Mitochondrial metabolism in primary and metastatic human kidney cancers

Summary Most kidney cancers display evidence of metabolic dysfunction1–4 but how this relates to cancer progression in humans is unknown. We used a multidisciplinary approach to infuse 13C-labeled nutrients during surgical tumour resection in over 70 patients with kidney cancer. Labeling from [U-13C]glucose varies across cancer subtypes, indicating that the kidney environment alone cannot account for all metabolic reprogramming in these tumours. Compared to the adjacent kidney, clear cell renal cell carcinomas (ccRCC) display suppressed labelling of tricarboxylic acid (TCA) cycle intermediates in vivo and in organotypic slices cultured ex vivo, indicating that suppressed labeling is tissue intrinsic. Infusions of [1,2-13C]acetate and [U-13C]glutamine in patients, coupled with respiratory flux of mitochondria isolated from kidney and tumour tissue, reveal primary defects in mitochondrial function in human ccRCC. However, ccRCC metastases unexpectedly have enhanced labeling of TCA cycle intermediates compared to primary ccRCCs, indicating a divergent metabolic program during ccRCC metastasis in patients. In mice, stimulating respiration in ccRCC cells is sufficient to promote metastatic colonization. Altogether, these findings indicate that metabolic properties evolve during human kidney cancer progression, and suggest that mitochondrial respiration may be limiting for ccRCC metastasis but not for ccRCC growth at the site of origin.

3 mutations in metabolic enzymes like fumarate hydratase (FH) and succinate 1 dehydrogenase (SDH) are initiating events in FH deficient renal cell cancer (RCC) 13 and 2 SDH-deficient RCC 14 , respectively. Although many tumours originating in the kidney 3 display mitochondrial dysfunction, it is unclear how these mitochondrial anomalies 4 impact nutrient metabolism in humans. 5 Intra-operative infusion of 13 C-labeled nutrients and subsequent metabolite 6 extraction and analysis of 13 C labelling from surgically-resected samples can reveal 7 metabolic differences between tumours and adjacent tissue and among different 8 tumours from the same organ 15,16 . We previously reported suppressed contribution of 9 glucose carbon to TCA cycle intermediates in five human ccRCCs, implying reduced 10 glucose oxidation in these tumours. Here we studied why this phenotype occurs in 11 human ccRCC, whether it characterizes kidney tumours more generally, and whether 12 metabolic properties evolve during ccRCC progression to distant metastatic disease in 13 patients. We infused patients with 13 C-glucose, 13 C-acetate and 13 C-glutamine, 14 capitalizing on the complementary views of the TCA cycle provided by these nutrients to 15 produce a detailed analysis of mitochondrial metabolism in human cancer.  Table 2). The labeling ratio of citrate m+2 (i.e. the fraction of citrate molecules 30 containing two 13 C nuclei) to pyruvate m+3 was lower in ccRCC samples compared to 31 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023.  Table 3). Labeling in the 3 tumours and renal cortex (hereafter, adjacent kidney) samples was variable, reflecting 4 both inter-patient variability and regional labeling differences among samples from the 5 same patient (Fig. 1D). When tumour labelling was compared to the adjacent kidney 6 labelling in the same patient, only 1 of 28 patients with ccRCC displayed a statistically 7 significant increase in the citrate m+2/pyruvate m+3 ratio in the tumour (Extended Data 8 Fig. 1B). Ten patients did not have matched adjacent kidney tissue available for 9 analysis. In addition to suppressed citrate m+2/pyruvate m+3 labeling ratios, total 10 labeling of citrate and other TCA cycle intermediates (1-(m+0)) was also suppressed in  Acetate and glutamine supply the TCA cycle in ccRCC 24 We next infused 12 ccRCC patients with [1,2-13 C]acetate (m+2), which can be 25 converted to acetyl-CoA m+2 by acetyl-CoA synthetases (ACSS1/2, Fig. 2A). This 26 tracer is useful for two reasons in this context. Unlike pyruvate, which can enter the TCA 27 cycle through both acetyl-CoA and oxaloacetate (OAA) and produces complex labeling 28 on even the first TCA cycle turn 18 , acetate only enters the TCA cycle through acetyl- 29 CoA. This exclusively produces m+2 labeling in the first turn. Second, [1,[2][3][4][5][6][7][8][9][10][11][12][13] C]acetate 30 transmits 13 C to the TCA cycle independently of PDH, and so it is an informative 31 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.06.527285 doi: bioRxiv preprint 5 complement to tracers like [U-13 C]glucose that produce 13 C-pyruvate, the substrate of 1 PDH. The conditions we used to infuse [1,[2][3][4][5][6][7][8][9][10][11][12][13] C]acetate did not alter acetyl-CoA levels 2 in tumours or adjacent kidneys and produced similar levels of acetyl-CoA labeling in 3 both tissues (Fig. 2B,C, see Extended Data Table 4 for full isotopologue distributions). 4 Fractional enrichments of m+2 TCA cycle intermediates in ccRCC tumours were also 5 similar to adjacent kidney (Fig. 2D), indicating similar contributions to the TCA cycle 6 under these infusion conditions. However, total labeling (1-(m+0)) of these metabolites 7 revealed decreased labeling in tumours compared to kidney, consistent with reduced 8 labeling beyond turn 1 of the TCA cycle ( Fig 2E). 9 We then examined TCA cycle turnover in three complementary ways. First, the 10 high enrichment in acetyl-CoA (average of 20-25%) allowed us to observe higher-order 11 labeling in TCA cycle intermediates from subsequent rounds of incorporation of acetyl-

12
CoA m+2 (Fig. 2F). The ratio of citrate m+4/m+2, a marker of 13 C retention through two 13 cycles, was reduced by about half in tumours relative to kidneys (Fig. 2G). Second, we 14 examined labeling of TCA cycle intermediates in fresh mitochondria isolated from these 15 resected tissues and cultured with [U-13 C]pyruvate. Both the citrate m+2/pyruvate m+3 16 and citrate m+2/citrate m+4 ratios were decreased in the ccRCC mitochondria 17 compared to kidney mitochondria (Fig. 2H), indicating that these metabolic properties 18 are intrinsic to ccRCC mitochondria.

19
Third, we examined positional 13 C labeling in glutamate, which exchanges with α-20 ketoglutarate and is classically used as a reporter of TCA cycle metabolism ( Fig. 2A)  we determined that [4,5-13 C]glutamate, which appears in the first turn of the cycle (Fig.   28 2A), accounts for a much higher fraction of glutamate m+2 in tumours compared to 29 adjacent kidney (Fig 2I). Therefore, most glutamate m+2 in ccRCC tumours comes from 30 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made To assess the TCA cycle using a third tracer, we infused seven ccRCC patients 3 with [U-13 C]glutamine. Glutamine is the most abundant amino acid in the circulation, and 4 its uptake in the tumour microenvironment is reported to be dominated by malignant 5 cells 23 . Glutamine's contributions to the TCA cycle involve conversion to alpha-6 ketoglutarate (α-KG) followed by either oxidation through α-KG dehydrogenase or 7 reductive carboxylation by isocitrate dehydrogenase-1 or -2 24,25 . In cell culture, labeling 8 through reductive metabolism is enhanced by processes that suppress pyruvate 9 oxidation, including VHL loss, PDH suppression and mitochondrial defects 26-28 . Isotope 10 labeling in citrate and other TCA cycle intermediates can discriminate which pathway is 11 being utilized (Fig. 3A). The patient infusions produced the same glutamine m+5 12 enrichment in tumours and adjacent kidneys (30-35%, Fig. 3B). Labeling of glutamate 13 m+5 and TCA cycle intermediates from the first turn of the oxidative TCA cycle (m+4) 14 were also similar between tumour and kidney (Fig. 3B). However, total labeling (1-15 (m+0)) of these metabolites was higher in the tumours (Fig. 3C, see Extended Data 16 Table 5 for full isotopologue distributions). The additional labeling in TCA cycle 17 metabolites from the tumours involved enhanced contributions from the reductive 18 pathway, as indicated by high citrate m+5 labeling in most fragments (Fig. 3D). This 19 level of labeling far exceeded labeling in plasma citrate, indicating that it resulted from 20 metabolism in the tumour (Extended Data Fig. 4A). The tumours also contained 21 relatively high levels of malate m+3, indicating further metabolism along the reductive 22 pathway (Fig. 3E). Therefore, glutamine is a carbon source in human ccRCC, and its 23 metabolism results in oxidative and reductive labeling of TCA cycle intermediates.  CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. 7 cohort both display reduced mRNA expression of ETC-related genes in ccRCC tumours 1 relative to adjacent kidney, whereas many glycolytic genes are overexpressed in the 2 tumours (Extended Data Fig. 1A). However, none of these analyses directly assessed 3 coupled respiration in mitochondria from tumours and kidneys. We therefore measured 4 oxygen consumption rates (OCR) of mitochondria immediately after harvesting them 5 from fresh, surgically-resected kidney and tumour tissues. We used a differential 6 centrifugation protocol to isolate mitochondria, then assessed ADP-stimulated (State III) 7 and unstimulated (State IV) respiration. Mitochondria from both the kidney and ccRCC 8 had normal respiratory control ratios (RCR, defined as State III/State IV respiration) 9 when supplied with Complex I substrates, indicating that the preparation produced Complex I, II, and IV was always lower in mitochondria isolated from tumours (Fig. 4B).

17
Mitochondria from other RCC subtypes displayed low state III respiration at 18 Complex I, but variable activities of other ETC components (Fig. 4A). Chromophobe 19 tumours and oncocytomas contain mutations in genes encoding Complex I subunits, 20 and accordingly both had low Complex I activity relative to adjacent kidney ( Fig 4A).

21
The RCRs of these mitochondria were also low when provided with Complex I 22 substrates (Extended Data Fig. 5D). However, absolute state III OCRs for Complex II 23 and IV were variable, and in mitochondria from oncocytomas, they exceeded rates from 24 kidney mitochondria. Therefore, oncocytomas and chromophobe tumours display the 25 expected defects in Complex I, with relative preservation of some other ETC  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023.  To directly examine metabolism in metastatic ccRCC, [U-13 C]glucose was 5 infused in 10 patients undergoing metastasectomy. Metastatic tumours in 9 of these 10 6 patients had higher citrate m+2/pyruvate m+3 ratios than the average citrate 7 m+2/pyruvate m+3 ratio from primary ccRCCs from the kidney (Fig. 5A). Two patients 8 with a primary ccRCC and a synchronous adrenal metastasis underwent concurrent 9 nephrectomy and adrenalectomy, allowing both lesions to be sampled during the same 10 infusion. Compared to the primary lesion, the metastatic adrenal tumours trended 11 towards higher citrate m+2/pyruvate m+3 ratios compared to the primary tumour ( Fig.   12 5B). In patient 2, two different regions of the primary tumour were sampled, with one 13 region having a reduced citrate m+2/pyruvate m+3 ratio relative to the other region; both 14 these regions had somewhat lower ratios than the metastasis (Fig 5B). Two patients 15 with metastatic tumours were infused with [1,2-13 C]acetate, and these tumours also  The low apparent oxidative metabolism in primary ccRCCs provided an 26 opportunity to test whether stimulating respiration would enhance metastatic spread. 27 We expressed the yeast mitochondrial NADH dehydrogenase NDI1 in VHL-deficient  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. labeling, but NDI1-expressing cells had among the top 5% of citrate labeling (Fig. 5E).

6
To assess the impact of NDI1 on metastatic colonization, the cells were engineered to 7 express dsRed-luciferase and transplanted into immune compromised mice via the tail 8 vein. Bioluminescence imaging revealed that NDI1 expression induced a large increase 9 in lung colonization and growth as compared to 786-O cells expressing the empty 10 vector ( Fig 5F). Therefore, human metastatic ccRCCs display evidence of enhanced 11 mitochondrial metabolism in patients, and ccRCC cells engineered to have increased 12 oxidative phosphorylation display increased metastatic colonization in mice. with primary ccRCCs having suppressed TCA cycle turnover relative to adjacent kidney.

25
While PDH suppression is a well-known effect of HIF-1α activation, we also report 26 dysfunction of multiple ETC components manifesting as reduced mitochondrial 27 respiration. This finding may be related to suppressed mtDNA copy number in ccRCC 30 , 28 and it predicts that activating PDH would not be sufficient to normalize oxidative 29 metabolism in ccRCC. Second, we report higher contributions of glucose to the TCA 30 cycle in metastatic ccRCC compared to primary ccRCC. This was observed in both 31 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.06.527285 doi: bioRxiv preprint synchronous and asynchronous metastases, in multiple metastatic sites, and it implies 1 an evolution or selection of mitochondrial function during ccRCC metastasis in patients.  Other studies that did not focus explicitly on metastasis have also reported the 16 differential importance of oxidative phosphorylation in advanced cancers. In a mouse 17 model of pancreatic ductal adenocarcinoma, oxidative phosphorylation underlies 18 relapse and outgrowth after genetic ablation of the oncogenic driver. In these mice, 19 relapse is suppressed and survival is enhanced by inhibiting the ETC 54 . In acute 20 myelogenous leukemia, human-derived mouse models with robust oxidative 21 phosphorylation display resistance to cytotoxic chemotherapy, and this resistance is  Efforts to suppress cancer progression by targeting mitochondrial metabolism will 30 benefit from understanding the basis of the relationship between the mitochondria and 31 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. metastasis. It is unclear why ccRCC metastases in patients bear hallmarks of enhanced 1 mitochondrial function, because neither the mtDNA content nor the levels of transcripts 2 related to oxidative phosphorylation differed between primary and metastatic ccRCC in 3 our cohort. Perhaps the most interesting and important challenge arising from this work 4 is to determine which metabolic effects of mitochondrial function support metastasis.

5
The ETC supports efficient ATP production from nutrient oxidation, and this may be 6 essential to survive the reduced nutrient uptake that accompanies loss of 7 anchorage 38,57 . But the ETC also supports the maintenance of a favorable redox Acknowledgements 15 We are grateful to Gerardo Guevara for his efforts on this project. We thank Aron Jaffe  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023.   CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023.  Table 1.   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.06.527285 doi: bioRxiv preprint separated on an Acquity UPLC® HSS T3 column (1.8 μm, 2.1 x 150 mm, Waters, MA). 1 The column was kept at room temperature. Mobile phase A composition was 0.1% 2 formic acid in water and mobile phase B composition was 0.1% formic acid in 100%  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023.
15 precursor ion scans were acquired at a resolving power of 120,000 full width at half-    (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023. ; https://doi.org/10.1101/2023.02.06.527285 doi: bioRxiv preprint After surgery, kidney cortex and tumour fragments were embedded in 0.1% 1 agarose and sliced into ~300 µM thick sections using a microtome (Precisionary 2 Instruments, Copresstome, VF-300). These tissues were then transferred and 3 maintained on hydrophilic PTFE cell culture inserts in human plasma like medium 4 (HPLM) supplemented with 10% dialyzed human serum. Prior to tracing assays, tissues 5 were washed twice with 0.9% saline and medium was replaced with HPLM containing 6 [U-13 C]glucose for 3 hours. Slices were maintained in an incubator with 5% CO2, 5% O2, 7 and 90% N2.

No substrate
Isolation buffer (IB) . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made were washed twice with 0.9% saline and medium was replaced with RPMI-1640 27 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023.
containing [U-13 C]glucose supplemented with 5% dialyzed FBS for 6 hours. Cells were 1 rinsed in ice cold 0.9% saline and lysed with three freeze thaw cycles in cold 80% 2 methanol. Samples were then prepared for GC/MS analysis.   (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted February 7, 2023.  represent mean ± standard deviation. Statistical significance was assessed using 10 unpaired two tailed parametric t-tests. Adr, adrenal gland; LN, lymph node. ns P>0.05, 11 *P < 0.05, **P < 0.01, ***P < 0.001, ****P<0.0001.                       (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made