RUVBL1/RUVBL2 ATPase Activity Drives PAQosome Maturation, DNA Replication and Radioresistance in Lung Cancer

SUMMARY

RUVBL1 and RUVBL2 (collectively RUVBL1/2) are essential AAA+ ATPases that function as co-chaperones and have been implicated in cancer. Here we investigated the molecular and phenotypic role of RUVBL1/2 ATPase activity in non-small cell lung cancer (NSCLC). We find that RUVBL1/2 are overexpressed in NSCLC patient tumors, with high expression associated with poor survival. Utilizing a specific inhibitor of RUVBL1/2 ATPase activity, we show that RUVBL1/2 ATPase activity is necessary for the maturation or dissociation of the PAQosome, a large RUVBL1/2-dependent multiprotein complex. We also show that RUVBL1/2 have roles in DNA replication, as inhibition of its ATPase activity can cause S-phase arrest, which culminates in cancer cell death via replication catastrophe. While in vivo pharmacological inhibition of RUVBL1/2 results in modest antitumor activity, it synergizes with radiation in NSCLC, but not normal cells, an attractive property for future preclinical development.

In Brief

Yenerall et al. identified a specific inhibitor of RUVBL1/2 ATPase activity, compound B, and demonstrate that RUVBL1/2 ATPase activity is required for PAQosome maturation/dissociation. Compound B kills non-small cell lung cancer (NSCLC) by inhibiting DNA replication. In addition, compound B radiosensitizes NSCLC, but not normal cells, an attractive property for future development.

Graphical Abstract

INTRODUCTION

In humans, RUVBL1 and RUVBL2 (also referred to as pontin/TIP49 and reptin/TIP48, respectively) are paralogous proteins of the AAA+ ATPase family with homology to the bacterial RuvB helicase (Parsons and West, 1993). RUVBL1 and RUVBL2 depend upon each other for stability and form ring-shaped heterohexamers that display ATPase activity (Lakomek et al., 2015; Lopez-Perrote et al., 2012). Due to their obligate nature, we refer to RUVBL1 and RUVBL2 collectively as RUVBL1/2. RUVBL1/2 are involved in the formation of two multiprotein complexes, the PAQosome (Houry et al., 2018) and the INO80 family of chromatin remodelers (Jha et al., 2013; Jonsson et al., 2004). Through the PAQosome, RUVBL1/2 controls the stability of the phosphatidylinositol 3-kinase-related kinases (PIKK) family of proteins (Izumi et al., 2010), and small nuclear ribonucleoprotein (snRNP) and small nucleolar ribonucleoprotein (snoRNP) biogenesis (Bizarro et al., 2015; McKeegan et al., 2009). Through the INO80 family of chromatin remodelers, RUVBL1/2 affects nucleosome positioning via INO80 (Jonsson et al., 2004), histone acetylation via TIP60 (Jha et al., 2013), and histone composition via SRCAP (Wong et al., 2007). RUVBL1/2 are essential for these complexes; however, the role of RUVBL1/2 ATPase activity in these complexes remains unknown.

RUVBL1/2 are essential for cellular proliferation from yeast (Qiu et al., 1998) to humans (Munoz et al., 2016) and have been implicated as oncogenic proteins in various cancers (Breig et al., 2014; Fan et al., 2017; Guo et al., 2018; Lauscher et al., 2007; Osaki et al., 2013; Rousseau et al., 2007; Yuan et al., 2016). In addition, an inhibitor of RUVBL1/2 has recently been developed for cancer therapy, highlighting their interest as therapeutic targets in this disease (Assimon et al., 2019). The precise mechanisms underlying the essentiality of RUVBL1/2 for these cancers, however, is unclear. In addition, our mechanistic understanding of the functional roles for RUVBL1/2 remain scant: What are the roles of ATP binding and hydrolysis for RUVBL1/2 activity, the molecular basis for the essential role of RUVBL1/2 in growth, and are RUVBL1/2 tractable targets in cancer?

To better understand the functions of RUVBL1/2 ATPase activity in cancer, we examined the consequences of inhibiting RUVBL1/2 ATPase activity in non-small cell lung cancer (NSCLC) by using a specific and potent inhibitor of RUVBL1/2. Identification of mutations that confer resistance to this inhibitor identified regions within RUVBL1/2 that may be crucial for ATP hydrolysis and demonstrated on-target specificity. Looking at RUVBL1/2-dependent complex formation, we found that the ATPase activity of RUVBL1/2 is necessary for the dissociation and/or maturation of the PAQosome but not the INO80-family chromatin remodelers. We then probed the mechanism underlying the essentiality of RUVBL1/2 ATPase activity for NSCLC and show that most patient-derived NSCLC lines require RUVBL1/2 ATPase activity for S-phase progression. In these cell lines, inhibition of RUVBL1/2 initially decreases the total number of active replication forks. Prolonged inhibition, however, results in cell death via replication catastrophe due to loss of ATR. Therapeutically, RUVBL1/2 inhibition as a monotherapy in NSCLC xenografts provides modest benefit due to a narrow therapeutic window. However, the combination of RUVBL1/2 inhibition with ionizing radiation (IR) is synergistic in killing tumor, but not normal cells. Thus, the greatest therapeutic potential for targeting RUVBL1/2 in NSCLC may lie in combination with IR.

RESULTS

High Expression of RUVBL1/2 in NSCLC Tumors Is Associated with Poor Prognosis

We first examined RUVBL1/2 expression in NSCLC patient tumors and their association with patient prognosis. While RUVBL1/2 are not frequently amplified or mutated, RUVBL1/2 mRNAs are overexpressed in NSCLC (Figure S1A). KMplotter (Szasz et al., 2016) showed that high RUVBL1 or RUVBL2 expression was associated with poor overall survival (Figure S1B) and multivariate analyses indicated this was independent of tumor histology, grade, and patient gender (p = 0.0012 for RUVBL1, p = 0.0004 for RUVBL2, Cox regression). We next validated the performance of an antibody against RUVBL1 (Figure S1C) and then analyzed RUVBL1 protein levels in two independent clinically annotated NSCLC tissue microarrays using immunohistochemistry (Figures S1D and S1E). Similar to high RUVBL1/2 mRNA levels (Figure S1B), high RUVBL1 protein levels were associated with poorer overall survival in NSCLC (Figure S1F). In addition, RUVBL1 protein level was independent of tumor stage and oncogenotype but was higher in squamous NSCLCs and tumors from smokers (Figures S1G–S1J).

RUVBL1/2 ATPase Activity Is Required for NSCLC Growth

To test the dependency of NSCLC on RUVBL1/2 for growth, we depleted RUVBL1 or RUVBL2 using small interfering RNAs (siRNAs) and endoribonuclease-prepared siRNAs (esiRNAs) (Kittler et al., 2004, 2007; Yang et al., 2002) in 24 NSCLC lines and measured cell growth. We found that the growth of all NSCLC lines was affected by RUVBL1/2 depletion (Figure 1A), and this was independent of their total level of RUVBL1/2 protein (Figure S1K). RUVBL1 or RUVBL2 knockdown produced similar effects (Figure S1L). In addition, by stably expressing an RNAi-resistant RUVBL1 or RUVBL2 cDNA, we were able to rescue the effects of all RNAi reagents utilized (Figures S1M and S1N).

Figure 1. RUVBL1/2 ATPase Activity Is Required for NSCLC Viability.

(A) Effect of RUVBL1/2 knockdown on the growth of NSCLC lines. Values for RNAI-RUVBL1 or RNAI-RUVBL2 represent the average relative number of cells remaining 5 days after knockdown of RUVBL1 or RUVBL2, respectively, between an siRNA and an esiRNA, in comparison with control oligonucleotides. Knockdowns performed with ≥2 biological replicates, and error bars represent the SD between siRNAs and esiRNAs

(B) Expression of wild-type doxycycline-inducible siRNA-resistant RUVBL1/2 cDNA can rescue growth following 5 days of RUVBL1/2 knockdown (left) but ATPase-dead cannot (right) in NSCLC lines. Dox = 2 μg/mL doxycycline. Values are averages ±SD of ≥2 biological replicates.

Molecularly, RUVBL1/2 require their ATPase activity for some (Izumi et al., 2010; Zaarur et al., 2015), but not all (Jha et al., 2013; Jonsson et al., 2004), of their functions. Phenotypically, however, RUVBL1/2 ATPase activity may be essential for proliferation (Gentili et al., 2015; Grigoletto et al., 2013). To determine if NSCLC requires RUVBL1/2 ATPase activity for growth, we depleted endogenous RUVBL1 or RUVBL2, expressed siRNA-resistant wild-type or ATPase-dead RUVBL1/2, and measured their ability to rescue cell growth. Only wild-type RUVBL1/2 protein was able to rescue growth following RUVBL1/2 knockdown, despite protein levels being restored by both cDNAs (Figure S2A), demonstrating the necessity of RUVBL1/2 ATPase activity for NSCLC growth (Figure 1B).

Characterization of Compound B as a Specific RUVBL1/2 ATPase Inhibitor

A custom library (Daiichi-Sankyo) of ~400,000 compounds was screened for their ability to inhibit the ATPase activity of purified RUVBL1 and RUVBL2 protein. The lead compound from this screen was further modified, and these molecules were rescreened (Ebisawa et al., 2017). This secondary screen yielded compound B (Figure 2A, top, compound 40 from WO 2015/125785 A1), which has a half maximal inhibitory concentration (IC₅₀) of 59 nM for purified RUVBL1/2 (Figure S2B), and an inactive enantiomer of compound B, compound C (Figures 2A and S2B). RNA sequencing (RNA-seq) after 24 h of compound C treatment induces no significant transcriptional responses (Figure S2C). Therefore, here we use compound C as an enantiomer control for compound B. Compound B inhibits RUVBL1/2 in cells, as treatment with compound B depleted PIKK-family members DNA-PK_CS (PRKDC), ATM, and ATR (Figure 2B), all of which require RUVBL1/2 ATPase activity for stability (Izumi et al., 2010).

Figure 2. Development and Validation of Compound B, an Orally Bioavailable RUVBL1/2 Inhibitor.

(A) Structure of compound B (active RUVBL1/2 inhibitor, top) and compound C (inactive control compound, bottom).

(B) Immunobloting of PIKK family members following 3 days of 100 nM Compound B (abbreviated Comp. B) or Compound C (abbreviated Comp. C) and 4 days of siRNA-mediated RUVBL1 or RUVBL2 knockdown in H2009.

(C–E) Orally delivered compound B inhibits RUVBL1/2 in tumors in vivo. (C and D) Immunoblot of tumor protein extracts from mice bearing H2009 (C) or H596

(D) xenografted subcutaneously into NOD/SCID mice and treated with vehicle or 175 mg/kg/day Compound B by oral gavage for 3 days. Each number is an independent mouse. (E) Part of the H2009 tumors from Figure 2C were analyzed for tumor-cell specific levels of ATM by IHC. Values are averages ±SD of the percent of ATM-positive tumor cells in 2 mice. Scale bars, 200 μm.

(F) PCR-mutagenesis identifies amino acid substitutions in RUVBL1 or RUVBL2 that confer resistance to compound B. Lollipop diagrams generated using cBioPortal (https://www.cbioportal.org/mutation_mapper) and RUVBL1/2 domains were overlaid. Enrichment refers to the frequency a residue was mutated in cells surviving compound B relative to the parental population. Substitutions >10-fold enriched are shown. See also Table S1.

(G) Compound B resistance-conferring mutations cluster at RUVBL1/2 interfaces. RUVBL1/RUVBL2 from Chaetomium thermophilum (PDB: 4WVY). Domains are depicted in cartoon and colored as in Figure 2F. ATP, black stick; resistance-conferring mutations, colored spheres. The RUVBL2/RUVBL1 interaction surface (RUVBL1, salmon) includes nine resistance-conferring mutations (red spheres), while RUVBL1/RUVBL2 interaction surface (RUVBL2, pink) includes three resistance-conferring mutations (red spheres). Remaining resistance-conferring mutations are white.

(H) Superposition of the ATP/apo (PDB: 4WVY) and ADP/ADP (PDB: 4WW4) bound states of the RUVBL1/RUVBL2 interface highlights nucleotide-dependent switch loop conformations in RUVBL1 (magenta/pink thick loops for ATP/ADP states, respectively) and RUVBL2 (red/salmon thick loops for apo/ADP states, respectively) that span both interfaces. Compound B resistance mutations are spheres.

(I) A flexible RUVBL2 N-terminal gatekeeper (yellow thick loop) from human RUVBL1/RUVBL2 bound to R2TP (PDB: 6QI8) is revealed by superposition of RUVBL1/RUVBL2 heterodimers from the nucleotide bound (ordered switch and gatekeeper loops) and apo (ordered elements in transparent surface with disordered loops) states. The ordered N terminus completes the RUVBL1/RUVBL2 interaction surface and compound B resistance mutations are spheres.

(J) Identified mutations in RUVBL1 rescue the ability of compound B to inhibit RUVBL1/2 in vitro. Values are averages ±SD of three biological replicates.

(K and L) Doxycycline-inducible expression of compound B resistance-conferring mutations in RUVBL1/2 confers resistance to compound B in viability-based assays. (K) RUVBL1 and (L) RUVBL2. See also Figures S2H and S2I for an additional cell line.

Next, we assessed compound B in vivo. Compound B accumulated in plasma in a dose-dependent manner, with linear increases in C_max (Figure S2D) and area under the curve (Figure S2E) for increasing doses of orally administered compound B. Administration of compound B by oral gavage to mice bearing subcutaneous NSCLC xenografts inhibited RUVBL1/2, as compound B treatment reduced the levels of ATM, ATR, and DNA-PK_CS (Figures 2C and 2D). In addition, immunohistochemistry (IHC) for ATM confirmed that ATM was depleted in tumor cells (Figure 2E).

We next sought to determine if compound B had significant off-target activity. For compound B-like molecules from our secondary screen, there was a relationship between target inhibition and growth inhibition (Figure S2B). There was also a significant correlation between sensitivity to genetic and chemical inhibition of RUVBL1/2 (Figure S2F). Since RUVBL1/2 hydrolyzes ATP, we also tested the ability of compound B to inhibit the in vitro activity of kinases, which also hydrolyze ATP. Compound B did not significantly inhibit any of the kinases tested (Figure S2G).

To definitively determine whether compound B kills cancer cells in an on-target manner, we sought to identify mutations in RUVBL1 or RUVBL2 that may confer resistance to compound B. To do this, we performed a screen where we randomly mutagenized RUVBL1 or RUVBL2 cDNAs using error-prone Taq polymerase, expressed these mutant alleles in H2009, and then treated these cells with a lethal dose of compound B. For surviving cells, their RUVBL1 or RUVBL2 cDNA was sequenced, and the frequency of single nucleotide variants (SNVs) in RUVBL1 or RUVBL2 between parental and surviving cells was compared. Similar to another recently reported RUVBL1/2 inhibitor, a structurally similar daughter molecule of compound B, known as CB-6644 (Assimon et al., 2019), we found that mutations in RUVBL1 or RUVBL2 could confer resistance to compound B, with some overlap (RUVBL1-A62T, RUVBL1-R117Q, RUVBL1-R276Q, and RUVBL1-R317Q) between studies (Table S1).

These resistance-conferring mutations do not obviously cluster in 2D space (Figure 2F); however, when mapped to the 3D RUVBL1/RUVBL2 structure, they cluster predominantly at the interfaces between RUVBL1/RUVBL2 and RUVBL2/RUVBL1 (Figure 2G, 12 mutations). Comparison of the ATP- and ADP-bound states of RUVBL1/2 reveal mutations in and surrounding a flexible switch loop in both RUVBL1 and RUVBL2 that interacts at the RUVBL1/2 interface (Figure 2H, nine mutations). This switch loop adopts nucleotide-dependent conformations, which is interesting given that CB-6644, a daughter molecule of compound B, only binds to ATP-bound RUVBL1/2 (Assimon et al., 2019). Similarly, two of the remaining mutations from RUVBL1 map near the flexible N terminus of RUVBL2 (Figure 2I), which is reported to act as a gatekeeper to allow nucleotide exchange in the R2TP complex (Munoz-Hernandez et al., 2019). Thus, most putative resistance-conferring mutations that could be mapped to the RUVBL1/RUVBL2 structure (15 of 18) map to the RUVBL1/RUVBL2/RUVBL1 interfaces or position flexible interface loops whose orientation is dependent on the nucleotide-bound state. Multiple mechanisms may explain how these mutations confer resistance to compound B. It is possible that some of these mutations disfavor the ATP-bound state to which the drug presumably binds, as some resistance-conferring mutations (such as RUVBL1-E110K and RUVBL1-K265E) form hydrogen bonds in the ATP-bound state not present in the ADP-bound state and may destabilize the ATP-bound conformation. Alternatively, some solvent-exposed mutations (such as RUVBL1-P296S and RUVBL1-G63E) may alter the physical binding site of compound B.

To experimentally validate these putative resistance-conferring mutations, we introduced some of these mutations into RUVBL1 or RUVBL2. For two mutations, we purified mutant RUVBL1 protein and found that these mutants, when mixed with wild-type RUVBL2, rescued the in vitro ATPase activity of RUVBL1/2 after compound B treatment (Figure 2J). To test whether these mutations could rescue cell growth following compound B treatment, we cloned five RUVBL1 and two RUVBL2 mutations into doxycycline-inducible vectors, transduced them into H2009 cells, and then measured sensitivity to compound B with or without doxycycline. All these mutations in RUVBL1 (Figure 2K) or RUVBL2 (Figure 2L) were able to rescue the growth defect from compound B treatment in H2009, as well as in an additional NSCLC cell line (Figures S2H and S2I), strongly suggesting that compound B kills NSCLC due to the inhibition of RUVBL1/2.

The ATPase Activity of RUVBL1/2 May Be Necessary for Most Functions of RUVBL1/2

To assess the importance of RUVBL1/2 ATPase activity in cellular functions at a global level, we analyzed the transcriptomic effects of RUVBL1 knockdown or RUVBL1/2 ATPase inhibition via RNA-seq in two NSCLC lines, H2009 and H596. Treatment of H596 with compound B or siRUVBL1 caused fewer genes to be differentially expressed (DE) than each respective treatment in H2009 (Figure 3A). Regardless, we found considerable overlap in DE genes, both between siRUVBL1 and compound B in H2009 (Figure 3B), and between H2009 and H596 following treatment with compound B (Figure 3C). This suggests that many cellular functions of RUVBL1/2 require ATPase activity, but does not rule out the possibility of ATPase-independent functions.

Figure 3. Molecular Consequences of Acute RUVBL1/2 ATPase Inhibition.

(A) MA plots of RNA-seq data from H2009 and H596 (top and bottom panels, respectively) treated with compound B for 24 h (left two panels) or siRUVBL1 for 3 days (right two panels). Colored dots, differentially expressed (DE) genes (RPKM >1, p < 0.01, false discovery rate [FDR] < 0.05, and fold change >2). Upregulated DE genes, red; downregulated DE genes, green. Values are averages of two biological replicates.

(B and C) Venn diagrams of DE genes (upregulated left, downregulated right) from (A) after compound B and siRUVBL1 treatment in (B) H2009 and (C) H596. Significance assessed by hypergeometric test.

(D) Inhibition of RUVBL1/2 ATPase increases PAQosome association with RUVBL1/2. Top: RUVBL1-V5 was immunoprecipitated from H2009 cells after 12 h of 100 nM of compound C or compound B treatment, and coimmunoprecipitated proteins were quantified by mass spectrometry. Values represent average relative abundance between compound B and compound C treatment ±SD of three biological replicates. Bottom: validation by independent immunoprecipitation-immunoblotting. See also Table S3.

(E) Proposed model of how RUVBL1/2 ATPase activity affects maturation and/or disassembly of the PAQosome complex.

Acute Inhibition of RUVBL1/2 ATPase Activity Inhibits PAQosome Maturation

Because RUVBL1/2 are required for the function of many multiprotein complexes, we interrogated the effects of RUVBL1/2 ATPase inhibition on the formation of RUVBL1/2-containing protein complexes. To do this, we immunoprecipitated RUVBL1 from cells following compound B or compound C treatment and then subjected coimmunoprecipitating proteins to mass spectrometry (Table S2). We found that components of the PAQosome complex and the TTT complex all had increased interactions with RUVBL1 after compound B treatment (Figure 3D, top). This suggests that the PAQosome can assemble after the inhibition of RUVBL1/2 ATPase activity, but is unable to dissociate and/or mature. While compound B increases the binding of PAQosome components to RUVBL1/2, as well as some PAQosome substrates, it does not affect the binding of the INO80 family of chromatin remodelers to RUVBL1/2 (Figure S3). Immunoprecipitation of RUVBL1-V5 after various lengths of compound B treatment followed by immunoblotting confirmed these findings and showed that this effect occurred in a time-dependent manner (Figure 3D, bottom). In addition, following compound B treatment, we found that PIKK-family proteins, which require the PAQosome for their stability, increased binding to RUVBL1/2 while conversely being depleted from total cellular lysate in a time-dependent manner (Figure 3D, bottom). We propose that RUVBL1/2 ATPase activity is necessary for the proper maturation and/or dissociation of the PAQosome complex and its client proteins, as illustrated in Figure 3E.

RUVBL1/2 Have Roles in DNA Replication

We sought to understand the mechanisms underlying the essential role of RUVBL1/2 for cellular viability in NSCLC. Because RUVBL1/2 have been implicated in numerous essential cellular processes, we first conducted unbiased analyses to nominate which pathways may be the most integral to the role RUVBL1/2 plays in supporting viability in NSCLC. First, we performed gene set enrichment analysis (GSEA) (Subramanian et al., 2005) on RNA-seq data following compound B or siRUVBL1 treatment and found that transcripts involved in DNA replication were frequently downregulated after genetic or pharmacological inhibition of RUVBL1/2 (Figure 4A). Next, we assessed the potential function of RUVBL1/2 in NSCLC patient tumors by gene coexpression analysis, which can infer putative gene function because coexpressed genes often act within the same pathway (Lee et al., 2004). GSEA analysis of the correlation between the expression of RUVBL1/2 and every gene in the TCGA NSCLC RNA-seq dataset showed that RUVBL1/2 were coexpressed with genes involved in DNA replication (Figure 4B). In addition, by immunoprecipitating RUVBL1-V5 or RUVBL2-V5 and performing mass spectrometry on coimmunoprecipitating proteins, we found an enrichment of replication factors in the proteins that coimmunoprecipitate with RUVBL1/2 (Figure S4A, Table S3). As a final correlative analysis, we compared the pattern of sensitivity, in regard to cellular growth, of RUVBL1/2 depletion to the pattern of sensitivity to depletion of cellular processes in lung cancer. To do this, we correlated the redundant siRNA activity (RSA) score of RUVBL1/2 in a small hairpin RNA (shRNA)-based drop-out screen, Project DRIVE (McDonald et al., 2017), to the score of every other gene tested in lung cancer cell lines, rank ordered this list, and analyzed it with GSEA. We found that the pattern of sensitivity to RUVBL1/2 knockdown in lung cancer cell lines strongly correlated with the pattern of sensitivity to knockdown of DNA replication factors (Figure 4C). In totality, these results suggest NSCLC requires RUVBL1/2 for viability due to their roles in DNA replication.

Figure 4. RUVBL1/2 Have Roles in DNA Replication.

(A) GSEA of RNA-seq data following compound B treatment (top) or RUVBL1 knockdown (bottom) in H2009. NES, normalized enrichment score; FDR, false discovery rate.

(B) RUVBL1/2 are coexpressed with DNA replication genes in NSCLC patient tumors. GSEA of the Pearson correlation between RUVBL1 (top) or RUVBL2 (bottom) expression and the expression of every gene in the TCGA NSCLC RNA-seq dataset.

(C) Sensitivity to knockdown of RUVBL1/2 correlates with sensitivity to knockdown of DNA replication factors. GSEA of the Pearson correlation between the average RSA score of RUVBL1 and RUVBL2 and the RSA score of 7,726 genes in 63 lung cancer cell lines from Project DRIVE.

(D) Fitted viability-based dose-response curves for cancer cell lines treated with compound B (67 cell lines) or compound C (32 cell lines) for 4 days. All data represent the average ±SD and most assays have ≥2 biological replicates. See Table S4 for more information.

(E) NSCLC lines that are more sensitive to compound B tend to arrest in S phase. Cell-cycle analysis by DNA content following 48 h of exposure to 100 nM compound B. Representative of at least two biological replicates. See Figure S4B for additional NSCLC lines.

(F) Time course of the effects of 100 nM compound B on various proteins by immunoblot in H2009 (left) and H596 (right). See Figure S4G for immunoblot of same proteins following RUVBL1/2 knockdown.

(G) Top: representative images from immunofluorescent microscopy of H2009 (left) and H596 (right) after 100 nM compound B or compound C for 48 h. DAPI, blue; γH2AX, red. Bottom: quantification of pan-γH2AX-positive cells, defined as nuclei covered in γH2AX staining. Values represent averages ±SD from three biological replicates. Scale bars, 20 μm. See also Figure S4F for γH2AX and 53BP1 foci data.

(H) Compound B slows progression through S phase. Top: schematic of experiment outline. Bottom: flow cytometric analysis of BrdU incorporation and DNA content (PI, propidium iodide) in H2009 cells following release into S phase from a double thymidine block. Representative of three biological replicates. See Figure S5A for analysis of BrdU only.

In line with RUVBL1/2 being essential for cellular viability, the ATPase activity of RUVBL1/2 is also broadly essential for cellular viability, as compound B reduced the viability of 67 cancer cell lines from diverse lineages with IC₅₀ values ranging from 11 to 198 nM (Figure 4D, Table S4). Because our analyses implicated RUVBL1/2 in DNA replication, we performed DNA content analysis after compound B treatment. This showed that NSCLC lines that were more sensitive to compound B displayed S-phase arrest (Figures 4E and S4B). Genetic depletion of RUVBL1 also resulted in impaired S-phase progression (Figure S4C). We confirmed that cells displaying S-phase arrest by DNA content analysis did not reach mitosis, as they had decreased levels of phosphorylated histone H3 (S10) (Figures S4D and S4E). Because loss of RUVBL1/2 impaired S-phase progression, we measured the levels of phosphorylated H2AX (S139, referred to as γH2AX), a general marker of DNA damage or replication stress. Compound B treatment increased the total level of γH2AX, but only after 48 h of treatment in cell lines that displayed S-phase arrest (H2009, Figure 4F). Compound B did not increase the number of γH2AX or 53BP1 foci, both of which are generally indicative of DNA double-strand breaks (DSBs) (Figure S4F). However, there was a significant increase in the number of cells displaying nuclear pan-positive γH2AX staining (H2009, Figure 4G), which is generally indicative of cells undergoing replicative stress (Toledo et al., 2011). The S-phase arrest following compound B treatment eventually culminates in apoptosis, as indicated by the sub-G1 population in DNA content analysis (Figure 4E) and cleaved PARP1 in immunoblots (Figure 4F). Similarly, knockdown of RUVBL1/2 in H2009 cells resulted in cells with sub-G1 DNA content (Figure S4C), increased total γH2AX levels, and increased cleavage of PARP1 (Figure S4G). To further demonstrate that RUVBL1/2 inhibition impairs DNA replication, we synchronized H2009 cells at the G1/S-phase boundary, treated with compound B or compound C, released them into S phase, and monitored S-phase progression by BrdU incorporation and DNA content. While cells treated with compound B initially replicate their DNA at a normal rate (see 6-h time point, Figure 4H), they ultimately continue to synthesize DNA and stay in S phase for a longer period of time than compound C-treated cells (see 12- and 18-h time points, Figure 4H). The finding that compound B did not inhibit bulk DNA replication suggests that replication forks may stall or not fire properly at certain regions of the genome when RUVBL1/2 ATPase activity is inhibited.

Compound B Treatment Reduces the Total Number of Replication Forks

While synchronized compound B-treated cells displayed a prolonged S phase, BrdU incorporation rate did not change dramatically (Figure S5A). To better understand how RUVBL1/2 inhibition affects DNA replication, we measured the amount of single-stranded DNA (ssDNA) in individual cells. In general, ssDNA is rapidly coated by the Replication Protein A (RPA) complex, and the amount of RPA-bound ssDNA usually increases after replication stress, with high levels triggering the replication stress response (Zou and Elledge, 2003). Naked ssDNA is predominantly found either transiently at the replication fork or during extreme cases of replication stress (Toledo et al., 2013). Cells in S/G2/M that were treated for 12 h with compound B had decreased amounts of ssDNA (bottom gates, Figure 5A). However, upon longer treatments, these cells began to display high levels of ssDNA and progress into apoptosis (high ssDNA, sub-G1 DNA content, top gates, Figure 5A). Analysis of RPA-bound chromatin showed similar results (Figure 5B).

Figure 5. RUVBL1/2 Inhibition Decreases the Number of Active Replication Forks and Causes Replication Catastrophe.

(A) Compound B treatment lowers, then increases, the amount of ssDNA. H2009 cells were analyzed for ssDNA and total DNA (PI) by flow cytometry after 100 nM of compound B or compound C treatment for indicated times. Top graphs: numbers in upper gate represent the percentage of cells with high levels of ssDNA and bottom gates capture cells in S/G2/M. Bottom graphs: amount of ssDNA in cells in S/G2/M phase (from top graph). MFI, median fluorescent intensity; HU + ATRi, 2 mM hydroxyurea (HU) for 4 h followed by 500 nM VE-822 (ATR inhibitor) for 2 h. Numbers are average ±SD of three biological replicates.

(B) RPA loading onto chromatin parallels ssDNA. H2009 cells were pre-extracted and analyzed for chromatin-bound RPA2 and DNA content (PI) by flow cytometry after 100 nM compound B or compound C treatment for indicated times. Top graphs: numbers in upper gate represent the percentage of cells with hyperloaded RPA and bottom gates capture cells in S/G2/M. Bottom graphs: amount of chromatin-bound RPA2 in cells in S/G2/M phase (from top graph). Hydroxyurea, 2 mM for 2 h. Numbers are average ±SD of three biological replicates.

(C) Complete loss of ATR causes replication catastrophe after RUVBL1/2 inhibition. H2009 cells were pre-extracted and analyzed for chromatin-bound RPA2 and DNA content (PI) by flow cytometry after being treated with 100 nM compound B or compound C ± 80 nM VE-822. Numbers are average ±SD of three biological replicates. See Figure S5B for the same experiment with a CHEK1 inhibitor.

(D) Compound B treatment increases the speed of individual replication forks. H2009 cells were treated with 100 nM compound C or compound B, labeled with IdU and CldU, and the length of IdU and CldU tracks were measured. Only the total length of dual IdU/CldU-positive tracks were measured. Data from two biological replicates in which >100 tracks were quantified for each condition. Statistical significance from one-way ANOVA with a post-hoc Sidak’s multiple comparisons test.

(E) The abundance of core replication factors at replisomes does not change after compound B treatment. H2009 was treated with 100 nM compound C or compound B for 12 h, active replisomes were purified by iPOND, and eluted proteins were analyzed by mass spectrometry. Values are averages ± SEM from three biological replicates. See also Table S5 for the complete list of identified proteins.

(F) POLE abundance increases, whereas histone abundance decreases, at active replisomes after compound B treatment. Data are from same experiment as in.

(E). Values are averages ± SEM from three biological replicates. See also Table S5.

(G) iPOND immunoblot confirms that compound B increases the abundance of POLE2 but decreases the abundance of histones at active replisomes. No click samples lack the biotin conjugation step. Chase samples contain only chromatin-bound proteins. Asterisk, nonspecific protein; arrow, specific protein.

We sought to understand why compound B-treated cells progress from reduced to increased levels of ssDNA and RPA-bound chromatin. At 48 h, when RPA becomes hyperloaded onto chromatin, the PIKK-family member protein ATR, which is a guardian of replication forks and the major sensor of replication stress (Zeman and Cimprich, 2014), is strongly depleted (Figure 4F). We hypothesized that initially after compound B treatment, the majority of cells are arrested in S phase, potentially due to stalled replication forks or unfired origins (i.e., fewer active replication forks); after 48 h, when ATR is strongly depleted, unfired/suppressed origins fire aberrantly, resulting in the hyperloading of RPA and generation of ssDNA. This phenomenon, where replication fork stalling and subsequent ATR inhibition results in RPA hyperloading, ssDNA generation, and apoptosis, has been termed replication catastrophe (Toledo et al., 2013). To test this hypothesis, we treated cells for 24 h with compound B, then added an ATR inhibitor transiently. This induced a level of RPA hyperloading nearly identical to 48 h of compound B treatment (Figure 5C). Transient treatment with a CHEK1 inhibitor produced nearly identical effects (Figure S5B) and co-treatment of compound B with a CHEK1 inhibitor was able to induce premature apoptosis (Figure S5C). This suggests that RUVBL1/2 inhibition may initially inhibit replication fork progression or origin firing, and subsequent loss of the ATR/CHEK1 signaling axis results in replication catastrophe.

We next determined what changes occur directly at the replication fork after the inhibition of RUVBL1/2. First, we measured the speed of individual replication forks using DNA fiber analysis and found that compound B treatment increased fork velocity as early as 6 h after treatment (Figure 5D). Fork speed and the number of active replication forks are inversely correlated (Rodriguez-Acebes et al., 2018). This, combined with our data showing that compound B treatment initially decreases levels of ssDNA and RPA-bound chromatin, suggests that compound B treatment initially decreases the total number of active replication forks.

Next, we determined if compound B treatment changed the abundance of proteins directly at active replication forks in an unbiased manner. To do this, we treated cells with compound B or compound C, performed isolation of proteins on nascent DNA (iPOND) (Sirbu et al., 2012), and then quantified and compared purified replisome components between compound B and compound C treatment by mass spectrometry. Unlike hydroxyurea treatment (Dungrawala et al., 2015), we found that compound B treatment had very little effect on the abundance of most of the core components of the replisome and many replication stress-related proteins (Figure 5E). Interestingly, the abundance of POLE, which is believed to be the leading strand DNA polymerase but can replicate either strand (Yeeles et al., 2017), increases after compound B treatment, whereas the abundance of core histone proteins decreases (Figure 5F). These results were confirmed by iPOND followed by immunoblotting (Figure 5G).

RUVBL1/2 Inhibition Impairs NSCLC Tumor Growth In Vivo

Because RUVBL1/2 are ubiquitously required for cellular growth, it is unknown if RUVBL1/2 inhibitors would provide an ample therapeutic window (i.e., desired effects without unacceptable toxicities). First, to assess the necessity of RUVBL1/2 for the in vivo growth of NSCLC, we depleted RUVBL1 using doxycycline-inducible shRNAs in H1299 xenografts. Knockdown of RUVBL1 attenuated tumor growth (Figure S6A), suggesting that RUVBL1/2 are required for NSCLC tumor growth in vivo.

Next, we assessed the therapeutic potential of compound B to inhibit NSCLC tumor growth in vivo. H2009 and H596 cells were xenografted and mice were randomized to receive either vehicle or compound B at the maximum tolerated dose. Treatment with compound B significantly reduced tumor growth in both models (Figures 6A and S6B) but had a larger effect in H2009 tumors, which were also more sensitive to compound B and RUVBL1/2 knockdown in vitro (Figure 1F and Table S4). H2009 tumors treated with compound B displayed stressed and vacuolated nuclei by H&E staining (Figure S6B). While compound B was able to reduce tumor growth, it caused transient bodyweight loss in mice (Figure 6B), despite there being no marked changes in the tissue architecture or cell morphology of the kidney, liver, spleen, or small intestine (Figure S6C). These results suggest that compound B inhibits NSCLC tumor growth in vivo but may only offer a moderate therapeutic window as a monotherapy in a subset of NSCLCs.

Figure 6. Inhibition of RUVBL1/2 Therapeutically Synergizes with Ionizing Radiation.

(A) Orally delivered compound B can inhibit NSCLC tumor growth in vivo. NSCLC lines that were relatively sensitive (H2009, left) or relatively resistant (H596, right) in vitro to compound B were implanted subcutaneously into NOD/SCID mice and randomized to receive 175 mg/kg/day compound B or vehicle once tumor volumes reached ~150 mm³. Arrows, treatment days; values are averages ± SEM. Statistical significance assessed with two-way repeated measures ANOVA. n represents the number of mice. See Figure S6B for tumor pictures, weights, and H&Es.

(B) Body weight of NOD/SCID mice from (A). Values are averages ± SEM. See Figure S6C for histology of mice organs.

(C) GSEA analysis of the RNA-seq data following compound B (top) or siRUVBL1 (bottom) treatment in H2009 (from Figure 4A) shows downregulation of gene sets important to the response to IR.

(D) Doxycycline-inducible shRNAs against RUVBL1 or RUVBL2 radiosensitize NSCLC. Cells were treated for 3 days with 62.5 ng/mL doxycycline, plated as single cells, irradiated at various doses, and left in the presence of doxycycline until colonies were formed, stained, and counted. Values are average ± SD of two biological replicates.

(E) Compound B radiosensitizes NSCLC. Cells were treated for 3 days with compound B or compound C, plated as single cells, irradiated at various doses, and left in the presence of drug until colonies were formed, stained, and counted. Values are average ± SD of two biological replicates.

(F) Compound B does not radiosensitize normal human bronchial epithelial cells (HBECs). Assay performed as in (E).

RUVBL1/2 Inhibition Synergizes with Ionizing Radiation In Vitro

We searched for clinically relevant therapies that may synergize with RUVBL1/2 inhibition to help improve this modest therapeutic window. GSEA analysis of our RNA-seq data following compound B treatment or RUVBL1 knockdown showed that cell-cycle genes that were upregulated after IR were downregulated following compound B treatment, as well as genes involved in base excision repair (Figure 6C), an integral pathway in the recovery from IR (Chaudhry, 2007). In addition, our proteomic analysis suggested that RUVBL1/2 interacting proteins were enriched for proteins involved in nucleotide excision repair (Figure S4A). Further, RUVBL1/2 is necessary for the assembly of the INO80 family of chromatin modelers (Jha et al., 2013; Jonsson et al., 2004) and the abundance of the Fanconi anemia core complex (Rajendra et al., 2014), both of which are important for the recovery from IR (Dong et al., 2014; Ikura et al., 2000; Morrison et al., 2004; Walden and Deans, 2014). Thus, we speculated that RUVBL1/2 inhibition may enhance IR, which is frequently used to treat patients with NSCLC.

To test if RUVBL1/2 affects the response to IR in NSCLC, we first examined the ability of RUVBL1/2 depleted cells to survive IR. Knockdown of RUVBL1 or RUVBL2 synergistically decreased the clonogenic potential of cells following IR in two NSCLC lines (Figure 6D). Utilizing compound B, we extended these findings to eight NSCLC lines (Figure 6E) and two non-transformed human bronchial epithelial cell lines (HBECs) (Figure 6F, also see Table S6). Importantly, while all NSCLC lines were radiosensitized by compound B, two models of non-transformed lung epithelial cells, HBEC3KT and HBEC30KT, were not radiosensitized. Of note, HCC4017, an NSCLC cell line that was radiosensitized, was derived from the same patient as HBEC30KT, a normal bronchial epithelial cell line that was not radiosensitized, demonstrating that compound B may synergize with IR in NSCLC but not normal lung epithelial cells.

To better understand the differential ability of compound B to radiosensitize NSCLC, but not HBECs, we treated both H2009 and HBEC3KT with 12.5 nM compound B (Figures 6E and 6F), delivered 10 Gy of radiation, and then monitored the activation of DNA repair proteins. In H2009, compound B lowered the total levels of DNA-PK_CS and ATM, as well as their activation, following IR (Figure 7A). In contrast, these proteins were not affected in HBEC3KT (Figure 7B). Titration of compound B on H2009 and HBEC3KT showed that key DNA repair proteins like DNA-PK_CS and ATM are modestly more sensitive to compound B in H2009 than HBEC3KT (Figure 7C). This may, in part, explain why these cell lines are differentially radiosensitized.

Figure 7. RUVBL1/2 Inhibition Impedes DNA Repair In Vitro and In Vivo, Potentially by Destabilizing ATM Preferentially in Tumor Cells.

(A) Compound B treatment blunts activation of ATM and DNA-PK_CS following IR In H2009. Cells were treated for 3 days with 12.5 nM compound B or compound C, Irradiated with 10 Gy, and harvested at indicated times post-radiation for immunoblot.

(B) Compound B treatment does not affect ATM or DNA-PK_CS activation following IR in HBEC3KT, a normal lung epithelial cell line. Cells were treated as in (A).

(C) PIKK-family proteins are destabilized at lower concentrations of compound B in H2009 than HBEC3KT. Cells were treated for 3 days with the indicated concentrations.

(D and E) Inhibition of RUVBL1/2 delays DSB resolution in NSCLC but not in HBECs. Cell lines were treated with drug for 3 days, irradiated with 2 Gy, grown in the presence of drug until indicated time points, then fixed and prepared for immunofluorescent microscopy. (D) Representative images before or after 2 Gy radiation in H2009 and HBEC3KT treated with 25 nM compound B or 50 nM compound C. Blue, DNA; green, 53BP1; red, γH2AX. Scale bars, 20 μm. (E) Left: quantification of the percentage of remaining μH2AX foci after 2 Gy radiation. Right: percentage of remaining 53BP1 foci after 2 Gy radiation. Values normalized to the number of foci 15 min after 2 Gy radiation and are average ±SD of two biological replicates.

(F) Inhibition of RUVBL1/2 potentiates radiation in vivo without toxicity. Top: experiment outline. Bottom: H1299 was xenografted subcutaneously into female Nude mice and once tumor volumes reached ~150 mm³ mice were randomized to receive indicated treatments. Bottom left: tumor volumes. Statistical significance between vehicle and compound B, and between vehicle + IR and compound B + IR assessed using two-way repeated measures ANOVA. Bottom right: body weight of mice. Values are average ± SEM, and n indicates the number of mice.

(G) NSCLC patient-derived explants (PDExs) are radiosensitized by compound B. Left: experiment outline. NSCLC patient tumors were diced into small pieces, grown on a gelatin sponge, treated for 3 days with 150 nM compound C or compound B, irradiated with 2 Gy, and after 8 h prepared for IHC. Middle: representative γH2AX staining. Blue, DAPI; red, γH2AX. Scale bars, 20 μm. Right: quantification of γH2AX staining in tumor cells from five patient explants. Mean fluorescent intensity in samples that received radiation was normalized to their respective drug treated, no radiation control sample, and lines connect samples from the same patient.

We next studied the ability of NSCLCs and HBECs pretreated with compound B to repair DSBs resulting from IR by monitoring the appearance and resolution of γH2AX and 53BP1 foci. Regardless of the cell line or reagent used, all cells displayed a similar number of DSBs 15 min after 2 Gy IR (Figure S7A). However, RUVBL1/2 inhibition delayed DSB repair kinetics in NSCLC, but not HBECs (Figures 7D and 7E). In addition, genetic depletion of RUVBL1/2 also delayed DSB repair kinetics (Figures S7B and S7C).

In order to evaluate the therapeutic potential of RUVBL1/2 inhibition for radiosensitization, we tested the effects of IR and compound B in vivo. We xenografted H1299 and treated mice with vehicle, compound B, and/or radiation. For these experiments, 125 mg/kg of compound B was used, as it did not change mice bodyweight (Figure 7F, right panel) but was able to reduce ATM protein levels (Figures S7D and S7E). Treatment with compound B at this reduced dose had no effect on H1299 tumor growth. However, compound B significantly potentiated the effects of IR, without obvious toxicity (Figure 7F). These results suggest that RUVBL1/2 inhibition can enhance the therapeutic window afforded by IR.

To extend these findings to additional preclinical models, we utilized patient-derived explant (PDEx) models, where primary tumor tissue specimens are obtained directly from NSCLC patients and grown ex vivo for a short period of time (Centenera et al., 2013). Tumor tissue obtained from five different NSCLC patients was treated with compound B or compound C, irradiated, and IHC for γH2AX was performed 8 h after radiation. Relative to their respective drug treatments, all five human NSCLC explants treated with compound B + IR had more residual γH2AX staining than explants treated with compound C + IR, demonstrating that compound B impairs the recovery from IR in primary NSCLC patient tumors (Figure 7G).

DISCUSSION

RUVBL1/2 are essential proteins that interact with and stabilize a diverse array of multiprotein complexes. Similar to their interacting partner, HSP90 (Trepel et al., 2010), this property complicates studying their role in normal and cancerous physiology because their diverse sets of client proteins may have varying expression levels, importance, and roles among different cell types and cancers. In addition, the dearth of RUVBL1/2 ATPase inhibitors has delayed the dissection of the role of their ATPase activity in these processes. Here, we present and have utilized compound B, a specific RUVBL1/2 inhibitor, to investigate the role of the ATPase activity of RUVBL1/2 in multiprotein complex regulation, supporting the viability of NSCLC and its utility as a therapeutic target in NSCLC.

While RUVBL1/2 are essential for the formation of the PAQosome and the INO80 family of chromatin remodelers, the role of RUVBL1/2 ATPase activity in these complexes was previously unknown. We show that inhibition of RUVBL1/2 ATPase activity results in the failure of the PAQosome complex to either mature client proteins and/or dissociate, without affecting the assembly or disassembly of the INO80 family of chromatin remodelers. Precisely why the PAQosome requires RUVBL1/2 ATPase activity, while the INO80 family may not, awaits further investigation.

Multiple unbiased analyses suggested that RUVBL1/2 have important roles in DNA replication in NSCLC. We found that the inhibition of RUVBL1/2 results in S-phase arrest and eventual replication catastrophe, primarily in NSCLC lines that are sensitive to compound B. Our data suggest that immediately following compound B treatment there may be fewer total active replication forks. The number of active replication forks is dictated both by the number of fired origins and the processivity of existing replication forks. Our iPOND data show that, for active replication forks, there are not significant changes in the abundance of core replication proteins after compound B treatment. Thus, if RUVBL1/2 ATPase activity affects fork processivity, these effects likely only occur at a subset of replication forks. It is also possible that the firing of certain, potentially problematic origins may be inhibited by compound B. Unfired or suppressed origins may also explain why long-term compound B treatment and subsequent ATR depletion results in replication catastrophe. It is important to note that we do not believe RUVBL1/2 play a direct role in regulating DNA replication, since replication defects are undetectable until RUVBL1/2 ATPase activity has been inhibited for at least 6 h. Delineating the precise effects of RUVBL1/2 on DNA replication, and the intermediary players, requires further study.

RUVBL1/2 have received increased attention as therapeutic targets in cancer, and the recently described CB-6644 RUVBL1/2 inhibitor has been shown to be effective as a monotherapy in a Burkitt’s lymphoma and a multiple myeloma cell line (Assimon et al., 2019). We found that compound B as a monotherapy in NSCLC in vivo exhibited a modest therapeutic window. However, our data suggest that the inhibition of RUVBL1/2 delays DSB repair and results in the synergistic killing of NSCLC, but not normal bronchial epithelial cells, following IR. Mechanistically, this may be because the stability of crucial DNA repair proteins such as ATM and DNA-PK_CS are more sensitive to PAQosome inhibition in tumor cell lines than normal cell lines, possibly due to the strained protein buffering capacity of HSP90 that is documented in cancer (Jaeger and Whitesell, 2019). The combination of compound B and IR may be attractive therapeutically because compound B has systemic antitumor activity, whereas IR is delivered locally. Mechanistically, it is interesting to note that our data suggest that the antitumor effects may stem from inhibiting DNA replication, whereas the enhancement of radiation may stem from inhibiting the PAQosome complex. Future work aimed at further exploring and optimizing this combination, determining if this combination is immunogenic, and identifying response biomarkers will be necessary to elucidate the therapeutic potential of inhibiting RUVBL1/2 ATPase activity in NSCLC.

STAR★METHODS

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Ralf Kittler (ralf.kittler@utsouthwestern.edu). All unique/stable reagents generated in this study are available from the Lead Contact without restriction, except for Compound B and Compound C, which require a Materials Transfer Agreement from Daiichi-Sankyo.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell Lines

The following cell lines were established by Drs. John Minna and Adi Gazdar at the NIH (referred to as Hxxxx) or UT Southwestern Medical Center (referred to as HCCxxxx or HBECxxKT) (1981–2019) (M = male, F = female) and are available through ATCC or the Hamon Center for Therapeutic Oncology at UT Southwestern Medical Center: H647 (M), H2009 (F), HBEC3KT (F), HBEC30KT (F), H1993 (F), H838 (M), H1703 (M), H727 (F), HCC4017 (F), H522 (M), H1648 (M), H460 (M), HCC1833 (F), H810 (M), H226 (M), HCC15 (M), HCC95 (M), HCC4019 (M), H1819 (F), H1155 (M), H1299 (M), H2073 (F), H1666 (F), H157 (M), H1650 (M), H596 (M), H1373 (M), H1048 (F) H1355 (M) H1473 (M), H1581 (M), H1693 (F), H1755 (F), H1944 (F), H1975 (F), H2030 (M), H2085 (M), H2087 (M), H2126 (M), H2347 (F), H2882 (F), H3122 (M), H358 (M), H661 (M), H820 (M), HCC1195 (M), HCC1719 (M), HCC193 (F), HCC2814 (M), HCC4190, HCC44 (F), HCC78 (M), HCC827 (F), and H1792 (M). DFCI-024 (F) was obtained from Dana-Farber Cancer Institute. OVCAR5 (F), PANC-1 (M), PC-3 (M), SK-LU-1 (F), MIA PaCa-2 (M), DU 145 (M), EBC-1 (M),ASPC-1 (F), A-427 (M), A549 (M), HOP-62 (F), Calu-1 (M), Calu-3 (M), Calu-6 (F) IMR-90 (F) and Ramos (M) were obtained from ATCC. Lenti-X 293T cells were obtained from Takara (catalog# 632180). All cancer cell lines were grown in RPMI-1640 (MilliporeSigma, catalog# R8758) supplemented with 5% FBS (ThermoFisher Scientific, catalog# 26140079). HBECs were grown in KSFM-SFM (ThermoFisher Scientific, catalog # 17005042). All cells were maintained in a 37°C incubator (NuAire, model NU-5810) with 5% CO2 and 95% relative humidity. Cell lines were routinely checked for mycoplasma contamination (E-myco, catalog # 25233, Bulldog Bio) and DNA fingerprinted for provenance (PowerPlex Fusion24, catalog# DC2402, Promega).

Animal Studies

Mice were housed by the UTSW Animal Resource Center at 68–79° F, 30–70% humidity, in individually ventilated cages, with no more than 5 mice per cage on 12 hour on:off light:dark cycles. Mice were screened for and found free of MHV, Sendai virus, MPV, EDIM, MVM, PVM<TMEV-/GD-7, REO-3 virus, Mycoplasma pulmonis, pinworms, fur mites, LCMV, ECTRO, MAV, and K virus and had unrestricted access to RO chlorinated water and irradiated 2916 Teklab global diet (Envigo, catalog# 2916). For xenografts with Compound B monotherapy, NOD/SCID females (UTSW breeding core, NOD.CB17-Prkdc^scid/J) aged 10–12 weeks, average weight ~21g, were used. For irradiation + Compound B experiments, female nude mice (Charles Rivers, Crl:NU-Foxn1^nu, catalog# 553) aged 10–12 weeks, average weight ~23g, were used. For pharmacokinetic studies, 7-week-old male BALB/c mice were used. All protocols performed at UTSW were approved and monitored by the Institutional Animal Care and Use Committee.

METHOD DETAILS

Immunohistochemistry (IHC)

For IHC of RUVBL1 in patient tumors, two tissue microarrays (TMAs) were prepared using three 1.0-mm tissue cores obtained from the center, middle and periphery of formalin-fixed and paraffin-embedded (FFPE) histological sections from 752 surgically resected primary NSCLC tumors. All NSCLC tissues were evaluated, and patients underwent surgical resection at The University of Texas MD Anderson Cancer Center, following informed consent under protocols approved by the university’s Institutional Review Board. From each FFPE sample, 4-micron sections were cut in a microtome and mounted on charged glass slides. All staining was performed automatically using the Leica Bond Max Autostainer system (Leica Biosystems). Antigen retrieval was performed with Epitope Retrieval Solution 1 BOND (Leica Biosystems, catalog# AR9961) and staining performed using the Bond Polymer Refine Detection kit (Leica Biosystems, catalog# DS9800) with RUVBL1 antibody at 1:1000 (MilliporeSigma, catalog# SAB4200194). Bond Polymer Refine Detection kit (Leica Biosystems) was employed as a detection system and diaminobenzidine (DAB) was used as chromogen for the visualization of the IHC staining. The slides were then counterstained with hematoxylin, dehydrated and coverslipped. Fallopian tube tissue was used as a positive control. After quality control on all slides, slides were scanned in an Aperio AT2 scanner (Leica Biosystems). IHC scoring was performed by a pathologist (P.V.) using a brightfield microscope and the data was presented as the average of the H-scores (which ranges from 0 to 300 and incorporates both the intensity and percent of stained tumor cells) for both nuclear and cytoplasmic staining. The final data, images and report was reviewed by an independent pathologist (J.R.).

For ATM IHC, tumors were fixed in 10% neutral buffered formalin (VWR, catalog# 89497–305) for 48 hours. Tumors were then embedded in paraffin, sectioned into 4-micron slices and placed onto positively charged slides. All staining was performed automatically using the Leica Bond Max Autostainer system (Leica Biosystems). Antigen retrieval was performed with Epitope Retrieval Solution 2 BOND (Leica Biosystems, catalog# AR9640) and Bond Polymer Refine Detection kit (Leica Biosystems, catalog# Catalog# DS9800) was employed as a detection system with 1:300 ATM antibody (Abcam, catalog# ab32420). We used a modified protocol that removed the post-primary antibody step to avoid background from mouse tissue IgG. Diaminobenzidine (DAB) was used as chromogen for the visualization of the IHC staining. The slides were then counterstained with hematoxylin, dehydrated and coverslipped. Human breast cancers that were positive for ATM was used as a control. After quality control on all slides, slides were scanned in an Aperio AT2 scanner (Leica Biosystems) and scored using Aperio Toolbox Image Analysis software (Leica Biosystems) applying a nuclear algorithm by a pathologist (P.V.) and data was presented as a percentage of positive tumor cell nuclei. Final data, images and report was reviewed by an independent pathologist (J.R.).

For γH2AX IHC, tumors were fixed in ice-cold 4% paraformaldehyde (Electron Microscopy Sciences catalog # 15710) for 48 hours with slight agitation at 4°C, then embedded in paraffin, sectioned into 4-micron slices and placed onto positively charged slides. Slides were heated at 60°C for 10 minutes, deparaffinzed and rehydrated. Antigen retrieval was performed with a Biocare Medical Decloaking Chamber at 110°C for 17 minutes using 10 mM sodium citrate pH 6 then allowed to cool to room temperature for 30 minutes. Slides were washed for 5 minutes with PBS, incubated in ice-cold 10% methanol for 10 minutes at room temperature, washed with de-ionized H20 for 5 minutes, then tissue was blocked for 30 minutes using SignalStain Antibody Diluent (Cell Signaling Technology, catalog # 8112). Blocking buffer was removed, then γH2AX antibody (Cell Signaling Technology, catalog # D7T2V) was diluted 1:1000 in SignalStain Antibody Diluent and added to samples at 4°C overnight with agitation. Samples were washed 3 times with .05% PBST for 5 minutes, once with PBS for 5 minutes, then secondary antibody solution (Vector Labs, ImmPRESS HRP Anti-Mouse IgG Polymer Detection Kit made in Goat, catalog # MP-7452) was added for 30 minutes with agitation at room temperature. Samples were washed once with .2% PBST for 5 minutes, twice with .5% PBST for 5 minutes, then a solution of 1:250 Opal570 (PerkinElmer, catalog # FP1488001KT) in 1X Plus Amplification Diluent (PerkinElmer, catalog # FP1498) was made and added to samples for 3 minutes. After 3 minutes, the solution was quickly aspirated and washed with .05% PBST 3 times, then washed with PBS + 2 mM EDTA for 10 minutes. Coverslips were mounted using Fluoroshield Mounting Medium with DAPI (Abcam, catalog # ab104139) and sealed with nail polish. Images were captured at 20X using a Keyence BZ-X710. For quantification, obvious debris and non-tumor tissue was first masked, then nuclear γH2AX signal was quantified using Fiji by splitting the multicolor images into three channels, selecting DAPI positive nuclei (“Find Edges”, “Make Binary”, “Close-“, “Fill Holes”, and “Create Selection”), applying this selection to the γH2AX channel, and measuring the median fluorescent intensity.

Inhibitors/Drugs Used

Compound B (RUVBL1 inhibitor) and Compound C (Compound B control) were synthesized by Daiichi-Sankyo. VE-822 (ATR inhibitor) was purchased from Selleckchem (catalog # S7102). LY2603618 (CHEK1 inhibitor) was purchased ApexBio (catalog # A8638). Puromycin was purchased from InvivoGen (catalog #ant-pr-1). Geneticin (G418) was purchased from ThermoFisher Scientific (catalog # 11811031). Doxycycline was purchased from MilliporeSigma (catalog # D9891).

Immunoblots

Cells were washed twice with ice-cold PBS and then scraped on ice. Cells were lysed with a modified RIPA buffer (50 mM Tris, 150 mM NaCl, .1% SDS, 1% IGEPAL CA-630, 1% sodium deoxycholate, 2 mM MgCl₂, pH 8) with 1 unit/μL benzonase (MilliporeSigma, catalog # E1014), protease inhibitors (MilliporeSigma, catalog # P8340) and phosphatase inhibitors (MilliporeSigma, catalog # 4906845001) by rotating lysates at 4°C for 2 hours. Lysates were then cleared by spinning at max speed for 10 minutes, quantified using BCA (ThermoFisher Scientific, catalog #23225), mixed with 2X Laemmli buffer (BioRad, catalog # 1610737EDU) and boiled for 5 minutes immediately prior to loading. For ATR, ATM, and DNA-PK_CS, 20–25 μg of protein was ran on a NuPAGE 3–8% Tris-Acetate gel (ThermoFisher Scientific, catalog # EA0375BOX) using NuPAGE Tris-Acetate SDS Running Buffer (ThermoFisher, catalog # LA0041) at 150V. For all other proteins, 20–25 μg of protein was ran on a 4–20% Mini-PROTEAN TGX gel (BioRad, catalog # 4561096) at 220 V. Samples were transferred using the Trans-Blot Turbo RTA Mini Nitrocellulose transfer kit (Biorad, catalog # 1704270) on the Trans-Blot Turbo Transfer System (Biorad, catalog # 1704150) using manufacturers suggested protocols. For all chemiluminescent blots (Figures S1F, S1H, S1J, and S6A), blocking/antibody incubation steps were done with 5% milk (Biorad, catalog # 1706404XTU) in .1% in PBST. All other blots were done using near infrared fluorescence, and blocking/antibody incubation steps were done with TBS Odyssey Blocking (LiCor, catalog # 927–50100) with 0% / .2% TBST, respectively, and imaged using a LiCor Odyssey Fc. Antibody information is listed in the Key Resources Table.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal RUVBL1 (Immunoblot 1:1000) (IHC 1:1000)	MilliporeSigma	Cat# SAB4200194
Rabbit polyclonal RUVBL2 (Immunoblot 1:1000)	Abcam	RRID: AB_2301439 Cat# ab36569
Rabbit polyclonal phosphoHistone H3 (s10) (Immunoblot 1:2000) (Immunofluorescence 1:500)	Abcam	RRID: AB_304763 Cat# ab5176
Rabbit monoclonal ATM (Immunoblot 1:1000)	Abcam	Cat# ab201022
Rabbit monoclonal H2AX (Immunoblot 1:4000)	Abcam	RRID: AB_10971675 Cat# ab124781
Rabbit polyclonal phosphoCHEK1 (s317) (Immunoblot 1:1500)	Cell signaling	RRID: AB_331488 Cat# 2344
Rabbit polyclonal DNA-PKCS (Immunoblot 1:2000)	Santa Cruz	RRID: AB_2172845 Cat# sc-9051
Mouse monoclonal CHEK1 (Immunoblot 1:1000)	Santa Cruz	RRID: AB_627257 Cat# sc-8408
Mouse monoclonal HSP90 (Immunoblot 1:2000)	Santa Cruz	RRID: AB_675659 Cat# sc-13119
Mouse monoclonal BrdU (Flow cytometry 1:200)	Santa cruz	RRID: AB_626766 Cat# sc-32323
Mouse monoclonal V5 (Immunoprecipitation, 15ug for 20million cells) (Immunoblot 1:2000)	Genscript	RRID: AB_2622216 Cat# A01724
Rabbit polyclonal V5 (Immunoblot 1:2000)	Bethyl Labs	RRID: AB_67586 Cat# A190-120A
Mouse IgG1 isotype control (Immunoprecipitation, 15ug for 20million cells)	Cell signaling	RRID: AB_10829607 Cat# 5415S
Mouse monoclonal a-Tubulin (Immunoblot 1:2000)	Cell signaling	RRID: AB_1904178 Cat# 3873
Mouse monoclonal phosphoH2AX (S139) (Immunoblot 1:4000) (Immunofluorescence 1:1000)	MilliporeSigma	RRID: AB_310795 Cat# 16-193
Rabbit monoclonal PARP1 (Immunoblot 1:1000)	Cell signaling	RRID: AB_659884 Cat# 9532
Mouse monoclonal RPA2 (Immunoblot 1:1000) (Flow cytometry 1:100) (Immunoprecipitation 1ug per 2 million cells)	Abcam	RRID: AB_302873 Cat# ab2175
Rabbit monoclonal GAPDH (Immunoblot 1:2000)	Cell signaling	RRID: AB_561053 Cat# 2118
Goat anti-Rabbit HRP conjugate (Immunoblot 1:2000)	Cell signaling	RRID: AB_2099233 Cat# 7074
Horse anti-Mouse HRP conjugate (Immunoblot 1:2000)	Cell signaling	RRID: AB_330924 Cat# 7076
Goat polyclonal anti-Mouse IRDye 680RD conjugate (Immunoblot 1:15000)	LiCor	RRID: AB_2651128 Cat# 925-68070
Goat polyclonal anti-Rabbit IRDye 800CW conjugate (Immunoblot 1:15000)	Licor	RRID: AB_2651127 Cat# 925-32211
Goat anti-mouse Alexa Fluor 488 conjugate (Flow cytometry 1:200) (Immunofluorescence 1:500)	ThermoFisher Scientific	RRID: AB_2534088 Cat# A-11029
Rabbit monoclonal ATM (IHC 1:300)	Abcam	RRID: AB_725574 Cat# ab32420
Donkey anti-Rabbit Alexa Fluor 555 conjugate (Immunofluorescence 1:500)	ThermoFisher Scientific	RRID: AB_162543 Cat# A-31572
Rabbit polyclonal 53BP1 (Immunofluorescence 1:500)	Cell signaling	RRID: AB_10694558 Cat# 4937
Rabbit polyclonal RPAP3 (Immunoblot 1:1000)	ProteinTech	Cat# 23741-1-AP
Rabbit polyclonal PIH1D1 (Immunoblot 1:1000)	ProteinTech	Cat# 19427-1-AP
Rabbit polyclonal PFDN6 (Immunoblot 1:1000)	ThermoFisher	RRID: AB_2645423 Cat# PA5-61657
Mouse monoclonal phosphoH2AX (S139) (IHC 1:1000)	Cell Signaling Technology	RRID: AB_2799949 Cat# D7T2V
Goat anti-mouse IgG-Peroxidase polymer detection kit	Vector Laboratories	RRID: AB_2744550 Cat# MP-7452
Rabbit polyclonal PDRG1 (Immunoblot 1:1000)	ProteinTech	RRID: AB_2162082 Cat# 16968-1-AP
Rabbit polyclonal TTI1 (Immunoblot 1:1000)	Bethyl Labs	RRID: AB_10953982 Cat# A303-451A
Rabbit polyclonal TTI2 (Immunoblot 1:1000)	Bethyl Labs	RRID: AB_10948973 Cat# A303-476A
Rabbit polyclonal TELO2 (Immunoblot 1:1000)	ProteinTech	RRID: AB_2203337 Cat# 15975-1-AP
Rabbit monoclonal RPA1 (Immunoblot 1:1000)	Abcam	RRID: AB_1603759 Cat# ab79398
Rabbit polyclonal RPA3 (Immunoblot 1:1000)	Abcam	RRID: AB_10695955 Cat# ab97436
Mouse monoclonal pATM (Ser1981) (Immunoblot 1:500)	Santa Cruz	RRID: AB_781524 Cat# sc-47739
Rabbit polyclonal pDNA-PKcs (Ser2056) (Immunoblot 1:1000)	Abcam	RRID: AB_869495 Cat# ab18192
Mouse monoclonal ATR (Immunoblot 1:1000)	GeneTex	RRID: AB_368158 Cat# GTX70109
Mouse monoclonal ATM (Immunoblot 1:500)	Santa Cruz	Cat# sc-377293
Rat monoclonal CldU/BrdU (DNA fiber analysis)	Accurate Chemical and Scientific	RRID: AB_2313756 Cat# OBT0030
Mouse monoclonal IdU/BrdU (DNA fiber analysis)	BD Biosciences	RRID:AB_400326 Cat# 347580
Goat anti-rat IgG Alexa Fluor 488 (DNA fiber analysis)	ThermoFisher Scientific	RRID:AB_2534074 Cat# A-11006
Goat anti-mouse IgG Alexa Fluor 568 (DNA fiber analysis)	ThermoFisher Scientific	RRID:AB_2534072, Cat# A-11004
Rabbit polyclonal POLE2 (Immunoblot 1:500)	ThermoFisher Scientific	RRID:AB_2645746, Cat# PA5-55654
Rabbit polyclonal POLD2 (Immunoblot 1:1000)	ThermoFisher Scientific	RRID:AB_2620518, Cat# A304-322A
Rabbit polyclonal RFC4 (Immunoblot 1:2000)	Abcam	RRID:AB_10687312, Cat# ab96852
Mouse monoclonal PCNA (Immunoblot 1:1000)	Santa Cruz	RRID:AB_628110, Cat# sc-56
Bacterial and Virus Strains
BL21 Star DE3 E. coli	ThermoFisher Scientific	Cat# C601003
NEB Stable E. coli	New England Biolabs	Cat# C3040H
ElectroMAX Stbl4 E. coli	Invitrogen	Cat# 11635018
Biological Samples
NSCLC Tissue Microarray	Ignacio Wistuba, M.D. Anderson Cancer Center
NSCLC Patient Tumor Explants	Ganesh Raj, UT Southwestern Medical Center
Chemicals, Peptides, and Recombinant Proteins
Compound B (RUVBL1/2 inhibitor)	Daiichi-Sankyo	Synthesized by Daiichi-Sankyo
Compound C (Compound B control)	Daiichi-Sankyo	Synthesized by Daiichi-Sankyo
VE-822 (ATR inhibitor)	SelleckChem	Cat# S7102
LY2603618 (CHEK1 inhibitor)	ApexBio	Cat# A8638
Puromycin	InvivoGen	Cat# ant-pr-1
Geneticin (G418)	ThermoFisher Scientific	Cat# 11811031
Benzonase	MilliporeSigma	Cat# E1014
PhosStop	MilliporeSigma	Cat# 4906845001
Protease Inhibitors	MilliporeSigma	Cat# P8340
Polybrene	SantaCruz	Cat# sc-134220
CellTiter 96 Aqueous One Solution Cell Proliferation Assay (MTS)	Promega	Cat# G3582
Doxycycline	MilliporeSigma	Cat# D9891
Poly(ethylene glycol) 200	MilliporeSigma	Cat# 88440
Protein G-Gamma Sepharose	GE Healthcare	Cat# 17088501
V5-peptide	MilliporeSigma	Cat# V7754
Coomassie	Biorad	Cat# 1610436
RNase A	Qiagen	Cat# 19101
Propidium Iodide	ThermoFisher Scientific	Cat# 3566
Bovine serum albumin (BSA)	Jackson ImmunoResearch	Cat# 001-000-173
Thymidine	MilliporeSigma	Cat# T1895
BrdU	MilliporeSigma	Cat# B5002
Sodium tetraborate	MilliporeSigma	Cat# 221732
Epitope Retrieval Solution 1 BOND	Leica Biosystems	Cat# AR9961
Epitope Retrieval Solution 2 BOND	Leica Biosystems	Cat# AR9640
Vectashield	Vector Laboratories	Cat# H-1200
Crystal Violet	MilliporeSigma	Cat# C6158
16% Paraformaldehyde	Electron Microscopy Sciences	Cat# 15710
SignalStain Antibody Diluent	Cell Signaling Technology	Cat# 8112
Carbenicillin	Sigma	Cat# C1613
Glutathione Sepharose 4 Fast Flow	GE Healthcare	Cat# 17513201
PreScission Protease	GE Healthcare	Cat# 27-0843-01
EdU	Cayman Chemicals	Cat# 20518
Glycine	Sigma-Aldrich	Cat# 50046
Biotin Azide	Click Chemistry Tools	Cat# 1265-6
Sodium Ascorbate	SantaCruz	Cat# sc-215877
CuSO4	SantaCruz	Cat# sc-203009A
Streptavidin agarose beads	Millipore	Cat# 161-0436
Opal570	PerkinElmer	Cat # FP1488001KT
1X Plus Amplification Diluent	PerkinElmer	Cat # FP1498
Fluoroshield Mounting Medium with DAPI	Abcam	Cat # ab104139
Critical Commercial Assays
e-Myco kit	Bulldog Bio	Cat# 25233
PowerPlex Fusion24	Promega	Cat# DC2402
BCA kit	ThermoFisher Scientific	Cat# 23225
Trans-Blot Turbo RTA mini nitrocellulose transfer kit	BioRad	Cat# 1704270
Effectene	Qiagen	Cat# 301425
Lipofectamine RNAiMAX	ThermoFisher Scientific	Cat# 13778150
Bond Polymer Refine Detection kit	Leica Biosystems	Cat# DS9800
APO-BrdU™ TUNEL Assay Kit, with Alexa Fluor™ 488 Anti-BrdU	ThermoFisher Scientific	Cat# A23210
RNeasy Plus Mini Kit	Qiagen	Cat# 74134
iScript cDNA synthesis kit	Biorad	Cat# 1708890
iTaq Universal SYBR Green Supermix	Biorad	Cat# 1725120
ADP-Glo	Promega	Cat# V9101
MyTaq DNA polymerase	Bioline	Cat# BIO-21105
MinElute PCR Purification Kit	Qiagen	Cat# 28004
ElectroLigase	New England Biolabs	Cat# M0369S
Plasmid Midiprep	Qiagen	Cat# 12145
DNeasy Blood & Tissue Kit	Qiagen	Cat# 69504
Phusion DNA polymerase	New England Biolabs	Cat# M0530S
QIAquick PCR purification kit	Qiagen	Cat# 28104
KinomeScan scanEDGE	DiscoverX
Deposited Data
RNA-sequencing data	NCBI GEO	GSE107637
DNA-sequencing data	NCBI GEO	GSE107637
Experimental Models: Cell Lines
H647 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1574
H2009 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1514
HBEC3KT (normal bronchial epithelium)	John Minna and Adi Gazdar	RRID: CVCL_X491
HBEC30KT (normal bronchial epithelium)	John Minna and Adi Gazdar	RRID: CVCL_AS83
H1993 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1512
H838 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1594
H1703 (Male, Squamous cell, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1490
H727 (Female, Carcinoid, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1584
HCC4017 (Female, Large Cell, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_V579
H522 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1567
H1648 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1482
H460 (Male, Large Cell, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_0459
HCC1833 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID:CVCL_5129
H810 (Male, Large Cell, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1590
H226 (Male, Mesothelioma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1544
HCC15 (Male, Squamous Cell, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_2057
HCC95 (Male, Squamous, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_5137
HCC4019 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_V581
H1819 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1497
H1155 (Male, Large cell, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1456
H1299 (Male, Large cell, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_0060
H2073 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1521
H1666 (Female, Squamous, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1485
H157 (Male, Squamous, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_0463
H1650 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1483
H596 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1571
H1373 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1465
H1792 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1495
H1048 (Female, Small Cell Lung Cancer)	John Minna and Adi Gazdar	RRID: CVCL_1453
H1355 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1464
H1437 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1472
H1581 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1479
H1693 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1488
H1755 (Female, Neuroendocrine, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1492
H1944 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1508
H1975 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1511
H2030 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1517
H2085 (Male, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1523
H2087 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1524
H2126 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1532
H2347 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1550
H2882 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_5158
H3122 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_5160
H358 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1559
H661 (Male, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1577
H820 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_1592
HCC1195 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_5127
HCC1719 (Male, Squamous, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_L089
HCC193 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_5130
HCC2814 (Male, Squamous, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_V586
HCC4190 (NSCLC)	John Minna and Adi Gazdar
HCC44 (Female, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_2060
HCC78 (Male, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_2061
HCC827 (Female, Adenocarcinoma, NSCLC)	John Minna and Adi Gazdar	RRID: CVCL_2063
MIA PaCa-2 (Male, Epithelial Carcinoma, Pancreas	ATCC	RRID: CVCL_0428 Cat# CRM-CRL-1420
OVCAR5 (Female, Ovarian Cancer or Upper GI)	ATCC	RRID: CVCL_1628
PANC-1 (Male, Pancreatic Ductal Adenocarcinoma)	ATCC	RRID: CVCL_0480 Cat# CRL-1469
PC-3 (Male, Adenocarcinoma, Prostate)	ATCC	RRID: CVCL_0035 Cat# CRL-1435
SK-LU-1 (Female, Adenocarcinoma, NSCLC)	ATCC	RRID: CVCL_0629 Cat# HTB-57
DFCI-024 (Female, Adenocarcinoma, NSCLC)	Dana-Farber Cancer Institute	RRID: CVCL_A762
A549 (Male, Adenocarcinoma, NSCLC)	ATCC	RRID: CVCL_0023 Cat# CCL-185
HOP-62 (Female, Adenocarcinoma, NSCLC)	ATCC	RRID: CVCL_1285
Calu-1 (Male, Squamous, NSCLC)	ATCC	RRID: CVCL_0608 Cat# HTB-54
Calu-3 (Male, Adenocarcinoma, NSCLC)	ATCC	RRID: CVCL_0609 Cat# HTB-55
Calu-6 (Female, NSCLC)	ATCC	RRID: CVCL_0236 Cat# HTB-56
IMR-90	ATCC	RRID: CVCL_0347 Cat# CCL-186
Lenti-X 293T (Female, kidney)	Takara	Cat# 632180
Ramos (Male, Burkitt Lymphoma)	ATCC	RRID: CVCL_1646 Cat# CRL-1923
A-427 (Male, NSCLC)	ATCC	RRID: CVCL_1055
ASPC-1 (Female, Adenocarcinoma, Pancreas)	ATCC	RRID: CVCL_0152 Cat# CRL-1682
EBC-1 (Male, Squamous, NSCLC)	ATCC	RRID: CVCL_0152
DU 145 (Male, Epithelial Carcinoma, Prostate)	ATCC	RRID: CVCL_0105 Cat# HTB-81
Experimental Models: Organisms/Strains
NOD.CB17-PrkdcSCID/J Female mice	UTSW Breeding Core	RRID: IMSR_JAX:001303
Athymic (Nu/Nu) Female mice	Charles Rivers	Cat# 553
BALB/cByJJcl Male mice	CLEA Japan Inc	RRID:IMSR_JAX:001026
Oligonucleotides
siRUVBL1, 5’-CAGAGCTAACTCCGTGTGAGA-3’	MilliporeSigma	Custom synthesis
siRUVBL2, On-Target Plus, 5’-UAACAAGGAUUGAGCGAAU-3’	GE Healthcare	Cat# J-012299-05
siNTC, On-Target Plus non-targeting siRNA #1	GE Healthcare	Cat# D-001810-01
esiRUVBL1 (primers used for amplification) 5’- TCACTATAGGGAGAGGAGGCATGTGGCGTCATAGT-3’ 5’- TCACTATAGGGAGAGTCAAGGTCGAATTCTGTGGCA-3’	See (Kittler et al., 2007) for protocol	Made in house
esiRUVBL2 (primers used for amplification) 5’- TCACTATAGGGAGAGACGGAGATCATCGAAGGGGA-3’ 5’- TCACTATAGGGAGAGTCGTCGATGAACAGCACTCC-3’	See (Kittler et al., 2007) for protocol	Made in house
esiGFP (primers used for amplification) 5’-TCACTATAGGGAGAGACGTAAACGGCCACAAGTTC-3’ 5’-TCACTATAGGGAGACGCGGATCTTGAAGTTCACCT-3’	See (Kittler et al., 2007) for protocol	Made in house
RUVBL1f 5’- AGAGCACTACGAAGACGCAG-3’ (qPCR)	MilliporeSigma	Custom synthesis
RUVBL1r 5’- TATGACGCCACATGCCTCTC-3’ (qPCR)	MilliporeSigma	Custom synthesis
RUVBL2f 5’- AACCGTTACAGCCACAACCA-3’ (qPCR)	MilliporeSigma	Custom synthesis
RUVBL2r 5’- GCACCGATTCGCTCAATCCT-3’ (qPCR)	MilliporeSigma	Custom synthesis
Recombinant DNA
pIRESneo3-RUVBL1-RNAiR-V5	pIRESneo3 backbone from Clontech	Cat# 631621
pIRESneo3-RUVBL2-RNAiR-V5	pIRESneo3 backbone from Clontech	Cat# 631621
pCMV-dR8.91	Robert Weinberg lab
pMD2.G	Addgene	Plasmid #12259
pLVX-TRE3G-RUVBL1-RNAiR-WT-IRES	pLVX-TRE3G-IRES backbone from Clontech	Cat# 631362
pLVX-TRE3G-RUVBL1-RNAiR-ATPaseDead-IRES	pLVX-TRE3G-IRES backbone from Clontech	Cat# 631362
pLVX-TRE3G-RUVBL2-RNAiR-WT-IRES	pLVX-TRE3G-IRES backbone from Clontech	Cat# 631362
pLVX-TRE3G-RUVBL2-RNAiR-ATPaseDead-IRES	pLVX-TRE3G-IRES backbone from Clontech	Cat# 631362
pLVX-TRE3G-RUVBL1mutants-IRES (all)	pLVX-TRE3G-IRES backbone from Clontech	Cat# 631362
pLVX-TRE3G-RUVBL2mutants-IRES (all)	pLVX-TRE3G-IRES backbone from Clontech	Cat# 631362
pLVX-EF1A-TET3G	Clontech	Cat# 631359
pTRIPZ-shNTC, sequence (stem-loop underlined) 5’-ATCTCGCTTGGGCGAGAGTAAGTAGTGAAGCCACAGATGTACTTACTCTCGCCCAAGCGAGAG	pTRIPZ from Dharmacon	Cat# RHS4696
pTRIPZ-shRUVBL1, sequence (stem-loop underlined) 5’-ACCGGCCAACTTGCTTGCTAAATAGTGAAGCCACAGATGTATTTAGCAAGCAAGTTGGCCGGG-3’	pTRIPZ from Dharmacon	Cat# RHS4696
pTRIPZ-shRUVBL2, sequence (stem-loop underlined) 5’-ACGAGAAAGACACGAAGCAGATTAGTGAAGCCACAGATGTAATCTGCTTCGTGTCTTTCTCGC-3’	pTRIPZ from Dharmacon	Cat# RHS4696
pGEX-6P-1	GE Healthcare	Cat# 28-9546-48
Software and Algorithms
FlowJo	FlowJo LLC	RRID: SCR_008520
Image Studio Lite	LiCor	RRID: SCR_014211
Central Proteomics Facilities Pipeline	(Trudgian and Mirzaei, 2012)	RRID: SCR_012043
X!Tandem	(Craig and Beavis, 2004)
Open MS Search Algorithm	(Geer et al., 2004)
SINQ normalized spectral index software	(Trudgian et al., 2011)
Gene Set Enrichment Analysis (GSEA)	(Subramanian et al., 2005)	RRID: SCR_003199
Enrichr	(Kuleshov et al., 2016)	RRID: SCR_001575
Cell Profiler (v1.0.9717)		RRID: SCR_007358
BZ-X Analyzer (v1.3.1.1)	Keyence
Sequest	Thermo Fisher Scientific	RRID:SCR_014594
Fiji Image Software	(Schindelin et al., 2012)
ZEN Digital Imaging for Light Microscopy	Zeiss	RRID:SCR_013672
Other
Dounce Homogenizer	VWR	Cat# KT885300-0002
Bioruptor Plus Water Bath Sonicator	Diagenode
70 μm nylon filter	Sigma-Aldrich	Cat# CLS431751

siRNA Transfections

siRNA and esiRNA sequences are listed in the Key Resources Table. For all cell lines, for all assays involving siRNA or esiRNAs, cells were first trypsinized, spun down and resuspended as a single cell solution. For each well in a 6-well tissue culture treated plate, 2–3 μL of RNAiMAX (ThermoFisher Scientific, catalog # 13778150) was first incubated with 500 μL of serum-free RPMI 1640 for 5 minutes, then mixed with 3 μL of 20 μM siRNA or esiRNA (Kittler et al., 2007) for 15 minutes. From the single cell solution, 50,000–150,000 cells (counted using a Beckman Coulter Z2 Particle Counter at >12 microns) in 2.5 mL of RPMI 1640 with 5% FBS were then mixed and left overnight. In the morning, media was changed. For cell growth assays, 5 days after transfection cells were trypsinized and counted using a Beckman Coulter Z2 particle counter at >12 microns and normalized to respective controls. RNAi-RUVBL1 or RNAi-RUVBL2 represents the average (mean) of the percent of cells remaining after siRUVBL1/2 knockdown and esiR-UVBL1/2 knockdown (respectively), in comparison to control oligonucleotides (siNTC or esiGFP). For immunoblots, the cells were washed twice with ice-cold PBS, scraped on ice, and then treated and analyzed following the Immunoblots methods outlined in this manuscript. For experiments involving siRNA + doxycycline addition (Figure 2C), 2 μg/mL doxycycline (MilliporeSigma, catalog# D9891) was added immediately following transfection. When multiple siRNAs were transfected, final siRNA concentration was kept at 20 nM.

Plasmid Transfection

To establish H1299 and H2009 cells stably expressing RNAi-resistant wild type RUVBL1 and RUVBL2 proteins, rat RUVBL1 cDNA (Dharmacon, catalog # MRN1768–202782702, IMAGE Clone ID 7304521) and rat RUVBL2 cDNA (Dharmacon, catalog # MRN1768–202784379, IMAGE Clone ID 7457010) were mutated via site-directed mutagenesis at all wobble positions to share no homology to siRUVBL1 or siRUVBL2 and then cloned from start codon to stop codon into pIRESneo3 with an added C-terminal V5 tag. The resulting pIRESneo3-RUVBL1-RNAiR-V5 or pIRES-neo3-RUVBL2-RNAiR-V5 plasmids were forward transfected into 2 million H1299 or H2009 cells in a T75 using 2 μg plasmid DNA, 16 μL enhancer and 20 μL Effectene (Qiagen, catalog # 301425) following manufacturers protocol. Cells stably expressing the constructs were selected for by adding .6 mg/mL G418 (ThermoFisher Scientific, Catalog# 11811031) for 2 weeks.

Lentivirus Generation and Transduction

To generate lentiviral particles, 2 million Lenti-X 293T cells (Clontech, catalog #632180) were forwarded transfected with 9 μg pCMV-dR8.91, 3 μg pMD2.G and 3 μg of LentiPlasmid-of-interest (i.e. pLVX-EF1A-TET3G, pLVX-TRE3G-IRES or pTRIPZ) using FuGene6 (Promega, catalog#E2691) following the manufacturers protocol. After 12 hours, media was changed and viral supernatant was collected every day for 3 days and filtered through a .45micron syringe filter (Corning, catalog #431220). For each infection in each cell line, viral supernatant was titrated onto cells and mixed with 6 μg/mL polybrene (Santa Cruz, catalog# sc-134220) and incubated with cells overnight. Only cells receiving concentrations of viral supernatant that resulted in low infectivity (i.e. >50% of the cells did not survive selection) were used to control for multiple integrations.

qPCR

Cells were lysed directly in 6-well plates and RNA was isolated using the RNeasy Plus Mini Kit (Qiagen, catalog# 74134) following manufacturers protocol. cDNA was synthesized from 1 μg of RNA using the iScript cDNA synthesis kit (Biorad, catalog# 1708890) following manufacturers protocols. qPCR was performed using iTaq Universal SYBR Green Supermix (Biorad, catalog# 1725120) using manufacturers protocols on an Applied Biosystems 7300 Real-Time PCR system (Thermo Fisher Scientific). qPCR primers for RUVBL1: 5’- AGAGCACTACGAAGACGCAG-3’ and 5’- TATGACGCCACATGCCTCTC-3’. qPCR primers for RUVBL2: 5’- AACCGTTACAGCCACAACCA-3’ and 5’- GCACCGATTCGCTCAATCCT-3’. RNA expression was determined using the delta delta Ct method (Schmittgen and Livak, 2008) and all primers were verified for efficiency.

PCR Mutagenesis Strategy to Identify Compound B-resistance Conferring Mutations

RUVBL1 or RUVBL2 open reading frames were amplified from the pIRESneo3-RUVBL1-RNAiR or pIRESneo3-RUVBL2-RNAiR plasmids using standard error-prone Taq polymerase (Bioline MyTaq catalog # BIO-21105) and 30 cycles, with addition of flanking 5’ NotI and 3’ NdeI restriction sites added via primers. These conditions were chosen as 91% of PCR products should contain at least one mutation. PCR products were purified (Qiagen MinElute PCR Purification Kit catalog # 28004), digested with NotI and NdeI, then ligated into pLVX-TRE3G-IRES (digested with NotI and NdeI) using ElectroLigase (NEB catalog # M0369S). Ligated products were electroporated into ElectroMAX Stbl4 E. coli (Invitrogen catalog # 11635018) and allowed to grow for 60 minutes before 1/20^th of the total volume was plated onto LB + carbenicillin (Sigma catalog # C1613) plates. All reactions were performed such that > 200 colonies were on this plate the next day, suggesting that there were at least 2000 individual transformants, and thus at least 1820 mutated RUVBL1/RUVBL2 alleles were made (2000 transformants * 91% PCR error rate). The remaining transformant mixture was grown overnight in LB + carbenicillin, then plasmid DNA was isolated (Qiagen Plasmid Midiprep catalog # 12145). Virus was packaged from these pLVX-TRE3G-RUVBL1/2-IRES plasmids and H2009 cells were transduced at .3 MOI, such that >100,000 cells received virus, and this number of cells was maintained throughout passaging. For Compound B selection, 500,000 H2009 cells/dish were plated into 4 15 cm dishes. Two plates received 1 μg/mL doxycycline for two days, and after two days 500 nM Compound B +/− doxycycline (to respective plates) was added. Media was changed every three days until large resistant colonies emerged in the plates with doxycycline. These cells were harvested via trypsinization, genomic DNA prepared (Qiagen DNeasy Blood & Tissue Kit, catalog # 69504), and three ~500bp barcoded amplicons that covered the RUVBL1/RUVBL2 open reading frame were made via nested PCR (NEB Phusion, catalog # M0530S), pooled together, and these PCR products were isolated (Qiagen QIAquick PCR purification kit catalog # 28104). DNA was run with the D1000 ScreenTape on the Agilent Tapestation 4200 and quantified by Qubit. Libraries were prepared with the KAPA Hyper Prep Kit. Samples were end repaired, 3’ ends adenylated and barcoded with multiplex adapters. Samples were then size selected prior to PCR to remove small fragments. Two cycles of PCR were used to amplify the libraries, which were then purified with AmpureXP beads and validated on the Agilent Tapestation 4200. Samples were normalized, pooled and then quantified by Qubit prior to being run on the MiSeq Reagent Kit v3 (600 cycle) chemistry. Reads obtained from pooled samples were separated into sample specific readIDs based on the barcode information. For each read in the pooled sample fastq file, 25bp containing the barcode and primer was trimmed from the 5’ end. Using the trimmed fastq file and readIDs obtained in the first step, sample specific fastq files were generated using BBMap (v 38.11). These reads were mapped to the RUVBL1 or RUVBL2 cDNA sequence using BWA (v. 0.7.15). Mismatched reads were separated from exact match reads using “NM:i:0” tag in the bam file. Read pileup information was obtained from these mismatched reads using the samtools (v.0.1.19) mpileup command. Finally, from the read pileup information, per base mismatch information was obtained using pileup2base tool. Per base error ratios were computed using an in-house Perl script.

Immunoprecipitation

For RUVBL1-V5 or RUVBL2-V5 immunoprecipitation, for each condition, 2 million H2009 cells stably expressing pIRESneo3-RUVBL1-RNAiR-V5 or pIRESneo3-RUVBL2-RNAiR-V5 were reverse transfected with 50 mL Lipofecatmine RNAiMAX and 20 nM siRUVBL1 or 20 nM siRUVBL2 (per dish) in 4 15cm dishes following manufacturer’s incubation times. For immunoprecipitation experiments involving drug treatment, 3 days after siRNA transfection cells were treated with 100 nM Compound B or Compound C for 12 hours, cells were washed twice with ice-cold PBS, scraped on ice and technical replicates were combined. For immunoprecipitation experiments without drug, cells were harvested as above 3 days after siRNA treatment. After scraping cells, all subsequent steps were performed on ice or at 4°C. Cells were lysed in 1 mL high salt NP-40 buffer (25 mM HEPES, 420 mM NaCl, 5 mM EDTA, .4% IGEPAL CA-630, 10% glycerol, pH 7.5) with protease inhibitors (MilliporeSigma, catalog # P8340) and phosphatase inhibitors (MilliporeSigma, catalog #4906845001) for 1 hour while rotating at 4°C, then spun at max speed for 15 minutes. For Compound B and Compound C treated samples, protein concentrations were determined and normalized using BCA (ThermoFisher Scientific, catalog # 23225), 5% of the sample was saved for an Input sample, then the normalized lysate was mixed with 3 volumes dilution buffer (25 mM HEPES, 2% glycerol, pH 7.5) and filtered through a .45 micron PVDF filter (MilliporeSigma, catalog# SLHV033RS). For samples involving an IgG control (no drug treatment), 1/3^rd volume from siRUVBL1 and 1/3^rd volume from siRUVBL2 samples were mixed together for the IgG sample. For samples involving drug treatment and mass spectrometry, no IgG control was used as immunoprecipitated proteins from Compound B were normalized to Compound C. 100 μLof Protein G-Gamma Sepharose beads (GE Healthcare, catalog# 17088501) per sample were washed 3 times with 1 mL of 3:1 dilution:lysis buffer, resuspended in 200 μL 3:1 dilution:lysis buffer and then added for 30 minutes for pre-clearing. 15 μg V5 antibody (Genscript catalog #A01724) or 15 μg Mouse IgG (Cell Signaling, catalog# 5415) was incubated with respective samples overnight. In the morning, 50 μL of Protein G-Gamma Sepharose beads per sample were washed 3 times in 1 mL 3:1 dilution:lysis buffer and then resuspended in 100 μL 3:1 dilution:lysis buffer, added to each sample, and rotated for 2 hours. Beads were then washed 3 times for 5 minutes using 1 mL 3:1 dilution:lysis buffer, and proteins were eluted using 50 μL of 500 μg/mL V5 peptide (MiliporeSigma, catalog# V7754) by shaking at 12°C at 900 RPM for 3 hours. Beads were pelleted, supernatant collected, and supernatant was passed through an insulin syringe (BD Biosciences, catalog# 324909). Samples were spun again, and supernatant was mixed with 4X Laemmli buffer (BioRad, catalog #1610747), boiled for 5 minutes, loaded onto a 4–20% Mini-PROTEAN TGX gel (BioRad, catalog # 4561096) and ran a distance of 2 cm. The gel was washed 3 times for 5 minutes with deionized water, stained with Coommassie (Biorad, catalog# 1610436) for 15 minutes and destained (45% methanol, 10% acetic acid, 45% deionized water) for 1 hour. Gels were imaged using the Licor Odyssey Fc. Lanes were cut, cubed into 1 mm³ pieces, then prepped for mass spectrometry (below).

Mass Spectrometry

For all samples, mass spectrometry was performed at UTSW’s proteomics core using the following protocol: gel pieces were digested overnight with trypsin (Pierce) following reduction and alkylation with DTT and iodoacetamide (MilliporeSigma). The samples then underwent solid-phase extraction cleanup with an Oasis HLB microelution plate (Waters) and the resulting samples were analyzed by liquid chromatograph followed by tandem mass spec (LC-MS/MS) using an Orbitrap Fusion Lumos mass spectrometer (ThermoElectron) coupled to an Ultimate 3000 RSLC-Nano liquid chromatography system (Dionex). Samples were injected onto a 75 μm i.d., 50-cm long EasySpray column (Thermo), and eluted with a gradient from 1–28% buffer B over 60 min. Buffer A contained 2% (v/v) ACN and 0.1% formic acid in water, and buffer B contained 80% (v/v) ACN, 10% (v/v) trifluoroethanol, and 0.1% formic acid in water. The mass spectrometer operated in positive ion mode with a source voltage of 2.4 kV and an ion transfer tube temperature of 275°C. MS scans were acquired at 120,000 resolution in the Orbitrap and up to 10 MS/MS spectra were obtained in the ion trap for each full spectrum acquired using higher-energy collisional dissociation (HCD) for ions with charges 2–7. Dynamic exclusion was set for 25 seconds after an ion was selected for fragmentation. Raw MS data files were converted to a peak list format and analyzed using the central proteomics facilities pipeline (CPFP), version 2.0.3 (Trudgian and Mirzaei, 2012). Peptide identification was performed using the X!Tandem (Craig and Beavis, 2004) and Open MS Search Algorithm (OMSSA) (Geer et al., 2004) search engines against the human protein database from Uniprot, with common contaminants and reversed decoy sequences appended (Elias and Gygi, 2007). Fragment and precursor tolerances of 20 ppm and 0.6 Da were specified, and three missed cleavages were allowed. Carbamido-methylation of Cys was set as a fixed modification and oxidation of Met was set as a variable modification. Label-free quantitation of proteins across samples was performed using SINQ normalized spectral index Software (Trudgian et al., 2011).

To identify true RUVBL1/2 interacting proteins in Figure S4A we required co-immunoprecipitating proteins in both IP:RUVBL1-V5 and IP:RUVBL2-V5 samples to have an average enrichment of >3 fold over IgG samples (“Selected” genes) using SINQ normalized spectral indices (Trudgian et al., 2011). We then called the overlapping proteins from these two lists (IP:RUVBL1-V5 and IP:RUVBL2-V5) as true RUVBL1/2 interacting proteins. RUVBL1/2 co-immunoprecipitating proteins were ran through the gene enrichment program Enrichr (Kuleshov et al., 2016) using default settings, and gene sets were called as enriched if they were found using both the KEGG and Reactome databases. The final list of identified peptides and results from Enrichr can be found in Table S3.

Chemical Screening Strategy for Compound B and RUVBL1/2 ATPase Activity

The ATPase activity of RUVBL1/RUVBL2 were evaluated using recombinant RUVBL1 and RUVBL2 proteins. For screening, RUVBL1 and RUVBL2 cDNAs were purchased from Open Biosystems and cloned into the pGEX-6P-1 plasmid (GE Healthcare, catalog # 28–9546-48) for expression in BL21 STAR (DE3) E. coli (ThermoFisher Scientific, catalog# C601003). To express and purify RUVBL1 mutant proteins, rat RNAi-resistant RUVBL1 cDNA from the pIRESneo3-RUVBL1-RNAiR plasmid were cloned via PCR into pGEX-6P-1. E. coli were sonicated in 20 mM Tris-HCl (pH 8), 1 mM EDTA, 1% Triton X-100,1mM PMSF, 5 mM DTT, with a protease inhibitor cocktail (MilliporeSigma, catalog # P8340), and then supernatants were mixed with Glutathione Sepharose 4 Fast Flow beads for 2 hours while stirring at 4°C (GE Healthcare, catalog# 17513201). Beads were loaded onto a column, washed with 20 mM Tris-HCl (pH 8), 1% Triton X-100 and 500 mM NaCl, and then washed with PreScission Buffer (50 mM Tris-HCl, 150 mM NaCl and 1 mM DTT, pH 7). The column was then incubated with PreScission Buffer with PreScission Protease (GE Healthcare, catalog# 27–0843-01) overnight at 4°C with shaking. The column was then eluted with PreScission Buffer, reloaded onto the column, and eluted again. For chemical screening, a 2-fold assay buffer (100 mM Tris-HCl pH 7.5,100 mM NaCl, 40 mM MgCl2, 2 mM DTT, 20 μM ATP, and 0.2 mg/mL BSA) was dispensed at 24.5 μL/well to a 384-well assay plate. Various concentrations of the test compound solution was dispensed thereto at 1 μL/well. Recombinant RUVBL1 and RUVBL2 were diluted to a concentration of 50 μg/mL using sterilized water. The diluted solution was dispensed at 24.5 μL/well to the 384-well assay plate, mixed using a plate mixer, and left standing at room temperature for 1.5 hours. 5 μL of the solution in each well was transferred to a white assay plate, and ADP-Glo Reagent (Promega Corp., catalog# V9101) was dispensed thereto at 5 μL/well. After the completion of the reaction, the plate was left standing at room temperature for 40 minutes. Kinase Detection Reagent was further dispensed thereto at 5 μL/well, the plate was left standing at room temperature for 30 minutes, and luminescence intensity was measured using a plate reader.

The % ATPase inhibitory activity for each compound was calculated according to the expression given below for luminescent intensities, where A = luminescence in well with recombinant RUVBL1/2 protein and test compound, B = luminescence in well with recombinant RUVBL1/2 protein, and C = luminescence in well with neither test compound nor recombinant RUVBL1/2 protein.

$$\text{\%}\text{ATPase}\text{inhibitory}\text{activity}\text{=}\text{100}\text{*}\text{[(A}\text{−}\text{C)}\text{/}\text{(B}\text{−}\text{C)]}$$

Curve fitting was performed using GraphPad Prism 4 from the concentration of each of serially diluted test compound and the % ATPase inhibitory activity at this concentration. This curve was then used to determine the concentration at which the ATPase activity was inhibited by 50% (IC₅₀ value).

Pharmacokinetics for Compound B

Compound B suspended in 0.5% methylcellulose was administered orally to 7-week-old male BALB/c mice purchased from CLEA Japan, Inc and housed by Daiichi-Sankyo. Plasma samples were obtained from blood collected from each animal into tubes containing sodium heparin at 0.5, 1, 2, 4, 7 and 23 hours after administration. The concentration of Compound B in mice plasma samples was determined using API4000 LC-MS/MS system (Applied Biosystems/MDX SCIEX). Pharmacokinetic parameters of Compound B were calculated by non-compartmental analysis using the computer software BioBook (Ver. 9.4.0, ID Business Solutions Ltd.).

Dose-Response Curves

For all experiments involving dose-response curves with viability readouts, 1000–2000 cells were plated into each well of a 96-well tissue culture treated plate. After 24 hours, drugs were added as 8-point serial dilutions. After 4 more days, viability was determined using CellTiter 96 Aqueous One Solution Cell Proliferation Assay (MTS) (Promega, catalog# G3582) on a Molecular Devices SpectraMax 190 microplate reader. Curve fitting was performed in GraphPad Prism using “[inhibitor] vs normalized response with variable slope”.

RNA-Sequencing

For H2009 siRUVBL1 experiments, RNA was harvested 2.5 days after transfection; RNA for H596 siRUVBL1 experiments was harvested 3 days after transfection to ensure similar levels of protein knockdown and cell doublings. For Compound B experiments, RNA was extracted 24 hours after treatment in both cell lines. RNA was extracted and genomic DNA discarded using the RNeasy Plus Mini Kit (Qiagen, catalog# 74134). All RNA had a RIN score >9 prior to sequencing as determined by an Agilent Tapestation 4200. Libraries were assembled using 4 μg of poly adenylated RNA with Illumina’s TruSeq Stranded mRNA library prep kit (Illumina, catalog # RS-122–2101) following manufacturers protocols. Libraries were sequenced on the Illumina NextSeq 500 using V2 reagents and 75 nucleotide single-end reads for an average of 30–40 million reads. TopHat (v 2.0.12), Cufflinks (v 2.2.1) and Cuffdiff (v2.2.1) were used for read mapping (using hg19), transcript calling, and differential expression analysis, respectively. For all differential expression analyses, Compound B was compared to Compound C, Compound C was compared to untreated cells, and siRUVBL1 was compared to siNTC. All RNA-seq datasets have been deposited under GEO accession number GSE107637.

GSEA Analysis

For GSEA of RNA-seq data, RPKM values for genes with RPKM >1 in treated and untreated samples were fed into GSEA (V 2.2.2). For GSEA of RUVBL1/2 correlated genes and GSEA of genes correlated to RUVBL1/2 RSA scores in Project DRIVE, rank ordered correlation values were fed into GSEA (v 2.2.2). The following gene sets from MSigDB were used in all GSEA analyses: H, C1, C2 (CGP, CP, CP:KEGG, CP:Reactome), C5, and C6.

Xenograft Studies

For studies involving Compound B alone, 1 million H2009 or H596 cells in 100μL ice-cold PBS were injected subcutaneously through a 25G needle into the right rear flank of ~10-week-old female NOD/SCID (NOD.CB17-Prkdc^SCID/J) mice under isoflurane. All mice were pre-conditioned to oral gavage (gavaged orally without drug) twice prior to treatment, and the gavage tip was always dipped in a 10% sucrose solution to reduce animal stress (Hoggatt et al., 2010). Once average tumor size of all mice reached ~150 mm³, (calculated length * width² * .52, where length is the longest axis), mice were randomized into equal groups, weighed routinely and treated via oral gavage with 175 mg/kg/day Compound B using flexible gavage tips (Instech, catalog# FTP-20–38). The treatment schedule was as follows: 4 days on treatment, 9 days off, 3 days on, 7 days off, 3 days on. On treatment days, mice were orally gavaged in the morning (9 A.M.) and evening (5 P.M.) with 100uL of vehicle (Polyethylene glycol 200, MilliporeSigma, catalog# 88440) or Compound B twice per day, such that the total dose per day was 175 mg/kg Compound B. Compound B was dissolved in polyethylene glycol 200 by douncing vigorously using a PFTE pestle and glass tube (MilliporeSigma, catalog# P7859). Tumor tissue for immunoblot, H&E and IHC was harvested after 3 days of treatment.

For studies involving Compound B and IR, 2 million H1299 cells in 100 μL ice-cold PBS were injected subcutaneously through a 25G needle into the right leg/low hip of ~10-week-old female Nude (Nu/Nu) mice under isoflurane. Mice were pre-conditioned to oral gavage, tumors measured, mice weighed, orally gavaged and treated with vehicle or Compound B as stated above for Compound B alone treatment, except that the total dose was 125 mg/kg/day on treatment days, and using the following treatment schedule: mice were treated at 9 A.M. with drug or vehicle, irradiated with 2 Gy at 12 P.M. (if applicable), then given drug or vehicle at 5 P.M. for four days. The fifth day, mice received only 2 Gy irradiation (if applicable). Irradiation was delivered specifically to the tumor using a PXI Precision X-ray X-RAD 320 Series machine while mice were under isoflurane.

For studies involving shRUVBL1 and shNTC, 1 million H1299 cells transduced with pTRIPZ doxycycline-inducible shNTC or shRUVBL1 in 100uL ice-cold PBS were injected subcutaneously through a 25G needle into the right rear flank of ~8-week-old female NOD/SCID (NOD.CB17-Prkdc^SCID/J) mice under isoflurane. Once average tumor volume reached ~150 mm³, 2 mg/mL doxycycline (MilliporeSigma, catalog# D9891) and 2% sucrose was added to mice drinking water until the time of sacrifice. Tumor tissue was harvested for immunoblots after 14 days of doxycycline treatment.

Isolation of Proteins on Nascent DNA (iPOND)

We modified the protocol of (Sirbu et al., 2012) to increase sonication efficiency and ensure equal amounts of protein lysate were utilized for pull down. The same protocol was utilized for iPOND-mass spec and iPOND-immunoblot, except that iPOND-immunoblot included additional controls (No Click and Chase samples). For each condition (No Click, Chase, 100 nM Compound C 12 hours, 100 nM Compound B 12 hours), 7 million H2009 cells were plated into 6 15cm dishes, as well as one additional plate for cell counting. Cells attached overnight and were then treated with 100 nM of Compound B or Compound C for 12 hours. After 11hours, cells on the extra plate were trypsinized and counted to determine cell numbers in each condition. EdU (Cayman Chemicals, catalog #20518) at a final concentration of 10 μM was added to Chase samples for 10 minutes, cells were washed with a solution of R5 (media) + 10 μM thymidine, then maintained in R5 + 10 μM thymidine for 1 hour. After 12 hours of drug treatment, Compound C and Compound B treated cells were treated with 10 μM of EdU for 10 minutes, cells were cross-linked with 1% formaldehyde (Sigma-Aldrich catalog # 252549) in PBS for 20 minutes, cross-links were quenched using 1.25M glycine (Sigma-Aldrich catalog # 50046), washed once with ice-cold PBS, scraped in ice-cold PBS on ice and technical replicates were combined. After 1 hour of R5 + 10 μM thymidine treatment, Chase cells were harvested as above. For No Click cells, cells were treated with 10 μM EdU for 10 minutes then harvested as above. Cells were washed twice in ice-cold PBS, then resuspended in 1 mL nuclear release buffer (pH 7.9 25 mM HEPES, 10 mM KCl, 1.5 mM MgCl₂, .5 mM DTT and protease inhibitors, MilliporeSigma, catalog # P8340), rotated at 4°C for 30 minutes, then dounced 10 times in a 2 mL Kontes Dounce Homogenizer (VWR catalog # KT885300–0002). Nuclei were pelleted, supernatant aspirated, and cells were resuspended in permeabilization buffer (.25% Triton X-100 in PBS) such that there were 10 million cells/mL. Cells were slowly agitated at room temperature for 30 minutes, spun down, supernatant discarded, resuspended in ice-cold .5% BSA (Jackson ImmunoResearch, catalog # 001–000-173) in PBS and spun down. Supernatant was aspirated, cells were washed with PBS, then Chase, Compound B and Compound C treated samples were resuspended in click solution at 100 million cells / 5 mL, which consists of 10 μM biotin azide (Click Chemistry Tools, catalog # 1265–6), 10 mM sodium ascorbate (SantaCruz catalog # sc-215877) and 2 mM CuSO₄ (SantaCruz, catalog # sc-203009A), added to PBS in that order, for 2 hours at room temperature with agitation. No Click cells were resuspended in a similar solution, but with biotin azide replaced with an equal volume of DMSO. Cells were spun down, supernatant aspirated, cells resuspended in ice-cold .5% BSA in PBS, spun down, washed with PBS, spun down, and supernatant aspirated. Cells were resuspended in lysis buffer (1% SDS in 50mM Tris, pH 8 with protease inhibitors) at a concentration of 10 million cells / 100 μL, moved to a 15 mL polystyrene conical, rotated for 15 minutes at 4°C, then sonicated 3 times for 15 minutes each on HIGH using a Diagenode Bioruptor Plus water bath sonicator, with fresh ice-cold water added after each 15 minute run. Lysate was transferred to a 1.5 mL eppindorf tube, spun at 16000g for 10’ at 4°C, the top lipid bilayer was aspirated, then supernatant was passed through a 70 μm nylon filter (Sigma-Aldrich, catalog # CLS431751). Protein lysate was quantified and normalized using BCA (ThermoFisher Scientific, catalog # 23225). From this protein lysate, 5 μL was decrosslinked by adding 90 μL H2O and 4 μL 5 M NaCl, incubating at 65°C overnight, adding 1 μL RNase A, incubating at 37°C for 30 minutes, adding 2 μL .5 M EDTA, 4 μL 1 M Tris and 1 μL proteinase K and incubating at 45°C for 2 hours. De-crosslinked DNA was isolated using QIAquick PCR purification kit (Qiagen, catalog # 28104), then ran on a .75% agarose gel to ensure that sonication yielded 100–300bp fragments. To the normalized protein lysate, 1 mL ice-cold PBS with protease inhibitors was added. Per condition, 100 μL streptavidin agarose beads (Millipore, catalog # 161–0436) were prepared by washing beads twice with 1 mL ice-cold lysis buffer and once with 1 mL ice-cold PBS + protease inhibitors. Streptavidin agarose beads were resuspended in 100 μL PBS + protease inhibitors, added to protein lysates, and rotated at 4°C for 20 hours. Beads were washed with ice-cold lysis buffer for 5 minutes, spun down, washed with room temperature 1 M NaCl for 5 minutes, spun down, then washed two more times with ice-cold lysis buffer as done previously. Beads were collected, supernatant aspirated, 60 μL of 2X Laemmli Sample Buffer (Biorad, catalog #1610737) added, and beads were boiled for 25 minutes to elute proteins. Eluted proteins were collected and either used directly for immunoblot following the protocol outlined in this manuscript or loaded onto a 4–20% Mini-PROTEAN TGX gel (BioRad, catalog # 4561096) and ran a distance of 2 cm for mass spectrometry. The gel was washed 3 times for 5 minutes with deionized water, stained with Coommassie (Biorad, catalog# 1610436) for 15 minutes and destained (45% methanol, 10% acetic acid, 45% deionized water) for 1 hour. Gels were imaged using the Licor Odyssey Fc, then lanes were cut (without the steptravidin band), cubed into 1 mm³ pieces, and prepped for mass spectrometry as described above in this manuscript. For analysis, because Compound B iPOND samples consistently pulled down less overall non-contaminant proteins, the abundance of each protein in Compound B treated samples was multiplied by (total abundance of proteins in Compound B treated samples / total abundance of proteins in Compound C treated samples). For each individual protein in each replicate, the normalized abundance of the Compound B treated protein was then divided by the abundance of the Compound C treated protein and biological replicates were averaged to find the fold enrichment/depletion.

Flow Cytometry Experiments

Cell cycle analyses by DNA content were performed by pelleting 400,000 cells and washing twice with ice-cold PBS. Cells were resuspsended in 1 mL of ice-cold PBS, then 2.5 mL of 100% ice-cold ethanol was added dropwise while vortexing cells slowly. The sample was then stored at −20°C at least overnight, pelleted, washed with ice-cold PBS, and then resuspended in 500 μL of PI staining solution consisting of .1% Triton X-100 in PBS, 50 μg/mL propidium iodide (ThermoFisher Scientific, catalog# P3566), 100 μg/mL RNase A (Qiagen, catalog# 19101), and 2% FBS (ThermoFisher Scientific, catalog# 26140079) and placed in a 37°C waterbath for 1 hour protected from light. Samples were washed once with ice-cold PBS+2% FBS, resuspended in 500 μL ice-cold PBS+2% FBS, then filtered into a polystyrene flow tube with a 35 μm strainer (Corning, catalog# 352235) and analyzed with a BD LSRFortessa flow cytometer using BD FACSDIVA software, such that 10,000 single, live cells were gated and collected per run. The resulting gated FCS2 files were then analyzed using FlowJo software.

For synchronization/BrdU incorporation assays, 250,000 H2009 cells were plated in a T75 and allowed to adhere overnight, before 2 mM thymidine (MilliporeSigma, T1895) was added for 18 hours. Cells were then washed with PBS and allowed to grow in normal media for 9 hours, after which 2 mM thymidine was added for 10 hours. After 10 hours, 100 nM Compound B or Compound C was added for 6 hours, then cells were washed with PBS (released from thymidine block) and kept in 100 nM Compound B or Compound C for the duration of the assay. At indicated time points post release (0, 3, 6, 12, 18 hours), 50 μM BrdU (MilliporeSigma, catalog# B5002) was added for 30 minutes to media before harvesting cells. Cells were harvested by pelleting 1,000,000 cells, washing twice with ice-cold PBS, resuspending the pellet in 1 mL ice-cold PBS, and then 2.5 mL 100% ice-cold ethanol was added drop wise while gently vortexing cells. Samples were stored at −20°C at least overnight, then pelleted and washed with PBS. Cells were permeabilized and DNA denatured by resuspending in 1 mL of 2 M HCl + .5% Triton X-100 for 30 minutes, then spun down and resuspended in 1mL of .1 M sodium tetraborate (MilliporeSigma, catalog# 221732). Cells were washed with .1% TBST, then blocked in 500 μL 5% BSA (Jackson ImmunoResearch, catalog# 001–000-173) in .1% TBST for 1 hour, and resuspended in 250 μL 5% BSA + 5 μL anti-BrdU antibody (Santa Cruz, catalog# sc-13119) overnight at 4°C. In the morning, cells were washed 3 times with .1% TBST, then resuspended in 250 μL BSA+ 1.25 μL anti-mouse Alexa Fluor 488 (ThermoFisher Scientific, Cat# A-11029) and rotated protected from light for 1 hour. Cells were washed 3 times with .1% TBST for 5 minutes, then stained with propidium iodide and analyzed using the cell cycle protocol outlined above (starting at the addition of 500uL PI staining solution), except that at least 30,000 single, live cells were gated and collected per sample and overlapping fluorescent signals were mitigated via compensation.

For native BrdU assays (detecting ssDNA) cells were first labeled with 10 μM BrdU (MilliporeSigma, catalog# B5002) for 48 hours. Then, following treatment with Compound C, Compound B or hydroxyurea + VE-822 in the presence of 10 μM BrdU for indicated times, 3,000,000 cells were pelleted and prepared for flow cytometry following a protocol identical to the above synchronization/BrdU assay, except with the exclusion of the 2 M HCl and .1 M sodium tetraborate steps. During flow cytometry, a total of 50,000 single, live cells were gated and collected per sample and overlapping fluorescent signals were mitigated via compensation.

RPA loading assays were performed by pelleting 3,000,000 cells and washing twice with ice-cold PBS. Cells were pre-extracted on ice by permeabilizing with 1 mL of ice-cold .2% Triton X-100 in PBS for 1 minute. This solution was then immediately diluted with 10 mL ice-cold PBS and spun down at 4°C. Cells were resuspended in 1 mL 4% paraformaldehyde (Fischer Scientific, catalog# 50–980-487), fixed in a 37°C waterbath for 10 minutes, kept on ice for 1 minute, and then spun down at 4°C. Cells were resuspended in 1 mL ice-cold PBS, then 9 mL of 100% ice-cold methanol was added drop wise while gently vortexing cells. This solution was stored at least overnight at −20°C, spun down, transferred to a polypropylene tube, washed with 1 mL .1% TBST, and blocked using 500 μL of 5% BSA (Jackson ImmunoResearch, catalog# 001–000-173) in .1% TBST for 1 hour while rotating. Cells were resuspended in 250 μL 5% BSA + 2.5 μL anti-RPA2 antibody (abcam, Cat# ab2175) and rotated overnight at 4°C. Cells were washed 3 times for 5 minutes with .1% TBST, then resuspended in 250 μL 5% BSA+ 1.25 μL anti-mouse Alexa Fluor 488 (ThermoFisher Scientific, Cat# A-11029) and rotated protected from light for 1 hour. Cells were washed 3 times with .1% TBST for 5 minutes, then stained with propidium iodide and analyzed using the cell cycle protocol outlined above (starting at the addition of 500 μL PI staining solution), except that at least 30,000 single, live cells were gated and collected per sample and overlapping fluorescent signals were mitigated via compensation.

For TUNEL assays, cells were synchronized and treated the same as the BrdU incorporation assay (above), except that 100 nM of the CHEK1 inhibitor LY2603618 (ApexBio, catalog# A8638) was added following the second (final) release from thymidine. At specified time points, cells were harvested by pelleting 500,000 cells, washing twice with ice-cold PBS and fixing with 1% paraformaldehyde (Fischer Scientific, catalog# 50–980-487) for 15 minutes on ice. Cells were then resuspended in 1 mL ice-cold PBS, and 2.5 mL of ice-cold 100% ethanol was added dropwise while slowly vortexing cells. This was stored at least overnight at −20°C, spun down, washed with PBS, and then stained using the APO-BrdU TUNEL Assay Kit, with Alexa Fluor 488 Anti-BrdU (ThermoFisher Scientific, catalog# A23210) only using polypropylene tubes and following manufacturers protocol with the following modifications: PBS + 2% BSA (Jackson ImmunoResearch, catalog# 001–000-173) was used instead of wash buffer to prevent cell sticking, and DNA staining and analysis was performed using the protocol outlined above for cell cycle analyses, instead of the kits solution, starting at the addition of 500 μL PI staining solution. At least 30,000 single, live cells were gated and collected per sample and overlapping fluorescent signals were mitigated via compensation.

Immunofluorescence

Cells were plated into glass chamberslides (ThermoFisher Scientific, 154526) and allowed to adhere overnight. Cells were then treated as indicated, media was aspirated, cells were fixed with 4% paraformaldehyde (Fischer Scientific, catalog# 50–980-487) for 10 minutes, washed 3 times with TBS, then permeabilized with .5% Triton-X 100 on ice for 10 minutes. After 3 TBS washes, cells were blocked with 5% BSA (Jackson ImmunoResearch, catalog# 001–000-173) in .1% TBST for 1 hour. Following blocking, cells were incubated with primary antibody + 5% BSA, using anti-γH2AX (1:1000, MilliporeSigma, catalog# 16–193), anti-phospho-Histone H3 (1:500, Abcam, catalog# ab5176), or anti-53BP1 (1:500, Cell Signaling, catalog# 4937) antibodies, for 2 hours. Cells were then washed 3 times for 5 minutes with .1% TBST and incubated with secondary antibody + 5% BSA, either anti-mouse Alexa Fluor 488 (1:500, ThermoFisher Scientific, catalog# A-11029) or anti-rabbit Alexa Fluor 555 (1:500, ThermoFisher Scientific, catalog# A-31572), for 1 hour protected from light. Chambers were then washed 3 times for 5 minutes with .1% TBST, chambers were removed, and glass coverslips (VWR, catalog# 48404–133) were mounted using Vectashield with DAPI (Vector Laboratories, catalog #H-1200). Images were captured at 40X using a Leica DM5500 or a Keyence BZ-X710 and analyzed for foci using Cell Profiler (1.0.9717) or BZ-X Analyzer (v1.3.1.1, Keyence), respectively, with similar results. Pan-γH2AX quantification was performed manually on 10X magnification.

DNA Fiber Analysis

Cells were labeled sequentially with 100 μM iododeoxyuridine (IdU) for 10 min and chlorodeoxyuridine (CldU) for 20 min. DNA fibers were spread as described (Merrick et al., 2004). Briefly, labeled cells were resuspended in PBS at 1 × 10⁶ cells/mL, lysed with SDS lysis buffer (0.5% SDS, 200 mM Tris-HCl, pH 7.4, 50 mM EDTA) and this solution was placed onto a glass slide. This slide was then placed at an angle and solution was allowed to run down the slide for 8 minutes. DNA fibers were fixed with methanol/acetic acid (3:1) and stained with primary antibodies (mouse anti-BrdU/IdU, BD Bioscience, catalog # 347580; rat anti-BrdU/CldU, clone BU1/75, Accurate Chemical & Scientific, catalog # OBT0030) and fluorescence-conjugated secondary antibodies (anti-rat Alexa Fluor 488, Invitrogen, catalog # 11006; anti-mouse Alexa Fluor 568, Invitrogen, catalog # A-11004). Fibers were imaged using the Zeiss AxioImagerM2 and measured using the Zen 2.5 lite software. Fiber length was only determined on red-green (i.e. only on dually IdU-CldU labeled) tracks. Two biological replicates were performed with >100 tracks measured in each condition in each replicate.

Surviving Fraction Curves

NSCLC and HBEC cells were trypsinized and re-suspended as single cell suspensions. Cells were counted (Z2 Particle Counter, Beckman Coulter) and a fixed number of cells was plated for individual doses of radiation across experimental groups, with increasing number of cells for higher levels of radiation. Counted cells were seeded in 60 mm tissue culture dishes in triplicate for each dose of radiation, allowed to attach to the dish for 6–8 hours, then irradiated at various doses using a 137Cs irradiator (Mark 1–68 irradiator, J.L. Shepherd and associates). Irradiated plates were then incubated until colonies formed, where a colony is defined as ~50 cells. The colonies were fixed and stained with 4% formaldehyde (Fischer Scientific, catalog# 50–980-487) in PBS containing 0.05% crystal violet (MilliporeSigma, catalog# C6158), and colonies were counted manually using a light microscope. The surviving cell fraction was calculated as: (Mean colony counts) / [(cells plated) X (plating efficiency)], in which plating efficiency was defined as (Mean colony counts) / (cells plated for unirradiated control). Curve fitting was performed using the Liner Quadratic equation in SigmaPlot (Systat Software Inc.), and SF2 values are from this fitted curve.

QUANTIFICATION AND STATISTICAL ANALYSIS

All statistical analyses were performed with GraphPad Prism (Version 7) unless otherwise stated. Differences between normal and tumor for RUVBL1/2 mRNA in TCGA data (Figure S1A) was evaluated by 1-way ANOVA with post-hoc Dunnett’s correction. Analysis of Kaplan-Meier plots for RUVBL1/2 mRNA levels (Figure S1B) was done using log-rank test and default settings on KMplot (Szasz et al., 2016). Differences in RUVBL1 staining in different patient subsets (Figures S1G–S1J) was done using 1-way ANOVA with post-hoc Tukey correction. Analysis of Kaplan-Meier survival curves for RUVBL1 protein levels (Figure S1F) was done using log-rank (Mantel-Cox) test. Correlations were performed using linear regression (Figures S1L, S2B, and S2D–S2F). Overlap p-values for Figures 3B and 3C were computed using hypergeometric tests. Statistics in xenograft studies were calculated by repeated measures 2-way ANOVA. Where appropriate, figure legends define n, which indicates the number of patients (for clinical data), the number of mice utilized (for xenograft data) or the number of biological replicates (for all other data).

DATA AND CODE AVAILABILITY

Sequencing data was deposited in GEO under accession number GSE107637.

SIGNIFICANCE.

RUVBL1/2 are known to be required for the assembly of various multiprotein complexes, have ATPase activity, and have been implicated in cancer, but the role of their ATPase activity and their utility as therapeutic targets has not been assessed. Utilizing a potent and specific RUVBL1/2 ATPase inhibitor, compound B, we show that RUVBL1/2 ATPase activity is required for the maturation or disassembly of the PAQosome complex but not for the assembly or disassembly of the INO80 family of chromatin remodelers. RUVBL1/2 are known to be essential for cellular growth, and here we show that most patient-derived non-small cell lung cancer (NSCLC) lines require RUVBL1/2 ATPase activity for DNA replication. Molecularly, inhibition of RUVBL1/2 ATPase activity initially decreases the number of replication forks; however, the subsequent loss of ATR causes this to progress into cancer cell apoptosis via replication catastrophe. Therapeutically, we show that treatment of NSCLC xenografts with compound B results in modest efficacy due to a narrow therapeutic window. However, RUVBL1/2 inhibition enhances the efficacy of radiation in cancerous cells but not normal cells, affording a larger therapeutic window. Thus, the combination of RUVBL1/2 inhibition with radiation may represent the best path forward for future potential clinical translation of RUVBL1/2 inhibitors.

Highlights.

• Validation of compound B as a specific RUVBL1/2 ATPase inhibitor

• RUVBL1/2 ATPase activity is required for PAQosome, not INO80-family, complexes

• NSCLC requires RUVBL1/2 ATPase activity for DNA replication

• Compound B radiosensitizes NSCLC, but not normal lung epithelial cells