Reverse-phase Protein Lysate Arrays Research Articles

IGF-1R and mTOR Blockade: Novel Resistance Mechanisms and Synergistic Drug Combinations for Ewing Sarcoma.

Therapies cotargeting insulin-like growth factor receptor 1 (IGF-1R) and mammalian target of rapamycin (mTOR) have demonstrated remarkable, albeit short-lived, clinical responses in a subset of Ewing sarcoma (ES) patients. However, the mechanisms of resistance and applicable strategies for overcoming drug resistance to the IGF-1R/mTOR blockade are still undefined. To elucidate predominant mechanism(s) of acquired drug resistance while identifying synergistic drug combinations that improve clinical efficacy, we generated more than 18 ES cell lines resistant to IGF-1R- or mTOR-targeted therapy. Two small-molecule inhibitors of IGF-1R were chosen, NVP-ADW-742 (IGF-1R-selective) and OSI-906 (a dual IGF-1R/insulin receptor alpha [IR-α] inhibitor). Reverse-phase protein lysate arrays (RPPAs) revealed proteomic changes linked to IGF-1R/mTOR resistance, and selected proteins were validated in cell-based assays, xenografts, and within human clinical samples. All statistical tests were two-sided. Novel mechanisms of resistance (MOR) emerged after dalotuzumab-, NVP-ADW-742-, and OSI-906-based targeting of IGF-1R. MOR to dalotuzumab included upregulation of IRS1, PI3K, and STAT3, as well as p38 MAPK, which was also induced by OSI-906. pEIF4E(Ser209), a key regulator of Cap-dependent translation, was induced in ridaforolimus-resistant ES cell lines. Unique drug combinations targeting IGF-1R and PI3K-alpha or Mnk and mTOR were synergistic in vivo and vitro (P < .001) as assessed respectively by Mantel-Cox and isobologram testing. We discovered new druggable targets expressed by chemoresistant ES cells, xenografts, and relapsed human tumors. Joint suppression of these newfound targets, in concert with IGF-1R or mTOR blockade, should improve clinical outcomes.

Abstract LB-235: IGF-1R/mTOR blockade and mechanisms of resistance in Ewing sarcoma

Abstract Ewing sarcoma (ES), the second most common bone cancer in adolescents and young adults, is dependent upon the insulin-like growth factor receptor 1 (IGF-1R) and mammalian target of rapamycin (mTOR) cell-signaling axis for malignancy, survival, and metastasis. Therapies co-targeting IGF-1R and mTOR have demonstrated remarkable, although ephemeral, clinical responses in a subset of ES patients. However, the mechanisms of resistance and applicable strategies for overcoming drug resistance to IGF-1R/mTOR blockade are still undefined. To elucidate predominant mechanism(s) of acquired drug resistance, while identifying synergistic drug combinations that improve clinical efficacy, we generated ES cell lines resistant to IGF-1R or mTOR targeted therapy. Two small-molecule inhibitors of IGF-1R were chosen, NVP-ADW-742 (IGF-1R-selective) and OSI-906 (a dual IGF-1R/insulin receptor alpha [IR-α] inhibitor). Chronic ridaforolimus exposure led to mTOR-resistant ES cell lines. Reverse phase protein lysate arrays (RPPA) revealed proteomic changes linked to IGF-1R/mTOR resistance and selected proteins were validated using in vitro cell-based assays, xenografts, and human clinical samples. Novel mechanisms of resistance (MOR) occurred in NVP-ADW-742-, OSI-906-, and ridaforolimus-resistant cells. Of particular clinical relevance given OSI-906's ongoing phase 1b/2 clinical trials, the MOR varied depending upon whether IGF-1R or both IGF-1R and IR-α were targeted. Key proteomic alterations present in vitro also occurred in ES xenografts and human tumor specimens exposed to these agents. Novel synergistic drug combinations (e.g., inhibitors of Mnk and mTOR) were identified in vitro that are currently undergoing prospective validation. In conclusion, we discovered new druggable targets expressed by chemo-resistant ES cells, xenografts, and relapsed human tumors. Joint suppression of these newfound targets with IGF-1R or mTOR enhanced the antineoplastic effect of IGF-1R/mTOR blockade and suggests the need for in vivo validation. Citation Format: Salah-Eddine LAMHAMEDI CHERRADI, Brian Menegaz, Vandhana Ramamoorthy, Joseph Ludwig. IGF-1R/mTOR blockade and mechanisms of resistance in Ewing sarcoma. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr LB-235.

Abstract P6-03-08: Androgen receptor (AR): A novel target for radiosensitization and treatment in triple-negative breast cancers (TNBC)

Abstract Background: Increased rates of locoregional recurrence have been observed in TNBC despite the use of chemotherapy and radiation (RT). Thus, approaches that result in radiosensitizaton of TNBC are critically needed. We have previously characterized the radiation response of 21 breast cancer cell (BCC) lines using clonogenic survival assays. We now pair this data with high-throughput drug screen data available through cancer cell line encyclopedia studies to identify AR as a top target for radiosensitization and assess AR inhibition as a radiosensitization strategy for TNBC. Methods: Clonogenic survival assays were performed to determine the intrinsic RT sensitivity of 21 BCC lines (0-8 Gy RT). IC50 values were determined for 130 clinically available compounds and correlation coefficients were calculated using IC50 values (for drug sensitivity) and SF-2Gy (for radiation sensitivity). Gene expression was measured using Affymetrix microarrays and protein expression was measured using reverse-phase protein lysate arrays (RPPA) of human tumor samples (n=2,061) and BCC lines (n=51). AR function was assessed using siRNA knockdown or inhibition with MDV3100 (enzalutamide). Kaplan-Meier analysis was performed to determine the clinical impact of AR expression on local control and survival. A Cox proportional hazards model was constructed to identify potential factors of survival, and multivariate analysis was used to determine variables most significantly associated with LRF survival. Results: Our radiosensitizer screen nominated bicalutamide as one of the most effective drugs in treating radioresistant BCC lines (R2= 0.46, p-value &amp;lt;0.001). Recognizing that a subgroup of TNBC includes AR expressing tumors, we interrogated the expression of AR in &amp;gt;2000 human breast tumor samples and found signifi[not]cant heterogeneity in AR expression with an increase in TNBC (35% of tumors) compared to non-TNBC (28% of tumors). This same heterogeneity was also identified in human BCC lines. There was a strong correlation between AR RNA expression and protein expression (R2= 0.72, p &amp;lt;0.0001). Inhibition of AR using both siRNA and MDV3100 induced radiation sensitivity in vitro with an enhancement ratio (ER) of 1.35-1.42 in AR-positive TNBC lines. No such radiosensitization was seen in AR-negative TNBC or ER-positive, AR-negative BCC lines. Radiosensitization was at least partially dependent on impaired dsDNA break repair mediated by DNAPKcs. In vivo assessment of tumor growth inhibition with RT and anti-AR strategies are currently underway. Clinically, analyses of patients with TNBC showed that patients whose tumors had high expression of AR had markedly higher rates of LR after RT than patients with low expression of AR (HR for LR 2.9-3.2, p-value &amp;lt;0.01, 2 independent datasets). There was no difference in LR in TNBC patients not treated with RT when stratified by AR expression status. In multivariate analysis, AR expression was the variable most significantly associated with worse LRF survival after RT with a HR of 3.58 (p-value &amp;lt; 0.01). Conclusion: Our results implicate AR as a mediator of radioresistance in breast cancer and support the rationale for developing clinical strategies to inhibit AR as a novel radiosensitizing target in TNBC. Citation Format: Corey Speers, Shuang G Zhao, Meilan Liu, Joseph Evans, Prasanna Alluri, Daniel F Hayes, Felix Y Feng, Lori J Pierce. Androgen receptor (AR): A novel target for radiosensitization and treatment in triple-negative breast cancers (TNBC) [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P6-03-08.

Protein Expression Clusters Can Differentiate Leukemia Subtypes in Pediatric Leukemia

Background Although AML and ALL are thought to arise from different progenitor cells, (common myeloid vs. common lymphoid) they share many clinical features and pathophysiological characteristics, such as excessive proliferation in association with blocks in differentiation, that raise the question of whether they might share mechanistic commonalities despite separate ontologic origins. This in turn would suggest common therapies to utilize in cases with similar mechanisms of action. Using reverse-phase protein lysate arrays (RPPA), we strove to determine if RPPA could 1) determine protein activation differences between pediatric ALL and AML, and 2) determine if protein expression clusters were affected by sample collection method, an issue critical to assessing protein activation pathways in the setting of cooperative group clinical trials.Methods We generated a custom RPPA with 132 leukemia enriched samples. The array included 59 AML (3 APL, 1 TMD, 1 relapse) and 73 ALL (58 pre-B, 15 T-ALL, 6 relapsed) patients. Median age was 9.1 years (range 0.0 – 22.1 years). Samples were enriched for leukemia cells using ficoll followed by magnetic bead separation, either CD3/CD19 depletion (AML), or B/T-cell isolation (ALL). The RPPA was probed with 194 strictly validated antibodies (139 antibodies assessing total protein expression, 36 phosphoproteins, 6 cleaved forms, and 3 methylation sites). Samples were collected in either heparin tubes (n=101) or CellSave preservations tubes (Veridex) (n=97), with 57 samples having both tube types.Results Clustering demonstrated that AML and ALL have highly different protein expression and activation patterns. Pre-treatment samples collected in Cell Save preservation tubes revealed two protein clusters dividing the patients into two groups based on the levels of 62/194 significantly differentially expressed proteins (p=0.01, FDR = 0.027). An AML-dominant cluster contained 31 AML and 4 ALL among its 35 members. The ALL-dominant cluster had 62 ALL and a lone AML sample among 63 members. Using heparin tubes, four protein clusters were observed which divided patients into three clusters based on the differential expression of 83 proteins (p=0.01, FDR<0.023). The ALL-dominant cluster, composed of 53 members, had 51 ALL and 2 AML samples. The two AML patients in the ALL group included a patient with “difficult to treat” AML that achieved CR after 3 cycles of standard AML therapy with the proteasome inhibitor bortezomib, and a patient with 11q23 ALL that subsequently underwent a lineage switch to AML. There were two AML clusters, one with 85% AML (middle cluster in figure) with 17 AML and 3 ALL and the other with 27 patents (far left cluster) consisting of 70% AML (18 AML and 1 TMD) along with 8 ALL samples. In the far left cluster, of the 8 ALL cases, 5 had changes involving chromosome 21, including four with t(12;21), and 1 with trisomy 21. ALL samples in total had only 8/73 with the t(12;21) translocation, indicating a statistically significant difference (p=0.002, two-tailed Fisher’s exact test).Twenty-seven proteins were different in both the CellSave (CS) and heparin analyses. An additional 34 proteins, mostly epigenetic modifiers and apoptosis pathway proteins (Bcl2, BAD, BAX), were seen in the CS samples. An additional 52 proteins were detected in the heparin group, many related to cell stress pathways.Conclusion Protein expression profiles strongly divided pediatric ALL from AML. Since this classification scheme is dependent on protein expression and functional activation states, it may be possible to further identify patients with different risk characteristics based on protein expression profiles. However, at least 111 of 194 proteins showed similar expression in both AML and ALL suggesting that there may be commonalities in mechanisms shared between leukemias of different lineages. Further analysis of protein functional groups and pathway utilization are in progress to define both differences and commonalities. [Display omitted] DisclosuresOff Label Use: bortezomib: off label use in pediatric AML as part of a COG clinical trial.

Proteomic classification of breast cancer.

Being a significant health problem that affects patients in various age groups, breast cancer has been extensively studied to date. Recently, molecular breast cancer classification has advanced significantly with the availability of genomic profiling technologies. Proteomic technologies have also advanced from traditional protein assays including enzyme-linked immunosorbent assay, immunoblotting and immunohistochemistry to more comprehensive approaches including mass spectrometry and reverse phase protein lysate arrays (RPPA). The purpose of this manuscript is to review the current protein markers that influence breast cancer prediction and prognosis and to focus on novel advances in proteomic classification of breast cancer.

Reno: regularized non-parametric analysis of protein lysate array data

The reverse-phase protein lysate arrays have been used to quantify the relative expression levels of a protein in a number of cellular samples simultaneously. To avoid quantification bias due to mis-specification of commonly used parametric models, a nonparametric approach based on monotone response curves may be used. The existing methods, however, aggregate the protein concentration levels of replicates of each sample, and therefore fail to account for within-sample variability. We propose a method of regularization on protein concentration estimation at the level of individual dilution series to account for within-sample or within-group variability. We use an efficient algorithm to optimize an approximate objective function, with a data-adaptive approach to choose the level of shrinkage. Simulation results show that the proposed method quantifies protein concentration levels well. We show through the analysis of protein lysate array data from cell lines of different cancer groups that accounting for within-sample variability leads to better statistical analysis. Code written in statistical programming language R is available at: http://odin.mdacc.tmc.edu/~jhhu/Reno

Proteomic Profiling Identifies Pathways Dysregulated in Non-small Cell Lung Cancer and an Inverse Association of AMPK and Adhesion Pathways with Recurrence

The identification of key pathways dysregulated in non-small cell lung cancer (NSCLC) is an important step toward understanding lung pathogenesis and developing new therapeutic approaches. Toward this goal, reverse-phase protein lysate arrays (RPPA) were used to compare signaling pathways between NSCLC tumors and paired normal lung tissue from 46 patients and assess their association with clinical outcome. After RPPA quantification of 63 proteins and phosphoproteins, tissue pairs were randomized to a training set (n = 25 pairs) and test set (n = 21 pairs). In the training set, 15 protein markers were differentially expressed between tumors and normal lung (p ≤ 0.01), including markers in the PI3K/AKT and p38 MAPK signaling pathways (e.g., p70S6K, S6, p38, and phospho-p38), as well as caveolin-1 and β-catenin. A four-protein signature (p70S6K, cyclin B1, pSrc(Y527), and caveolin-1) independent of histology classified specimens as tumor versus normal with a predicted accuracy of 83%, sensitivity of 67%, and specificity of 100%. The signature was validated in the test set, correctly classifying all normal tissues and 14 of 21 tumor tissues. RPPA results were confirmed by immunohistochemistry for caveolin-1 and p70S6K. In tumors from patients with resected NSCLC, expression of proteins in the energy-sensing AMPK pathway (pLKB1, AMPK, p-Acetyl-CoA, pTSC2), adhesion, EGFR, and Rb signaling pathways was inversely associated with NSCLC recurrence. These data provide evidence for dysregulation of several pathways including those involving energy sensing and adhesion that are potentially associated with NSCLC pathogenesis and disease recurrence.

Comprehensive Mapping of the Human Kinome to Epidermal Growth Factor Receptor Signaling

Disregulation of epidermal growth factor receptor (EGFR) signaling directly promotes bypass of proliferation and survival restraints in a high frequency of epithelia-derived cancer. As such, much effort is currently focused on decoding the molecular architecture supporting EGFR activation and function. Here, we have leveraged high throughput reverse phase protein lysate arrays, with a sensitive fluorescent nanocrystal-based phosphoprotein detection assay, together with large scale siRNA-mediated loss of function to execute a quantitative interrogation of all elements of the human kinome supporting EGF-dependent signaling. This screening platform has captured multiple novel contributions of diverse protein kinases to modulation of EGFR signal generation, signal amplitude, and signal duration. As examples, the prometastatic SNF1/AMPK-related kinase hormonally upregulated Neu kinase was found to support EGFR activation in response to ligand binding, whereas the enigmatic kinase MGC16169 selectively supports coupling of active EGFR to ERK1/2 regulation. Of note, the receptor tyrosine kinase MERTK and the pyrimidine kinase UCK1 were both found to be required for surface accumulation of EGFR and subsequent pathway activation in multiple cancer cell backgrounds and may represent new targets for therapeutic intervention.

IGFBP2 is a candidate biomarker for Ink4a-Arf status and a therapeutic target for high-grade gliomas

The levels of insulin-like growth factor-binding protein 2 (IGFBP2) are elevated during progression of many human cancers. By using a glial-specific transgenic mouse system (RCAS/Ntv-a), we reported previously that IGFBP2 is an oncogenic factor for glioma progression in combination with platelet-derived growth factor-beta (PDGFB). Because the INK4a-ARF locus is often deleted in high-grade gliomas (anaplastic oligodendroglioma and glioblastoma), we investigated the effect of the Ink4a-Arf-null background on IGFBP2-mediated progression of PDGFB-initiated oligodendroglioma. We demonstrate here that homozygous deletion of Ink4a-Arf bypasses the requirement of exogenously introduced IGFBP2 for glioma progression. Instead, absence of Ink4a-Arf resulted in elevated endogenous tumor cell IGFBP2. An inverse relationship between p16(INK4a) and IGFBP2 expression was also observed in human glioma tissue samples and in 90 different cancer cell lines by using Western blotting and reverse-phase protein lysate arrays. When endogenous IGFBP2 expression was attenuated by an RCAS vector expressing antisense IGFBP2 in our mouse model, a decreased incidence of anaplastic oligodendroglioma as well as prolonged survival was observed. Thus, p16(INK4a) is a negative regulator of the IGFBP2 oncogene. Loss of Ink4a-Arf results in increased IGFBP2, which contributes to glioma progression, thereby implicating IGFBP2 as a marker and potential therapeutic target for Ink4a-Arf-deleted gliomas.

Constitution and quantity of lysis buffer alters outcome of reverse phase protein microarrays

Application of novel technology to clinical samples requires optimization of procedures. Reverse phase protein lysate arrays use femtomolar quantities of tissue lysate from clinical samples with which to profile biochemical events happening in the tumor. We analyzed the effects of different tissue solubilization buffers on frozen ovarian tumor samples in order to identify the system with the best signal intensity dynamic range, reproducibility, tissue solubility, and signal consistency. A modified RIPA-like buffer supplemented with DTT and SDS was deemed optimal.

Non-parametric quantification of protein lysate arrays

Proteins play a crucial role in biological activity, so much can be learned from measuring protein expression and post-translational modification quantitatively. The reverse-phase protein lysate arrays allow us to quantify the relative expression levels of a protein in many different cellular samples simultaneously. Existing approaches to quantify protein arrays use parametric response curves fit to dilution series data. The results can be biased when the parametric function does not fit the data. We propose a non-parametric approach which adapts to any monotone response curve. The non-parametric approach is shown to be promising via both simulation and real data studies; it reduces the bias due to model misspecification and protects against outliers in the data. The non-parametric approach enables more reliable quantification of protein lysate arrays. Code to implement the proposed method in the statistical package R is available at: http://odin.mdacc.tmc.edu/jhu/lysatearray-analysis/

Transcript and protein expression profiles of the NCI-60 cancer cell panel: an integromic microarray study

To evaluate the utility of transcript profiling for prediction of protein expression levels, we compared profiles across the NCI-60 cancer cell panel, which represents nine tissues of origin. For that analysis, we present here two new NCI-60 transcript profile data sets (A based on Affymetrix HG-U95 and HG-U133A chips; Affymetrix, Santa Clara, CA) and one new protein profile data set (based on reverse-phase protein lysate arrays). The data sets are available online at http://discover.nci.nih.gov in the CellMiner program package. Using the new transcript data in combination with our previously published cDNA array and Affymetrix HU6800 data sets, we first developed a "consensus set" of transcript profiles based on the four different microarray platforms. Using that set, we found that 65% of the genes showed statistically significant transcript-protein correlation, and the correlations were generally higher than those reported previously for panels of mammalian cells. Using the predictive analysis of microarray nearest shrunken centroid algorithm for functional prediction of tissue of origin, we then found that (a) the consensus mRNA set did better than did data from any of the individual mRNA platforms and (b) the protein data seemed to do somewhat better (P = 0.027) on a gene-for-gene basis in this particular study than did the consensus mRNA data, but both did well. Analysis based on the Gene Ontology showed protein levels of structure-related genes to be well predicted by mRNA levels (mean r = 0.71). Because the transcript-based technologies are more mature and are currently able to assess larger numbers of genes at one time, they continue to be useful, even when the ultimate aim is information about proteins.

Proteomic prediction in hormone receptor-positive breast cancer

541 Background: Hormone receptor-positive breast cancer (HR+BC) is heterogeneous. Attempts to use transcriptional analysis such as Oncotype Dx facilitate therapy planning but fail to identify patients who will relapse or potential therapy targets. Methods: We used reverse phase protein lysate arrays and Microvigene software to quantify expression of estrogen receptor (ERα) and 36 total/activated components of the HER2, phosphatidylinositol-3-kinase (PI3K), mitogen-activated protein kinase (MAPK), and STAT pathways in 64 HR+BCs and 40 breast cancer cell lines. Clustering was performed with Xcluster and Treeview. Results: 47/64 HR+BC patients were treated with adjuvant hormone therapy and 43/64 with chemotherapy. There were 12 recurrences including 5 patients diagnosed with metastases within 0–3 months of diagnosis. Unsupervised analysis using the expression of all 37 proteins revealed two large subclusters of HR+BCs. One large cluster was composed of tumors with lower ERα expression levels and was driven by an antibody group composed mostly of phosphoproteins indicative of activated growth factor signaling pathways. Thus, there were significant inverse correlations between ERα expression and the expression and activation of components of the PI3K/MAPK pathways including EGFR, src, AKT, 4EBP1, and PKCα (p≤0.05 for each). Similar inverse correlations were seen in 40 assayed breast cancer cell lines. The clinicoproteomic predictors of relapse among HR+BCs were nuclear grade (p=0.001), low expression of ERα (p=0.04), low p38 phosphorylation (p=0.02), and high p53 (p=0.02). There also was a trend (p≤0.1) to the association of low MAPK and S6 phosphorylation, low p27, and high cyclin B1 with relapse. Using these antibodies to perform a supervised analysis, we found a small group of p53-high, cyclin B1-high, ERα-low HR+BCs with a 75% likelihood of relapse, significantly greater than in other tumors (p&lt;0.003). Since 10/12 relapses occurred in 26 grade 3 tumors, we searched for and found a ‘grade 3’ HR+BC 6 antibody signature associated with a recurrence-free survival at 20 months of 17% compared to 100% in other patients (p=0.002). Conclusion: Although patient numbers are small and the findings require validation, kinase signaling profiles have potential in breast cancer prediction. No significant financial relationships to disclose.

Pathway alterations during glioma progression revealed by reverse phase protein lysate arrays

The progression of gliomas has been extensively studied at the genomic level using cDNA microarrays. However, systematic examinations at the protein translational and post-translational levels are far more limited. We constructed a glioma protein lysate array from 82 different primary glioma tissues, and surveyed the expression and phosphorylation of 46 different proteins involved in signaling pathways of cell proliferation, cell survival, apoptosis, angiogenesis, and cell invasion. An analysis algorithm was employed to robustly estimate the protein expressions in these samples. When ranked by their discriminating power to separate 37 glioblastomas (high-grade gliomas) from 45 lower-grade gliomas, the following 12 proteins were identified as the most powerful discriminators: IBalpha, EGFRpTyr845, AKTpThr308, phosphatidylinositol 3-kinase (PI3K), BadpSer136, insulin-like growth factor binding protein (IGFBP) 2, IGFBP5, matrix metalloproteinase 9 (MMP9), vascular endothelial growth factor (VEGF), phosphorylated retinoblastoma protein (pRB), Bcl-2, and c-Abl. Clustering analysis showed a close link between PI3K and AKTpThr308, IGFBP5 and IGFBP2, and IBalpha and EGFRpTyr845. Another cluster includes MMP9, Bcl-2, VEGF, and pRB. These clustering patterns may suggest functional relationships, which warrant further investigation. The marked association of phosphorylation of AKT at Thr308, but not Ser473, with glioblastoma suggests a specific event of PI3K pathway activation in glioma progression.

Mitochondrial proteome: altered cytochrome c oxidase subunit levels in prostate cancer.

Laser capture microdissection was combined with reverse phase protein lysate arrays to quantitatively analyze the ratios of mitochondrial encoded cytochrome c oxidase subunits to nuclear encoded cytochrome c oxidase subunits, and to correlate the ratios with malignant progression in human prostate tissue specimens. Cytochrome c oxidase subunits I-III comprise the catalytic core of the enzyme and are all synthesized from mitochondrial DNA. The remaining subunits (IV-VIII) are synthesized from cellular nuclear DNA. A significant (P < 0.001, 30/30 prostate cases) shift in the relative concentrations of nuclear encoded cytochrome c oxidase subunits IV, Vb, and VIc compared to mitochondrial encoded cytochrome c oxidase subunits I and II was noted during the progression of prostate cancer from normal epithelium through premalignant lesions to invasive carcinoma. Significantly, this shift was discovered to begin even in the premalignant stage. Reverse phase protein lysate array-based observations were corroborated with immunohistochemistry, and extended to a few human carcinomas in addition to prostate. This analysis points to a role for nuclear DNA encoded mitochondrial proteins in carcinogenesis; underscoring their potential as targets for therapy while highlighting the need for full characterization of the mitochondrial proteome.