Starving Ovarian Cancer Cells of Cysteine May Boost Carboplatin Response

Ovarian cancer cells appear unusually vulnerable to the loss of a single amino acid, and that vulnerability may be turned against them. In a new study published in Cancer & Metabolism, researchers report that depriving ovarian cancer cell lines of cyst(e)ine—a term covering cysteine and its oxidized dimer cystine—made them markedly more responsive to carboplatin, the platinum chemotherapy at the backbone of standard ovarian cancer treatment. The effect held in carboplatin-resistant cells grown in the laboratory and, in an early preclinical test, in tumors implanted in mice. The work remains far from the clinic, but it offers a systematic map of how cancer cells from different tissues depend on different amino acids, and a concrete example of how those dependencies might be exploited.

The study began with an unusually broad question: what happens to cancer cells when you remove each of the twenty proteinogenic amino acids from their growth medium, one at a time? To answer it, the team profiled the viability of 60 cancer cell lines drawn from six tissue origins—breast, colon, lung, ovarian, pancreatic, and gastric cancer—after 48 hours of deprivation of each individual amino acid. Cells were grown in customized RPMI-based media lacking a single amino acid, and viable cells were quantified with a colorimetric assay, with careful controls to minimize plate-to-plate variation, including a co-seeded reference cell line whose response to glutamine starvation served as a quality gate.

The screening revealed patterns that clustered by both amino acid and lineage. Principal component analysis separated cyst(e)ine into a cluster of its own, distinct from the essential amino acids and from the remaining non-essential amino acids, hinting at its singular metabolic role. Statistical analysis showed cancer type-specific differences in deprivation sensitivity: valine, cyst(e)ine, glutamine, and arginine deprivation all produced lineage-associated differences in viability. After correction for multiple testing across the twenty amino acids, glutamine remained significant (FDR = 0.032) and cyst(e)ine showed a borderline trend (FDR = 0.084), while valine and arginine did not survive correction. Notably, cyst(e)ine deprivation reduced viability by more than 50% in certain lung and ovarian cancer cell lines, including those of poor-prognosis subtypes such as large cell carcinoma and serous carcinoma.

That relative sensitivity of ovarian cancer cells to cyst(e)ine loss prompted the key therapeutic question. The researchers tested cyst(e)ine deprivation alongside three standard ovarian cancer drugs—carboplatin, paclitaxel, and doxorubicin—across nine ovarian cancer cell lines under varying cystine concentrations. All three drugs showed some increase in responsiveness under cyst(e)ine starvation, but carboplatin exhibited the most pronounced and consistent effect. When viability data were normalized to account for the effect of deprivation alone, the combination of carboplatin and low cyst(e)ine substantially reduced relative viability, whereas the effects with paclitaxel and doxorubicin were modest and inconsistent across cell lines.

The combination was then pushed into harder models. Four ovarian cancer cell lines with relatively low baseline carboplatin responsiveness—SNU251, SNU8, SK-OV-3, and OVCAR-3—showed enhanced carboplatin-induced cell death under cyst(e)ine starvation, assessed by apoptotic protein markers and by flow cytometry for Annexin V and propidium iodide staining. The dominant mode of cell death varied by line: SK-OV-3 displayed a clear shift toward apoptotic death, while SNU251 and SNU8 showed mixed apoptotic and non-apoptotic features. More striking still, the team generated a carboplatin-resistant version of the TOV-21G cell line through roughly nine months of stepwise drug exposure, producing cells with a 4.7-fold higher carboplatin IC50 and elevated expression of resistance markers including BRCA1, ERCC1, and BCL2. Under cyst(e)ine starvation, these resistant cells showed a marked reduction in viability and elevated cleaved caspase-3 and PARP-1—responding even more strongly to cyst(e)ine loss than their parental counterparts, even without carboplatin present.

Because cyst(e)ine is a building block of proteins, dietary restriction in patients is not a realistic route. The researchers instead turned to pharmacology, deploying two inhibitors that attack cyst(e)ine metabolism at different nodes: sulfasalazine, which blocks the SLC7A11 (xCT) transporter that imports cystine, and DL-propargylglycine, which inhibits cystathionine γ-lyase, an enzyme of the transsulfuration pathway that synthesizes cysteine de novo. In SK-OV-3 and OVCAR-3 cells, adding both inhibitors to carboplatin reduced viability more than carboplatin alone, and combination index values below 1 at fraction affected values above 0.4 indicated synergistic interaction by the Chou–Talalay method. The triple treatment also increased cell death markers, and critically, both the loss of viable cells and the apoptotic marker induction could be rescued by supplying excess cyst(e)ine—linking the drug effects to cyst(e)ine availability itself.

The mechanism appears to run through redox chemistry. Cyst(e)ine feeds glutathione synthesis, the cell’s principal antioxidant buffer, so its loss should sensitize cells to the oxidative and DNA-damaging stress that platinum drugs impose. Consistent with this, the inhibitor combination lowered the ratio of reduced to oxidized glutathione (GSH/GSSG), and the triple treatment increased reactive oxygen species measured by DCFDA staining, along with phosphorylation of the DNA damage response proteins CHK1 and H2AX. The antioxidant N-acetylcysteine attenuated these responses, though the authors caution that NAC can also serve as an intracellular cysteine donor, so its rescue effects may reflect restored cyst(e)ine supply as well as redox buffering.

In a xenograft model, SK-OV-3 cells implanted subcutaneously in nude mice were treated for five weeks with carboplatin alone or with the two inhibitors added. Carboplatin reduced tumor volume relative to vehicle, and co-treatment with sulfasalazine plus propargylglycine suppressed tumor growth further, with no significant changes in body weight across groups, suggesting no apparent gross toxicity under the tested conditions. This in vivo validation, however, rested on a single subcutaneous model with five mice per group—far from orthotopic or patient-derived settings that would better reflect the tumor microenvironment of human ovarian cancer, which the authors themselves flag as a key limitation.

To probe the underlying molecular response, the team performed RNA sequencing on SK-OV-3 cells treated with vehicle, carboplatin, the inhibitor pair, or all three. The triple combination produced the most distinct transcriptional profile, with enriched signatures of p53 pathway, apoptosis, DNA repair, and unfolded protein response. Transcription factor enrichment analyses converged on ATF3, a stress-responsive gene whose mRNA rose with each treatment and climbed highest under the triple combination. When the researchers knocked down ATF3 with siRNAs, levels of cleaved caspase-3 and cleaved PARP-1 partially declined, indicating that ATF3 contributes, at least in part, to the apoptotic response—though only partially, implying other parallel pathways. The JNK signaling pathway also appeared involved: the triple treatment increased JNK phosphorylation and ATF3 expression, both reduced by cyst(e)ine supplementation or NAC, and a JNK inhibitor partially blunted ATF3 induction and apoptotic markers.

The authors are careful about what these results do and do not establish. The screening used cell lines with uneven subtype representation, and responses varied substantially even within lineages, suggesting that static expression levels of cyst(e)ine metabolism genes correlate only weakly and inconsistently with starvation sensitivity. The cystine concentrations used in vitro—0.5 and 200 micromolar, roughly 1 and 400 micromolar cysteine equivalents—were intended to represent discrete metabolic states rather than physiological thresholds, and reported human plasma total cysteine concentrations of approximately 200–300 micromolar sit near the high end of the tested range, while tumor microenvironment concentrations may vary substantially. Validation in patient-derived samples, more physiologically relevant xenografts, and clearer mechanistic dissection of apoptotic versus non-apoptotic death pathways all remain to be done. Still, as a proof of concept, the study demonstrates a systematic framework for finding amino acid dependencies and shows that pharmacological targeting of cyst(e)ine metabolism can re-sensitize ovarian cancer models—including resistant ones—to a workhorse chemotherapy.

Subject of Research: Cyst(e)ine metabolism as a metabolic target for enhancing carboplatin responsiveness in ovarian cancer

Article Title: Metabolic targeting of cyst(e)ine enhances carboplatin responsiveness in ovarian cancer

Article References: Yoon, I., Oh, S., Park, S. Y., Park, J., Jang, H., Park, E., You, K. A., Kim, K., Hwang, S., Kim, H. Y., Cho, S., Han, J. M., & Kim, S. (2026). Metabolic targeting of cyst(e)ine enhances carboplatin responsiveness in ovarian cancer. Cancer & Metabolism, 14(1), Article 34. https://doi.org/10.1186/s40170-026-00450-5

Image Credits: AI Generated

DOI: 10.1186/s40170-026-00450-5

Keywords: ovarian cancer, cysteine metabolism, carboplatin, amino acid deprivation, drug resistance, sulfasalazine, propargylglycine, ATF3, glutathione, xenograft model, cancer metabolism, preclinical study