2024.01.16.575611v1.full-html.html

CPT1a regulates the delivery of extracellular fatty acids for cardiolipin turnover in

prostate cancer cells.

Nancy T Santiappillai

, Mariam F Hakeem-Sanni

, Anabel Withy

, Lisa M Butler

2,3

, Lake-Ee

Quek

, Andrew J Hoy

1,5

The University of Sydney, Charles Perkins Centre, School of Medical Sciences, Sydney, New South Wales, 2006,

Australia.

South Australian Immunogenomics Cancer Institute and Freemasons Centre for Male Health and Wellbeing,

University of Adelaide, Adelaide, SA 5005, Australia

South Australian Health and Medical Research Institute, Adelaide, SA 5000, Australia

The University of Sydney, Charles Perkins Centre, School of Mathematics and Statistics, Sydney, New South

Wales, 2006, Australia.

Lead contact. Tel: +61 2 9351 2514; Email: andrew.hoy@sydney.edu.au

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Abstract

Mitochondrial fatty acid oxidation (FAO) has been proposed to be a major bioenergetic

pathway in prostate cancer. However, this concept fails to consider FAO relative to other

mitochondrial substrates. Here, we found extracellular long-chain fatty acids (LCFAs),

including palmitate, stearate, oleate, linoleate, linolenate, are minor sources of carbon entering

the TCA cycle compared to glucose and glutamine in prostate cancer cells, despite being

assimilated in the mitochondria as acyl-carnitines. In contrast, cardiolipins were a prominent

LCFAs sink, with some species achieving greater than 50%

C-labelling within 6 hours,

suggesting high cardiolipin turnover using extracellular LCFAs. Knockdown of CPT1a, the

rate-limiting enzyme of LCFA entry into mitochondria, reduced the incorporation of

extracellular linoleate into cardiolipins. These results demonstrate that FAO is not a major input

for the TCA cycle and provide evidence for an underappreciated role for CPT1a in regulating

LCFAs entry into mitochondria for cardiolipin remodelling.

Keywords

Fatty acid oxidation, parallel stable isotope tracing, TCA cycle, cardiolipin, prostate cancer,

metabolomics

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Introduction

Prostate cancer (PCa) display unique metabolic features compared to many other solid

tumours, as it typically does not exhibit the “Warburg effect”. A hallmark of PCa is alterations

in lipid metabolism, which arise from reports that PCa overexpress lipid metabolism genes

1-3

have enhanced fatty acid (FA) oxidation (FAO) rates

4-6

, increased sensitivity to inhibition of

FAO and FA synthesis pathways

7,8

, lipid profile alterations that associate with PCa

progression

, and increased citrate oxidation for lipogenesis

9,10

compared to benign epithelial

cells and tissues. Of these features, it is the broader interpretations of the higher FAO rates that

are poorly supported. In that, it has been long proposed that FAO is the predominant

bioenergetic source in PCa

11,12

, yet there have been no quantifications of FAO rates in relation

to other major mitochondria substrates, such as glucose and glutamine. PCa cells consume non-

trivial amounts of glucose and glutamine

6,13

and this shifts along a spectrum of glucose and

lipid utilisation from benign to castration resistant PCa

14,15

. However, it is unclear whether

FAO is a major carbon source for the TCA cycle, and thereby support the widely held belief

that it is the predominant bioenergetic source, compared to glucose and glutamine in PCa cells.

Cytosolic LCFA-CoAs enter the mitochondria through the carnitine shuttle system

, with the

contemporary view being that mitochondria LCFA-CoAs are shunted into

-oxidation to

produce acetyl-CoA for the TCA cycle, and NADH and FADH

for the electron transport chain

(ETC)

. As such, CPT1 is viewed as the rate-limiting step of FAO, which is supported by

evidence that loss-of-function decreases CO

and ATP production and causes cell death in

cancer cells

5,18-20

. This CPT1a-FAO-ATP-cell death cascade requires FAO to be the

predominant bioenergetic process, and that glucose and glutamine metabolism are unable to

sustain ATP levels in the setting of impaired FAO. Significant doubts about this cascade arose

from Yao

et al

who reported that pharmacological inhibition of FAO with low concentrations

of the CPT1 inhibitor etomoxir did not affect the proliferation rates of various cancer cells.

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Secondly, they showed that CPT1a loss-of-function in BT549 breast cancer cells decreased

mitochondrial lipid levels and altered mitochondrial morphology

. Similar observations have

been reported in T cells

and mouse heart

. Together, these results raise the possibility that

intramitochondrial LCFA-CoAs are not a major TCA cycle source of carbons, and that CPT1

is essential for maintaining intramitochondrial lipid homeostasis and ultimately questions the

sole commitment of mitochondria LCFA-CoAs into beta-oxidation. In this study, we aimed to

determine whether FAO provides a significant carbon source to the TCA cycle in PCa cells and

whether extracellular LCFAs contribute to other significant mitochondrial fates.

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Results

Glucose and glutamine are the major TCA cycle carbon sources in prostate cancer cells.

PCa is characterised by higher rates of FAO compared to prostate epithelial cells

4-6

, however,

it remains to be determined how these rates relate to the use of other sources. We aimed to

quantify, in parallel, the conversion of extracellular glucose, glutamine and LCFAs into TCA

cycle metabolites in a panel of PCa cells spanning the spectrum of disease, from benign PNT1

cells to late-stage PC-3 cells. Consistent with previous reports

4-6

, PCa cells had 2-4- fold greater

FAO rates compared to PNT1 benign cells (Fig 1a). These differences in baseline FAO rates

could not be explained by active mitochondria amount (Fig. 1b, c), nor by CPT1a protein levels

(Fig. 1d, e). Next, we performed parallel [U-

C] isotope tracing experiments to quantify the

relative incorporation of uniformly labelled palmitate (FA 16:0), glucose, and glutamine

carbons into TCA cycle metabolites

Palmitate was chosen as the representative LCFA as it is

the most abundant FA circulating in humans

. We determined that 6 hours was the optimal

endpoint for our tracing experiments as isotopic steady state was reached, and media substrates

were not depleted (>50% of starting levels; Extended Data Fig. 1a). Similar studies using

C-

glucose and -glutamine in cultured cells have shown that 3 to 6 hours of label exposure is

sufficient for isotopic steady state

25,26

. We separately noted that doubling of media volume to

2 mL was necessary for a 12 to 24 hour experiment, in order to maintain media substrates above

50% (Extended Data Fig. 1b). A notable pattern was the consumption of media lactate by cells

that had depleted glucose (Extended Data Fig. 1a), thereby, demonstrating metabolic plasticity

of PCa cells.

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Fig. 1: Palmitate is a minor TCA cycle substrate source compared to glucose and

glutamine.

(A)

Fatty acid [1-

C]-FA 16:0 oxidation rates. N=3-5 per cell line. FAO rates were compared to PNT1

cells.

(B)

Active mitochondria as a fraction of mitochondrial potential to mitochondrial content normalised

to PNT1 cells. N=3-4 per cell line.

(C)

Correlation of active mitochondria content to FAO excluding PNT1 cells. Error bars represent



SD. P and r

values by linear regression.

(D)

Quantification of CPT1a protein expression normalised to PNT1 cells. N=3 per cell line. A

representative immunoblot is shown.

(E)

Correlation of CPT1a protein levels to FAO. Error bars represent



SD. P and r

values by linear

regression.

(F)

Fractional contribution of [U-

C]-glucose to pyruvate (red), [U-

C]-glutamine to intracellular

glutamine (blue), [U-

C]-FA 16:0 to palmitoyl-carnitine (orange) after 6 hours of labelling. [U-

C]-glucose, glutamine, or 16:0 to citrate, oxoglutarate and malate. N=9 per cell line. Level of

natural enrichment of metabolite standards indicated.

100

(G)

Fatty acid [1-

C]-16:0 oxidation rates of MR42D and MR49F treatment resistant cell lines

101

compared to treatment sensitive LNCaP cells. N=3 per cell line.

102

(H)

Fractional contribution of [U-

C]-glucose, glutamine, and FA 16:0 to citrate in MR42D, MR49F

103

and LNCaP cells. N=9 per cell line.

104

(I)

Fractional contribution of [U-

C]-FA 16:0 to citrate in LNCaP, C4-2B and 22RV1 3D spheroid

105

models. N=3 per cell line.

106

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Graphs show mean



SEM unless stated otherwise. *p<0.05 by One-way ANOVA with Dunnett’s

107

multiple comparisons test, unless stated otherwise.

108

We achieved sufficient labelling of the respective intracellular TCA cycle precursor pools after

109

6 hours (pyruvate >65%, glutamine >85%, and palmitoyl-carnitine >60%; Fig. 1f) and

110

confirmed the uptake of extracellular labelled substrates. The fractional contribution of [U-

111

C]-glucose and [U-

C]-glutamine, calculated as the weighted sum by the number of labelled

112

carbons

, combined ranged from 60 to 80% for all measured TCA cycle metabolites (Fig. 1f).

113

In contrast, extracellular [U-

C]-palmitate contributed less than 15% of carbons to TCA cycle

114

metabolites (Fig, 1f and Extended Data Fig. 1c-d). 22Rv1 and AD1 cells had the greatest

115

enrichment of

C-palmitate, which was consistent with their higher FAO rates (Extended Data

116

Fig. 1c). In general, the fractional contribution of palmitate to TCA cycle metabolites weakly

117

correlated with FAO and, somewhat surprisingly, inversely correlated with active mitochondria

118

content (Extended Data Fig. 1d-e). Overall, non-trivial amounts of carbon from extracellular

119

palmitate oxidation entered the TCA cycle, but this influx was dwarfed by the contribution of

120

extracellular glucose and glutamine.

121

The treatment resistant PCa cells MR42D and MR49F

have been reported to exhibit enhanced

122

FA uptake and oxidation

29,30

and so we next assessed the contribution of extracellular glucose,

123

glutamine and palmitate to TCA cycle intermediates in these models. In our hands, FAO rates

124

for both MR42D and MR49F cells were not different to treatment sensitive LNCaP cells (Fig.

125

1g). Likewise, the fractional contribution of palmitate to citrate in MR42D and MR49F cells

126

was similar to LNCaP cells (4-8.5%), and consistent with our other results, citrate was largely

127

derived from glucose (35%) and glutamine (20%; Fig. 1h and Extended Data Fig. 1g). We

128

repeated our parallel tracing approach using 3D PCa spheroids as they exhibit a more oxidative

129

phenotype compared to 2D cell cultures

31-33

. Again, we observed minimal contribution of

130

palmitate into TCA cycle metabolites in LNCaP, C4-2B and 22RV1 3D spheroids (<6%; Fig.

131

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

1i and Extended Data Fig. 1j). Importantly, we confirmed that media glucose, glutamine, and

132

palmitate were sufficiently plentiful prior to harvesting cells at 6 hours (Extended Data Fig. 1a,

133

h), and that substantial

C labelling of precursor pools were achieved in both treatment

134

resistant cells and 3D spheroids (Extended Data Fig. 1f, i). Altogether, we show that

135

extracellular glucose and glutamine are the major TCA cycle substrates compared to exogenous

136

palmitate in a range of prostate cells, regardless of disease stage pathophysiology, treatment

137

responsiveness, or multicellular microenvironment.

138

Palmitate oxidation increases in the absence of glucose and glutamine but is insufficient

139

to maintain TCA cycle function.

140

One concern we had was that the low enrichment values (Fig. 1) were, in part, due to a technical

141

or biological issue. As such, we set out to meaningfully increase the enrichment of extracellular

142

palmitate into TCA cycle intermediates. Others have shown that limiting glucose availability

143

increased FAO rates in MCF7 and T47D breast cancer cells

and so we next determined

144

whether FAO in PCa cells is increased by withdrawing glucose from the media, and hence lead

145

to greater palmitate incorporation into the TCA cycle. Compared to basal FAO rates measured

146

in the presence of 5 mM glucose, FAO in glucose-free conditions was 1- to 5- fold greater

147

across the panel of prostate cells (Fig. 2a). We also measured FAO in a range of glucose

148

concentrations (0 mM to 5 mM) and observed a dose-dependent suppression of FAO (Extended

149

Data Fig. 2a). These data demonstrate that glucose levels, and glycolytic activity, impacts FAO

150

rates.

151

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

152

Fig. 2: Palmitate alone is insufficient at maintaining TCA cycle activity.

153

(A)

[1-

C]-FA 16:0 oxidation rates in the presence (blue) or absence (red) of glucose. N=3-5 per cell line.

154

Multiple unpaired t-tests.

155

(B)

Fractional contribution of [U-

C]-FA 16:0 to citrate with or without glucose or glucose and glutamine. N=9

156

per cell line.

157

(C)

Schematic tracing conversion of [U-

C]-FA 16:0 into citrate m+2 and m+4 isotopologues.

158

(D)

Enrichment fraction (%) of [U-

C]-FA 16:0 to m+2 and m+4 citrate isotopologues in +glucose/+glutamine,

159

- glucose/+glutamine, and -glucose/-glutamine media conditions. N=6 per cell line.

160

(E)

Fractional contribution of [U-

C]-FA 16:0 to glutamate and aspartate +glucose/+glutamine, -

161

glucose/+glutamine, or -glucose/-glutamine. N=5-6 per cell line.

162

(F)

Relative abundance (normalised to +glucose/+glutamine) of [U-

C]-FA 16:0 to citrate. N=4-9 per cell line.

163

Graphs show mean



SEM. P<0.05 by One-way ANOVA with Dunnett’s multiple comparisons test, unless stated

164

otherwise. *compared to +glucose/+glutamine, ^compared to -glucose/+glutamine.

165

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

We next quantified the extent that extracellular palmitate provides carbons to TCA cycle

166

metabolites in the absence of glucose or glucose and glutamine. There was no significant

167

impact on cellular metabolic activity (MTT) or cell viability in glucose-free and glucose-/

168

glutamine-free media conditions at 6 hours (Extended Data Fig. 2b, c). Consistent with the

169

increase in FAO rates (Fig. 2a), the fractional contribution of palmitate to citrate was increased

170

in glucose-free and glucose-/ glutamine-free conditions, but it remained remarkably minimal

171

(~4% to ~6-8%; Fig. 2b). Notably, PNT1 benign and AD1 PCa cells had increased palmitate

172

labelling of citrate only in the absence of both glutamine and glucose, but not glucose alone

173

(Fig. 2b), suggesting that these cells mobilised glutamine first instead of palmitate to

174

compensate for glucose deprivation. In the absence of glucose and glutamine, palmitate

175

produced markedly more m+2 and m+4 citrate isotopologues, which represent palmitate

176

utilisation in the first and second turns of the TCA cycle, respectively (Fig. 2c, d). This suggests

177

that palmitate carbons were retained in the TCA cycle for longer under glucose and glutamine

178

deprivation. We also observed other dynamic changes in the usage of palmitate-derived carbons

179

during glucose and glutamine deprivation, especially into cataplerosis-derived metabolites

180

glutamate and aspartate (Fig. 2e), which are produced from oxoglutarate and oxaloacetate

181

respectively (Fig. 2c). Firstly, we observed essentially no incorporation of

C-palmitate into

182

glutamate and aspartate in cells cultured in media containing physiological levels of glucose

183

and glutamine (Fig. 2e). In glucose-free, and most strikingly in glucose-/glutamine-free

184

conditions, we saw significant

C-palmitate labelling into glutamate and aspartate (Fig. 2c, e).

185

As aspartate synthesis is essential for cell proliferation

and glutamate for glutathione

186

synthesis and redox homeostasis

, our results suggest that prostate cells, in the absence of

187

glucose or both glucose and glutamine, upregulate cataplerotic reactions in an attempt to

188

maintain cellular function and viability. Together, these data demonstrate the dynamic

189

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

adaptability of prostate cell palmitate metabolism to supply carbons to the TCA cycle when

190

deprived of glucose only or glucose and glutamine.

191

FAO has been proposed to be the dominant bioenergetic source for PCa cells

11,12

. As such, we

192

reasoned that FAO would be capable of sustaining TCA cycle activity in glucose or glucose-

193

glutamine deprived conditions. In striking contrast to

C-palmitate enrichment patterns, citrate,

194

malate and oxoglutarate abundance were reduced in cells cultured in glucose-free and glucose-

195

/ glutamine-free settings for 6 hours compared to glucose and glutamine replete settings (Fig.

196

2f and Extended Data Fig. 2d). Specifically, citrate levels were decreased by 43-70% in most

197

cells, except for C4-2B and 22Rv1 cells, which were not impacted by glucose- and

198

glucose/glutamine-free conditions (Fig. 2f). We did observe striking reductions in malate and

199

oxoglutarate levels (between 53-95%) in all cells in response to glucose and glutamine

200

depletion (Extended Data Fig. 2d). Importantly, these results preceded the cell death which

201

occurred between 6 and 24 hours of culturing (Extended Data Fig. 2b, c). Combined, these data

202

show that even with the increased FAO rates to incorporate more palmitate carbon into the TCA

203

cycle as well as into aspartate and glutamate, these changes were insufficient to maintain

204

metabolite abundance and so TCA cycle activity that were followed by cell death.

205

Extracellular LCFAs are minor TCA cycle substrates.

206

Since palmitate is a minor carbon source for the TCA cycle (Fig. 1 and 2), we questioned

207

whether this was a palmitate-specific phenotype.

To address this, we repeated our experiments

208

with a panel of non-essential (16:0, 18:0, 18:1) and essential (18:2, 18:3) LCFAs of varying

209

saturation and carbon chain lengths. There was a similar trend of minimal contribution (2-18%)

210

to TCA cycle metabolites observed in our panel of prostate cells, despite the diversity of LCFAs

211

(Fig. 3a). 22Rv1 and AD1 cells had the greatest palmitate labelling and FAO rates amongst

212

prostate cells (Fig. 1a and Extended Data Fig. 1c), and consistently exhibited greater utilisation

213

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

of LCFAs by the TCA cycle compared to an unlabelled control (Fig. 3a). In general, there was

214

no clear substrate preference between palmitate and other LCFAs to feed carbons to the TCA

215

cycle (Fig. 1c), despite substantial labelling of respective acyl-carnitine species (0.8-1, ratio

216

/m0+m

; Fig. 3b). Collectively, we show that in the presence of physiological glucose and

217

glutamine, diverse extracellular LCFAs were not major carbon sources for the TCA cycle,

218

despite substantial acyl-carnitine precursor labelling.

219

220

Fig. 3: LCFAs of varying saturations are minor contributing substrate sources to the TCA

221

cycle.

222

(A)

Fractional contribution of [U-

C]-FA 16:0, 18:0, 18:1, 18:2, 18:3 to citrate, oxoglutarate, malate. N=9 per

223

cell line.

224

(B)

Fractional contribution of acyl-carnitines associated with the [U-

C]-FA substrates. N=9 per cell line.

225

Graphs show mean



SEM unless stated otherwise. P<0.05 by Two-way ANOVA with Tukey’s multiple

226

comparisons test, *compared to unlabelled control (dotted line), #compared to [U-

C]-FA 16:0.

227

One possible reason for the disconnect between the substantial conversion of LCFA to acyl-

228

carnitines and the lack of conversion to TCA cycle intermediates could be incomplete FAO,

229

which generates medium chain acyl-carnitine species instead of fully oxidising LCFAs to

230

acetyl-CoA

37,38

. In agreement with this hypothesis, FA 16:0, 18:0, and 18:1 greatly enriched

231

shortened acyl-carnitine species (Extended Data Fig. 3). Specifically, FA 16:0 significantly

232

enriched shortened CAR 14:0 (0.04-0.7 fraction) to CAR 10:0 (0-0.45 fraction), FA 18:0

233

enriched CAR 16:0 (0.01-0.38 fraction) to CAR 12:0 (0.04-0.4 fraction), and FA 18:1 enriched

234

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

CAR 16:1 (0.03-0.79 fraction) to CAR 12:1 (0-0.1 fraction). The greatest amounts of

235

exogenous FA enrichment into shortened acyl-carnitines were observed in PNT1, 22Rv1, AD1

236

and PC-3 cells (Extended Data Fig. 3), which were cells that also had the greatest incorporation

237

of LCFAs into precursor acyl-carnitines (Fig. 3b). These data show that labelled extracellular

238

LCFAs are not completely converted to acetyl-CoA, and thereby provides new insights into the

239

relationships between acyl-carnitine synthesis, LCFA entry into mitochondria, and FAO.

240

Extracellular LCFAs supports the rapid turnover of cardiolipin pools.

241

Our observations showed that extracellular LCFAs were not completely oxidised to acetyl-CoA

242

and minimally contributed carbons to the TCA cycle, despite being substantially present as

243

acyl-carnitines (Fig. 3 and Extended Data Fig. 3). As such, we asked whether there are other

244

pathways that could consume extracellular LCFAs after entering the mitochondria via CPT1a.

245

This question arises from the work of Yao and colleagues

as they were the first to report a

246

reduction of mitochondrial complex lipids in BT549 breast cancer cells with CPT1a knocked

247

down. Of their observed lipidome results, cardiolipins (CLs) piqued our interest as they are

248

exclusively located within the inner mitochondrial membrane (IMM) and account for 10-20%

249

of mitochondria phospholipid content

. CLs are composed of four fatty acyl chains of different

250

saturation and chain lengths

and provide essential structural functions to mitochondria

251

membrane and cristae

, and ETC complexes I-IV for ATP production

42-44

. As CL remodelling

252

functions to mature nascent CLs by the reacylation from donated acyl-CoAs

, it was

253

conceivable that this process could be a significant consumer of mitochondria LCFA-CoAs.

254

Therefore, we next investigated the contribution of extracellular LCFAs to CL synthesis and

255

remodelling (Fig. 4a). First, we examined the enrichments of CL pools from extracellular [U-

256

C]-16:0, 18:0, 18:1, and 18:2 LCFAs. Surprisingly, cells cultured in media containing labelled

257

LCFAs for 6 hours had >50% enrichment of many CL species, indicating substantial LCFA

258

turnover of CL pools (Fig. 4b), in stark contrast to the marginal arrival of LCFA label into the

259

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

TCA cycle (Fig. 1-3). CL has been shown to be a very slow turnover lipid class with CL half-

260

life reported to be approximately two days in isolated rat cardiomyocytes

and cultured rat

261

H9c2 cardiomyoblast cells

. These turnover rates are much slower than other

262

glycerophospholipids, including PC

46-49

. Here, we saw time-dependent enrichment of [U-

C]-

263

FA 16:0 into PC 32:1 and 34:1 of C4-2B and PC-3 cells (Extended Data Fig. 4a) and that this

264

enrichment was greater than related FA 16:0 containing CL species (CL 68:2 and CL 70:2) at

265

6 hours (Extended Data Fig. 4b). As such, C4-2B and PC-3 cells incorporate extracellular FAs

266

into PCs at a higher proportion relative to CLs.

267

FAs, following activation into FA-CoAs, are substrates for elongation and desaturation

268

reactions

. In our experiments, we saw that exogenous FAs 16:0, 18:0, and 18:2 were directly

269

incorporated into the acyl-chains of CLs (i.e. unmodified via elongation or desaturation;

270

Extended Data Fig. 4c). For instance, CL 68:2 is composed of 16:0 and 18:1 acyl-chains and

271

was predominantly labelled by FA 16:0, while CL 70:3 is also composed of 16:0 and 18:1 acyl-

272

chains but was mainly labelled by FA 18:1 (Fig. 4c). Similarly, CL 72:5 is composed of 18:1

273

and 18:2 chains and was almost entirely labelled by FA 18:2 within 6 hours (Fig. 4c). We

274

observed some differences between PCa cells and benign PNT1 cells; however, no obvious

275

pattern could be discerned (Fig. 4c). Finally, to compare the relative incorporation of the four

276

FAs into CLs, we calculated an overall percentage of labelled fatty acyl chains among the 15

277

quantified CL species. In general, a greater proportion of FA 18:2 was labelled compared to

278

FAs 16:0, 18:0, and 18:1 in our panel of prostate cells at 6 hours exposure to the respective

279

substrates (Extended Data Fig. 4d). This suggests 18:2-containing CLs have the greatest

280

turnover, which was also reported for cardiomyocytes and HeLa cells

51,52

. This may associate

281

with the fact a majority of CL acyl chains is 18:2. The exceptions were LNCaP and C4-2B

282

cells, which showed greater FA 16:0 13C enrichment among the CLs (32% and 14.2%,

283

respectively) compared to FA 18:2 (9.5% and 8.2%, respectively; Extended Data Fig. 4d).

284

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

287

Fig. 4: Extracellular LCFAs are directly and rapidly incorporated into cardiolipin pools.

288

(A)

Schematic depicting fates of FAs entering the mitochondria including FAO and CL pathways.

289

(B)

Fractional enrichment of [U-

C]-FA 16:0, 18:0, 18:1, 18:2 (150



M) tracing to CL species for 6 hours. N=9

290

per cell line.

291

(C)

Fractional enrichment of [U-

C]-FA 16:0, 18:0, 18:1, 18:2 to CL 68:2, CL 70:3, CL 72:5 species. N=9 per

292

cell line.

293

Graphs show mean



SEM. P<0.05 by One-way ANOVA with Tukey’s multiple comparisons test, *PCa cells vs

294

PNT1 cells for each FA.

295

Reduced CPT1a protein levels decreased FAO and incorporation of LCFAs into CL.

296

The high incorporation of extracellular LCFAs into CLs in both PCa cells and PNT1 cells was

297

in stark contrast to the magnitude of enrichment into TCA cycle metabolites (Fig. 3 and 4).

298

LCFAs are incorporated into CLs via multiple starting points, not exclusively competing with

299

FAO for mitochondria FAs that are supplied via CPT1a. While these intramitochondrial acyl-

300

CoAs are substrates for CL remodelling acyltransferases MLCAT1

and ALCAT1

, CLs are

301

also assembled by the

de novo

pathway producing nascent CLs from phosphatidylglycerol

302

(PG)

, and then remodelled via acyl-chain exchange with PC and phosphatidylethanolamine

303

by the actions of Tafazzin

. To tease out the association between CPT1a and CL remodelling,

304

we quantified the incorporation of extracellular FA 18:2 in cells with CPT1a knocked down,

305

since 18:2 has been reported to be predominantly incorporated into CLs via remodelling

306

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

pathways

45,51

(Fig. 5a). We included FA 16:0 in our experiments as it can be incorporated via

307

both

de novo

synthesis and remodelling

56,45

. Further, we used C4-2B cells for this experiment

308

as they had the greatest level of CPT1a protein in our panel of prostate cells (Fig. 1d). As

309

expected, CPT1a knockdown (CPT1a

; reduced by 89%) in C4-2B cells (Fig. 5b) decreased

310

palmitate oxidation (Fig. 5c) and the abundance of

C-labelled 18:2 and 16:0 acyl-carnitines

311

compared to scrambled non-targeting control cells (Fig. 5d). CPT1a

decreased FA 16:0

312

enrichment of citrate from 3.5% to 3% (delta 0.5%) but did not impact the contribution from

313

FA 18:2 (Fig. 5e).

314

In contrast, we found CPT1a

decreased FA 18:2 enrichment of CLs. The overall proportion

315

of labelled FA 18:2 among the CL species was decreased from 19% to 15%, but this reduction

316

was not reciprocated by FA 16:0 (Fig. 5f, Extended Data Fig. 5a). Namely, we saw all

317

individual CL species had less FA 18:2 enrichment, but the effects were mixed for FA 16:0

318

enrichment (Fig. 5g). The latter suggests FA 16:0 is incorporated into CLs largely via the

319

novo

pathway, which is independent of CPT1a and mitochondrial acyl-CoAs

. As expected,

320

there was no difference in

C-FA 18:2 and

C-FA 16:0 PG enrichment between CPT1a

and

321

scrambled control cells (Fig. 5h, Extended Data Fig. 5b, c), which suggests that the reduction

322

in CPT1 activity did not divert LCFAs towards the endoplasmic reticulum to increase PG and

323

CL synthesis. Altogether, we showed that that CPT1a regulates the incorporation of FA 18:2

324

into CL, thereby demonstrating that CL remodelling is a pathway downstream of CPT1a

325

alongside FAO and the TCA cycle (Fig. 5i).

326

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

327

Fig. 5: CPT1a knockdown reduced FA 18:2 derived CL remodelling.

328

(A)

Schematic depicting expected FA 18:2 to CL remodelling downstream of CPT1a and FA 16:0 to CL synthesis

329

pathways derived from the endoplasmic reticulum.

330

(B)

CPT1a knockdown by reverse siRNA transfection in C4-2B cells for 72 hours quantified by immunoblot.

331

Representative immunoblots shown. N=3 per group.

332

(C)

[1-

C]-FA 16:0 oxidation rates in non-targeting scrambled (scr) and siCPT1a C4-2B cells. N=3 per group.

333

(D)

Intensity of [U-

C]-FA 18:2 or 16:0 to carnitine species following siCPT1a treatment. N=9 per group.

334

Statistically significant differences between labelled (m16 orange or m18 purple) intensities (*) or unlabelled

335

(m0 grey) intensities (#).

336

(E)

Fractional contribution of unlabelled or [U-

C]-FA 18:2 or 16:0 to citrate in scr or siCPT1a C4-2B cells. N=9

337

per group.

338

(F)

Percentage of 18:2 or 16:0 acyl chain contained in CLs labelled by exogenous [U-

C]-FA. N=9 per group.

339

(G)

Fractional contribution of [U-

C]-FA 18:2 or 16:0 to PG and CL species. Annotated n values on the x-axis

340

indicate the number of 18:2 or 16:0 containing acyl-chains. N=9 per group. Unpaired t-tests.

341

(H)

Delta differences between fractional contribution of [U-

C]-FA 18:2 or 16:0 to PG (N=6) and CL species

342

(N=17). Statistically significant differences within [U-

C]-FA 18:2 or 16:0 groups to 0 difference determined

343

by Nonparametric Wilcoxon signed rank test. Error bars represent



min to max values.

344

(I)

Schematic depicting major carbon contributions from glucose and glutamine into the TCA cycle in

345

comparison to the substantial contributions of LCFA-derived acyl-CoAs for CLs either by CL remodelling

346

downstream of CPT1a or CL synthesis.

347

Graphs show mean



SEM unless stated otherwise. ns not significant, *p<0.05 between siCPT1a and scr control

348

groups by student’s t-test unless stated otherwise.

349

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Discussion

350

PCa do not exhibit classical Warburg metabolism but displays a more lipid dependent

351

phenotype

8,57

. One aspect of this lipid phenotype that is widely accepted is the assumption that

352

FAO is the predominant bioenergetic pathway in PCa, which likely arises from reports of

353

enhanced FAO rates in PCa cells compared to benign cells

4-6

, low glycolytic activity at early

354

disease stages

14,15

, and increased sensitivity to FAO targeting

. However, this belief that FAO

355

is a dominant bioenergetic pathway in PCa, first proposed by Liu

in 2006, failed to consider

356

the contribution of other major respiration substrates. CPT1a has been implicated as the rate-

357

limiting step of FAO because its inhibition results in decreased ATP production and cancer cell

358

death

5,18-20

, and assumes that LCFAs are entirely diverted to FAO and the TCA cycle. Here, we

359

generated evidence disputing this dogmatic view by showing extracellular LCFAs are minor

360

carbon sources to the TCA cycle compared to glucose and glutamine in PCa cells, despite their

361

increased FAO rates compared to benign PNT1 cells. We also showed that LCFAs are not

362

completely oxidised to acetyl-CoA and produce shorter acyl-carnitine species. Most strikingly,

363

there was substantial incorporation of LCFAs into CLs within 6 hours, and that CPT1a

364

regulated the assimilation of FA 18:2 into CLs. Together, our data show a substantial role for

365

CPT1a to regulate CL turnover in PCa cells and propose that observations of a CPT1a-

366

dependent reduction of FAO and mitochondrial function can be explained, in part, by perturbed

367

CL maturation and reduced oxidative phosphorylation efficiency.

368

We and others have shown that FAO in PCa cells is faster than benign prostate epithelial cells

369

, and that inhibiting CPT1 impacts cell viability

18-20

. These observations, combined with

370

similar studies and the hypothesis article by Liu

, has resulted in the widely held view that

371

FAO is a dominant bioenergetic pathway in PCa. Here, we present comprehensive evidence

372

that challenges this. In a series of experiments, we show that extracellular LCFAs are a minor

373

contributor of carbons to TCA cycle metabolites, compared to glucose and glutamine, which

374

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

may, in part, be explained by incomplete oxidation. Further, FAO could not meet the

375

anaplerotic demand to maintain TCA cycle activity in the absence of glucose alone or glucose

376

and glutamine, leading to cell death. We also determined that extracellular FAs essentially do

377

not contribute to glutamate and aspartate synthesis in the presence of extracellular glucose and

378

glutamine, but do so in their absence. These results were seen across the spectrum of PCa

379

progression, from benign to late-stage, 2D and 3D models. Importantly, the magnitude of

380

exogenous FA enrichment into citrate was comparable to similar studies, including ~15% in

381

vivo

cultured malignant Patient-Derived Xenografts (500

M [U-

C]palmitate, 4 hours)

and

382

BT549 cells (100

M [U-

C]palmitate, 24 hours)

, <15% in HCC1954 and SKBR3 breast

383

cancer cells (100

M [U-

C]palmitate

, 24

hours)

, and ~10% in MCF-7 and MDA-MB-231

384

breast cancer cells cultured in glucose-free media (100

M [U-

C]palmitate, 4 hours)

385

Therefore, these observations raise questions regarding the many studies demonstrating that

386

CPT1 loss-of-function leads to cell death

19,60

. One possible explanation is that many studies

387

used etomoxir, an irreversible inhibitor of CPT1, at concentrations that have been reported to

388

have significant off-target effects

, and often failed to identify the minimum concentration of

389

etomoxir to maximally suppress FAO. Another reason could be due to the focus on

390

demonstrating a dose-dependent reduction in cell viability and thereby determining IC

in the

391

absence of measures of FAO or CPT1 activity. Finally, the other technical consideration when

392

comparing our results to others is the use of the Seahorse XF Analyser Palmitate Oxidation

393

Kit, which measures oxygen consumption in saturating conditions of palmitate and carnitine

394

but in media that is free of glucose, pyruvate, and glutamine. Further, the recommendation to

395

culture cells in substrate limited conditions overnight will impact cellular metabolism. We

396

observed striking increases in FAO rates, ranging from a doubling to a 5-fold increase, as well

397

as substantial changes in intracellular partitioning of LCFA carbons in glucose-free conditions.

398

As such, any observed changes in FAO performed in conditions free of major TCA cycle

399

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

substrates must be interpreted with caution and likely explains the differences between our

400

observations and other published studies. Unlike other studies, we showed that CPT1a

401

substantially decreased acyl-carnitine levels and the synthesis from extracellular LCFAs, and

402

had a minor impact the contribution of LCFAs to TCA cycle metabolites in media containing

403

5 mM glucose and 1 mM glutamine. Overall, we conclude that extracellular LCFAs are not

404

completely oxidised, and are a minor source of carbons for TCA cycle metabolites in a panel

405

of PCa cells and benign PNT1 cells, therefore that FAO is not a major bioenergetic pathway.

406

In general, much of the literature discussing CPT1 function and mitochondrial FA metabolism

407

state that CPT1 is rate-limiting for FAO and that FAO is the sole destination for

408

intramitochondrial LCFA-CoAs following facilitated transport by the CPT1 –

carnitine-

409

acylcarnitine translocase – CPT2 system. This perception fails to consider that there are other

410

destinations, especially given the evidence that CPT1 inhibition reduces mitochondrial lipid

411

levels in cancer cells

21,61

and alters mitochondrial morphology in BT549 cells

, T cells

, and

412

rat heart muscle

. In our studies, we focused on the mitochondria exclusive CL as it is essential

413

for mitochondria function and bioenergetics

. To our surprise, we saw rapid incorporation of

414

extracellular LCFAs into CLs in PCa cells (>50% after 6 hours), which was much faster than

415

what has been reported in isolate rat cardiomyocytes (2 days)

and cultured rat H9c2

416

cardiomyoblast cells (~54 hours)

. That said, we did determine that FA 18:2 was

417

predominantly incorporated into the acyl-chains of CLs compared to other LCFAs, which is

418

similar to previous reports

51,52

and that the rate of incorporation of LCFAs into other lipid

419

classes was faster than CL

. These results were especially important as cancer cells that

420

substantially incorporate exogenous 18:2 into CL acyl-chains have enhanced stimulation of

421

ETC complex I activity

and reduced susceptibility to apoptosis

. These results showing

422

significant enrichment of CLs with extracellular LCFAs provided a platform to determine

423

whether CPT1 regulates LCFA incorporation into CLs, especially given that CPT1 loss-of-

424

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

function reduces CL and mitochondrial lipids

21,61

. Consistent with our hypothesis, depletion of

425

CPT1a reduced

C-labelled FA 18:2 incorporation into CL in C4-2B PCa cells, and so suggest

426

that CPT1 influences CL remodelling which may influence mitochondrial respiration and

427

apoptosis

51,63

. Taken together, we propose that CLs, alongside FAO, are a prominent fate for

428

extracellular LCFAs entering the mitochondria via CPT1a.

429

A limitation of our study is that we only quantified the contribution of extracellular substrates

430

to mitochondrial pathways, which accounted for less than 65% of carbons to citrate. We

431

focused on the relationship between FAO and TCA cycle metabolites, in part, to complement

432

our previous reports that extracellular LCFAs as the major precursors for PCa lipids

. It is likely

433

that the unaccounted for sources of carbons for TCA cycle metabolites include lactate

434

branched chain amino acids

, as well as intracellular sources of FAs and amino acids.

435

Quantifying the totality could make new insights into carbon source for the TCA cycle and

436

how differs in other cancer types, disease stage and impacted by the tumour microenvironment.

437

We also provide evidence for incomplete FAO in PCa cells; however, further investigation into

438

the roles of shortened acyl-carnitines may provide insights into other fates of LCFAs, alongside

439

our new insights into CLs and TCA cycle biology. Finally, we showed for the first time that

440

CPT1a impacts CL remodelling in PCa cells. Future experiments are required to link CPT1a

441

function, CL homeostasis and ETC complex structure

and other aspects of mitochondrial

442

function, beyond FAO. This is especially true given that altered CL function has similar effects

443

on mitochondrial morphology, function, and cell viability

as CPT1a loss-of-function

444

experiments

5,18-22

445

This study provides evidence that extracellular LCFAs are rapidly incorporated into CLs but

446

are not the predominant carbon source to the TCA cycle in prostate cells compared to glucose

447

and glutamine. Further, we conclude that CPT1a influences CL remodelling alongside FAO

448

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Methods

452

Key Resources Table

453

REAGENT or RESOURCE

SOURCE

IDENTIFIER

Antibodies

Mouse monoclonal anti-CPT1A

Abcam

Cat#Ab128568; RRID:
AB_11141632

Rabbit monoclonal anti-GAPDH

Cell Signalling
Technology

Cat#2118; RRID:
AB_561053

Chemicals, peptides, and recombinant proteins

Enzalutamide MDV3100

Selleckchem

Cat#S1250

Lipofectamine RNAiMAX transfection reagent

Thermo Fisher Scientific

Cat#13778075

[1-

C] Palmitic acid

Perkin Elmer

Cat#NEC075H250UC

Palmitic acid

Sigma-Aldrich

Cat#P0500

D-Glucose Sigma-Aldrich

Cat#273053

L-Glutamine Gibco

Cat#25030081

Fatty acid free BSA

Bovogen

Cat#BSAS0.10

L-carnitine hydrochloride

Sigma-Aldrich

Cat#C0283

Mitotracker Green

Thermo Fisher Scientific

Cat#M7514

Mitotracker Red

Thermo Fisher Scientific

Cat#M7512

FCCP Sigma-Aldrich

Cat#SML2959

DTT Sigma-Aldrich

Cat#D0632

Protease and Phosphatase Inhibitor Cocktail, EDTA free

Astral Scientific

Cat#T-2496

Thiazolyl blue tetrazolium bromide MTT reagent

Sigma-Aldrich

Cat#M2128

Linoleic acid

Sigma Aldrich

Cat#L1012

D-Glucose (U-

, 99%)

Cambridge Isotope
Laboratories, Inc.

Cat#CLM-1396

L-Glutamine (U-

, 99%)

Cambridge Isotope
Laboratories, Inc.

Cat#CLM-1822-H

Palmitic acid (U-

, 98%)

Cambridge Isotope
Laboratories, Inc.

Cat#CLM-409

Stearic acid (U-

, 98%)

Cambridge Isotope
Laboratories, Inc.

Cat#CLM-6990

Linoleic acid (U-

, 98%)

Cambridge Isotope
Laboratories, Inc.

Cat#CLM-10487

Linolenic acid (U-

, 98%)

Cambridge Isotope
Laboratories, Inc.

Cat#8386

Critical commercial assays

Pierce BCA Protein assay kit

Thermo Fisher Scientific

Cat#23225

Immobilon Crescendo Western HRP substrate Merck

Millipore Cat#WBLUR0500

CellCarrier Spheroid ULA 96-well microplates

Perkin Elmer

Cat#6055330

Experimental models: Cell lines

PNT1 ECACC

95012614

LNCaP ATCC

CRL-1740

C4-2B ATCC

CRL-3315

22Rv1 ATCC

CRL-2505

CWR-R1-AD1

A/Prof. Luke Selth,
Flinders University

Nyquist 2013

, Li 2013

PC-3 ATCC

CRL-1435

MR42D

Prof. Lisa Butler,
SAHMRI

Bishop 2017

MR49F

Prof. Lisa Butler,
SAHMRI

Bishop 2017

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Oligonucleotides

ON-TARGETplus Human CPT1A siRNA SMARTpool

Horizon Discovery

Cat#L-009749-00-0005

ON-TARGETplus Non-targeting Control pool

Horizon Discovery

Cat#

D-001810-10-05

Software and algorithms

GraphPad Prism V9

GraphPad Software

https://www.graphpad.co
m/

Image Lab 6.1

Bio-Rad

https://www.bio-
rad.com/en-
au/product/image-lab-
software

FlowJo 2

FlowJo LLC

https://www.flowjo.com/

MATLAB Mathworks

https://au.mathworks.com
/products/matlab.html

MSConvert N/A

Chambers

2012

MSDial Riken

http://prime.psc.riken.jp/c
ompms/msdial/main.html

Biorender Biorender

https://biorender.com/

454

Method Details

455

Cell lines and culture conditions

456

Human benign prostate epithelial cell line PNT1 and human prostate carcinoma cells lines

457

LNCaP (androgen receptor (AR) +, hormone sensitive), C4-2B, 22Rv1, AD1 (AR+, androgen

458

independent), and PC-3 (AR-, hormone resistant) cells were cultured in RPMI1640 medium

459

(Life Technologies), supplemented with 10% FBS (Cytiva Hyclone), and 1%

460

penicillin/streptomycin (Gibco). MR42D and MR49F (hormone resistant, treatment resistant)

461

cells were cultured in RPMI 1640 medium (Life Technologies), supplemented with 10% FBS,

462

1% penicillin/streptomycin, and 10



M enzalutamide. All cells were incubated at 37



C and 5%

463

464

Spheroid culturing

465

Cells were seeded in 96-well Spheroid microplates, at a density of 5 000 cells/well (LNCaP),

466

8 000 cells/well (22Rv1), and 2 000 cells/well (C4-2B), in RPMI medium in 100



L per well.

467

An additional 100



L of medium was added to cells at 24h and left un-agitated for a further

468

24h until spheroids formed.

469

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Cell Transfection

470

C4-2B cells were transfected for 72 hours using RNAiMAX transfection reagent and 25 pmol

471

pooled CPT1a siRNA or non-targeting scrambled control, according to the manufacturer’s

472

instructions. After 72 hours, cells were used for subsequent radiolabelling, western

473

immunoblotting, and

C-labelling experiments.

474

C-labelling

475

Cells were washed in warm PBS and incubated in 600



L media containing 0.1 mmol/L cold

476

palmitate, [1-

C]-palmitate (0.1



Ci/mL) conjugated to 2% (wt/vol) FA-free BSA and 1

477

mmol/L -carnitine in RPMI glucose-free medium (Gibco) for 4 hours. Fatty acid oxidation

478

rates were determined from

production as previously described

and supplemented with

479

0 mM or 5 mM glucose to measure metabolic flexibility. Medium for dose dependent glucose

480

deprivation results were supplemented with 0 mM, 1 mM, 3 mM, 5 mM glucose. Protein

481

content was measured by BCA protein assay and absorbances read at 540 nm using a TECAN

482

infinite M1000 pro plate reader.

483

Mitotracker

484

Cells were seeded in 12-well plates at a density of 2x10

cells/well in RPMI medium. After 24

485

hours, cells were washed in warm PBS and obtained by trypsinisation and collected in serum-

486

free RPMI medium. The cell suspension was centrifuged at 1500 rpm for 5 min in fluorescence-

487

activated cell sorting (FACS) tubes. Cells were subsequently stained with 200 nM Mitotracker

488

Red and Mitotracker Green at 37



C for 30mins. Additionally, unstained, red only, green only,

489

and 2.5



M FCCP treated cells were included as experimental controls. Cells were centrifuged

490

at 1500 rpm for 5min and fixed with 4% paraformaldehyde at 37



C for 10min. Cells were

491

washed with warm PBS and centrifuged at 1500 rpm for 5min and resuspended in 150



L PBS

492

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

before the samples were read on a BD LSR Fortessa flow cytometer. Data analysis was

493

performed using FlowJo software.

494

Immunoblotting

495

Cells were seeded in 6-well plates (5x10

). After 24 hours, cells were lysed in RIPA buffer

496

containing 1% DTT and 0.1% protease and phosphatase inhibitor cocktail. Protein content was

497

measured by BCA protein assay and absorbances read at 540 nm using a TECAN infinite

498

M1000 pro plate reader. Cell lysates were subjected to SDS-PAGE, transferred to

499

polyvinylidene difluoride (PVDF) membranes (Merck Millipore). Ponceau staining or

500

GAPDH was used as a loading control, and then membranes were immunoblotted for CPT1a.

501

Chemiluminescence was performed using Western HRP Substrate and imaged using the Bio-

502

Rad ChemiDoc MP Imaging System (Bio-Rad Laboratories) and Image Lab software.

503

Cell viability

504

Cells were seeded in 96-well plates, in 6 well replicates (3x10

cells/well). After 24 hours the

505

media was removed and replaced with glucose- and glutamine- free media (Gibco)

506

supplemented with 5 mM glucose, 1 mM glutamine and/or 150



M palmitate, conjugated to

507

2% (wt/vol) FA-free BSA. For glucose-deprived conditions DMEM no glucose, no glutamine

508

media was supplemented with 1mM glutamine and 150



M palmitate; for glucose and

509

glutamine-deprived conditions media was supplemented with 150



M palmitate only. Media

510

(5 mM glucose, 1 mM glutamine, 150



M palmitate) supplemented with 10% FBS was used

511

as a control. At 0, 6, 12 and 24 hours, 50



L of MTT reagent was added to wells. MTT reactions

512

were incubated at 37°C and 5% CO

for 4 hours, and then stopped by replacing media with

513

100



L DMSO (Sigma–Aldrich) per well. Absorbances were read at 540 nm using a TECAN

514

infinite M1000 pro plate reader. For live cell counts, cells were seeded in 6-well plates at 5x10

515

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

cells/well. After 24 hours, media was replaced as described for the MTT assay, and cells

516

counted by Trypan blue (Gibco) exclusion at 6 and 24 hours.

517

U-

C stable isotope tracing

518

Cells were seeded in 6-well plates (5x10

) for 2D culture, or as above for 3D spheroid models.

519

After 24 hours, wells were washed with warm PBS and media replaced with 600



L of DMEM

520

no glucose (0 mM), no glutamine (0 mM) media (Gibco), supplemented with 5 mM glucose, 1

521

mM glutamine and 150



M FAs replaced with their respective [U-

C] forms (Palmitate (16:0),

522

Oleate (18:0), Stearate (18:1), Linoleate (18:2), or Linolenate (18:3)). For glucose-deprived

523

conditions, DMEM no glucose, no glutamine media was supplemented with 1mM glutamine

524

and [U-

C]-150



M palmitate; for glucose and glutamine-deprived conditions, media was

525

supplemented with [U-

C]-150



M palmitate only. FAs were conjugated to 2% (wt/vol) FA-

526

free BSA containing media for 24 hours at 37



C prior to treating cells. Cells were incubated in

527

C containing medium for 6 hours, or up to 48 hours for time course experiments. For spheroid

528

experiments, 200



L of spheroids was pooled and collected and centrifuged at 1400 rpm for 5

529

min at room temperature. The supernatant medium was collected and used for further

530

extracellular substrate experiments. The remaining cell pellet was washed with 2 mL cold PBS

531

and centrifuged again. The cell pellet was collected for subsequent LC-MS analysis.

532

Extracellular substrate sampling

533

Cells were seeded in 6-well plates (5x10

) for 2D culture of as above for 3D spheroid models.

534

After 24 hours, wells were washed with warm PBS and media replaced with 1 mL or 2 mL of

535

DMEM no glucose, no glutamine media (Gibco) supplemented with 5 mM glucose, 1 mM

536

glutamine and 150



M FAs. 100



L of extracellular media were collected from wells at 3, 6,

537

12, 24 hour timepoints for cells incubated in 1 mL of media, and 12, 24 hours for cells incubated

538

in 2mL of media. For spheroid experiments,

C labelled medium was collected from cells at 0

539

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

and 6 hours timepoints. Media samples were centrifuged at 16 000xg for 5 min at 4



C and the

540

supernatant collected for subsequent LC-MS analysis.

541

Cell lysate extraction for LC-MS

542

Media was aspirated from plates and the cells were washed with 2 mL ice-cold 0.9% (w/v)

543

NaCl. Cells were scraped in 300



L of extraction buffer, EB, (1:1 LC/MS methanol:water

544

(Optima) + 0.1x internal standards (non-endogenous polar metabolites) and transferred to a 1.5

545

mL microcentrifuge tube. A further 300



L of EB was added to the cells and combined in the

546

tube. 600



L of chloroform (Honeywell) was added before vortexing and incubating on ice for

547

10 min. Tubes were vortexed briefly and centrifuged at 15 000g for 10min at 4



C. The aqueous

548

upper layer was collected and dried without heat, using a Savant SpeedVac (Thermo Fisher)

549

for metabolomics LC-MS analysis. The lower layer was collected and dried under nitrogen

550

flow for lipidomics LC-MS analysis. For time course lipidomics experiments, extractions were

551

carried out as previously described

by quenching with 0.1M formic acid, neutralising with

552

ammonium bicarbonate, and using 40:40:20 acetonitrile:methanol:water with 0.1 M formic

553

acid as the solvent system.

554

Measuring stable-isotope labelled metabolites by LC-MS

555

Dried aqueous upper phase samples were resuspended in 40



L Amide buffer A (20 mM

556

ammonium acetate, 20 mM ammonium hydroxide, 95:5 HPLC H

O: Acetonitrile (v/v)) and

557

vortexed and centrifuged at 15, 000g for 5min at 4



C. 20



L of supernatant was transferred to

558

HPLC vials containing 20



L acetonitrile for LCMS analysis of amino acids and glutamine

559

metabolites. The remaining 20



L of resuspended sample was transferred to HPLC vials

560

containing 20



L LC-MS H

O for LCMS analysis of glycolytic, pentose phosphate pathway

561

(PPP), and TCA cycle metabolites. Amino acids and glutamine metabolites were measured

562

using Vanquish-TSQ Altis (Thermo) LC-MS/MS system. Analyte separation was achieved

563

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

using a Poroshell 120 HILIC-Z Column (2.1x150 mm, 2.7



m) (Agilent) at ambient

564

temperature. The pair of buffers used were Amide buffer A and 100% acetonitrile (Buffer B),

565

flowed at 200



L/min; injection volume of 5



L. Glycolytic, PPP and TCA cycle metabolites

566

were measured using 1260 Infinity (Agilent)-QTRAP6500+ (AB Sciex) LC-MS/MS system.

567

Analyte separation was achieved using a Synergi 2.5



m Hydro-RP 100A LC Column (100x2

568

mm) at ambient temperature. The pair of buffers used were 95:5 (v/v) water:acetonitrile

569

containing 10 mM tributylamine and 15 mM acetic acid (Buffer A) and 100% acetonitrile

570

(Buffer B), flowed at 200



L/min; injection volume of 5



L. Raw data from both LC-MS/MS

571

systems were extracted using MSConvert

and in-house MATLAB scripts.

572

Measuring extracellular substrates by LC-MS

573



L of collected extracellular media was mixed with 80



L water, vortexed. 10



L of diluted

574

media was mixed with 90



L of extraction buffer containing 1:1 (v/v) acetonitrile and methanol

575

+ 1x internal standards (non-endogenous standards) at -30



C. The mixture was centrifuged at

576

12 000xg for 5 min at 4



C and transferred into HPLC vials for LC-MS analysis measured using

577

the Vanquish-TSQ Altis as described above. Raw data from both LC-MS/MS systems were

578

extracted using MSConvert

and in-house MATLAB scripts.

579

Measuring stable isotope labelled lipids by LC-MS

580

Dried lower phase samples were resuspended in 100



L of 4:2:1 (v/v) isopropanol: methanol:

581

chloroform containing 7.5 mM ammonium formate. Cardiolipin and acyl-carnitine lipids were

582

measured using a Thermo Scientific Q-Exactive-HF-X Hybrid Quadrupole Orbitrap LC-

583

MS/MS system.

584

For cardiolipins, analyte separation was achieved using an Agilent Poroshell 120, EC-C18,

585

2.1x150mm, 2.7



m column. The pair of buffers used were 60:40 acetonitrile: water (v/v) with

586

10 mM ammonium formate and 0.1% formic acid (mobile phase A), and 90:10 isopropanol:

587

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

acetonitrile (v/v) with 10 mM ammonium formate and 0.1% formic acid (mobile phase B),

588

flowed at 200



L/min on negative mode. MS1 data was acquired with the following settings:

589

3.5kV, capillarity temperature 300



C, 120 000, injection time 100 ms, AGC 1x10

, scan range

590

1100-1650. For ddMS2 data acquirement, the following settings were used: top 20, resolution

591

30,000, 200-2000, isolation 1.0m/z, nce 30, AGC target 1x10

, intensity threshold 5x10

592

dynamic exclusion 20 s. Raw data from both LC-MS/MS systems were extracted using

593

MSConvert

, in-house MATLAB scripts, and MSDial.

594

For acyl-carnitines, analyte separation was achieved using an Agilent Poroshell 120, HILIC,

595

2.1x100 mm, 2.7



m column. The pair of buffers used were 0.1% (v/v) formic acid and 10 mM

596

ammonium formate in H

O and 0.1% (v/v) formic acid in acetonitrile, flowed at 200



L/min

597

on positive mode. MS1 data were acquired with the following settings: 3.5kV, capillarity

598

temperature 300

℃

, 120 000, injection time 100ms, AGC 1x10

, scan range 200-1000. For

599

ddMS2 data acquirement, the following settings were used: top 5, resolution 30,000, 200-2000,

600

isolation 1.0m/z, nce 30, AGC target 1x10

, intensity threshold 5x10

, dynamic exclusion 20s.

601

Raw data from both LC-MS/MS systems were extracted using MSConvert

and in-house

602

MATLAB scripts.

603

Quantifying 13C mass shifts from high-resolution MS1 and MS2 data

604

The extraction of mass spectrometry data was performed in MATLAB. First, retention times

605

of CL, acyl carnitine, PC and PG species were confirmed using accurate precursor mass and

606

MS2 fragmentation patterns. Accurate masses for the monoisotopic and expected mass shifts

607

were calculated using ion formulas. The resulting ion cluster was used to visualise the analyte’s

608

MS1 chromatogram and to integrate the area under the peak. These extracted MS1 ion

609

intensities were then reported as

C fractional enrichments

610

A labelled acyl chain could be incorporated into CLs with or without modification (e.g.,

611

elongation or desaturation). Thus, MS2 data was used to verify localisation of

C-FA among

612

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

the acyl chains. From the MS2 data we can derive the ratio of labelled to unlabelled product

613

ions fragmented from a labelled precursor. Using relative ion intensities of labelled and

614

unlabelled acyl anion [FA - H]

, the ion intensity of a labelled precursor (from MS1) is

615

apportioned to its constituent acyl chains in terms of their labelled and unlabelled forms. The

616

percentage labelling of a given acyl chain among CL or PG pools was then calculated by

617

summing up these labelled and unlabelled constituents across all quantified CL or PG species.

618

Quantification and statistical analysis

619

Student’s t-test, one- or two-way ANOVA with Dunnett’s or Tukey’s multiple comparisons,

620

Nonparametric Wilcoxon signed rank test, and linear regression statistical tests were performed

621

with GraphPad Prism 9.0 (GraphPad Software). P < 0.05 was considered significant. Data are

622

reported as mean ± standard error of the mean (SEM), standard deviation (SD), or min to max

623

values of at least three independent determinations as indicated in figure legends. Schematic

624

diagrams were created with Biorender.com.

625

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Extended Data

626

Extended Data Fig. 1. Palmitate is a minor carbon source to the TCA cycle.

627

(A)

Percentage of glucose, glutamine, and FA 16:0 relative to starting amount, and amount of lactate in culture

628

media over 24 hours in 1mL culture medium. N=9 per cell line.

629

(B)

Percentage of glucose and glutamine relative to starting amount in 1 mL (blue) or 2 mL (red) culture medium.

630

N=9 per cell line.

631

(C)

Fractional contribution of [U-

C]-FA16:0 to citrate, oxoglutarate, succinate, fumarate, and malate. N=9 per

632

cell line. Statistically significant differences between PCa cells to PNT1 cells are shown.

633

(D)

Correlations of [U-

C]-FA 16:0 to citrate (plotted), oxoglutarate, succinate, fumarate, and malate to FAO.

634

Error bars represent



SD. P and r

values by linear regression.

635

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

(E)

Correlations of [U-

C]-FA 16:0 to citrate (plotted), oxoglutarate, succinate, fumarate, and malate to active

636

mitochondria content. Error bars represent



SD. P and r

values by linear regression.

637

(F)

Fractional contribution of [U-

C]-glucose to pyruvate (red), [U-

C]-glutamine to intracellular glutamine

638

(blue), [U-

C]-FA 16:0 to palmitoyl-carnitine (orange) after 6 hours of labelling in LNCaP, MR42D and

639

MR49F cells. N=9 per cell line.

640

(G)

Fractional contributions of [U-

C]-glucose, glutamine, or FA 16:0 to citrate, oxoglutarate and malate. N=9

641

per cell line.

642

(H)

Percentage of glucose, glutamine, and FA 16:0 m0 and m16 relative to starting amount, comparing unlabelled

643

and [U-

C]-FA 16:0 media at 0 and 6 hours in representative C4-2B spheroids. N=3 per group.

644

(I)

Fractional contribution of [U-

C]-FA 16:0 to palmitoyl-carnitine in 3D spheroid LNCaP, C4-2B, and 22RV1

645

models. N=4-5 per cell line.

646

(J)

Fractional contribution of [U-

C]-FA 16:0 to succinate, fumarate, and malate in 3D spheroid models. N=3

647

per cell line.

648

Graphs show mean



SEM unless stated otherwise. *p<0.05 by One-way ANOVA with Dunnett’s multiple

649

comparisons test, unless stated otherwise.

650

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Extended Data Fig. 2. Palmitate cannot maintain TCA cycle activity under glucose and

651

glutamine deprivation.

652

(A)

[1-

C]-FA 16:0 oxidation rates in the presence of 5 mM, 3 mM, 1 mM, or 0 mM glucose. N=3 per cell line.

653

* compared to 5 mM glucose, p<0.05 by One-way ANOVA with Dunnett’s multiple comparisons test.

654

(B)

Impact of glucose- or glucose and glutamine- deprivation on metabolic activity assessed by MTT assay to 48

655

hours, normalised to day 0 results. Dotted line at 24 hours indicates replenishment of media.

656

(C)

Impact of glucose- or glucose and glutamine- deprivation on cell viability assessed by live cell counts at 0, 6

657

and 24 hours normalised to day 0 results.

658

(D)

Relative abundance (normalised to +glucose/+glutamine) of [U-

C]-FA 16:0 to oxoglutarate or malate. N=3-

659

9 per cell line.

660

Graphs show mean



SEM. *compared to +glucose/+glutamine, ^compared to -glucose/+glutamine, p<0.05 by

661

One-way ANOVA with Dunnett’s multiple comparisons test.

662

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

667

Extended Data Fig, 4. FA 18:2 predominantly enriches CL pools compared to other FAs.

668

(A)

Enrichment fraction of PC 32:1 and 34:1 in PC3 and C4-2B cells over 48 hours exposure to[U-

C]-FA

669

16:0. N=1 per cell line.

670

(B)

Comparison of fractional contribution of [U-

C]-FA 16:0 to PC and CL species at 6 hours in PC3 and C4-

671

2B cells. N=3-9 per cell line. P<0.05 by two-way ANOVA with Tukey’s multiple comparisons test.

672

(C)

Fractional contribution of [U-

C]-FA 16:0, 18:0, 18:1, 18:2 (150



M) to CL species containing the same

673

acyl chains at 6 hours. N=3 per cell line.

674

(D)

Overall percentage of fatty acyls (of the same length and saturation as labelled substrate) contained in CL

675

species labelled by [U-

C]-FA 16:0, 18:0, 18:1 or 18:2. Unlabelled

C controls show level of background

676

labelling. N=3 per cell line.

677

Graphs show mean



SEM unless stated otherwise.

678

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

679

Extended Data Fig. 5. CPT1a

reduces FA 18:2 incorporation to CLs.

680

(A)

Percentage of 18:2 or 16:0 acyl chain in CLs respectively labelled by [U-

C]-FA 18:2 or 16:0 in scrambled

681

(scr) and siCPT1a cells. N=9 per group.

682

(B)

Fractional contribution of [U-

C]-FA 18:2 or 16:0 to 18:2 or 16:0 acyl chains of CL species. N=9 per group.

683

(C)

Delta differences between fractional contribution of [U-

C]-FA 18:2 or 16:0 to acyl chains of CL species.

684

N=7-8 per group. Statistically significant differences within [U-

C]-FA 18:2 or 16:0 groups to 0 difference,

685

determined by nonparametric Wilcoxon signed rank test. Error bars represent



min to max values.

686

Graphs show mean



SEM unless stated otherwise. ns, not significant, *p<0.05 by unpaired t-tests between

687

siCPT1a and scr control groups unless stated otherwise.

688

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Acknowledgements

689

N.T.S. is supported by the Australian Rotary Health/Rotary Club of Blacktown City ‘Mel Grey’

690

PhD scholarship. L.M.B. and A.J.H. acknowledge grant support from The Movember

691

Foundation/Prostate Cancer Foundation of Australia (MRTA3). A.J.H. is supported by a

692

Robinson Fellowship from the University of Sydney and funding from the University of

693

Sydney. L.M.B. is supported by a Principal Cancer Research Fellowship produced with the

694

financial and other support of Cancer Council SA’s Beat Cancer Project on behalf of its donors

695

and the State Government of South Australia through the Department of Health and was

696

supported by a Future Fellowship from the Australian Research Council (FT130101004). We

697

thank the Sydney Mass Spectrometry facility for access to LC-MS instruments; Julia Scott and

698

Dr Zeyad Nassar in the Butler lab (University of Adelaide) for assistance with 3D spheroid and

699

mitotracker experiment development; Dr Helen McGuire and the Sydney Flow Cytometry

700

facility for assistance with mitotracker experiments; and Prof. Lisa Horvath and members of

701

the Butler lab for scientific discussions.

702

Author contributions

703

Conceptualisation by A.J.H. and L.E.Q. Project administration, visualisation, writing-original

704

draft by N.T.S. Investigation by N.T.S., M.F.H.S., and A.W. Methodology by L.E.Q. Software

705

by N.T.S. and L.E.Q. Formal analysis by N.T.S., L.E.Q and A.J.H. Writing-editing & reviewing

706

by N.T.S., L.E.Q., A.J.H., and L.M.B. Funding acquisition, resources by A.J.H. Supervision by

707

A.J.H., L.E.Q., and L.M.B.

708

Declaration of interests

709

The authors declare that they have no competing interests.

710

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

References:

711

Swinnen, J. V., Esquenet, M., Goossens, K., Heyns, W. & Verhoeven, G. Androgens

712

stimulate fatty acid synthase in the human prostate cancer cell line LNCaP.

Cancer Res.

713

, 1086-1090 (1997).

714

2 Yue,

et al.

Cholesteryl ester accumulation induced by PTEN loss and PI3K/AKT

715

activation underlies human prostate cancer aggressiveness.

Cell Metab.

, 393-406,

716

doi:10.1016/j.cmet.2014.01.019 (2014).

717

3 Migita,

et al.

Fatty acid synthase: a metabolic enzyme and candidate oncogene in

718

prostate cancer.

J. Natl. Cancer Inst.

101

, 519-532, doi:10.1093/jnci/djp030 (2009).

719

Nassar, Z. D.

et al.

Human DECR1 is an androgen-repressed survival factor that

720

regulates PUFA oxidation to protect prostate tumor cells from ferroptosis.

Elife

721

doi:10.7554/eLife.54166 (2020).

722

5 Balaban,

et al.

Extracellular Fatty Acids Are the Major Contributor to Lipid

723

Synthesis in Prostate Cancer.

Mol. Cancer Res.

, 949-962, doi:10.1158/1541-

724

7786.Mcr-18-0347 (2019).

725

6 Fidelito,

et al.

Multi-substrate Metabolic Tracing Reveals Marked Heterogeneity and

726

Dependency on Fatty Acid Metabolism in Human Prostate Cancer.

Mol. Cancer Res.

727

, 359-373, doi:10.1158/1541-7786.Mcr-22-0796 (2023).

728

7 Iglesias-Gato,

et al.

The Proteome of Primary Prostate Cancer.

Eur. Urol.

, 942-

729

952, doi:10.1016/j.eururo.2015.10.053 (2016).

730

Butler, L. M.

et al.

Lipidomic Profiling of Clinical Prostate Cancer Reveals Targetable

731

Alterations in Membrane Lipid Composition.

Cancer Res.

, 4981-4993,

732

doi:10.1158/0008-5472.CAN-20-3863 (2021).

733

Ahmad, F., Cherukuri, M. K. & Choyke, P. L. Metabolic reprogramming in prostate

734

cancer.

Br. J. Cancer

125

, 1185-1196, doi:10.1038/s41416-021-01435-5 (2021).

735

Costello, L. C. & Franklin, R. B. The intermediary metabolism of the prostate: a key to

736

understanding the pathogenesis and progression of prostate malignancy.

Oncology

737

269-282, doi:10.1159/000012183 (2000).

738

Liu, Y. Fatty acid oxidation is a dominant bioenergetic pathway in prostate cancer.

739

Prostate Cancer Prostatic Dis.

, 230-234, doi:10.1038/sj.pcan.4500879 (2006).

740

Liu, Y., Zuckier, L. S. & Ghesani, N. V. Dominant uptake of fatty acid over glucose by

741

prostate cells: a potential new diagnostic and therapeutic approach.

Anticancer Res.

742

369-374 (2010).

743

Van Heijster, F. H. A., Breukels, V., Jansen, K. F. J., Schalken, J. A. & Heerschap, A.

744

Carbon sources and pathways for citrate secreted by human prostate cancer cells

745

determined by NMR tracing and metabolic modeling.

Proc. Natl. Acad. Sci. U. S. A.

746

119

, doi:10.1073/pnas.2024357119 (2022).

747

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

14 Dueregger,

et al.

Differential Utilization of Dietary Fatty Acids in Benign and

748

Malignant Cells of the Prostate.

PLoS One

, e0135704,

749

doi:10.1371/journal.pone.0135704 (2015).

750

Mah, C. Y., Nassar, Z. D., Swinnen, J. V. & Butler, L. M. Lipogenic effects of androgen

751

signaling in normal and malignant prostate.

Asian J. Urol

, 258-270,

752

doi:

https://doi.org/10.1016/j.ajur.2019.12.003

(2020).

753

Fritz, I. B. & Yue, K. T. Long-chain carnitine acyltransferase and the role of

754

acylcarnitine derivatives in the catalytic increase of fatty acid oxidation induced by

755

carnitine.

J. Lipid Res.

, 279-288 (1963).

756

Carracedo, A., Cantley, L. C. & Pandolfi, P. P. Cancer metabolism: fatty acid oxidation

757

in the limelight.

Nat. Rev. Cancer

, 227-232, doi:10.1038/nrc3483 (2013).

758

Ricciardi, M. R.

et al.

Targeting the leukemia cell metabolism by the CPT1a inhibition:

759

functional preclinical effects in leukemias.

Blood

126

, 1925-1929, doi:10.1182/blood-

760

2014-12-617498 (2015).

761

Schlaepfer, I. R.

et al.

Lipid Catabolism via CPT1 as a Therapeutic Target for Prostate

762

Cancer.

Mol. Cancer Ther.

, 2361-2371, doi:10.1158/1535-7163.mct-14-0183

763

(2014).

764

Pike, L. S., Smift, A. L., Croteau, N. J., Ferrick, D. A. & Wu, M. Inhibition of fatty acid

765

oxidation by etomoxir impairs NADPH production and increases reactive oxygen

766

species resulting in ATP depletion and cell death in human glioblastoma cells.

Biochim

767

Biophys Acta

1807

, 726-734, doi:

https://doi.org/10.1016/j.bbabio.2010.10.022

(2011).

768

21 Yao,

C.-H.

et al.

Identifying off-target effects of etomoxir reveals that carnitine

769

palmitoyltransferase I is essential for cancer cell proliferation independent of

770

oxidation.

PLoS Biol.

, e2003782, doi:10.1371/journal.pbio.2003782 (2018).

771

O'Connor, R. S.

et al.

The CPT1a inhibitor, etomoxir induces severe oxidative stress at

772

commonly used concentrations.

Sci. Rep.

, 6289, doi:10.1038/s41598-018-24676-6

773

(2018).

774

23 He,

et al.

Carnitine palmitoyltransferase-1b deficiency aggravates pressure overload-

775

induced cardiac hypertrophy caused by lipotoxicity.

Circulation

126

, 1705-1716,

776

doi:10.1161/circulationaha.111.075978 (2012).

777

Carta, G., Murru, E., Banni, S. & Manca, C. Palmitic Acid: Physiological Role,

778

Metabolism and Nutritional Implications.

Front. Physiol.

, 902,

779

doi:10.3389/fphys.2017.00902 (2017).

780

Ahn, W. S. & Antoniewicz, M. R. Parallel labeling experiments with [1,2-13C]glucose

781

and [U-13C]glutamine provide new insights into CHO cell metabolism.

Metab Eng.

782

34-47, doi:

https://doi.org/10.1016/j.ymben.2012.10.001

(2013).

783

26 Duan,

et al.

13C tracer analysis suggests extensive recycling of endogenous CO2 in

784

vivo.

Cancer Metab.

, 11, doi:10.1186/s40170-022-00287-8 (2022).

785

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Buescher, J. M.

et al.

A roadmap for interpreting (13)C metabolite labeling patterns

786

from cells.

Curr. Opin. Biotechnol.

, 189-201, doi:10.1016/j.copbio.2015.02.003

787

(2015).

788

Bishop, J. L.

et al.

The Master Neural Transcription Factor BRN2 Is an Androgen

789

Receptor–Suppressed Driver of Neuroendocrine Differentiation in Prostate Cancer.

790

Cancer Discov.

, 54-71, doi:10.1158/2159-8290.Cd-15-1263 (2017).

791

29 Joshi,

et al.

CPT1A Supports Castration-Resistant Prostate Cancer in Androgen-

792

Deprived Conditions.

Cells

, 1115 (2019).

793

Nassar, Z. D.

et al.

Fatty Acid Oxidation Is an Adaptive Survival Pathway Induced in

794

Prostate Tumors by HSP90 Inhibition.

Mol. Cancer Res.

, 1500-1511,

795

doi:10.1158/1541-7786.mcr-20-0570 (2020).

796

Sutherland, R. M., McCredie, J. A. & Inch, W. R. Growth of Multicell Spheroids in

797

Tissue Culture as a Model of Nodular Carcinomas2.

J Natl Cancer Inst.

, 113-120,

798

doi:10.1093/jnci/46.1.113 (1971).

799

32 Jones,

et al.

3D Growth of Cancer Cells Elicits Sensitivity to Kinase Inhibitors

800

but Not Lipid Metabolism Modifiers.

Mol. Cancer Ther.

, 376-388,

801

doi:10.1158/1535-7163.MCT-17-0857 (2019).

802

Tidwell, T. R., Røsland, G. V., Tronstad, K. J., Søreide, K. & Hagland, H. R. Metabolic

803

flux analysis of 3D spheroids reveals significant differences in glucose metabolism

804

from matched 2D cultures of colorectal cancer and pancreatic ductal adenocarcinoma

805

cell lines.

Cancer Metab.

, 9, doi:10.1186/s40170-022-00285-w (2022).

806

34 Ma,

et al.

Functional analysis of molecular and pharmacological modulators of

807

mitochondrial fatty acid oxidation.

Sci. Rep.

, 1450, doi:10.1038/s41598-020-58334-

808

7 (2020).

809

Sullivan, Lucas B.

et al.

Supporting Aspartate Biosynthesis Is an Essential Function of

810

Respiration in Proliferating Cells.

Cell

162

, 552-563,

811

doi:

https://doi.org/10.1016/j.cell.2015.07.017

(2015).

812

36 Le,

et al.

Glucose-Independent Glutamine Metabolism via TCA Cycling for

813

Proliferation and Survival in B Cells.

Cell Metab.

, 110-121,

814

doi:10.1016/j.cmet.2011.12.009 (2012).

815

37 Koves,

et al.

Peroxisome Proliferator-activated Receptor-

Co-activator 1

816

mediated Metabolic Remodeling of Skeletal Myocytes Mimics Exercise Training and

817

Reverses Lipid-induced Mitochondrial Inefficiency.

J. Biol. Chem.

280

, 33588-33598,

818

doi:10.1074/jbc.m507621200 (2005).

819

38 Koves,

et al.

Mitochondrial Overload and Incomplete Fatty Acid Oxidation

820

Contribute to Skeletal Muscle Insulin Resistance.

Cell Metab.

, 45-56,

821

doi:10.1016/j.cmet.2007.10.013 (2008).

822

Colbeau, A., Nachbaur, J. & Vignais, P. M. Enzymic characterization and lipid

823

composition of rat liver subcellular membranes.

Biochim Biophys Acta

249

, 462-492,

824

doi:10.1016/0005-2736(71)90123-4 (1971).

825

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Falabella, M., Vernon, H. J., Hanna, M. G., Claypool, S. M. & Pitceathly, R. D. S.

826

Cardiolipin, Mitochondria, and Neurological Disease.

Trends Endocrinol Metab.

827

224-237, doi:10.1016/j.tem.2021.01.006 (2021).

828

Ikon, N. & Ryan, R. O. Cardiolipin and mitochondrial cristae organization.

Biochem

829

Biophys Acta Biomembr.

1859

, 1156-1163,

830

doi:

https://doi.org/10.1016/j.bbamem.2017.03.013

(2017).

831

Schwall, C. T., Greenwood, V. L. & Alder, N. N. The stability and activity of respiratory

832

Complex II is cardiolipin-dependent.

Biochim Biophys Acta

1817

, 1588-1596,

833

doi:

https://doi.org/10.1016/j.bbabio.2012.04.015

(2012).

834

43 Dudek,

et al.

Cardiac-specific succinate dehydrogenase deficiency in Barth

835

syndrome.

EMBO Mol. Med.

, 139-154, doi:10.15252/emmm.201505644 (2016).

836

Eble, K. S., Coleman, W. B., Hantgan, R. R. & Cunningham, C. C. Tightly associated

837

cardiolipin in the bovine heart mitochondrial ATP synthase as analyzed by 31P nuclear

838

magnetic resonance spectroscopy.

J. Biol. Chem.

265

, 19434-19440 (1990).

839

Taylor, W. A. & Hatch, G. M. Identification of the human mitochondrial linoleoyl-

840

coenzyme A monolysocardiolipin acyltransferase (MLCL AT-1).

J. Biol. Chem.

284

841

30360-30371, doi:10.1074/jbc.M109.048322 (2009).

842

Zachman, D. K.

et al.

The role of calcium-independent phospholipase A2 in cardiolipin

843

remodeling in the spontaneously hypertensive heart failure rat heart.

J. Lipid Res.

844

525-534, doi:

https://doi.org/10.1194/jlr.M000646

(2010).

845

Xu, Y. & Schlame, M. The turnover of glycerol and acyl moieties of cardiolipin.

Chem.

846

Phys. Lipids

179

, 17-24, doi:

https://doi.org/10.1016/j.chemphyslip.2013.10.005

847

(2014).

848

Landriscina, C., Megli, F. M. & Quagliariello, E. Turnover of fatty acids in rat liver

849

cardiolipin: Comparison with other mitochondrial phospholipids.

Lipids

, 61-66,

850

doi:10.1007/BF02532585 (1976).

851

49 Wahjudi,

et al.

Turnover of nonessential fatty acids in cardiolipin from the rat

852

heart.

J. Lipid Res.

, 2226-2233, doi:10.1194/jlr.M015966 (2011).

853

Nagarajan, S. R., Butler, L. M. & Hoy, A. J. The diversity and breadth of cancer cell

854

fatty acid metabolism.

Cancer Metab

, 2, doi:10.1186/s40170-020-00237-2 (2021).

855

51 Oemer,

et al.

Fatty acyl availability modulates cardiolipin composition and alters

856

mitochondrial function in HeLa cells.

J. Lipid Res.

, 100111,

857

doi:10.1016/j.jlr.2021.100111 (2021).

858

Hoch, F. L. Cardiolipins and biomembrane function.

Biochimica et Biophysica Acta

859

(BBA)

1113

, 71-133, doi:10.1016/0304-4157(92)90035-9 (1992).

860

Cao, J., Liu, Y., Lockwood, J., Burn, P. & Shi, Y. A Novel Cardiolipin-remodeling

861

Pathway Revealed by a Gene Encoding an Endoplasmic Reticulum-associated Acyl-

862

CoA:Lysocardiolipin Acyltransferase (ALCAT1) in Mouse.

J. Biol. Chem.

279

, 31727-

863

31734, doi:10.1074/jbc.m402930200 (2004).

864

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Lu, Y.-W. & Claypool, S. M. Disorders of phospholipid metabolism: an emerging class

865

of mitochondrial disease due to defects in nuclear genes.

Frontiers in Genetics

866

doi:10.3389/fgene.2015.00003 (2015).

867

Xu, Y., Malhotra, A., Ren, M. & Schlame, M. The Enzymatic Function of Tafazzin*.

868

Biol. Chem.

281

, 39217-39224, doi:

https://doi.org/10.1074/jbc.M606100200

(2006).

869

Hostetler, K. Y., Galesloot, J. M., Boer, P. & Van Den Bosch, H. Further studies on the

870

formation of cardiolipin and phosphatidylglycerol in rat liver mitochondria. Effect of

871

divalent cations and the fatty acid composition of CDP-diglyceride.

Biochim Biophys

872

Acta

380

, 382-389, doi:10.1016/0005-2760(75)90106-x (1975).

873

57 Zadra,

et al.

Inhibition of de novo lipogenesis targets androgen receptor signaling in

874

castration-resistant prostate cancer.

Proc. Natl. Acad. Sci. U. S. A.

116

, 631-640,

875

doi:10.1073/pnas.1808834116 (2019).

876

58 Parida,

et al.

Limiting mitochondrial plasticity by targeting DRP1 induces

877

metabolic reprogramming and reduces breast cancer brain metastases.

Nature Cancer

878

, 893-907, doi:10.1038/s43018-023-00563-6 (2023).

879

Yang, J. H.

et al.

Snail augments fatty acid oxidation by suppression of mitochondrial

880

ACC2 during cancer progression.

Life Science Alliance

, e202000683,

881

doi:10.26508/lsa.202000683 (2020).

882

Flaig, T. W.

et al.

Lipid catabolism inhibition sensitizes prostate cancer cells to

883

antiandrogen blockade.

Oncotarget

, 56051-56065, doi:10.18632/oncotarget.17359

884

(2017).

885

61 Zhang,

et al.

Lipidomics reveals carnitine palmitoyltransferase 1C protects cancer

886

cells from lipotoxicity and senescence.

J Pharm Anal

, 340-350,

887

doi:10.1016/j.jpha.2020.04.004 (2021).

888

Paradies, G., Paradies, V., De Benedictis, V., Ruggiero, F. M. & Petrosillo, G.

889

Functional role of cardiolipin in mitochondrial bioenergetics.

Biochim Biophys Acta

890

1837

, 408-417, doi:

https://doi.org/10.1016/j.bbabio.2013.10.006

(2014).

891

63 Zhong,

et al.

Mitochondrial control of apoptosis through modulation of cardiolipin

892

oxidation in hepatocellular carcinoma: A novel link between oxidative stress and

893

cancer.

Free Radic. Biol. Med.

102

, 67-76,

894

doi:

https://doi.org/10.1016/j.freeradbiomed.2016.10.494

(2017).

895

64 Faubert,

et al.

Lactate Metabolism in Human Lung Tumors.

Cell

171

, 358-371.e359,

896

doi:10.1016/j.cell.2017.09.019 (2017).

897

65 Neinast,

et al.

Quantitative Analysis of the Whole-Body Metabolic Fate of

898

Branched-Chain Amino Acids.

Cell Metab.

, 417-429.e414,

899

doi:

https://doi.org/10.1016/j.cmet.2018.10.013

(2019).

900

Kiebish, M. A., Han, X., Cheng, H., Chuang, J. H. & Seyfried, T. N. Cardiolipin and

901

electron transport chain abnormalities in mouse brain tumor mitochondria: lipidomic

902

evidence supporting the Warburg theory of cancer.

J. Lipid Res.

, 2545-2556,

903

doi:10.1194/jlr.M800319-JLR200 (2008).

904

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint

Nyquist, M. D.

et al.

TALEN-engineered AR gene rearrangements reveal endocrine

905

uncoupling of androgen receptor in prostate cancer.

Proc. Natl. Acad. Sci. U. S. A.

110

906

17492-17497, doi:doi:10.1073/pnas.1308587110 (2013).

907

68 Li,

et al.

Androgen receptor splice variants mediate enzalutamide resistance in

908

castration-resistant prostate cancer cell lines.

Cancer Res.

, 483-489,

909

doi:10.1158/0008-5472.Can-12-3630 (2013).

910

69 Chambers,

et al.

A cross-platform toolkit for mass spectrometry and proteomics.

911

Nat. Biotechnol.

, 918-920, doi:10.1038/nbt.2377 (2012).

912

70 Lu,

et al.

Metabolite Measurement: Pitfalls to Avoid and Practices to Follow.

Annu.

913

Rev. Biochem.

, 277-304, doi:10.1146/annurev-biochem-061516-044952 (2017).

914

915

CC-BY-NC-ND 4.0 International license

perpetuity. It is made available under a

preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in

The copyright holder for this

this version posted January 20, 2024.

;

https://doi.org/10.1101/2024.01.16.575611

doi:

bioRxiv preprint