutations found in our cohort of 146 pediatric T-ALL patients. The Four-point-one, Ezrin, Radixin, Moesin (FERM), Src homology-2 (SH2), pseudokinase and kinase domains are indicated. (B) Overlay of the JAK1 pseudokinase (blue) and kinase (orange) structures on the highly homologous TYK2 pseudokinase-kinase structure (grey). Indicated in green are two predicted contact sites between the pseudokinase and the kinase domains: the kinase domain loop (residues 893 through 904) interacts with the hinge region of the pseudokinase domain (residues 666 through 736); the helical domain (residues 564 through 571) in the C-lobe of the pseudokinase domains, which was not resolved in the crystal structure for TYK2, may create a second interaction site with the kinase domain. (C–F) Top view (C) and details of the kinase domain loop (residues 893 through 904) (D), bottom view (E) and details of the region of the C-helix (F) of the JAK1 complex. JAK1 mutated residues found in our cohort are shown in red.

A second mutation hotspot in JAK molecules involves residues that form the so-called F-F-V triad in the C-lobe of the pseudokinase domain. This conserved triad (F575-F636-V658 in JAK1 and F537-F595-V617 in JAK2) is believed to act as a structural switch that controls the catalytic activity of JAK kinases.9 Mutations in this triad, including the JAK1 V658F mutation in T-ALL6 (and the equivalent JAK2 V617F mutation in polycythemia vera patients), disrupt the structural switch, leading to constitutive kinase activity. In JAK3, the M511I mutation is prevalent in T-ALL1312 and is two amino acids away from residue F513, which corresponds to the F575 residue in JAK1. The L624YPILKV mutation is located within a loop that connects a β-strand to the N-terminus of the C-helix (Figure 1F). These inserted residues extend the loop upstream of the C-helix and may compromise the F-F-V triad, thereby leading to de-repression of the kinase domain (Figure 1C–F). Our model of the JAK1 pseudokinase-kinase complex predicts a second interaction site between the C-lobes of the pseudokinase and kinase domains; this interaction is formed by a helical domain (residues 564 through 571) in the pseudokinase domain, which was not resolved in the crystallographic structure of TYK2 (Figure 1B). As modeled in Figure 1B, this helical domain has significant van der Waals clashes with the kinase domain, indicating that the helical domain may actually fold into the catalytic pocket of the kinase domain, thereby repressing kinase activity. The helical domain lies just upstream of the F575 residue in the F-F-V triad. Our results support a model in which the triad-controlled position of the C-helix functions as a switch that can alter the interaction between the helical domain (residues 564 through 571) and the kinase domain.

A second point of discussion concerns the frequency of TYK2 mutations in T-ALL: Springuel et al.1 cited an incorrect frequency of TYK2 mutations in 21% of T-ALL patients based on findings by Sanda et al.,14 who reported five activating TYK2 mutations in four out of 17 T-ALL cell lines, but none in 45 primary T-ALL patient samples. In contrast to frequent mutations in JAK1, JAK2, and JAK3 in hematological malignancies,62 incidental mutations in TYK2 have been reported to date,1615 with only two out of 93 T-ALL cases reported by Kalender et al.17 Neither of these mutations revealed TYK2 autophosphorylation or increased transforming activity compared to wild-type TYK2 in Ba/F3 cells, so it remains unclear whether these are activating mutations or not. In support of other large sequencing screens,1813 we found no mutations in TYK2 in 146 pediatric T-ALL patient samples. Interestingly, the TYK2 mutations in four T-ALL cell lines identified by Sanda et al.14 are different from those in the same cell lines reported by Kalender et al.,17 who provide evidence that TYK2 mutations arise during culture over time. Therefore, it is likely that the TYK2 mutations acquired in T-ALL cell lines during culture do not represent mutations in T-ALL patients.

Despite a lack, or very low frequency, of activating TYK2 mutations in T-ALL, TYK2 activity could be an important factor. The survival of primary T-ALL cells seems to require TYK2 signaling, which activates STAT1 and BCL2.14 Therefore, TYK2-specific inhibitors may have therapeutic potential, as suggested by Sanda et al.14 In a commentary on that report, Fontan and Melnick19 proposed two possible mechanisms to explain TYK2 dependency in T-ALL: i) a somatic activating mutation occurs in TYK2 itself, or ii) a signaling pathway upstream of TYK2 becomes constitutively active.19 Since TYK2 mutations are so rare in T-ALL patients, the second explanation seems more likely. In general, given the high frequency of JAK1, JAK2, and JAK3 mutations, the very low frequency of TYK2 aberrations in the entire spectrum of hematologic malignancies is highly intriguing. Additional research into the role of TYK2 in normal hematopoiesis may reveal why it is rarely a target for mutagenesis during malignant transformation.