JAK inhibitors for the treatment of myeloproliferative neoplasms and other disorders

Abstract

JAK inhibitors have been developed following the discovery of the JAK2V617F in 2005 as the driver mutation of the majority of non- BCR-ABL1 myeloproliferative neoplasms (MPNs). Subsequently, the search for JAK2 inhibitors continued with the discovery that the other driver mutations ( CALR and MPL) also exhibited persistent JAK2 activation. Several type I ATP-competitive JAK inhibitors with different specificities were assessed in clinical trials and exhibited minimal hematologic toxicity. Interestingly, these JAK inhibitors display potent anti-inflammatory activity. Thus, JAK inhibitors targeting preferentially JAK1 and JAK3 have been developed to treat inflammation, autoimmune diseases, and graft-versus-host disease. Ten years after the beginning of clinical trials, only two drugs have been approved by the US Food and Drug Administration: one JAK2/JAK1 inhibitor (ruxolitinib) in intermediate-2 and high-risk myelofibrosis and hydroxyurea-resistant or -intolerant polycythemia vera and one JAK1/JAK3 inhibitor (tofacitinib) in methotrexate-resistant rheumatoid arthritis. The non-approved compounds exhibited many off-target effects leading to neurological and gastrointestinal toxicities, as seen in clinical trials for MPNs. Ruxolitinib is a well-tolerated drug with mostly anti-inflammatory properties. Despite a weak effect on the cause of the disease itself in MPNs, it improves the clinical state of patients and increases survival in myelofibrosis. This limited effect is related to the fact that ruxolitinib, like the other type I JAK2 inhibitors, inhibits equally mutated and wild-type JAK2 (JAK2WT) and also the JAK2 oncogenic activation. Thus, other approaches need to be developed and could be based on either (1) the development of new inhibitors specifically targeting JAK2V617F or (2) the combination of the actual JAK2 inhibitors with other therapies, in particular with molecules targeting pathways downstream of JAK2 activation or the stability of JAK2 molecule. In contrast, the strong anti-inflammatory effects of the JAK inhibitors appear as a very promising therapeutic approach for many inflammatory and auto-immune diseases.

Introduction

Janus kinases (JAKs) play a central role in the regulation of hematopoiesis as being mandatory for signaling by receptors for hematopoietic/immunological cytokines ¹. They control survival, proliferation, and differentiation of hematopoietic cells as well as the function of mature cells by binding to hematopoietic type I and type II cytokine receptors, which are devoid of catalytic activity ( Figure 1). JAKs pre-associated to these receptors to form a functional signaling complex ¹. Cytokine binding induces or re-orients receptor dimerization, such that JAK kinase domains face each other in a productive conformation for transactivation and phosphorylation of the cytokine receptor cytoplasmic tails ¹. The latter and JAKs themselves become scaffolds for signaling molecules, particularly for the members of the signal transducer and activator of transcription (STAT) family, which in turn are phosphorylated and homo/hetero-dimerize before translocating to the nucleus. JAK activation also initiates activation of mitogen-activated protein kinase (MAPK), phosphatidylinositol-3′-kinase (PI3K), and AKT/mammalian target of rapamycin (mTOR) ( Figure 2) ².

Figure 1. Janus kinases (JAKs) and the cytokine receptor superfamily.

Schematic representation of type I and type II cytokine receptor subfamilies based on the extracellular domain sequence homologies. The different JAKs (JAK1, JAK2, JAK3, and TYK2) are employed by each class of receptors, as indicated. Type I receptors can form homodimers (α/α), heterodimers (α/β), or oligomers (gp130/α/gp130);(α/β/γ), although the α chain is mainly responsible for cytokine binding. Cytokine receptor complexes composed of two or more different chains activate at least two different JAKs, while single-chain receptors such as homodimeric receptors activate JAK2 only (although TpoR/MPL and G-CSFR/CSF3R can also use TYK2 and JAK1, respectively). The myelopoiesis-related cytokine receptors are denoted in red, and the lymphopoiesis-related cytokines receptors are denoted in green.

Figure 2. Principal signaling pathways activated by homodimeric cytokine receptors.

Cytokine binding to the extracellular domain of receptors induces conformation changes that enable cross-phosphorylation of the appended Janus kinases (JAKs), which then can activate each other. As a result, JAK molecules phosphorylate tyrosine residues on the intracellular part of the receptor, which then can serve as docking sites for SH2 domain containing signaling molecules such as signal transducer and activator of transcription (STAT) but also proteins from the phosphatidylinositol-3′-kinase (PI3K) and mitogen-activated protein kinase (MAPK) pathways.

The JAK/STAT pathway is frequently dysregulated in malignant diseases and in disorders with an abnormal immunological response ³.

The discovery that classic BCR-ABL1–negative myeloproliferative neoplasms (MPNs) are constantly associated with abnormal JAK2 activation due to different mutations, has paved the way for the development of JAK inhibitors in the therapy of these disorders as well as of other diseases with either genetic alterations in the JAK pathway or JAK-induced activation by autocrine and paracrine cytokine loops ^{4, 5}. Here, we focus on the role of JAKs as potential therapeutic targets, the development of JAK inhibitors and their limitations, and potential new strategies targeting the JAKs.

Janus kinases

The human genome codes for four JAKs: JAK1, JAK2, JAK3, and TYK2 ¹. Like the god Janus, who has two faces and opens doors, JAKs possess two kinase domains—one catalytically active domain at the C-terminus and an upstream pseudokinase domain that binds ATP—but does not phosphorylate substrates (except weakly itself). At the N-terminus, JAKs possess a FERM (band four-point-one, ezrin, radixin, moesin)-like domain and an Src homology 2 (SH2)-like domain ( Figure 3A). The non-covalent attachment of JAKs to cytokine receptor tails is specific and depends on the FERM domain and on the juxtamembrane receptor sequence containing a proline-rich Box 1 and a motif denoted Box 2 with both hydrophobic and negatively charged residues as well as the sequences stretching from Box 1 to Box 2 ^{6, 7}. JAK2 associates with the three homodimeric receptors: erythropoietin receptor (EPOR), thrombopoietin receptor (MPL/TPOR), and the granulocyte colony-stimulating factor receptor (G-CSFR/CSF3R). JAK1, JAK2, and TYK2 can also associate with heterodimeric or trimeric receptors, and JAK3 only with receptors containing the γ-chain with solely JAK1 as partner ( Figure 1) ¹. The precise structure of receptor/JAK complexes remains unknown. Recently, the structures of the kinase and pseudokinase domains of JAK1, JAK2, and TYK2 were obtained and represent a major asset for the identification of novel (allosteric) inhibitors (see below) ^{8– 10}. Interestingly, JAKs also play a role of chaperones for traffic and stability at the cell surface of several cytokine receptors ¹¹.

Figure 3. The kinase domain of Janus kinase 2 (JAK2) has been commonly targeted by the current inhibitors.

( A) Secondary structure of JAK molecules. They are composed (from the N- to C-terminal end) of a FERM domain (band 4.1, erzin, radixin, moeisin domain), a pseudo-SH2 domain (ψ-SH2), a pseudokinase domain (JH2), and a kinase domain (JH1). While the major function of the N-terminus region is to specifically bind to cytokine receptor intracellular tail, the carboxyl-terminal region contains the catalytically active kinase domain, JH1, and the regulatory domain, JH2, of which the exact function is still matter of debate. ( B) Three-dimensional representation of JAK2 kinase domain in its active conformation (PDB: 3KRR ¹²) bound to type 1 inhibitor, ruxolitinib (show in yellow spheres). The binding of ruxolitinib has been modelled on the basis of the co-crystal of ruxolitinib-bound Scr (PDB: 4U5J ¹³). The activation loop, colored in pink, adopts an ‘open’ (active) conformation stabilized by phosphorylation of tyrosine residues 1007 and 1008. The αC is colored in orange. ( C) Three-dimensional representation of JAK2 kinase domain in its inactive conformation bound to a type 2 inhibitor, NVP-BTT594 (PDB: 3UGC ¹⁴). The activation loop, colored in pink, adopts a ‘closed’ (inactive) conformation. The compound is shown in cyan spheres, and the αC in orange. (D) Illustration of the structures of the main compounds discussed in this review.

Disease associated with abnormal JAK activation

Autonomous activation of the JAK/STAT pathway is central in several pathologies. Genetic alterations targeting this signaling pathway are associated mainly with hematologic malignancies. Pathological JAK activation also occurs in diseases linked to abnormal cytokine stimulation.

Hematologic malignancies

In many myeloid and lymphoid malignancies, driver mutations leading to constitutive JAK activation can be found. The paradigm is represented by BCR-ABL1–negative MPNs as they are all related to mutations activating JAK2, which in more than 50% of the cases both initiate and drive the disease phenotype. Thus, the development of JAK2 inhibitors is particularly critical for MPNs.

Classic BCR-ABL1–negative MPNs. The classic BCR-ABL1–negative MPNs (hereafter, MPNs) include three different disorders—essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (PMF)—and are caused by constitutive activation of the cytokine receptor/JAK2 pathway due to acquired somatic mutations in three major genes ¹⁵. JAK2V617F is the most prevalent mutation in MPNs associated with the three disorders (65–70%) and is present in 95% of PVs. Mutations in the exon 12 of JAK2 are found in around 2% of PV, which are negative for the JAK2V617F mutation (reviewed in 15). Interestingly, these two JAK2-activating mutations are not located in the kinase domain of the protein but involve amino acid changes in the pseudokinase (JH2) domain ( JAK2V617F) and the SH2-JH2 linker domain ( JAK2 exon 12) ( Figure 3A and see below). Mutations in the thrombopoietin receptor ( MPL) gene are much rarer than JAK2 mutations and are present in 3% of MPNs (4–5% of ET and myelofibrosis [MF]). The recurrent mutations are located in the exon 10. The most frequent are substitutions of the W515 residue to many other amino acids, mostly L or K but also R or A (reviewed in 15). In MPNs, the MPLS505N mutation is also found but less frequently than in hereditary thrombocytosis. Rarer mutations of MPL have been described in both the extracellular and intracellular domains. The third gene found frequently mutated in MPNs is calreticulin ( CALR), which is affected by mutations leading to a +1 frameshift in the exon 9 ^{16, 17}. As for MPL, CALR mutations are associated with ET and MF but with a higher frequency (25%). CALR is not a molecule directly involved in activation of JAK2, but the new C-terminus common to all mutants allows the CALR mutants to tightly bind and activate MPL and JAK2 ^{18– 21}.

All of these activating mutations mimic the effects of hematopoietic cytokines by inducing constitutive signaling via the STAT, PI3K, and ERK/MAPK pathways. To achieve this, JAK2V617F must be bound to cytokine receptors, more particularly homodimeric receptors (EPOR, CSF3R/G-CSFR, and MPL/TPOR), which allow its dimerization and activation ²². CALR mutants specifically activate MPL and (to a lesser extent) G-CSFR, and MPL mutations result in an active conformation of the receptor ¹⁸. Loss-of-function mutations in SH2B3, a negative regulator of JAK2, have been described as drivers in MPNs and idiopathic erythrocytosis ^{23, 24}, but there is evidence that in the majority of the cases it is either a secondary mutation or a germ-line mutation predisposing to MPNs ²⁵.

In 20–30% of patients with MPNs, mutations besides driver JAK2-activating mutations have been identified in genes involved in splicing ( SRSF2 and U2AF1) and epigenetic regulation ( TET2, DNMT3A, ASXL1, EZH2, and IDH1/IDH2) and in tumor suppressor genes ( TP53) ^{15, 26, 27}. They could predate or follow the driver mutations ^{28– 30}. Mutations, such as in TET2, DNMT3A, or EZH2, favor clonal dominance and disease initiation ^{28, 31}. Of note, these associated mutations promote progression to MF or leukemic transformation ( ASXL1, IDH1/2, EZH2, and TP53) ^{32, 33}. Almost always, one such mutation is present in PMF, and the number of such mutations correlates with the severity of the disease ^{32, 33} and might modulate the response to JAK2 inhibition.

Other myeloid malignancies

MPN associated with JAK2 fusion proteins

Four fusion proteins—ETV6 (TEL)-JAK2 (t(9;12) (p24;p13), PCM1-JAK2 (t(8;9) (p22;p24), BCR-JAK2 (t(9;22) (p24;q11.2), and RPN1-JAK2 (t(3;9) (q21;p24)—have been described in some MPNs—PMF, chronic eosinophilic leukemia (CEL), atypical chronic myeloid leukemia (aCML), and unclassified MPNs—or in mixed MPN/myelodysplastic syndrome (MPN/MDS) ^{34– 37}. The translocation partner of JAK2 sets the expression level and induces dimerization of JAK2 kinase domains.

Chronic neutrophilic leukemia, an MPN associated with G-CSFR ( CSFR3 ) mutations

Acquired activating CSF3R mutations have been found in the great majority of chronic neutrophilic leukemias (CNLs). Most CNLs carry a T618I mutation (T595I if the first counted residue is after the peptide signal sequence) ³⁸. The T615N (T592N) mutation has also been described but more rarely. CSF3R mutations have also initially been described in aCML but are much less frequent than in CNL.

JAK2V617F and (to a lower extent) MPL and CALR mutations are extremely prevalent in refractory anemia with ring sideroblasts and thrombocytosis and are associated with SF3B1 mutations ³⁹.

Mutations in JAK2 or cytokine receptors are rare in the other myeloid malignancies. They can be found in chronic myelomonocytic leukemia (CMML) but always associated with other mutations.

In acute myeloid leukemia (AML), mutations in JAKs are rare and, when present, mostly involve JAK1. CSF3R mutations have also been described but as late events. The only AML associated frequently with JAK mutations is Down syndrome acute megakaryoblastic leukemia (AMKL), where JAK1, JAK2, and also JAK3 mutations are found in around 20% of cases ^{40, 41}.

In acute myeloid leukemia (AML), mutations in JAKs are rare and, when present, mostly involve JAK1. CSF3R mutations have also been described but as late events. The only AML associated frequently with JAK mutations is Down syndrome acute megakaryoblastic leukemia (AMKL), where JAK1, JAK2, and also JAK3 mutations are found in around 20% of cases ^{40, 41}.

Lymphoid malignancies. JAK2 activation may play a more significant role in the pathogenesis of B and T neoplasms than previously thought, although they might be only secondary events. JAK/STAT activation occurs via either mutations/translocations or cytokine paracrine/autocrine loops.

Acute lymphoblastic leukemia

In B-cell acute lymphoblastic leukemia (B-ALL), a new subtype has emerged called BCR-ABL1-like subtype (15% of pediatric ALL and 50% of ALL with Down syndrome), as it exhibits a transcriptional profile similar to that of BCR-ABL1 –positive ALL with poor prognosis ⁴². Half of them have JAK1 and JAK2 mutations and rearrangement of CRLF2, a gene encoding a cytokine receptor ^{43, 44}. The most frequent JAK2 mutations target the R683 (R683G/S), a residue present in the DIREED motif located in the hinge between the N- and C-lobes of the pseudokinase domain of JAK2 ^{44, 45}. Fusion with partners ETV6, BCR, PAX, and SSBP2 results in the activation of JAK2 kinase domain. Mutations in JAK1 are relatively rare in B-ALL in comparison with T-cell ALL (T-ALL) ⁴⁶ and are located in the FERM domain and the pseudokinase domain. For instance, JAK1V658F is the equivalent of JAK2V617F ⁴⁷. In most cases, JAK2 mutations are associated with an aberrant expression of CRLF2, a cytokine receptor chain which associates with IL-7RA to bind the thymic stromal lymphopoietin (TSLP) ⁴⁸. The F232C-activating mutation in CRLF2 induces homodimerization of CRLF2 and is detected in 10% of cases overexpressing CRLF2 ⁴⁹. Activating mutations in the IL-7RA have been described either as point mutation (S185C) or as an insertion-deletion in the transmembrane domain inducing homodimerization of the IL-7RA ⁵⁰. Truncating rearrangements of the EPOR have also been identified ⁵¹.

In T-ALL, mutations in JAK1, JAK2, JAK3, and the IL-7R are found in around 15% of cases, more particularly in early T-cell progenitor (ETP)-ALL ⁵². Targeting JAKs in these two types of ALL appears to be a valuable approach.

Hodgkin lymphoma and primary mediastinal B-cell lymphoma

Hodgkin lymphoma and primary mediastinal B-cell lymphoma (PMBL) have some common mechanisms of lymphomagenesis, being driven by alterations in the nuclear factor-kappa B (NFκB) and JAK/STAT pathways ^{53, 54}. In rare cases (3%), a JAK2 fusion protein (SEC31A-JAK2) is present ⁵⁵.

Other lymphoma

In T-cell pro-lymphocytic leukemia, mutations in JAK1 (8%) and JAK3 (30%) (more particularly M511I) have been described ⁵⁶. JAK3 mutations (particularly A572V and A573V) are frequent (35%) in natural killer/T-cell lymphomas ⁵⁷. Mutations or silencing of negative regulators of JAKs ( PTPN2 and SOCS1) is also frequent in diffuse large B-cell lymphoma, follicular lymphoma, and peripheral T-cell lymphoma.

An autocrine/paracrine cytokine loop induced by IL-6, IL-10, and IL-13 activates JAKs. This occurs in several types of lymphoma as well as in chronic lymphocytic leukemia and Waldenström macroglobulinemia ⁵⁸.

Disorders other than hematologic malignancies

Inherited disorders of the hematopoietic system. The majority of inherited thrombocytosis are related to spontaneous activation of the MPL/JAK2 pathways due to MPL or JAK2 mutations ^{59, 60}. The other mechanism is related to an excess of plasma thrombopoietin (TPO) due either to an excessive synthesis or to a defect in its clearance as a consequence of MPL mutations affecting receptor trafficking. This excess of TPO induces JAK2 activation ^{59, 60}.

Inherited erythrocytosis are related to excess erythropoietin (EPO) synthesis or to gain-of-function mutations in EPOR that activate JAK2 ⁶⁰. Hereditary neutrophilia carry activating mutations of CSF3R that activate JAK2 ⁶¹.

Inflammatory and autoimmune diseases. A large spectrum of diseases of the immune system involves an activation of JAKs through autocrine or paracrine cytokine loops ⁶². In inflammatory pathologies, such as rheumatoid arthritis (RA), psoriasis, inflammatory bowel disease, and alopecia areata, the most important JAK to be targeted is JAK1 ⁶². In autoimmune disease, JAK1 is also the main JAK to be targeted as both the IL-6 and the type 1 interferon (IFN) pathways are involved. Blocking JAK1/2 in graft-versus-host disease (GVHD) is useful as type II IFN and IL-6 are pathogenic ⁶³. Lastly, JAK inhibition can be useful to curb oncogenic inflammatory responses in a wide range of solid tumors.

JAK inhibitors

Several types of inhibitors exist according to their mechanism/region targeted in JAKs ⁶⁴.

Type I inhibitors

Type I inhibitors target the ATP-binding site of the JAKs under the active conformation of the kinase domain ( Figure 3B) ⁶⁴. All clinically tested inhibitors are type I. They differ in their specificity for each JAK. Many inhibitors target JAK2, JAK1, and eventually TYK2 (ruxolitinib, momelotinib, AZD1480, and baricitinib) or JAK3 and JAK1 (tofacitinib). Some are pan-JAK inhibitors (gandotinib, XL019, NVP-BSK805, peficitinib, and pyridone 6). Less frequently, they target only JAK2 (NS-018, pacritinib, CEP-33779, NVP-BVB808, TG101209, fedratinib, and AZD960), JAK1 (filgotinib and itacitinib), or JAK3 (decernotinib, janex1, and JAK3-IN-1). However, they also target other kinases, in particular FLT3 (pacritinib, NVP-BVB808, TG101209, and fedratinib), Src (NS018), or Aurora A (AZD1480) ( Table 1 and Figure 3B).

Table 1. JAK inhibitors, their targets, and their applications to pathologies.

Inhibitors	Selectivity	Off- target	Diseases	Clinical phases
Type I
Ruxolitinib	JAK2>JAK1>JAK3		MF and hydroxyurea resistant or intolerant PV Refractory leukemia (post-MPN leukemia) Pancreatic cancers Corticosteroid refractory-GVHD Psoriasis Alopecia Vitiligo	FDA-approved 77– 79 Phase 2 80 Phase 2 in combination with capecitabine (after gemcitabine failure) 81 In evaluation 82 Phase 2 83 Open-label clinical trial 84 Case report 85
Momelotinib (CYT-387)	JAK2>JAK1>JAK3	ALK-2 TBK1 IKKε	PMF Post PV/ET MF PV/ET	Phase 3 - SIMPLIFY-1/2 (Stopped) 86 Phase 2 (terminated) 87
AZD1480	JAK2>JAK1	Aurora A FGFR1 FLT4	PMF Post PV/ET MF B-ALL Solid tumors	Phase 1 (completed) 88 In evaluation (preclinic) 89 Phase 1 (terminated) 90
Baricitinib (INCB-028050)	JAK2>JAK1		Rheumatoid arthritis Psoriasis	Phase 3 (FDA approval in process, EMA-approved) 91 Phase 2 92
Tofacitinib	JAK1>JAK3		Methothrexate-resistant rheumatoid arthritis Crohn’s Psoriasis Alopecia areata Dermatomyositis, vitiligo	FDA-approved 93 Phase 2 94 Phase 3 95 Phase 2 96 Case report 96
Gandotinib (LY2784544)	Pan-JAK JAK2V617F>JAK2		JAK2V617F-positive MF, ET and PV patients	Phase 1 97 Phase 2 (in progress)
XL019	Pan-JAK		PV, MF	Phase 1 (terminated) 98
NVP-BSK805	JAK2		JAK2V617F	Cellular models 99
NS-018	JAK2V617F>JAK2	Src	PMF, post PV/ET MF patients JAK2V617F selective	Phase 2 12 In vitro 100
Pacritinib (SB11518)	JAK2	FLT3	MF	Phase 3 PERSIST-1 101, PERSIST-2 102 PAC203 study evaluating the effect of lower doses
CEP-33779	JAK2		Rheumatoid arthritis, colorectal cancer, lupus nephritis	Preclinical mouse models 103
NVP-BVB808	JAK2	FLT3	MPN	Cell lines 104
TG101209	JAK2	FLT3	MPN, systemic sclerosis	Cellular models 105
Fedratinib (TG101348)	JAK2	FLT3 BRD4	MF	Phase 3 JAKARTA 106 FDA removed the clinical hold in August 2017
AZ960	JAK2		ATL, other leukemia	Cell lines
Filgotinib (GLPG0634)	JAK1>JAK2		Rheumatoid arthritis Bowel and Crohn’s diseases Lupus and psoriasis	Phase 3 107 Phase 2 108 Phase 2 109
Itacitinib (INCB-039110)	JAK1		MF Psoriasis Non-small cell lung cancer GVHD B-cell lymphoma BRAF-mutant melanoma and other solid tumors	Phase 2 (alone or in combination with low-dose of ruxolitinib) 110 Phase 2 111 Phase 2 (combination with EGFR inhibitor, osimertinib) (in progress) Phase 3 (combination with corticosteroids) (in progress) Phase 1/2 (combination with BTK inhibitor, ibrutinib) (in progress) Phase 1 (in combination with MAPK inhibitors, dabrafenib or trametinib) (in progress)
INCB52793	JAK1		Advanced malignancies	Phase 1 (in progress)
PF-04965842	JAK1		Moderate to severe psoriasis	Phase 2 112
Upadacitinib (ABT-494)	JAK1		Rheumatoid arthritis	Phase 2 113 Phase 3 (in progress)
Decernotinib (VX-509)	JAK3		Rheumatoid arthritis	Phase 2/3 114
WHI-P131/ JANEX-1	JAK3		GVHD	Preclinical mouse model 115
JAK3-IN-1	JAK3		N/A	N/A
Peficitinib (ASP015K)	JAK3		Psoriasis Rheumatoid arthritis	Phase 2 116 Phase 2 117
Type II
NVP-BBT594	JAK2	BCR-ABL KDR FLT3 RET		Cellular models 14
NVP-CHZ868	JAK2	KIT, PDGFR VEGFR	MPN B-ALL	Preclinical mouse models 75
Allosteric inhibitors
LS104	JAK2	BCR- ABL	MPN	JAK2V617F cell lines 118
ON044580	JAK2	BCR- ABL	MPN	BCR-ABL cell lines 119

ATL, adult T-cell leukemia; B-ALL, B-cell acute lymphoblastic leukemia; BTK, Bruton’s tyrosine kinase; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; ET, essential thrombocythemia; FDA, US Food and Drug Administration; GVHD, graft-versus-host disease; JAK, Janus kinase; MAPK, mitogen-activated protein kinase; MF, myelofibrosis; MPN, myeloproliferative neoplasm; N/A, not applicable; PMF, primary myelofibrosis; PV, polycythemia vera.

The differences in specificities for JAK are the basis for the different trials: JAK2 specificity for MPNs and certain malignant disorders ^{65, 66} and JAK1 and JAK3 for inflammation and auto-immune diseases ⁶⁷. The clinical toxicity can be related to the precise JAK protein that is inhibited—hematological toxicity, eventually immune suppression for JAK2 ⁶⁶,—immune suppression and long-term effects on hematopoietic stem cells (HSCs) for JAK1 and JAK3 ⁶⁸. It can be also due to an off-target inhibition. It has been suggested that the gastrointestinal toxicity was related to FLT3 inhibition ⁶⁹ and the Wernicke encephalitis observed in rare patients treated with fedratinib to inhibition of thiamine uptake ⁷⁰.

Currently, only three JAK inhibitors are US Food and Drug Administration (FDA) or European Medicines Agency (EMA)-approved for treatment: ruxolitinib for the treatment of MF and hydroxyurea (HU)-resistant or -intolerant PVs ⁶⁵, tofacitinib and Baricitinib for the treatment of methotrexate-resistant RA ⁷¹.

The major limitation of type I inhibitors that bind to active state kinases is that while they block catalysis they allow increased phosphorylation of the activation loop on Y1007, which upon overexpression of JAK2 or other JAKs can create heteromeric JAK complexes that re-set signaling ⁷². This might explain why resistance to JAK2 inhibition is not related to mutations but to functional inhibition ⁷³.

Type II inhibitors

The type II inhibitors bind to the ATP-binding pocket of kinase domains in inactive conformation, and the F of the DFG pocket is in an out conformation ( Figure 3C) ^{64, 74}. Inhibition is more efficient and is not reversed by drug detachment ¹⁴. The best example is imatinib and the second generation of BCR-ABL inhibitors ⁷⁴. Two type II JAK2 inhibitors (NVP-BBT594 and NVP-CHZ868) have been developed. NVP-CHZ868 has been used in preclinical models and was very effective ^{75, 76}. Both inhibitors were not amenable for drug development. Owing to their powerful activity, the type II JAK2 inhibitors present the risk of inducing profound cytopenia, limiting its future use in PV or ET.

Allosteric inhibitors

Allosteric inhibitors are molecules that do not bind to the active kinase site but to another site ⁶⁴. Theoretically, they would be more specific than an ATP-pocket inhibitor given the high homology of ATP-binding sites. Allosteric inhibition could be interesting to specifically target only the mutated JAKs ( JAK2V617F: 60–70% of MPNs and JAK2 exon 12: 1%) as JAK2WT is indispensable for normal hematopoiesis. An efficient JAK2WT inhibition will always lead to a cytopenia.

Within allosteric inhibitors, type III bind to a site close to the ATP-binding site while type IV bind to an allosteric site distant from the ATP-binding site. This means that in the case of JAK2V617F it might be important to target the pseudokinase domain (type IV inhibitors, see last section). To our knowledge, there is currently no allosteric inhibitor of JAK2V617F in development.

Other types of inhibitors

For kinases other than JAKs, additional types of inhibitors have been developed ⁶⁴:

Type V. Such inhibitors reversibly bind to two sites of the kinase domain.

Covalent. They irreversibly bind on a nucleophilic residue (usually a cysteine) located in the ATP-binding site.

JAK inhibitor therapies

BCR-ABL –negative MPNs

Triggered by the discovery of JAK2V617F, JAK2 inhibitors have been developed, but none of them is specific to the mutant protein and most of them also target other kinases such as JAK1 and FLT3. One advantage is that their use can be extended to the JAK2V617F-negative MPNs or other pathologies associated with JAK2 activation that include inflammatory diseases and certain other cancers. Theoretically, it is not conceivable to completely inhibit JAK2 in the long term, because this will lead to a profound cytopenia and eventually aplastic anemia. Thus, these inhibitors can be used because they only partially inhibit JAK2 in vivo for different reasons (pharmacokinetics and resistance). This explains why such inhibitors give similar clinical results in MF’s improving quality of life, decreasing the splenomegaly, and improving survival. The evolution of the disease is not changed, nor apparently is the rate of MPN transformation. The differences between inhibitors concern the side effects that may essentially be related to distinct off-targets ( Table 1).

Ruxolitinib. Ruxolitinib ( Figure 3B and D) was the first JAK2 inhibitor approved for therapy of MF (high and intermediate risks). It is now the only JAK inhibitor for which a long-term follow-up has been reached. In intermediate-2 and high-risk MF, a very significant effect is found in more than 50% of the patients with a reduced spleen size (clinical trials COMFORT-1 and COMFORT-2) at any point in the trial and even a more marked effect on the general symptoms, in particular the pruritus ^{77– 79}. These results have been extended to low- and intermediate-1 risk MF ¹²⁰. In HU-refractory PVs, ruxolitinib allows the control of hematocrit in more than 60% of patients and induces a spleen volume reduction in 38% of cases at 32 weeks. In addition, a molecular decrease of JAK2V617F allele burden reaching a mean of 40% at 208 weeks was observed ^{78, 121}. In HU-resistant or -intolerant ET, there are divergent results. In one study, ruxolitinib offered no advantage compared with other therapies in the control of the thrombocytosis and disease complications but did alleviate general symptoms and pruritus ¹²². In the other ¹²³, which was an open-label phase 2 trial, ruxolitinib induced a meaningful reduction in platelet levels and attenuated ET-related symptoms. These preliminary results seemed superior to historically observed results, but this study was done in the absence of a comparison with another treatment.

Overall, ruxolitinib is a well-tolerated oral treatment with approximately 25–33% of adverse effects. The main toxicities are hematological, moderate anemia that may correct with time, and thrombocytopenia, which can be very severe in high-risk MF. Weight gain is also observed with possible abnormalities in lipid metabolism. Middle-term toxicity is an immune suppression that may be responsible for reactivation of viral infections, particularly herpes zoster and HIV1 and bacterial infections such as pneumonia, tuberculosis reactivation and urinary tract infections ^{124, 125}. Long-term monitoring will be important because ruxolitinib decreases natural killer cell functions with a potential risk of solid tumor and lymphoma development ^{126, 127}. This is particularly important if indications are extended to low-risk MF, PV, and ET. Analysis of patients treated for several years with ruxolitinb indicates an increase in survival in MF, but progression to leukemia is not significantly different ¹²⁸. It is possible that most pro-survival effects derive from its palliative anti-inflammatory effects ⁶⁵.

Other JAK2 inhibitors. Momelotinib (CYT38) ( Figure 3D) is a JAK1/JAK2 inhibitor that has shown activity resembling ruxolitinib with respect to spleen size reduction and constitutional symptom alleviation ^{129– 131}. Importantly, momelotinib was shown to ameliorate anemia, which is a major concern in MF. The mechanism appears to be the reduction of hepcidin production through a direct inhibition of the activin receptor-like kinase-2 (ALK-2) ¹³². Thus, momelotinib was thought to be an alternative to ruxolitinib for patients with anemia. These promising results led to the opening of a phase III trial for the SIMPLIFY-1 and -2 studies in MF. However, the results of the two clinical trials did not show a major advantage of momelotinib on ruxolitinib, although momelotinib was associated with a decrease in transfusion requirement ⁸⁶. Momelotinib development has been stopped.

NS-018 ( Figure 3D) is a JAK2/Src inhibitor that has been assessed in patients with JAK2V617F-positive MF, ET, and PV. NS-018 shows an apparent increased potency for the JAK2V617F mutant in mouse models, possibly leading to less immunosuppressive effects ¹⁰⁰. It was tested in MF with symptom improvement but minor impact on the numbers of JAK2V617F cells ¹³³. Gandotinib (LY2784544) ( Figure 3D) is a potent JAK2 inhibitor, which also exhibits a certain selectivity toward JAK2V617F. It was evaluated for safety and tolerability in ET, PV, and MF ⁹⁷.

Pacritinib (SB1518) ( Figure 3D) is a JAK2/FLT3 inhibitor. Promising results were obtained in phase 1–2 clinical trials. It showed good activity in patients with less immunosuppressive effects ¹³⁴. Pacritinib could be administered to patients with low platelet levels, as it does not induce thrombocytopenia. The reasons behind this feature are unclear; they could be linked to reduced specificity for MPL/JAK2 complexes. Subsequently, two phase 3 clinical trials (PERSIST 1 and 2) were started with different doses of pacritinib. In 2016, the FDA put a complete clinical hold on the trials because of an excess of intracranial hemorrhage and cardiac failure in treated patients ¹³⁵. This hold was lifted in 2017 ¹³⁶. As a consequence, CTI BioPharma has just launched the PAC203 study evaluating the effect of pacritinib at different doses.

Fedratinib (TG101348) ( Figure 3D) was assessed during the JAKARTA trials with interesting clinical results, including fibrosis reduction ^{137– 139}, but rare patients developed Wernicke encephalopathy, which led to its stop ¹³⁹. It was assumed that it was related to an inhibition of thiamine uptake, although fedratinib does not lead to inhibition of thiamine uptake in rats ^{70, 140}. FDA removed the clinical hold in august 2017 and clinical trials are being planned in 2018.

JAK1 inhibition has also been proposed in MPNs, as an anti-inflammatory strategy, and an alternative to JAK2 inhibitors to avoid anemia and thrombocytopenia ^{65, 66}. A preliminary study has produced mixed results with a modest reduction in spleen size ¹¹⁰. Recently, it has been demonstrated that JAK1 plays essential and non-redundant roles at the level of HSC and therefore long-term JAK1 inhibition might have negative effects on HSCs ⁶⁸.

Other malignant disorders

Ruxolitinib was also used in refractory leukemia, including post-MPN leukemia, and 3 out of 18 patients achieved complete remission ⁸⁰. Ruxolitinib was unexpectedly used in combination with chemotherapeutic agents in solid cancers (pancreatic and lung cancer) and there was some benefit ¹⁴¹. This was based on JAK1 involvement in the signaling of several cytokine receptors, inflammation, and possibly progression of the malignant clone ^{81, 142}.

Ruxolitinib has been used both in myeloid malignancies with JAK2 fusion proteins and in CNL with a very good initial response but was inefficient to cure the diseases ^{143– 146}.

Itacitinib (INCB39110), which is also a selective JAK1 inhibitor, is being tested in non-small cell lung cancer in combination with an epidermal growth factor receptor inhibitor. Another JAK1 inhibitor, INCB52793, is also in development in advanced malignancies.

Inflammatory and autoimmune diseases

JAK1 inhibitors have been used in inflammatory/immune diseases ^{67, 147}. Tofacitinib was authorized by the FDA in RA ¹⁴⁸ but is currently being tested in Crohn’s disease, psoriasis, and other diseases ^{149, 150}. It decreases inflammation particularly by lowering T-cell and macrophage infiltrates. Filgotinib is currently in phase 3 clinical trials in RA and Crohn’s disease and in phase 2 clinical trials in lupus and psoriasis. Itacitinib (INCB39110) is being tested in phase 2 trials in psoriasis and in MF with meaningful improvements.

Ruxolitinib, also a good inhibitor of JAK1, was shown to reduce GVHD in mice and in patients with corticosteroid-refractory GVHD ^{82, 151}. Indeed, in GVHD, severe complications are due to high levels of proinflammatory cytokines that are inhibited by ruxolitinib. Itacitinib will be assessed as monotherapy in GVHD.

A JAK1 inhibitor, PF-04965842, is also in clinical trials in atopic dermatis and severe psoriasis ¹¹². Both topical tofacitinib and ruxolitinib are tested for alopecia and vitiligo.

Baricitinib (Olumiant), a JAK1/2 inhibitor (half maximal inhibitory concentration [IC ₅₀] values of 5.9 and 5.7 nM, respectively), has been initially identified by Incyte and subsequently developed by Eli Lilly and Company for RA. It also inhibits TYK2 (IC ₅₀ of around 53 nM). A phase 3 clinical trial was conducted, and the molecule was approved by the EMA but not yet by the FDA ^{148, 152}. It may be of interest in MPN treatment given its similarities with ruxolitinib but with a longer half-life (12.5 hours). Upadacitinib (ABT-494, AbbVie) is a highly selective JAK1 inhibitor that will enter phase 3 trials for RA, psoriasis, and ulcerative colitis.

Two JAK3 inhibitors were evaluated. ASP015K, also designated JNJ-54781532, displays a moderate selectivity on JAK3 over JAK1 and JAK2. It was shown to induce efficacy and safety in psoriasis ¹¹⁶. It has demonstrated efficacy in preclinical models of RA and dermatitis. Decernolitinib is also a potent and selective inhibitor of JAK3 developed as a second-generation inhibitor in autoimmune diseases, particularly in RA. However, at high doses, it leads to anemia, indicating that its selectivity in vivo could be different ¹⁵³. In MPNs, these types of inhibitors could also be useful to decrease inflammation, especially in MF.

The major drawback of these JAK1 and JAK3 inhibitors affecting the inflammatory response is that they can induce autoimmune diseases (thyroiditis or myocarditis) or can prime the development of many infections in patients with MPN. Furthermore, it appears that for the IL-2/IL-4/IL-7/IL-9/IL-15/IL-21 complexes, JAK1 is the initiating kinase for signaling, and a JAK3 that is inhibited can still fulfill its scaffolding role in the complex, its inhibition not giving results comparable to the absence of JAK3 by autosomal mutation ¹⁵⁴.

Perspectives in JAK inhibition

The goal of JAK2 inhibition in MPNs has switched from a curative therapy to a symptomatic and anti-inflammatory therapy with certain clinical benefits ⁶⁵. However, this aim is far from what is expected for a targeted therapy ⁶⁹. Although several in vitro studies identified potential kinase domain mutations that would give resistance to JAK2 inhibitors ^{155, 156}, no such mutations were identified in patients ⁷², further arguing that the current inhibitors are weak. Thus, there is a need for new inhibitors or combination of therapies.

New inhibitors

First, a new generation of JAK2-specific compounds, including allosteric inhibitors targeting unique sequences in JAK2, should come to light. Such inhibitors would provide more specificity toward JAK2 or JAK2V617F, thereby ameliorating normal JAK2 inhibition-based immune suppression. Two inhibitors have been developed (LS104 and ON044580) that inhibit kinase activities in a non-ATP-competitive manner ^{118, 119}. LS104 preferentially inhibits JAK2V617F kinase and can synergize with ATP-competitive inhibitors ¹¹⁸, whereas ON044580 inhibits BCR-ABL and its T315I mutant ¹¹⁹. The precise targeted residues remain unknown.

Identification of allosteric sites in enzymes has been accomplished for several years and is the basis for the development of a new class of pharmaceuticals. One example is the development of the BCR-ABL allosteric inhibitor GNF-2 that can overcome the effect of resistant mutations and also exhibits an increased potency when used in combination with classic ATP-competitive inhibitors ¹⁵⁷. Another example is the development of MEK allosteric inhibitors. It was shown that analogues of PD184352 could specifically bind to a unique region adjacent to the ATP pocket created by the displacement of the helix αC of the kinase in the active conformation ¹⁵⁸.

As the main conformational difference between JAK2V617F and JAK2WT is around the helix αC of the pseudokinase JH2 domain ¹⁵⁹, a rational design of similar molecules targeting this region could be efficient ( Figure 4). Recent structural and mechanistic data on JH2 V617F might help in designing such small molecules. Moreover, targeting JH2 might become feasible as shown by the recent publication of co-crystals of JH2 with several compounds. Future therapeutics should target small conformational changes that are specific for a particular mutant protein or its constitutive activity.

Figure 4. Graphic representation of the V617F-activation mechanism in Janus kinase2 (JAK2).

The V617F activation has been suggested to derive from a combination of several molecular events triggered from a region surrounding the JH2 αC but also involving the SH2-JH2 linker. Phenylalanine 617 interacts with F594 and F595 from the JH2 αC and F537 from the SH2-JH2 linker, as supported by structural data ^{9, 159}, then induces putative conformational changes that are transmitted to the adjacent catalytic kinase domain where activation is initiated. Targeting the ATP-binding pocket of JH2 that is spatially close the αC represents an appealing approach for specific targeting of the mutant JAK2V617F. Small molecules, such as BI-D1870 (as represented here ¹⁶⁰), have been co-crystallized as bound to the JH2 ATP-binding site. The use of amendable compounds targeting the JH2 pocket has recently become a tantalizing concept and will represent the future challenge for drug design.

The activating mutation, JAK2V617F, is located in the pseudokinase domain of JAK2 and allosterically regulates the activity of JAK2 kinase domain. Another strategy would be to interrupt these specific intramolecular changes induced by V617F on the kinase domain of JAK2. The aromatic phenylalanine at position 617 interacts with the αC helix phenylalanines 594 and 595 ¹⁶¹ and with phenylalanine 537 from the SH2-JH2 linker ⁹ to produce an autonomous activation to the adjacent catalytically active domain, JH1 ( Figure 4). Both of these aberrant hydrophobic contacts induce critical conformational changes that can be visualized on the JH2 V617F crystal structures ^{9, 159}. Any small molecules that would interfere with these specific atomic configurations exclusively present in the conformation of JAK2V617F would most likely generate mutant-specific inhibitors. Evidence was provided that V617F activates JAK2 through a community of residues across JAK2 (from JH2 to JH1 through the SH2-JH2 linker) and that reversing the charge of the negative residue, E596, located on the solvent-exposed face of the JH2 αC helix, can efficiently uncouple JAK2V617F activation from cytokine-induced activation, thereby restoring auto-inhibition of JAK2.

Finally, methotrexate, a well-known drug used in autoimmune disease was shown to also inhibit the JAK/STAT pathway and in theory could be tested for selective effects on the MPN clones ¹⁶². An advantage of this treatment is its low cost, while its side effects are well known.

Combination of therapies

The goal of certain combinations is to improve the anemia of MF, which can be worsened by ruxolitinib ⁶⁵. Trials have been performed in combination with androgen without benefit. Trials are ongoing with pomalidomide or thalidomide and also with sotartercept, an activin receptor IIa ligand trap ¹⁶³. Treatment with erythropoiesis-stimulating agents is also conceivable, although they are theoretically antagonistic with JAK2 inhibitors ¹⁶⁴.

Downstream JAK2 are activations of STATs, PI3K-AKT/mTOR, and RAS-MAPK ERK1/2 ( Figure 2) ¹⁶⁵. Specific inhibitors of STAT5 and STAT3 are now being tested in preclinical studies ¹⁶⁶. A combination of JAK2 and STAT5 inhibitors might be effective in MPNs. In preclinical models, JAK2 and pan type I PI3K and mTOR inhibitors synergize to block JAK2V617F-induced proliferation ^{167, 168} but less on mutant CALR-induced proliferation ¹⁸. Other molecules involved in the inhibition of the PI3K, AKT, or mTOR pathway were tested in preclinical models or clinical trials.

The HSP90 chaperone is involved in JAK2 stability and is controlled via its acetylation status ¹⁶⁹. HSP90 inhibitors or histone deacetylase inhibitors were tested ¹⁶⁹. Notably, panobinostat and pracinostat were studied in combination with ruxolitinib in a phase 2 trial with a better spleen response in MF ¹⁷⁰. Furthermore, a preliminary trial using an HSP90 inhibitor showed some clinical benefit ¹⁷¹. Combinations of ruxolitinib with CDK4/6 inhibitors, PIM1 kinase inhibitors, BH3 mimetics, or MDM2 inhibitors appear logical because they target molecules, which are downstream of JAK2 signaling, thus curbing any residual activation due to incomplete JAK2 inhibition due to short half-life of inhibitor and type I inhibition mechanism that allows rapid reactivation. Intriguingly, it has been reported that an association of ruxolitinib with pegylated interferon alpha was synergistic in a preliminary clinical trial, and a new larger clinical trial is ongoing ¹⁷². Normally, a JAK1 inhibitor should decrease interferon alpha signaling and should limit the effects of pegylated interferon; thus, the synergy might reflect how inefficiently ruxolitinib can actually inhibit JAKs in vivo. The main risk of all these different associations is to induce important cytopenia.

In MF, it has also been suggested to combine ruxolitinib with MEK inhibitors and PRM-151, a molecule that inhibits differentiation of fibrocytes ¹⁷³, which are implicated in fibrosis development ¹⁷⁴. Both drugs exhibit an effect on bone marrow fibrosis in preclinical studies or in clinical trials ¹⁷⁵. An inhibitor of hedgehog signaling pathway (LDE225) was also tested in phase I in ET/PV and MF with disappointing results ¹⁷⁶. However, recent evidence points to targeting Gli1 in association with ruxolitinib because Gli1 ⁺ mesenchymal cells may play a central role in fibrosis development ¹⁷⁷.

JAK2V617F can increase the protein methylation demonstrated by activation of PRMT5 arginine methyltransferase ¹⁷⁸. It also prevents the binding of heterochromatin factor HP1 to chromatin ¹⁷⁹. It is possible that some key genes could be hypermethylated as in many cancers. Therefore, it was postulated that demethylating agents such azacytidine and decitabine may have some impact. However, administered alone, they showed minor responses in two clinical trials ^{180, 181}. They were thus combined with ruxolitinib in a few patients with MF and the clinical response was good.

Conclusions

After the discovery of JAK2V617F and the demonstration that BCR-ABL–negative MPNs are driven by abnormal JAK2 activation, there were curative expectations for JAK inhibitors. Despite nearly 10 years of development in MPNs, only one JAK2 inhibitor (ruxolitinib) has been clinically approved. Most other inhibitors had their development stopped because of neurotoxicity or the absence of superiority compared with ruxolitinib. Only pacritinib ( Figure 3B) is still in phase 3 clinical testing and fedratinib is being reevaluated. Although ruxolitinib offers clear benefits for patients, its effects are quite limited on the disease itself in MF. On one hand, this may be explained by the additional mutations detected by MF clones, which act independently from JAK2 and which place MF at the boundary between MPN and MDS. On the other hand, ruxolitinib has a short half-life and acts as a type I inhibitor, which means that each time inhibitor is consumed, re-activation of JAK2 will occur.

Thus, it seems important in PV and ET, which are JAK2-dependent MPNs, to obtain specific inhibitors of JAK2V617F or even to preferentially target the constitutive active JAK2 over cytokine-activated JAK2 in the cases of mutated MPL and CALR ET and MF. It is expected that such inhibitors will be less toxic and will really target the clonal disease. In MF, the low-hanging fruit of combination therapies could be a valuable approach that holds a risk for significant toxicities.

An important finding of the clinical trials in MPNs was the discovery that JAK inhibition is a valuable approach for treatment of inflammatory diseases. Thus, it can be expected that one of the main applications of JAK inhibitors will be for inflammatory diseases, autoimmune diseases, and possibly other diseases, which may include an inflammatory response, including neurodegenerative disorders or cancers where inflammation contributes to oncogenesis.

Editorial Note on the Review Process

F1000 Faculty Reviews are commissioned from members of the prestigious F1000 Faculty and are edited as a service to readers. In order to make these reviews as comprehensive and accessible as possible, the referees provide input before publication and only the final, revised version is published. The referees who approved the final version are listed with their names and affiliations but without their reports on earlier versions (any comments will already have been addressed in the published version).

The referees who approved this article are:

• Srdan Verstovsek, Division of Cancer Medicine, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, USA

• Ann Mullally, Division of Hematology, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, USA

Funding Statement

This work was supported by grants from Ligue Nationale Contre le Cancer (‘Equipe labellisée 2016’ to WV and IP), Institut National du Cancer (PLBIO2015 to IP), MPN Research Foundation (IP), and Institut National de la Santé et de la Recherche Médicale (Inserm). The Laboratory of Excellence Globule Rouge-Excellence (IP and WV) is funded by the program ‘Investissements d’avenir’. SNC has received funding from the Ludwig Institute for Cancer Research, FRS-FNRS, Salus Sanguinis, Fondation contre le cancer, Project Action de Recherche Concerte of the Université Catholique de Louvain ARC10/15-027, the PAI (Interuniversity Support Program) Belgian Medical Genetics Initiative, and the Foundation ‘Les avions de Sébastien’ (Belgium).

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.