Abstract

Objectives

   A high proportion of women with advanced epithelial ovarian cancer
   (EOC) experience weakness and cachexia. This relationship is associated
   with increased morbidity and mortality. EOC is the most lethal
   gynecological cancer, yet no preclinical cachexia model has
   demonstrated the combined hallmark features of metastasis, ascites
   development, muscle loss and weakness in adult immunocompetent mice.

Methods

   Here, we evaluated a new model of ovarian cancer-induced cachexia with
   the advantages of inducing cancer in adult immunocompetent C57BL/6J
   mice through orthotopic injections of EOC cells in the ovarian bursa.
   We characterized the development of metastasis, ascites, muscle
   atrophy, muscle weakness, markers of inflammation, and mitochondrial
   stress in the tibialis anterior (TA) and diaphragm ∼45, ∼75 and ∼90
   days after EOC injection.

Results

   Primary ovarian tumour sizes were progressively larger at each time
   point while severe metastasis, ascites development, and reductions in
   body, fat and muscle weights occurred by 90 Days. There were no changes
   in certain inflammatory (TNFα), atrogene (MURF1 and Atrogin) or GDF15
   markers within both muscles whereas IL-6 was increased at 45 and 90 Day
   groups in the diaphragm. TA weakness in 45 Day preceded atrophy and
   metastasis that were observed later (75 and 90 Day, respectively). The
   diaphragm demonstrated both weakness and atrophy in 45 Day. In both
   muscles, this pre-severe-metastatic muscle weakness corresponded with
   considerable reprogramming of gene pathways related to mitochondrial
   bioenergetics as well as reduced functional measures of mitochondrial
   pyruvate oxidation and creatine-dependent ADP/ATP cycling as well as
   increased reactive oxygen species emission (hydrogen peroxide).
   Remarkably, muscle force per unit mass at 90 days was partially
   restored in the TA despite the presence of atrophy and severe
   metastasis. In contrast, the diaphragm demonstrated progressive
   weakness. At this advanced stage, mitochondrial pyruvate oxidation in
   both muscles exceeded control mice suggesting an apparent metabolic
   super-compensation corresponding with restored indices of
   creatine-dependent adenylate cycling.

Conclusions

   This mouse model demonstrates the concurrent development of cachexia
   and metastasis that occurs in women with EOC. The model provides
   physiologically relevant advantages of inducing tumour development
   within the ovarian bursa in immunocompetent adult mice. Moreover, the
   model reveals that muscle weakness in both TA and diaphragm precedes
   severe metastasis while weakness also precedes atrophy in the TA. An
   underlying mitochondrial bioenergetic stress corresponded with this
   early weakness. Collectively, these discoveries can direct new research
   towards the development of therapies that target pre-atrophy and
   pre-severe-metastatic weakness during EOC in addition to therapies
   targeting cachexia.

   Keywords: Ovarian cancer cachexia, Mitochondria, Skeletal muscle,
   Metastasis

Highlights

     * •
       This is the first inducible orthotopic model of metastatic ovarian
       cancer cachexia in adult immunocompetent mice.
     * •
       Diaphragm and limb muscle weakness precedes severe metastasis and
       atrophy during ovarian cancer.
     * •
       Skeletal muscle mitochondrial oxidative and redox stress occur
       during pre-severe-metastatic stages of ovarian cancer.
     * •
       Specific muscle force, mitochondrial pyruvate oxidation and
       creatine metabolism demonstrate compensation in later stages.
     * •
       Ovarian cancer has heterogeneous effects on distinct muscle types
       across time.

1. Introduction

   Cancer-induced cachexia is a multifactorial syndrome characterized by
   muscle loss and weakness [[54]1]. Severe cachexia is linked to
   reductions in quality of life, tolerance to anticancer therapies and
   overall survivability [[55][2], [56][3], [57][4]]. The prevalence of
   cachexia varies widely (20–80%) across different types and severity of
   cancer [[58]5,[59]6]. Raising the possibility that cachexia may have
   both ubiquitous and distinct mechanisms related to the host organ. With
   growing awareness that cancer itself can induce cachexia even in the
   absence of cytotoxic cancer therapies, it is imperative to develop
   pre-clinical models for each type of cancer cachexia that captures
   critical phenotypic hallmarks of this disease in humans.

   Dozens of clinical investigations examining prospective therapeutics to
   treat cancer cachexia have been conducted to date, but none have
   resulted in an approved treatment [[60]7]. It has been suggested that
   such pitfalls could be mitigated by developing new pre-clinical models
   of cancer cachexia. Such models would consider the complex and
   multifactorial aspects of cancer cachexia including the interactions
   between cancer and the host organ, the immune system, metastasis, rate
   of cancer development, and the influence of age. Some pre-clinical
   models develop tumours spontaneously in adulthood but can require large
   animal colonies and expenses. Other models involve either a genetic
   mutation, whereby mice are born with cancer, or injections of cancer
   cells under the skin. While these approaches are generally used to
   research many types of cancer [[61][8], [62][9]], it is unclear if the
   mechanisms underlying their cachectic phenotype are limited in their
   translational potential. Thus, researchers have questioned the utility
   of existing models in regard to pre-clinical therapy development and
   elucidating underlying mechanisms of cachexia [[63][10], [64][11],
   [65][12]]. Indeed, the lack of appropriate preclinical models has been
   a strong criticism for all forms of cancer cachexia – not just ovarian
   cancer - over the last number of years [[66]11,[67]13,[68]14]. Models
   with cancer in the host organ that can be induced at any age in
   immunocompetent mice, that also demonstrate metastasis, has been
   suggested to be important for improving the predictive power of in vivo
   models of cancer cachexia [[69][10], [70][11], [71][12]].

   With regards to epithelial ovarian cancer (EOC) cachexia – the most
   lethal gynecological cancer in women [[72]15] - several models to date
   have shown robust primary outcomes of muscle atrophy, muscle weakness,
   and loss of adipose tissue. Some of the existing models used for
   ovarian cancer cachexia use injection techniques that are ectopic to
   the ovaries, in immunodeficient mice, and can develop cachexia at rapid
   rates (as little as 8 days in some cases) [[73][16], [74][17],
   [75][18]]. Furthermore, none of these preclinical models have
   demonstrated the presence of metastasis [[76][16], [77][17], [78][18]].
   The development of a new preclinical model would ideally capture
   metastasis given this defining event is associated with severe cachexia
   and reduced survival rates during advanced stages of ovarian cancer
   [[79]19]. Indeed, when ovarian cancer is detected at early stages and
   before metastasis, the cure rate is estimated to be as high as 90%
   [[80]20] in contrast to much lower survival rates once metastasis has
   occurred. As more than 70% of ovarian cancer cases are diagnosed at
   late stages, improving our understanding of how cachexia develops could
   lead to new insight into improved patient management and perhaps early
   detection of ovarian cancer itself [[81]19].

   Characterizing cachexia warrants careful consideration of how changes
   in muscle mass and force production occur over time as the tumour
   develops, and in relation to underlying mechanisms regulating both
   aspects of muscle quality. Recently, we reported that muscle weakness
   precedes atrophy in the C26 (Colon-26) colorectal mouse [[82]21]. This
   pre-atrophy weakness also occurred without any changes in expression of
   classic atrophy-related gene programs suggesting muscle weakness during
   cancer could also be caused by unknown atrophy-independent mechanisms.
   In this study, a strong relationship was found between pre-atrophy
   weakness and mitochondrial pathway-specific reprogramming as an
   apparent early metabolic stress response to the initial appearance of
   tumours. Remarkably, once locomotor muscle atrophy occurred,
   mitochondria appeared to adapt by increasing pyruvate oxidation, which
   was related to an unexpected restoration of mass-specific force
   production [[83]21]. Of interest, this relationship was heterogeneous
   across different types of muscles suggesting the effects of cancer on
   one muscle type do not necessarily predict the response in another
   muscle. This phenomenon demonstrates the value of comparing muscle
   force to muscle mass across time and between muscle types in relation
   to tumour size. In this regard, muscle weakness during the pre-atrophy
   and atrophy (cachexia) phases of ovarian cancer have not been
   investigated, nor in relation to metastasis or metabolic dysfunction.
   Understanding this could lead to more precise understanding of ovarian
   cancer cachexia pathology to aid better mechanism elucidation and
   therapy development.

   The purpose of this study was to identify the time-dependent and
   muscle-specific development of weakness and atrophy in a novel model of
   ovarian cancer cachexia in relation to metabolic reprogramming. In this
   model, spontaneously transformed ovarian epithelial cells from the same
   mouse strain (syngeneic) were injected into the ovarian bursa
   (orthotopic) in immunocompetent mice with the intention of retaining
   the normal immune response to this type of cancer. These results
   demonstrate a new in vivo model of ovarian cancer cachexia that
   captures metastasis characteristic of advanced stages of EOC seen in
   women [[84]22,[85]23] that also has considerable utility for
   identifying new relationships between the development of muscle
   weakness, atrophy, and metabolic reprogramming across time during
   ovarian cancer.

2. Materials and methods

2.1. Animal care and ID8 inoculation

   Two cohorts of 48 (n = 12 per group; [86]SFigure 1A and B) female
   C57BL/6 mice were ordered at 7–9 weeks of age from Charles River
   Laboratories. These mice were housed at the University of Guelph in
   accordance with the Canadian Council on Animal Care. Tumours were
   induced as described previously at the University of Guelph [[87][24],
   [88][25], [89][26], [90][27]]. Briefly, ID8 cells (epithelial ovarian
   cancer cells; 1.0 × 10^6 in 5 μL) were injected directly under the left
   ovarian bursa of C57BL/6J mice generating an orthotopic, syngeneic,
   immunocompetent cancer mouse model. ID8 cells were originally gifted by
   Drs. Paul Terranova and Kathy Roby from Kansas State University to Dr.
   Jim Petrik at the University of Guelph. ID8 cells are transformed
   murine ovarian surface epithelial cells that do not contain mutations,
   however, after orthotopic implantation and interaction with the ovarian
   microenvironment, ID8 cells that metastasize spontaneously develop a
   gain-of-function p53 mutation that is identical to the mutation seen in
   95% of women with high-grade serous carcinoma [[91]28]. Control mice
   were sham injected with equivalent volumes of sterile phosphate
   buffered saline (PBS). Two weeks after ID8 inoculation, mice were
   transported from University of Guelph to York University where they
   were housed for the remainder of the study in accordance with the
   Canadian Council on Animal Care. All mice were provided access to
   standard chow and water ad libitum.

   Control (CTRL) and 75 Day mice were injected at 9–11weeks old with PBS
   and ID8 cells respectively and aged for 72–78 days post injection. 45
   Day mice were injected at 16–17 weeks old and aged for 42–48 days post
   injection. 90 Day mice were injected at 9–10 weeks old and aged for
   83–107 days post injection ([92]SFigure 1A and 1B). These ranges were
   chosen given force and mitochondrial assessments limit daily
   experimental throughput, and health metrics used to determine the date
   of euthanasia were variable in the more advanced stages of cancer.
   Specifically, at the 90 Day time point, mice were euthanized upon
   presentation of some of the following endpoint criteria: >10% body
   weight loss, >20 mL of ascitic volume collected during paracentesis, >5
   ascites paracentesis taps completed, and/or subjective changes in
   behavioural patterns consistent with removal criteria as per animal
   care guidelines (self-isolation, ruffled fur/poor self-grooming and
   irregular gait). Ascitic taps were necessary to pro-long the survival
   of mice as it would occur in human ovarian cancer and to reach >10%
   body weight loss which is highly suggestive of a cancer cachexia
   phenotype. Staggering the age at which mice received cancer cells
   permitted a consistent age at euthanasia for all mice (20–24 weeks old)
   to reduce aging effects on all measures.

2.2. Volitional wheel running & forelimb grip strength

   72 h before euthanasia, a subset of mice were placed in individual
   cages with a 14 cm diameter running wheel and rotation counter (VDO m3
   bike computer, Mountain Equipment Co-Op, Vancouver, Canada) as done
   previously [[93]29]. Any mice that had ascites within the 90 Day group
   were tapped immediately prior to introduction in the cage with a
   running wheel to reduce the potential interference of ascites volume on
   running performance. 24 h later, distance and time ran were recorded
   and mice were placed in separate caging with no running wheel. Muscle
   measurements were made 48 h thereafter. On the day of euthanasia, mice
   were removed from cages and brought towards a metal grid until such
   time the mice grasped the grid with the forepaws. Upon grasping,
   animals were pulled away from the grid until the grasp was released.
   Peak tension was recorded, and this was repeated twice more with the
   maximum peak tension of 3 trials was used for analyses as done
   previously [[94]29].

2.3. Tissue collection procedure

   Mice were anesthetized with isoflurane and hearts were removed for
   euthanasia. All hindlimb muscles, inguinal subcutaneous fat and spleens
   were weighed and snap-frozen in liquid nitrogen and stored at −80^οC.
   Primary ovarian tumours were also collected by removing the ovary and
   tumour at the site of injection and carefully separating the tumour
   mass from the ovary mass and stored in liquid nitrogen. Hindlimb
   muscles were also embedded in optimal cutting temperature (OCT) medium
   and frozen (see section below). Tibialis anterior (TA) and diaphragm
   muscle were placed in BIOPS containing (in mM) 50 MES Hydrate,
   7.23 K[2]EGTA, 2.77 CaK[2]EGTA, 20 imidazole, 0.5 dithiothreitol, 20
   taurine, 5.77 ATP, 15 PCr, and 6.56 MgCl[2]·6H[2]O (pH 7.1) to be
   prepared for mitochondrial bioenergetic assays.

2.4. Sectioning, histochemical & immunofluorescent staining

   Tibialis anterior and diaphragm muscle samples were embedded in OCT
   medium (Thermo Fisher Scientific) and frozen in 2-methylbutane. These
   muscles were then sectioned into 10 μm sections with a cryostat (HM525
   NX, Thermo Fisher Scientific) maintained at −20^οC on Fisherbrand
   Superfrost Plus slides (Thermo Fisher Scientific). Hematoxylin and
   eosin (H&E) staining was used to assess mononuclear cell infiltration.
   Images were taken using EVOS M7000 imager (Thermo Fisher Scientific)
   using 20× magnification and analyzed on ImageJ. Immunofluorescent
   analysis of myosin heavy chain (MHC) expression was completed as
   previously described [[95]30]. Images were taken with EVOS M7000
   equipped with standard red, green and blue filter cubes. Fibers that
   did not fluoresce were considered IIx fibers. A total of 25–50 fibers
   were then randomly selected and measured. Type IIb fibers in the
   diaphragm strips were in low abundance, therefore, 5–27 fibers were
   measured and used for analysis. Type I fibers were extremely low in
   abundance in the TA and thus were not analyzed [[96]30]. These images
   were also analyzed for cross sectional area (CSA) on ImageJ in a
   blinded fashion. Immunofluorescent analysis of embryonic myosin heavy
   chain (eMHC) were adapted from previous literature [[97]31]. Briefly,
   sections were fixed with 10% formalin, blocked with 10% goat serum,
   followed by mouse IgG block (BML 2202; Vector Laboratories Inc.,
   Burlingame, CA), and incubated with anti-eMHC (15 μg/mL; DHSB F1.652)
   overnight. Secondary Alexa Fluor 647 IgG (1:1000; Abcam, ab150107) was
   then used to fluoresce eMHC primary antibody. Sections were then
   re-blocked once again and incubated with wheat-germ agglutinin (WGA;
   1:1000; Invitrogen [98]W11261) pre-conjugated to Alexa Fluor 488. Last,
   samples were mounted with DAPI mounting medium (Abcam, ab104139).
   D2.mdx muscle tissue saved from previous studies in our lab was used to
   evaluate the efficacy of the antibodies ([99]SFigure 2).

2.5. In situ tibialis anterior force and in vitro diaphragm force

   In situ TA force production was partially adapted from previous
   literature [[100]32,[101]33]. Mice were anesthetized with isoflurane
   and the distal tendon of the TA was exposed by incision at the ankle.
   The distal tendon was tied with suture thread as close to the muscle
   attachment as possible. Once the knot was secured the distal tendon was
   severed. Small cuts were made up the lateral side of the TA to expose
   the muscle for needle electrode placement. The knot was tied to an
   Aurora Scientific 305C (Aurora Scientific Inc., Aurora, ON, Canada)
   muscle lever arm with a hook. The foot of the mouse was secured with
   tape and the knee was immobilized with a needle and set screw with the
   length of the limb parallel to the direction of force. The two needle
   electrodes were placed in the gap of fascia between the TA and tibia to
   stimulate the common peroneal nerve (10–50 mA). The mouse was heated
   with a heating pad or heat lamp throughout force collection. Optimal
   resting length (L[o]) was determined using single twitches (pulse
   width = 0.2 ms) at 1 Hz stimulation frequency with 1 min rest in
   between contractions to avoid fatigue. Once L[o] was established, a
   ruler was used to determine length before the start of force-frequency
   collection (1, 10, 20, 30, 40, 50, 60, 80, 100, 120 and 200Hz with
   1 min rest in between contractions). Force production was normalized to
   the calculated CSA of the muscle strip (m/l∗d) where m is the muscle
   mass, l is the length, and d is mammalian skeletal muscle density
   (1.06  mg mm^3).

   In vitro force production for diaphragm muscle was done as completed
   previously [[102]21,[103]34,[104]35]. Briefly, the diaphragm strip was
   carefully sutured in Ringer's solution (containing in mM: 121 NaCl, 5
   KCl, 1.8 CaCl[2], 0.5 MgCl[2] 0.4 NaHPO[4], 24 NaHCO[3], 5.5 glucose
   and 0.1 EDTA; pH 7.3 oxygenated with 95% O[2] and 5% CO[2]) such that
   the thread secured to the central tendon and ribs. The strip was then
   placed in an oxygenated bath filled with Ringer's and maintained at
   25^οC. The suture secured to the central tendon was then attached to
   the lever arm and the loop secured to the ribs was attached to the
   force transducer. The strip was situated between flanking platinum
   electrodes driven by biphasic stimulator (model 305C; Aurora Scientific
   Inc.). L[o] determined using single twitches (pulse width = 0.2 ms) at
   1 Hz stimulation frequency with 1 min rest in between contractions to
   avoid fatigue. Once L[o] was determined, the strip acclimatized for
   30 min in the oxygenated bath. L[o] was re-assessed and measured with a
   ruler and the start of the force-frequency protocol was initiated (1,
   10, 20, 40, 60, 80, 100, 120, 140 and 200Hz with 1 min rest in between
   contractions). Force production was normalized to CSA of the muscle
   strip (m/l∗d) where m is the muscle mass, l is the length, and d is
   mammalian skeletal muscle density (1.06  mg mm^3).

2.6. Western blotting

   A frozen piece of TA and diaphragm from each animal was homogenized in
   a plastic microcentrifuge tube with a tapered Teflon pestle in ice-cold
   buffer containing (in mM) 20 Tris/HCl, 150 NaCl, 1 EDTA, 1 EGTA, 2.5
   Na[4]O[7]P[2], and 1 Na[3]VO[4] and 1% Triton X-100 with PhosSTOP
   inhibitor tablet (Roche; 4906845001) and protease inhibitor cocktail
   (Sigma Aldrich; P8340) (pH7.0) as published previously
   [[105]21,[106]36]. Protein concentrations were determined using a
   bicinchoninic acid assay (life Technologies, Thermo Fisher Scientific).
   15–20 μg of denatured and reduced protein was subjected to 10%–12%
   gradient SDS-PAGE followed by transfer to low-fluorescence
   polyvinylidene difluoride membrane. Membranes were blocked with Odyssey
   Blocking Buffer (Li-COR) and immunoblotted overnight (4^οC) with
   antibodies specific to each protein. A commercially available
   monoclonal antibody was used to detect electron transport chain
   proteins (rodent OXPHOS Cocktail, ab110413; Abcam, Cambridge, UK, 1:250
   dilution), including V-ATP5A (55 kDa), III-UQCRC2 (48 kDa), IV-MTCO1
   (40 kDa), II-SDHB (30 kDa), and I-NDUFB8 (20 kDa). A commercially
   available monoclonal antibody was used to detect mitochondrial creatine
   kinase (mtCK C-1 43 kDa; Santa Cruz 376320, 1:500 dilution).

   After overnight incubation in primary antibodies, membranes were washed
   3 times for 5 min in TBS-Tween and incubated for 1 h at room
   temperature with the corresponding infrared fluorescent secondary
   antibody (LI-COR IRDye 680RD 925–68020, 1:20 000).

2.7. Preparation of permeabilized muscle fibers

   The assessment of mitochondrial bioenergetics was performed as
   described previously in our publications [[107]21,[108]29,[109][37],
   [110][38], [111][39]]. Briefly, the TA and diaphragm from the mouse was
   removed and placed in ice cold BIOPS. Muscle was separated gently along
   the longitudinal axis to from bundles that were treated with 40 μg/mL
   saponin in BIOPS on a rotor for 30 min at 4^οC. Following
   permeabilization the permeabilized muscle fiber bundles (PmFBs) for
   respiration were blotted and weighed in 1.5 mL of tared prechilled
   BIOPS for normalization of respiratory assessments. The remaining PmFBs
   for mitochondrial H[2]O[2] (mH[2]O[2]) were not weighed at this step as
   these data were normalized to fully recovered dry weights taken after
   the experiments. All PmFBs were then washed in Buffer Z on a rotator
   for 15 min at 4^οC to remove the cytoplasm. Buffer Z contained (in mM)
   105 K-MES, 30 KCl, 10 KH[2]PO[4], 5 MgCl[2]·6H[2]O, 1 EGTA and 5 mg/mL
   BSA (pH 7.1).

2.8. Mitochondrial respiration

   High-resolution respirometry (O[2] consumption) were conducted in 2 mL
   of respiration medium (Buffer Z) using the Oroboros Oxygraph-2k
   (Oroboros Instruments, Corp., Innsbruck, Austria) with stirring at
   750 rpm at 37 °C. Buffer Z contained 20 mM Cr to saturate mtCK and
   promote phosphate shuttling through mtCK or was kept void of Cr to
   prevent the activation of mtCK [[112]40]. All experiments were
   conducted in the presence of 5 μM blebbistatin (BLEB) in the assay
   media to prevent spontaneous contraction of PmFB, which has been shown
   to occur in response to ADP at 37 °C that alters respiration rates
   [[113]40]. Complex I-supported respiration was stimulated using 5 mM
   pyruvate and 2 mM malate (NADH) followed by a titration of ADP
   concentrations from physiological ranges (25 μM, 100 μM; [[114]41]) to
   high submaximal (500 μM) and saturating to stimulate maximal coupled
   respiration (5000 μM in the presence of creatine and 7000 μM in the
   absence of creatine). 10 mM glutamate (further NADH generation) was
   added at the end of the ADP titration. Cytochrome c was then added to
   test mitochondrial outer membrane integrity. Experiments with low
   ADP-stimulated respiration (bundles that did not respond to ADP) with
   high cytochrome c responses (>15% increase in respiration) were removed
   from analysis (13 of 370 bundles). Last, 20 mM Succinate (FADH[2]) was
   added to stimulate complex-II supported respiration. These protocols
   were designed to understand the regulation of respiration coupled to
   oxidative phosphorylation of ADP to ATP (Adenosine triphosphate).

2.9. mH[2]O[2]

   mH[2]O[2] was determined spectrofluorometrically (QuantaMaster 40,
   HORIBA Scientific) in PmfB placed in a quartz cuvette with continuous
   stirring at 37 °C in 1 mL of Buffer Z supplemented with 10 μM Amplex
   Ultra Red, 0.5 U/ml horseradish peroxidase, 1 mM EGTA, 40 U/ml
   Cu/Zn-SOD1, 5 μM BLEB and 20 mM Cr to saturate mtCK. No comparisons
   were made to PmFB in the absence of creatine due to tissue limitations.
   State II mH[2]O[2] (maximal emission in the absence of ADP) was induced
   using the Complex I-supporting substrates (NADH) pyruvate (10 mM) and
   malate (2 mM) mH[2]O[2] to generate forward electron transfer
   (FET)-supported electron slip at Complex I [[115]42] as described
   previously [[116]29]. These PmFBs were incubated with 35 μM CDNB during
   the 30-minute permeabilization to deplete glutathione and allow for
   detectable rates of mH[2]O[2]. Following the induction of State II
   mH[2]O[2], a titration of ADP was employed to progressively attenuate
   mH[2]O[2] as it occurs when membrane potential declines during
   oxidative phosphorylation [[117]43]. A separate PmFB was used to
   stimulate electron slip at Complex I through reverse electron transfer
   (RET) from complex II using succinate (FADH[2]) [[118]42]. followed by
   ADP titrations as used in the previous protocol. After the experiments,
   the fibres were lyophilized in a freeze-dryer (Labconco, Kansas City,
   MO, USA) for > 4h and weighed on a microbalance (Sartorius Cubis
   Microbalance, Gottingen Germany). The rate of mH[2]O[2] emission was
   calculated from the slope (F/min) using a standard curve established
   with the same reaction conditions and normalized to fibre bundle dry
   weight.

2.10. Serum GDF15

   Blood was collected through cardiac puncture and allowed to clot at
   room temperature for 30 min. Blood was then spun at 1000g for 10 min
   and serum was collected. GDF15 (Growth differentiation factor 15)
   levels were analyzed in serum using the mouse GDF-15 DuoSet ELISA kit
   according to the manufacturer's instructions (R&D Systems DY6385).

2.11. RNA isolation and Rt-PCR

   To perform reverse transcription-polymerase chain reaction (Rt-PCR),
   tissue was used from a separate cohort of mice. These mice were
   ID8-inoculated as done previously and housed at the University of
   Guelph. These mice had cancer for similar times (∼45, ∼75, ∼90 days),
   and mice at the 45-day time point were 15–16 weeks old at the time of
   euthanasia. RNA isolation was performed twice for two separate
   analyses.

   In the first analysis, RNA isolation was performed at the University of
   Guelph using a TRIzol (Invitrogen) and RNeasy (Qiagen) hybrid protocol.
   Briefly, snap frozen TA tissue was homogenized in 1 mL of TRIzol
   reagent according to the manufacturer instructions. The RNA mixture was
   transferred to a RNeasy spin column (Qiagen) and processed according to
   the RNeasy kit instructions. RNA was quantified spectrophotometrically
   at 260 nm using a NanoDrop (ND1000, ThermoFisher Scientific INC.) and
   used for RNA sequencing.

   In the second analysis, RNA isolation was performed at York University
   on a separate cohort of tissue as previously described [[119]35]. TA
   and diaphragm samples were lysed using TRIzol reagent (Invitrogen) and
   RNA was separated to an aqueous phase using chloroform. The aqueous
   layer containing RNA was then mixed with isopropanol and loaded to
   Aurum Total RNA Mini Kit columns (Bio-Rad, Mississauga, ON, Canada).
   Total RNA was then extracted according to the manufacturer's
   instructions. RNA was quantified spectrophotometrically using the
   NanoDrop attachment for the Varioskan LUX Multimode Microplate reader
   (Thermo Scientific). Reverse transcription of RNA into cDNA was
   performed by M-MLV reverse transcriptase and oligo(dT) primers (Qiagen,
   Toronto, ON, Canada). cDNA was then amplified using aCFX384 Touch
   Real-Time PCR Detection Systems (Bio-Rad) with a SYBR Green master mix
   and specific primers ([120]STable 1). Gene expression was normalized to
   β-actin (Actb) and relative differences were determined using the ΔΔCt
   method. Values are presented as fold changes relative to the control
   group.

2.12. RNA sequencing

   RNA libraries were prepared using the NEBNext Ultra II Directional RNA
   Library Prep Kit for Illumina (NEB, E7760) according to manufacturer's
   polyA mRNA workflow at the Advanced Analysis Center at the University
   of Guelph (Guelph, Ontario, Canada). Libraries were normalized,
   denatured, diluted, and sequenced on an Illumina 2 × 100 bp NovaSeq S4
   flowcell usingv1.5 chemistry according to manufacturer's instructions.

   After demultiplexing, fastq files were uploaded into Partek and pre-
   and post-alignment quality control (QC) was performed in Partek
   (average phred quality score >35). Paired-end reads (100bp or 150bp)
   were aligned using STAR 2.7.8a [[121]44] with mm39 – RefSeq Transcripts
   98 (05/05/2021) in Partek. A minimum read cutoff of 20 was applied; all
   other settings were default. Data were normalized using counts per
   million (CPM), and Limma-voom [[122]45] was used for differential gene
   expression analysis: sham vs 45d, sham vs 75d, and sham vs 90d. Raw
   p-values were adjusted for false discovery rate (FDR) using the FDR
   step-up procedure. Gene Ontology and Reactome pathway enrichment
   analysis was completed using Enrichr and ConsensusPathDB with a
   background list, using DEGs with p < 0.05. Significance threshold for
   volcano plots were set at P < 0.01. Results are avilable in the NCBI's
   Gene Expression Omnibus database.

2.13. Statistics

   Results are expressed as mean ± SD. The level of significance was
   established at p < 0.05 for all statistics. The D'Agostino-Pearson
   omnibus normality test was first performed to determine whether data
   resembled a Gaussian distribution, and all data were subject to the
   ROUT test (Q = 0.5%) to identify and exclude outliers which was a rare
   occurrence. When data fit normal distributions, standard one way and
   two-way ANOVAs were performed. When data did not fit a Gaussian
   distribution for analysis with one independent variable, the
   Kruskal–Wallis test was used. Moreover, when data did not fit a
   Gaussian distribution for analysis with two independent variables, data
   was first log transformed then analyzed using a standard two-way ANOVA
   (See [123]SFigure 8 for log transformed analysis of [124]Figure 6,
   [125]Figure 7,K) but data was still presented in the Results as
   non-transformed data. Respective statistical tests are provided in
   figure legends. When significance was observed with an ANOVA, post-hoc
   analyses were performed with a two-stage set-up method of Benjamini,
   Krieger and Yekutieli for controlling false discovery rate (FDR) for
   multiple-group comparisons. With this method, all reported p values are
   FDR adjusted (traditionally termed “q”). All statistical analyses were
   performed in GraphPad Prism 10 (La Jolla, CA, USA).

Figure 6.

   [126]Figure 6
   [127]Open in a new tab

   Pyruvate & malate supported mitochondrial respiration in tibialis
   anterior and diaphragm muscle of epithelial ovarian cancer (EOC)
   injected mice. ADP-stimulated (State III) respiration supported by
   pyruvate (5 mM) and malate (2 mM) generating NADH was assessed in the
   absence (-creatine) and presence (+creatine) of creatine within
   tibialis anterior and diaphragm PmFBs of EOC injected mice.
   Mitochondrial respiration in the absence of creatine was assessed at
   submaximal concentrations (25 μM, 100 μM and 500 μM) in tibialis
   anterior of EOC injected (A). Mitochondrial respiration in the presence
   of creatine was also assessed at submaximal concentrations (25 μM,
   100 μM and 500 μM) in tibialis anterior of EOC injected (B). A
   schematic representative summary of changes in -Creatine/+Creatine
   pathways is depicted (C). This was repeated for the diaphragm (D–F)
   Results represent mean ± SD. n = 9–12. α p < 0.05 Control versus 45
   Day; β p < 0.05 Control versus 75 Day; δ p < 0.05 Control versus 90
   Day; θ p < 0.05 45 Day versus 90 Day; λ p < 0.05 75 Day vs 90 Day;
   ∗p < 0.05 Control versus all time points; & p < 0.05 45 Days versus all
   time points; #p < 0.05 75 Days versus all time points; $ p < 0.05
   90 Days versus all time points. All ANOVAs were followed by a two-stage
   step-up method of Benjamini, Krieger and Yukutieli multiple comparisons
   test. Voltage dependent anion channel (VDAC); adenine nucleotide
   translocator (ANT); mitochondrial creatine kinase (mtCK); adenosine
   diphosphate (ADP); adenosine triphosphate (ATP); phosphocreatine (PCr);
   creatine (Cr); creatine-independent phosphate shuttling (-Creatine);
   creatine-dependent phosphate shuttling (+Creatine). All data was
   analyzed suing a two-way ANOVA (main effects shown only). C57BL/6J
   female mice ∼75 days post PBS injection as controls (CTRL); C57BL/6J
   female mice ∼45 days post ovarian cancer injection (45 Days); C57BL/6J
   female mice ∼75 days post ovarian cancer injection (75 Days); C57BL/6J
   female mice ∼90 days post ovarian cancer injection (90 Days).

Figure 7.

   [128]Figure 7
   [129]Open in a new tab

   Complex I forward and reverse electron transfer emissions in tibialis
   anterior and diaphragm muscle of epithelial ovarian cancer (EOC)
   injected mice. Complex I forward electron transfer (FET) and complex I
   reverse electron transfer (RET) is schematically depicted (A & B). In
   FET mitochondrial H[2]O[2] emission was supported by pyruvate (10 mM)
   and malate (2 mM) to generate maximal rates and with ADP to assess
   H[2]O[2] emission during OXPHOS. This experiment was repeated to assess
   RET H[2]O[2] emission by using succinate (10 mM) as opposed to pyruvate
   and malate. FET and RET H[2]O[2] emissions were assessed in the TA of
   EOC injected mice and a summary of changes compared to control is
   depicted (C–G). This was repeated in the diaphragm (H–L). Results
   represent mean ± SD. Lettering denotes statistical significance when
   different from each other (p < 0.05). β p < 0.05 Control versus 75 Day;
   λ p < 0.05 75 Day vs 90 Day; δ p < 0.05 Control versus 90 Day. A
   one-way ANOVA or Kruskal–Wallis test was used when data did not fit
   normality in figure C, E, H and J. A two-way ANOVA was used in figured
   D, F, I and K. All ANOVAs were followed by a two-stage step-up method
   of Benjamini, Krieger and Yukutieli multiple comparisons test.
   Oxidative phosphorylation (OXPHOS); manganese superoxide dismutase
   (MnSOD); electron (e−); superoxide (O[2]^−). C57BL/6J female mice ∼75
   days post PBS injection as controls (CTRL); C57BL/6J female mice ∼45
   days post ovarian cancer injection (45 Days); C57BL/6J female mice ∼75
   days post ovarian cancer injection (75 Days); C57BL/6J female mice ∼90
   days post ovarian cancer injection (90 Days).

3. Results

3.1. Orthotopic epithelial ovarian cancer induces metastasis, ascites,
impaired functional capacity and body weight loss at advanced stages

   Immunocompetent C57BL6J mice received orthotopic injections of murine
   ID8 epithelial ovarian cancer cells ([130]Figure 1A). Tumours were
   allowed to develop for ∼45 days, ∼75 days and ∼90days post tumour
   injection to evaluate muscle responses across tumour development. This
   study was completed in two cohorts of mice to obtain sufficient tissue
   to complete all experiments. Weekly body weights (BW) in both cohorts
   were tracked post tumour injection ([131]SFigure 1A and B). BW, tumour
   weight, ascitic volume, muscle mass and spleen mass data were then
   merged as these data were collected in both cohorts.

Figure 1.

   [132]Figure 1
   [133]Open in a new tab

   The effects of transformed epithelial ovarian cancer cells (ID8)
   implantation underneath the ovarian bursa of C57BL6 mice. 1 × 10^6 ID8
   cells were injected underneath the ovarian bursa (A) and developed for
   42–48, 72–78 and 83–107 days (45 Day, 75 Day and 90 Day time points
   respectively). Control mice were injected with identical volumes of PBS
   and aged for 72–78 days. Primary ovarian tumour mass was measured at
   sacrifice (B, n = 21–24). Noticeable metastasis of ovarian cancer cells
   occurred by the 90-day time point and were photographed for qualitative
   assessment (C). Hematoxylin & eosin staining was used to assess
   mononuclear cell infiltration as an index of metastasis (D, n = 4–7, E
   Representative images; original magnification, ×20). Mice developed
   ascites after ∼75 days of ovarian cancer (F) and were tapped to prolong
   their survival (G & H, n = 24). Volitional wheel running (I, n = 8–11)
   and grip strength (J, n = 11–12) were used to assess voluntary motor
   function. Body weights were also measured every week and the delta
   weekly body weight (BW) was analyzed (K, n = 22–24). Tibia length (L,
   n = 11–12), peak body weight (M, n = 22–24), and final primary ovarian
   tumour-free body weight (N, n = 22–24) were also assessed. Percent
   change from peak body weight to final body weight was analyzed (O,
   n = 22–24). Results represent mean ± SD. Lettering denotes statistical
   significance when different from each other (p < 0.05). All data was
   analyzed using a one-way ANOVA and followed by a two-stage step-up
   method of Benjamini, Krieger and Yukutieli multiple comparisons test.
   Data that was not normally distributed was analyzed with a
   Kruskal–Wallis test followed by the same post-hoc analysis. C57BL/6J
   female mice ∼75 days post PBS injection as controls (CTRL); C57BL/6J
   female mice ∼45 days post ovarian cancer injection (45 Days); C57BL/6J
   female mice ∼75 days post ovarian cancer injection (75 Days); C57BL/6J
   female mice ∼90 days post ovarian cancer injection (90 Days).

   Primary ovarian tumour mass grew progressively, reaching a maximum of
   ∼400 mg by ∼90 days ([134]Figure 1B). Metastatic tumour spread to the
   diaphragm was noted at the 90 day timepoint ([135]Figure 1C) observed
   with abundant mononuclear cell infiltration ([136]Figure 1D and E).
   Another common secondary complication in ovarian cancer patients is the
   development of ascites fluid within the intraperitoneal space. Ascites
   developed as early as 77 days post ID8-inoculation (data not shown);
   and there was a significant increase in the amount of ascites
   paracentesis taps performed and total volume of ascites collected per
   mouse by 90 days post-inoculation ([137]Figure 1F–H). At this time,
   decreases in voluntary wheel running and grip strength were noted
   ([138]Figure 1I,J). Change in weekly BW, tibia length and peak body
   mass demonstrate how all groups grew similarly post cancer injection,
   as there were no significant differences between groups
   ([139]Figure 1K–M). However, final primary tumour-free BW and % change
   of peak BW to final BW were significantly decreased in the 90 Day group
   indicative of cachexia ([140]Figure 1N,O). All BW measurements were
   made after ascitic taps in the 90 Day group. We acknowledge that
   primary-tumour free BW does not accurately represent the full tumour
   load as it was not feasible to measure all metastasized cells. However,
   assessing final BW without subtracting primary tumour yielded the same
   statistical results (data not shown).

3.2. Muscle loss and fat loss occur at advanced stages of ovarian cancer with
no increases in GDF15, inflammatory markers or atrogenes

   In addition to BW loss, muscle mass and fat mass loss are hallmarks of
   cancer cachexia. Muscle mass was lower in the 90 Day group in the
   extensor digitorum longus (EDL; −11%), plantaris (PLT; −18%), tibialis
   anterior (TA; −19%), gastrocnemius (GA; −13%), and quadriceps
   (Quad; −13%) muscles compared to control (CTRL) whereas Soleus (SOL)
   mass did not change ([141]Figure 2A). Adipose tissue from the inguinal
   fat pad was lower in the 45 Day and 90 Day groups post ovarian cancer
   inoculation ([142]Figure 2B). Serum GDF15 – a recently identified
   cachexia regulator [[143]46] – did not change ([144]Figure 2C). Spleen
   mass was greater in the 75 and 90 Day groups suggesting an increased
   inflammatory response ([145]Figure 2D). We then measured cytokines and
   atrophy markers known to be elevated in certain clinical and
   preclinical models of cancer cachexia [[146]13,[147]46]. In the TA,
   Interleukin 6 (IL-6) mRNA was not different between groups while tumour
   necrosis factor alpha (TNF-α) was lower at 45 days compared to control
   ([148]Figure 2E). In addition, atrogin and muscle RING-finger protein-1
   (MURF-1) mRNA followed similar patterns whereby mRNA levels were lower
   at early time points with no differences compared to control in the 90
   Day group ([149]Figure 2F). In the diaphragm, IL-6 mRNA was higher in
   the 45 and 90 Day groups, while TNF-α was lower at the 45 and 75 Day
   groups ([150]Figure 2G). Atrogin mRNA was lower than control in the 90
   Day group while MURF-1 was not different between groups
   ([151]Figure 2H). It is important to note that these cytokines could be
   elevated in the serum of these mice throughout cancer development, but
   we were unable to make this assessment due to limited serum following
   GDF15 analysis. Future studies should measure the protein content of
   the cytokines in both serum and muscle.

Figure 2.

   [152]Figure 2
   [153]Open in a new tab

   The effects of ID8 implantation on muscle mass, fat mass, spleen mass,
   GDF15 and gene expression of inflammation and atrogenes. Analysis of
   muscle mass at all time points in hindlimb muscles was completed (A,
   n = 22–24; soleus (SOL), extensor digitorum longus (EDL), plantaris
   (PLA), tibialis anterior (TA), gastrocnemius (GA) and quadriceps
   (QUAD)). Subcutaneous adipose mass in the inguinal fat depot (B,
   n = 9–12), serum GDF15 (C, n = 8–11) and spleen mass (D, n = 21–22)
   were also analyzed. mRNA content of inflammatory and atrophy markers
   interleukin-6 (IL-6), tumour necrosis factor – alpha (TNF-α), atrogin
   and muscle RING-finger protein-1 (MURF-1) were measured using
   quantitative PCR in the TA and diaphragm of all groups (E-H, n = 6–8).
   Results represent mean ± SD. Lettering denotes statistical significance
   when different from each other (p < 0.05). C57BL/6J female mice ∼75
   days post PBS injection as controls (CTRL); C57BL/6J female mice ∼45
   days post ovarian cancer injection (45 Days); C57BL/6J female mice ∼75
   days post ovarian cancer injection (75 Days); C57BL/6J female mice ∼90
   days post ovarian cancer injection (90 Days). All data was analyzed
   using a one-way ANOVA or Kruskal–Wallis test when data did not fit
   normality. All ANOVAs were followed by a two-stage step-up method of
   Benjamini, Krieger and Yukutieli multiple comparisons test.

3.3. Muscle atrophy in the absence of muscle regeneration occurs earlier in
the diaphragm compared to TA throughout ovarian cancer development

   Muscle atrophy is also a hallmark of cancer cachexia. TA and diaphragm
   muscles were sectioned and tagged for MHC isoforms I, IIA and IIB;
   black fibers were assumed IIx. In the TA, fiber CSA was not different
   in the 45 Day group in any isoform or when isoforms were pooled
   ([154]Figure 3A–C). In the 75 Day group, TA exhibited lower CSA in
   pooled fibers (−11%) with specific reductions in MHCIIx isoforms
   ([155]Figure 3A–C). In the 90 Day group, TA exhibited an exacerbated
   reduction in pooled fiber CSA (−23%) with specific reductions in all
   isoforms compared to control ([156]Figure 3A–C). This is further
   exemplified when pooled fibers are presented in a histogram as a higher
   frequency of smaller fibers exist in the TA in the 90 Day group
   compared to control ([157]Figure 3D). We also demonstrate muscle
   atrophy was occurring in the absence of apparent regeneration by
   measuring eMHC to evaluate if new fibers were developing. In the TA no
   eMHC fibers were identified in any groups ([158]Figure 3E,F). We also
   performed a fiber type distribution analysis in the red TA between the
   CTRL and 90 Day group to evaluate if a fiber type shift occurred
   throughout cancer development. No differences were found between groups
   ([159]SFigure 3).

Figure 3.

   [160]Figure 3
   [161]Open in a new tab

   Evaluation of tibialis anterior and diaphragm fiber type atrophy and
   fiber regeneration in epithelial ovarian cancer injected mice. Analysis
   of fiber histology on myosin heavy chain (MHC) isoforms and eMHC was
   performed in control and EOC mice. Cross-sectional area (CSA) of MHC
   isoforms were evaluated in the tibialis anterior (A-C, n = 7–10). All
   fiber types were also pooled, binned and averaged based off fiber area
   and plotted by frequency distribution at each time point compared to
   control (D, n = 7–10). Embryonic MHC (eMHC) was tagged in separate
   sections to evaluate the presence of new fibers (E & F, n = 7–10). This
   was repeated within the diaphragm (G-L, n 10 = 14). Results represent
   mean ± SD. Lettering denotes statistical significance when different
   from each other (p < 0.05). α p < 0.05 Control versus 45 Days; β
   p < 0.05 Control versus 75 Days; δ p < 0.05 Control versus 90 Days. A
   one-way ANOVA was used for figures A, B, G and H. Data that was not
   normally distributed was analyzed with a Kruskal–Wallis test. A two-way
   ANOVA was used for figures D and J (interactions shown only). All
   ANOVAs were followed by a two-stage step-up method of Benjamini,
   Krieger and Yukutieli multiple comparisons test. C57BL/6J female mice
   ∼75 days post PBS injection as controls (CTRL); C57BL/6J female mice
   ∼45 days post ovarian cancer injection (45 Days); C57BL/6J female mice
   ∼75 days post ovarian cancer injection (75 Days); C57BL/6J female mice
   ∼90 days post ovarian cancer injection (90 Days).

   In the diaphragm, muscle atrophy was evident earlier than in the TA. In
   the 45 Day group, diaphragm fibers exhibited lower CSA stained with
   MHCIIa and MHCIIb isoforms as well as lower CSA in pooled fibers (−12%)
   ([162]Figure 3G–I). In the 75 Day group, fiber CSA remained lower in
   pooled fibers (−13%) specifically in MHCIIa and MHCIIb isoforms
   compared to control and were similar to the 45 Day group
   ([163]Figure 3G–I). In the 90 Day group, diaphragm CSA exhibited
   extensive reductions when isoforms were assessed separately or pooled
   (−24%) ([164]Figure 3G–I). This is further demonstrated when pooled
   fibers are displayed in a histogram as a higher frequency of smaller
   fibers were present at 90 days ([165]Figure 3J). This atrophy in the
   diaphragm also occurred in the absence of apparent regeneration as no
   eMHC fibers were identified ([166]Figure 3K,L). As muscle strips were
   used for diaphragm histology, a fiber type distribution analysis could
   not be performed as done on the TA.

3.4. Early muscle weakness is further reduced in the diaphragm throughout
ovarian cancer progression but gradually recovers in the TA

   Muscle weakness is another hallmark of cancer cachexia. In the 45 Day
   group, TA specific force production was lower compared to control
   ([167]Figure 4A,B). In contrast, force production increased
   progressively by 75 days and 90 days, while still remaining lower
   compared to control (main effect) ([168]Figure 4A,B). The rate of
   contraction (Df/dt) at 1Hz stimulation frequency was lower at 45 Day
   group compared to control but not different at 100Hz ([169]Figure 4C).
   In addition, the half relaxation time (HRT) was significantly longer in
   the 90 Day group at both stimulation frequencies ([170]Figure 4D). This
   suggests that at 45 days the TA exhibits a slower rate of contraction
   at lower stimulation frequencies, and at 90 days the TA exhibits slower
   relaxation. We also measured mRNA content of ryanodine receptors (RyR1)
   and sarcoendoplasmic reticulum calcium ATPase (SERCA) as these proteins
   are integral for the regulation of calcium release and reuptake that
   regulates contraction. RyR1 mRNA content was not different across the
   time points, however, SERCA1 mRNA content was higher in the 90 Day
   group compared to control ([171]Figure 4E).

Figure 4.

   [172]Figure 4
   [173]Open in a new tab

   The effects of epithelial ovarian cancer (EOC) on tibialis anterior and
   diaphragm force production, contractile properties and calcium handling
   gene expression. In situ tibialis anterior force production was
   assessed using the force-frequency relationship (A, n = 9–10; B,
   Representative twitches at 1 Hz and 100Hz). Rate of twitch contractions
   along with the half relaxation time were also assessed at 1Hz and 100Hz
   (C & D, n = 18–22). mRNA expression of ryanodine receptors (RyR1) and
   sarcoplasmic/endoplasmic reticulum ATPase (SERCA; SERCA1 used for
   tibialis anterior (fast twitch) and SERCA2 for diaphragm (slow twitch)
   was also measured (E, n = 8). This was repeated for the diaphragm (F-J,
   n = 8–22) Results represent mean ± SD. ∗p < 0.05 Control versus all
   time points; & p < 0.05 45 Days versus all time points; #p < 0.05
   75 Days versus all time points; $ p < 0.05 90 Days versus all time
   points. Lettering denotes statistical significance at an alpha set at
   p < 0.05. A two-way ANOVA was used for figures A and J (main effects
   shown only) and all other data was analyzed using a one-way ANOVA or
   Kruskal–Wallis test when data did not fit normality. All ANOVAs were
   followed by a two-stage step-up method of Benjamini, Krieger and
   Yukutieli multiple comparisons test. C57BL/6J female mice ∼75 days post
   PBS injection as controls (CTRL); C57BL/6J female mice ∼45 days post
   ovarian cancer injection (45 Days); C57BL/6J female mice ∼75 days post
   ovarian cancer injection (75 Days); C57BL/6J female mice ∼90 days post
   ovarian cancer injection (90 Days).

   Diaphragm force production was also significantly lower compared to
   control in the 45 Day group as a main effect (interaction at 40Hz
   onward; not shown), with no changes in contractile properties but
   decreases in RyR1 and SERCA2 mRNA contents ([174]Figure 4F–J).
   Interestingly, in the 75 Day group, diaphragm specific force
   transiently increased compared to the 45 Day group while still
   remaining lower than control as a main effect (interaction at 40Hz
   onward; not shown), with no changes in the rate of contraction, time to
   relaxation or mRNA content of RyR1 or SERCA2 ([175]Figure 4F–J). In the
   90 Day group, force production was further lowered compared to all time
   points as a main effect (interaction at 40Hz onward; not shown) with
   longer half relaxation time at 100Hz stimulation frequency
   ([176]Figure 4F–I). This decrease in force production and increase in
   half relaxation time was coupled to decreases in RyR1 and SERCA2 mRNA
   content ([177]Figure 4J). These data demonstrate that each muscle
   demonstrates unique contractile adaptations to ovarian cancer over
   time.

3.5. Mitochondrial genes are downregulated in the TA during early and
advanced stages of epithelial ovarian cancer

   Given there were no changes in markers of atrophy-related mechanisms
   that have been found in other cachexia models (GDF15, TNF-α, Atrogin
   and MURF-1 [[178]13,[179]46], we examined the potential role for
   mitochondrial stress responses that have been identified in other
   models [[180]21,[181]47]. In the TA, RNA-seq revealed 691, 795 and 3402
   differentially expressed genes (DEGs) in the 45, 75, and 90 Day groups
   compared to control, respectively. Some of these DEGs are shared among
   time points ([182]Figure 5A). We then used a volcano plot to
   demonstrate DEGs that were upregulated or downregulated compared to
   control ([183]Figure 5B–D). Gene ontology (GO) enrichment analyses was
   then performed on DEGs to investigate up and down regulated biological
   processes altered across time points compared to control
   ([184]Figure 5E–G). Several pathways were significantly enriched, with
   the majority of downregulated pathways being mitochondrial-related. The
   two most significantly different upregulated and downregulated pathways
   from the enrichment analysis were represented in a chord plot to
   exemplify the specific genes changing across ovarian cancer progression
   ([185]Figure 5H–J). These results largely demonstrate how more
   mitochondrial genes are down regulated as ovarian cancer progresses
   with increases in certain muscle contraction related genes in the 90
   Day group ([186]Figure 5H–J).

Figure 5.

   [187]Figure 5
   [188]Open in a new tab

   RNA sequencing analysis of tibialis anterior muscle in epithelial
   ovarian cancer (EOC) injected mice. Number of differentially expressed
   genes (DEGs) in each comparison were exemplified in a Venn diagram (A).
   Volcano plot showing -log10 p-value and log2 fold changes of DEGs for
   each comparison were also completed (B–D). Top 3 upregulated and top 3
   downregulated biological processes enriched in DEGs at each time point
   were also analyzed and graphed (E–G) Top two upregulated and down
   regulated biological processes were also used to generate a chord plot
   with the corresponding DEGs and respective log fold changes at each
   time point (H–J). n = 6. C57BL/6J female mice ∼75 days post PBS
   injection as controls (CTRL); C57BL/6J female mice ∼45 days post
   ovarian cancer injection (45 Days); C57BL/6J female mice ∼75 days post
   ovarian cancer injection (75 Days); C57BL/6J female mice ∼90 days post
   ovarian cancer injection (90 Days).

3.6. Decreases in carbohydrate supported mitochondrial respiration occur in
early stages of ovarian cancer progression but is restored by late-stage
disease

   Using permeabilized muscle fibres, mitochondrial respiration was
   assessed by stimulating complex I with NADH generated by pyruvate
   (5 mM; carbohydrate substrate) and malate (2 mM) across a range of ADP
   concentrations to challenge mitochondria over a spectrum of metabolic
   demands. The ADP titrations were repeated without (-Creatine;
   [189]Figure 6A,D) and with creatine (+Creatine; [190]Figure 6B,E) in
   the assay media to model the two main theoretical mechanisms of high
   energy phosphate shuttling from the mitochondria to the cytosol
   [[191]43,[192][48], [193][49], [194][50]]. Briefly, in the absence of
   creatine, ATP is exported across the double membranes while in the
   presence of creatine, matrix-derived ATP crosses the inner membrane and
   is used by mitochondrial creatine kinase in the intermembrane space to
   phosphorylate creatine. The phosphocreatine product is then exported
   across the outer membrane which is then used by cytosolic creatine
   kinases to re-phosphorylate ADP local to ATP consuming proteins.
   Previous studies have shown that ADP/ATP flux is much slower than
   creatine/phosphocreatine due to the diffusion limitations of ATP/ADP vs
   phosphocreatine/creatine [[195]51]. Prior modeling experiments have
   posited that up to 80% of the phosphate exchange between mitochondria
   and cytoplasm (in muscle) is likely comprised of the creatine-dependent
   system. 25 μM, 100 μM and 500 μM ADP were selected to reflect creatine
   sensitivity as this is within the predicted range that is sensitive to
   the effects of mitochondrial creatine kinase (mtCK) [[196]29,[197][51],
   [198][52], [199][53]]. Maximal ADP-stimulated respiration was unchanged
   in both muscles, in each condition, at all-time points
   ([200]SFigure 5A-D) indicating that cancer alters the regulation of
   mitochondrial pyruvate oxidation stimulated by submaximal ADP
   concentrations.

   Within the 45 and 75 Day groups, TA and diaphragm exhibit a decrease in
   mitochondrial respiration in the -Creatine condition with an apparent
   supercompensation by 90 days ([201]Figure 6A,D). In the +Creatine
   condition, early decreases in mitochondrial respiration in both muscles
   returned to normal by the 90-day time point ([202]Figure 6B,E). A
   summary of all respiration changes across time and between conditions
   as a main effect versus control is provided for both muscles
   ([203]Figure 6C,F). All changes in mitochondrial respiratory control
   were not influenced by changes in mitochondrial electron transport
   chain (ETC) content as there were no changes in ETC subunit content
   estimated via western blot ([204]SFigure 4A & 4B).

   Another approach for evaluating changes in the relative control exerted
   by mitochondrial creatine kinase on ADP-stimulated respiration, or
   ‘mitochondrial creatine sensitivity’, is to calculate the ratio of
   respiration at submaximal ADP concentrations in
   both +Creatine/-Creatine conditions
   [[205]21,[206]29,[207]34,[208]51,[209]54]. In the TA, creatine
   sensitivities were unchanged across time points indicating alterations
   in respiration were similar between both high energy phosphate
   shuttling systems ([210]SFigure 5E). While both -Creatine and +Creatine
   respiration changed across time points, the relative changes were
   disproportionate within the diaphragm, particularly in the 90 Day group
   compared to control. More specifically, the -Creatine system exhibited
   a greater increase in respiration compared wo the +Creatine system,
   thus, mitochondrial sensitivity to creatine was decreased compared to
   control in the 90 Day group, but this does not necessarily reflect
   compromised energy transfer from mitochondria to cytoplasm given both
   systems nonetheless improved. ([211]SFigure 5F), indicating the
   creatine-dependent phosphate shuttling system is selectively impaired
   compared to the creatine-independent system. This effect was not
   explained by mtCK protein contents given they were unchanged at all
   time points ([212]SFigure 5G & 5H).

   We also measured fat oxidation in each muscle in order to determine if
   these changes in respiration were unique to pyruvate stimulation of
   respiration through Complex I (NADH). In the TA, there were no changes
   in palmitoyl CoA-supported respiration ([213]SFigure 6B & 6D)
   suggesting that the decreases in pyruvate oxidation were not due to a
   dysfunction that impacted all substrates. Likewise, there were no
   changes in Complex II-supported respiration (succinate; FADH[2]) and
   glutamate-supported respiration (further NADH generation; Complex I) in
   both +Creatine/-Creatine conditions ([214]SFigure 7A-D). This suggests
   that the decrease in pyruvate oxidation were also not due to Complex I
   impairments per se nor ETC components downstream of both Complex I and
   II (see discussion). Interestingly, there was an increase in state II
   pyruvate/malate respiration in the 90 Day group in the absence of
   creatine suggesting increased proton leak, but no changes were observed
   in the presence of creatine ([215]SFigure 7E & 7F) nor in response to
   palmitoyl CoA ([216]SFigure 6A).

   The diaphragm also exhibited no changes in palmitoyl CoA
   ([217]SFigure 6C & 6D) respiration. However, Complex II-supported
   respiration (succinate; FADH[2]) in the -Creatine condition was higher
   in the 90 day group along with glutamate-supported respiration (further
   NADH generation; Complex I) in both creatine conditions
   ([218]SFigure 7G-J). There were no changes in state II
   pyruvate/malate-supported respiration ([219]SFigure 7K & 7L).

3.7. Increased mitochondrial H[2]O[2] emissions occur in complex I forward
and reverse electron transfer in this EOC model

   We stimulated complex I-supported mH[2]O[2] with forward electron
   transfer (pyruvate and malate (2 mM) to generate NADH) ([220]Figure 7A)
   and reverse electron transfer (succinate to generate FADH[2])
   ([221]Figure 7B) [[222][55], [223][56], [224][57], [225][58]]. These
   substrate-specific maximal mH[2]O[2] kinetics were followed by
   titration of ADP to determine the ability of ADP to attenuate mH[2]O[2]
   during oxidative phosphorylation (OXPHOS). In the TA, pyruvate/malate
   and succinate supported maximal mH[2]O[2] was not different at any time
   point compared to control ([226]Figure 7C,E). However, pyruvate/malate
   supported mH[2]O[2] during OXPHOS were increased in the 75 Day group
   which returned to control levels by the 90 Day group, while succinate
   supported mH[2]O[2] was increased at the 75 and 90 Day groups
   ([227]Figure 7D,F). There were no changes in diaphragm pyruvate/malate
   and succinate supported maximal mH[2]O[2] at any time point
   ([228]Figure 7H,J). The diaphragm exhibited no changes in pyruvate &
   malate-supported H[2]O[2] during OXPHOS but exhibited higher
   succinate-supported mH[2]O[2] during OXPHOS in the 75 Day group that
   returns to baseline by the 90 Day group ([229]Figure 7I,K). A summary
   of mH[2]O[2] changes across time and between conditions as a main
   effect versus control is provided for the TA ([230]Figure 7G) and
   diaphragm ([231]Figure 7L).

   A comprehensive summary of all changes between the TA and diaphragm
   captured within this study design is provided ([232]Figure 8).

Figure 8.

   [233]Figure 8
   [234]Open in a new tab

   Summary of changes in a metastatic epithelial ovarian cancer cachexia
   model. When mice were injected with epithelial ovarian cancer, at a
   pre-metastasis time point (45 days post-injection) early muscle
   weakness was associated with decreases in pyruvate oxidation in both
   the tibialis anterior and diaphragm muscles. At this time, the tibialis
   anterior muscle did not exhibit muscle atrophy while the diaphragm did.
   With the exception of IL-6 in the diaphragm, there were no increases in
   TNF-
   [MATH: <mrow><mi>α</mi></mrow> :MATH]
   and atrophy markers of cancer cachexia at this time. During severe
   metastasis (90 days post-injection) both muscles exhibited muscle
   atrophy and muscle weakness, however the tibialis anterior recovered
   specific force production. Moreover, both muscles exhibited
   compensatory increases in submaximal pyruvate oxidation. Last, with the
   exception of IL-6 there were still no increases in TNF-
   [MATH: <mrow><mi>α</mi></mrow> :MATH]
   and atrophy markers of cancer cachexia.

4. Discussion

   Epithelial ovarian cancer is the most lethal gynecological cancer in
   women. Advanced stages of this disease cause severe muscle weakness,
   yet the mechanisms remain unknown due in part to the inherent
   challenges of studying clinical population as well as a limited
   selection of pre-clinical models. Moreover, it has been suggested that
   the paucity of metastatic, orthotopic models for cancer cachexia has
   contributed to failures in clinical trials of therapies designed to
   preserve muscle mass and/or function that were otherwise based on
   evidence from other types of preclinical models [[235]11]. Modelling
   cachexia in a metastatic context is believed to greatly improve the
   predictive power of preclinical models for identifying mechanisms and
   therapy development [[236]11]. Here, we developed a new immunocompetent
   mouse model of metastatic cancer cachexia reflective of late-stage
   ovarian cancer while retaining the clinically relevant aspects of
   tumour growth in the nascent organ that can be induced during
   adulthood. Importantly the findings of this study demonstrate that
   early muscle weakness precedes clinical signs of ovarian cancer
   including metastasis and ascites formation as well as atrophy. The
   eventual development of atrophy seemingly triggers an adaptive response
   whereby specific force is restored in a sustained manner within limb
   muscle but only transiently in diaphragm. These pathological and
   adaptive responses in muscle quality during ovarian cancer coincided
   with dynamic alterations in pyruvate oxidation and mRNA contents
   related to numerous mitochondrial pathways but without increases in the
   cachexia-regulating atrogene programs, TNF-α or GDF15 that have been
   attributed in this process for other cancers [[237]13,[238]46].

   Collectively, the time-dependent responses in force production and
   fibre size in this new model of ovarian cancer-induced cachexia serve
   as a foundation for exploring numerous potential mechanisms underlying
   the development of atrophy-independent and -dependent weakness in
   relation to metastasis. The findings also highlight how mitochondrial
   stress is a defining feature of muscle weakness during ovarian cancer
   [[239]21,[240]47].

4.1. A novel atrophy-independent weakness in locomotor muscle during ovarian
cancer

   At the earliest time point assessed (45 Day group), both TA and
   diaphragm demonstrate lower specific force production. As this
   measurement is normalized to the size of muscle, any occurrence of
   atrophy cannot explain this observation. In fact, while modest atrophy
   was observed in the diaphragm, fibre size was unchanged in the TA at
   this timepoint. Therefore, the TA demonstrated a pre-atrophy weakness
   similar to our previous findings in the C26 colorectal cancer model
   that weakness precedes atrophy in the quadriceps and the diaphragm
   [[241]21]. Hence, pre-atrophy weakness has now been reported in two
   distinct models suggesting it is a common phenomenon during cancer.
   This finding is important because it raises questions regarding the
   atrophy-independent mechanisms of muscle weakness during cancer – a
   topic that is considerably understudied in contrast to the literature
   on atrophy-dependent muscle weakness during cachexia. While it is
   possible that weakness precedes atrophy in the diaphragm of the current
   EOC model, future studies would need to examine an earlier time course
   design.

4.2. Reduced mitochondrial pyruvate oxidation is associated with early muscle
weakness during ovarian cancer

   In the TA, RNAseq identified nuclear genes encoding mitochondrial
   proteins as the most dominant gene expression stress response early
   during cancer (45 Day group) when tumours were just appearing, and well
   before severe metastasis or atrophy developed. High resolution
   respirometry revealed that this stress response corresponds with
   reduced pyruvate oxidation – an index of carbohydrate oxidation – but
   with no corresponding changes in the capacity for long chain fatty acid
   oxidation. This finding of a substrate-specific change in respiration
   was similar in diaphragm at this early time point. There were also no
   changes in the oxidation of the amino acid-derived substrate glutamate
   or succinate (generation of FADH[2] at complex II). As both pyruvate
   and glutamate generate NADH through their respective dehydrogenases
   (pyruvate dehydrogenase (PDH) and glutamate dehydrogenase (GDH)), the
   findings suggest that the ability of Complex I to oxidize NADH may not
   have been altered in the TA. This suggests that the unique reduction in
   pyruvate oxidation at 45-days may be due to changes in PDH itself.
   While there were no changes in PDH or PDH phosphatases mRNA expression
   identified with RNAseq at this time point (data not shown), future
   studies could determine if isolated PDH activity is inhibited similar
   to indications from previous reports in C26 mice [[242]59].

   Considering that measurements of oxygen consumption reflect reduction
   of O[2] at Complex IV downstream of both Complex I and II, the lack of
   decreases in both glutamate and succinate oxidation in both TA and
   diaphragm indicates that the integrated function of the electron
   transport chain was not altered, at least as could be detected within
   the physiologically relevant context of ADP-stimulated (coupled)
   respiration. This finding is interesting given that protein contents of
   specific subunits of ETC complexes were not changed across time points.
   In the 75 Day and 90 Day groups, the protein contents of ETC protein
   subunits measured by western blot were not changed, but mRNA content of
   these subunits were decreased (data not shown; C1 - NDUFB8, CII – SDHB,
   CIII – UQCRC2, CV – ATP5A). While speculative, these findings suggest a
   number of possibilities including increased ETC protein stability or
   that reductions in mRNA reflect increased translation rather than
   decreased gene expression [[243]60].

   Given that ovarian cancer does not reduce oxidation of the other
   substrates explored at this early timepoint, it is also difficult to
   define the unique reductions in pyruvate oxidation as a ‘mitochondrial
   dysfunction’ per se. This is a critical outcome of the present
   investigation and highlights the importance of comparing substrates to
   each other and to defined primary outcomes of myopathy tracked over
   time, and across muscle types. This approach determines whether
   mitochondrial stress responses affect a central governance of oxidative
   phosphorylation or an adaptive reprogramming - a concept and
   perspective we have proposed previously for the study of myopathies
   [[244]61].

4.3. Recovery of specific force during the development of atrophy is more
sustained in limb muscle versus diaphragm

   As atrophy developed by the 75 Day group in both muscles, specific
   force partially recovered despite the appearance of atrophy in the TA
   with even greater atrophy in the diaphragm. As specific force is a
   measure that is independent of muscle mass, this finding raises
   questions regarding the mechanism of intrinsic improvements within the
   atrophied muscle itself. This remarkable adaptive response in both
   muscles diverged in the 90 Day group. Particularly, specific force
   recovered even further in the TA, whereas diaphragm force plummeted to
   very low levels. Interestingly, severe metastasis occurred on the
   diaphragm in the 90 Day group, suggesting that metastases to the
   diaphragm may contribute directly to the progressive loss of force
   production at advanced stages of ovarian cancer cachexia.

   Unlike the 45 Day group, changes in force production were not
   consistently related to changes in pyruvate oxidation given this
   function remained low as force recovered at 75 days, with increases in
   pyruvate oxidation at 90 days being positively or inversely related to
   muscle force production in the TA and diaphragm, respectively.
   Nonetheless, decrements in pyruvate oxidation were more strongly
   related to pre-atrophy weakness early during cancer. This finding
   warrants further examination with targeted approaches that determine
   whether altered glucose metabolism contributes to weakness uniquely at
   early stages of ovarian cancer.

   As the mechanisms underlying the progressive vs transient increase in
   force production in both muscles require further investigation, RNAseq
   analyses in the TA demonstrated mRNA contents related to chromatin
   regulation and biosynthetic processes that can be explored for the
   generation of numerous hypotheses in future investigations. Likewise,
   increased mRNA contents corresponding to genes related to myofibril
   assembly and actomyosin structure organization during the pre-atrophy
   period at 45 days in the TA suggests potential turnover of sarcomeric
   structures may have occurred. Future studies could consider whether
   pre-atrophy weakness was due to declining quality of contractile
   machinery. Last, Reactome enrichment analysis identified 19 “muscle
   contraction” genes upregulated at the 90-day time point in the TA. Some
   of these genes are related to calcium handling (ATP2b4, ITPR2, RyR1,
   and ATP2a1), suggesting increases in force production could be related
   to calcium regulation. While RNAseq was not performed in the diaphragm,
   Rt-PCR analysis did identify decreases in RyR and SERCA mRNA content
   concurrent with a decrease in force-frequency production. Functional
   calcium handling measures could be performed in the future to explore
   potential relationships between mitochondrial ATP supply supported by
   carbohydrate oxidation and the energetic cost of contraction [[245]62].

4.4. Mitochondrial-cytoplasmic phosphate shuttling: two systems, two
different responses during ovarian cancer

   This study was also designed to consider how mitochondria shuttle
   phosphate to the cytoplasm in the form of both PCr (Phosphocreatine)
   and ATP in order to gain deeper insight into the precise mechanisms by
   which skeletal muscle mitochondria demonstrate metabolic reprogramming
   (as explained in Results and in [[246]51]). These modeling approaches
   identify early reductions in pyruvate oxidation in both TA and
   diaphragm that were observed more consistently in the dominant
   creatine-dependent pathway (PCr export) yet both phosphate shuttling
   systems (ATP and PCr export) improved over time. The late stage
   increases in both systems, with an apparent supercompensation in the
   creatine-independent (ATP) shuttling system, indicate a mitochondrial
   ‘hormesis’ consistent with a perspective that mitochondria attempt to
   enhance the supply of ATP to a failing muscle fibre as the stress of
   cancer intensifies.

   Collectively, this experimental design led to findings that can guide
   pre-clinical therapy development to treat cancer-induced muscle
   weakness. For example, the relationships would support further
   investigation into therapies that preserve pyruvate oxidation or
   creatine-dependent metabolism early in cancer could be explored to
   determine if the pre-atrophy weakness can be prevented.

4.5. mH[2]O[2] emission: a delayed relationship with weakness?

   mH[2]O[2] emission was not elevated at the early 45 Day timepoint
   corresponding to muscle weakness in both muscles. Therefore, there was
   a stronger relationship between early weakness and reduced pyruvate
   oxidation and creatine-dependent respiration in both muscles than to
   oxidative stress. Rather, mH[2]O[2] emission was increased by the 75
   Day group in both muscles. By the 90 Day group, mH[2]O[2] emission
   returned to control levels in both muscles although this depended on
   the pathway assessed. While mH[2]O[2] emission derived from the reverse
   electron transfer pathway was more consistently elevated in both
   muscles, the unique time-dependent patterns of both systems further
   highlight the complexities of mitochondrial reprogramming that would
   not be captured by traditional single pathway analyses. Collectively,
   there is a clear increase in mH[2]O[2] in mid to late stages of cancer
   corresponding to atrophy. Therefore, the findings serve as a basis for
   examining the potential roles of mitochondrial-derived redox signals in
   regulating muscle fibre size distinct from mechanisms governing earlier
   muscle weakness that was more strongly related to changes in oxidative
   phosphorylation in this model. The findings also suggest that
   mitochondrial-targeted antioxidants could be tested to determine if
   these mH[2]O[2] responses are partially contributing to atrophy during
   ovarian cancer. Indeed, a previous study in C26 cancer mice
   demonstrated the mitochondrial cardiolipin-targeting small peptide
   SS-31 prevented atrophy at later stages of development in relation to
   lower mH[2]O[2] emission [[247]63].

4.6. Mechanisms regulating cachexia in an orthotopic, metastatic, epithelial
ovarian cancer model appear to differ from other pre-clinical models

   Contemporary theories propose muscle wasting during cancer cachexia is
   induced by circulating factors generated by the host or tumour which
   trigger protein degradation and loss of myofibrillar proteins
   [[248]4,[249]64,[250]65]. Several genes and cytokines are thought to
   regulate this skeletal muscle degradation but atrogin, Murf-1, TNF-a,
   IL-6, and GDF15 are perhaps most commonly identified and measured
   [[251]13,[252]46]. However, the current investigation using an
   immunocompetent, orthotopic, metastatic model of ovarian cancer does
   not demonstrate robust activation of these pathways. Indeed, with the
   exception of IL-6 in the diaphragm, the factors regulating muscle
   atrophy seem to be different within the current model. The
   time-specific increases in IL-6 in the diaphragm could be explored
   given this cytokine is integral for the development of muscle loss in
   the Apc^min/+ mouse (genetic spontaneous colorectal cancer model)
   [[253]66,[254]67].

   The absence of atrogene responses is similar to a patient-derived
   xenograft (PDX) model whereby, pancreatic ductal adenocarcinoma (PDAC)
   tumours from cancer patients are orthotopically injected into
   immunodeficient NSG mice [[255]68]. Within this model, the TA
   demonstrates up-regulation in canonical atrophy-associated pathways
   (ubiquitin-mediated protein degradation), while the diaphragm
   demonstrates an up-regulation in genes related to the inflammatory
   response [[256]68]. This could suggest that orthotopic models
   demonstrate distinct cachexia profiles between muscles that are unique
   to ectopic models.

   Reductions in food intake and physical activity are thought to
   contribute partially to cachexia [[257]69,[258]70]. The degree to which
   these patterns contribute to pre-atrophy weakness or atrophy itself in
   the current study requires further investigation. However, previous
   work in C26 mice has shown that reductions in food intake did not
   contribute to reduced muscle weights, fiber CSA, or muscle force given
   pair-fed mice retained normal muscle parameters compared to tumour
   bearing mice [[259]71].

4.7. Perspectives, limitations and conclusions

   The discovery that muscle weakness and mitochondrial stress precedes
   severe metastasis and ascites accumulation in ovarian cancer raises the
   question of whether pre-atrophy weakness could be an early diagnostic
   marker of cancer, given that ovarian malignancy is largely undetectable
   at premetastatic stages. Indeed, most women are diagnosed with ovarian
   cancer at stage III – a time where metastasis/ascites have already
   started, and survival rates are low [[260]22,[261]23,[262]72]. Thus,
   early cancer detection is suggested to be one of the best strategies
   for cancer prevention [[263]73]. Exploring this possibility would be
   complex given muscle weakness and altered mitochondrial functions could
   occur in other health conditions such as ageing and muscle disuse
   [[264]74,[265]75]. These findings also position mitochondrial
   reprogramming as a potential therapeutic target in pre-atrophy weakness
   and cachexia during metastatic ovarian cancer.

   While we measured monocular cell infiltration to the diaphragm, we
   could not repeat these measures in all organs within the
   intraperitoneal space (e.g. stomach, colon, liver, spleen, etc.) to
   confirm the lack of metastasis at the earlier time points with high
   confidence. In this regard, we cannot definitively state that muscle
   weakness precedes metastasis per se, but rather severe metastasis as
   evidenced by the visual confirmation in the diaphragm and lack of
   obvious tumours by sight throughout the abdominal cavity (data not
   shown). Future studies could use dye-based imaging methods to track and
   characterize metastatic disease as done previously [[266]76] to
   identify if mitochondrial stress and muscle weakness precede the
   earliest onset of metastasis.

   Of interest, chemoresistance during ovarian cancer is commonly
   associated with disease recurrence after platinum-based treatments
   [[267]77] at which point cachexia can become more evident. Although not
   investigated in the current study, it is worth highlighting that this
   EOC model is responsive to platinum-based chemotherapies [[268]78].
   Thus, future studies could consider using this model to investigate the
   interactions of chemotherapy and ovarian tumour-related stressors on
   cachexia.

   In conclusion, this is the first mouse model of epithelial ovarian
   cancer-induced muscle weakness that offers the advantages of orthotopic
   injections of EOC cells into the ovarian bursa that can be performed in
   immunocompetent mice during adulthood. The model also demonstrates the
   critical clinical feature of metastasis in the abdominal cavity similar
   to what occurs in women with late-stage ovarian cancer. The
   identification of an early muscle weakness that precedes both atrophy
   and severe metastasis provide a new direction for research in
   understanding the atrophy-independent mechanisms of muscle weakness
   during ovarian cancer. We identified substrate-specific alterations in
   mitochondrial oxidative phosphorylation and increases in mitochondrial
   reactive oxygen species that coincide with early pre-metastatic
   weakness. The model also demonstrates late-stage apparent compensatory
   relationships between mitochondrial metabolism and specific force
   restoration in limb muscle suggesting a remarkable adaptive mechanism
   that appears to be muscle-specific. The time-dependent and
   muscle-specific relationships described in this new model will support
   continued efforts in defining atrophy-independent and -dependent
   mechanisms of weakness during ovarian cancer in relation to metastasis
   and for guiding the design of pre-clinical therapy development
   investigations.

CRediT authorship contribution statement

   Luca J. Delfinis: Writing – review & editing, Writing – original draft,
   Visualization, Validation, Project administration, Methodology,
   Investigation, Formal analysis, Conceptualization. Leslie M. Ogilvie:
   Writing – review & editing, Methodology, Investigation,
   Conceptualization. Shahrzad Khajehzadehshoushtar: Writing – review &
   editing, Methodology, Investigation, Conceptualization. Shivam Gandhi:
   Writing – review & editing, Investigation. Madison C. Garibotti:
   Writing – review & editing, Investigation. Arshdeep K. Thuhan: Writing
   – review & editing, Investigation. Kathy Matuszewska: Methodology.
   Madison Pereira: Methodology. Ronald G. Jones: Writing – review &
   editing, Formal analysis, Data curation. Arthur J. Cheng: Writing –
   review & editing, Validation. Thomas J. Hawke: Writing – review &
   editing, Validation. Nicholas P. Greene: Writing – review & editing,
   Validation. Kevin A. Murach: Writing – review & editing, Validation,
   Formal analysis. Jeremy A. Simpson: Writing – review & editing,
   Methodology, Conceptualization. Jim Petrik: Writing – review & editing,
   Methodology, Conceptualization. Christopher G.R. Perry: Writing –
   review & editing, Writing – original draft, Visualization, Validation,
   Supervision, Project administration, Methodology, Funding acquisition,
   Conceptualization.

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgment