Individual‐specific variation in the respiratory activities of HMECs and their bioenergetic response to IGF1 and TNFα

Metabolic reprograming is a hallmark of cancer cells. However, the roles of pre‐existing differences in normal cells metabolism toward cancer risk is not known. In order to assess pre‐existing variations in normal cell metabolism, we have quantified the inter‐individual variation in oxidative metabolism of normal primary human mammary epithelial cells (HMECs). We then assessed their response to selected cytokines such as insulin growth factor 1 (IGF1) and tumor necrosis factor alpha (TNFα), which are associated with breast cancer risk. Specifically, we compared the oxidative metabolism of HMECs obtained from women with breast cancer and without cancer. Our data show considerable inter‐individual variation in respiratory activities of HMECs from different women. A bioenergetic parameter called pyruvate‐stimulated respiration (PySR) was identified as a key distinguishing feature of HMECs from women with breast cancer and without cancer. Samples showing PySR over 20% of basal respiration rate were considered PySR+ve and the rest as PySR−ve. By this criterion, HMECs from tumor‐affected breasts (AB) and non‐tumor affected breasts (NAB) of cancer patients were mostly PySR−ve (88% and 89%, respectively), while HMECs from non‐cancer patients were mostly PySR+ve (57%). This suggests that PySR−ve/+ve phenotypes are individual‐specific and are not caused by field effects due to the presence of tumor. The effects of IGF1 and TNFα treatments on HMECs revealed that both suppressed respiration and extracellular acidification. In addition, IGF1 altered PySR−ve/+ve phenotypes. These results reveal individual‐specific differences in pyruvate metabolism of normal breast epithelial cells and its association with breast cancer risk.


| INTRODUCTION
Metabolic individuality and its relevance to health risks are currently being explored (Suhre et al., 2011a(Suhre et al., , 2011b. Cellular bioenergetics can vary in a given cell type from individual-to-individual. This variation can arise due to genetic, epigenetic, aging, and environmental exposures. Bioenergetic impairments are linked with various human diseases including cancer (Wallace, 2005(Wallace, , 2013. Cancer cells reprogram their metabolism to survive, grow, and proliferate. The metabolic reprograming involves changes in bioenergetics and redox balance to support enhanced biosynthesis of macromolecules such as lipids, nucleotides, and nonessential amino acids (DeBerardinis & Chandel, 2016). Alterations in oncogenes and tumor suppressors are considered as the underlying cause of metabolic reprograming in cancer cells. However, it is possible that bioenergetic differences, which in normal cells can be due to genetic variation or environmental exposures, precede cancer development.
Mitochondria play key roles in cellular bioenergetics by carrying out oxidative phosphorylation. Oxidative phosphorylation depends on supply of substrates to the respiratory chain and ATP demand. In most cells, glucose is the primary bioenergetic fuel. After its entry into cells, glucose is converted into pyruvate via the reactions of glycolysis. Next pyruvate is either oxidized inside mitochondria or converted into lactate within cytosol. Pyruvate oxidation inside mitochondria generates NADH and FADH 2, which support cellular respiration. The secretion of lactate in the extracellular medium causes acidification, which is often used as a surrogate for assessing the rate of glycolysis (Wu et al., 2007). Because the respiratory chain is functionally linked with the TCA cycle, the bicarbonate produced by it also contributes to medium acidification (Mookerjee, Goncalves, Gerencser, Nicholls, & Brand, 2015). Therefore, relative contributions of lactate and bicarbonate to extracellular medium acidification provide additional insights into overall cellular bioenergetics (Mookerjee, Nicholls, & Brand, 2016).
In this study, we employed respirometry to assess bioenergetic individuality of normal breast epithelial cells from different women and determined if a pre-existing bioenergetic difference may be linked with breast cancer risk. Further, we assessed bioenergetic responses of breast epithelial cells to treatments with host factors, such as, insulin growth factor 1 (IGF1) and tumor necrosis factor alpha (TNFα) because they were associated with breast cancer risk in women (Kaaks et al., 2014;Szlosarek, Charles, & Balkwill, 2006;To, Knower, & Clyne, 2013). A bioenergetic parameter called pyruvate-stimulated respiration (PySR) was found to distinguish cells from women with breast cancer and without cancer. The cells from women with breast cancer were mostly PySR −ve (89%). While both cytokines had overall suppressive effects on cellular respiration and acidification, their effects were variable in different individuals. The effects of IGF1 were more consistent compared to TNFα. The PySR −ve/+ve phenotype was also altered by IGF1 to a larger extent than TNFα.
Mitochondrial bioenergetics positively regulates tumor suppressor protein p53 and alters radiation sensitivity of cells (Compton et al., 2011). Therefore, we tested the radiation response of HMECs from different individuals in the presence and absence of IGF1. The effects of IGF1 varied in cells from different individuals. These data underscore individual-specific differences in pyruvate metabolism and they suggest that pre-existing differences in pyruvate metabolism may contribute to breast cancer risk. The analyses also demonstrate that cytokines can alter breast epithelial cells metabolism.

| Materials
The reagents were procured from following sources. DMEM/F12 and hydrocortisone were obtained from Corning/Cellgro (Manassas, VA).
Cholera toxin and collagenase were obtained from EMD-Millipore (Billerica, MA) and Gibco/Thermo Fisher, respectively. Insulin, hyaluronidase, IGF1, TNFα, and all other reagents were purchased from Sigma-Aldrich unless otherwise indicated.

| Study outline
This study was approved by Baystate Medical Center Institutional Review Board (IRB#324059). The study subjects were women enrolled in the Rays of Hope Breast Research Patient Registry. All subjects were consented to provide excess tissue not needed for diagnostic purposes, and clinical data.
Surgery was performed at Baystate Medical Center, Springfield, MA. The procured normal (benign) breast tissue was fresh with less than 1 hr anoxic time. Sampled tissues were from subjects with cancer undergoing unilateral or bilateral mastectomy for cancer, or reduction mammoplasty from subjects with no cancer history (Table 1). In tumor affected breast the normal breast tissue was procured away from the tumor. Normal HMECs obtained from tumor-affected breasts (AB) and paired non-affected breasts (NAB) were designated as AB or pAB-and pNAB-HMECs, respectively, ( Figure 1). The women undergoing reduction mammoplasty had no history of breast cancer, and HMECs obtained from them were designated rNAB.
The pNAB and rNAB were grouped as NAB (unless otherwise specified). In this study only normal/benign HMECs irrespective of their origin in breasts with or without tumors were used. Three types of comparisons were made between AB-versus NAB-HMECs. One involved all AB-HMECs versus all NAB-HMECs irrespective of their origin. The second involved pAB versus pNAB-HMECs obtained from bilateral mastectomies. The third involved pNAB-versus rNAB-HMECs from breast cancer and reduction mammoplasty patients, respectively. Bioenergetics of both AB-and NAB-HMECs with and without IGF1 and TNFα treatments were assessed by in situ respirometry. Respirometry data were analyzed to quantify (i) interindividual variation, (ii) bioenergetic differences in AB-versus NAB-HMECs, and (iii) their responses to IGF1 and TNFα treatments. Figure 1 provides a schematic outline of this study.

| HMECs preparation
Tissues were finely minced and digested at 37°C overnight in mammary digestion medium (DMEM/F12 supplemented with 5 µg/ml insulin, 2% BSA, 5 µg/ml hydrocortisone, 10 ng/ml cholera toxin, 300 U/ml collagenase, 100 U/ml hyaluronidase, and 1× antibiotic/ antimycotic mixture). Any undigested tissue was removed, and the tissue suspension was centrifuged at 80g for 10 min. The pellet was washed in 10 ml of cold Hanks balanced salt solution containing 5% fetal bovine serum (HBF) and re-centrifuged. Next, the pellet was incubated with 2 ml of 0.25% trypsin/EDTA for 5 min at room temperature, and washed with HBF and centrifuged. The cells were treated with 2 ml dispase (2 mg/ml) SCHNEIDER ET AL. | 2751 and 20 U of DNase-I for 5 min at room temperature before HBF wash and centrifugation. Cells were passed through 100 and 40 μm cell strainers and centrifuged for 5 min at 100g. Resulting cells were designated as HMECs, plated in 10% fetal bovine serum to allow adherence and then switched to Mammocult medium for culture with designation Passage zero (P0). Subsequently, HMECs were passaged at 3-4 day intervals, and all experiments were carried out with HMECs at passage two unless otherwise noted. Table 1 shows the list of HMECs used in this study with associated age and body mass index (BMI) of the subjects.
Antibiotics-free medium was used for 48 hr culture to remove any suppressive effects of antibiotics on mitochondrial protein synthesis.
The IGF1 (20 ng/ml) and TNFα (10 ng/ml) were added to cells 24 hr before respiration assays and 24 hr post-seeding. The respiratory activity of cells was assayed in 700 µl low-K + buffer [LKB: 3.5 mM KCl, 10 mM KH 2 PO 4 , 1 mM Na 2 SO 4 , 2 mM MgCl 2 , 1.3 mM CaCl 2 , 120 mM NaCl, 15 mM glucose, pH 7.4] containing 0.4% fatty acids-free bovine serum albumin (BSA). Cells were washed twice with 700 µl LKB and then incubated in a non-CO 2 incubator at 37°C for ∼30-60 min. Prehydrated XF24 cartridges for 24 hr were calibrated according to the manufacturer's (Agilent-Seahorse) instructions after loading injection ports with the indicated compounds (75 µl). After calibration of the sensor cartridges, the V7 culture plates with cells were loaded into the XF24-3 analyzer. Repeated cycles of mixing, waiting, and measurements were performed as described (Gerencser et al., 2009;Wu et al., 2007). After four basal respiration rate measurements, oligomycin (2 µg/ml), carbonylcyanide p-trifluoromethoxy phenylhydrazone (FCCP, 2 µM), pyruvate (10 mM), and rotenone plus antimycin A Prefix p in pAB and pNAB indicates paired AB-and NAB-HMECs obtained from the tumor affected (AB) and non-affected breasts (NAB) of cancer patients. Prefix r in rNAB indicates that NAB-HMECs came from reduction mammoplasty. HMECs from single mastectomy are denoted as AB as they came from tumor affected breast (AB). a HMECs used for radiation-induced death.
(1 µM each) were added. At least three respiration rates were measured following the addition of each compound ( Figure 2). The obtained respiratory profiles were used to derive different parameters as in Table 2. Means of four basal rates and three rates after treatments (n = 4-5 well/group) were used for data analysis (see below The extracellular acidification rates (ECAR) obtained from the XF24-3 Analyzer were used to calculate total, respiratory, and glycolytic proton production rates (PPRtotal, PPRresp, and PPRglyc, respectively) as described . Buffering power (0.0322) of the respiratory medium was used to determine the fractions of PPRresp and PPRglyc as described

| Metabolic phenotyping/fingerprinting
Metabolic phenotyping was carried out using 96-well PM-M1 phenotyping microarray for mammalian cells (BioLog Inc., Hayward, CA; (Bochner et al., 2011)). PM-M1 microarray was preloaded with different carbon substrates. Cells (10,000/well) were seeded in 50 µl FIGURE 2 A representative respiratory profile. It shows the experimental scheme, pyruvate-stimulated respiration (PySR) and selected bioenergetic parameters. Indicated compounds were added sequentially from left to right. See Table 2 for individual bioenergetic parameters, and corresponding abbreviations, that can be derived from such a respiratory profile. Averages of at least three oxygen consumption rates (OCR) for each parameter were used in different analyses. On x-axis each dot represents one rate at corresponding time FIGURE 1 Schematic outline of the study. AB and NAB refer to tumor-affected or non-affected breasts, respectively. HMECs designations are: rNAB, reduction mammoplasty-derived; AB, single mastectomy-derived; pAB and pNAB, paired bilateral mastectomy-derived. For more details see Table 1 SCHNEIDER ET AL.

| Radiation sensitivity of HMECs
HMECs were plated in 8-well glass chamber slides (Falcon, BD Biosciences) and allowed to settle overnight. Half the cultures were pretreated with 20 ng/ml IGF-1 for 1 hr and then they were exposed to 5 Gy of γ-radiation using a 137 Cs irradiator (Gammator-B, Radiation Machinery Corporation, Parsippany, NJ). Next 24 hr after irradiation, the cells were checked for viability using the Live/Dead assay kit (Invitrogen). Briefly, 2 μM for Calcein-AM and 4 μM for EthD-1 (Invitrogen/Thermo Fisher) were added to the cells in PBS and incubated at 37°C for 3 hr. Then cells were imaged by fluorescent microscopy. Image J was used for counting of live and dead cells. The percentage of cell death in control and IGF1-treated conditions was compared using GraphPad Prism 5 software.
Corresponding controls versus treated (IGF1 or TNFα) comparisons were made using paired Student's t-test. Averages of four basal rates and three rates of treated conditions after the addition of indicated compound(s) were used in analyses. Statistical analyses were performed mostly using GraphPad Prism 5 and Microsoft Excel.
Data are shown either as mean ± SD or mean ± sem as specified.
Means of each bioenergetic parameter between two groups were compared by Student's t-test as indicated above. Statistical significance was calculated at p* ≤ 0.05, **0.01, ***0.001 with 95% confidence interval. The frequency of PySR −ve or PySR +ve phenotypes in pNAB-versus rNAB-HMECs samples was analyzed by Chi-square test using GraphPad Prism 5. The 20% cut-off for PySR +ve phenotype was arbitrarily set above the average standard error of controls, IGF1and TNFα-treated rNAB-HMECs, because mostly they showed the PySR +ve phenotype and they were from women who did not have cancer.

| Inter-individual variation in AB-and NAB-HMECs bioenergetics
The variation in bioenergetics of HMECs from different women was assessed by respirometry. Figure 2 shows a typical respiratory profile with sequential additions of oligomycin, FCCP, pyruvate, and rotenone plus antimycin. Such profiles were obtained for HMECs from different individuals and used to derive various bioenergetic parameters as indicated in Table 2. Briefly, basal respiration has three components: ATP synthesis (ATPR)-and H + -leak (PLR)-supporting respirations, and non-mitochondrial respiration (NMR) (Jekabsons & Nicholls, 2004). The respiration medium, LKB, contained 15 mM glucose from the start. Thus basal respiration was primarily supported by glucose and any internal substrates that might be present inside cells. The oligomycin was added to block ATP synthesis and to estimate ATPR.
The oligomycin-insensitive respiration after subtracting the NMR gave an estimate of PLR. FCCP was used to induce maximal respiration by dissipating the H + gradient across the mitochondrial inner membrane.
Thus FCCP addition after oligomycin reports respiratory capacity on Bioenergetic parameters derived from a respiratory profile as shown in Figure 2 Parameters descriptions-> Abbreviation Basal respiration (see Figure 2) BR Non-mitochondrial respiration (rotenone + antimycin-A insensitive; see Figure 2) NMR

Mitochondrial basal respiration (BR-NMR) mBR
Oligomycin-insensitive respiration (see Figure 2) OIR ATP synthesis supporting respiration (oligomycin-sensitive, BR-OIR) ATPR glucose (RCo). Blocking mitochondrial synthesis may limit respiratory capacity due to ATP limitation. Therefore, to overcome any limitation in respiratory capacity, 10 mM pyruvate was added after FCCP. By using this experimental scheme, as shown in Figure 2, it was possible to assess respiratory capacity of cells on glucose versus glucose + pyruvate as substrates (i.e., before and after pyruvate addition, respectively) within the same experiment and assess individual-specific responses to exogenous pyruvate.
Pyruvate metabolism plays a central role in cellular bioenergetics.
Therefore, we compared respiratory responses of HMECs from different women to exogenous pyruvate. Figure 3a  showing PySR by ≥20% of basal respiration were considered PySR +ve and the rest were considered as PySR −ve . By this criterion 31% NAB-HMECs (n = 5/16) were PySR +ve (Figure 3c). These data suggest that in the majority of NAB-HMECs, glucose was sufficient to support maximal respiration and exogenous pyruvate did not increase respiration any further. Therefore, pyruvate addition did not alter mean spare respiratory capacity (SRC) significantly in either AB-or NAB-HMECs (Table 3). However, there were individual-specific differences. In some samples, SRC values were negative due to a decline of respiration following FCCP addition. In these samples, the addition of pyruvate did not rescue the respiratory capacity. This could be either due to defective respiratory chain, redox homeostasis, or pyruvate delivery to mitochondria. These data support existence of individual-specific variation in pyruvate metabolism of breast epithelial cells.
To determine whether the PySR −ve or PySR +ve phenotype was more common in breast epithelial cells from women without cancer, we compared pNAB-versus rNAB-HMECs. There was a striking difference in PySR −ve versus PySR +ve frequencies of both groups ( Figure 3c). The majority of pNAB-HMECs were PySR −ve . On the other hand, the majority of rNAB-HMECs (57%; n = 4/7) were PySR +ve . The average age of women undergoing mammoplasty was significantly lower than the women undergoing mastectomy (37.43 ± 6.13 vs. 50.57 ± 6.20 years, respectively, mean ± SD, unpaired t-test, FIGURE 3 Individual-specific variation in the respiratory activity. a) A respiratory profile of rNAB-HMECs (SS206) from a woman displaying robust response to exogenous pyruvate. b) A respiratory profile of rNAB-HMECs (SS229) from a woman displaying no response to exogenous pyruvate. c) Inter-individual variation in PySR in pNAB-versus rNAB-HMECs. The pNAB-and rNAB-HMECs were derived from bilateral (prophylactic) mastectomy (n = 9) and reduction mammoplasty (n = 7), respectively. See Figure 1 and Table 1  41.15% ± 50.50%, mean ± SD, n = 7 vs. 9, t-test p = 0.029; Figure 3e). These data suggest that PySR −ve phenotype may be associated with breast cancer risk. This is supported by predominantly PySR −ve phenotype of AB-HMECs (n = 14/16).
To determine the degree of PySR −ve/+ve phenotype switching in cells from the same women, we compared pAB-versus pNAB-HMECs.
Together, these data suggest that there is a good agreement in PySR −ve phenotype of breast epithelial cells from cancer patients, whether they come from the tumor affected or non-affected breasts. These data also reveal individual-specific bilateral differences in local environments due to the presence of tumor.
Next, we determined fold variation ( Under basal conditions 30-36% acidification was due to respiration-associated bicarbonate secretion (Figure 4d). In oligomycintreated cells, the respiratory acidification declined to ∼2% in both AB-and NAB-HMECs. This is expected because the TCA cycle will be slowed down in the presence of oligomycin (Kim et al., 2014).
Addition of FCCP after oligomycin increased the respiratory acidification to 29-39% of total. This is expected because the TCA cycle activity will increase to supply NADH/FADH 2 to the respiratory chain. Addition of pyruvate after FCCP did not increase either respiratory or glycolytic acidification any further (Figure 4d, data not shown). Together, the above data demonstrate the existence of a considerable degree of variation in respiration and acidification rates of normal breast epithelial cells from different women. The coupling efficiency (CE) informs about the degree of basal respiration supporting ATP synthesis inside mitochondria. Therefore, we assessed variation in coupling efficiency, which ranged from 85% to 105% in AB-HMECs and 84% to 110% in NAB-HMECs. Its values above 100% are due to over-estimation of NMR. Overall the NMR was 21 ± 3% and 23 ± 3% of basal respiration in AB-and NAB-HMECs (mean ± sem, n = 16 each group, Table 4). These values were obtained in the presence of oligomycin and FCCP added as shown in Figure 2.
Like AB-versus NAB-HMECs, no significant difference between pABversus pNAB-HMECs was observed (not shown). Between pNAB-and rNAB-HMECs, the only notable difference was the effect of exogenous pyruvate, which is discussed above. Although, the glycolytic PPR in rNAB-HMECs was relatively lower than pNAB-HMECs, it was not statistically significant (4.98 ± 7.21 rNAB n = 7 vs. 11.94 ± 8.29 pNAB n = 9, mean ± SD). Therefore, we conclude that overall mitochondrial bioenergetics of benign epithelial cells from cancer affected and nonaffected breasts are similar despite individual-specific variation.

| Bioenergetic response of HMECs to IGF1 treatment
We tested IGF1 as a potential host-factor that could affect cellular bioenergetics, because IGF1 is known to be associated with breast cancer risk (Kaaks et al., 2014). Cells were treated with IGF1 for 24 hr before respiratory assays were performed. Respirometry profiles of control and PySR informs about differences in exogenous pyruvate oxidation.
Therefore, we determined how IGF1 treatment affected PySR +ve/−ve phenotype. We observed an increase in the fraction of samples showing PySR +ve phenotype in both AB-and NAB-HMECs following treatment with IGF1. In AB-HMECs, the PySR +ve fraction increased from 13% (n = 2/16) to 63% (n = 10/16; Table 5). In NAB-HMECs, it increased from 31% (n = 5/16) to 60% (n = 11/16; Table 5). The increase in PySR +ve fraction affected the mean PySR values significantly in both AB-and NAB categories (Figure 5g,h). These and above mentioned data suggest that: (i) IGF1 can suppress respiratory activity, (ii) alter respiratory response to exogenous pyruvate, and (iii) affect mitochondrial bioenergetics of epithelial cells from both tumor-affected and non-affected breasts.
To determine if the suppressive effect of IGF1 on respiration correlated with reduced extracellular acidification, we compared the proton production rates (PPR) in control versus IGF1-treated cells.
Under basal condition, IGF1 significantly reduced respiratory PPR in both AB-and NAB-HMECs (Figure 5i,j). This correlated with a significant reduction in total PPR in AB-HMECs only. However in oligomycin treated condition, IGF1 significantly reduced glycolytic PPR that correlated with reduction in total PPR in both AB-and NAB-HMECs (Figure 5k,l). This suggests that mitochondrial ATP synthesis supports glycolysis in IGF1-treated cells to a larger extent than control cells. Under FCCP-treated conditions, both respiratory and glycolytic acidifications contributed toward reduced total acidification (Table S1). Unlike in AB-HMECs, in the presence of exogenous pyruvate, glycolytic PPR was not significantly affected by IGF1 in NAB-HMECs (Table S1). In terms of percent contribution of respiratory and glycolytic PPRs, the AB-HMECs were different from the NAB-HMECs (Table S2). These data suggest that IGF1 suppresses respiratory activity of HMECs by suppressing glycolysis. Further, in terms of extracellular acidification, there is a potential difference in the metabolism of breast epithelial cells from tumor-affected and nonaffected breasts in response to IGF1.
The effects of TNFα on the proton production rate (PPR) in control versus TNFα-treated cells were compared. In both AB-and NAB-HMECs, TNFα reduced respiratory PPR under basal condition (Figure 6i,j).
However, this correlated with significant reduction in total PPR only in AB-HMECs (Figure 6i). Under oligomycin-treated condition, TNFα significantly suppressed glycolytic PPR only in AB-HMECs and it correlated with a reduction in total PPR (Figure 6k). TNFα did not have a notable effect on PPR in oligomycin treated NAB-HMECs (Figure 6l).
Under FCCP-treated conditions, the reduced total acidification correlated with reduced glycolytic PPR (Table S3). Unlike AB-HMECs, the respiratory PPR was reduced in FCCP-treated NAB-HMECs, and the difference in total PPR was significant only in the presence of exogenous pyruvate.
These data suggest that TNFα effect on extracellular acidification was more prominent in AB-HMECs compared to NAB-HMECs (Table S3). In terms of percent respiratory and glycolytic PPRs, there were no significant differences between control and TNFα treated cells, which may be attributed to individual-specific variation (Table S4). These data suggest that TNFα differentially affects metabolism of AB-versus NAB-HMECs.
While in AB group glycolysis is affected in NAB group the respiration is more affected. Thus bioenergetic responses of epithelial cells from tumoraffected and non-affected breasts are different to TNFα.

| Analysis of the IGF1 effects on carbon substrates by metabolic fingerprinting
To determine how IGF1 alters carbon substrate utilization in treated cells, we used 96-well microarrays with different carbon sources PM-M1 from BioLog Inc. (Hayward, CA, Table S5). Use of individual

c) Effect of
TNFα on mitochondrial bioenergetics of AB-HMECs (n = 16, mean ± sem). d) Effect of TNFα on mitochondrial bioenergetics of NAB-HMECs (n = 16, mean ± SD, OCR/pmoles/µg protein). e) Effect of TNFα on respiration rates supporting different bioenergetic parameters in AB-HMECs (n = 16, mean ± SD, OCR/pmoles/µg protein, *p ≤ 0.05). f) Effect of TNFα on respiration rates supporting different bioenergetic parameters in NAB-HMECs (n = 16, mean ± SD, OCR/pmoles/µg protein, **p ≤ 0.01, *** ≤ 0.001). g) Effect of TNFα on PySR in AB-HMECs (n = 16, mean ± sem). h) Effect of TNFα on PySR in NAB-HMECs (n = 16, mean ± sem). i) Effect of TNFα on basal PPR in AB-HMECs (n = 16, mean ± SD, Control vs. TNFα *p ≤ 0.05). j) Effect of TNFα on basal PPR in NAB-HMECs (n = 16, mean ± SD, Control vs. TNFα, ** ≤ 0.01). k) Effect of IGF1 on PPR in AB-HMECs after oligomycin addition (n = 16, mean ± SD, Control vs. TNFα *p ≤ 0.05). l) Effect of TNFα on PPR in NAB-HMECs after oligomycin addition (n = 16, mean ± SD, Control vs. TNFα) substrates were scored by monitoring the absorbance of redox dye MA, which develops purple color in the presence of NADH and NADPH without affecting cell viability (Bochner et al., 2011). It does not develop color in the presence of NAD + and NADP + (data not shown, N. Yadava & B. Bochner). Therefore, it informs about overall anabolic and catabolic metabolism of a given substrate indirectly by NADH and NADPH production. Figure 7a shows representative maps for control and IGF1-treated cells. In IGF1-treated cells, glucose use was significantly reduced as revealed by reduced absorbance at 590 nm ( Figure 7b). We also compared the rate of dye reduction by control and IGF1-treated HMECs from 6 women within 30-60 min of dye addition. Four of these 6 samples showed 12-31% reduced rate of dye reduction. This suggests that IGF1 reduces glucose-supported NADH/NADPH production in the majority of HMECs. Together these data suggest that IGF1 can alter glucose use in normal breast epithelial cells, and its response can vary in different individuals.

| Effects of IGF1 on radiation-induced cell death
To determine the biological relevance of cytokines effects on metabolism of HMECs, we determined radiation-induced death in control versus IGF1 treated cells. We measured the response of IGF1 on radiation-induced cell death on a set of randomly selected samples, which consisted of 3 AB and 4 NAB samples. Our data demonstrates that IGF1 significantly suppresses radiation-induced death in 57% (4/7) of samples, and there was a trend toward a reduction in all but  hydratase (FH) provide strong evidence for the role of pre-existing metabolic differences in susceptibility to tumorigenesis (Baysal, Rubinstein, & Taschner, 2001;Evenepoel et al., 2015).

Mitochondrial metabolism can vary from one individual to
another. This is because a large number of genes control structure, function, and regulation of the enzymes involved in the TCA cycle and Cells from individual donors may differ in many respects. One of these is differences in metabolism as revealed by this study. Parity and genetic background may play a role in individual differences. Differences in cell populations could also be a factor to consider (Linnemann et al., 2015). Thus there is a possibility that the percentage of basal to luminal cells differ in the early passage HMECs from women with cancer versus without cancer. Such variation in HMECs preparation may contribute to noted individual-specific metabolic differences.
Thus, future studies looking at purified subpopulations may be required to tease out this potential confounding factor before deciphering the role of genetic variations in nuclear and mitochondrial components of the oxidative metabolism.
In most cells, glucose is the primary fuel as well as the carbon source for biosynthesis. If glucose is completely oxidized for ATP synthesis, then all its carbons will be released as CO 2 . Therefore, cells balance the use of glucose carbon for bioenergetics versus biosynthesis. At the center of this balance lies the pyruvate metabolism that connects all three aspects of metabolic reprograming, that is, bioenergetics, redox balance, and biosynthesis. Cytosolic NAD + regeneration from NADH depends on pyruvate to lactate conversion, NADH redox shuttles, and its oxidation inside mitochondria to support respiratory chain function.
Changes in NAD + /NADH redox within cytosol and mitochondria can alter pyruvate production, secretion, and oxidation (Titov et al., 2016).  (Bricker et al., 2012;Herzig et al., 2012). MPC1 is often lost in cancer cells, and its expression is linked with anti-proliferative phenotype (Schell et al., 2014). The cells with reduced MPCs rely on glutamine anaplerosis to feed the TCA cycle without any difference in glucose and glutamine uptake. They also secret pyruvate and reprogram metabolism to promote lipid synthesis and branched chain amino acids oxidation. In many cancers the PK-M2 isoform, which is negatively regulated by post-translational modifications via growth factor signaling, is four-to six-fold higher than the PK-M1 isoform. Thus cells expressing PK-M2 isoform can limit the production of pyruvate and divert it away from oxidation to support biosynthesis. In PySR +ve cells either pyruvate production from glucose is reduced or pyruvate is diverted away from the oxidation. Because exogenous pyruvate increases respiration, a low mitochondrial pyruvate carrier is not expected in PySR +ve cells for the limitations in the respiratory capacity on glucose as sole fuel. On the other hand, in PySR −ve cells glucose itself is able to support maximal respiratory capacity. That means the cells are also not limited in mitochondrial pyruvate carrier.
Alterations in pyruvate oxidation have been linked with susceptibility to oncogene-induced senescence, a protective mechanism against tumorigenesis (Kaplon et al., 2013). Therefore the observation of difference in pyruvate oxidation in cells from women with and without cancer is highly relevant to breast cancer risk. One of the potential mechanisms for breast cancer susceptibility due to BRCA1 mutations is by alterations in cellular metabolism. A recent study suggests that BRCA1 haploinsufficiency (BRCA1 +/− ) decreases intracellular pyruvate by 78% and favors biosynthesis of lipids and branched-chain amino acids (Cuyas et al., 2016). The exit of pyruvate as citrate for lipid synthesis could be a potential cause for reduced intracellular pyruvate. The PySR −ve/+ve state of BRCA1 +/− cells is unknown. These cells may be PySR −ve if exogenous pyruvate will be directed toward anabolism without producing a significant difference in respiratory response. It may also result in altered glutamine metabolism. Decreased pyruvate flux in mitochondria is linked with increased glutamine use (Le et al., 2012;Metallo et al., 2011).
The difference in exogenous pyruvate oxidation and respiratory capacity of cells clearly indicates variation in mitochondrial metabolism.
This can involve oxidative metabolism at the levels of physical contents of the complexes of oxidative phosphorylation system, the TCA cycle, and redox metabolism. The impairments of oxidative metabolism can impair signaling pathways that promote tumorigenesis (Yadava, Schneider, Jerry, & Kim, 2013). Respiratory chain impairments suppress tumor suppressor protein p53 and provide protection against radiationinduced death (Compton et al., 2011). Respiratory chain impairments also genetically inactivate p53 in neural stem cells (Bartesaghi et al., 2015). P53 is also connected to mitochondria by estrogen signaling, which plays a key role in breast tumorigenesis (Wickramasekera & Das, 2014). Since p53 pathway is the most potent tumor suppressing pathway, its suppression can predispose to breast tumorigenesis in epithelial cells with compromised respiratory activity. Mitochondrial metabolism is also linked with suppression of tumorigenic property of metastatic cells (Kaipparettu et al., 2013). Because pyruvate metabolism influences oncogene-induced senescence, a protective mechanism against tumorigenesis (Kaplon et al., 2013), the pre-existing differences in pyruvate metabolism is relevant to breast cancer susceptibility.
In summary, our study supports the existence of inter-individual variation in cellular bioenergetics. This is primarily reflected in respiratory activity and its response to exogenous pyruvate. The difference in exogenous pyruvate oxidation detected as pyruvatestimulated respiration (PySR) is an interesting finding of the metabolic differences present in normal breast epithelial cells from different women. Based on the relative frequency of the PySR −ve phenotype between cells from cancer patients and women without any cancer history, we predict that PySR −ve phenotype is linked with breast cancer risk. Furthermore, cytokines such as IGF1 and TNFα can alter mammary epithelial cells bioenergetics as host factors. The response to cytokines is also variable among breast epithelial cells from different women. Together, the bioenergetic variation and its response to cytokines may alter susceptibility to oncogenic transformation.