Coinhibition of the deubiquitinating enzymes, USP14 and UCHL5, with VLX1570 is lethal to ibrutinib- or bortezomib-resistant Waldenstrom macroglobulinemia tumor cells

The survival of Waldenstrom macroglobulinemia (WM) tumor cells hinges on aberrant B-cell receptor (BCR) and MYD88 signaling. WM cells upregulate the proteasome function to sustain the BCR-driven growth while maintaining homeostasis. Clinically, two treatment strategies are used to disrupt these complementary yet mutually exclusive WM survival pathways via ibrutinib (targets BTK/MYD88 node) and bortezomib (targets 20 S proteasome). Despite the success of both agents, WM patients eventually become refractory to treatment, highlighting the adaptive plasticity of WM cells and underscoring the need for development of new therapeutics. Here we provide a comprehensive preclinical report on the anti-WM activity of VLX1570, a novel small-molecule inhibitor of the deubiquitinating enzymes (DUBs), ubiquitin-specific protease 14 (USP14) and ubiquitin carboxyl-terminal hydrolase isozyme L5 (UCHL5). Both DUBs reside in the 19 S proteasome cap and their inhibition by VLX1570 results in rapid and tumor-specific apoptosis in bortezomib- or ibrutinib-resistant WM cells. Notably, treatment of WM cells with VLX1570 downregulated BCR-associated elements BTK, MYD88, NFATC, NF-κB and CXCR4, the latter whose dysregulated function is linked to ibrutinib resistance. VLX1570 administered to WM-xenografted mice resulted in decreased tumor burden and prolonged survival (P=0.0008) compared with vehicle-treated mice. Overall, our report demonstrates significant value in targeting USP14/UCHL5 with VLX1570 in drug-resistant WM and carries a high potential for clinical translation.


INTRODUCTION
Waldenstrom macroglobulinemia (WM) is an indolent yet incurable B-cell lymphoma, in which aberrant B-cell receptor (BCR), Tolllike receptor and MYD88 signaling cooperatively drive tumor cell survival. This observation has been clinically exploited with the BTK inhibitor, ibrutinib. Although capable of inducing remission iñ 91% of WM patients, the majority of patients on ibrutinib therapy exhibit partial response, with no complete remissions observed. 1 Moreover, the phenomenon of ibrutinib resistance is increasingly being reported in other B-cell cancers, [2][3][4][5][6][7] and although WM patients demonstrate durable responses to ibrutinib, it is anticipated that when these patients relapse (or become refractory to the drug), their survival outcome will be poor. 8,9 These observations underscore the plasticity of WM cells to adapt to targeted therapeutic stress through modulation of other oncogenic survival mechanisms, signifying a need for continued development of new treatment strategies.
To sustain BCR/Toll-like receptor-driven growth and maintain optimal cellular homeostasis, WM cells upregulate proteasome function. 10,11 Clinical benefit derived through proteasome inhibitor (PI) therapy has been clearly established in several B-cell cancers, including WM. 12 PI-based management of disease yields responses in~85% of relapsed/refractory WM patients. [13][14][15][16] Current PI (bortezomib, carfilzomib and ixazomib) are designed to preferentially inhibit the β-subunit catalytic activity of the 20 S proteasome core. However, the 26 S proteasome contains several distinct components (with exclusive functions), each whose roles in supporting tumorigenesis remain underexplored and for which no therapeutic agents are currently approved. The 19 S regulatory particle of the mammalian proteasome contains deubiquitinating enzymes (DUBs) that unfold and remove ubiquitin moieties from proteins before their entry into the 20 S β-catalytic core. 17 Two of these, ubiquitin-specific protease 14 (USP14) and ubiquitin carboxyl-terminal hydrolase isozyme L5 (UCHL5), reversibly associate with the proteasome, 18 and, collectively, their dysregulated activity has been linked to tumor cell survival, 19 metastasis 20 and poor clinical outcome. 21,22 We have recently described the effects of a chemical inhibitor (b-AP15), which preferentially targets USP14 and UCHL5 and induces robust tumor cell death in drug-resistant lymphoplasmacytic cells in vitro. 23 Herein, we extended our investigations to define the in vitro and in vivo antineoplastic activity of a clinical grade DUB inhibitor, VLX1570, notably in WM cells that are resistant to β5-targeting PI or BTK inhibition. Compared with b-AP15, VLX1570 displays enhanced solubility, stability and target residence time and is currently being investigated in relapsed/ refractory multiple myeloma patients.

MATERIALS AND METHODS
Primary WM cells, cell lines, cell culture and reagents WM cell lines and derived bortezomib-resistant (BR) suclones, 24,25 as well as ibrutinib-resistant (IR) subclones, 26,27 were used in experiments, as reported previously. 23,25 Primary WM patient cells (CD19+/CD138+ sorted) were obtained from the Predolin Biobank (Mayo Clinic, Rochester MN, USA) after approval by the Mayo Clinic Institutional Review Board. Heparinized peripheral blood from healthy human donors was obtained and peripheral blood mononuclear cells (PBMCs) were extracted, as described previously. 28 All cells were cultured according to conditions previously described by us. 23 VLX1570 and b-AP15 were provided as gifts from Vivolux AB (Uppsala, Sweden). RPMI, penicillin, streptomycin, tetramethylrhodamine, methyl ester (TMRM) and fetal bovine serum were purchased from Life Technologies (Carlsbad, CA, USA). Ibrutinib and bortezomib were purchased from Sellekhem (Houston, TX, USA). Annexin-V/Propidium Iodide Apoptosis Staining Kit was purchased from BD Biosciences (San Jose, CA, USA).
Cell death, proliferation and apoptosis assays MTS assay was used to determine the half-maximal inhibitory concentration values and proliferation rate/viability; apoptosis was determined using the Annexin-V/Propidium Iodide Binding Assay Kit from BD Biosciences (San Diego, CA, USA) according to the manufacturer's instructions and previously described methods. 23,25 Determination of MOMP Cells were treated with VLX1570 for 12 h and assessed for mitochondrial outer membrane permeability (MOMP) using TMRM (Life Technologies) in a manner similar to that reported by us previously. 23,25 CXCR4 surface receptor analysis For staining of CXCR4 on WM tumor cells, cells were washed two times with cold phosphate-buffered saline and suspended in 300 μl of binding buffer (phosphate-buffered saline solution with 2% fetal bovine serum). Cell were divided in three tubes: unstained, isotype control and those with antibody against CD184/CXCR4 (BioLegend, San Diego, CA, USA). Five microliters of antibody was added and cells were incubated for 30 min at room temperature. Tumor cells were washed two times with cold phosphate-buffered saline and suspended in 100 μl of 4% paraformaldehyde in phosphate-buffered saline solution (Affymetrix Inc., Santa Clara, CA, USA; 19943 1LT) followed by an analysis using a BD Accuri C6 Flow Cytometer (Franklin Lakes, NJ, USA). FCS Express 4 (De novo Softwares, Los Angeles, CA, USA) was used to analyze the data.

Human WM xenograft model
Animal experiments were performed with the approval of the Institutional Animal Care and Use Committee of Mayo Clinic. A xenograft model of WM was established as described previously. 27 Calculation of drug combination effects The pharmacobiological interaction between VLX1570 and ibrutinib was examined in WM cells based on the Chou-Talalay principle 29 using the CompuSyn software program (ComboSyn Inc., Paramus, NJ, USA). WM cells were exposed to different concentrations of VLX1570 and ibrutinib for 24 h in quadruplicate and viability was examined using the CellTiter Glo assay (Promega Corp., Madison, WI, USA), according to the manufacturer's instructions. Cell viability data were expressed as the fraction of antiproliferative activity (Fraction affected, Fa) by the individual drugs or the combination in drug-treated cells from which a combination index (CI) value was derived. An additive effect is indicated by a CI of 1, an antagonistic effect is denoted by CI 41 and a synergistic effect is indicated by a CI o 1.

Statistical analysis
Statistical evaluation was performed using GraphPad Prism 6 (GraphPad Software Inc., La Jolla, CA, USA). Standard error bars were derived from the standard deviation of values from average values from three or more experiments. P-values were calculated using a Student's t-test of two data array sets in triplicates with one-tailed and two-sample unequal variance. The Kaplan-Meier method was used to visualize overall survival differences between VLX1570-treated mice and mice given vehicle only. A log-rank statistic was used to establish the statistical significance of the difference between survival distributions.
Please see the Supplementary Materials and methods for additional information.

RESULTS
USP14 and UCHL5 is highly expressed in WM cells USP14 and UCHL5 are known to be expressed in a variety of cell types including hematopoietic cells. 17,30 We first quantified their expression in WM cell lines as well as in CD19+/CD138+-sorted primary WM tumor cells. Compared with PBMCs from healthy donors, USP14 mRNA expression was significantly elevated in WM tumor cell lines and primary WM cells (5.4-fold median increase; range, 2.8-19-fold) (Figure 1a). For UCHL5, we found variable expression, where compared with healthy donors, UCHL5 mRNA was higher only in IR WM cells (BCWM.1/IR, 1.4-fold increase) and primary cells from a WM patient with relapsed/refractory disease (WM2, 3-fold increase) ( Figure 1b). These findings are largely consistent with our previous report on USP14/UCHL5 protein expression in BR WM cell lines. 23 Immunostaining of bone marrow biopsies from patients with a confirmed diagnosis of WM (Figure 1c, WM6 and WM3) demonstrated increased staining intensity and overall distribution of both USP14 and UCHL5 compared with marrow from normal healthy donors (Figure 1c, N5 and N6). Altogether, these findings affirmed that both proteasome-associated DUBs are upregulated and their expression is increased in human WM cells.

VLX1570 binds and inhibits the activity of USP14 and UCHL5
We have recently reported on the lead optimization of b-AP15, from which VLX1570 emerged as the most promising analog, suitable for clinical use. 31 We modeled VLX1570 and b-AP15 side by side and performed a comparative in silico docking analysis on their binding capabilities with USP14 and UCHL5 (Figures 2a and  b; b-AP15 in turquoise and VLX1570 in orange). VLX1570 preferentially binds the Cys88 residue of UCHL5 and interfaces with a thiol of USP14 at residue Cys114 via a 1,4-Michael's addition reaction, potentially forming a covalent bond. These binding interactions are nearly identical to those of b-AP15; however, notable differences between the compounds account for superiority of VLX1570 over b-AP15. b-AP15 has two nitro moieties (one per phenyl ring), whereas VLX1570 has nitro moieties and also carries a fluoro on each phenyl at the para-position for a connection with a methylene carbon. b-AP15 has a central nitrogen-substituted hexyl moiety, whereas VLX1570 has a nitrogen-substituted heptyl moiety (compare Figures 2c and d). Improved affinity for its targets is most likely due to the electrostatics from the two fluoro atoms. Additionally, although both compounds have an α-enamide at the warhead position, the nitrogen-substituted alkane ring is slightly modified for VLX1570 so that a better α-enamide position is available for approaching the Cys residue in the active site of the target proteins. The precovalent docked poses from VLX1570 are superior to the precovalent docked poses of b-AP15, indicating that the initial approach of the VLX1570 is superior for forming optimal geometry of orientation required for the thiol addition reaction. In general, VLX1570 is better able to form covalent docking owing to better position before the reaction. Thus, these chemical differences predict for VLX1570 to be significantly more stable and soluble with improved in vivo clearance.
Next, to determine the binding capacity of VLX1570 in a cellular system, we used an enzyme-based competition assay, which uses a DUB suicide-substrate probe (HA-UB-VS) that competes with active DUBs for binding to other molecules. We observed at 30 min that VLX1570 inhibited the activity of both USP14 and UCHL5 in a manner similar to b-AP15, as noted by a decrease in UbVS-labeled DUBs (Figure 2e). To determine if this effect occurred at an earlier time point, we probed for labeled and   Table 3). Contrastingly, exposure of PBMCs from healthy donors to VLX1570 or b-AP15 resulted in negligible cell death. These data indicate that VLX1570 is minimally toxic to nontumor cells from the peripheral blood of healthy donors but induces significant cell death in malignant WM cells, irrespective of their resistance to β5-targeting PI bortezomib or ibrutinib.
VLX1570 modulates ER stress machinery and perturbs mitochondrial transmembrane integrity leading to activation of the intrinsic apoptotic cascade USP4/UCHL5 inhibition has been shown to have a prominent effect on ER and cell stress proteins. 33,34 Therefore, we looked residues that are critical for binding affinity, notably for the covalent docked linkage at the -SH thiol from Cys114. (b) Similarly, overlay of b-AP15 and VLX1570 in UCHL5 shows their relative alignment and optimal positioning to covalently dock at the -SH thiol from Cys88. Threedimensional (3D) chemical renderings of (c) b-AP15 and (d) VLX1570 reveal key structural differences that result in enhanced activity of VLX1570. (e) A competitive enzyme assay using the HA-Ub-VS DUB labeling probe was carried out in BCWM.1 cells to identify whether the DUB inhibitors reduced labeling of USP14/UCHL5 in live WM cells. BCWM.1 cells were treated with DMSO, b-AP15 or VLX1570 for 3 h and thereafter incubated with the HA-Ub-VS probe for 30 min. After immunoblotting with USP14 or UCHL5 antibodies, we observed the fraction of labeled USP14 or UCHL5 was notably reduced in DUB inhibitor-treated vs DMSO-treated cells. (f) In a similar manner, we labeled DMSO or VLX1570-treated BCWM.1 cells with HA-Ub-VS for 1, 10 and 20 min to determine if reduction in DUB activity occurred at an earlier stage. We noted that labeled USP14 and UCHL5 were markedly decreased between 10 and 15 min after suicide-substrate labeling.
at ER stress-associated proteins, and in line with prior observations, 30,34 we noted ER stress machinery such as XBP-1 (and its spliced active form, XBP-1 s), which are primary effectors of ER stress induction, 35 to be upregulated by VLX1570 in WM cell lines (Figure 4a). IRE-1α levels did not show any consistent pattern of alteration following VLX1570 treatment. These differential changes (in IRE-1α) may reflect underlying differences in the degree of basal ER stress present across the different WM cell lines and thus their response to VLX1570.
Disruption of the mitochondrial transmembrane potential (Δψ m ), through an increase in MOMP, can occur in response to a strong (but ultimately futile) ER stress response and signifies engagement of the intrinsic apoptotic cascade. 36,37 We found MOMP to be significantly induced in WM cell lines exposed to  Table 4). Activation of the intrinsic apoptotic pathway was also verified by cleavage of caspase-9 (Figure 4c).
To further understand the mitochondrial bioenergetics preceding these events, we measured mitochondrial oxygen consumption rate and noted that mitochondrial respiration is triggered as early as 2 h after VLX1570 treatment (Figure 4d). Increasing concentrations of VLX1570 caused elevation of basal respiration and ATP production, as well as uncoupler induced maximal respiration (Figure 4e). Maximal mitochondrial respiratory activity was reached by VLX1570 250 nM and higher concentrations resulted in collapse of all monitored parameters, likely due to mitochondrial exhaustion, ultimately leading to mitochondriamediated apoptosis.
VLX1570 downregulates BCR-signalosome components and their end effectors, as well as CXCR4 expression The 26 S proteasome regulates bioavailability of several kinases and those of the TEC family, which includes BTK. 38 Bortezomib has been shown to downregulate the expression of BTK as well as other BCR-signalosome components such as MYD88. 38,39 We addressed the question of whether this phenomenon could be similarly elicited by blocking the DUB (USP14/UCHL5) machinery (USP14/UCHL5) of the 19 S proteasome with VLX1570. As anticipated, exposure of BCWM.1/IR cells to VLX1570 (3 h treatment) resulted in decreased transcription of BTK and MYD88 along with their downstream effector NFATC2 (Figure 5a). This was then validated by examining the corresponding protein levels, where immunoblot analysis conducted on VLX1570-treated BCWM.1/IR cells demonstrated reduced expression of BTK, MYD88 and NFATC2. BCR signaling ultimately terminates in the activation of nuclear factor-κB (NF-κB). 40 Thus, we interrogated the impact of VLX1570 on NF-κB and by using western blot analysis we observed that VLX1570 decreased the nuclear translocation of phosphorylated-NF-κB (p-p65) (Figure 5c, one representative cell line shown).
WM cells are known to rely on CXCR4 signaling for homing into the bone marrow, and interestingly, 41 CXCR4 receptor expression is regulated by USP14. 20 As such, we examined its expression in WM cells after treatment with VLX1570 (250 nM, 3 h treatment) and observed not only decreased CXCR4 mRNA but also reduced surface receptor expression (Figures 5d and e). Taken in concert, these analyses reveal for the first time that proteasome inhibition at the 19 S lid alters survival-promoting BCR signaling and chemokine receptor (CXCR4) availability, particularly in IR WM cells.
WM-bearing mice treated with VLX1570 show significant reduction in tumor burden and increased survival Although b-AP15 has been examined in murine models of various cancers 30 including MM, 33 the in vivo activity of its lead optimized analog, VLX1570, has been described in only two disease models thus far. 42,43 Thus, using our murine xenograft model of WM, 44 we administered VLX1570 via intraperitoneal injection to WM-bearing mice. Xenografted tumors were allowed to grow in size until detection by bioluminescent imaging confirmed congruent tumors (~200 mm 3 ) in all mice. Thereafter, mice were randomized into two groups, receiving either vehicle or VLX1570 (4.4 mg/kg) every alternate day. After 21 days of treatment, compared with the vehicle-treated group, we observed a significant reduction in tumor volume and growth as detected by both bioluminescent radiance (Figures 6a-c) and direct caliper measurements (Figures 6d and e). As a secondary systemic marker of response, we measured human immunoglobulin M (IgM), which is secreted by the xenotransplanted WM cells into the mice sera. Significantly lower human IgM levels were noted in VLX1570-treated mice versus those in the vehicle-treated group (Figure 6f). VLX1570 treatment was well tolerated with insignificant weight loss observed in drug-treated mice by day 37 and no other signs of systemic toxicity (Figure 6g). Importantly, a significant survival (d) CXCR4 mRNA levels were assessed in BCWM.1/IR cells in a manner similar to (a) and were found to be reduced by 3.9-fold relative to DMSO-treated cells. (e) Decrease in CXCR4 was further affirmed by assessing expression of the surface receptor. BCWM.1/IR cells were treated with DMSO or VLX1570 (250 nM) for 6 h and incubated with anti-CD184 antibody followed by flow cytometry analysis. Histogram demonstrates CXCR4 receptor expression was reduced by 21-28% in VLX1570-treated cells relative to DMSO-treated counterparts (one representative histogram shown). Black line represents isotype control (DMSO-treated), green line indicates CXCR4 in DMSO-treated cells and red line indicates its expression in VLX1570-treated cells.
advantage in mice treated with VLX1570 was noted as compared with mice receiving vehicle alone (median survival, 62 vs 44 days, respectively, P = 0.0008) (Figure 6h).
Pharmacodynamic effects of therapy were studied using immunohistochemistry analysis on mice tumor tissues and showed increased apoptosis (cleaved caspase-3) as well as decrease in WM cell proliferation (Ki-67) and secretion of human IgM in VLX1570-treated mice (Figure 7a). In vivo target engagement was assessed in a more specific manner by staining for K-48linked ubiquitinated protein accumulation and CXCR4, which were increased and decreased, respectively, in mice treated with VLX1570 relative to vehicle-treated mice (Figure 7b). These data provide first evidence on the in vivo antitumor activity profile, safety and survival benefit associated with VLX1570 treatment of WM-xenografted mice, as well as the drugs ability to modulate proteins critical for WM cell survival.
VLX1570 and ibrutinib display synergistic activity in decreasing IR WM cell viability Compared with monotherapy, drug combination therapy is known to yield improved antitumor activity as well as counter drug (f) A third method of quantifying tumor burden was used by measuring human IgM secreted in the mouse serum from the xenografted Luc-RPCI-WM1 cells, which was significantly lower in VLX1570-treated mice (P = 0.029). (g) Body weight measurements were conducted everyday and showed insignificant weight loss in some mice by~day 37, with the overall (VLX1570) treatment regimen tolerated by mice. (h) Kaplan-Meir analysis was conducted to determine overall survival and showed that mice treated with VLX1570 survived an average of~25 days longer than mice treated with vehicle alone (P = 0.00082).
resistance. We addressed the question of whether combining VLX1570 with another agent could induce cytotoxicity in IR WM cells and overcome their insensitivity to single-agent ibrutinib. We treated RPCI-WM1/IR cells with different concentrations of VLX1570+ibrutinib for 24 h followed by measurement of cell viability. This combination induced a synergistic reaction (Figure 8a) in five out of nine dose permutations tested (median CI: 0.3; range: 0.2-0.9) (Figure 8b) and significantly reduced tumor cell viability with the fraction of affected cells (Fa) ranging between 0.3 and 0.81 (30-81% cell death; median Fa: 0.49) (Figure 8c). These data suggest the ability of VLX1570 and ibrutinib to successfully synergize at various combinatorial concentrations to induce optimal cell death in WM cells that are resistant to ibrutinib monotherapy.

DISCUSSION
There are~98 DUBs encoded by the human genome, which are distributed into six families based on sequence and structural similarities. 45 The 19 S regulatory particle of the proteasome harbors three DUBs, each exhibiting notable differences in homology. 17 Of the three, USP14 and UCHL5 reversibly associate with the proteasome, whereas the third, RPN11/POH1, is an intrinsic component of the lid subcomplex of the 19 S cap. 18 The role of USP14 and UCHL5 in maintaining cancer cell viability and providing growth advantage is increasingly being recognized in a variety of cancers, including hematologic malignancies. 30 Although the precise molecular contributions made by these two DUBs towards tumor cell survival continue to be studied, 34,[46][47][48][49][50] efforts to target their dysregulated (and often increased) activity are underway, notably in MM and WM. 23,30 Herein, we extend our initial observations that USP14 and UCHL5 are highly expressed in drug-resistant WM tumor cells including those from WM patients as compared with PBMCs or resident bone marrow cells from healthy donors. Affirming their functional significance, we observed that targeted codisruption of both DUBs with the first in class USP14/UCHL5 inhibitor VLX1570 leads to robust in vitro, ex vivo and in vivo WM-specific apoptotic cell death.
VLX1570 was developed from b-AP15, which was identified in a screen for compounds that induce the lysosomal apoptosis pathway. 30 To make b-AP15 (a piperdine-based chemical) suitable for administration in humans, med-chem techniques were used to create several novel derivatives. It was found that analogs containing a central azapene ring possessed greater biological activity, of which VLX1570 emerged as the most potent. 31 Our in silico docking analysis uncovered potential structural differences between the two molecules (b-AP15 and VLX1570) that account for the improved activity of VLX1570. VLX1570 maintains better pipi-cloud planarity with nearest neighbors that exhibit pi-cloud orbital mechanics and its overall position has a better pocket occupancy than b-AP15. From energy assessment of its structure (minimization and foldX), the dihedral and thiol (-SH) arrangement appear more stable in VLX1570. Lesser water would be allowed to penetrate into the region where the alkane/alkene from the drug abuts to the protein's (USP14 and UCHL5) hydrophobic region of the binding pocket. Indeed, we observed enhanced performance of VLX1570 in virtually all cell-based assays (apoptosis, MOMP and protein modulation) where lower concentrations (~250 nM) sufficiently induced changes that were nearly equivalent to when twice the concentration of b-AP15 (~500-1000 nM) was used in identical assays for the same duration. Notably, this phenomenon was preserved in WM cells resistant to bortezomib or ibrutinib.
The activity profile of VLX1570 in BR WM cells is intriguing and the exact nature of why these cells retain sensitivity to proteasome blockade at the 19 S cap but resist downstream 20 S core inhibition remains enigmatic. One potential reason for this differential activity may be because of the type of polyubiquitinated proteins that accumulate in the cytosol of VLX1570-vs bortezomib-treated cells. Our own comparative analysis herein, using the two drugs side by side and measuring ubiquitin-conjugated protein in BR WM cells, appeared consistent with prior reports that USP14/UCHL5 inhibition (with b-AP15) produces buildup of higher-molecular-weight (⩾400 kDa) proteins than does bortezomib (~260-290 kDa). 30,33 It is conceivable that buildup of these higher kDa proteins after VLX1570 treatment overload the cells clearance capacity in a much shorter duration (within 3 h), thereby committing the cells towards apoptosis-an effect that is not achieved by the same concentration/time exposure with bortezomib.
The activity of VLX1570 was also observable in IR WM cells, an encouraging and highly desirable property for a new drug (being developed for WM) to possess. Notably, these IR WM cells lack both BTK C481S or CXCR4 WHIM mutations and display~5-fold apoptosis resistance towards ibrutinib. 26,27,51 Although 20 S PI such as bortezomib have been shown to modulate levels of TEC family kinases, including BTK, ITK, Bmx and Tec, as well as other BCR-signalosome components such as MYD88, 38,39 we questioned whether this effect would be similar with 19 S (USP14/UCHL5) inhibition. As anticipated, coinhibition of the 19 S DUBs reduced the transcriptional and translational expression of BTK and MYD88. Both are actively involved in transmission of signaling towards their end effectors NFATC2 and NF-κB, 40 levels of which are decreased following VLX1570 treatment. Although we examined NF-κB translocation as a downstream function of BCR signaling, it may be possible that the inverse holds true where BTK levels are reduced owing to downregulation of NF-κB by VLX1570. NF-κB signaling regulates BTK expression and Yu et al. 38 have shown that BTK uses a positive autoregulatory feedback mechanism to stimulate transcription from its own promoter via NF-κB. The effect of NF-κB downregulation on MYD88 is more elusive. However, a screen for genes 52 that when perturbed result in decreased MYD88 transcription revealed both Rela and Relb (NF-κB components) as two of several genes that negatively regulate MYD88 levels (Supplementary Table 5). Ongoing analysis will delineate the significance of this finding and to the best of our knowledge we are first to show this phenomenon occuring in malignant B cells. Overall, these mechanisms are in alignment with our previous findings on USP14/UCHL5 inhibition and its direct downregulation of NF-κB gene activation. 23 The reduced expression of CXCR4 is also of interest and its deubiquitination by USP14 is necessary for the chemokine receptors degradation and bioavailability. 20 This would explain decreased surface expression; however, we also observed a significant decrease in USP14 mRNA levels. A plausible explanation for this event once again may be linked to NF-κB, where the p50 and p65 subcomplexes bind the CXCR4 gene promoter region and regulate its activation resulting in tumor cell migration. 53 Yet, whether NF-κB is directly or indirectly responsible for decreasing CXCR4 in malignant B cells remains to be further studied.
To assess the in vivo anti-WM capabilities of VLX1570, we used our xenograft murine model of aggressive human WM, using the RPCI-WM1 cell line. RPCI-WM1 was developed from a bortezomib/ rituximab refractory WM patient and displays inherent apoptosis resistance towards ibrutinib in vitro. 44,54 As described in a prior report, robust activity of USP14/UCHL5 inhibition with b-AP15 was observed in delaying the outgrowth of tumor cells in MMxenotransplanted mice. 33 Whereas Tian et al. 33 administered b-AP15 at 4 mg/kg on a consecutive day schedule for 14 days, owing to the improved chemistry of VLX1570, we opted for a similar dose but on an alternate day schedule consisting of 12 treatments. This regimen was well tolerated by the mice and systemic drug activity was observed by several parameters including reduction of circulating human IgM in mice sera, decreased expression of CXCR4 and K-48-linked ubiquitinated protein levels, with the latter two attesting to in vivo target specificity of VLX1570. In concert, these data provide further validation of USP14 and UCHL5 as bonafide targets in WM and the ability of VLX1570 to reduce WM cell growth in tumor bearing mice while extending their survival.
Our preclinical analyses demonstrate the robust activity of VLX1570 alone as well as in combination with another therapeutic. We opted to test pharmacologic synergy of VLX1570 with the standard of WM care agent ibrutinib as it stood to reason not only from a clinical developmental standpoint but also from a biological perspective. Each agent effectively disrupts separate modules of a common axis that results in downregulation of NF-κB; thus, we hypothesized that simultaneous disruption of this oncogenic axis at the 19 S proteasome (via VLX1570) and directly at the BCR/MY88 L265P module (via ibrutinib) would result in enhanced tumor kill. A second interest was to determine if this combination would be effective in WM cells resistant to ibrutinib. Our experiments demonstrate synergy between VLX1570 and ibrutinib, supporting the potential benefit of adding VLX1570 to an ibrutinib-based regimen to potentially delay (or prevent) onset of resistance to ibrutinib as well as benefit existing patients already on ibrutinib but which begin to exhibit insensitivity to the BTK inhibitor.
In summary, we present compelling data on the targetability of the 19 S proteasome-associated DUBs, USP14 and UCHL5 with the use of VLX1570, which induces tumor-specific cell death, particularly in WM cells resistant to bortezomib or ibrutinib. The proteasome remains a well-validated, yet underdeveloped target in oncology. VLX1570 is a first in class USP14/UCHL5-specific inhibitor, which modulates biocellular pathways (i.e. BCR signaling components) that are essential for WM cell survival and whose cellular effects are in many ways distinct from 20 S PI such as bortezomib. In vitro, in vivo and in silico data presented in this report lay the framework for further development of VLX1570 for clinical use in patients with WM.