Nano-delivery of RAD6/Translesion Synthesis Inhibitor SMI#9 for Triple-negative Breast Cancer Therapy

Triple-negative breast cancers (TNBC) account for approximately 15% of all the breast cancers and because they lack estrogen receptor (ER), progesterone receptor (PGR), and Her2/neu amplification, they are not treatable with therapies targeting these molecules. TNBCs present with high histologic grade and patients with TNBC have poorer outcomes compared with other breast subtypes (1). The triple-negative phenotype is most commonly observed in patients with BRCA1/BRCA2 mutations (2). BRCA1-associated and sporadic TNBCs share features such as high-grade cytokeratin expression, p53 mutation, and aberrant DNA repair pathways (3). Because clinicopathologic features of TNBCs overlap with BRCA-related breast cancers, BRCA1 is considered an important player in TNBC biology. BRCA1 is required for homologous recombination–mediated DNA repair. BRCA1-deficient cells are unable to repair lesions and resort to error-prone DNA repair that increase the risk of cancer progression and chemoresistance. Although patients with TNBC initially respond better to chemotherapy, patients with TNBC, regardless of their BRCA1 status, have decreased progression-free and overall survival rates (4). Thus, strategies that target TNBCs with and without BRCA1 mutation constitute an important component of TNBC treatment.

RAD6 is a fundamental component of the translesion synthesis (TLS) DNA repair pathway. It is critical for the continuation of replication on damaged DNA templates, thereby preventing replication fork stalling and resultant cell-cycle arrest and ultimately cell death (5). Because the RAD6 pathway is essential for cell survival in the face of a variety of genotoxic insults, it plays an important role in DNA damage tolerance (5, 6). However, because the RAD6-mediated TLS process involves low-fidelity TLS polymerases, it is potentially an error-prone process. Thus, its usage needs to be tightly regulated because it is potentially mutagenic. The E2 ubiquitin–conjugating (UBC) activity of RAD6 is essential for TLS (7). We have previously reported that RAD6B, a.k.a. UBE2B or HHR6B, is weakly expressed in normal breast cells and overexpressed in metastatic and chemoresistant breast carcinomas (8–10). Constitutive RAD6B overexpression in normal human breast epithelial cells induces tumorigenesis and chemoresistance (8, 11, 12), whereas RAD6B depletion compromises TLS, rendering cells chemosensitive (5). We have reported the development of a RAD6-selective small-molecule inhibitor SMI#9 that inhibits UBC activity of RAD6 (13). SMI#9 treatment suppresses proliferation and migration of breast cancer cells while sparing normal breast cells (13). We recently demonstrated that intratumorally administered SMI#9 restores cisplatin sensitivity and inhibits growth of cisplatin-resistant TNBC cells. We showed that this inhibition results from SMI#9 inhibition of TLS-mediated repair of cisplatin-induced DNA damage (14).

SMI#9 has poor aqueous solubility that limits its therapeutic efficacy. To improve SMI#9 solubility and uptake, we previously developed SMI#9-tethered gold nanoparticles (GNP) using a chemistry that allows its intracellular release, and showed that the SMI#9 released from the GNP conjugate behaves similarly to the parent-free drug and inhibits TNBC cell proliferation without affecting normal cells (15). GNPs have been widely used in cancer diagnostics (16–18) and targeted delivery (19, 20) because of their versatile surface chemistry and biological inertness. In this study, we analyzed the pharmacokinetic, biodistribution, and therapeutic properties of SMI#9-GNP on BRCA1 wild-type and BRCA1-mutant TNBC growth. Pharmacokinetic analysis showed that SMI#9 is slowly released from the GNP conjugate allowing for a more stable build-up of the drug as compared with unconjugated SMI#9. SMI#9-GNP was well tolerated without overt signs of toxicity. Systemically administered SMI#9-GNP inhibited growth of both BRCA1 wild-type and mutant TNBCs as effectively as intratumorally injected unconjugated SMI#9. Quantitative GNP biodistribution analysis showed that despite highest accumulation of GNPs in the tumors and livers of blank-GNP and SMI#9-GNP–treated mice, tumor inhibition occurred only in SMI#9-GNP–treated mice. Metabolic transformations typified by switch to aerobic glycolysis and profound mitochondrial reprogramming are widely considered to support the cancer phenotype, and TNBCs show increased glycolysis compared with other breast cancer subtypes (21, 22). RAD6 is implicated in control of mitochondrial turnover (23). To determine whether RAD6 inhibition impacts TNBC metabolism, we performed ¹H-NMR–based metabolomics analysis. Our results showed that SMI#9-GNP selectively decreased pathways regulating lactate, glutamate, and glycine metabolism in both BRCA1 wild-type and mutant TNBCs. These data reveal novel metabolic vulnerability impacted by RAD6 inhibition, further strengthening the therapeutic potential of targeting RAD6 for treatment of TNBCs.

Materials are described under Supplementary Data.

SMI#9 was synthesized as described previously (13). SMI#9 was chemically modified to enable formation of a biodegradable ester bond between the drug and mercaptosuccinic acid (MSA)-capped GNP carrier (15). Synthesis of MSA-capped GNP (blank-GNP) and MSA-capped GNP carrying SMI#9 (SMI#9-GNP) followed the same procedure as described previously (15). Thermogravimetric analysis (TGA) was performed to determine the amount of SMI#9 in the nanoconjugate. To improve the biocompatibility and colloidal stability, thiolated PEG (Molecular weight 3,000) was added to MSA-capped GNP or SMI#9-GNP. The nanoconjugates were characterized by atomic force microscopy (AFM), transmission electron microscopy (TEM), UV-vis, and dynamic light scattering (DLS) as described previously (15).

MDA-MB-231 (BRCA1 wild-type) and MDA-MB-468 (BRCA1 wild-type) human TNBC cells were purchased from ATCC, and SUM1315 (BRCA1 mutant) TNBC cells were purchased from Asterand. All lines were maintained in DMEM/F12 (Invitrogen) containing 5% FBS. Following authentication by cell bank (short tandem repeat PCR) and Mycoplasma screening (MycoAlert, Lonza), several aliquots of cells were frozen and used within 10 passages.

MDA-MB-231 or SUM1315 cells were treated overnight with SMI#9-GNP (5 μmol/L SMI#9 equivalent dose) or the equivalent amount of blank-GNP. Cells were trypsinized and reseeded at 150 cells/well in 24-well plates. Colony formation was assessed with crystal violet staining and colony-forming efficiency was expressed relative to blank GNP–treated cells.

TNBC cell responses to SMI#9-GNP and PEGylated SMI#9-GNP were compared by MTT assay as described previously (15). SUM1315 (5 × 10³) cells seeded in triplicates in 96-well plates were treated for 72 hours with 0–8 μmol/L of SMI#9-GNP, PEGylated SMI#9-GNP, or blank GNP. Results are expressed relative to control from three independent experiments.

MDA-MB-231 or SUM1315 (1 × 10⁴) cells seeded on coverslips were treated with SMI#9-GNP (5 μmol/L SMI#9 equivalent dose) or blank GNP in triplicates for 24 hours. Cultures were stained with acridine orange/ethidium bromide (each 25 μg/mL) and quantitated as described previously (13).

MDA-MB-231 or SUM1315 (1.2 × 10⁶) cells seeded in 60-mm dishes were treated overnight with 0.5–5 μmol/L SMI#9-GNP, equivalent amounts of blank-GNP, or left untreated. Details of metabolite extraction are described under Supplementary Data. ¹H-NMR was conducted using Agilent DD2-600 MHz. Metabolite peaks within 0–10 ppm spectral width were captured and processed simultaneously using ACD/Spec Manager 7.00 (Advanced Chemistry Development, Inc.) to avoid variations during processing. The spectra were phased, base line corrected, and digitized into 1,000 bins by intelligent binning.

The table of integrals was imported into SIMCA P+ software (Umetrics Academy) and multivariate data analysis (MVDA) performed. Pareto scaling was performed and spectral regions corresponding to peaks from DSS, imidazole, water, and exchangeable protons were removed before MVDA. Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) score plots were generated to reveal metabolites perturbed by SMI#9-GNP. CHENOMX NMR suite (CHENOMX Inc.) was used to identify and quantify the metabolites using DSS as the reference. Metabolite concentrations (μmol/L) were normalized to protein content and expressed from three independent experiments.

Metabolic pathways perturbed by SMI#9-GNP in the TNBC cells were identified by MetaboAnalyst 4.0 software, which is based on KEGG (http://www.genome.jp/kegg) and the Human Metabolome (http://www.hmdb.ca/) databases (24).

Female Ncr nude mice were intravenously injected with a single dose of SMI#9 (5 mg/kg body weight in PBS/2.5% DMSO) or PEGylated SMI#9-GNP (SMI#9 equivalent dose 5 mg/kg body weight in PBS). Blood samples were collected prior to injection and at defined time points postinjection, and SMI#9 concentrations in plasma were determined by LC/MS-MS (15).

A total of 1 × 10⁶ MDA-MB-468 or SUM1315 cells were suspended in 100 μL of Matrigel and injected into the bilateral inguinal mammary gland fat pads of female Ncr nude mice. When the tumors reached approximately 150 mm³, mice were randomly assigned to vehicle, SMI#9, PEGylated SMI#9-GNP, or blank GNP groups (n = 6 mice/group) and treated twice weekly with intratumoral injections of SMI#9 (1.5 mg/kg body weight in PBS/0.5% DMSO) or vehicle, or intraperitoneal injections of PEGylated SMI#9-GNP (SMI#9 equivalent dose of 0.85 mg/kg body weight in PBS) or blank GNP. Bidimensional tumor size and body weights were measured twice weekly. Mice were sacrificed at 37 or 44 days postimplantation of SUM1315 or MDA-MB-468 cells, respectively, 3 days after the last drug administration. Tumors were harvested, and the mean tumor burden mass for each group was determined. Animals were evaluated for gross abnormalities of organs by necropsy. Excised tumors and vital organs were fixed in buffered formalin and paraffin-embedded for histologic analysis. Portions of tumors were snap-frozen for LC/MS-MS analysis of SMI#9. Organs and tumor fragments were processed for inductively coupled mass spectrometry analysis of gold. The in vivo studies were conducted in accordance with the Institutional Animal Care and Use Committee (IACUC) guidelines of Wayne State University (Detroit, MI).

SMI#9 concentrations in the plasma and tumors were determined by LC/MS-MS as described previously (15). Briefly, tumors (50–100 mg) were homogenized, and plasma or tissue homogenates were extracted with ethyl acetate. The organic phase was dried down and reconstituted in the mobile phase [methanol/0.45% formic acid in water (60:30, v/v)]. LC/MS-MS was performed on a Waters Xevo TQ-XS LC/MS-MS system with a Waters Xterra MS C18 column and isocratic elution with the mobile phase. SMI#9 was monitored at the positive electrospray ionization mode at m/z 366.69 > 150.1 mass transitions (15).

Organs and tumors were digested with aqua regia (hydrochloric acid: nitric acid, 3:1 v/v) and analyzed using an Agilent 7700× inductively coupled plasma mass spectrometry (ICP-MS) equipped with an Agilent ASX-500 series autosampler. External calibration using standards with internal standard correction of bismuth (Bi) was performed prior to the measurements. GNP biodistribution data were expressed as a percentage of injected dose of gold/gram dry tissue.

Results are presented as the mean ± SD or SEM, and were analyzed by two-tailed Student t test and one-way ANOVA. Statistical significance was accepted at a value of P < 0.05.

GNPs with a 5-nm core diameter were functionalized with modified SMI#9 prior to PEGylation (15). GNPs were characterized by AFM, TEM, and UV-vis as described previously (15) and the results are summarized in Table S1 in Supplementary data. The resulting SMI#9-GNP conjugate had a hydrodynamic size of 41 nm, and approximately 26% of the MSA-GNP reactive sites were occupied by SMI#9.

We have previously reported that SMI#9 inhibits TNBC colony formation (13). To determine whether SMI#9-GNP similarly affects clonogenic potentials of TNBC cells, MDA-MB-231 or SUM1315 TNBC cells were treated overnight with 5 μmol/L SMI#9-GNP or blank-GNP. SMI#9-GNP treatment significantly decreased the colony-forming efficiencies of both lines compared with controls (P <.0.01; Fig. 1A and B). SMI#9-GNP–induced apoptosis was detected by differential uptake of acridine orange and ethidium bromide. Early apoptosis marked by intercalated acridine orange within fragmented DNA (25) and late-stage apoptosis marked by apoptotic body separation and reddish-orange color due to acridine orange binding to fragmented DNA (26) were observed by 24 hours in >85% of TNBC cells treated with SMI#9-GNP (Fig. 1C and D, long arrow). GNP presence in the apoptotic cells is indicated by a short arrow in Fig. 1C and D. Blank GNP–treated cells were minimally affected as >98% of the cells showed a green intact nuclear structure (Fig. 1C and D), thus verifying the nontoxicity of GNPs and the anticancer activity of the conjugated SMI#9.

Because SMI#9-GNP treatment compromises mitochondrial membrane potential (15), we performed ¹H-NMR spectroscopy to determine whether SMI#9-GNP induces metabolic changes. Fourier-transformed spectra with the major metabolite peaks identified from MDA-MB-231 and SUM1315 cells are shown in Fig. 1E and G. PCA plots (Fig. 1F and H) and the corresponding loading plots for SUM1315 and MDA-MB-231 (Supplementary Fig. S1A and S1B, respectively) show the regions responsible for spectra separation between blank GNP and SMI#9-GNP groups (Fig. 1F and H). Consistent with the differences in responses to blank GNP versus SMI#9-GNP, PLS-DA score plots showed clear spectral separations between blank GNP and SMI#9-GNP–treated cells (Supplementary Data, Fig. S2A and S2B). Spectral regions corresponding to lactate, glutamate, and glycine (quantified and confirmed by CHENOMX) were similarly affected by SMI#9-GNP treatment in SUM1315 (Fig. 2A) and MDA-MB-231 (Fig. 2B) cells. Phosphocholine, Sn-glycero-3-phosphocholine, proline, creatinine, and alloisoleucine were also decreased by SMI#9-GNP in SUM1315 cells (Fig. 2C) but not in MDA-MB-231 cells (Fig. 2D). To identify the pathways perturbed by SMI#9, the metabolic profiles were analyzed using MetaboAnalyst 4.0. Pathways regulating glycine, pyruvate, and alanine/aspartate/glutamate metabolism were found to be most impacted by SMI#9-GNP in both TNBC models. A summary of pathway analysis is shown in Fig. 3A and C, and the tabulated results are shown in Fig. 3B. Results for the individual cell models are shown in Supplementary Tables S2 and S3 (Supplementary Data). These data show that SMI#9-GNP perturbs common metabolic pathways in BRCA1 wild-type and BRCA1-mutant TNBC cells.

SMI#9 concentrations in plasma were measured by LC/MS-MS at 5 minute to 24 hours after a single intravenous injection of SMI#9 or PEGylated SMI#9-GNP. Following injection of the unconjugated drug, SMI#9 achieved maximum plasma concentration (C_max) of approximately 74 nmol/L at 5 minutes, and was rapidly cleared by 30 minutes (Fig. 4A). In contrast, following injection of PEGylated SMI#9-GNP, SMI#9 was slowly released and achieved C_max of approximately 1,200 nmol/L at 8 hours (Fig. 4B). These data confirm that SMI#9 is released from the PEGylated SMI#9-GNP conjugates and its slow release helps achieve higher systemic exposure as compared with the unconjugated SMI#9.

Postsynthesis surface modification of SMI#9-GNP with PEG was performed as PEG improves colloidal stability and prevents aggregation (27, 28). Furthermore, PEG reduces adsorption of cellular proteins and increases the circulation time of nanoparticles. We first determined whether PEGylated SMI#9-GNP similarly sensitizes TNBC cells in vitro as non-PEGylated SMI#9-GNP. SUM1315 cells were treated with various concentrations of SMI#9-GNP or PEGylated SMI#9-GNP (or blank-GNP) and the responses measured by MTT assays. SUM1315 cells responded similarly to both formulations with GI₅₀ values (based upon SMI#9 concentration) of approximately 0.75–1 μmol/L, verifying that tethering of PEG did not significantly affect the activity of the SMI#9 payload (Fig. 4C). To compare the in vivo effects of SMI#9 and PEGylated SMI#9-GNP on TNBC xenografts, MDA-MB-468 or SUM1315 TNBC cells were orthotopically implanted in female nude mice. When the tumors reached approximately 150 mm³, mice were randomly assigned for twice weekly treatments with vehicle, SMI#9, PEGylated SMI#9-GNP, or blank GNP. To facilitate easy tumor access, unconjugated SMI#9 (or vehicle) was administered intratumorally. To assess tumor accessibility and bioavailability of conjugated SMI#9, PEGylated SMI#9-GNP (or blank GNP) was injected intraperitoneally. Vehicle, blank GNP, and unconjugated SMI#9-treated groups initially tracked similar growth, however, continued treatment with unconjugated SMI#9 significantly retarded MDA-MB-468 and SUM1315 tumor growth (Fig. 4D and F; P < 0.0001). PEGylated SMI#9-GNP treatment significantly retarded growth of MDA-MB-468 (P < 0.0001; Fig. 4D) and SUM1315 (P = 0.0038; Fig. 4F) tumors as compared with blank GNP, and was slightly more efficacious than unconjugated SMI#9 (P < 0.01). Comparison of the excised tumor masses upon sacrifice confirmed that both SMI#9 and PEGylated SMI#9-GNP treatments significantly inhibited MDA-MB-468 and SUM1315 tumor growth as compared with the corresponding controls (P ≤ 0.05; Fig. 4E and G).

To evaluate the effects of unconjugated and GNP-conjugated SMI#9 on tumor morphology, the tumors from control and treated groups were analyzed by hematoxylin and eosin (H&E) staining. Compared with the robust tumors in control mice, SMI#9 and PEGylated SMI#9-GNP–treated MDA-MB-468 (Fig. 5A) and SUM1315 (Fig. 5B) tumors were sparsely populated and contained several apoptotic cells marked by nuclear pyknosis and apoptotic bodies (arrows in insets of Fig. 5A and B) as scored by H&E diagnostic criteria for apoptosis (29). These data are consistent with the data in Fig. 1C and D that showed robust apoptosis induction by SMI#9-GNP. To determine whether PEGylated SMI#9-GNP treatment affected vital organs, H&E-stained tissues were scored for organ damage. Tissues harvested from untreated nontumor-bearing normal mice were used as reference for detecting GNP-induced alterations. No differences in tissue color and appearance were observed after GNP treatments. H&E-stained tissues showed normal morphologies for lung, liver, spleen, kidney, heart, and small intestines in blank-GNP and SMI#9-GNP–treated mice with no evidence of atrophy or inflammation, and were indistinguishable from normal control mice (Fig. 5C). Mice treated with unconjugated SMI#9, PEGylated SMI#9-GNP, or blank-GNP show similar body weights as the vehicle control group (Fig. 5D). Taken together, these data confirm that blank- or SMI#9-GNP treatments are tolerated well with negligible histopathologic changes and organ toxicity.

To verify whether the tumor growth inhibition in SMI#9-GNP–treated mice resulted from SMI#9 release from the nanoconjugate, we quantitated SMI#9 levels in tumor fragments collected at sacrifice by LC/MS-MS. Compared with the control groups, measurable amounts of SMI#9 were detected in SUM1315 and MDA-MB-468 tumors treated with SMI#9 or SMI#9-GNP (Fig. 6A and B). However, there was variability in the detected levels, which could have potentially resulted from tumor tissue samplings as tissue fragments rather than whole tumors were used for analysis or from loss of SMI#9 resulting from therapy-induced tumor cell death in the responsive tumors.

Next, we determined the biodistribution and retention of GNPs in tumor and organ tissues collected at sacrifice (Fig. 6C) by ICP-MS. The injected dosage of gold was calculated on the basis of the TGA data: weight ratio of gold to SMI#9 equivalent to 1:0.083 in the nanoconjugate (15). Thus, a SMI#9 dosage of 0.85 mg/kg body weight corresponds to gold dosage of 10.24 mg. On the basis of 0.02 kg average weight of mouse, the gold dosage per injection is 0.2 mg and the total gold injected is 1.8 mg after nine injections. Among the tissues analyzed, liver and tumors showed the highest accumulation of gold ranging from 0.19% to 0.4%; however, there were no significant differences in gold content between livers and tumors of blank GNP and SMI#9-GNP groups (Fig. 6C). Despite multiple injections, this accumulation was comparable with those reported for a single injection (30). GNP bioaccumulation in liver and spleen are regulated by the reticuloendothelial system (RES; ref. 31). The low levels of GNP accumulation in the spleen potentially reflect the low GNP doses administered in our study and/or the time intervals between the injections. Furthermore, because the tissue samples were collected 3 days after the last GNP injection, larger increases in gold depositions were perhaps not observed because of active renal clearance. The GNPs used here are 5-nm particles with a hydrodynamic size of 41 nm following SMI#9 conjugation. Because the kidney filtration threshold is 6–8 nm (32), GNPs that were stripped off their SMI#9 cargo would be small enough to be cleared by the kidneys leaving behind the nonbiodegraded SMI#9-GNPs. Because tumors contained higher levels of gold, it is likely that enhanced permeability and retention (EPR) effects from leaky vasculature in the tumors contributed to accumulation of gold in the tumors (ref. 33; Fig. 6C).

In this article, we evaluated the toxicity, biodistribution, and therapeutic activity of systemically administered PEGylated SMI#9-conjugated GNPs to assess their utility as a drug delivery platform for treatment of BRCA1 wild-type and BRCA1-mutant TNBCs. We demonstrate that the tethered SMI#9 prodrug is released from the GNP conjugate and inhibits growth of both BRCA1 wild-type and mutant TNBCs. We also demonstrate that PEGylated SMI#9-GNP inhibits TNBC growth as effectively as intratumorally administered unconjugated SMI#9. Intratumoral injection of SMI#9, a clinically unsuitable mode of drug administration, was necessary because of its poor solubility and poor pharmacokinetic properties. Delivery of SMI#9 as a prodrug–GNP conjugate not only greatly improved its circulation stability and drug exposure, but also allowed for a clinically acceptable mode, that is, systemic administration of the drug because of improved solubility.

Surface functionalization of GNP with PEG is commonly used to prevent rapid clearance by the RES. PEGylation prevents nonspecific GNP interactions with serum proteins improving GNP circulation time (27, 28). Solid tumors are characterized by defective vasculature and impaired lymphatic drainage/recovery system that promote EPR effect (33, 34). Consequently, tumor accumulation of GNPs increases with increased blood circulation time and a leaky vasculature. MDA-MB-468 and SUM1315 TNBC growth is strongly inhibited by systemically administered PEGylated SMI#9-GNP compared with blank-GNP. This is supported by our data from ICP-MS analysis that showed high accumulation of gold in the tumor tissues, and further corroborated by LC/MS-MS analysis that showed SMI#9 in the residual tumors of SMI#9-GNP–treated mice. SMI#9 presence in the tumors establishes SMI#9-GNP tumor uptake, lysosomal processing of the prodrug (SMI#9)–GNP conjugate, and the hydrolytic release of the tethered SMI#9 (15). Our data confirm that SMI#9 released from the PEGylated–GNP conjugate is therapeutically active as it potently inhibits tumor growth with similar efficacy as the intratumorally injected unconjugated SMI#9.

Perrault and colleagues (35) tested GNPs with size range of 10–100 nm and found that although GNPs with hydrodynamic diameter of 60–100 nm with PEGylation efficiently utilize EPR effect for enhanced tumor accumulation, these larger GNPs penetrated weakly into the tumors and localized in the perivascular regions. GNPs are generally considered nontoxic when the size is larger than 2 nm, whereas ultrasmall nanoparticles are found to be toxic (36). GNPs of larger size (35–50 nm) were found to enter cells more efficiently without toxicity than small nanoparticles (1.4 nm; refs. 37–39). Consistent with these data, our results show that 41-nm SMI#9-GNPs are taken up efficiently by tumor cells and display no apparent toxicity. Quantitative analysis of GNP biodistribution by ICP-MS analysis showed higher accumulation of gold in tumors and liver as compared with other organs of SMI#9-GNP and blank GNP–treated mice. These data are in agreement with other studies that showed maximal accumulation of nanoparticles in the liver regardless of their size, shape, dosage, or composition (31, 40, 41). However, our data suggest that the GNPs are actively cleared because the amounts of gold deposited in the liver and other tissues despite repeated injections resemble those in mice receiving a single dose. Because the GNPs used in our study have a 5-nm core diameter and a hydrodynamic size of 41 nm for the SMI#9 nanoconjugate, our data suggest that the majority of GNPs are stripped off their SMI#9 payload rendering them small enough for renal clearance (32). These data are consistent with those reported by Perrault and colleagues who showed rapid clearance of PEGylated 20-nm and 40-nm GNPs from circulation without accumulation in the spleen and liver (35). Regardless of the coating layer, nanoparticles always form a “protein corona” when encountering physiologic conditions (42, 43). This corona phenomenon potentially balances out any differences between blank GNP and SMI#9-conjugated GNP during circulation and may explain the similar patterns of biodistribution data. It is noteworthy that despite high accumulation of gold in the tumors of blank GNP–treated mice, tumor growth was unaffected in these mice providing further support for SMI#9 therapeutic activity and the nontoxicity of GNPs. Although tumor growth was robustly inhibited by SMI#9-GNP, the treatment did not affect body weights or induce overt toxicity as determined by necropsy and histopathologic analysis of the major vital organs.

TNBC cells are characterized by high glycolytic flux and decreased mitochondrial respiration (44), and are primed to switch to a glycolytic program regardless of the oxygen status unlike nontransformed cells (44, 45). A siRNA screen showed that TNBC cells are dependent on lactate dehydrogenase activity for elevated glycolysis as compared with non-TNBC cells (46). Using ¹H-NMR–based spectroscopy, we analyzed the impact of RAD6 inhibition on metabolic profiles of BRCA1 wild-type and BRCA1-mutant TNBC cells. Using unsupervised analysis, four discriminating metabolites (lactate, glycine, alanine, and glutamate) were found to be perturbed by SMI#9-GNP in both BRCA1 wild-type and BRCA1-mutant TNBC cells. It is likely that a greater number of metabolites would have been identified by mass spectrometry because of the sensitivity limitations of NMR spectroscopy. It is noteworthy, however, that against the backdrop of myriads of metabolic changes, NMR spectroscopy helped identify quantitative and reproducible changes in key metabolic determinants perturbed by the RAD6 inhibitor. TNBC cells are characterized by high levels of lactate, glutamate, and glycine and have been correlated with high glycolytic activity and tumor aggressiveness (47). Glycine is involved in the synthesis of proteins, nucleotides, and one-carbon metabolism, and high levels of glycine have been shown to correlate with poor prognosis in breast cancer (48). Compared with ER/PGR-positive tumors, TNBCs contain lower levels of glutamine and higher levels of glutamate, which might result from increased glutaminolysis (49). Aberrant phosphatidyl-choline metabolism associated with increases in phosphocholine has been reported in breast cancer (50). However, although SMI#9-GNP treatment inhibited tumor growth in both MDA-MB-468 and SUM1315 TNBC models, it decreased phospholipid levels only in SUM1315 cells. These data suggest that the SMI#9-GNP–inhibitory effects may not be universally linked with phosphatidyl-choline metabolism.

The UBC activity of RAD6, also known as UBE2A, is implicated in maintaining healthy mitochondria. RAD6, in combination with Parkin E3 ubiquitin ligase, regulates mitochondrial protein ubiquitination to facilitate autophagic clearance of dysfunctional mitochondria (23). Consistent with these data, RAD6-deficient human and Drosophila cells show defective mitochondria turnover (23). We have shown previously that SMI#9 in both free and GNP-conjugated forms induces accumulation of dysfunctional mitochondria corroborating the role of Rad6 in mitochondrial function (15). Interestingly, loss of mitochondrial function in SMI#9-treated TNBC cells did not result in concomitant increase in metabolites associated with Warburg effect but rather caused reductions in glycolytic mediators. These data suggest that the increase in glycolytic flux seen in TNBC cells may not be directly related to mitochondrial function. Because RAD6 inhibition can concurrently perturb mitochondrial function and glycolytic potential besides its traditional role in TLS, our data suggest that targeting Rad6 could offer a promising approach for TNBC treatment.

In summary, we demonstrated the therapeutic utility of a GNP-based platform for delivering Rad6 inhibitor for treating BRCA1 wild-type and BRCA1-mutant TNBCs, a breast cancer subtype with poor prognosis and no targeted therapies. SMI#9 delivery as a prodrug–GNP conjugate not only improved its pharmacokinetic properties but also allowed for a clinically acceptable (systemic) mode of drug administration that inhibited TNBC growth. There was no evidence of blank or SMI#9-GNP–induced toxicity as measured by their impact on animal body weight and organ histopathology. SMI#9-GNP–induced sensitization of BRCA1 wild-type and BRCA1-mutant TNBCs was associated with perturbation of a core set of metabolic pathways revealing markers of metabolic vulnerabilities that can be targeted for TNBC management.

No potential conflicts of interest were disclosed.

Conception and design: N. Saadat, F. Liu, L.A. Polin, M.P. Shekhar

Development of methodology: N. Saadat, F. Liu, X. Bao, J. Li, L.A.Polin, S. Gupta, G. Mao, M.P. Shekhar

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Saadat, F. Liu, P. Nangia-Makker, X. Bao, L.A. Polin, G. Mao, M.P. Shekhar

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Saadat, F. Liu, B. Haynes, X. Bao, L.A. Polin, S. Gupta, G. Mao, M.P. Shekhar

Writing, review, and/or revision of the manuscript: N. Saadat, F. Liu, B. Haynes, X. Bao, J. Li, L.A. Polin, S. Gupta, G. Mao, M.P. Shekhar

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N. Saadat, L.A. Polin, G. Mao, M.P. Shekhar

Study supervision: S. Gupta, G. Mao, M.P. Shekhar

This work was been supported by NCI (grant number R21CA178117, to M.P. Shekhar), National Science Foundation (grant number CHE1404285, to G. Mao), NIH (grant number R01HD031550, to G. Mao), and Molecular Therapeutics Program of Karmanos Cancer Institute (to M.P. Shekhar and G. Mao). The Pharmacology Core and Animal Model and Therapy Evaluation Core facilities are supported by NIH Center grant P30 CA022453 to the Karmanos Cancer Institute at Wayne State University. B. Haynes was supported by Initiative for Maximizing Student Diversity award from NIH to Wayne State University (R25 GM058905) and Ruth L. Kirschstein National Research Service Award T32-CA009531 training grant from NIH.

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