Clofarabine Commandeers the RNR-α-ZRANB3 Nuclear Signaling Axis

Article

Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis Graphical Abstract

Authors Marcus J.C. Long, Yi Zhao, Yimon Aye

Correspondence [email protected]

In Brief Regulators of Ras oncogenic signaling, particularly non-essential genes, are critically important targets for drug discovery. Long et al. show that ZRANB3, a non-essential gene, is required for HRas-induced focus formation. This process is inhibited by an RNRa-targeting drug, clofarabine that promotes RNR-a nuclear translocation, thereby ushering previously unrecognized therapeutic opportunities.

Highlights d

ZRANB3 is required for DNA synthesis in various cell types

d

Drugs promoting nuclear translocation of enzyme RNR-a inhibit this ZRANB3 function

d

ZRANB3 is an example of non-oncogene addiction during H-rasG12V-driven transformation

d

Upregulating the levels of nuclear RNR-a offers a means to target oncogenesis

Long et al., 2020, Cell Chemical Biology 27, 1–12 February 20, 2020 ª 2019 Elsevier Ltd. https://doi.org/10.1016/j.chembiol.2019.11.012

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

Cell Chemical Biology

Article Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis Marcus J.C. Long,2,3 Yi Zhao,1,3 and Yimon Aye1,4,* 1Swiss

Federal Institute of Technology Lausanne (EPFL), Lausanne 1015, Switzerland Burton, Beverley, East Riding of Yorkshire HU17 8QH, UK 3These authors contributed equally 4Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.chembiol.2019.11.012 2Bishop

SUMMARY

Ribonucleotide reductase (RNR) is an essential enzyme in DNA biogenesis and a target of several chemotherapeutics. Here, we investigate how antileukemic drugs (e.g., clofarabine [ClF]) that target one of the two subunits of RNR, RNR-a, affect noncanonical RNR-a functions. We discovered that these clinically approved RNR-inhibiting dATP-analogs inhibit growth by also targeting ZRANB3—a recently identified DNA synthesis promoter and nuclear-localized interactor of RNR-a. Remarkably, in early time points following drug treatment, ZRANB3 targeting accounted for most of the drug-induced DNA synthesis suppression and multiple cell types featuring ZRANB3 knockout/knockdown were resistant to these drugs. In addition, ZRANB3 plays a major role in regulating tumor invasion and H-rasG12Vpromoted transformation in a manner dependent on the recently discovered interactome of RNR-a involving select cytosolic-/nuclear-localized protein players. The H-rasG12V-promoted transformation— which we show requires ZRANB3-supported DNA synthesis—was efficiently suppressed by ClF. Such overlooked mechanisms of action of approved drugs and a previously unappreciated example of nononcogene addiction, which is suppressed by RNRa, may advance cancer interventions.

INTRODUCTION Ribonucleotide reductase (RNR) is a dual-subunit enzyme essential for deoxyribonucleotide triphosphate (dNTP) synthesis and maintaining absolute and relative dNTP levels (Aye et al., 2015; Chabes and Thelander, 2003; Nordlund and Reichard, 2006). One subunit, RNR-b, is present only in S-phase. The other subunit, RNR-a, is present throughout the cell cycle (Thelander, 2007). RNR reductase activity occurs by dint of a catalytically essential cysteine (C429) within RNR-a, which is activated via a proton-coupled electron transfer process initiated by a metallocofactor within RNR-b (Stubbe and van Der Donk, 1998). An

alternative RNR-b subunit, RNR-p53b, is constitutively present but exists only at low levels (Thelander, 2007). Because both proofreading and misincorporation rates are acutely sensitive to relative and absolute dNTP levels (Haber, 2013), dNTP pool maintenance regulated by RNR—beyond its role in dNTP provision—is essential for genome integrity (Mathews, 2015). RNR-a is an excellent sensor of (d)NTPs. This (deoxy)nucleotidesensing function is coupled to reductase activity: ATP is an allosteric promoter of RNR enzyme activity; dATP is an allosteric downregulator (Nordlund and Reichard, 2006). The dATP-driven downregulation of RNR activity is functionally coupled to RNRa-specific hexamerization (Ahmad and Dealwis, 2013; Aye et al., 2015; Cooperman and Kashlan, 2003; Hofer et al., 2012). Several US Food and Drug Administration (FDA)-approved deoxyadenosine (dA)-mimetic nucleoside antimetabolites, e.g., clofarabine (ClF) (Figure 1), also cause RNR-a hexamerization (Aye and Stubbe, 2011; Wisitpitthaya et al., 2016). RNR-a oligomers (endogenous and ectopic) have been isolated from HEK293T, HeLa, and COS-7 cells after ClF exposure (Aye et al., 2012a). Electron microscopy showed hexameric species similar to those formed with recombinant RNR-a in vitro (Aye et al., 2012a, Brignole et al., 2018). Not unexpectedly, nucleoside metabolites, such as ClF, which is FDA approved to treat leukemias but is not clinically used against solid tumors, exhibit numerous bioactivities, including misincorporation, DNA damage, apoptotic signaling, and possibly affecting unfolded protein response (Bonate et al., 2006; Christopherson et al., 2014; Ewald et al., 2008; Galmarini et al., 2001; Jordheim et al., 2013). However, expression of RNR-a(D57N), a hexamerization-defective but reductase-active mutant resistant to ClF, abrogates ClFinduced cytotoxicity (Wisitpitthaya et al., 2016). Conversely, gemcitabine (F2C) (a nucleoside antimetabolite approved to treat pancreatic cancer, which functions as a mechanism-based inactivator of RNR-a [Stubbe and van Der Donk, 1998], but does not cause RNR-a-specific hexamerization) (Fu et al., 2014; Fu et al., 2013; Wisitpitthaya et al., 2016) and triapine (3-AP, which targets RNR-b [Aye et al., 2012b]) (Figure S1A) are both only weakly attenuated by RNR-a(D57N) expression (Wisitpitthaya et al., 2016). Thus, RNR-a is a key early target in the pharmaceutical profile of RNR-inhibiting nucleoside antimetabolite drugs, and in the case of ClF, RNR-a hexamer induction is required for bioactivity. RNR-a is linked with numerous unexpected activities, including tumor suppression, for which there is little link to RNR’s canonical reductase activity (Aye et al., 2015). These data hint at

Cell Chemical Biology 27, 1–12, February 20, 2020 ª 2019 Elsevier Ltd. 1

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

A

B

C

Figure 1. Early DNA Synthesis Suppression Driven by Anti-leukemic Nucleotides ClF/CLA/FLUMP in HeLa Cells Is Dependent on ZRANB3, Requiring ZRANB3-RNR-a Interaction in the Nucleus (A) This study for the first time interrogates the potential interplay between clofarabine (ClF), cladribine (CLA), and fludarabine monophosphate (FLUMP), and the recently discovered ZRANB3-RNR-a signaling axis (also see Figure S1B). (B) Representative images of PLA showing HeLa cells treated with DMSO, CLA (50 mM), and ClF (5 mM). Scale bars, 10 mm. (C) Quantitation of cells with one or more punctum in the nucleus (for two or more puncta per nucleus, see also Figure S1E). Concentrations (mM) used are indicated in parentheses. Data are presented as mean ± SEM (DMSO, n = 34; CIA, n = 29; ClF, n = 23; FLUMP, n = 12; 3-AP, n = 10). Relates to Figures 2, 3, and 4, S1–S5, and S6A–S6E.

non-canonical roles for this protein. Interestingly, knockdown of RNR-a promotes tumor growth and transformation (Fan et al., 1997), indicating that non-canonical activities of RNR-a may be, in some instances, dominant over the need to supply nucleotides. However, the necessity of RNR-a for dNTP provision means that correlating RNR-a levels and expression in tumors is overall a complex process. Nonetheless, there is some evidence that RNR-a expression is positively correlated with remission and survival (Aye et al., 2015), for which there remains little explanation. Surprisingly, little basic science work has set about understanding these phenomena. Change in protein quaternary state is an established means by which many enzymes acquire moonlighting functions (Jeffery, 1999). Thus, we hypothesized that, upon hexamerization, RNR-a gains ancillary functions independent of its role in ribonucleotide reduction. We recently discovered that hexameric RNR-a translocates into the nucleus within
30 min following cell stimulation by hexamer inducers, e.g., dA and ClF (Figure S1B). Nuclear RNR-a—regardless of oligomerization state—binds ZRANB3 (Fu et al., 2018), a poorly characterized protein that was primarily associated with the DNA damage response (DDR) (Neelsen and Lopes, 2015; Poole and Cortez, 2017). In the canonical model of ZRANB3 biology, following DNA damage, ZRANB3 associates with ubiquitinated proliferating cell nuclear antigen (PCNA) to coordinate template switching to reinitiate stalled replication forks and bypass DNA lesions (Ciccia et al., 2012; Weston et al., 2012; Yuan et al., 2012); however, we found no indication of DDR upregulation in our studies (Fu et al., 2018). We further found that, in cells subjected to DNA damage, the recently identified nuclear RNRa-ZRANB3 interaction is ablated. This result implies that nuclear RNR-a regulates an overlooked ZRANB3 function occurring 2 Cell Chemical Biology 27, 1–12, February 20, 2020

exclusively in non-damaged cells. We subsequently showed that ZRANB3 promotes DNA synthesis and passage into Sphase. Our findings agree with initial reports that ZRANB3 and PCNA interact in the absence of DNA damage, although the function of this interaction was unknown (Ciccia et al., 2012). Specifically, we found that knockdown of ZRANB3 elicits 30%–40% decrease in DNA synthesis rates in S-phase cells (Fu et al., 2018). The basal function of ZRANB3 requires PCNA association, and nuclear RNR-a binds ZRANB3 competitively with PCNA. Indeed, the ZRANB3 nuclear RNR-a interaction occurs through the APIM domain of ZRANB3, one of the two functionally coupled PCNA-binding sites within ZRANB3. However, many key questions underlying the recently discovered RNR-a interactome remain unanswered, both in terms of the RNR-a-ZRANB3 signaling axis and the role of ZRANB3 in promoting DNA synthesis in non-damaged cells. For instance, the nuclear content of RNR-a is increased by ClF. Nuclear RNR-a inhibits the DNA synthesis-promoting function of ZRANB3. However, ZRANB3 is required for rapid proliferation, which is a hallmark of cancer cells. Thus, one major therapeutically relevant question constitutes the extent to which ZRANB3 plays a role in the clinical action of ClF (Figure 1A). Our previous studies (Fu et al., 2018) explored this link using dA. However, dA is highly pleiotropic and acutely toxic, and must be administered at high concentrations (mM). Such concerns raise questions concerning the role of adverse/secondary effects following dA stimulation that are not easily parsed from on-target effects. Here, we exploit the principle of genetic epistasis (Phillips, 2008), classic knockdown/knockout/ rescue, and drug treatment to directly establish the involvement of ZRANB3 in ClF bioactivity in several independent lines. These questions are particularly interesting as ClF, at low

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

concentrations (nanomolar to micromolar), induces rapid and persistent RNR-a hexamerization (Aye et al., 2012a; Aye and Stubbe, 2011) and consequent nuclear translocation (Fu et al., 2018) in various cell types. As target occupancy by drugs in vivo is typically moderate especially when clearance is taken into account (Lai and Crews, 2016), this mechanism could be important in the pharmaceutical programs choreographed by ClF and other RNR-a hexamer-inducing anti-leukemic/anti-lymphoma drugs (Wisitpitthaya et al., 2016), e.g., cladribine (CLA) and fludarabine monophosphate (FLUMP) (Figure 1A). Such evidence would add weight to the growing data showing the importance of RNR-a translocation in ZRANB3 regulation (Figure S1B). By extension, such evidence would also further highlight the importance of the unappreciated role of ZRANB3 modulating basal DNA synthesis rates (i.e., in cells not treated with DNA-damaging agents). We were equally curious about how the unexplained tumor suppression roles of RNR-a (Aye et al., 2015) could be related to the ZRANB3-dependent signaling axis for DNA synthesis promotion in non-stressed cells (Figure S1B). In this regard, it is paramount to understand the roles of nuclear RNR-a and ZRANB3 in tumor suppression and how these roles are linked. Here, we investigate how these nucleoside antimetabolites intersect with the ZRANB3 nuclear RNR-a signaling axis (Figure 1A), using ZRANB3 knockdown and knockout cells and nuclear RNR-a overexpression. Our data document that ZRANB3 function is indirectly modulated by ClF and its variants in HEK293T, U2OS, and NIH3T3 cells, via RNR-a hexamerization-dependent RNR-a nuclear signaling. Our further experiments suggest that ZRANB3 is required for H-rasG12Vinduced focus formation, which we demonstrate is suppressed by ClF. RESULTS

ZRANB3-specific small interfering RNAs (siRNAs), relative to a non-targeted siControl (Figure S1D) (note: this result further validates the ZRANB3-siRNAs used in this work). WB analysis showed similar knockdown efficiency. We conclude that the nuclear signal from this ATLAS antibody is due to a specific interaction with ZRANB3. Notably, this antibody also shows cytosolic staining that is not decreased following siZRANB3 treatment (Figure S1D). This non-specific cytosolic staining does not bias our analysis of the ZRANB3 nuclear-specific protein. We proceeded to the PLA using the ATLAS anti-ZRANB3/ mouse anti-RNR-a antibody combination. DMSO treatment gave weak background staining of the nucleus with some elevated signal in the nucleolar region in HeLa cells (Figure 1B). A few cells showed nuclear puncta, likely reflecting that ZRANB3 and RNR-a interact without stimulation. Treatment of cells with nuclear translocation inducers (ClF, CLA, and FLUMP [Fu et al., 2018]) increased the number of nuclear puncta observed and increased the number of cells with more than one nuclear punctum (Figure S1E). 3-AP–an RNR inhibitor that does not elicit RNR-a hexamerization/translocation (Aye et al., 2012a) (Figure S1A)—did not increase nuclear puncta (Figures 1C and S1E). Unsurprisingly, as the antiZRANB3 antibody shows some cytosolic staining (Figure S1D) and the hexamer inducers ClF/CLA/FLUMP alter the cytosolic interactome of RNR-a (namely, IRBIT and importin-a1) (Figure S1B), as well as increasing the nuclear proportion of RNR-a, we also identified (likely RNR-a-dependent/ZRANB3independent) an increase in puncta in the cytosol. Because nuclear-ZRANB3 is the predominant source of the signal derived from the ZRANB3 antibody (Figure S1D), the elevation of nuclear puncta that selectively occurs upon treatment only with multiple RNR-a nuclear translocation inducers (and not with 3-AP) is strongly consistent with our previous data.

Proximity Ligation Assay Confirms the Association of Endogenous ZRANB3 and Endogenous RNR-a in Cells We have shown the ZRANB3-RNR-a interaction: directly, through immunoprecipitation (IP); indirectly, through the loss of ectopic ZRANB3-PCNA interaction by both IP and puncta formation using immunofluorescence (IF) (Fu et al., 2018). We thus first validated endogenous ZRANB3-RNR-a association in an in-cell assay. We used the proximity ligation assay (PLA) (Soderberg et al., 2006). Although PLA is not a direct measurement of association, it is locale specific and it can analyze endogenous proteins, with no need for epitope tagging or overexpression in cells. PLA requires two antibodies of different species that (1) detect endogenous proteins and (2) are IF compatible. We previously disclosed RNR-a antibodies from both mouse and rabbit that function well in IF and western blot (WB) (Aye et al., 2012a, 2012b; Fu et al., 2018). However, we have had difficulty validating a ZRANB3 antibody that functions in IF (Fu et al., 2018): our previous antibody that showed, among several non-specific bands, one specific ZRANB3 band by WB (Bethyl laboratories), gave almost exclusively cytosolic staining by IF (Figure S1C). We here identified a rabbit anti-ZRANB3 antibody (ATLAS HPA035234) that functioned sufficiently well for IF: the nuclear signal observed from this antibody was reduced around 70% by three different

ZRANB3 Is One of the Earliest Inhibition Events following ClF Treatment We next investigated how ClF affects DNA synthesis in siRNA knockdown control versus siZRANB3 knockdown cells, using three different ZRANB3-targeting siRNAs (validated above by IF [Figure S1D], and previously by WB [Fu et al., 2018]) in a dual-pulse whole-nuclei staining assay (DPA). DPA was chosen because it gives a very sensitive measure of DNA synthesis, over a specific time period (allowing discounting of cells entering DNA synthesis during the assay, and giving more confidence that each signal measured is specific). In cells not treated with ClF, all three ZRANB3 siRNAs suppressed DNA synthesis rates 40% relative to control siRNA (Figure 2A), consistent with our previous report (Fu et al., 2018). ClF treatment alone led to an almost 50% suppression in DNA synthesis. However, when siZRANB3 cells were treated with ClF, DNA synthesis was not affected relative to siZRANB3 cells treated with DMSO in place of ClF (Figure 2A). Such reduced (or hypomorphic) response to ClF in siZRANB3 cells indicates that ZRANB3 is also inhibited (directly or indirectly) by ClF and that this inhibition significantly contributes to the response seen in control cells. This finding is similar to our previous report on cells stimulated with dA (Fu et al., 2018). As DPA specifically measures DNA synthesis rates in S-phase, it is not affected by changes in cell-cycle Cell Chemical Biology 27, 1–12, February 20, 2020 3

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

A

B

Figure 2. Dual-Pulse DNA-Labeling Assays Show that ZRANB3 Mediates ClF/CLA/ FLUMP-Induced DNA Synthesis Suppression in Early Stage in HEK293T Cells; PLA in U2OS, ZRANB3 KO-35 U2OS, and ZRANB3 KO-35 U2OS Cell Lines Re-expressing ZRANB3wt Validates that ClF Promotes ZRANB3-RNR-a Interaction in the Nucleus

(A) HEK293T cells were treated with the indicated siRNA for 2 days then they were either treated with DMSO or ClF (5 mM) for 30 min. After this time, dualpulse labeling was carried out with a pulsing time of 30 min for each dT analog (5-ethynyl-20 -deoxyuridine [EdU], to clearly label S-phase cells, followed immediately by bromodeoxyuridine (BrdU), which was used to quantify DNA synthesis rate during the C D second pulse). After this, cells were washed with PBS, then fixed with methanol. Data were analyzed by confocal imaging and the density of the BrdU signal in EdU-positive cells was measured. Data are presented as mean ± SEM (for the cohort of DMSOtreated cells, siCont, n = 501; siZRANB3(1), n = 462; siZRANB3(2), n = 224; siZRANB3(3), n = 282; for the cohort of ClF-treated cells, siCont, n = 225; siZRANB3(1), n = 185; siZRANB3(2), n = 172; siZRANB3(3), n = 174). (B) Similar to (A), but cells were treated with the indicated compound (F2C and ClF 5 mM, CLA 50 mM, and FLUMP 200 mM) for 30 min in each case. Note: (A and B) are derived from separate batches of cells from different passage numbers treated on different days. Some variation is observed between runs. Also see Figure S1F, and S2A–S2B. Data are presented as mean ± SEM (for the cohort of DMSOtreated cells, siCont, n = 753; siZRANB3, n = 803; for the cohort of ClF-treated cells, siCont, n = 331; siZRANB3, n = 530; for the cohort of CLA-treated cells, siCont, n = 466; siZRANB3, n = 745; for the cohort of FLUMP-treated cells, siCont, n = 428; siZRANB3, n = 296; for the cohort of F2C-treated cells, siCont, n = 513; siZRANB3, n = 556). (C) U2OS, ZRANB3 KO-35 U2OS, and ZRANB3 KO-35 U2OS re-expressing ZRANB3wt were treated with either DMSO or 3 mM ClF for 30 min. Cells were then fixed with 20 C methanol and PLA was performed following the manufacturer’s protocol (Sigma Aldrich) using the primary antibodies: anti-ZRANB3 (ATLAS, 1:25) and anti-RNR-a (Proteintech, 1:300). Note: representative images are shown for each cell line either treated with DMSO (top) or ClF (bottom). Scale bars, 10 mm. (D) Quantification of data from (C) reporting cells with one or more nuclear punctum. Data are presented as mean ± SEM (for the cohort of native U2OS cells, DMSO, n = 19; 3 mM ClF, n = 21; for the cohort of KO-35 U2OS cells, DMSO, n = 19, 3 mM ClF, n = 18; for the cohort of KO-35 U2OS re-expressing ZRANB3wt cells, DMSO, n = 20; 3 mM ClF, n = 21). (E) Quantification of data from (C) reporting cells with two or more nuclear puncta. Data are presented as mean ± SEM (for the cohort of native U2OS cells, DMSO, n = 19; 3 mM ClF, n = 21; for the cohort of KO-35 U2OS cells, DMSO, n = 19, 3 mM ClF, n = 18; for the cohort of KO-35 U2OS re-expressing ZRANB3wt cells, DMSO, n = 20; 3 mM ClF, n = 21). In all figures, Student’s two-tailed t test was performed to determine significance. ns, non-significant (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Relates to Figures 1, 3, 4, S1–S5, and S6A–S6E.

distribution or growth rates that we have reported in these ZRANB3 knockdown lines (Fu et al., 2018), although changes in cell cycle/growth are relatively minor compared with the observed, almost completely penetrative loss of response to ClF treatment in ZRANB3-deficient cells. Finally, these data are also compelling as we employed siRNA and drug treatments, which give little scope for genetic compensation/selection that may occur during prolonged knockdown or knockout. 4 Cell Chemical Biology 27, 1–12, February 20, 2020

ZRANB3 Inhibition Is an Early Event Common to Hexamer-Inducing RNR-a Allosteric Modulators To further investigate the role of RNR-a translocation in this ZRANB3-mediated DNA synthesis suppression process, we replicated these assays using different RNR-inhibiting nucleosides. Beyond ClF, we examined additional RNR-a targeting reversible allosteric modulators (Wisitpitthaya et al., 2016) that cause RNR-a-specific nuclear translocation (Fu et al., 2018)

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

(namely, CLA and FLUMP) (Figure 1A). Results from these compounds were compared with those from the irreversible mechanism-based inactivator, gemcitabine (F2C) (Figure S1A, a compound that does not induce either RNR-a-specific hexamerization or nuclear translocation, and that is also a frontline chemotherapeutic [Aye et al., 2015]). siZRANB3-treated cells showed resistance to all nucleoside drugs relative to siControl cells (Figure 2B). However, resistance was most apparent when the drug treatment promoted RNR-a translocation, regardless of the magnitude of DNA synthesis suppression. Similar effects were observed upon treatment with natural hexamer inducer dA (Figures S2A and S2B), with and without the co-treatment with non-hexamer-inducing dG/dC (Figure S2B). These results demonstrate that the observed rescue from ClF/CLA/FLUMP-induced replication-inhibition in siZRANB3-treated cells (Figures 2A and 2B) is an endogenous function, and not an artifact of drug treatment. We cannot rule out that some other factors contribute weakly to the observed hypomorphic responses to reductase activity downregulating drugs. However, these data suggest that RNR-a allosteric modulators—which elicit RNR-a-specific nuclear translocation (Figure S1B)—selectively target ZRANB3 as part of their pharmaceutical spectrum. The mechanism-based irreversible inactivator F2C (Figure S1A)—which does not elicit RNR-a-specific hexamerization (Wisitpitthaya et al., 2016)/translocation (Fu et al., 2018)—does not target ZRANB3. A similar effect was observed in NIH3T3 cells expressing integrated shRNA control or shZRANB3 (Figure S1F), thereby confirming that the effect is not cell-line specific and is not unique to transformed cells. To investigate this phenomenon further, we investigated the effect of ZRANB3 knockout (KO) and rescue in U2OS cells, a human osteosarcoma line that has been used extensively for DNA damage research and specifically in studies involving ZRANB3 nuclear protein (Ciccia et al., 2012; Vujanovic et al., 2017). We first validated that RNR-a translocation occurred upon ClF treatment in U2OS cells (3 mM, 30 min) (Figure S2C). We also validated the ZRANB3 antibody (ATLAS HPA035234) for use in IF in this line, by comparing ZRANB3 signal in U2OS cells, ZRANB3 KO-35 U2OS (nuclear signal depleted), and ZRANB3 KO-35 U2OS re-expressing ZRANB3wt (nuclear signal restored, with a small percentage of cells hyperexpressing ZRANB3, as is common in such lines) (Figures S2D and S2E). Similar outcomes in terms of ZRANB3 levels were also observed by WB (Figure S2F). With these key validations in hand, PLA was conducted in U2OS and a ClF-dependent increase in nuclear puncta was observed (Figures 2C–2E). By contrast, in ZRANB3 KO-35 U2OS, a lower basal number of nuclear puncta was observed relative to U2OS; critically, there was no increase in nuclear puncta following ClF treatment. Both basal and ClF-dependent nuclear puncta formation was re-established in ZRANB3 KO-35 USO2 re-expressing ZRANB3. Subsequent execution of DPA in two different ZRANB3 KO U2OS lines as expected showed suppressed DNA synthesis in S-phase in non-stressed cells (Figures 3A and S3A–S3C [see DMSO-treated samples], and S4A–S4B). Re-expression of ZRANB3wt in one of these KO lines rescued this basal DNA synthesis rate (Figures 3A and S3B, images in top row, left, and middle panels). As predicted from PLA above, ZRANB3 KO-35 and KO-38 lines were also resistant to ClF treatment for

30 min relative to U2OS (Figure 3B, and cf. Figures S3A versus S3B), and ZRANB3 KO-35 U2OS re-expressing ZRANB3wt (Figures 3B and S3B, left versus middle panels). ZRANB3- or RNR-a expression was not affected by any of the different drug treatments in this line (Figures S4A–S4C), consistent with our previous data in other cell lines (Wisitpitthaya et al., 2016). These data further show the generality of ZRANB3’s sensitivity to ClF, and underscore that, although ZRANB3 is not absolutely required for DNA synthesis (since knockout cells can still synthesize DNA), this protein is an important adjuvant of DNA synthesis. Finally, although as noted above, knockout lines can show complications in terms of compensation effects, the ZRANB3 KO-35 U2OS cells re-expressing ZRANB3wt were able to recapitulate effects observed in U2OS, indicating that a ZRANB3-specific effect is sufficient to explain these data. We progressed to investigate how ZRANB3 knockdown affected late-stage susceptibility to the different drugs. For these assays, we used NIH3T3 and HEK293T cells. The extent of cell proliferation (assessed by PI50 values, Table S1) for ClF, CLA, F2C, or 3-AP differed by no more than 1.7-fold between the control and knockdown cells, in both NIH3T3 and HEK293T cells (Figure S5). Thus, ZRANB3-specific effects occur in the early time point and are not manifested at later stages following drug treatment, where a series of complex cytotoxic events involving DNA damage is known to ensue (Bonate et al., 2006). Critically, the ZRANB3-RNR-a interaction does not occur upon DNA damage (Fu et al., 2018). Thus, the late-stage toxicity could readily override the effects of RNR-a-specific nuclear translocation that occurs rapidly and reaches saturation within 30 min at 5 mM ClF (Fu et al., 2018). Prolonged treatment of cells with the drugs led to a non-detectable level of DNA synthesis and, thus, we are measuring a true reduction in DNA synthesis rate, as opposed to a persistent DNA synthesis in the knockdown line (Figure S6A). RNR-a Is Required for ClF-Induced ZRANB3 Inhibition We next examined responses to DNA synthesis upon combinations of siZRANB3, siRNR-a, and ClF treatment in HEK293T cells. siZRANB3 suppressed DNA synthesis. siRNR-a—which reduced both nuclear and cytosolic RNR-a levels (validated through IF, Figures S6B and S6C)—also suppressed DNA synthesis. However, siRNR-a lines did not show changes in DNA synthesis upon ZRANB3 knockdown (Figure 4A and S6D– S6E). This result suggests that RNR-a plays a role in the DNA synthesis suppression that accompanies ZRANB3 knockdown. One simple explanation for such an observation is that loss of RNR-a increases basal ZRANB3 activity (i.e., DNA synthesis promotion), rendering cells more able to sustain a partial loss of ZRANB3 following ZRANB3 knockdown (Figure S6). We also investigated how ZRANB3 knockdown and RNR-a knockdown affected response to ClF (Figure 4A). All lines, except the ZRANB3 knockdown line, showed downregulation in DNA synthesis following ClF treatment. The siRNR-a line showed moderate sensitization to ClF (and dA, Figures S2A and S2B) treatment. This observation may be expected given that these nucleosides also downregulate RNR reductase activity (which is intrinsically linked to DNA synthesis rate). However, DNA synthesis in the siZRANB3- and siRNR-a-treated cells was significantly reduced, relative to the ClF-treated Cell Chemical Biology 27, 1–12, February 20, 2020 5

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

A

B

Figure 3. Knockout and Rescue Experiments Demonstrate that ZRANB3 Is a Mediator of Early-Stage DNA Synthesis Suppression Induced by ClF in U2OS Cells

(A) U2OS, ZRANB3 KO-35 U2OS, ZRANB3 KO38 U2OS, and ZRANB3 KO-35 U2OS cells expressing ZRANB3wt were treated with DMSO for 30 min. Dual-pulse labeling was then carried out with a pulsing time of 30 min for each dT analog (EdU, to clearly label S-phase cells, followed immediately by BrdU, which was used to quantify DNA synthesis during the second pulse). Cells were then washed with PBS, fixed with methanol followed by DPA. Data were analyzed by confocal imaging and the density of the BrdU signal in EdUpositive cells was measured. Also, see Figures S3A–S3C, S4A, and S4B. Data are presented as mean ± SEM (native U2OS cells, n = 976; KO35 U2OS cells, n = 908; KO-38 U2OS cells, n = 160; KO-35 cells expressing ZRANB3wt, n = 579). (B) Similar to (A), but the indicated U2OS lines were treated with DMSO or different concentrations of ClF for 30 min. Note: the DMSO data in (B) are recapitulated from (A). Also, see Figures S3A–S3C, S4A, and S4B. Data are presented as mean ± SEM (for the cohort of native U2OS cells, DMSO, n = 976; 1 mM ClF, n = 309; 2.5 mM ClF, n = 224; 5 mM ClF, n = 54; for the cohort of KO-35 U2OS cells, DMSO, n = 908; 1 mM ClF, n = 129; 2.5 mM ClF, n = 296; 5 mM ClF, n = 137; for the cohort of KO-38 U2OS cells, DMSO, n = 160; 1 mM ClF, n = 43; 5 mM ClF, n = 110; for the cohort of KO-35 cells expressing ZRANB3wt, DMSO, n = 579; 1 mM ClF, n = 153; 2.5 mM ClF, n = 319; 5 mM ClF, n = 116). In all figures, Student’s two-tailed t test was performed to determine significance. ns, non-significant (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Relates to Figures 2, 4, and S2–S4.

siRNR-a-only lines. This rescue of sensitization to siZRANB3 in the presence of ClF could readily be explained as follows: knocking down RNR-a in the ZRANB3 knockdown line allows basal DNA synthesis to become ZRANB3 dependent (by reducing the extent of ZRANB3 inhibition by nuclear RNR-a, since the amount of RNR-a in the nucleus is reduced [Figures 4B and S6C]). This observation is also consistent with the fact that ZRANB3 knockdown does not promote DNA synthesis suppression in the RNR-a knockdown background (Figure 4A, DMSO set). Thus, upon ClF treatment, translocated RNR-a meets a relatively low concentration of ZRANB3, which is contributing to DNA synthesis. The translocated RNR-a is sufficient to affect DNA synthesis in this double-knockdown background (Figure 4A, ClF-treated set, and Figure 4B). RNR-a Nuclear Accumulation Suppresses ClF-Induced DNA Synthesis Downregulation Regardless of RNR Reductase Activity To examine whether nuclear-specific RNR-a plays a role in ClFinduced DNA synthesis inhibition, we used HEK293T T-REx monoclonal lines that, upon tetracycline treatment, express RNR-a bearing a nuclear localization sequence tag (NLS). We have shown that RNR-a-NLS (irrespective of wild type [WT]/ mutant) efficiently localizes to the nucleus, and inhibits ZRANB3’s DNA synthesis-promoting activity (Fu et al., 2018). Expression of RNR-a-NLS regardless of reductase activity (WT, catalytically active but hexamerization and nuclear translocation defective and ClF/dA inhibition-resistant mutant, RNRa(D57N), or reductase dead but hexamerization-capable mutant, RNR-a(C429S)) diminishes DNA synthesis (Figure 4C). However, these RNR-a-NLS-expressing cells were resistant to ClFinduced DNA synthesis inhibition, similarly to siZRANB3 lines (Figures 4C and S2B). Critically, this ClF treatment-dependent 6 Cell Chemical Biology 27, 1–12, February 20, 2020

suppression of DNA synthesis occurred regardless of RNR reductase activity, and was not significantly more in lines expressing RNR-a(D57N)-NLS that fully protects against ClF toxicity (Wisitpitthaya et al., 2016) when the protein is cytosolic (Figure 4C). Thus, nuclear RNR-a suppresses the response to ClF regardless of RNR reductase activity, ClF sensitivity, or RNR-a hexamerization capability. Furthermore, we have also shown that (1) overexpression of RNR-a-only weakly protects against late-stage toxicity of ClF, even though the drug-resistant point mutant (i.e., the reductase active but hexamerization-/translocation-defective RNRa(D57N)) is highly resistant (Wisitpitthaya et al., 2016) and (2) nuclear RNR-a-overexpressing cells are unable to inhibit DNA synthesis in ZRANB3-deficient cells (Fu et al., 2018), indicating that the nuclear role of RNR-a involves eliciting ZRANB3 loss of function. Taken together, we interpret these data to mean that ZRANB3 inhibition is an early event accompanying ClF treatment. ZRANB3 Knockdown Reduces Wound Healing in NIH3T3 Cells We proposed that ZRANB3 inhibition could be a mechanism through which RNR-a overexpression elicits oncogene-induced tumor suppression, an unexplained observation made over the past two decades. ClF has a panoply of ascribed bioactivities, although, at least in HEK293T T-REx cells, RNR-a inhibition is required for the majority of these responses. We thus asked the question: What phenotypic links are there between ClF/ CLA/FLUMP treatment and RNR-a overexpression (a process that we have shown elevates nuclear RNR-a levels; Fu et al., 2018)? Answering this question led us to find two similar functional intersections between the pharmacological action of

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

A

B

C

Figure 4. Knockdown and Overexpression Experiments in both HEK293T and Tetracycline-Inducible HEK293T T-REx Cells Show that ClFPromoted Early DNA Synthesis Suppression Is Dependent on RNR-a, Independent of Its Reductase Activity, and Mediated by ZRANB3 Nuclear Protein (A) HEK293T cells were treated with the indicated siRNA(s) for 2 days (note: for single siRNA treatments, an equal amount of siCont was added; for siCont alone treatment, 23 siCont was used), and subsequently, after a 30-min treatment with ClF (5 mM) or DMSO, cells were subjected to dual-pulse labeling after which BrdU incorporation in EdU-positive cells was measured. Data are presented as mean ± SEM (for the cohort of DMSO-treated cells, siCont, n = 655; siZRANB3, n = 509; siRNR-a, n = 589; siZRANB3+ siRNR-a, n = 749; for the cohort of ClF-treated cells, siCont, n = 680; siZRANB3, n = 660; siRNR-a, n = 372; siZRANB3+ siRNR-a, n = 325). (B) Model to understand the effects of different knockdowns on DNA synthesis (the same color code for RNR-a [green ellipse] and ZRANB3 [yellow rectangle] is used as in Figure S1B; see also Figure 1A). Top left panel: normal. In the cytosol, RNR-a (together with RNR-b, not shown) provides dNTPs that are required for DNA synthesis. Top right panel: knockdown of RNR-a thus decreases DNA synthesis by limiting dNTPs. RNR-a is also decreased in the nucleus, and hence active ZRANB3 is increased (or remains at a similar level, if ZRANB3 is in excess under normal conditions [top left]; however, our data are consistent with ZRANB3 being limiting, thus increase in active ZRANB3 is expected). Bottom left panel: in the nucleus, ZRANB3 is required for DNA synthesis. Knockdown of ZRANB3 decreases DNA synthesis (a recent report from another laboratory [Puccetti et al., 2019] shows that even loss of one ZRANB3 allele is phenotypic in some cases). Bottom right panel: when ZRANB3 and RNR-a are knocked down, there is a decrease in DNA synthesis relative to basal (top left), but relative to wild-type cells, there is no RNR-a to inhibit ZRANB3, so ZRANB3 knockdown has a minimal effect. (C) HEK293T T-REx cells with single integrated copies of RNR-a (C429S)-NLS (left), RNR-a-NLS (middle), or RNR-a(D57N)-NLS (right) were exposed to the specified conditions (tetracycline exposure was for 2 days; ClF concentration was 5 mM; siRNA exposure was 2 days); the EdU signal in EdU-positive cells was recorded (performed this way because these cells have lower basal DNA synthesis and are more susceptible to ClF than parent HEK293T cells). Data are presented as mean ± SEM in each graph. RNR-a (C429S)-NLS (left), for the cohort of DMSO-treated cells, Tetracycline, n = 1,067; +Tetracycline, n = 669; for the cohort of ClF-treated cells, Tetracycline, n = 746; +Tetracycline, n = 830; RNR-a-NLS (middle), for the cohort of DMSO-treated cells, Tetracycline, (legend continued on next page)

Cell Chemical Biology 27, 1–12, February 20, 2020 7

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

Figure 5. Wound-Healing Assays Show that ZRANB3 Knockdown Mirrors the Phenotype Induced by RNR-a Overexpression (Which Increases Nuclear RNR-a Levels) NIH3T3 cells expressing the indicated shRNA were grown on separate glass-backed plates each until a lawn of cells covered the whole plate surface. An autoclaved glass pipette was then used to draw two scratches perpendicular to each other on each plate. The width of each scratch was measured as a function of time. The widths of each starting scratch were similar; quantitation (left) and representative images (right). Scale bar, 300 mm. Data are presented as mean ± SEM (n = 4 biological replicates for all data points in the graph). Relates to Figures 6, S6F, and S6G.

RNR-inhibiting nucleos(t)ide drugs and the genetic consequences of overexpressing RNR-a, namely, lower invasiveness and tumor suppression. It is worth noting that shRNA knockdown of ZRANB3 in most cell types, including NIH3T3, proved challenging. In NIH3T3, shZRANB3 was predominantly toxic. As alluded to above, we were able to establish one shZRANB3 line with reduced ZRANB3 protein levels. In addition to the shControl line, we made use of shIRBIT cells. We previously showed that IRBIT knockdown elevates nuclear RNR-a and displays phenotypes consistent with elevated nuclear RNR-a (Fu et al., 2018) (Figure S1B). In wound-healing assays, we observed that both shZRANB3 and shIRBIT lines exhibited reduced wound closure compared with shControl lines (Figure 5). This result is consistent with previous assays on RNR-a-overexpressing lines that showed reduced invasion in a Boyden chamber assay (Cao et al., 2003; Fan et al., 1997; Gautam et al., 2003; Zhou et al., 1998). These data demonstrate that ZRANB3 knockdown and IRBIT knockdown show wound closure/invasion suppression phenotypes similar to the known behavior of RNR-a overexpression. ZRANB3 Knockdown Reduces Focus Formation in NIH3T3 Cells We next investigated the role of ZRANB3 nuclear RNR-a in rasinduced transformation. Indeed, ClF/CLA/FLUMP have been previously shown to suppress anchorage-independent growth (Celik et al., 2018). We first investigated how ClF affected focus formation in NIH3T3 lines. Significant suppression of focus formation was observed even at the PI10 (concentration at 10% proliferation inhibition) of the non-transformed lines, indicating that ClF is likely more effective against transformed cells than against non-transformed cells, similar to how ZRANB3 behaves (Figure 6A). We then validated the previous report that RNR-a overexpression suppresses H-rasG12V-induced focus formation in NIH3T3 fibroblasts (Figure S6F). Nuclear RNR-a, regardless of catalytic activity, also suppressed focus formation similar to that shown by RNR-a global overexpression (Fig-

ure 6B). Consistent with this observation, shIRBIT lines—which have elevated nuclear RNR-a levels (Fu et al., 2018) (Figure S1B), but have elevated dNTP-producing capacity (Arnaoutov and Dasso, 2014)—also showed limited ability to form foci (Figure 6C, top row). This outcome documents that nuclear RNR-a is likely the principal cause of RNR-a overexpression-promoted tumor suppression and further indicates that inhibition of basal ZRANB3 activity could be the(a) root cause of the observed tumor suppression. We also examined the effect of PCNA overexpression. Overexpression of PCNA rescued the effect of RNR-a-promoted tumor suppression (Figure 6B, lower row), similar to what we observed for the nuclear RNR-a-ZRANB3 interaction-driven DNA synthesis inhibition. These data together predict that ZRANB3 knockdown would suppress focus formation in NIH3T3 cells. We transfected our sh-ZRANB3 line with H-rasG12V and measured the resulting number of foci relative to two different control lines. These data showed that ZRANB3 knockdown strongly reduced focus formation; RNR-a overexpression had little effect on the focus content in these cells (Figure 6C, lower row). DISCUSSION Nucleos(t)ide drugs are a venerable class of antimetabolites that share numerous chemical similarities (Aye et al., 2015; Jordheim et al., 2013). Several of these molecules show no statistically differentiable efficacy in ‘‘Phase IV’’ clinical trials (Muluneh et al., 2015). However, there is certainly evidence that these drugs do behave differently, e.g., the clinical efficacy of CLA and F2C are quite divergent (Chow et al., 2011); and, of course, F2C shows efficacy against solid tumors that is not manifest by ClF, CLA, or FLUMP (Aye et al., 2015; Bonate et al., 2006). Understanding the factors responsible for the variable mechanism(s) of action is of significant interest. Unfortunately, gaining mechanistic insights is difficult because it is widely understood that nucleos(t)ide drugs are pleiotropic.

n = 731; +Tetracycline, n = 487; for the cohort of ClF-treated cells, Tetracycline, n = 431; +Tetracycline, n = 410; RNR-a(D57N)-NLS (right), for the cohort of DMSO-treated cells, tetracycline and siCont, n = 548; tetracycline and siZRANB3, n = 587; +Tetracycline and siCont, n = 396; for the cohort of ClF-treated cells, tetracycline and siCont, n = 499; tetracycline and siZRANB3, n = 483; +Tetracycline and siCont, n = 425. In all figures, Student’s two-tailed t test was performed to determine significance. ns, non-significant (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Relates to Figures 2, 3, S1–S5, and S6A–S6E.

8 Cell Chemical Biology 27, 1–12, February 20, 2020

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

A

B

C

Figure 6. Focus Formation Assays Show that ClF Suppresses H-rasG12V-Driven Tumor Foci, and ZRANB3 Knockdown Phenocopies RNR-a Overexpression-Induced Tumor Suppression Activity (A) NIH3T3 cells were transfected with either H-rasG12V or an empty vector plasmid before being treated with different concentrations of ClF (0, PI10 [30 nM], PI30 [80 nM], and PI60 [500 nM] as in Figure S5 for shControl-1). The cells were cultured with ClF-supplemented medium, maintaining the same ClF concentration, for another 2 weeks before fixing with 2% paraformaldehyde in PBS, and foci were stained by crystal violet followed by water destaining. (B) NIH3T3 cells were either transfected with empty plasmid and transfection agent, or a plasmid expressing H-rasG12V, and the indicated RNR-a construct (the table on the right shows wild type [wt] versus respective functional mutants employed), either empty vector and/or PCNA-GFP. Absence of GFP-PCNA was balanced using an empty vector. After 3 weeks, cells were analyzed as in (A). (C) Similar experiment to (B), except that NIH3T3 cells expressing the stated shRNA were co-transfected with H-rasG12V. After 3 weeks, cells were analyzed as in (A).

From our work, and that of others, it is clear that RNR reductase activity inhibition is a key early step along with the multifactorial mode of action underpinning several nucleoside antimetabolites (Aye et al., 2015). However, following RNR-a inhibition, other processes are critical (Bonate et al., 2006; Ewald et al., 2008;

Galmarini et al., 2001; Jordheim et al., 2013). The importance of each independent or coordinated mechanism to drug efficacy is challenging to assess, and hence there could easily be many unknown factors contributing to drug efficacy and clinical applications that remain to be found. Characterizing additional Cell Chemical Biology 27, 1–12, February 20, 2020 9

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

modes of drug action is thus of both fundamental and translational relevance and could potentially lead to unconventional avenues of therapeutic discovery/hint at the mode(s) of action of utilized, but unannotated, drugs. We were able to score a common mode of action for CLA, ClF, and FLUMP that involves eliciting reductase activity downregulated RNR-a hexamers. Critically, all of these compounds are analogs of the native nucleoside dA, the triphosphate of which (dATP) binds an allosteric site on RNR-a. This mode of action (both site selectivity and RNR-a-specific hexamerization) is not shared by 3-AP and F2C (Wisitpitthaya et al., 2016), which were used as comparisons in this study. Our data show that, at least in the early time points, almost all of the DNA synthesis inhibition for ClF, CLA, and FLUMP is directly associated with inhibition of ZRANB3. Indeed, in cells devoid of ZRANB3, some of these drugs increase DNA synthesis, as opposed to blocking it. Thus, the non-DNA damage-associated activity of ZRANB3 is an early casualty of RNR-a hexamer inducer drugs. Conversely, in the late stage of toxicity, shZRANB3 lines are similarly affected by nucleoside antimetabolites as in control lines. We believe that this is most likely because late-stage cytotoxicity is dominated by DNA damage induction, futile damage/ repair cycles, base-pair misincorporation, breakdown of mitochondria integrity, and changes in cytosine methylation, among others (Eadon et al., 2013; Ewald et al., 2008; Galmarini et al., 2001; Jordheim et al., 2013). Under these conditions, PCNA is ubiquitinated (Ciccia et al., 2012; Weston et al., 2012; Yuan et al., 2012), rendering ZRANB3 less able to interact with RNRa. This may, at first glance, suggest that RNR-a-ZRANB3 interaction is not critical for ClF action. However, we do not know which conditions—early or extreme endpoint—reflects the mode of action of ClF/CLA/FLUMP in patients (but we do know that the early state is essential for drug action). Furthermore, there is some evidence that RNR-a hexamer inducers modulate DDR: ClF exerts larger effects on DDR than 5-fluorouracil (Yamauchi et al., 2001). Indeed, both FLUMP and ClF are inhibitors of the DDR, which could be explained by impairment of ZRANB3. Further investigations into the relationships we uncover could readily be targeted in this direction. Regardless, our data imply that the nuclear RNR-a-ZRANB3 axis has been inadvertently tapped into by these FDA-approved drugs. Our study is thus a starting point to highlight that this action may be pursuable in terms of a hitherto-unrecognized intervention. We postulated that more insights could be gleaned from investigating the link between the ZRANB3-RNR-a axis and focus formation. We here disclose the unexpected finding that ZRANB3 is required for invasion and H-rasG12V-induced focus formation. We thus implicate the RNR-a-mediated ZRANB3 inhibition as the cause of RNR-a-driven reduction of tumor foci, reported seminally by Fan et al. (1997), and later by other research groups (Aye et al., 2015). Importantly, our data do not show that ZRANB3 is an oncogene; it is most likely that H-rasG12Vtransformed cells are more reliant on ZRANB3 to promote/sustain growth. We thus consider this observation as an example of non-oncogene addiction (Luo et al., 2009). Our data, in fact, are strongly consistent with independent work showing that both heterozygous and knockout ZRANB3 mice are more prone to myc-induced tumorigenesis than WT mice (Puccetti et al., 2019). Clearly, examples of non-oncogene addiction are ripe 10 Cell Chemical Biology 27, 1–12, February 20, 2020

for anti-cancer drug discovery, rendering it likely that ZRANB3 is an unheralded cancer drug target. We will thus in the future focus our efforts on developing small-molecule interventions that will mimic nuclear RNR-a and inhibit the DNA synthesis-promoting activities of ZRANB3. We hope that others will similarly take up the baton in this area. SIGNIFICANCE During the progression from a normal cell to a cancerous cell, the destiny of different proteins is reshaped considerably: often tumor suppressors are deleted or inactivated; (proto)oncogenes are activated. The resulting upregulation of stress, DNA damage, and pathway rewiring also alter how proteins that are neither tumor suppressors nor oncogenes behave. Some of these proteins are not essential for growth or even development, but they become essential players that cancer cells cannot do without. Such proteins and their pathways are ideal for drug design, and understanding modes of regulation of these pathways is crucial for personalized medicine and biomarker profiling. Here, we investigate how downregulation of ZRANB3—a protein that is important for promotion of normal cell growth but is not essential—affects cancer-like behavior induced by expression of the oncogene, H-ras(G12V). Surprisingly, although ZRANB3 knockdown/knockout cells survive and can be cultivated indefinitely, H-ras(G12V) expression does not cause uncontrolled growth in ZRANB3-deficient cells, as it does in controls. We have also identified that ribonucleotide reductase subunit a (RNR-a) is a negative regulator of ZRANB3. However, this regulatory pathway requires RNR-a to move from the cytosol into the nucleus, where it can bind ZRANB3. This translocation is complex and requires changes in RNR-a oligomerization, which is most readily caused by binding dATP, or analogs, such as the approved anti-cancer drug clofarabine nucleotide(s). Intriguingly we present evidence that, aside from inhibiting RNR-a activity, ClF nucleotide(s) also function(s) through inhibiting ZRANB3, both at early time points and in anti-cancer assays. These data show that ZRANB3’s role in cancer is an example of non-oncogene addiction that has been inadvertently tapped into by approved drugs. We propose that this druggable pathway is ripe for further investigations and designing hitherto-unappreciated interventions. Such interventions should be highly selective as ZRANB3’s function is dispensable in normal cells. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d d d

KEY RESOURCES TABLE LEAD CONTACT AND MATERIALS AVAILABILITY EXPERIMENTAL MODEL AND SUBJECT DETAILS B Cell Lines METHOD DETAILS B Cell Line Engineering B Cell-Based Assays B Dual Pulse Assay (DPA) B Proximity Ligation Assay (PLA)

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

d d d

B Transformation Assays QUANTIFICATION AND STATISTICAL ANALYSIS B Sample Size n for Figures with Quantification DATA AND CODE AVAILABILITY ADDITIONAL RESOURCES B Chemicals and Reagents B Antibodies B siRNAs B shRNAs B Equipment

SUPPLEMENTAL INFORMATION Supplemental Information can be found online at https://doi.org/10.1016/j. chembiol.2019.11.012. ACKNOWLEDGMENTS Project funding support: Novartis Foundation for Medical-Biological Research (18C200); Shared instrumentation, supplies, and personnel support: NIH Director’s New Innovator award (1DP2GM114850); National Centers of Competence in Research (NCCR) Chemical Biology; Swiss National Science Foundation (SNSF); and Swiss Federal Institute of Technology Lausanne (EPFL). We thank Professors Massimo Lopes (University of Zurich) and David Cortez (Vanderbilt School of Medicine) for the U2OS cell lines used in this study. We thank Dr. Yuan Fu for collaboration in plasmid construction. AUTHOR CONTRIBUTIONS Y.A. and M.J.C.L. designed the experiments. M.J.C.L. performed all experiments except those involving U2OS cells and focus formation assays involving NIH3T3 cells treated with ClF, which were performed by Y.Z. All authors analyzed the data. M.J.C.L. and Y.A wrote the manuscript that was subsequently refined by all authors. All authors wrote the experimental sections. All authors proofed the final manuscript. DECLEARATION OF INTERESTS The authors declare no competing interests. Received: August 3, 2018 Revised: August 30, 2019 Accepted: November 19, 2019 Published: December 10, 2019 REFERENCES Ahmad, M.F., and Dealwis, C.G. (2013). Chapter fourteen––The structural basis for the allosteric regulation of ribonucleotide reductase. In Progress in Molecular Biology and Translational Science, J. Giraldo and F. Ciruela, eds. (Academic Press), pp. 389–410. Arnaoutov, A., and Dasso, M. (2014). IRBIT is a novel regulator of ribonucleotide reductase in higher eukaryotes. Science 345, 1512–1515. Aye, Y., Brignole, E.J., Long, M.J., Chittuluru, J., Drennan, C.L., Asturias, F.J., and Stubbe, J. (2012a). Clofarabine targets the large subunit a of human ribonucleotide reductase in live cells by assembly into persistent hexamers. Chem. Biol. 19, 799–805. Aye, Y., Li, M., Long, M.J., and Weiss, R.S. (2015). Ribonucleotide reductase and cancer: biological mechanisms and targeted therapies. Oncogene 34, 2011–2021. Aye, Y., Long, M.J., and Stubbe, J. (2012b). Mechanistic studies of semicarbazone triapine targeting human ribonucleotide reductase in vitro and in mammalian cells: tyrosyl radical quenching not involving reactive oxygen species. J. Biol. Chem. 287, 35768–35778.

Aye, Y., and Stubbe, J. (2011). Clofarabine 5’-di and -triphosphates inhibit human ribonucleotide reductase by altering the quaternary structure of its large subunit. Proc. Natl. Acad. Sci. U S A 108, 9815–9820. Bonate, P.L., Arthaud, L., Cantrell, W.R., Jr., Stephenson, K., Secrist, J.A., 3rd, and Weitman, S. (2006). Discovery and development of clofarabine: a nucleoside analogue for treating cancer. Nat. Rev. Drug Discov. 5, 855–863. Brignole, E.J., Tsai, K.L., Chittuluru, J., Li, H., Aye, Y., Penczek, P.A., Stubbe, J., Drennan, C.L., and Asturias, F. (2018). 3.3-A resolution cryo-EM structure of human ribonucleotide reductase with substrate and allosteric regulators bound. Elife 7, https://doi.org/10.7554/eLife.31502. Cao, M.Y., Lee, Y., Feng, N.P., Xiong, K., Jin, H., Wang, M., Vassilakos, A., Viau, S., Wright, J.A., and Young, A.H. (2003). Adenovirus-mediated ribonucleotide reductase R1 gene therapy of human colon adenocarcinoma. Clin. Cancer Res. 9, 4553–4561. Celik, H., Sciandra, M., Flashner, B., Gelmez, E., Kayraklioglu, N., Allegakoen, D.V., Petro, J.R., Conn, E.J., Hour, S., Han, J., et al. (2018). Clofarabine inhibits Ewing sarcoma growth through a novel molecular mechanism involving direct binding to CD99. Oncogene 37, 2181–2196. Chabes, A., and Thelander, L. (2003). DNA building blocks at the foundation of better survival. Cell Cycle 2, 171–173. Chow, K.U., Boehrer, S., Bojunga, J., Stieler, M., Rummel, M.J., Fauth, F., Schneider, B., Martin, H., Hoelzer, D., Weidmann, E., et al. (2011). Induction of apoptosis by cladribine (2-CdA), gemcitabine and other chemotherapeutic drugs on CD34 +/CD38 + and CD34 +/CD38-Hematopoietic progenitor cells: selective effects of doxorubicin and 2-CdA with protection of immature cells. Leuk. Lymphoma 43, 377–384. Christopherson, R.I., Mactier, S., Almazi, J.G., Kohnke, P.L., Best, O.G., and Mulligan, S.P. (2014). Mechanisms of action of fludarabine nucleoside against human Raji lymphoma cells. Nucleosides Nucleotides Nucleic Acids 33, 375–383. Ciccia, A., Nimonkar, A.V., Hu, Y., Hajdu, I., Achar, Y.J., Izhar, L., Petit, S.A., Adamson, B., Yoon, J.C., Kowalczykowski, S.C., et al. (2012). Polyubiquitinated PCNA recruits the ZRANB3 translocase to maintain genomic integrity after replication stress. Mol. Cell 47, 396–409. Cooperman, B.S., and Kashlan, O.B. (2003). A comprehensive model for the allosteric regulation of Class Ia ribonucleotide reductases. Adv. Enzyme Regul. 43, 167–182. Eadon, M.T., Wheeler, H.E., Stark, A.L., Zhang, X., Moen, E.L., Delaney, S.M., Im, H.K., Cunningham, P.N., Zhang, W., and Dolan, M.E. (2013). Genetic and epigenetic variants contributing to clofarabine cytotoxicity. Hum. Mol. Genet. 22, 4007–4020. Ewald, B., Sampath, D., and Plunkett, W. (2008). Nucleoside analogs: molecular mechanisms signaling cell death. Oncogene 27, 6522–6537. Fan, H., Huang, A., Villegas, C., and Wright, J.A. (1997). The R1 component of mammalian ribonucleotide reductase has malignancy-suppressing activity as demonstrated by gene transfer experiments. Proc. Natl. Acad. Sci. U S A 94, 13181–13186. Fu, Y., Lin, H.Y., Wisitpitthaya, S., Blessing, W.A., and Aye, Y. (2014). A fluorimetric readout reporting the kinetics of nucleotide-induced human ribonucleotide reductase oligomerization. Chembiochem 15, 2598–2604. Fu, Y., Long, M.J., Rigney, M., Parvez, S., Blessing, W.A., and Aye, Y. (2013). Uncoupling of allosteric and oligomeric regulation in a functional hybrid enzyme constructed from Escherichia coli and human ribonucleotide reductase. Biochemistry 52, 7050–7059. Fu, Y., Long, M.J.C., Wisitpitthaya, S., Inayat, H., Pierpont, T.M., Elsaid, I.M., Bloom, J.C., Ortega, J., Weiss, R.S., and Aye, Y. (2018). Nuclear RNR-a antagonizes cell proliferation by directly inhibiting ZRANB3. Nat. Chem. Biol. 14, 943–954. Galmarini, C.M., Mackey, J.R., and Dumontet, C. (2001). Nucleoside analogues: mechanisms of drug resistance and reversal strategies. Leukemia 15, 875. Gautam, A., Li, Z.R., and Bepler, G. (2003). RRM1-induced metastasis suppression through PTEN-regulated pathways. Oncogene 22, 2135–2142.

Cell Chemical Biology 27, 1–12, February 20, 2020 11

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

Haber, J.E. (2013). Genome Stability: DNA Repair and Recombination (Garland Science).

Direct observation of individual endogenous protein complexes in situ by proximity ligation. Nat. Methods 3, 995–1000.

Hofer, A., Crona, M., Logan, D.T., and Sjoberg, B.M. (2012). DNA building blocks: keeping control of manufacture. Crit. Rev. Biochem. Mol. Biol. 47, 50–63.

Stewart, S.A., Dykxhoorn, D.M., Palliser, D., Mizuno, H., Yu, E.Y., An, D.S., Sabatini, D.M., Chen, I.S., Hahn, W.C., Sharp, P.A., et al. (2003). Lentivirusdelivered stable gene silencing by RNAi in primary cells. RNA 9, 493–501.

Jeffery, C.J. (1999). Moonlighting proteins. Trends. Biochem. Sci. 24, 8–11. Jordheim, L.P., Durantel, D., Zoulim, F., and Dumontet, C. (2013). Advances in the development of nucleoside and nucleotide analogues for cancer and viral diseases. Nat. Rev. Drug Discov. 12, 447. Lai, A.C., and Crews, C.M. (2016). Induced protein degradation: an emerging drug discovery paradigm. Nat. Rev. Drug Discov. 16, 101–114. Luo, J., Solimini, N.L., and Elledge, S.J. (2009). Principles of cancer therapy: oncogene and non-oncogene addiction. Cell 136, 823–837. Mathews, C.K. (2015). Deoxyribonucleotide metabolism, mutagenesis and cancer. Nat. Rev. Cancer 15, 528–539. Muluneh, B., Buhlinger, K., Deal, A.M., Zeidner, J.F., Foster, M.C., Jamieson, K.J., Bates, J., and Van Deventer, H.W. (2015). A comparison of clofarabinebased (GCLAC) and cladribine-based (CLAG) salvage chemotherapy for relapsed/refractory AML. Blood 126, 1342. Neelsen, K.J., and Lopes, M. (2015). Replication fork reversal in eukaryotes: from dead end to dynamic response. Nat. Rev. Mol. Cell. Biol. 16, 207–220. Nordlund, P., and Reichard, P. (2006). Ribonucleotide reductases. Annu. Rev. Biochem. 75, 681–706. Phillips, P.C. (2008). Epistasis––the essential role of gene interactions in the structure and evolution of genetic systems. Nat. Rev. Genet. 9, 855–867. Poole, L.A., and Cortez, D. (2017). Functions of SMARCAL1, ZRANB3, and HLTF in maintaining genome stability. Crit. Rev. Biochem. Mol. Biol. 52, 696–714. Puccetti, M.V., Adams, C.M., Kushinsky, S., and Eischen, C.M. (2019). Smarcal1 and Zranb3 protect replication forks from myc-induced DNA replication stress. Cancer Res. 79, 1612–1623. Soderberg, O., Gullberg, M., Jarvius, M., Ridderstrale, K., Leuchowius, K.J., Jarvius, J., Wester, K., Hydbring, P., Bahram, F., Larsson, L.G., et al. (2006).

12 Cell Chemical Biology 27, 1–12, February 20, 2020

Stubbe, J., and van Der Donk, W.A. (1998). Protein radicals in enzyme catalysis. Chem. Rev. 98, 705–762. Thelander, L. (2007). Ribonucleotide reductase and mitochondrial DNA synthesis. Nat. Genet. 39, 703–704. Vujanovic, M., Krietsch, J., Raso, M.C., Terraneo, N., Zellweger, R., Schmid, J.A., Taglialatela, A., Huang, J.W., Holland, C.L., Zwicky, K., et al. (2017). Replication fork slowing and reversal upon DNA damage require PCNA polyubiquitination and ZRANB3 DNA translocase activity. Mol. Cell 67, 882–890.e5. Weston, R., Peeters, H., and Ahel, D. (2012). ZRANB3 is a structure-specific ATP-dependent endonuclease involved in replication stress response. Genes. Dev. 26, 1558–1572. Wisitpitthaya, S., Zhao, Y., Long, M.J., Li, M., Fletcher, E.A., Blessing, W.A., Weiss, R.S., and Aye, Y. (2016). Cladribine and fludarabine nucleotides induce distinct hexamers defining a common mode of reversible RNR inhibition. ACS. Chem. Biol. 11, 2021–2032. Yamauchi, T., Nowak, B.J., Keating, M.J., and Plunkett, W. (2001). DNA repair initiated in chronic lymphocytic leukemia lymphocytes by 4-hydroperoxycyclophosphamide is inhibited by fludarabine and clofarabine. Clin. Cancer Res. 7, 3580–3589. Yuan, J., Ghosal, G., and Chen, J. (2012). The HARP-like domain-containing protein AH2/ZRANB3 binds to PCNA and participates in cellular response to replication stress. Mol. Cell 47, 410–421. Zhou, B.-S., Tsai, P., Ker, R., Tsai, J., Ho, R., Yu, J., Shih, J., and Yen, Y. (1998). Overexpression of transfected human ribonucleotide reductase M2 subunit in human cancer cells enhances their invasive potential. Clin. Exp. Metastasis 16, 43–49.

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

STAR+METHODS KEY RESOURCES TABLE

REAGENT or RESOURCE

SOURCE

IDENTIFIER

AB81085; RRID: AB_1640848

Antibodies See Method Details for dilutions and use of each antibody Rabbit polyclonal anti-RNR-a

Abcam

Mouse monoclonal anti-RNR-a

Proteintech

60073-1-Ig; RRID: AB_2238496

Mouse monoclonal anti-RNR-b

Abcam

AB57653; RRID: AB_2253869

Mouse monoclonal anti-BrdU

Calbiochem/Merck

Clone MoBu-1; RRID: AB_10681767

Rabbit polyclonal anti-ZRANB3

Bethyl Laboratories

A303-033-A; RRID: AB_10773114

Rabbit polyclonal anti-ZRANB3

Protein Atlas

HPA035234; RRID: AB_2674530

Goat anti-rabbit Alexa-488

Invitrogen

A11008; RRID: AB_143165

Donkey anti-rabbit Alexa-647

Abcam

Ab150075; RRID: AB_2752244

Donkey anti-mouse Alexa-568

Abcam

Ab175472; RRID: AB_2636996

Anti-mouse-HRP secondary

Abcam

Ab6789; RRID: AB_955439

Anti-rabbit-HRP secondary

Abcam

Ab97051; RRID: AB_10679369

Agilent Technologies

Cat. No. 200315

Bacterial and Virus Strains XL-10 competent cells Chemicals, Peptides, and Recombinant Proteins Sulfo-cyanine5 azide

Lumiprobe

Cat. No. B3030

TransIT-2020

Mirus Bio LLC

Cat. No. MIR5400

TransIT-LT1

Mirus Bio LLC

Cat. No. MIR2300

Phusion HotStartII DNA polymerase

Thermo Fisher Sci

Cat. No. F549L

Clofarabine (ClF)

AK Scientific

Cat. No. G787

Triapine (3-AP)

This paper

Synthetic procedure described previously (Aye et al., 2012b)

Cladribine (CLA)

AK Scientific

Cat. No. E997

Gemcitabine (F2C)

AK Scientific

Cat. No. F287

Fludarabine monophosphate (FLUMP)

Selleckchem

Cat. No. S1491

Hygromycin B (50 mg/ml)

Invitrogen

Cat. No. 10687010

Tetracycline

Sigma Aldrich

Cat. No. T7660

Zeocin (100 mg/ml)

Invitrogen

Cat. No. R25001

Puromycin Dihydrochloride (10 mg/ml)

Invitrogen

Cat. No. A1113803

Crystal violet

Sigma Aldrich

Cat. No. C6158

Minimal Essential Media+GlutamaxTM (MEM)

Gibco

Cat. No. 41090-028

Opti-MEM

Gibco

Cat. No. 51985-026

Dulbecco’s PBS

Gibco

Cat. No. 14190-169

DMEM media (without sodium pyruvate supplemented)

Gibco

Cat. No. 41965-039

DMEM media (with sodium pyruvate supplemented)

Gibco

Cat. No. 41966-029

100X sodium pyruvate

Gibco

Cat. No. 11360-070

100X non-essential amino acids

Gibco

Cat. No. 11140-035

100X penicillin streptomycin

Gibco

Cat. No. 15140-122

Fetal Bovine Serum (FBS)

Sigma Aldrich

Cat. No. F0392

Newborn Calf Serum (NBCS)

Gibco

Cat. No. 16010159

Tris-(carboxyethyl)phosphine hydrochloride (TCEP)

ChemImpx

Cat. No. 23004

Deoxyguanosine (dG)

ChemImpx

Cat. No. 33054

Deoxyadenosine (dA)

ChemImpx

Cat. No. 22121

2’-Deoxy-5-ethynyluridine (EdU)

Abcam

Cat. No. Ab146186 (Continued on next page)

Cell Chemical Biology 27, 1–12.e1–e5, February 20, 2020 e1

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

Continued REAGENT or RESOURCE

SOURCE

IDENTIFIER

Polybrene (Hexadimethrine bromide)

Sigma Aldrich

Cat. No. 107689

Critical Commercial Assays AlamarBlue Cell Viability Assay

Thermo Fisher Sci

Cat. No. DAL1100

Duallink PLA Kit (Mouse/Rabbit specific secondary antibody)

Sigma Aldrich

Cat. No. DUO92101

MycoGuard Mycoplasma PCR detection Kit

Genecopoeia

Cat. No. MP004

Western-blot ECL substrate and

Pierce

Cat. No. 32106

ECL-Plus substrate

Pierce

Cat. No. 32132

Human: HEK293T cells

ATCC

Cat. No. CRL-11268TM

Experimental Models: Cell Lines Human: Flp-In T-REx HEK293 cells

Thermo Fisher

Cat. No. R75007

Mouse: NIH-3T3 cells

ATCC

Cat. No. CRL-1658TM

Human: HeLa cells

ATCC

Cat. No. CCL-2TM

Human: U2OS cells (native)

Prof. Massimo Lopes and Prof. David Cortez

Vujanovic et al., 2017

Human: U2OS cells with ZRANB3 knock out (clone 35)

As above

As above

Human: U2OS cells with ZRANB3 knock out (clone 38)

As above

As above

Human: U2OS cells with ZRANB3 knock out and ZRANB3wt re-expression (generated from clone 35)

As above

As above

siControl

Santa Cruz Biotechnology

Cat. No. sc-37007

siControl

Santa Cruz Biotechnology

Cat. No. sc-44230

Oligonucleotides See Method Details for procedures for siRNA and shRNA knockdowns

siZRANB3 or siZRANB3(1)

Dharmacon

Cat. No. D-010025-03-0005

siZRANB3(2)

Invitrogen

Cat. No. s38487

siZRANB3(3)

Invitrogen

Cat. No. s38488

siRNR-a or siRNR-a(1)

Dharmacon

Cat. No. D-004270-01-0005

siRNR-a(2)

Dharmacon

Cat. No. D004270-02-0002

siRNR-a(3)

Dharmacon

Cat. No. D004270-03-0002

siRNR-b

Dharmacon

Cat. No. D-010379-05-0005

shControl target for lac Z (CGCGATCGTAATCACCCGAGT)

Dr. Andrew Grimson

N/A

shControl target for GFP (GTCGAGCTGGACGGCGACGTA)

Dr. Andrew Grimson

N/A

shIRBIT target for mouse IRBIT (CCAGAAAGTTGATTCTTTA)

Sigma Aldrich

Cat. No. TRCN0000317932

shZRANB3 target for mouse ZRANB3 (TGGTCTTTGCGCACCATTTA)

Sigma Aldrich

Cat. No. TRCN0000239044

Recombinant DNA pCMVR8.74

Addgene (gift from Didier Trono)

Cat. No. 22036

pCMV-VSV-G

Addgene (Stewart et al., 2003)

Cat. No. 8454

ImageJ

NIH

imagej.nih.gov/ij/

GraphPad Prism V8.2

GraphPad Software

graphpad.com/scientific-software/prism/

Zen Imaging Process software

Zeiss

zeiss.com/microscopy/int/products/ microscope-software/zen.html

Software and Algorithms

e2 Cell Chemical Biology 27, 1–12.e1–e5, February 20, 2020

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

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, Yimon Aye ([email protected]) All unique/stable reagents/cell lines generated in this study are available from the Lead Contact and may require a completed Materials Transfer Agreement, in accordance with the authors’ institutional policy. Plasmids generated in this study are being deposited to Addgene and in due course can be found at the following URL: https://www.addgene.org/Yimon_Aye/. EXPERIMENTAL MODEL AND SUBJECT DETAILS Cell Lines Sex of Cell Lines or Primary Cultures No primary cultures were used in this study. HEK293T, HeLa, and NIH-3T3 cells used in this study were from ATCC, whereas for FlpIn T-REx HEK293 cells, were from Thermo (See Key Resources Table for catalogue numbers). Information regarding the origin of U2OS cells, which were a kind gift of Prof. Massimo Lopes (University of Zurich, Switzerland) and Prof. David Cortez (Vanderbilt University, USA) (See Key Resources Table), may be available from the indicated scientists. Details on all of these cell types HeLa, HEK293(T), NIH-3T3, U2-OS are available on ATCC (https://www.lgcstandards-atcc.org/en.aspx); however, please note that gender information is not always listed on the manufacturer’s website. Cell Culture Maintenance HEK293T, Flp-In T-REx HEK293, and HeLa cells were cultured in 1X MEM+GlutamaxTM media (Gibco, 41090-028) supplemented with 10% FBS (Sigma, F0392), 1X NEAA (Gibco, 11140-035), 1X sodium pyruvate (Gibco, 11360-070) and 1X Pen-Strep (Gibco, 15140-122). NIH-3T3 cells were cultured in 1X DMEM (Gibco, 41965-039) supplemented with 10% NBCS (Newborn Calf Serum, Gibco, 16010159). U2OS cells were cultured in 1X DMEM (Gibco, 41966-029) supplemented with 10% FBS (Sigma, F0392) and 1X Pen-Strep (Gibco, 15140-122). All cell lines were grown in a humidified, 5% CO2 incubator at 37 C. Mycoplasma testing for all the cell lines in this work was performed using MycoGuard Mycoplasma PCR detection Kit every 3-months. METHOD DETAILS Cell Line Engineering Generation of Flp-In T-REx HEK293 Flp-In T-REx HEK293 cells were cultured in MEM supplemented with 10% FBS, 1X pyruvate and non-essential amino acids, 100 mg/ml zeocin and 1X penicillin/streptomycin in a humidified atmosphere at 37 C with 5% CO2. To generate the line, HEK293TREX cells were co-transfected with the POG44 vector encoding the Flp recombinase and pcDNA5/FRT/TO-RNR-a(D57N)-NLS-2XFlag using Mirus 2020 in 6 well plates. Two days after transfection, the cells were split and passaged into 10 cm dishes, then after 1.5 d the cells were changed to selective media containing 150 mg/mL Hygromycin B. After 3 weeks, single colonies had formed, and these were harvested using cloning cylinders. Individual clones were grown in 12 well plates and verified for expression using western blot or imaging. To induce expression, cells were exposed to 100 ng/mL tetracycline. Generation of Knockdown Cell Lines Using Lentivirus HEK293T packaging cells were seeded and grown overnight in antibiotic-free media in 6 well plates. At 80% confluence, each well was transfected with packaging plasmid (pCMV-R8.74psPAX2, 500 ng), envelope plasmid (pCMV-VSV-G, 50 ng) and pLKO vector (500 ng) using TransIT.LT1 as per the manufacturer’s protocol. After 18 h media were removed and replaced with 20% serum-containing media. After 24 h, media were collected, spun down and passed through a 0.45-micron filter and used directly. Cells in log phase were treated with 1 ml of virus supernatant (from above) in 8 mg/ml polybrene in a total of 6 ml of media in a 6-well plate. After 24 h, media were removed and replaced with media containing 2 mg/ml puromycin (which was 100% toxic to all lines used in this study). Cells were cultured till plate was confluent, then cells were split and moved to a 10 cm dish in 2 mg/ml puromycin containing media and grown till confluent again. At this point, the line was considered to be ‘‘selected’’, and expression of target gene was analyzed by western blot and compared to shRNA controls. Up to passage 5 was used for these cells and they were typically grown in 1.5 mg/ml puromycin. Cell-Based Assays Immunofluorescence Assay Cells were grown in 35 mm glass-bottom plates to appropriate confluence and treated as described. At the conclusion of the experiment cells were fixed by addition of 2 ml MeOH (–20 C) for 20 min (– 20 C). MeOH was removed and PBS was added and stored at 4 C for days to weeks. NOTE: fixing with MeOH is critical for appropriate visualization of RNR-a. Cells were then washed 3 times with PBS and carried straight on to blocking (2 ml imaging blocking buffer: 3% BSA & 0.2% Triton x-100 in PBS at 37 C for 1 h). Plates were washed 2 times with PBS (1 ml), then treated with primary antibody, in antibody incubation buffer: 1% BSA and 0.02% Triton x-100 in PBS, using 120–150 ml to cover only the glass portion of the plate. Plates were incubated at rt for 2 h. Primary antibody was

Cell Chemical Biology 27, 1–12.e1–e5, February 20, 2020 e3

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

removed and plates were washed two times with PBS (1 ml). Then secondary antibody (120–150 ml; 1:1000 dilution in antibody incubation buffer) was added (1 h). Plates were washed one time in PBS containing 1 mg/ml DAPI, then once with PBS and stored at 4 C until ready to image. Dual Pulse Assay (DPA) For imaging BrdU/EdU-treated cells: directly before blocking (see immunofluorescence assay above), cells were treated with 1 M HCl at 4 C for 10 min, then 2 M HCl at rt for 10 min, then warmed to 37 C for 10 min. After this time cells were treated with PBS two times, then 50 mM HEPES pH 7.6. Cells then were treated with a freshly made mix of 1 mM CuSO4 (from a 100 mM CuSO4 stock in water), 10 mM sodium ascorbate (made as a 100 mM stock in 500 mM HEPES pH 7.6) all in 50 mM HEPES pH 7.6 and 15 mM Cy5-azide. Cells treated with this mixture were left for 20 min in the dark. Note: this mix must be made just before use. Then cells were treated with imaging blocking buffer for 1 h, then washed twice with PBS, then primary antibody buffer was applied and subsequent steps were carried out as described above, starting from the primary antibody addition step. Note: Clone MoBu-1 anti-BrdU is essential for the DPA (dual pulse imaging assay). Proximity Ligation Assay (PLA) PLA was carried out as suggested by Sigma Aldrich using the procedure for Doulink in situ red starter kit with mouse/rabbit, primary antibody incubations were as described above. AlamarBlue Cell Proliferation Assay For NIH-3T3 cells, 4000 cells were plated in clear 96-well plates (total volume 100 ml), then 100 ml 2X compound was added in complete growth media and cells were allowed to grow for 48 h. After this time, 100 ml media were removed and AlamarBlue (10 ml) was added and cells were maintained at 37 C for 3-7 h. For HEK293T cells, 10 cm plates were transfected with the stated siRNA and Dharmafect1 for 12 h, then 4000 cells were split into every 96 well plates (total volume 100 ml), then 100 ml 2X compound was added in complete growth media and cells were allowed to grow for 2 days. At the appropriate time point, number of cells was measured using fluorescence as per the manufacturer’s protocol. Data shown are mean +/- standard deviation of at least 8 independent replicates. Transformation Assays Focus Formation and Scratch/Wound Healing Assays in NIH-3T3 Cells For focus formation assays, NIH-3T3 cells were transfected with H-rasG12V (0.5 mg) and indicated transgenes and/or empty vector (2.5 mg) in a 6 well plate using lipofectamine LTX and plus reagent. After 2 days, cells were split and equal numbers of cells (0.5– 0.7 million) were plated in a 6 cm dish. Cells were grown to confluence with medium exchanges every 2–3 days (after 4 days 4% serum media were used) for 2-3 weeks, until foci became visible, then cells were fixed in 4% cracked PFA for 30 min, then stained with crystal violet solution and destained with water as required. For scratch/wound healing assay, cells on glass plates were grown to confluence then scratched using a glass pipette to introduce a central linear scratch across the surface (two scratches were made perpendicular to each other and analyzed separately). 2D cell migration into the wound was measured by imaging on a Biotek Cytation III plate reader. The plot was generated by measuring the initial wound width and dividing the width of the wound as a function of time by the initial width for each set. Wound widths varied by less than 25% across all data sets. QUANTIFICATION AND STATISTICAL ANALYSIS Sample Size n for Figures with Quantification n for imaging experiments represents the number of single cells or in a few instances, panels quantified from at least five fields of view from at least two plates. n for AlamarBlue assays represents the number of individual wells under identical experimental conditions. In all figures, Student’s two-tailed t test was performed to determine significance. ns = non-significant (p > 0.05), *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. DATA AND CODE AVAILABILITY The published article includes all datasets generated or analyzed during this study. ADDITIONAL RESOURCES Chemicals and Reagents All chemicals were from Sigma-Aldrich and were of the highest grade unless otherwise stated. Cyanine5(Cy5)-azide was from Lumiprobe; ClF, CLA, and F2C were from AK Scientific. Fludarabine monophosphate (FLUMP) was from Selleckchem. TCEP, deoxyguanosine (dG), deoxyadenosine (dA), adenosine (A) and EdU (working concentration 20 mM) were from ChemImpex. Deoxycytidine was from Sigma-Aldrich. Triapine (3-AP) was synthesized in house as previously described. Hygromycin B was from Invitrogen. AlamarBlue was from Invitrogen, used as the manufacturer’s instructions. Minimal Essential Media, RPMI, Opti-MEM, Dulbecco’s e4 Cell Chemical Biology 27, 1–12.e1–e5, February 20, 2020

Please cite this article in press as: Long et al., Clofarabine Commandeers the RNR-a-ZRANB3 Nuclear Signaling Axis, Cell Chemical Biology (2019), https://doi.org/10.1016/j.chembiol.2019.11.012

PBS, 100X pyruvate (100 mM), 100X non-essential amino acids (11140-050) and 100X penicillin streptomycin (15140-122) were from Gibco. Transfection: Cells were transfected using Mirus 2020 (all small scale and growth/proliferation experiments, used according to manufacturer’s instructions) or Mirus TransIT.LT1 (for virus production). MycoGuard Mycoplasma PCR Detection Kit was from Genecopoeia. Cells were typically left for 24–48 h without changing media prior to harvest/assay. ECL substrate or ECL-Plus substrate were from Pierce and used as directed. Acrylamide, ammonium persulfate, TEMEDA, Precision Plus protein standard was from Bio-Rad. All lysates were quantified using the Bio-Rad Protein Assay (Bio-Rad), relative to BSA as a standard (Bio-Rad). PCR was carried out using Phusion Hot start II (Thermo Scientific) as per manufacturer’s requirements. All plasmid inserts were validated by sequencing at Cornell Biotechnology sequencing core facility. All sterile cell culture plastic-ware was from CellTreat, except for glassbottomed dishes used for imaging that was from In Vitro Scientific. The cloning cylinders for harvesting cell colonies (3166-6) were from Corning. The PLA kit was purchased from Sigma Aldrich (Doulink in situ red starter kit with mouse/rabbit). Antibodies RNR-a [AB81085 (Rb), AbCam. 1:1000 (WB), 1:200 (IF); 60073-1-Ig (Ms), Proteintech, 1:2000 (WB), 1:200 (IF), 1:300 (PLA for both HeLa and U2OS cells)]; RNR-b [AB57653 (Ms), AbCam, 1:1000 (WB), 1:200 (IF)]; BrdU [NA61, Calbiochem, 1:1000 (Clone MoBu-1, DPA, selective for BrdU over EdU)]; ZRANB3 [A303-033-A (Rb), Bethyl Laboratories. 1:2500 (WB); 1:200 (IF); HPA035234 (Rb), ATLAS, 1:200 (WB for U2OS cells and IF for HEK293T cells), 1:300 (PLA for HeLa cells), 1:25 (IF and PLA for U2OS cells)]; goat anti-rabbit Alexa-488 [A11008, Invitrogen. 1:1000 (IF)]; donkey anti-rabbit Alexa 647 [Ab150075, AbCam, 1:1000 (IF)]; donkey anti-mouse Alexa 568 [Ab175472, AbCam, 1:1000 (IF)]. WB = western blot; IF = immunofluorescence; PLA proximity ligation assay. Ms = mouse. Rb = rabbit. siRNAs siRNA was delivered using dharmafect as per the manufacturer’s protocol. Unless otherwise indicated in figure legends, cells were transfected at 25% confluence, and typically left for 2 days prior to experimentation. siZRANB3(1) siRNA was purchased from Dharmacon (D-010025-03-0005). siZRANB3(2) and siZRANB3(3) were purchased from Invitrogen (s38487 and s38488). Two negative control siRNAs were purchased from Santa Cruz Biotechnology (sc-37007 and sc-44230). The sequences of the 3 different ZRANB3-targeting siRNAs were as follows: siZRANB3 or siZRANB3(1): GAUCAGACAUCACACGAUU siZRANB3(2): CAAGAGAUAUCAUCGAUUAtt siZRANB3(3): GAUUCGAUCUAAUAACAGUtt Invitrogen designed their siRNAs by adding "tt" to the 3’-ends to improve the stability of the siRNA. RNR-a siRNA (D-004270-01-0005, D004270-02-0002, and D004270-03-0002) and RNR-b siRNA (D-010379-05-0005) were purchased from Dharmacon. The sequences of the RNR-a siRNAs were as follows: siRNR-a or siRNR-a(1): GAACACAUACGACUUUA siRNR-a(2): GGACUGGUCUUUGAUGUGU siRNR-a(3): GCACAGAAAUAGUGGAGUA. shRNAs The following shRNA sequences were used in pLKO1 vector. The shRNA plasmids for control were gifts from Dr. Andrew Grimson (Cornell University, Ithaca, NY). The shRNA plasmids for ZRANB3 (of which one was functional) and IRBIT were purchased from Sigma-Aldrich. For details, see lentivirus production and infection method sections. shControls: (1) CGCGATCGTAATCACCCGAGT (lac Z); (2) GTCGAGCTGGACGGCGACGTA (GFP). shIRBIT (1) CCAGAAAGTTGATTCTTTA (mouse) shZRANB3 (1) TGGTCTTTGCGCACCATTTA (mouse) Equipment Plate reader assays were carried out on a Biotek Cytation III plate reader. All confocal imaging experiments except those involving U2OS cells were carried out on a Zeiss LSM710 confocal microscope (Cornell Biotechnology Imaging Core Facility), equipped with an Inverted Axio Observer.Z1 microscope with 405, 458, 488, 514, 543, 561, and 633 nm laser lines. All U2OS cell-imaging experiments were performed on a Zeiss LSM700 confocal microscope (EPFL Bioimaging and Optics Platform), equipped with an Inverted Axio Observer.Z1 microscope with 405, 488, 555, and 639 nm laser lines.

Cell Chemical Biology 27, 1–12.e1–e5, February 20, 2020 e5