Abstract

   Small molecule inhibitors that target the E3 ligase activity of
   MDM2-MDM4 have been explored to inhibit the oncogenic activity of
   MDM2-MDM4 complex. MMRi62 is a small molecule that was identified using
   an MDM2-MDM4 E3 ligase-based high throughput screen and a
   cell-death-based secondary screen. Our previous studies showed that
   MMRi62 promotes MDM4 degradation in cells and induces p53-independent
   apoptosis in cancer cells. However, MMRi62 activity in solid tumor
   cells such as melanoma cells, especially in BRAF inhibitor resistant
   melanoma cells, have not been explored. Although its promotion of MDM4
   degradation is clear, the direct MMRi62 targets in cells are unknown.
   In this report, we show that MMRi62 is a much more potent
   p53-independent apoptosis inducer than conventional MDM2 inhibitors in
   melanoma cells. A brief structure-activity study led to development of
   SC-62-1 with improved activity. SC-62-1 potently inhibits and
   eliminates clonogenic growth of melanoma cells that acquired resistance
   to BRAF inhibitors. We developed a pair of active and inactive SC-62-1
   probes and profiled the cellular targets of SC-62-1 using a chemical
   biology approach coupled with IonStar/nano-LC/MS analysis. We found
   that SC-62-1 covalently binds to more than 15 hundred proteins in
   cells. Pathways analysis showed that SC-62-1 significantly altered
   several pathways including carbon metabolism, RNA metabolism, amino
   acid metabolism, translation and cellular response to stress. This
   study provides mechanistic insights into the mechanisms of action for
   MMRi62-like quinolinols. This study also suggests multi-targeting
   compounds like SC-62-1 might be useful for overcoming resistance to
   BRAF inhibitors for improved melanoma treatment.

   Keywords: MMRi62, quinolinol, MDM2-MDM4, E3 ligase, cell death,
   targets, chemical robe

1. Introduction

   Small-molecule inhibitors targeting cancer-driver protein kinases and
   enzymes constitute a majority of the targeted therapies approved by FDA
   in the past decades [[42]1]. These targeted therapeutics are designed
   to target the interfaces in ATP binding pockets, domains of kinase
   activation, catalytic domains of enzymes or ligand binding for
   inactivation of the kinase activity that drives the cancerous cell
   proliferation. Although significantly improved cancer treatment for
   many cancer types, these therapies without exception face three
   challenges: low responsive rate, short-lived efficacy and drug
   resistance. Low responsive rate may be caused by additional genetic
   alterations in addition to the alterations in the intended drug targets
   in different patients. Short-lived efficacy is often intertwined with
   drug resistance that includes intrinsic and acquired drug resistance.
   The intrinsic drug resistance can be mediated by tumor cell
   heterogeneity, differential expression of efflux pump, and tumor
   microenvironment, while acquired resistance of initial responsive
   cancer cells can be gained through distinct mechanisms including gene
   mutation and non-genomic alteration of the targeted signaling pathways
   [[43]2]. For example, acquired resistance to first and
   second-generation EGFR inhibitors in lung cancer can be conferred by
   point mutations such EGFR T790M mutation and resistance to
   third-generation EGFR inhibitors can be conferred by EGFR C797S
   mutation [[44]3]. In the case of mutant BRAF inhibitors in melanoma,
   the acquired resistance is conferred by mechanisms involving genetic
   and epigenetic changes that activate different signaling pathways such
   as MAPK to bypass the effect of BRAF inhibition [[45]4]. Likewise,
   inhibitors targeting MDM2-p53 interaction for p53-based therapies face
   resistance mechanisms involving mutations in p53 and other components
   of the p53 pathway [[46]5,[47]6,[48]7] including MDM4 overexpression
   [[49]8,[50]9,[51]10].

   Despite our understanding of the various resistance mechanisms, the
   root cause of the limited efficacy of targeted therapies is lack of
   potent cell killing effect. This is because most drug targets for
   targeted therapies are mutated or deregulated kinases or enzymes, whose
   inactivation most likely results in reduced proliferation rather than
   cell death. While reduced proliferation prolongs progression-free
   survival, it allows cancer cells to rewire to alternative pathways or
   the treatment pressure selects resistant cancer populations.
   Eventually, the once-shrank tumors adapt the blockade by the targeted
   therapies leading to regrowth of more aggressive subtypes. In case of
   targeting MDM2 for p53-based cancer therapy, drug response to MDM2
   inhibitors may differ significantly in terms of whether apoptosis or
   growth arrest is the major endpoint. This is because p53 plays a
   differential effect in apoptosis and cell cycle arrest in different
   cell types, with mainly apoptosis in leukemic or lymphoma cells and
   mainly growth arrest in fibroblasts and other cell types [[52]11].
   Indeed, MDM2 inhibitors induced a mixture of apoptosis and growth
   arrest in cancer patients in clinical trials [[53]12,[54]13].

   To identify inhibitors that target the oncogenic activity of MDM2 with
   better cell killing activity, we used a unique approach for screening
   the lead compound. Our strategy was to first identify small molecules
   that target the RING domain of MDM2 which carries the E3 ligase
   activity and then use apoptosis assay in the secondary screen of
   primary hits identified by the biochemical screening [[55]14]. This
   strategy is based on findings that the RING domains of both MDM2 and
   MDM4 are essential for p53 regulation in vivo [[56]15,[57]16,[58]17]
   and that the E3 ligase activity of MDM2-MDM4 RING heterodimers is
   required for both p53 regulation in vivo and p53-independent cell cycle
   regulation [[59]18]. Using this strategy, we identified MMRi62 that
   modulates the E3 ligase activity of MDM2-MDM4, promotes MDM4
   degradation in cells and potently induces p53-independent apoptosis in
   leukemic cells [[60]19] and ferroptosis in pancreatic cells [[61]20].
   Although MMRi62 hits the intended targets of MDM2-MDM4 complex, it is
   not clear how MMRi62 potently induces cancer cell death in different
   death mode. In this study, we report that MMRi62 has advantages over
   conventional MDM2 inhibitors in cell death induction in melanoma cells
   and SC-62-1 as MMRi62 derivatives has improved activity and is useful
   in eliminating melanoma cells with acquired resistance. We also report
   the results of SC-62-1-targeted pathways using SC-62-1 chemical probes
   and quantitative proteomics.

2. Results and Discussion

2.1. MMRi62 Has Advantage of Preferential Induction of Apoptosis over
MDM2-p53 Inhibitors in Melanoma Cells

   Our previous study showed that MMRi62 induces MDM4 degradation and
   p53-independent apoptosis in leukemic cells [[62]19]. We asked if this
   pro-apoptotic activity of MMRi62 also takes place in solid tumor cells
   such as melanoma cells. We used A375 and p53-knockdown shp53-A375 cells
   and treated them with either MMRi62 or AMG232, a potent MDM2 inhibitor
   that specifically disrupts MDM2-p53 interaction [[63]21] as a
   comparison in anti-proliferation assays. Our results showed that MMRi62
   and AMG232 had comparable IC[50] values in A375 cells, but MMRi62’s
   anti-melanoma activity was not affected by p53 knockdown since A375 and
   shp53-A375 had similar IC[50]s ([64]Figure 1A,B). In contrast to
   MMRi62, AMG232 showed a 40-fold reduced activity in shp53-A375 cells as
   compared to A375 cells ([65]Figure 1B). These data demonstrated that
   the anticancer activity of AMG232 is p53-dependent and MMRi62’s
   activity is p53-independent, although both compounds induced p53
   accumulation in A375 cells ([66]Figure 1C,D). In contrast to AMG232,
   MMRi62-treated cells showed strong MDM4 downregulation in both A375 and
   shp53-A375 cells. MMRi62 treatment induced fast and strong apoptotic
   PARP cleavage (cPARP) in A375 cells at 24 h while AMG232 induced cPARP
   in A375 cells only after 48 h treatment ([67]Figure 1C–E).
   MMRi62-induced cPARP was much weaker in shp53-A375 cells than in A375
   cells, suggesting a major p53-dependent apoptosis mechanism was
   triggered in A375 cells. However, the MMRi62-induced p53-independent
   apoptosis was sustained up to 72 h in MMRi62-treated shp53-A375 cells
   ([68]Figure 1C,F). Given that MMRi62 had similar IC[50] values in A375
   and shp53-A375 cells but with different levels of apoptosis, MMRi62’s
   anti-melanoma activity must involve multiple death modes including
   p53-dependent apoptosis, p53-independent apoptosis and non-apoptotic
   cell death.

Figure 1.

   [69]Figure 1
   [70]Open in a new tab

   Westen Blot analysis of MMRi62 and AMG232 for the effect on MDM2, MDM4,
   p53 and cleaved PARP induction. (A,B) Dose response curves in the
   presence of MMRi62 (A) or AMG232 (B) for A375 and shp53-A375 cells.
   IC[50]s are shown at the bottom. The representative data of three
   independent experiments were shown and the IC[50] values were obtained
   from three independent experiments. (C,D) WB analysis of MDM2, MDM4,
   p53 and cleaved PARP (cPARP) induction in A375 and shp53-A375 cells
   treated with MMRi62 (C) or AMG232 (D) for 24 h (E). WB analysis of
   MDM2, MDM4, p53 and cleaved PARP (cPARP) induction in A375 treated with
   AMG232 or MMRi62 for 48 h and 72 h (F), WB analysis of cleaved PARP
   (cPARP) and p53 in shp53A375 cells treated with AMG232 or MMRi62 for 48
   h and 72 h. Tubulin is the protein loading control. The intensities of
   cPARP and tubulin bands in (C,F) were quantified by ImageJ1.54P
   software and fold increases in cPARP over untreated control shown under
   cPARP WB were obtained after normalization against tubulin.

2.2. SC-62-1, an MMRi62 Derivative with Improved Activity

   A previous SAR study on MMRi62 and MMRi67 identified the critical role
   of the quinolinol’s hydroxyl moiety in anti-proliferation activity of
   these molecules [[71]22]. Optimization of MMRi62 led to development of
   SC-62-1. We synthesized the methoxy counterparts of MMRi62 (MMRi62Me)
   and SC-62-1 (SC-62-1Me) and tested them together in A375 melanoma
   cells. Our results reached the same conclusion that hydroxyl group of
   MMRi62/SC-62-1 is essential for its activity since their methoxy
   counterparts lost activity ([72]Figure 2). The calculated CLogP of
   SC-62-1 is 2.13, placing it in the realm of the mean value of ClogP for
   approved drugs, which has stayed consistently at 2.3–2.6 over a long
   period of time [[73]23]. By replacing a chloro substituent with its
   bioisosteric nitrile [[74]24] in the structure of SC-62-1, we have
   improved the water solubility over MMRi62 whose calculated CLogP is
   3.42. To assess whether SC-62-1 is less toxic to non-cancer cells, we
   tested SC-62-1 in immortalized human keratinocytes HaCaT cells and
   obtained an IC[50] of 0.61 ± 0.13 μM which is 5-fold higher than that
   of A375 cells, suggesting that SC-62-1 has limited degree of
   selectivity for melanoma cells over immortalized keratinocytes. We
   noticed that HaCaT cells grew as fast as A375 cells in culture which
   might suggest that SC-62-1 toxicity to HaCaT cells may be related to
   its cancer-like fast proliferation.

Figure 2.

   [75]Figure 2
   [76]Open in a new tab

   Chemical structure and antiproliferative activity of MMRi62, SC-62-1
   and their inactive counterparts. Top: Chemical structures of MMRi62,
   MMRi62Me, SC-62-1 and SC-62-1Me. Bottom: The IC[50] of MMRi62 and
   SC-62-1 in A375 cells.

2.3. SC-62-1 Prevents Appearance of Drug Resistance and Effectively Kills
Melanoma Cells with Acquired Resistance to BRAF Inhibitors

   Like most kinase inhibitors, BRAF inhibitors (BRAFi) such as
   Vemurafinib (Vem) potently inhibit growth of BRAFV600E mutant melanoma
   cells but do not eliminate them by cell death. A substantial population
   of resistant cells or drug-resistant persisters survives the BRAFi
   treatment and lead to disease progression [[77]25,[78]26]. To know how
   SC-62-1 affects the drug-resistant persisters in melanoma, we performed
   experiments with continuous drug exposure of A375 cells for 2 weeks
   with either Vem or SC-62-1 in cell culture by replacing the medium with
   fresh drug-containing medium every 3 days. The drug-resistant
   persisters always existed in Vem-treated plates. Vem up to 4 μM (33×
   IC[50] dose) could not eliminate the persister cells. In contrast,
   SC-62-1 at 1 μM eliminated persisters ([79]Figure 3A, microscopy images
   of crystal violet stained cells, top vs. middle). SC-62-1-Vem
   combination at 1:1 ratio eliminated Vem-resistant persisters at 2 and 4
   μM but failed at 1 μM, suggesting synergism at doses higher than 2 μM.
   We readily established Vem-resistant cell lines (A375VemR cells) by a
   stepwise increase in Vem concentration in cell culture. The A375VemR
   cells acquired Vem resistance by a ~180-fold increase in the IC[50]
   value. However, the A375VemR cells showed only little resistance to
   SC-62-1 with just a ~1.3-fold IC[50] change in the IC[50] value
   compared to A375 cells ([80]Figure 3B). Importantly, A375VemR cells
   remained sensitive to SC-62-1. Exposure of A375VemR cells to SC-62-1 or
   SC-62-1-Vem combo at >4 μM for 8 days completely wiped out these cells,
   but Vem alone at 4-to-16 μM failed to do so ([81]Figure 3C). The
   results of the antiproliferation assay indicated that the SC-62-1-Vem
   combinations generated strong synergisms in the A375VemR cells in a
   wide range of drug concentrations with inhibition levels higher than
   40% ([82]Figure 3D). The combination reached strong synergism level at
   effect levels >90% inhibition with a CI < 0.3. These data suggest that
   SC-62-1 can prevent the development of resistance and eliminate the
   resistant to Vem treatment cells on its own or with Vem at doses higher
   than 2 μM.

Figure 3.

   [83]Figure 3
   [84]Open in a new tab

   Effect of SC-62-1 and its combinations with Vemurafinib (Vem) on
   survival and proliferation of melanoma cells and Vem-resistant cells in
   vitro. (A) Microscopy images of crystal violet stained A375 cells after
   2 week exposure to 1, 2, 4 μM Vem, SC-62-1 or their combinations. (B)
   Dose response curves of Vem-resistant A375 cells (A375VemR) with
   acquired resistance. (C) Microscopy images (top) and whole plates of
   crystal violet stained A375VemR cells after 8 day exposure to 4, 8, and
   16 μM Vem, SC-62-1 or their combinations. (D) Combination index diagram
   obtained for SC-62-1-Vem combinations. CI < 1, synergism, CI = 1,
   additive, CI > 1, antagonism. Arrowheads indicate that the higher the
   CI values above 1, the more antagonistic they are, while the smaller
   the CI values below 1 toward zero, the more synergistic they are.

2.4. Design, Synthesis and Chracterization of SC-62-1 Probes

   The capability of MMRi62 and SC-62-1 in inducing cancer cell death made
   them attractive tool compounds for uncovering the molecular basis of
   cancer cell killing. To identify MMRi62/SC-62-1’s direct targets beyond
   the induced MDM4 degradation, we used an unbiased approach for
   profiling cellular targets with MMRi62/SC-62-1 probes coupled with
   quantitative proteomics.

   Quinolinol Betti bases like MMRi62/SC-62-1 are synthesized via a three
   component Betti reaction involving condensation of 8-hydroxyquinoline,
   an aromatic aldehyde and 2-aminopyridine ([85]Figure 4A). We
   hypothesized that quinolinols act as covalent inhibitors as their
   primary mechanism of action (MOA) via a
   retro-aza-Michael-quinone-methide mechanism [[86]27,[87]28]. The
   reactive quinone methides target cysteine side chains of cellular
   proteins for covalent bonding which may alter the enzymatic activity or
   structural function of their targets ([88]Figure 4B). The potential
   targets of quinone methides can be broad including key enzymes in
   various pathways required for cancer cell growth and survival whose
   inactivation should trigger cell death due to cellular stress in
   different cancer types. This establishes a rationale that the
   quinolinol Betti bases may compensate the limitations of kinase
   targeted therapies by targeting multiple pathways to trigger cell death
   as an endpoint. In fact, about 30% of the marketed drugs are covalent
   inhibitors including tamoxifen and other targeted therapies
   [[89]29,[90]30,[91]31]. To identify targets for covalent binding of
   SC-62-1 predicted by our hypothetical MOA, we designed and synthesized
   the SC-62-16/SC-62-16Me pair of probes where SC-62-16 is the active
   compound and SC-62-16Me is the inactive counterpart ([92]Figure 4C).
   This pair of probes used an alkyl azide on C(5) for a “Click” reaction
   to crosslink the probe with alkyne-functionalized beads for pulldown of
   probe-bound proteins [[93]32]. As expected, SC-62-16 was active in A375
   cells with an IC[50] of 1.36 μM and promotes MDM4/MDM2 degradation
   while SC-62-16Me was inactive with an IC[50] of 134 μM without inducing
   MDM4/MDM2 degradation ([94]Figure 4D,E). Therefore, SC-62-16Me is a
   good negative control for identifying off-targets of SC-62-1 not
   involved in anticancer mechanisms. As expected, downregulation of MDM4
   by SC-62-16 is accompanied with p53 accumulation up to 5 μM. However,
   both MDM4 and p53 were downregulated by 10 μM SC-62-16. We speculate
   that lower concentrations of SC-62-16 below 5 μM induced specific
   effect on MDM4-p53 axis while high concentrations of SC-62-16 at 10 μM
   or above caused inhibition of both MDM4 and p53 expression. A moderate
   degree of apoptotic PARP cleavage was induced by 5 μM and 10 μM of
   SC-62-16 without clear dose-dependency, suggesting that apoptosis is
   only a part of involved death mechanisms and higher concentrations of
   the compound tend to trigger non-apoptotic endpoints.

Figure 4.

   [95]Figure 4

   [96]Figure 4
   [97]Open in a new tab

   Synthesis of SC-62-1 and azide probes, hypothetical mechanisms of
   action of SC-62-1 and molecular characterization of azide probes. (A)
   Synthesis of quinolinol Betti base SC-62-1. (B) Hypothetical MOA for
   SC-62-1 as covalent inhibitors of target proteins via quinone methide
   to alkylate target proteins. (C) Synthesis routes for SC-62-16 and
   SC-62-16Me. (D) Anti-proliferation assays in A375 cells treated with
   SC-62-16 and SC-62-16Me for 72h. IC[50s] values of the indicated
   compounds are shown. (E) Western blot analysis of SC-62-16 and
   SC-62-16Me effect on MDM4/MDM2 degradation and apoptotic PARP cleavage
   induction in treated A375 cells for 24 h. GAPDH is the protein loading
   control.

2.5. Identification of SC-62-16 Bound Proteins and Implicated Biological
Pathways

   We optimized a pulldown protocol in which biotin-PEG4-alkyne was used
   in click reaction [[98]33,[99]34] to tag a biotin moiety to
   SC-62-16/SC-62-16Me molecules that are covalently bound to proteins in
   treated cells, followed by pulldown of the biotin-tagged probe and its
   bound proteins by streptavidin-magnetic beads. The identities of
   proteins bound to the probes were then identified by quantitative
   IonStar/nano-LC/MS ([100]Figure 5A). We treated A375 cells at 10 μM for
   4 h before apoptosis occurs. Using 10 μM of the compounds ensured
   catching all possible SC-62-1 targets since this concentration induced
   strong MDM4 degradation and cell death at 24 h post-treatment
   ([101]Figure 4E).

Figure 5.

   [102]Figure 5
   [103]Open in a new tab

   Identification of SC-62-16 bound proteins and targeted pathways. (A)
   Diagram for the optimized procedure of pulldown of SC-62-1-bound
   proteins. (B) Number of SC-62-16-bound proteins and their distributions
   in different categories. (C) Results of pathway analysis of the
   SC-62-16-bound proteins. The top 11 significantly altered pathways in
   using KEGG and REACTOME database are shown. (D) List of 11 pathways
   that are significantly impacted by SC-62-1.

   Our approach worked well since the negative control compound SC-62-16Me
   did not bring down any proteins while SC-62-16 brought down more than
   1500 proteins ([104]Figure 5B). More than 600 of the SC-62-16-bound
   proteins are enzymes. These proteins are classified in several
   biochemical/biological categories including RNA metabolism,
   mitochondrial function, ubiquitin proteasome system, protein
   translation, DNA repair and checkpoint, cell migration, cell cycle and
   mitosis and DNA replication. When the sequences of these proteins were
   subjected to pathway analysis using KEGG [[105]35] and REACTOME
   database [[106]36], it was revealed that the significantly altered
   pathways include those for carbon metabolism, RNA metabolism, amino
   acid metabolism, RNA splicing, protein translation and cellular stress
   response ([107]Figure 5C,D).

   When the SC-62-16-bound proteins were grouped by biological processes
   or functions in a common pathway and ranked by the protein numbers of
   each group, they could be classified into six groups as shown in
   [108]Table 1. RNA metabolism was in the number one ranked category with
   a total of 124 proteins including 87 RNA binding proteins such as SUB1,
   FUS, EWSR1, RBMX, RRP43, DDX17, …etc. ([109]Table 1), and 37 proteins
   in mRNA splicing including SF3B1, SF3A1, RBM39, SF3B4, PRPF8, U2AF1,
   …etc. ([110]Table 1).

Table 1.

   SC-62-16 covalent binding proteins.
   Pathway Ranking Pathway Name Proteins Number PROTEIN Names (Some Are
   Partial List)
   1 RNA metabolism 124 87 RNA binding proteins such as SUB1, FUS, EWSR1,
   RBMX, RRP43, DDX17,
   DDX9, ADAR, SRB, RNH1, LUC7L2, ILF3, UBAP2L, SERBP1, KHDRBS1, NONO,
   IGF2BP2, SRRT, FTSJ3, MED23, RBM4, RCL1, EXOSC3, SYNCRIP, DHX30, EBP2,
   REXO2, ZFR, DDX1, …
   37 mRNA splicing such as SF3B1, SF3A1, RBM39, SF3B4, PRPF8, U2AF1,
   SCAF11,
   SF1, SFRS19, SRSF9, SRSF4, SFRS18, SF4, SFRS14, SFPQ, SRSF1, SRSF2,
   SRSF7,
   SRSF3, SPF45, SAP145, PRPF4, SRRM2, …
   2 Mitochondrial function 94 50 metabolic enzymes such as NDUFV1,
   MTHFD2, PYCR1, MCAT, GPD2, GSR,
   PPA2, PC, PCCA, MDH2, SHMT2, GOT2, ECHS1, IDH3A, CS, MCCC1, UQCRC1,
   SUCLG1, ALDH2, ECH1, HAGH, FH, IVD, ACAT1, ACADVL, DBT,
   SUCLG2, PDHB, CPOX, DLAT, AK2, PDHA1, COQ7, PDHX, HIBADH,
   SDHB, PRDX3, ATP5F1A, UQCRC2, ATP5F1B, NDUFS1, PGAM5,
   ATP5PB, HSD17B10, HADHB, GLS, ATP5PO, DLST, MECR, OGDH,….
   10 for mitochondrial tRNA metabolism: VARS2, VARSL, MARS2, IARS2,
   MRM1, CARS2, REXO2, TARS2, RPUSD3, WARS2.
   34 proteins for mitochondrial maintenance: APEX1, HAP1, WARS2, AFG3L2,
   TOMM70, PMPCB, CLPB, HSPD1, HSPA9, SLC25A24, SCAMC1, MTCH2,
   TUFM, ETFA, LRPPRC, SLC25A3, SSBP1, SLC25A22, HSPE1, DAP3, VDAC2,
   VDAC1, MRPS27, CCDC51, MTX1, MRPS5, TMEM126B, TMEM11, IMMT,
   MINOS2, LETM1, MAVS, CYCS, DNM1L.
   3 Protein degradation 55 18 ubiquitin E3 ligases:
   TRIM25, ARIH1, ARIH2, RNF20, CBL, HECTD1, HUWE1, TRIM18, RAD18,
   RBBP6, RNF113A, RNF149, RNF220, RNF34, RNFT2, TRIP12, UBR1, UBR4
   5 ubiquitin-conjugating enzymes: UBE2D2, UBE2D3, UBE2L3, UBE2N, UBE2Z,
   4 deubiquitinases: USP47, USP5, USP7, USP9
   28 proteasome components: PSMD4, PSMB2, PSMB6, PSMB4, PSMB1, ECPAS,
   PSMA6, PSMA1, PSMB7, PSMD2, PSMA3, PSMD7, PSMD13, PSMA2, PSMC4,
   PSMC6, PSMD8, PSMB5, PSMA5, PSMD11, PSMC3, PSMD1, PSMC1, PSMD4,
   PSMA7, PSMC2, PSMA4, PSMD12.
   4 Protein translation 50 18 translation initiation factors: EIF3A,
   EIF6, EIF3G, EIF5A, EIF4G1, EIF5, EIF3I,
   EIF2S2, CDC123, EIF5B, EIF2S3, EIF3D, EIF3H, EIF3B, EIF3K, EIF4G3,
   EIF4G2,
   EIF4B.
   22 tRNA ligases, RARS1, VARS1, YARS1, AARS1, IARS1, EPRS1, TARS2,
   QARS1, HARS1, MARS1, WARS2, FARSB, IARS2, MARS2, AARS2, TARS1,
   SARS1, CARS1, NARS1, GARS1, WARS1, FARSA.
   10 ribosome proteins: GCN1, RRBP1, BMS1, NIP7, BOP1, MRTO4, LRRC59,
   GCN1, TMA7B, EEF1D.
   5 DNA repair and 18 RAD50, RAD23B, TREX1, MSH6, ERCC6L, RAD18, ERCC5,
   XRCC6, MRE11,
   checkpoint response APEX1, MMS19, BUB3, SMC2, MDC1, TP53BP1, FANCI,
   BRAT1, TOP1.
   6 Cell migration 12 AGRIN, RAC1, MIG8, NUDC, MIG10, MYH9, RHOC, GIT1,
   ARHGAP34,
   ARHGAP17, ARHGDIA, MYL6B.
   [111]Open in a new tab

   The second highest group impacted mitochondrial function, with a total
   of 94 proteins including 50 metabolic enzymes, namely, NDUFV1, MTHFD2,
   PYCR1, MCAT, GPD2, GSR, PPA2, …etc. ([112]Table 1), and 44 proteins for
   mitochondrial tRNA metabolism and mitochondrial maintenance including
   VARS2, VARSL, MARS2, APEX1, HAP1, IARS2, WARS2, AFG3L2, MRM1, CARS2,
   REXO2, …etc. ([113]Table 1).

   The third group targets protein degradation, with a total of 55
   proteins that include 18 ubiquitin E3 ligases, TRIM25, ARIH1, ARIH2,
   RNF20, CBL, HECTD1, HUWE1, TRIM18, RAD18, RBBP6, RNF113A, RNF149,
   RNF220, RNF34, RNFT2, TRIP12, UBR1, UBR4, 5 E2 ubiquitin-conjugating
   enzymes, UBE2D2, UBE2D3, UBE2L3, UBE2N, UBE2Z, 4 deubiquitinases,
   USP47, USP5, USP7, USP9 and 28 proteasome components including PSMD4,
   PSMB2, PSMB6, PSMB4, PSMB1, ECPAS, PSMA6, PSMA1, PSMB7, PSMD2, PSMA3,
   PSMD7, PSMD13, PSMA2, PSMC4, PSMC6, PSMD8, PSMB5, …etc. ([114]Table 1).

   The fourth group impacts protein translation, with a total of 50
   proteins including 18 translation initiation factors including EIF3A,
   EIF6, EIF3G, EIF5A, EIF4G1, EIF5, …etc. ([115]Table 1), 22 tRNA ligases
   including RARS1, VARS1, YARS1, AARS1, IARS1, EPRS1, TARS2, QARS1, …etc.
   ([116]Table 1), and 10 ribosome proteins including GCN1, RRBP1, BMS1,
   NIP7, BOP1, MRTO4, LRRC59, GCN1, TMA7B, EEF1D. The fifth group contains
   18 proteins involved in DNA repair and the DNA damage checkpoint
   response (RAD50, RAD23B, TREX1, MSH6, ERCC6L, RAD18, ERCC5, XRCC6,
   MRE11, APEX1, MMS19, BUB3, SMC2, MDC1, TP53BP1, FANCI, BRAT1, TOP1) and
   the last group is involved in cell migration and contains 12 proteins
   (AGRIN, RAC1, MIG8, NUDC, MIG10, MYH9, RHOC, GIT1, ARHGAP34, ARHGAP17,
   ARHGDIA, MYL6B) ([117]Table 1).

   Although our findings suggest that SC-62-1 alkylates ~1500 proteins in
   cells, the effect of alkylation on individual proteins is difficult to
   assess without further study of their effect on downstream components
   of their involved pathways. We speculate that the effect on the
   function of individual proteins can vary significantly and not all
   these potential alkylation events have negative impact on the function
   of target proteins. For example, there were 28 proteasome components
   among the targets, the collective effect of their inactivation on the
   proteasome activity could be significant and result in severe defect in
   proteasomal degradation of ubiquitinated proteins such as MDM4.
   However, MMRi62 and SC-62-16 but not SC-16Me induced fast proteasomal
   degradation of MDM4 as we showed in this study ([118]Figure 1 and
   [119]Figure 4E). These results suggested that the proteasome function
   was not defective at least for the induced proteasomal degradation of
   MDM4, despite that many of the proteasome components were alkylated by
   MMRi62 and SC-62-16. This suggests that while this target profile of
   SC-62-1 provides a comprehensive list of potential drug targets, only a
   limited number of protein targets have a meaningful consequence upon
   alkylation and are responsible for the drug effect. It appears that
   SC-62-1-mediated alkylation does not occur by random contact with
   SH-group since simply mixing SC-62-1 and glutathione in deuterated
   DMSOd[6] and heating at 37 °C for 24 h did not produce any change in
   the ^1H NMR spectrum of SC-62-1, which may indicate that alkylation
   occurs primarily inside proteins, possibly facilitated by acid/base
   catalysis and “intramolecularity”.

   A genetic screen for enriched genes in cells surviving SC-62-1
   treatment might help sort out the targets that are responsible for the
   drug response [[120]37]. Nevertheless, we speculate that SC-62-1 must
   have inactivated multiple key proteins and touched multiple pathways
   that prevent cells from overcoming the collateral damage. Based on the
   phenotypical effect of cell killing and the target profile of SC-62-1,
   we propose a working model as a possible mechanism of action for
   SC-62-1, shown in [121]Figure 6. This model proposes that SC-62-1 acts
   as a multi-targeting alkylating agent. Through alkylating its target
   proteins, SC-62-1 not only induces proteasomal degradation of MDM4 and
   lysosomal degradation of FTH1, but also alters multiple pathways
   including RNA metabolism, mitochondrial function, protein translation
   and regulated protein degradation by proteasomes. The collective effect
   of its action creates overwhelming stressful state that cancer cells
   cannot overcome leading to p53-independent cell death in form of
   apoptosis in apoptosis-default cell types such leukemic cells or
   ferroptosis in ferroptosis-default cell types such as pancreatic and
   melanoma cells.

Figure 6.

   [122]Figure 6
   [123]Open in a new tab

   Proposed model for MOA of SC-62-1 in induction of cancer cell death.
   SC-62-1’s anticancer activity is via production of quinone methide
   which alkylates many target proteins. Its downstream effect includes
   induction of proteasomal degradation of MDM4 protein and lysosomal
   degradation of FTH1 and significant alteration of the pathways
   including RNA metabolism, carbon metabolism, mitochondrial function,
   protein translation and protein degradation. Collectively, induction of
   changes in these pathways by SC-62-1 creates overwhelming cellular
   stress that cannot be resolved, thus leading to p53-independent cell
   death.

   In summary, this study reports that quinolinol derivatives like MMRi62
   and SC-62-1 preferentially induce cancer cell death compared to
   specific MDM2 inhibitors and BRAF inhibitors. They induce proteasomal
   degradation of MDM4 [[124]19] and lysosomal degradation of FTH1
   [[125]20] but also target multiple other fundamental pathways for cell
   death induction. These compounds lack the specificity of conventional
   targeted therapies, a weakness measured by the concept of targeted
   therapies. However, pursuit of high specificity comes at the cost of
   short-lived efficacy and drug resistance, which is a well-established
   phenomenon for all targeted therapies. This limitation of targeted
   therapies lies in the capability of cancer cells easily bypassing the
   blockade of very specific drugs by mutating the well-defined interface
   or rewiring the affected signaling pathways. In fact, Mencher and Wang
   decades ago proposed that promiscuity of drugs can be a virtue compared
   to selective drugs [[126]38]. Promiscuity has also been explored to
   overcome drug resistance to HIV-1 protease inhibitors [[127]39]. In a
   recent study, 54 of 62 FDA-approved kinase inhibitors including Gleevec
   were reported to be promiscuous inhibitors with loose specificity of
   targeting at least two or more kinases. More interestingly, the study
   found that the more promiscuous type II inhibitors showed better
   pharmacodynamics than more specific type I inhibitors [[128]40].
   Therefore, it is not the specificity but the general toxicity and
   pharmacodynamics of a compound that determine its favorable therapeutic
   window. The toxicity of SC-62-1 is acceptable since its maximum
   tolerated dose in Scid mice is 60 mg/kg given every other day by
   intraperitoneal (I.P.) injection. We tested SC-62-1 for its in vivo
   efficacy in monotherapy in subcutaneous models of melanoma A375Luc in
   Scid mice at 57 mg/kg I.P. every other day and achieved ~50% reduction
   in tumor burden with body weight changes within 5%. However, due to
   small mouse number and large individual variation in tumor size, the
   results did not reach statistical significance. Future studies with a
   larger number of mice and development of derivatives with improved
   cancer cell selectivity will inform the potential of these quinolinol
   compounds as a new class of anticancer agents.

3. Materials and Methods

3.1. Representative Chemistry Methods

   (±)-2-Chloro-3-((8-hydroxyquinolin-7-yl)(pyridin-2-ylamino)methyl)
   benzonitrile (SC-62-1)

   To a 250 mL round bottomed flask equipped with a magnetic stir bar,
   2-chloro-3-cyanobenzaldehyde (1.00 g, 6.04 mmol) was added along with
   2-aminopyridine (568 mg, 6.04 mmol). 50 mL of absolute ethanol was
   added to the flask and the mixture was stirred until solids fully
   dissolved, at which point 8-hydroxyquinoline (1.05 g, 7.25 mmol) was
   added. The reaction flask was heated to 90 °C and refluxed for 24 h
   during which an off-white solid precipitated. The solution was allowed
   to cool to room temperature and to stand for several hours, then the
   resulting solid was isolated by filtration, giving SC-62-1 (1.75 g, 75%
   yield) as an off-white powder. Mp = 190–192 °C; ^1H NMR (300 MHz,
   CDCl[3]) δ 8.77 (d, J = 3.0 Hz, 1H), 8.13 (d, J = 7.2 Hz, 1H), 8.09 (d,
   J = 4.1 Hz, 1H), 7.97 (d, J = 7.9 Hz, 1H), 7.59 (d, J = 6.4 Hz, 1H),
   7.50–7.26 (m, 6H), 6.70 (d, J = 6.4 Hz, 1H), 6.66–6.57 (m, 1H), 6.36
   (d, J = 8.4 Hz, 1H), 5.55 (d, J = 6.2 Hz, 1H); ^13C NMR (75 MHz,
   CDCl[3]) δ 157.2, 149.8, 148.3, 141.5, 138.2, 137.7, 136.1, 136.0,
   133.0, 132.9, 127.9, 127.4, 127.0, 122.2, 120.6, 117.9, 116.2, 114.3,
   114.0, 107.1, 53.6; IR (neat film): 3346, 2923, 2233, 1600, 1571, 1517,
   1501 cm^−1; HRMS (ESI) calculated for C[22]H[16]ClN[4]O [M + H]^+:
   387.1016, found 387.1007.

   (±)-N-(2-azidoethyl)-2-(7-((2-chloro-3-cyanophenyl)(pyridin-2-ylamino)m
   ethyl)-8-hydroxyquinolin-5-yl)-N-methylacetamide (SC-62-16)

   (Note: The 3-component Betti reaction above did not work for this
   5-substituted-8-hydroxy quinoline, it is much less reactive than
   8-hydroxy quinoline). To a 10 mL reaction tube, the imine 3 (42 mg,
   0.18 mmol) was added, followed by the amide 2 (40 mg, 0.14 mmol, 0.8
   eq.) dissolved in PhCF[3] (0.5 mL). The squaramide catalyst 3 (8 mg,
   0.039 mmol, 0.1 eq.) was also added to the reaction tube. Flame-dried 4
   [MATH: <mrow><mrow><mtext> </mtext><mo>Å</mo></mrow></mrow> :MATH]
   molecular sieves (30 mg) were added, and the reaction tube was sealed
   and kept in a dark environment at 105 °C for 5 days. The reaction
   mixture was cooled to rt, diluted with CH[2]Cl[2], and filtered through
   Celite then concentrated to a crude mixture. The crude was then
   recrystallized from MeOH to yield 42 mg of SC-62-16 as light brown
   crystals in 40% yield. The NMR data indicates the presence of amide
   rotamers due to hindered rotation. Mp = 162–164 °C; ^1H NMR (400 MHz,
   CDCl[3]) δ 8.79 (dd, J = 4.2, 1.5 Hz, 1H), 8.36 (ddd, J = 14.3, 8.6,
   1.5 Hz, 1H), 8.08 (ddd, J = 5.0, 1.9, 0.8 Hz, 1H), 7.94 (dt, J = 7.9,
   2.5 Hz, 1H), 7.59 (dd, J = 7.7, 1.6 Hz, 1H), 7.49 (ddd, J = 8.6, 4.2,
   1.6 Hz, 1H), 7.45–7.30 (m, 2H), 7.25 (s, 1H), 6.69 (dd, J = 6.5, 2.4
   Hz, 1H), 6.61 (ddd, J = 7.2, 5.0, 0.9 Hz, 1H), 6.38 (d, J = 8.4 Hz,
   1H), 5.54 (d, J = 6.5 Hz, 1H), 4.08–3.91 (m, 2H), 3.51 (t, J = 5.4 Hz,
   2H), 3.47–3.33 (m, 3H), 3.07 (s, 2H), 2.92 (s, 1H); ^13C NMR (101 MHz,
   CDCl[3]) δ 170.5, 157.8, 149.3, 148.0, 147.8, 142.9, 138.5, 136.8,
   135.4, 133.7, 133.4, 132.8, 127.9, 127.5, 122.5, 121.7, 120.8, 115.9,
   113.8, 113.0, 108.8, 51.8, 51.7, 49.2, 48.7, 48.5, 47.0, 37.2, 36.7,
   35.8, 32.5; HRMS (ESI) calculated for C[27]H[24]ClN[8]O[2] [M + H]^+:
   527.1711, found 527.1704.

   The full synthesis and characterization details of all new compounds
   are provided in the [129]Supplemental Information (SI).

3.2. Biological Assays and Methods

3.2.1. Cell Culture

   Melanoma cell line A375 was cultured in DMEM-10% fetal bovine serum, 50
   U/mL penicillin and 50 μg/mL streptomycin. Cell line with stable p53
   knockdown shp53-A375 was established using pLKO.1-p53 (purchased from
   Addgene) (Plasmid #19119) [[130]41] followed by puromycin selection at
   1 μg/mL for 2 days, then clonal expansion in puromycin-free medium.

3.2.2. Western Blotting Analysis and Crystal Violet Staining

   The Western blotting procedure and the relevant antibodies was
   described previously [[131]19]. For crystal violet staining, cell
   plates were fixed with 4% (w/v) paraformaldehyde for 15 min followed by
   30 min of staining in 0.1% (w/v) crystal violet solution in 20% (v/v)
   Ethanol-80% (v/v) PBS. After rinsing with tap water, the plates were
   airdried and the images were taken on Olympus IX73 inverted
   fluorescence microscope at a magnitude of 20×. The images of whole
   plates were taken on iPhone.

3.2.3. IC[50] Measurement and Analysis

   The procedure for measuring IC[50] was described previously [[132]19].
   Some data were obtained by Chou-Median-Effect Equation using
   CompuSynversion 1.0 (CompuSyn.exe) software [[133]42] and some were
   obtained by GraphPad (Prism 8) using affected fractions of
   compound-treated wells normalized against no-drug control wells with
   non-linear regression model.

3.2.4. Affinity Pulldown and Identification of SC-62-16/SC-62-16Me Covalently
Bound Proteins

   The procedure uses 2% (w/v) SDS-PBS lysis buffer to lyse the treated
   cells followed by treatment with 10 mM DTT, at 70 °C for 15 min to
   dissociate all non-covalent protein–protein interaction and reduce all
   the disulfide bonds among peptides. Then, the proteins were
   precipitated with 100% (v/v) methanol and 20% (v/v) chloroform, then
   were re-dissolved in 2% (w/v) SDS-PBS buffer by sonication. Before the
   “click” reaction, the samples were treated with iodoacetamide to
   alkylate sulfhydryl groups from further forming disulfide bonds. After
   “click” reaction with Biotin-PEG4-alkyne, the biotin-tagged
   compound-bound proteins were pulled down with magnetic Streptavidin
   beads. The bead-bound proteins were then eluted by 25 mM biotin heated
   at 95 °C for 5 min.

   The click reaction conditions to link Biotin-PEG4-alkyne to SC-62-16
   and SC62-16Me azide were adapted from a report by A. J. Pradhan et al.
   [[134]43]. The conditions for biotin-streptavidin lysate preparation
   and biotin-streptavidin affinity purification and elution of
   compound-bound proteins were adapted from the method reported by J. S.
   Cheah et al. [[135]44]. Biotin-PEG4-alkyne was purchased from
   Sigma-Aldrich (St. Louis, MO, USA) (cat# 764213-5MG) and PuroMAG™
   Magnetic Beads-Streptavidin were purchased from Luna Nanotech (Markham,
   ON, Canada, Cat# MGB-STRP-10). Briefly, A375 cells (20 × 10 cm plates
   at 90% confluency) were treated with 10 μM SC-62-16 or SC-62-16Me for 4
   h. After collecting the cells in a 50 mL centrifuge tube, they were
   washed with PBS and resuspended in 25 mL PBS and added 20% (w/v) SDS to
   final 2% (w/v) SDS-PBS followed by 20 cycles of sonication
   (30”-on-30”-off at 30% intensity of output) (Fisher Scientific Sonic
   Dismembrator Model 100 from Fisher Scientific, Waltham, MA, USA). Then,
   the cell lysate was treated with 10 mM DTT at 70 °C for 15 min. After
   centrifugation of the lysate at 50,000× g at 20 °C for 2 h, the soluble
   proteins were precipitated with 1× volume of methanol and 1/5× volume
   of chloroform with 5 min vigorous vortexing and 10 min centrifugation.
   After re-dissolving the proteins in 2% (w/v) SDS-PBS aided with
   sonication, the soluble protein solution was treated with 20 mM
   iodoacetamide for 30 min in darkness. Then, the proteins were
   precipitated again with methanol-chloroform and washed with cold
   methanol and resuspended in 4% (w/v) SDS-PBS. The 10× click reaction
   stock reagents were then added, to make a final concentration of 1 mM
   CuSO[4], 1 mM TCEP, 0.25 mM TBTA and 0.25 mM Biotin-PEG4-alkyne. After
   a brief vortex, the mixture was incubated at 37 °C for 1 h. Then, the
   proteins were precipitated with methanol-chloroform followed by two
   washes with methanol with sonication. The proteins were re-dissolved in
   3 mL of biotin-streptavidin lysis buffer (50 mM Tris HCl pH 7.4, 150 mM
   NaCl, 0.4% (w/v) SDS, 1% (v/v) NP, 1 mM EGTA and 1.5 mM MgCl[2]) and
   underwent 10 cycles of 30 secs-on-30 secs-off sonication in Sonic
   Dismembrator followed by centrifugation at maximal speed in a
   microcentrifuge to remove insoluble proteins. Then, the biotin-labeled
   proteins were pulled down with PuroMAG™ Magnetic Beads-Streptavidin at
   100 μL of flurry beads/1 mL protein solution, rotate the sample tubes
   overnight at 4 °C. Then, the beads were washed twice with 1 mL of
   biotin-streptavidin lysis buffer with vigorous vortexing for 5 min
   followed by two washes with 2% SDS-PBS and then two washes with
   biotin-streptavidin lysis buffer. The proteins were eluted with two
   elution with 60 μL of 20 mM biotin in 50 mM Tris pH 8.3 and the
   identity of proteins were determined by high-quality and robust protein
   quantification by IonStar/nano-LC/MS at proteomics core facility run by
   Dr. Jun Qu [[136]45] at the New York State Center of Excellence
   Bioinformatics and Life Sciences (CBLS), University at Buffalo.

3.2.5. Pathway Analysis of SC-62-16/SC-62-16Me Covalently Bound Protein

   Genes encoding the proteins identified by mass spectrometry in SC-62-16
   samples were used to run pathway enrichment analysis using g:Profiler
   [[137]46] and functional annotation using DAVID [[138]47,[139]48]. For
   the pathway enrichment runs, KEGG [[140]35] and Reactome database
   [[141]36] were used for the pathway search. The pathways with
   Benjamini–Hochberg adjusted p-value < 0.05 were ranked by their p
   values with the lowest as the top hit.

Acknowledgments