Similar deficiencies, different outcomes: succinate dehydrogenase loss in adrenal medulla vs. fibroblast cell culture models of paraganglioma

Abstract

Heterozygosity for loss-of‐function alleles of the genes encoding the four subunits of succinate dehydrogenase (SDHA, SDHB, SDHC, SDHD), as well as the SDHAF2 assembly factor predispose affected individuals to pheochromocytoma and paraganglioma (PPGL), two rare neuroendocrine tumors that arise from neural crest-derived paraganglia. Tumorigenesis results from loss of the remaining functional SDHx gene copy, leading to a cell with no functional SDH and a defective tricarboxylic acid (TCA) cycle. It is believed that the subsequent accumulation of succinate competitively inhibits multiple dioxygenase enzymes that normally suppress hypoxic signaling and demethylate histones and DNA, ultimately leading to increased expression of genes involved in angiogenesis and cell proliferation. Why SDH loss is selectively tumorigenic in neuroendocrine cells remains poorly understood. In the absence of SDH-loss tumor-derived cell models, the cellular burden of SDH loss and succinate accumulation have been investigated through conditional knockouts of SDH subunits in pre-existing murine or human cell lines with varying degrees of clinical relevance. Here we characterize two available murine SDH-loss cell lines, immortalized adrenally-derived premature chromaffin cells vs. immortalized fibroblasts, at a level of detail beyond that currently reported in the literature and with the intention of laying the foundation for future investigations into adaptive pathways and vulnerabilities in SDH-loss cells. We report different mechanistic and phenotypic manifestations of SDH subunit loss in the presented cellular contexts. These findings highlight similarities and differences in the cellular response to SDH loss between the two cell models. We show that adrenally-derived cells display more severe morphological cellular and mitochondrial alterations, yet are unique in preserving residual Complex I function, perhaps allowing them to better tolerate SDH loss, thus making them a closer model to SDH-loss PPGL relative to fibroblasts.

(281 words)

Introduction

Succinate dehydrogenase (SDH), is a crucial enzyme complex that catalyzes the oxidation of succinate to fumarate as part of the tricarboxylic acid (TCA) cycle, coupled to reduction of ubiquinone to ubiquinol as Complex II of the mitochondrial electron transport chain [1, 2]. SDH consists of four subunits: SDHA, SDHB, SDHC, and SDHD, and involves assembly factors required for covalent flavinylation of SDHA and iron-sulfur cluster modification of SDHB [3,4,5]. Deleterious variants in any of the genes encoding these proteins can disrupt SDH enzyme activity, affecting the mitochondrial respiratory chain and normal cellular energy production [6]. While SDH-loss cells are presumably metabolically disabled, it is fascinating that SDH loss can lead to tumors such as gastrointestinal stromal tumor (GIST), renal cell carcinoma (RCC), pituitary neuroendocrine tumors [7] and pheochromocytoma and paraganglioma (PPGL) [8,9,10,11,12,13]. In these rare tumors SDH-loss metabolism appears to be the primary cause of neoplasia.

Several mechanistic hypotheses have been proposed to connect SDH loss to PPGL tumorigenesis [14,15,16,17]. When SDH is deficient, the normal flow of metabolites through the TCA cycle is disrupted, leading to the accumulation of succinate. According to the succinate accumulation hypothesis [18], mitochondrial succinate gains access to the cytosol and nucleus via metabolite transporters and acts as an oncometabolite [18, 19]. Because it is a byproduct in the conserved reaction mechanism of dozens of 2-oxoglutarate-dependent dioxygenases, accumulated succinate can act a competitive inhibitor of these enzymes, causing a wide variety of cellular impacts. These effects include inhibition of dioxygenases involved in the normal degradation of hypoxia-inducible factors (HIFs), and inhibition of demethylation of histones, DNA, and RNA [20]. There is also evidence for succinyl-CoA accumulation upon SDH loss in some contexts, driving post-translational lysine hypersuccinylation [20]. The resulting dysregulation of metabolism and epigenetics is tumorigenic in certain tissues [21].

Upon SDH loss, chromaffin and glomus cells of the paraganglia can give rise to PPGL [22]. The basis for this cell-specificity in PPGL tumorigenesis following SDH loss is unknown. Chromaffin cells are primarily found in the adrenal medulla and are responsible for the production and secretion of catecholamines such as adrenaline and noradrenaline [23]. Extra-adrenal paraganglia, which include glomus cells and other cell clusters derived from neural crest progenitors, also contribute to PPGL and can be found in various locations such as the head, neck, chest, abdomen, and pelvis [24, 25]. It remains unclear what cellular features unique to these tissues sustain viability and proliferation of SDH-loss cells during PPGL tumorigenesis. Insights may shed light on potential therapeutic interventions. Moreover, understanding PPGL tissue specificity is important in the quest for relevant preclinical models [26,27,28].

In a previous study by Kľučková et al., WT and SDHB-loss immortalized mouse chromaffin cell lines (imCCs) were compared to WT and SDHB-deficient non-chromaffin cells to discern possible metabolic differences [29]. We note that while the literature designation of imCCs implies chromaffin cell function, our characterization shows these cells to lack chromaffin morphology (e.g. secretory granules) or classic chromaffin cell markers by RNA expression profile. We thus consider the cells to be “immortalized adrenal medulla-derived cells” possibly representing precursors of mature chromaffin cells. Here we confirm and significantly extend these prior results by comparison of these previously studied wild type (WT) and SDHB-loss imCCs with our previously created SDHC-loss immortalized mouse embryonic fibroblasts (iMEFs). We characterize cell morphologies, cell cycles and replication times, gene expression, and detailed metabolism in normoxia and hypoxia. As previously suggested [29], we confirm that SDHB-deficient imCCs retain detectable Complex I function in the absence of Complex II and appear less dependent on glycolysis than SDH-loss fibroblasts. The metabolic resilience of SDHB-loss imCCs is particularly striking in hypoxia. These results suggest possible survival advantages for SDH-loss imCCs vs. other SDH-loss cells, possibly hinting toward similar survival mechanisms in PPGL tumors. Our detailed comparative characterization sets the stage for unbiased screens to seek vulnerabilities of SDH-loss chromaffin cells.

Materials and methods

Cell culture

Unless otherwise specified, cell culture was performed at 37 °C, 95% humidity in room air (21% O₂) with 5% CO₂. Culture media consisted of high-glucose DMEM (Gibco # 11965118) with GlutaMAX™ (Gibco #10566016), 10% heat-inactivated FBS (Gibco #10082147) and a 0.5 mg/mL final concentration of penicillin/streptomycin (Gibco #15140122). Additional supplements included 1 mM sodium pyruvate (Gibco #11140035), 10 mM HEPES buffer (Gibco #15630130) and nonessential amino acids [100 µM final concentration each of glycine, alanine, asparagine, aspartic acid, glutamic acid, proline, and serine (Gibco #11140035)]. Cells were supplied with fresh media every other day and replated based on their doubling rate when 80–90% confluence was reached. Cells for analysis of dioxygenase inhibition by succinate and hypoxia Seahorse studies were grown in 10% and 5% room air, respectively. for hypoxia-inducible factor analysis, cultured cells were seeded 24 h prior to hypoxia treatment, then exposed to 21% (room air) or 10% O₂ in an incubator, or 1% O₂ in a sealed hypoxia chamber placed in a heated incubator for 24 h. All experiments were performed within 7 passages from the initial seeding of frozen stocks.

Original generation of SDH-loss imCCs and iMEFs

Wild type and Sdhb^−/− imCC lines were originally generated at the Paris Cardiovascular Research Center (PARCC) as previously described [30]. Briefly, P1 (129S2/SvPas) ES cells were transfected through electroporation with a targeting vector that included Sdhb exon 2 surrounded by LoxP sites, along with a neomycin (neo) selection cassette bordered by FRT sites. Two positive embryonic stem clones were selected to be injected into C57BL/6J blastocysts and create chimeras for germline transmission. Male transmitters were bred into a Flipase-expressing background to allow excision of the neo cassette. Derived C57BL/6J-Sdhb^+/fl mice were bred to generate Sdhb^fl/fl animals, allowing isolation of mouse chromaffin cells (mCC) from the adrenal medullae of these Sdhb^fl/fl mice following previously described methods [31]. Cells were maintained in a standard culture environment at 37°C with 5% CO₂, using Dulbecco’s modified Eagle’s medium with glutamine and high glucose (Gibco #10566016). The medium was supplemented with 10% fetal bovine serum (Gibco #10437028) and 1% penicillin-streptomycin (Gibco #15140122). Remaining quiescent for 6 months, some cultures spontaneously initiated cells growing as immortalized clusters (imCCs). These immortalized cells were isolated and subjected to infection with 10⁷ plaque-forming units (pfu)/mL of an Ad-CMV-iCre adenovirus expressing Cre recombinase (Cell BioLabs #1045). Cloning by limited dilution yielded two Sdhb^−/− imCC clones referenced as clone 6 and clone 8. Sdhb^−/− and Sdhb^+/+ imCC lines were cultured in standard medium. PCR genotyping of Sdhb status was determined by PCR using LJM-6539 (5’-C₂ATGC₂AGTA₂C₄AGA₂T₂G₂-3’) and LJM-6540 (5’-GTC₃AC₂TCACTGTAG₂TCAC-3’) as primers.

Immortalized mouse embryonic fibroblasts (iMEFs) lines were isolated from mice harboring a loxP gene trap construct from Sanger Institute sandwiching exon 4 of Sdhc as previously reported [32, 33]. Sdhc^fl/fl mice were introduced to the C57BL/6J-R26M^2rtTA/M2rtTA background to create mice in which Sdhc rearrangement could be induced by doxycycline-dependent expression of Cre recombinase via a tet-inducible promoter. After inducing rearrangement of the Sdhc gene by doxycycline treatment of Sdhc^fl/fl and Sdhc^+/fl iMEF lines for 5 d, cells were diluted into DMEM (Gibco # 11965118) supplemented with 0.5 mg/mL penicillin/streptomycin antibiotics (Gibco #15140122), MEM NEAA [containing glycine, alanine, asparagine, aspartic acid, glutamic acid, proline, and serine at a concentration of 100 µM (Gibco #11140035)], sodium pyruvate [1 mM (Gibco #11140035)], and HEPES buffer [10 mM (Gibco #15630130)]. Cloning was achieved by limited dilution. Incubation was at 37°C for 2 weeks before screening for colonies. Recovered clones were then expanded gradually from 12-well plates to T25 flasks in growth medium of the same composition. A small number of cells from each clone underwent DNA extraction and PCR analysis to determine the Sdhc gene rearrangement status. Clones exhibiting homogeneous Sdhc rearrangement were selected for expansion and establishing stocks of Sdhc^+/− and Sdhc^−/− iMEFs. Genotyping for CRE-recombined Sdhc product was carried using LJM-4429 (5’-CT₂AGA₂CTGATC₄TGC₃-3’) and LJM-5125 (5’- C₂TG₂A₂CTAGA₂T₂AT₂GATG₂ATG-3’).

Western blot analysis

Western blotting was used to confirm the loss of SDHB and SDHC proteins in imCCs and iMEFs, respectively. Cell pellets from all four lines (3 million cells each) were lysed in 150 µL cold RIPA buffer containing protease and phosphatase inhibitor cocktail (Santa Cruz Biotechnology # sc-24948). Sample lysates were then incubated on ice for 30 min with gentle vortex mixing every 10 min. Following centrifugation at 15,000×g for 15 min, protein quantification was carried out using a BCA protein assay kit (Pierce #A55864). Samples were heated to 70 °C for 10 min after the addition of reducing agent and an appropriate volume of 4× LDS denaturation buffer (Invitrogen #NP0007). Denatured samples (35 µg) were subjected to electrophoresis through NuPAGE 10% Bis-Tris protein gels (Invitrogen #NP0315BOX) in MES-SDS running buffer (Invitrogen #NP0002) at 150 V for 1 h. PVDF membrane transfer was performed according to the manufacturer’s protocol in a Novex Western Transfer Apparatus in NuPAGE transfer buffer containing 20% methanol. The transferring process was at 4 °C (30 V, 245 mA) for 90 min. The quality of transfer was confirmed by Ponceau S staining. Membranes were then blocked with 3% non-fat milk for 1 h at room temperature, followed by three 5-min washes in TBST buffer. All antibodies were purchased from Abcam, and antibody combinations were tested to confirm lack of cross-reactivity. A dilution buffer was prepared using 7.5 mL TBST, 2.5 mL 4% BSA, and 250 µL 0.5% sodium azide. Antibody against and PCNA (Abcam #ab29, 1:10,000) was used as loading control. Anti-SDHB (Abcam #ab175225, 1:5000) and anti-SDHC (Abcam #ab155999, 1:1000) antibodies were used to detect their protein targets at 32 kDa and 19 kDa, respectively. After 48 h of incubation at 4 °C, blots were washed three times in TBST before staining with IRDye^® 680rd goat anti-rabbit IgG (LI-COR #926-68071, 1:15,000) or 800cw goat anti-mouse IgG secondary antibody (LI-COR #926-32210, 1:15,000) antibodies in TBST with 3% non-fat milk for 1 h at room temperature prior to imaging with Amersham Typhoon 5 Biomolecular Imager (Amersham Typhoon, Uppsala, Sweden).

For immunoblotting analysis of HIF1/2α, DNA damage and mitochondrial complexes, cultured cells were harvested by washing with ice cold PBS, dissociated with ice cold trypsin-EDTA (Gibco 25300054), and neutralized with ice cold culture media. Cells were pelleted at 700×g and washed with ice cold PBS. Total cellular protein was extracted on ice using RIPA lysis buffer (ChemCruz sc-24948) with additional phosphatase inhibitors (ThermoFisher #78420) and soluble protein lysate was isolated after clarifying the extract at 15,000×g. Quantification of total soluble protein concentration was performed using the BCA protein assay according to the manufacturer’s protocol (Pierce #23225). Protein lysates were denatured and reduced in reducing agent (Invitrogen #NP0004) and LDS buffer (Invitrogen #NP0007) at 95ºC for 5 min immediately after samples protein levels were quantified and again prior to gel loading. Lysate (45 µg protein per sample) was loaded on 4–12% Bis-Tris gels (Invitrogen #NP0335BOX) and subjected to SDS-PAGE in reducing MES or MOPS buffer (Invitrogen #NP0002, NP0001 and NP0005). Proteins were then transferred to a PVDF membrane (Bio-Rad #1620174) in transfer buffer (Invitrogen #NP00061) containing either 10% methanol and 0.01% SDS for HIF1α and HIF2α detection, or 20% methanol for all other proteins. The membrane was then briefly stained with Ponceau S, imaged on a Typhoon 5 biomolecular imager (Amersham Typhoon, Uppsala, Sweden), destained with deionized water, and blocked in 5% non-fat dairy milk dissolved in TBST for 1–2 h at room temperature. The membrane was washed with TBST before incubating in primary antibodies diluted into 5% BSA dissolved in TBST overnight with gentle rocking at 4ºC overnight. The membrane was then washed with TBST prior to incubation with fluorescent secondary antibodies (LI-COR, IRDye 680RD goat anti-rabbit #926-68071, 1:20,000 and IRDye 800CW goat anti-mouse #926-32210, 1:20,000) with gentle rocking for 15–20 min at room temperature. The membrane was washed with TBST, then with TBS prior to imaging on the Typhoon 5. Primary antibodies used were nti-HIF1α (Cayman Chemical rabbit polyclonal 10006421; 1:1000), anti-HIF2α (Novus Biologicals mouse monoclonal NB100-132; 1:500), anti-α-tubulin (abcam mouse monoclonal ab7291; 1:50,000), anti-H3 (abcam mouse monoclonal ab10799; 1:1000), anti-H3K9me3 (Active Motif rabbit polyclonal 39162; 1:5000), Anti-H2A.X (Active Motif rabbit polyclonal 39690; 1:2000), anti-γH2A.X (phospho S139) (MilliporeSigma mouse monoclonal 05-636; 1:1000), anti-COX10 (CST #24744, 1:1000), anti-NDUFS1 (CST 70264, 1:1000) and anti-cytochrome c (CST 11940, 1:1000).

Histone isolation

Cultured cells were harvested by washing with ice cold PBS, dissociated with ice cold trypsin-EDTA (Gibco 25300054), and neutralized with ice cold culture media. Cells were pelleted at 700×g and washed with ice cold PBS. Nuclei were extracted by lysing cells at 4 °C in TEB (PBS containing 0.5% Triton X-100 and protease inhibitor cocktail) for 10 min with gentle agitation, then pelleted at 2000×g at 4 °C. Pelleted nuclei were washed twice with TEB by centrifugation at 1500×g. Nuclei pellets were frozen at -20ºC for later histone extraction. Histones were obtained by acid extraction for 1 h with 0.4 N H₂SO₄ and clarified at 15,000×g at 4ºC. Histones were then precipitated from the supernatant with 100% TCA to achieve a final TCA concentration of 33% v/v and incubated on ice for 30 min. Histones were pelleted at 15,000×g at 4ºC, washed twice with ice cold acetone by centrifugation 15,000×g at 4ºC to remove the acid. Histone pellets were air dried and stored at -80ºC until use. Histone pellets were then resuspended in ultrapure H₂O and quantified using A₂₃₀ measurement, with final concentration being estimated using the formula:

$$histone\,mass\left( {{{mg} \mathord{\left/{\vphantom {{mg} {mL}}} \right.\kern-\nulldelimiterspace} {mL}}} \right) = {{{A_{230}}} \mathord{\left/{\vphantom {{{A_{230}}} {0.42.}}} \right.\kern-\nulldelimiterspace} {0.42.}}$$

(1)

Cell measurements

Cells were grown to 20–90% confluence in 6-well plates. Harvest was accomplished by incubating cells in 0.5 mL of 0.25% trypsin (Gibco #25200056) for 5 min (control lines) or 8 min (SDH-loss lines) with addition of light mechanical dissociation before trypsin quenching by addition of 1 mL media and viewing under a microscope. Average cell diameter was obtained by dividing the known diameter of a saturated field of view by the number of contained cells. Results were validated using a Countess II FL Automated Cell Counter instrument (Invitrogen, Carlsbad, CA). Doubling time for imCCs and iMEFs lines was obtained using an adaptation of a previously described method [32], replacing manual imaging with IncuCyte^® technology (Sartorius, Ann Arbor, MI, USA) according to the manufacturer’s protocol. In short, cells were seeded at a low density and incubated in the IncuCyte instrument for 5 d, supplied with fresh media every 2 d. Proliferation was determined using a label-free count method where phase objects were counted 2 h before the cultures reached confluence (defined as a cell density when the analysis software fails to distinguish individual phase objects). Doubling time (T_d) was calculated based on cell counts n₁ and n₂ at times t₁ and t₂, respectively, according to:

$${T_d} = ({t_2} - {t_1})\left[ {\frac{{\log 2}}{{\log \left( {\frac{{{n_2}}}{{{n_1}}}} \right)}}} \right]$$

(2)

Trypsin sensitivity

Three replicates of five million cells for each of the four lines were seeded in T-75 flasks and allowed to plate overnight prior to trypsinization assay. The following morning, cells were incubated with 5 mL of 0.25% trypsin (Gibco #25200056) for 5 min and then quenched with a 4× volume of growth media. The number of detached cells was determined by hemocytometer based on a sample of the supernatant. For each line, a counting control was used to calculate the final number of cells. This counting control involved trypsinization prolonged for 10 min with light mechanical shaking to calculate the total number of cells.

Cell cycle analysis

Three replicates of two million cells per sample for each of the four cell lines were harvested by aspirating culture medium followed by three washes in PBS. Cells were then fixed and permeabilized in ice cold 70% ethanol for 3 days in -20° C. On day 4, samples underwent 3 additional washes followed by resuspension in 1 mL PBS in flow tubes (1 million cells per mL). One drop Propidium Iodide Ready Flow™ Reagent (Invitrogen #R37169) per million cells was added with incubation in the dark for 45 min before sample analysis on a BD FACSymphony A3 Flow Cytometer at 50,000 cells per sample (Becton, Dickinson and Co, Vernon Hills, IL, USA). For Live cell cycle analysis, 3 replicate T-175 flasks per line were harvested at 80% confluency by 0.25% Trypsin (Gibco #25200056). The cells were then washed and suspended in PBS. One million cells were counted and diluted in a total of 1 mL PBS. Live cell staining was done using Hoechst 33,342 Ready Flow™ Reagent [(Invitrogen #R37165); 2 drops per 1 million cells] and the samples were incubated in 37 C for 60 min prior to cell sorting. Data analysis was performed using the univariate cell cycle model in FlowJo™ v10.10 (FlowJo, LLC, Ashland, OR, USA).

Metabolite quantitation

imCCs and iMEFs were grown to ~ 80–90% confluence (~ 1 million cells 100-cm dish). Cells were washed three times with PBS to remove media. Cells were frozen on dry ice and 1.5 mL chilled methanol (-20 °C) was added to each dish. Cells were then scraped into methanol using a sterile plastic scraper with a rubber head. The resulting slurries were transferred to 2-mL conical tubes and frozen on dry ice. Samples were stored at -80 °C prior to analysis. TCA cycle-related analytes were measured by gas chromatography/mass spectrometry (GC/MS) with technical assistance from the Mayo Clinic Metabolomics Core facility as previously described [34, 35].

Electron microscopy

Sample processing for electron microscopy was performed in the Mayo Clinic Microscopy and Cell Analysis Core. One million control or SDH-loss imCCs or iMEFs (4 samples total) were washed twice in PBS and fixed in 5 mL McDowell Trump’s fixative (Electron Microscopy Sciences # 18030-05) using gentle vortex mixing. Micrographs were acquired using a JEOL 1400 Plus transmission electron microscope (JEOL, Inc., Peabody, MA) at 80 kV equipped with a Gatan Orius camera (Gatan, Inc., Warrendale, PA) [36].

F-actin staining

F-actin staining was performed using ActinGreen AlexaFluor™ 488 phalloidin (Invitrogen # R37110) according to the manufacturer’s protocol. Briefly, 10,000–20,000 cells were seeded on glass-bottom plates and allowed to adhere overnight. Cells were then washed with PBS and fixed in 4% formaldehyde (Sigma Aldrich #1039991000) in PBS solution for 15 min followed by permeabilization in 0.1% Triton-X100 (Sigma Aldrich #X100-100ML) for 10 min. After an additional wash in PBS, cells were incubated with 1 µg/mL DAPI for 5 min followed by addition of 1 drop ActinGreen for 30 min. Cells were then imaged using Olympus BX50 DIC Fluorescence Microscope (Olympus Corporation of the Americas​​, Center Valley, PA, USA) and processed using ZEISS TIVATO 700 (ZEISS White Plains, NY, US).

Chromosome counting by fluorescence in situ hybridization (FISH)

FISH for determination of total mouse chromosome number was performed in the Genomics Culture Laboratory of the Mayo Clinic Division of Laboratory Genetics and Genomics. Counting and analysis was performed with the IKAROS and ISIS digital FISH imaging system (MetaSystems, Altlussheim, Germany).

Cell labeling and confocal imaging

imCCs and iMEFs were plated at a density of ~ 10,000 cells/well into 8-well Ibidi µ-slide plates (Ibidi #80826) and allowed to adhere. Cells were treated for 15 min with growth medium (37˚C) containing 100 nM MitoTracker^® Green FM (Invitrogen #M46750; excitation/emission wavelength 490/516 nm) followed by washing with Hanks’ Balanced Salt Solution [HBSS (Invitrogen #11575032)]. Cells were imaged by confocal microscopy using instrumentation previously described [37,38,39,40,41]. The dynamic range for imaging was set by first scanning a region containing no fluorescence signal and then a second region of interest containing maximum fluorescence. A series of 0.5 μm thick optical slices were acquired for each cell. This Z-stack was acquired and analyzed using NIS-Elements software (Nikon Instruments Inc., Melville, NY).

Estimation of mitochondrial volume density

Three-dimensional reconstructions were obtained and deconvolved using the algorithm available in NIS-Elements analysis software (Modified Richardson Lucy method; Point Scan Confocal modality; Nikon Instruments Inc.) [38, 40,41,42]. Deconvolution was applied to the image signal-to-noise ratio so as to improve contrast and edge detection. Voxel dimensions of each deconvolved optical slice were 0.207 × 0.207 × 0.5 μm. Cell boundaries were then delineated using ImageJ-Fiji 2.16.0 software (https://imagej.nih.gov/ij/). Using the mitochondrial analyzer module of ImageJ software, Z-stacks were then processed for background correction and ridge filter detection [39, 40, 43,44,45]. Mitochondria were identified by thresholding to create a binary image and then skeletonized for morphometric analysis. Using thresholded images, total mitochondrial volume was measured using the mitochondrial analyzer software module, where the number of voxels containing fluorescently labeled mitochondria within a single delineated cell was determined [39, 46]. Mitochondrial volume density was calculated as the ratio of mitochondrial volume within the cell to the total volume of the delineated cell [40,41,42, 47,48,49].

Quantitative histochemical estimation of SDH reaction velocity

Images were acquired using a Nikon Eclipse A1 laser scanning confocal microscope (RRID: SCR_020317) with a 60×/1.4 NA oil-immersion objective at 12-bit resolution into a 1,024 × 1,024-pixel array using the transmitted light channel was used for image acquisition. An interference filter with a peak emission wavelength of 570 nm was placed in the light path to limit the spectral range of the light source to the maximum absorption wavelength of NBT_dfz[36]. The measured gray level (GL) of the microscope was calibrated to known optical density (OD) units using a photographic density stepwedge tablet (0.04–2.20 OD units in increments of 0.15 OD; Stouffer Industries, Mishawaka, IN). The dynamic range of the microscope was adjusted to take advantage of the full range of OD while avoiding saturation of the images at both ends of the OD range [50]. The quantitative histochemical procedure for measuring SDH reaction velocity was then performed. This technique has been previously described in detail for single skeletal muscle fibers [38,39,40,41,42,43,44,45,46,47, 51] and airway smooth muscle cells [50, 52,53,54]. imCCs and iMEFs were plated in 8-well Ibidi µ-slide plates (Ibidi #80826) at a density of ~ 10,000 cells/well and incubated to allow for cell adherence. Cells were exposed to solutions containing 1.5 mM nitro blue tetrazolium (NBT; reaction indicator), 5 mM EDTA, 1 mM mPMS, and 0.75 mM sodium azide in 0.1 M phosphate buffer (pH: 7.4). In a separate series of experiments, the dependence of the SDH reaction on succinate concentration was assessed, and a concentration of 80 mM succinate was found to produce the maximum velocity of the SDH reaction in both imCCs and iMEFs. In the quantitative histochemical procedure, the progressive precipitation of colored NBT_dfz due to the reduction of NBT is used as the reaction indicator. Experiments were performed at room temperature (22 ± 1 °C), with similar parameters maintained across preparations. A series of 0.5 μm thick optical slices was acquired every 15 s across a 10-min period. The Z-stack was acquired and analyzed using NIS-Elements software (version 5.20.02; RRID: SCR_014329; Nikon Instruments Inc., Melville, NY). From the optical slices, the cells were delineated as the region of interest while the nuclei were excluded. NBT_dfz precipitation was measured as the progressive increase in OD as the SDH reaction proceeded. The change in average OD within the selected region of interest was measured every 15 s across a 10-min period. The SDH_max was then determined using the Beer-Lambert equation:

$$SD{H_{max}} = \frac{{d\left[ {NB{T_{dfz}}} \right]}}{{dt}} = \frac{{{\raise0.7ex\hbox{${dOD}$} \!\mathord{\left/{\vphantom {{dOD} {dT}}}\right.\kern-\nulldelimiterspace}\!\lower0.7ex\hbox{${dT}$}}}}{{kl}}$$

(3)

Where OD is measured based on the calibrated gray level of NBT_dfz accumulation, k is the molar extinction coefficient for NBT_dfz (26,478 M^− 1 cm^− 1), and l is the pathlength for light absorbance (0.5 μm optical slice thickness). SDH_max is expressed as mmol fumarate per liter of cell per min.

Multiple cells were visualized within a single microscopic field and the cells whose borders were not overlapping were analyzed. Typically, this selection process resulted in analyzing 3 to 4 cells per field.

Oxygen consumption rate (OCR)

imCCs and iMEFs were plated into Seahorse XFe96 cell culture microplates (Agilent Technologies # 204626-100) at three densities (30,000, 60,000 and 100,000 cells per well) to establish optimal signal-to-noise ratios. Because of the larger size of plated SDH-loss imCCs, an equal number of cells occupy more surface area and grow more slowly at high confluence. Thus, higher cell numbers are impractical. Wells were allowed to recover overnight. OCR was measured by the mitochondrial stress test assay performed using the Seahorse XFe96 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) according to the manufacturer’s instructions and as described previously [55]. Prior to the assay, growth media were replaced with Seahorse assay medium consisting of Seahorse XF DMEM base medium (Agilent #103335-100) supplemented with 17.5 mM glucose (Gibco #A2494001), 1 mM sodium pyruvate (Gibco #11360070), and 2 mM glutamine (Gibco #25030081) at pH 7.4, and incubated for 1 h in a non-CO₂ incubator. Optimal concentration of inhibitors and FCCP used for both cells were found to be: 1 µM oligomycin (ATP uncoupler), 1 µM FCCP (accelerates electron transport chain), and 0.5 µM antimycin A (Complex III inhibitor) with 1 µM rotenone (Complex I inhibitor). OCR measurements were normalized to total adherent cell count, assessed using 1 µg/mL Hoechst 33,342 Solution (Invitrogen #H3570; excitation/emission wavelength: 361/486 nm) and using the Seahorse XF Imaging and Normalization on a Cytation 1 (BioTek, Winooski, VT).

For physiologic O₂ studies, imCCs and iMEFs were adapted for three passages in 5% O₂/5% CO₂ using a Biospherix XVivo X3 incubator (Biospherix, Parish, NY) with media conditioned to 5% O₂/5% CO₂ with a Hypoxycool media conditioner (Baker, Sanford, ME). Following adaptation, cells were plated, and Seahorse assays performed as above with modifications. To enable O₂ control, the Seahorse XFe96 was housed in Hypoxia Glove Box (Coy Laboratory Products, Grass Lake, MI) and stabilized to 5% O₂/0% CO₂ for 24 h prior to the assay and Seahorse XF media was conditioned to 5% O₂/0% CO₂. Cells were not seeded into the final column of the XFe96 cell culture plate allowing these wells to serve as 0% O₂ control wells. Following replacement of growth media with Seahorse XF media (or calibrant in 0% O₂ control wells), cells were incubated in Seahorse XFe96 for 1 h to allow for degassing of CO₂. The Seahorse XFe96 was operated in Hypoxia Mode and injection ports for 0% O₂ control wells were loaded with 1 M sodium sulfite.

Electron transport chain complex-specific Seahorse extracellular flux analysis was performed as previously described [55]. In brief, imCCs and iMEFs were plated into Seahorse XFe96 cell culture microplates (Agilent # 204626-100). Prior to the assay, growth media was replaced with Mitochondrial Assay Solution (MAS, pH 7.4 at 37 C, 220 mM mannitol, 70 mM sucrose, 10 mM KH₂PO₄, 5 mM MgCl₂, 2 mM HEPES, 1 mM EGTA and 0.2% (w/v) fatty-acid free BSA, 1 mM malate, 10 mM pyruvate, 4 mM ADP, 1 nM XF Plasma Membrane Permeabilization (PMP) agent and 4 µM FCCP. Sequential injection of complex specific substrates and inhibitors allow the quantitation of complex activity, baseline (complex I activity), 2 µM rotenone (inhibit Complex I), 10 mM succinate (complex II), and 2 µM Antimycin A (inhibit Complex III downstream of Complex I and II).

RNA sequencing

Cell pellets from imCCs and iMEFs were collected from three replicates at 24, 48 and 72 h after seeding. RNA was extracted using RNeasy kits (Qiagen #74104) and DNA removed by addition of on-column DNase I treatment (Thermo Scientific #EN0521). Further RNA purification and concentration were performed using Concentrator-5 (Zymo #R1013) prior to analysis in the Mayo Clinic Medical Genome Facility where indexed deep sequencing libraries were prepared using the Illumina TruSeq mRNA v2 kit (Illumina #RS-122-2001). Deep sequencing was conducted on a NextSeq P2 flow cell (Illumina, San Diego, CA, USA) for an approximate total of 70 M reads/sample. For analysis, paired-end sequence reads were aligned to the mm9 mouse reference genome using the STAR fast read aligner [56]. Gene level counts were obtained using featureCounts v1.4.6 from the subRead package [57]. Differential expression analysis was performed with DESeq2 [58]. RStudio, Bioconductor 3.17 and R packages: ggplot2; Complex heatmap [59] were used for analysis and plotting.

Statistical analyses

Relevant statistical tests are described in individual figure legends.

Results

Phenotypic changes following SDH loss in imCCs and iMEFs

SDH loss influences cellular biochemistry, metabolism, gene expression and morphology, particularly due to succinate accumulation [60,61,62]. Our goal was to compare and contrast SDH loss effects in two different available cell models: WT vs. SDHB-loss imCCs and heterozygous vs. SDHC-loss iMEFs. SDH-loss lines were confirmed by genotyping (Supplemental Fig. S1A-D). Western blotting for SDHB and SDHC indicated that target gene knockout results in complete loss of the targeted SDH subunit with detectable traces of residual the non-targeted SDH subunit (Supplemental Fig. S2A). This confirms that disrupted SDH complex assembly destabilizes SDHB/C subunits [63]. Similarly, mitochondrial complex western blotting assessment of COX10, NDUFS1 (oxidoreductase core of Complex I) and cytochrome c suggest SDH loss affects other components of the Respiratory Chain beyond SDH inhibition (Supplemental Fig. S2B). Furthermore, RNAseq analysis involving mitochondrial fission and fusion markers showed differential expression in SDH-loss cells compared to their respective controls, most notably in imCCs. Certain genes like Mfn1, Mfn2, and Opa1 show more subtle changes in iMEFs compared to imCCs, suggesting greater effect on fusion genes more in imCCs. By contrast, there appears to be a stronger bias toward mitochondrial fission in SDH-loss iMEFs, as evidenced by robust upregulation of Fis1 and Drp1 (Supplemental Fig. S2C). We confirmed loss of SDH enzyme activity in SDH-loss lines vs. control lines using an imaging-based SDH activity assay (Fig. 1A). It is unknown if residual SDH activity in Sdhc^−/− iMEFs could be attributable to residual SDHB protein (Supplemental Fig. S2A) that might remain associated with membrane unbound SDHA catalytic subunit that, if properly flavinylated, might confer a trace of enzymatic function [64].

Fig. 1

SDH-deficiency induces phenotypic changes in imCCs and iMEFs. (A) Maximum velocity of SDH catalysis (SDH_max) calculated according to Eq. 3 (n = 20 cells per line). (B) Quantitation of cell doubling time based on Eq. 2 for imCCs and iMEFs (n = 6 wells per line). (C) Average cell diameter for control and SDH-loss imCCs and iMEFs. Data represent average calculated diameter for 6 fields per line. (D) Percentage of detached cells following trypsinization for 5 min compared to a counting control treated with excess trypsin and mechanical force for total detachment (100%, indicated by red line). P-values are calculated using student’s t-test and the number of asterisks indicates degree of significance. (E) Examples of Flowjo-generated cell cycle analysis histograms showing percentage of the indicated cells in each of the indicated cell cycle phases (G1, S or G2) based on DNA content (measured by fluorescence intensity) using a Watson analysis model (n = 50,000 cells per sample). (F) average percentage of cells in each of the indicated cell cycle phases (G1, S or G2) for each control or SDH-loss line shown as mean ± SEM (n = 3 samples per line processed in a single run). RMSD values for the depicted histograms are Sdhb^+/+imCCs: 4.94; Sdhb^−/−imCCs: 4.59; Sdhc^+/−iMEFs: 4.31 and Sdhc^−/−iMEFs: 5.77. These values are representative of the replicate samples. Totals do not all reach 100% because of the presence of cells with DNA content not assigned to any of the indicated phases. P-values are calculated using Bonferroni correction for multiple t-tests and the number of asterisks indicates degree of significance

Not surprisingly, SDH loss dramatically slows the proliferation rate of both iMEFs and imCCs, presumably due to reduced ATP production and TCA cycle truncation and derangement as described for other models of SDH-loss [65, 66]. Cell doubling time was increased more (~ 4-fold) for imCCs, compared to SDHC-loss iMEFs (~ 2-fold) (Fig. 1B). As previously noted for SDH-loss imCC lines [29], cell diameter was 35% larger in suspension. Cell enlargement was observed to a lesser extent in Sdhc^−/− iMEFs (Fig. 1C and Supplemental Fig. S3). This phenotype is also confirmed through flow cytometry forward scatter (FSC) and side scatter (SSC) signatures of SDH-loss cells compared to controls (Supplemental Fig. S3A), and is reflected as well in their adherent state (Supplemental Fig. S3B) Interestingly, SDH-loss cells showed increased intercellular and surface adherence as judged by higher resistance to trypsinization (Fig. 1D).

In light of the extended cell replication time, cell cycle analysis was performed on fixed cells using Propidium Iodide (PI). Results are shown in Fig. 1E-F. These data reveal cell-type specific cell cycle effects of SDH loss. Sdhb^−/− imCCs show a prolonged G1 phase and abbreviated S phase, possibly indicating changes in DNA repair pathways [67]. Similar results were seen in Sdhc^−/− iMEFs, in addition to a prolonged G2 phase (Fig. 1E-F), often taken as evidence of DNA damage. However, Western blotting results probing against the canonical γH2AX DNA damage marker did not demonstrate a significant constitutive DNA damage response in SDH-loss iMEFs (Supplemental Fig. S4C) though some previous evidence of such an effect has been reported [68]. Live cell analysis using Hoechst dye was also performed with varying reproducibility compared to fixed cell analysis due to different levels of PBS tolerance impacting WT lines (Supplemental Figs. S4A-B). While neither dye can reliably compare signal intensity across different cell types, data in Fig. 1E and Supplemental Fig. S4A demonstrate differences in total DNA content for SDH-loss lines vs. their respective controls, as validated below.

Consequences of succinate accumulation on TCA metabolites and dioxygenases

A unique signature of SDH-loss PPGL tumors is the accumulation of succinate both inside and outside of the mitochondrial matrix. Succinate accumulation is associated with an aggressive tumor phenotype and poor clinical outcomes [15, 69]. To investigate the effect of SDH deletion on succinate and other metabolites, a mass spectrometry-based quantitative metabolomic analysis was completed (Fig. 2A-C). The results indicate a striking ~ 130-fold increase in cellular succinate levels for both Sdhb^−/− imCCs and Sdhc^−/− iMEFs. Moreover, SDH-loss cells demonstrate a general trend in decreased levels of other TCA-related metabolites, particularly for SDHB-loss imCCs when normalized to cellular protein (Fig. 2B). The exception is that SDH-loss iMEFs preserve approximately normal levels of isocitrate, 2-hydroxyglutarate and glutamate. Importantly, SDH-loss iMEFs accumulate lactate, while SDH-loss imCCs do not. This suggests a greater reliance of SDH-loss iMEFs on glycolysis and lactate dehydrogenase activity for NAD⁺ regeneration. This result points to the surprising conclusion that SDH-loss imCCs maintain a degree of fitness without a Warburg-like shift to glycolysis [70]. It is noteworthy that normalizing TCA metabolites to cell counts suggests that most metabolites (except succinate) remain close to WT levels, implying that the metabolic activity per cell is similar, despite the increase in protein content. Overall, these results suggest that SDH-loss lines have lower metabolic activity per unit of protein per cell.

Fig. 2

Metabolite profiling and effects on DNA methylation. (A) Schematic overview of TCA cycle highlighting the loss of SDH function (red “X”). (B-C) Relative metabolite levels (log scale) in SDH-loss cells compared to their respective controls (n = 3 samples of million cells per line, data shown as mean ± SEM) normalized to protein content (B) or per cell (C). Cell normalization data were accounted for mathematically by calculating the number of cells in each sample using a standard curve for protein content per 1 M cells. Lac: lactate; Asp: aspartate; Mal: malate; Fum: fumarate; Succ: succinate; Glu: glucose; 2-hg, 2-hydroxyglutarate; AKG, 2-oxoglutarate; c-Acon: cis-aconitate; Isoc: isocitrate; Cit: citrate. (D) Western blot of H3K9me3, detected at 17 kDa, where hypermethylation serves as a marker of succinate inhibition of JMJD demethylases, a subfamily of OG-dependent dioxygenases. Anti-histone H3 is used as loading control, also detected at 17 kDa. (E) Western Blot analysis for HIF1/2α levels in various indicated oxygen concentrations. Protein extracts from the indicated cell lines were probed with anti-HIF1α or HIF2α, both detected at 120 kDa (two upper panels). α-tubulin was used as a loading control and detected at 50 kDa (lower panel)

As posited by the succinate accumulation hypothesis [19], succinate accumulation associated with SDH loss has implications for the function of 2-oxoglutarate (2-OG)-dependent dioxygenases where elevated succinate can cause enzyme competitive inhibition [71]. This is especially true for JMJD demethylases, which rely on 2-OG as a co-substrate to catalyze the removal of methyl groups from histones, a process critical for epigenetic regulation. High levels of succinate can disrupt this process, leading to accumulation of methylated histone residues in chromatin. Histone hypermethylation thus serves as an indicator of succinate inhibition. We confirmed this for both control and SDH-loss imCCs and iMEFs by monitoring H3K9me3 using western blotting (Fig. 2D). We also monitored levels of Hypoxia-Inducible Factor 1α and 2α by western blotting (Fig. 2E). These experiments were performed in various oxygen conditions according to reported methods and previously published reports [27, 72] Interestingly, oxygen concentration did not detectable increase HIF1α stabilization in WT or SDH-loss cells. Rather, SDH-loss cells appear to accumulate higher amounts of HIF1α isoforms with higher molecular weight (suggesting post-translational modifications) than the same proteins detected in control cells. Also unexpectedly, reduced oxygen concentration did not increase HIF2α stabilization. Again, SDH-loss cells demonstrate higher amounts of larger isoforms consistent with post-translational modification than HIF2α isolated from control cells, as noted by the the shift to increasing size in the double-band HIF2α pattern for SDH-loss imCC vs. control cells, and the predominance of the upper band of the doublet in SDH-loss iMEFs vs. control cells (Fig. 2E). These results confirm succinate inhibition of histone demethylases in SDH-loss cells, but reveal more complex effects on HIF proteins.

Morphological changes induced by SDH loss

Due to its key role in both the mitochondrial TCA cycle and mitochondrial electron transport chain, we predicted that the loss of SDH function would alter mitochondrial morphology [73, 74]. To test this hypothesis, we utilized electron microscopy for ultrastructural characterization of control and SDH-loss imCCs and iMEFs. Our results confirmed striking SDH-loss pathology in both imCCs and iMEFs, most impressively in the former (Fig. 3A and Supplemental Fig. S5A). SDH-loss imCCs exhibit dramatically enlarged, ‘ghostly’ mitochondria identifiable by faint residual internal cristae, and often characterized by a single electron-dense central condensate (Fig. 3A, bottom right). Some of these features were also observed in SDH-loss iMEFs, though more mitochondria appeared to maintain structural integrity, resulting in more diverse mitochondrial morphology in SDH-loss iMEFs. We observed an unusual pattern of F-actin surrounding enlarged SDH-loss mitochondria in both iMEFs (Fig. 3B and Supplemental Fig. S5B-C) and imCCs (Fig. 3B, especially bottom panel) [75]. We combined this study with independent assessment of three-dimensional mitochondrial volume density using MitoTracker Green staining. This dye binds mitochondria independent of membrane potential (Supplemental Fig. S6A-B). The results are shown in Fig. 3C. These results suggest that SDH-loss cells suffer an overall reduction in cellular mitochondrial volume despite the striking bloating of remaining mitochondria (Fig. 3A).

Fig. 3

Altered mitochondrial and cellular morphology following SDH loss. (A) Representative electron micrographs of mitochondria for the indicated cell lines at the indicated magnifications. Individual examples of mitochondria are indicated by red arrows in the far-right panels. (B) Staining of the same cell lines to show unusual circumferential F-actin staining (red arrows) around enlarged mitochondria in SDH-loss cells. (C) Mitochondrial volume density measurement by 3D microscopy for the four indicated lines (n = 20 cells per line). (D) Average chromosome counts per cell for each control and SDH-loss line (n = 20 cells per line). Red line symbolizes the normal chromosomal count of a diploid mouse cell. Error bars in panels C-D indicate mean ± SD. P-values are calculated using student’s t-test for each experimental and control pair and the number of asterisks indicate degree of significance. (E) Micrographs of single-cell FISH analysis with a mouse pan centromere probe (green) to mark the centromeres on DAPI stained chromosomes

Our three-dimensional microscopic analysis suggested that Sdhb^−/− imCCs display a ~ 3-fold volume increase, consistent with the cell diameter differences reported above (Supplemental Fig. S6C-E). Nuclear volume was also increased (Supplemental Fig. S6D). To investigate a potential link between increased nuclear volume and chromosome count [76] we used FISH to determine the number of chromosomes in the four study cell lines, noting that deviation from normal chromosomal count commonly accompanies spontaneous immortalization [77]. Indeed, our results demonstrate striking and heterogeneous polyploidy in both control lines, further increased upon SDH loss (Fig. 3D-E). Sdhb^−/− imCCs and Sdhc^−/− iMEFs carry ~ 2.5-fold and ~ 1.8-fold more chromosomes than the expected normal diploid mouse complement of 40 chromosomes, respectively. Initial inspection suggests an increase in chromosome fragments carrying centromeres (Fig. 3E). As previously noted by flow cytometry (Fig. 1E), the total mass of nuclear DNA substantially increased in SDH-loss cells.

Differential metabolic fitness of SDH-loss imCCs vs. iMEFs in normoxia and hypoxia

Loss of SDH function is predicted to disrupt the TCA cycle and reduce the flow of electrons through the mitochondrial electron transport chain, decreasing oxidative phosphorylation. As a result, SDH-loss cells are predicted to demonstrate decreased oxygen consumption and increased reliance on glycolysis [78, 79]. We employed sensitive Seahorse extracellular flux analysis to characterize metabolic phenotypes of control and SDH-loss imCCs and iMEFs in room air (20% oxygen) and more physiologically relevant conditions (5% oxygen).

In 20% O₂, basal oxygen consumption rate (OCR) measurements revealed that while both Sdhb^−/− imCCs and Sdhc^−/− iMEFs demonstrated lower resting levels of mitochondrial respiration than their control lines, Sdhb^−/− imCCs consistently showed higher residual oxidative metabolism than Sdhc^−/− iMEFs (Fig. 4A). This suggests a distinctive chromaffin cell-specific adaptation to SDH loss, as previously suggested for these cells [29]. This result is also consistent with the absence of lactate accumulation in Sdhb^−/− imCCs noted above (Fig. 2A). Thus, while both Sdhb^−/− imCCs and Sdhc^−/− iMEFs show deficits in oxidative metabolism, the effect is more severe for SDH-loss iMEFs. Interestingly, residual non-mitochondrial oxygen utilization (signal at ~ 70 min in Fig. 4A, B) appears lower for both SDH-loss cell lines than their control counterparts. A plausible explanation for this interesting observation is that dioxygenase intoxication by accumulated succinate in SDH-loss cells blocks non-mitochondrial oxygen utilization by this large family of enzymes [80].

Fig. 4

imCCs, but not iMEFs maintain substantial metabolic fitness following loss of SDH. (A-B) Seahorse mitochondrial stress test profiles for the indicated cell lines in 20% O₂ (A) or 5% O₂ (B). Rote/AA: rotenone and antimycin A. (C) Complex-specific Seahorse-based assay utilizing Complex I, II and III inhibitors or substrates to assess activities of Complexes I and II in the indicated cell lines. (D-E) Bar graphs representing data for Complex I (D) or Complex II (E) activity from panel C. Seahorse tests were performed in replicates of three (n = 3), error bars represent mean ± SEM. P-values are calculated using student’s t-test and the degree of significance is indicated with number of asterisks

Seahorse studies of oxygen utilization were repeated under physiologically relevant conditions (5% oxygen). Strikingly, basal oxygen utilization by Sdhb^−/− imCCs under hypoxia became essentially indistinguishable from that of WT Sdhb^+/+ imCCs (Fig. 4B). This indicates much higher basal oxidative fitness for SDH-loss imCCs than SDH-loss iMEFs under physiological oxygen concentratitons. This result again points to the possibility that the chromaffin cell context tolerates SDH loss better than the fibroblast context, supporting the observed tissue specificity of SDH-loss PPGL. Both SDH-loss cell lines are deficient with respect to maximum oxidative capacity following uncoupling with FCCP, suggesting a limited ability to respond to high energy demand (Fig. 4B).

As noted, the previous study of these SDH-loss imCCs by Kľučková et al. [29] attributed residual oxidative capacity of SDH-loss imCC to residual function of mitochondrial Complex I in the absence of SDH (Complex II). To directly test this hypothesis in our systems, we employed Seahorse instrumentation with a modified protocol in permeabilized cells enabling the supply of complex-specific substrates and inhibitors, including rotenone as a specific complex I inhibitor, followed by succinate as a complex II substrate, and concluding with Antimycin A as inhibitor of the entire electron transport chain [29, 55, 81]. This experiment (Fig. 4C-E) was performed in 20% O₂. Importantly, our data show that SDH-loss imCCs, but not SDH-loss iMEFs, substantially preserve Complex I activity and corresponding oxygen utilization, even in the complete absence of complex II activity (Fig. 4C-E). This is a remarkable confirmatory observation given the obviously compromised mitochondrial morphology in SDH-loss imCCs (Fig. 3A).

Strikingly different transcriptomic responses of imCCs and iMEFs to SDH loss

We compared RNA expression in our four cell lines of interest. The cellular response to SDH loss is anticipated to involve activation of cell type-specific signaling pathways related to stress and metabolic adaptation [82, 83]. Transcriptomic analyses of the four cell lines of interest are shown in Fig. 5 and Supplemental Fig. S8. The volcano plots in Fig. 5AB confirm dramatically altered transcriptomes upon SDH loss, as expected. Interestingly, with the exception of the targeted Sdhx genes themselves, SDH knockout did not alter transcription of the other Sdhx subunit genes (Fig. 5AB), suggesting that there is not transcriptional feedback regulation from Complex II status to Sdhx gene expression. We sought to compare the transcriptional responses of imCCs and iMEFs to SDH loss. These responses were strikingly uncorrelated (Fig. 5C). This result contrasts with the correlation between transcript changes upon SDH loss within a single cell type (for example, imCCs, Fig. 5D). This is an important result, underlining the cell type-specific transcriptional response to SDH loss. The result is further emphasized when a pathway analysis was undertaken to understand altered gene expression in the two cell lines (Supplemental Fig. S8). After accounting for cell type differences in the gene expression, differential gene expression analyses in imCCs and iMEFs detected 679 genes that were overexpressed and showed the same directionality across Sdhb^−/− and Sdhc^−/− cells and 249 genes that were underexpressed (Supplemental Fig. S9A-D). Quite remarkably, very few upregulated or downregulated pathways are shared in the responses of the two cell types to SDH loss. This suggests a surprising lack of commonality in the response of adrenal-derived cells vs. fibroblasts to SDH loss. This result raises the possibility that adrenal-derived cells are unique in their resilience to SDH loss.

Fig. 5

Distinct transcriptomic changes upon SDH loss in imCCs (A) or iMEFs (B). Data are displayed as volcano plots where genes shown in back are differentially expressed [p-value < 0.05, Log₂(FC) > 1.5]. (C) Scatter plot showing relatively poor correlation between Log₂(FC) values for RNA transcripts changed upon SDH loss in imCCs (x-axis) and iMEFs (y-axis). (D) For comparison, relatively strong correlation is observed for expressed transcripts between WT imCCs (x-axis) and SDH-loss imCCs (y-axis). E and F. Volcano plots show differential expression of RNAs encoding components of Complex I upon SDH loss in imCCs (E) or in iMEFs (F). Labelled genes are differentially expressed [p-value < 0.05, Log₂(FC) > 1.5]

Of particular interest in this regard is the apparent preserved level of oxidative phosphorylation in SDH-loss imCCs. Based on the results presented above, we attribute this at least in part to preserved Complex I activity. We examined the expression of Complex I subunits in response to SDH loss in the two cell types. The results are shown in Fig. 5E, F. It is notable that ~ 10 Complex I subunits are overexpressed by 2-fold or more upon SDH loss in imCCs, while that number is only ~ 3 for SDH-loss imCCs. It is likely that one or more of these differentially expressed Complex I subunits may account for sustained Complex I activity in the absence of Complex II in imCCs.

Discussion

Individuals heterozygous for defective variants affecting SDH subunits are at risk for stochastic loss of heterozygosity in susceptible neuroendocrine cells. This triggers development of PPGLs, neuroendocrine tumors that form specifically in the paraganglia, including adrenal chromaffin cells, and extra-adrenal paraganglia found in the carotid bodies and in scattered clusters of the head and neck, abdomen, pelvis, and thorax [63, 84, 85]. The present work continues to examine why SDH loss is tumorigenic, and why tumorigenesis is limited to very specific cell types.

Our study compares characteristics of control and SDH-loss cells derived from mouse adrenal medulla or mouse embryonic fibroblasts. Our comparison is based on cell lines available to the field, and we note that the comparison is imperfect because imCCs suffer SDHB loss while iMEFs suffer SDHC loss. Thus, not only is the cell type different, but the particular SDH subunit insult is different. SDHA carries the active site for succinate oxidation to fumarate, generating FADH₂ within the TCA cycle. SDHB contains three iron-sulfur clusters responsible for passing electrons from FADH₂ and routing them to membrane-soluble ubiquinone forming ubiquinol among the membrane anchoring SDHC and SDHD subunits [85]. Because all SDH subunits are required for complex stability and for proper function within the TCA cycle and as complex II of mitochondrial electron transport chain [63, 82, 83], we consider that SDHB loss and SDHC loss will have equivalent enzymatic effects. However, this assumption can be challenged. Stochastic somatic LOH results in loss of SDH tumor suppressor function in vulnerable cells of heterozygous individuals [66, 85,86,87]. While the enzymatic results of SDHB and SDHC loss should theoretically be identical (total loss of SDH function), subtle differences in residual SDH subunit inventory and activity could contribute to other differences between SDHB-loss imCCs and SDHC-loss iMEFs. For example, it remains a mysterious feature of SDH-deficient PPGL that SDHB loss appears more clinically penetrant than SDHC loss [65, 88]. Possible mechanisms continue to be pursued [88,89,90]. Our goal here has been to explore mechanisms contributing to cell type-specific vulnerability to tumorigenesis after SDH loss.

Pathology in cellular energy machinery is reasonably expected to affect vital processes such as cell division, mobility, adhesion, cytoskeletal rearrangement, post-translational protein modifications, epigenetic control and metabolite composition lose homeostasis [91,92,93]. Our present results suggest that common responses to SDH loss shared between imCCs and iMEFs include increased cell size and slowed proliferation with cell cycle alterations (Fig. 1A-E). Both cell types also accumulate succinate to a striking degree (Fig. 2B-C), confirming that (i) SDH function is compromised and (ii) that SDHB and SDHC are equally important for SDH enzyme activity. Intriguingly, SDH-loss iMEFs showed relatively normal levels of TCA metabolites isocitrate, 2-hydroxyglutarate and glutamate, but uniquely accumulated lactate. These results suggest a shift towards glycolysis in Sdhc^−/− iMEFs. We note that our ECAR measurements were not sufficiently sensitive to detect this effect (Supplemental Fig. S7). Additionally, we show that a hallmark of SDHx-loss cells, namely inhibition of dioxygenases such as JMJD demethylases is detectable in SDH-loss cells (Fig. 2D), confirming a canonical response in both cell types. We find that HIF transcription factor subunit responses to SDH loss in these cells are more complex (Fig. 2E).

In agreement with previously observed human tumor pathology, ultrastructure analysis (Fig. 3A) suggests shared aspects of mitochondrial pathology in SDH-loss imCCs and iMEFs, with the former showing more profound mitochondrial defects [94,95,96]. These results were coupled with abnormal reorganization of actin filaments with actin-enclosed mitochondria (Fig. 3B) [75]. Electron microscopy demonstrates severe alteration in mitochondrial morphology particularly in SDH-loss imCC. The swollen residual mitochondria demlnstrate only traces of cristae [97] (Fig. 3A). Defective mitochondria in SDH-loss imCC often include single electron-dense condensed bodies of unknown origin (Fig. 3A, bottom right). It is tempting to speculate that these deposits may correspond to nucleoid material. Our additional three-dimensional microscopy-based measurements suggest an overall decrease in mitochondrial volume density for SDH-loss cells. Previously published volume measurements with this imCC model reported increased volume per mitochondrion [29]. We suggest, based on their results as well as our own (Fig. 3A, D), that both SDH-loss models suffer an overall loss of mitochondrial count, with remaining mitochondria appearing pathologically swollen (Fig. 3A).

It remains unknown how our observations of profound succinate accumulation relate to previous studies of inflammatory responses [98]. Pro-inflammatory pathways may become upregulated at a critical commitment point involving mitochondria prior to apoptosis. As such, SDH loss presumably inhibits various normal roles of mitochondria [99]. While pathological mitochondrial morphology is observed in both SDH-loss imCCs and SDH-loss iMEFs, SDH-loss imCC appear more strongly affected (Fig. 3A).

Importantly, our findings confirm that Sdhb^−/− imCCs maintain a higher basal OCR than Sdhc^−/− iMEFs, in room air. This is consistent with the previous hypothesis [29] that chromaffin cells maintain residual mitochondrial Complex I in the absence of SDH (Complex II), whereas fibroblasts such as SDH-loss iMEFs do not [29]. Using a complex-specific Seahorse analysis, we directly confirm this residual Complex I activity in SDHB-loss imCCs (Fig. 4C-D). This is paradoxical because it suggests greater residual Complex I activity in the context of mitochondria that have more demonstrably severe pathology (SDH-loss imCCs) than in cells with less profoundly deranged mitochondria (SDH-loss iMEFs) (Fig. 3A).

The observation that Sdhb^−/− imCCs appear more metabolically fit than Sdhc^−/− iMEFs is consistent with the hypothesis that it is this residual metabolic fitness that makes chromaffin cells uniquely susceptible to PPGL tumorigenesis because of their ability to better proliferate after SDH loss. This possibility is particularly striking in light of our findings under physiologically-relevant 5% O₂ conditions where Sdhb^−/− imCCs appeared as metabolically competent as WT imCCs prior to the addition of FCCP uncoupling agent (Fig. 4B). These findings suggest a higher and physiologically-relevant adaptability of imCCs to SDH loss relative to Sdhc^−/−iMEFs.

Finally, the results of our RNA sequencing studies clearly demonstrate that transcriptional responses to SDH loss are cell type-specific, revealing strikingly different patterns between the two tested cases (Fig. 5). Though some common themes can be detected, the overall picture is one of idiosyncratic responses to SDH loss in each cell type. This result suggests the crucial importance of cell context in determining whether loss of the TCA cycle is tumorigenic.