Sperm-specific COX6B2 enhances oxidative phosphorylation, proliferation, and survival in lung adenocarcinoma

Cancer testis antigens (CTAs) are genes whose expression is normally restricted to the testis but anomalously activated in cancer. In sperm, a number of CTAs promote energy generation, however whether these proteins contribute to tumor cell metabolism is not understood. Here we describe COX6B2, a sperm-specific component of cytochrome c oxidase (complex IV). COX6B2 is frequently expressed in human lung adenocarcinoma (LUAD) and expression correlates with reduced survival time in patients. COX6B2, but not its somatic isoform COX6B1, enhances activity of complex IV, increasing mitochondrial oxidative phosphorylation (OXPHOS) and NAD+ generation. Consequently, COX6B2-expressing cells display a proliferative advantage, particularly in low oxygen conditions. Conversely, depletion of COX6B2 attenuates OXPHOS and collapses mitochondrial membrane potential leading to cell death or senescence. Furthermore, COX6B2 is both necessary and sufficient for growth of tumors in vivo. Our findings reveal a previously unappreciated, tumor specific metabolic pathway hijacked from one of the most ATP-intensive processes in the animal kingdom: sperm motility.

(LUAD) and expression correlates with reduced survival time in patients. COX6B2, but 23 not its somatic isoform COX6B1, enhances activity of complex IV, increasing 24 mitochondrial oxidative phosphorylation (OXPHOS) and NAD + generation. 25 Consequently, COX6B2-expressing cells display a proliferative advantage, particularly 26 in low oxygen conditions. Conversely, depletion of COX6B2 attenuates OXPHOS and 27 collapses mitochondrial membrane potential leading to cell death or senescence. 28 Furthermore, COX6B2 is both necessary and sufficient for growth of tumors in vivo. Our 29 findings reveal a previously unappreciated, tumor specific metabolic pathway hijacked 30 from one of the most ATP-intensive processes in the animal kingdom: sperm motility.

INTRODUCTION 33
Tumors frequently activate genes whose expression is otherwise restricted to testis; these 34 genes are known collectively as cancer-testis antigens (CT-antigens, CTAs). The biased expression pattern of these proteins to tumors and testis may offer an 46 extraordinarily broad therapeutic window if they can be targeted directly. 47 One unique aspect of mammalian sperm physiology is the tremendous energy demand 48 required for motility while preserving the integrity of their precious DNA cargo within the 49 hostile environment of the female reproductive tract. To meet this demand, sperm contain 50 a number of tissue specific protein isoforms for glycolysis and oxidative phosphorylation 51 (OXPHOS) that mediate the increase in ATP production. For example, lactate 52 dehydrogenase C, LDHC, is a testis-specific isoform of the terminal enzyme in glycolysis 53 that catalyzes the reduction of pyruvate to lactate essential for male fertility (Odet et al. 54 2008; Krisfalusi et al. 2006; Li et al. 1989). Similarly, COX6B2 is a testis-specific subunit 55 of cytochrome c oxidase (complex IV) (Huttemann, Jaradat, and Grossman 2003). mRNA 56 encoding of either LDHC or COX6B2 is undetectable in normal tissue, but both are 57 upregulated in a number of different tumor derived cell lines, classifying them as CTAs 58 ( (Maxfield et al. 2015), (CTpedia (http://www.cta.lncc.br/index.php)). However, it is 59 unclear whether these proteins support metabolic programs in tumor cells. 60 In a large-scale loss of function analysis to annotate the contribution of individual CTAs 61 to neoplastic behaviors, we found that COX6B2 is essential for survival of non-small cell 62 lung cancer (NSCLC) cell lines (Maxfield et al. 2015). COX6B2 is a nuclear encoded, 63 sperm-specific component of complex IV (Huttemann, Jaradat, and Grossman 2003). By 64 transferring electrons from reduced cytochrome c to O2, complex IV is the rate-limiting 65 step for ATP production by the electron transport chain (ETC). Thirteen subunits make 66 up complex IV: three are mitochondrial encoded and ten are derived from nuclear DNA. 67 Six of the ten nuclear encoded subunits have tissue-specific isoforms that permit 68 regulation of this complex in response to environmental cues (e.g. pH, hormones, metals, 69 ATP/ADP ratio etc.) (Kadenbach and Huttemann 2015). The somatic isoform of COX6B2 70 is COX6B1. These two proteins share 58% amino acid identity and ~80% similarity. 71 Based on structural information, it is apparent that COX6B1/2 are the only complex IV 72 subunits that are not membrane bound. Instead, their localization is confined to the 73 intermembrane space where cytochrome c associates. It is inferred that COX6B1/B2 74 participate in dimerization of complex IV and cytochrome c association (Sampson and 75 abundance of complex IV as judged by COXIV accumulation (Figure 3-figure  168 supplement 1A). This finding suggests that COX6B2 is a rate limiting subunit of 169 Recently, a number of reports have suggested that tissue specific complex IV 171 isoforms may promote supercomplex formation and selectively and specifically increase 172 ATP production in tumor cells (Maranzana et  Here, we observed that >90% of endogenous COX6B2 is found in in complex IV dimers 180 or incorporated into supercomplexes. In contrast, >98% COX6B1 is found in monomeric when either isoform is overexpressed ( >85% of COX6B2 is in complex IV in 183 dimers/supercomplexes versus 40% for COX6B1) ( Figure 3F). 184 We next tested the polymeric distribution of complex IV, using analysis of COXIV, 185 a core protein subunit of complex IV, following COX6B1-V5 or COX6B2-V5 expression. 186 Here, we found that upon expression of COX6B2-V5, but not COX6B1-V5, the 187 distribution of complex IV shifted from monomers to increased formation of 188 supercomplexes ( Figure 3G). Supercomplexes are hypothesized to enhance electron 189 transport activity without a compensatory increase in ROS (Maranzana et al. 2013). Mitochondrial OXPHOS has been linked to proliferation via regulation of the NAD + /NADH 207 ratio, which is elevated in COX6B2-V5 overexpressing cells ( Figure 2H) (Birsoy et al. 208 2015; Sullivan et al. 2015). Consistent with this model, supplementation with pyruvate, an 209 electron acceptor which drives NAD + production, largely diminishes the proliferative 210 effects of COX6B2-V5 overexpression ( Figure 4B). A similar proliferative advantage is 211 also observed when COX6B2-V5 cells are grown in the absence of extracellular matrix, 212 suggesting that COX6B2 promotes anchorage-independent growth ( Figure 4C). 213 Isoforms of complex IV subunits can be selectively expressed to permit adaptation to 214 tissue-specific needs or environmental conditions (Sinkler et al. 2017  conditions. Given this phenotype, we examined whether COX6B2 is regulated in 223 response to low oxygen. ChIP data from human breast cancer cells indicates that HIF-1 224 is bound near the COX6B2 promoter (Zhang et al. 2015). In addition, we also identified 225 11 putative HREs (A/G-CGTG) which are enriched at the HIF-1 peaks (Figure 4-figure  226 supplement 1D). Indeed, we observed a greater than two-fold increase in COX6B2 but 227 not COX6B1 mRNA following 12 hours of 2.5% oxygen ( Figure 4E). This accumulation 228 of mRNA was associated with stabilization of COX6B2 protein ( Figure 4F). Under both 229 normoxic or hypoxic conditions, we did not observe any significant alteration in COX6B2  promoting key tumorigenic behaviors including unrestrained proliferation, anchorage 235 independent growth and enhanced survival in hypoxia. 236

LUAD cells 239
Given the pro-proliferative phenotype following COX6B2 expression, we next examined supplement 1C). The reduction in OCR following COX6B2 depletion is accompanied by 249 a reduction in the NAD + /NADH ratio ( Figure 5C). In addition, total cellular ATP is 250 decreased following siRNA-mediated COX6B2 depletion with two independent siRNA 251 sequences across a panel of COX6B2-expressing LUAD cell lines ( Figure 5D). 252 Consistent with a loss of electron transport chain function, depletion of COX6B2 resulted 253 in decreased mitochondrial membrane potential ( Figure 5E) and increased intracellular 254 hydrogen peroxide ( Figure 5F). Collectively, these findings indicate that COX6B2 is 255 essential for respiration and mitochondrial integrity in tumor cells. Based on the observed decrease in OCR, we asked whether tumor cell viability was 260 diminished following COX6B2 depletion in LUAD cells. Depletion of COX6B2 for nine 261 days, results in a decrease in EdU incorporation ( Figure 6A). Furthermore, these cells 262 are severely compromised in their ability to grow in the absence of an extracellular matrix 263 ( Figure 6B). Both of these phenotypes were recapitulated in sgCOX6B2 cells (Figure 6-264 figure supplement 1A-B). We next examined the underlying mechanism(s) leading to 265 the observed loss of viability. In our original screen, we found that siRNA-mediated 266 depletion of COX6B2 enhances cleavage of caspase 3/7, a phenotype that is rescued in COX6B2 function has remained obscure in both tumor cells and sperm since its 303 discovery (Huttemann, Jaradat, and Grossman 2003). Our data indicate that COX6B2 304 enhances complex IV activity leading to an increase in mitochondrial oxidative 305 phosphorylation. The function of COX6B2 appears to differ from COX6B1, which did not 306 enhance OXPHOS when overexpressed, a finding consistent with previous biochemical 307 analyses that indicate removal of COX6B1 enhances complex IV activity (Weishaupt and 308 Kadenbach 1992). Genetic studies demonstrate that human mutations in COX6B1 reduce 309 assembly of complex IV and other ETC complexes. Based on this cumulative data, we 310 hypothesize that COX6B1 (and perhaps B2) are essential for assembly and regulation of 311 complex IV. In normal tissues, COX6B1 may regulate complex IV to prevent the 312 generation of excessive and unused energy and also damaging ROS. In sperm, where 313 ATP demand is high, COX6B2 expression could promote complex IV dimerization and 314 incorporation into supercomplexes to promote efficient OXPHOS and limit ROS 315 production that would otherwise damage DNA (Agarwal et al. 2014). This function may 316 be particularly important during early tumorigenesis, where a highly hypoxic environment 317 can lead to proliferative arrest and excessive ROS generation. We propose that tumor 318 cells that express and engage the COX6B2-based, sperm-specific mechanism can 319 overcome this innate barrier to transformation that would otherwise restrict cell survival. 320 One of the most surprising aspects of our study was the loss of tumor cell viability 321 following the depletion of COX6B2. As COX6B2 enhances OXPHOS, its loss would be 322 expected to reduce ATP levels and the NAD + /NADH ratio and slow proliferation, but not 323 necessarily kill tumor cells. Tumor cell glycolytic pathways should theoretically 324 compensate for reduced energy generation. Moreover, the presence of COX6B1 should 325 be able to support complex IV function. Instead, we observe ROS production and an 326 exit from the cell cycle either to death or senescence. This finding implies that these 327 cells are highly dependent upon OXPHOS for survival. In addition, the loss of COX6B2 328 may promote apoptosis due to a collapse of mitochondrial membrane potential and

Cell lines 365
All NSCLC cell lines were obtained from John Minna (UT Southwestern) between 2014 366 and 2018. NSCLC cells were cultured in RPMI media supplemented with 5% FBS at 37°C, 367 5% CO2 and 90% humidity. NSCLC cells were not passaged more than 10 times after 368 thawing. Cells cultured under hypoxia followed the methods described previously (Wright 369 and

Expression plasmids 458
Human COX6B2 and COX6B1 were obtained in pDONR223 (DNASU) and cloned into 459 pLX302 and pLX304 respectively using the Gateway ® Cloning system (Thermo Fisher 460 Scientific). psPAX2 and pMD2.G lentiviral constructs were purchased from Addgene.

Proliferation measurements 558
Cell proliferation was measured as described (Gui et al. 2016). HCC515 and H2122 cells 559 were plated in replicates in 96-well plate with an initial seeding density of 1,200 cells and 560 700 cells respectively. The proliferation rate was calculated using the following formula: 561 Proliferation Rate (Doublings per day) = Log2 (Final cell count (day 5)/Initial cell count 562 (day 1))/4 (days). 563 564

Soft-agar assays 565
Cells were seeded at a density of 50,000 cells (HCC515) or 5000 cells (H2122) per 12-566 well plate and treated as previously described (Gallegos et al. 2019). After three weeks, 567 colonies were stained overnight with 0.01% crystal violet. Images were captured with a 568 dissecting microscope, and quantitated by ImageJ software. 569

Gene expression 571
RNA was isolated using a mammalian total RNA isolation kit (RTN350, Sigma) and 572