Jak1–STAT3 Signals Are Essential Effectors of the USP6/TRE17 Oncogene in Tumorigenesis

Protein ubiquitination plays a central role in diverse cellular processes, including the DNA damage response, vesicular trafficking, and transformation (1). Reversal of ubiquitination is catalyzed by deubiquitinating enzymes (DUB), which comprise five subclasses: the ubiquitin-specific protease (USP), UCH, OTU, Josephin, and JAMM subfamilies (2, 3). The USP subfamily is the most populous, with over 50 members in humans, and a key goal is to elucidate their substrate specificity and biological functions.

Multiple USPs have been implicated in human tumorigenesis, but particularly notable is USP6/TRE17 (4, 5). USP6 is a recurrent target of chromosomal translocation in several bone and soft tissue tumors, including aneurysmal bone cyst (ABC) and nodular fasciitis (6–8). In all cases, rearrangement causes promoter swapping and high level expression of wild-type USP6. Notably, USP6 is absent or expressed at exceedingly low levels in most normal adult human tissues (9).

Remarkably little is known regarding ABC and nodular fasciitis pathogenesis. ABC is a benign but locally aggressive bone tumor that can cause bone destruction, fracture, neurologic symptoms, and soft tissue damage (10–12). It was long hypothesized to originate from a localized hemodynamic disturbance, leading to formation of a dilated vascular bed, inflammatory recruitment, and neovascularization. However, identification of USP6 translocation in approximately 70% of ABCs revealed that they are neoplasms. Translocated cells are disseminated throughout the lesion. While the affected lineage has not been definitely determined, candidate cells are pre-osteoblasts, fibroblasts, and mesenchymal progenitor cells (13). ABCs are typically treated by curettage, but recurrence is not uncommon. In addition, they can arise in inoperable locations, where they can cause significant morbidity. Development of targeted chemotherapeutic agents would therefore be of great value.

Nodular fasciitis was also long believed to be a reactive process, but our identification of USP6 translocation in approximately 90% of cases revealed its neoplastic origins (14, 15). The translocation drives overexpression of WT USP6 (6). Again, the cell lineage harboring the translocation has not been conclusively identified, but fibroblasts and myofibroblast progenitors are likely candidates. Nodular fasciitis manifests as a rapidly growing subcutaneous lesion. Because of its histology and rapid growth, it can be mistaken for a sarcoma (16), resulting sometimes in overtreatment of nodular fasciitis patients with radiation and chemotherapy. As USP6 translocation is not found in sarcomas, our findings provided a novel diagnostic marker for nodular fasciitis, preventing unnecessarily aggressive treatment.

While identification of USP6 translocation in ABC and nodular fasciitis provided a significant advance, until recently little was known about how USP6 overexpression contributes to tumor pathogenesis. Though initially cloned over twenty years ago (17), its molecular functions in normal physiology remain elusive, due in large part to the absence of cell lines or primary tissues expressing appreciable levels of USP6. However, we recently established cellular and animal models to investigate its functions in ABC and nodular fasciitis pathogenesis (18–20). Overexpression of USP6 in fibroblasts or preosteoblasts (candidate cell lineages for harboring the translocation) led to tumor formation in xenografted immunodeficient mice. Molecular, histologic, and clinical features of ABC and nodular fasciitis were recapitulated in these systems (18–20). USP6′s ability to activate NF-κB was essential for tumorigenesis (19). Furthermore, USP activity was required, as USP6 alleles lacking USP activity failed to induce tumor formation (20). However, a vital question that remained unanswered was the identity of USP6′s relevant substrates. Previous studies indicate that USP6 promotes both deubiquitination of itself and of clathrin-independent endocytic pathway cargoes in vivo (21, 22). However, it remained unknown whether any of these proteins are direct substrates of USP6, and what their relevance is to transformation. Thus, identifying substrates that mediate USP6 tumorigenesis has remained a critical outstanding goal.

Jak family tyrosine kinases play an essential role in cytokine signaling (23–25). They function to phosphorylate STAT family transcription factors, which triggers their dimerization and nuclear translocation. The STAT3 family member is activated in numerous cancers, where it often acts in concert with NF-κB to promote tumor cell proliferation and survival (26, 27). In addition, they cooperate in establishing a protumorigenic microenvironment by inducing angiogenesis and stimulating production of inflammatory cytokines, such as IL6 (28, 29).

In the current study, we identify Jak1–STAT3 as essential mediators of USP6′s pathogenic mechanism. USP6 activates this pathway by deubiquitinating Jak1, leading to its dramatic stabilization and STAT3 activation. Inactivation of this pathway by CRISPR-mediated gene editing or pharmacologic means dramatically attenuates tumorigenesis by USP6. This work identifies the first deubiquitinating enzyme for a Jak family member, and defines a novel target for treating neoplasms driven by USP6 translocation.

NIH3T3 and MC3T3 cell lines expressing USP6 alleles are the major biologically relevant cells used in this study (18–20). Both parental cell lines were purchased from the ATCC in 2008 and 2007, respectively; certificates of analysis were provided. Authenticity of MC3T3 cells as osteoblasts was confirmed by validating expression of osteoblastic markers upon differentiation. Both lines were confirmed as murine by qPCR of multiple genes, including GAPDH and Jak1.

CRISPR target sequences for JAK1 (GTGGTCCCTGAGCCTGGAG) and STAT3 (CAGCTGGACACACGCTACC) were designed with ZiFit Targeter software (http://zifit.partners.org/ZiFiT/; Zinc Finger Consortium) and inserted into pCR-Blunt II-TOPO. TOPO plasmids were cotransfected with a plasmid encoding CAS9 and G418 resistance (Dr. Paul Bates, University of Pennsylvania, Philadelphia, PA).

Cycloheximide (C104450) and bafilomycin A (B1793) were from Sigma-Aldrich. MG132 (#474790), Jak Inhibitor I (CAS 457081-03-7; #420099), and Src inhibitor PP2 (#529573) were from EMD. Doxycycline (#8634-1) was from Clontech. IL6 was from R&D Systems (#206-IL-010). Lipofectamine and Oligofectamine were from Life Technologies.

USP6 constructs were described previously (18, 20, 21, 30, 31). FLAG-tagged WT and kinase-dead Jak1/pRK5 were provided by Dr. Warren Leonard (NHLBI, Bethesda, MD). Myc-Ub was provided by Dr. Roger Greenberg (University of Pennsylvania, Philadelphia, PA). USP6 SMARTPool siGenome siRNA (M-006096-03-0050) was from Dharmacon.

Cell lysis, immunoprecipitation, cell fractionation, and EMSA were performed as described previously (18, 20, 21, 30, 31). Jak1 (#3332), Jak2 (#3230), phospho-STAT3 (#9145), STAT3 (#9139) antibodies were from Cell Signaling Technology. Actin (sc-8432), p65 (sc-372), c-Myc(9E10) (sc-40), and HDAC2 (sc-7899) antibodies were from Santa Cruz Biotechnology. FLAG (M2) beads (A2220) were from Sigma-Aldrich. USP6 antibody was described previously (20).

To monitor Jak1 ubiquitination in vivo, HeLa cells were cotransfected with FLAG-Jak1 and Myc-Ub, with indicated USP6 constructs. Cells were treated for 2.5 hours with MG132 (20 μmol/L), lysed in RIPA buffer, then immunoprecipitated with FLAG beads for 4 hours at 4°C. Samples were washed three times, fractionated by SDS-PAGE, and blotted with anti-Myc. For in vitro DUB assays, FLAG-Jak1 and Myc-Ub were cotransfected, and then subjected to FLAG immunoprecipitation. Samples were washed in RIPA buffer, then DUB assay buffer (20 mmol/L Tris pH 7.5, 100 mmol/L NaCl, 0.05% Tween-20, 0.5 mg/mL BSA, and 5 μmol/L β-mercaptoethanol). The sample was divided into equal portions, one of which was immediately boiled. The remaining replicate samples were incubated with GST or GST-tagged USP6 (WT or catalytically inactive mutant; Ubiquigent) for 1.5 hours at 37°C. Samples were washed in RIPA and then immunoblotted with Myc.

NOD-SCID or NOG-SCID mice (4–8 weeks) were injected subcutaneously in the flank with 2.5E6 cells; tumors were harvested after 3–4 weeks, with animals maintained on water containing doxycycline (1 mg/mL; BioWorld, 40410005-2) and 5% sucrose. Jak inhibitor CYT387 (50 mg/kg; Selleckchem, S2219) or vehicle was administered twice daily by oral gavage where indicated. All animal procedures were performed under Institutional Animal Care and Use Committee–approved protocols.

Immunohistochemical staining was performed at the Pathology Research Core of the Mayo Clinic on 40 primary tumor samples whose USP6 translocation status was confirmed. Formalin-fixed paraffin sections were probed with phospho-STAT3 (Clone D3A7, Cell Signaling Technology) at 1:400. The Polymer Refine Detection System (Leica) was used, which includes hydrogen peroxidase block, post primary and polymer reagent, DAB, and hematoxylin. Slides were counterstained with Schmidt hematoxylin.

Microarray analysis was performed on 36 tumors: 9 nodular fasciitis tumors with USP6 translocation, and 27 others [alveolar rhabdomyosarcoma (n = 3); dermatofibroma or benign fibrous histiocytoma (n = 3); dermatofibrosarcoma protubera (n = 3); gastrointestinal stromal tumor (n = 1); malignant peripheral nerve sheath tumor (n = 3); melanoma (n = 3); neurofibroma (n = 3); embryonal rhabdomyosarcoma (n = 3); synovial sarcoma (n = 3), and schwannoma (n = 2)]. Total RNA was extracted from formalin-fixed, paraffin-embedded material using miRNeasy FFPE kit (Qiagen). Microarray analysis was performed on 200 ng RNA using Human WG-DASL Assay with Human HT12 v4.0 BeadChips (Illumina), which contain 29,377 probes. Microarray analyses were performed using Illumina GenomeStudio and Partek Genomics Suite software. Preprocessing of data at the probe level was performed within GenomeStudio software using quantile normalization with no background subtraction. “Nonexpressed” probes were defined as those with Illumina detection P > 0.05 in all 36 samples. 20,818 probes remained after filtering, and log2-transformed expression values were analyzed using Partek Genomic Suite software tools. Differentially expressed genes in nodular fasciitis compared with other tumor types were identified using ANOVA to calculate P values and fold enrichments. To control the false discovery rates (FDR) from the multiple testing, we calculated the adjusted P values using the Benjamini–Hochberg method (32). According to Source of Variance analyses, the microarray batch (array slide of 12 samples) was also included as a variable in the ANOVA. Gene set enrichment analysis was performed using the Hallmark molecular signature database (MSigDB) from the Broad Institute, which contains well-defined pathways curated from other gene sets. Additional analysis was done using Canonical Pathway MSigDB, which contains pathways from well-known databases such as KEGG and Biocarta typically compiled by domain experts. For both analyses, the permutation type was set to "geneset" and the metric for ranking genes was set to "Signal2Noise" (33, 34). Microarray datasets can be accessed at http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=cnejeuggnxifnmx&acc=GSE78991, and in Supplementary Table S1.

We previously reported that USP6 activates NF-κB, and that NF-κB plays an essential role in USP6-mediated transformation (19, 20). Given that NF-κB and STAT3 are often activated coordinately during tumorigenesis, we examined whether STAT3 was activated by USP6 in our established cellular models of ABC and nodular fasciitis. USP6 was stably expressed in a doxycycline-inducible manner in NIH3T3 fibroblasts or MC3T3 pre-osteoblasts (18–20). Upon doxycycline treatment of USP6/NIH3T3 or USP6/MC3T3 cells, robust activation of STAT3 was observed, as measured by phosphorylation on tyrosine residue 705 (Y705; Fig. 1A).

To confirm STAT3 activation in human tumors harboring USP6 translocation, IHC was performed on 24 primary nodular fasciitis and ABC samples, as well as 16 other mesenchymal tumors. Moderate to strong nuclear P-STAT3 staining was observed in 22 of 24 nodular fasciitis and ABC tumors (Fig. 1B and Supplementary Fig. S1; Supplementary Table S2). In ABC, both neoplastic cells and other cells in the microenvironment exhibited P-STAT3 activation (Fig. 1B, left), suggesting that USP6 might induce paracrine signaling (as was confirmed below). In nodular fasciitis (Fig. 1B, right), all of the neoplastic cells (with their characteristic plump spindle morphology, bundled in short fascicles; ref. 6) stained positively for P-STAT3.

To confirm whether STAT3 is activated by USP6 translocation in an unbiased manner, microarray analysis was performed on nodular fasciitis tumors. The optimal cell type to use as a counterpart for comparison was unclear, as the cell of origin in nodular fasciitis has not been definitively identified. However, as the affected lineage is mesenchymally derived, transcriptome analysis was performed on 27 tumors predominantly of mesenchymal origin that lacked USP6 translocation, and an "averaged" expression profile was generated. An nodular fasciitis transcriptome (derived from 9 independent USP6-translocated nodular fasciitis tumors) was then compared with this averaged data set. Using this strategy, we sought to exclude genes that are general mesenchymal markers or common indicators of the transformed state, and instead identify those that are selectively induced by USP6. Gene-set enrichment analysis (GSEA) revealed that signatures reflective of IL6/Jak/STAT3 and NF-κB activation were activated in nodular fasciitis (Fig. 1C and D and Supplementary Fig. S2; Supplementary Tables S3–S4). This validates our cellular model systems, and indicates that USP6 expression in human tumors triggers STAT3 and NF-κB signaling responses.

We sought to dissect the mechanism by which USP6 activates STAT3. We speculated that STAT3 activation might arise as a secondary consequence of NF-κB activation, as NF-κB can induce expression of multiple STAT3 agonists. To test this, we examined STAT3 activation in USP6/NIH3T3 cells in which NF-κB was inhibited through expression of IκB-alpha super repressor. Activation of NF-κB by USP6 was completely suppressed in these cells (Fig. 2A, left), in accordance with our previous results (19). Nevertheless, activation of STAT3, as monitored by its nuclear localization, occurred unimpeded (Fig. 2A, right), indicating that its activation is not dependent on NF-κB.

Jak and Src family kinases are capable of phosphorylating STAT3 (23, 35). To determine which family mediated activation of STAT3, pharmacologic inhibitors were utilized. A pan-Jak inhibitor completely blocked USP6-induced phosphorylation of STAT3, while the pan-Src PP2 inhibitor was ineffectual (Fig. 2B). To exclude the possibility that the pan-Jak inhibitor nonspecifically affected Src or receptor tyrosine kinases, a range of doses was tested. As shown in Fig. 2C–E, the pan-Jak inhibitor blocked USP6-induced phosphorylation of STAT3 in a dose-dependent manner, but had no effect on tyrosine phosphorylation induced by Src or the EGF receptor. Together, these results demonstrate that USP6 activates STAT3 exclusively through a Jak family kinase.

We next tested whether a Jak family inhibitor could prevent tumor formation by USP6. For these studies, we used CYT387, a pan-Jak inhibitor, that is efficacious in vivo (36, 37). USP6/NIH3T3 cells were subcutaneously injected into NOD/SCID mice. One cohort of animals was subjected twice daily to treatment with CYT387, while the other was treated with vehicle. As we previously reported, after 3 weeks large, highly vascularized tumors were visible in the control group (Fig. 2F). In contrast, tumors harvested at the same time point from CYT387-treated animals treated were significantly reduced in size and avascular (Fig. 2F). Together, these results indicate that Jak–STAT3 is activated in neoplasms harboring USP6 translocation, and that targeting this pathway may be an effective approach for controlling growth of tumors driven by USP6 overexpression.

To determine which Jak kinase is responsible for STAT3 phosphorylation, we examined expression of the various family members in our cells. Strikingly, we found that Jak1 levels were significantly increased in NIH3T3 and MC3T3 cell lines expressing USP6, in a doxycycline-dependent manner (Fig. 3A and C). This effect was highly specific, as other Jak family members (Jak2, Jak3, and Tyk2) were unaffected (Fig. 3A–C).

Analysis of various USP6 alleles revealed that upregulation of Jak1 required USP activity. Jak1 levels were not elevated in NIH3T3 cells stably expressing an inactive point mutant with a critical cysteine residue mutated to serine (USP6/CS; Fig. 3D). Similar results were obtained in transiently transfected HeLa cells: neither USP6(CS) nor USP6(short), a naturally occurring C-terminally truncated isoform that lacks USP activity, were competent to upregulate Jak1 or activate STAT3 (Fig. 3E), confirming that USP6′s USP activity can activate Jak1–STAT3 in different cellular contexts. In contrast, a triple point mutant in the TBC domain that ablates USP6′s ability to activate Arf6 (USP6/A6-; ref. 18) promoted Jak1 upregulation comparably with WT USP6 (Fig. 3E).

RT-qPCR revealed that Jak1 mRNA levels were not increased by USP6 (Supplementary Fig. S3). We therefore explored whether USP6 acts through a posttranslational mechanism. To monitor USP6′s effects on Jak1 protein half-life, HeLa cells were cotransfected with Jak1 and various USP6 alleles, then treated for various times with the protein synthesis inhibitor, cycloheximide. In vector control samples, Jak1 was rapidly degraded, with minimal protein remaining after 2 hours (Fig. 4A). Coexpression with USP6 led to dramatic stabilization of Jak1. Stabilization required USP activity, as Jak1 half-life was comparable in cells expressing USP6 (short) and control vector (Fig. 4A). Conversely, we found that siRNA-mediated knockdown of USP6 in HeLa cells led to significant reduction in steady-state levels of Jak1 protein (Fig. 4B).

Cytosolic proteins are typically degraded via the proteasome, whereas membrane proteins are typically degraded mediated via the lysosome. While Jak1 is cytosolic, it has the capacity to associate with transmembrane receptors (24), and could potentially be degraded by either route. To determine its mode of degradation, Jak1 stability was monitored in cells treated with the proteasome and lysosome inhibitors, MG132 and bafilomycin A (BafA), respectively. MG132, but not BafA, significantly prolonged the half-life of Jak1 (Fig. 4C). Control blots of the EGFR, which is degraded via the lysosome, confirmed proper functioning of BafA (Fig. 4D). Thus, Jak1 turnover is controlled through proteasome-mediated degradation.

As activated kinases sometimes trigger their own ubiquitination and degradation, we explored whether Jak1 kinase activity affected its stability. Two approaches were taken: first, cells were transfected with a kinase inactive mutant of Jak1. As shown in Fig. 4E (top), USP6 was able to stabilize this mutant as well as it did WT Jak1. As a second approach, cells were treated with the pan-Jak inhibitor. USP6′s ability to prolong the half-life of Jak1 was unaffected by the Jak inhibitor (Fig. 4E, bottom). Together, these data indicate that Jak1 kinase activity does not modulate its half-life, either basally or in the presence of USP6.

Our data are consistent with a model whereby Jak1 is normally maintained at low levels through proteasome-mediated degradation, but becomes stabilized through deubiquitination by USP6. To test whether USP6 is capable of directly deubiquitinating Jak1 in vitro, ubiquitinated Jak1 was immunopurified from HeLa cells coexpressing FLAG-Jak1 and Myc-Ub using anti-FLAG. Immunoprecipitates were incubated with recombinant USP6, and ubiquitin remaining on Jak1 was detected by anti-Myc immunoblotting. The isolated catalytic domain of USP6 efficiently deubiquitinated Jak1, while a catalytically inactive point mutant and control GST protein did not (Fig. 5A). Furthermore, consistent with USP6 serving as a DUB for Jak1, USP6 coimmunoprecipitated with endogenous Jak1 in vivo (Fig. 5B).

To test whether USP6 can induce deubiquitination of Jak1 in vivo, HeLa cells were cotransfected with FLAG-Jak1, Myc-Ub, and HA-USP6 alleles. Jak1 was immunoprecipitated using anti-FLAG, and its ubiquitination monitored by anti-Myc immunoblotting. Coexpression of USP6, but not USP6 (short), completely abolished ubiquitination of Jak1 in vivo (Fig. 5C). Although the level of Jak1 ubiquitination in cells expressing USP6 (short) was clearly higher than that in cells expressing WT USP6, they were consistently reduced compared with vector control cells. The underlying reason is unclear, but may be due to interaction of USP6 (short) with endogenous USP6. Regardless, together, these results establish Jak1 as the first substrate identified for USP6.

We sought to determine whether the Jak1–STAT3 signaling axis is required for USP6-mediated transformation. Toward this end, Jak1 and STAT3 were deleted from USP6/NIH3T3 by CRISPR-Cas9–mediated genome editing. Clonal cell lines were established with complete or monoallelic deletion of Jak1 (Fig. 6A), or complete deletion of STAT3 (Fig. 6B). Immunoblotting confirmed that the Jak1 and STAT3 CRISPR constructs did not target other family members (Supplementary Fig. S4A). USP6-induced STAT3 phosphorylation was completely abrogated in cells depleted of Jak1 (USP6/J1#17). Strikingly, STAT3 activation was also significantly reduced in cells with monoallelic Jak1 deletion (USP6/J1#11; Fig. 6A). Jak1 in this clone was reduced to levels approximating those in control NIH3T3 cells (Supplementary Fig. S4A), demonstrating that upregulation of Jak1 by USP6 is required for full activation of STAT3 by USP6.

Next, we examined the role of Jak1 and STAT3 in signaling and survival downstream of USP6. Upon serum withdrawal, NIH3T3 cells undergo growth arrest followed by apoptosis, which can be monitored by Erk inactivation and PARP cleavage, respectively (Fig. 6C and D). USP6 was competent to induce Erk activation and prevent PARP cleavage upon serum starvation, and this effect was abrogated by depletion of Jak1 or STAT3 (Fig. 6C and D). The protective effects of USP6 were also compromised in USP6/J1#11 cells, again supporting the notion that upregulation of Jak1 is required. We also examined the effects of Jak1/STAT3 depletion on proliferation under serum-replete conditions. Interestingly, USP6/NIH3T3 proliferated more slowly than vector control cells, and growth rates were not further reduced by Jak1/STAT3 depletion or Jak inhibitor (Supplementary Fig. S4B and S4C).

Xenografting of the CRISPR cells into immunodeficient mice revealed a critical role for Jak1 and STAT3 in USP6-induced tumorigenesis. Strikingly, both complete and partial depletion of Jak1 significantly reduced the mass of USP6-induced tumors (Fig. 6E). Deletion of STAT3 attenuated USP6′s tumorigenic potential even more dramatically than Jak1 (Fig. 6E). While 10 of 10 mice injected with USP6/NIH3T3 cells developed tumors, only 5 of 10 injected with USP6/S3#1 did. Furthermore, tumors that did form in STAT3-deficient cells were drastically reduced in size. Together, these data confirm that Jak1 upregulation is an essential component of USP6′s transformation mechanism, and that STAT3 is a critical downstream target.

Simply elevating Jak1 levels would likely be insufficient to elicit its activation; rather, agonists would still be required. As STAT3 was potently activated in USP6/NIH3T3 under serum-free conditions, this suggested that the cells were producing autocrine/paracrine factors. We confirmed that conditioned medium (CM) from USP6/NIH3T3 but not control cells induced STAT3 activation in naïve cells, to levels on par with IL6 (Fig. 7A). Production of paracrine factors was abolished in Jak1- and STAT3-deleted cells, and significantly reduced in USP6/J1#11 heterozygously deleted cells (Fig. 7A).

In addition to regulating the production of cytokines, we surmised that Jak1 levels might also dictate sensitivity to them. To test this, we examined STAT3 activation in response to exogenous Jak1 agonists. We found that not only was basal phosphorylation of STAT3 increased in USP6/NIH3T3 cells, but its activation was also enhanced in response to IL6 and IL4 (Fig. 7B and C). Deletion of Jak1 reversed this enhancement in a dose-dependent manner (Fig. 7B and C). Together, this group of experiments indicates that upregulation of Jak1 by USP6 is required not only for the production of autocrine/paracrine factors, but also for heightened STAT3 activation in response to them.

Our signaling studies and analysis of primary human tumors identify Jak1–STAT3 as critical effectors of USP6 in oncogenesis. We identify Jak1 as the first direct substrate of USP6, and demonstrate that it is required for STAT3 activation by USP6. We further show that this pathway is essential for USP6-mediated tumorigenesis. Our data support a model in which Jak1 levels are increased through deubiquitination by USP6, rendering cells hypersensitive to low levels of Jak1 agonists present in the microenvironment. This leads to activation of STAT3, which drives production of autocrine/paracrine factors by USP6-expressing cells; the elevated Jak1 levels sensitize USP6-expressing cells to the autocrine factors, thereby amplifying STAT3 activation in a positive feedback loop. In addition, the paracrine factors induce STAT3 activation in neighboring nontranslocated cells in the tumor microenvironment. Whether autocrine/paracrine factor production involves pathways in addition to Jak1-STAT3 remains to be determined. This work summons the first targeted therapy (namely Jak-STAT3 inhibitors) for the treatment of neoplasms driven by USP6 translocation/overexpression.

Our studies illustrate an atypical mode of kinase regulation. Kinases are most commonly regulated through site-specific phosphorylation. While ubiquitination and proteasomal degradation have been reported for multiple kinases, they are typically triggered by the activated kinase as a means of auto-downregulation (38). However, degradation of Jak1 is not dependent on its activation: inhibition of Jak1 kinase activity neither affects basal turnover, or stabilization by USP6. This mode of regulation is reminiscent of another kinase, NIK, a central regulator of noncanonical NF-κB. Like Jak1, NIK is maintained at low levels through constitutive ubiquitination (39). Future studies will determine whether physiologic agents regulate Jak1 levels through modulation of USP6.

ABC and nodular fasciitis can be added to the growing list of neoplasms harboring Jak1 dysregulation. Activation of Jak1 has been reported in leukemias, hepatocellular carcinoma, and gynecologic cancers (40–43). In these cases, activation arose through mutation of Jak1, usually within the kinase domain. To our knowledge, this is the first example of Jak1 being regulated at the protein level, through modulation of its turnover.

Most importantly, our work identifies potential novel therapeutic strategies for the treatment of ABC and other neoplasms driven by USP6 overexpression. Operable cases of ABC are typically treated by curettage of the affected bone. However, there is no standard of care upon recurrence (which affects up to 70% of patients), or for inoperable cases (10, 11). Treatment strategies, which involve prevention of recurrence, repair of damaged tissue, and pain management, can be associated with serious complications or morbidity. Our current and prior work suggests that inhibitors of Jak kinases, STAT3, and NF-κB would be highly effective in the management of ABC, as inactivation of all of these effectors dramatically attenuated USP6′s tumorigenic potential (19). Notably, such inhibitors are being avidly explored for treatment of other cancers and inflammatory diseases (44–49), paving the way for their testing in ABC patients. We posit that regimens that simultaneously inhibit Jak-STAT3 and NF-κB might be particularly effective. Moreover, as USP activity is required for activation of Jak1–STAT3 and NF-κB, development of USP6-specific inhibitors would also be a highly desirable approach. Given USP6′s highly restricted expression in normal tissues (9), such an inhibitor would likely have minimal side effects. In sum, our work identifies several new avenues for the treatment of recurrent and inoperable ABC, and other neoplasms driven by USP6 overexpression.

No potential conflicts of interest were disclosed.

Conception and design: M.M. Chou

Development of methodology: L. Quick, X. Wang, M.M. Chou

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I.C. Henrich, X. Wang, A. Oliveira, M.M. Chou

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Quick, I.C. Henrich, Y.W. Asman, M.M. Chou

Writing, review, and/or revision of the manuscript: I.C. Henrich, Y.W. Asman, A. Oliveira, M.M. Chou

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Quick, I.C. Henrich, X. Wang

We thank Xavier Jaq and Jeannine Harrigan (Mission Therapeutics) for advice on DUB assays, and Gerd Blobel for reading of manuscript.

This work was supported by RO1 CA168452/R21 CA178601 (NCI/NIH; M.M. Chou) and Hauck Family Fund for Bioinformatics from Mayo Center for Individualized Medicine at Mayo Florida (Y.W. Asman).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.