In this study, we used MDA-MB-231 (hormonal-independent triple-negative breast cancer). Cells were cultured in DMEM-F12 (Biowest), consisting of 10% fetal bovine serum (Biowest), 1% antibiotic-antimycotic (ATCC), and 0.5% non-essential amino acids (GIBCO) in a humidified incubator at 37°C, in the presence of 5% CO₂. When the mixture had 80% confluency, the cells were separated with Versene (KCl, EDTA, NaCl, and Tris-HCl), and cell viability was determined through a Trypan blue assay (4%) with a Neubauer camera (Strober, 2001).

NaH₂PO₄, Na₂HPO₄, deuterium oxide (D₂O) (D, 99.98%), and cisplatin were purchased from Sigma-Aldrich (St. Louis, MO, United States). Trimethylsilyl propionic acid-d₄ sodium salt (TSP) was supplied by Merck (Darmstadt, Germany). CasIIgly (4,7-dimethyl, 1,10-phenanthroline, glycinato copper (II) nitrate, monohydrate) was synthetized as in patents (Ruiz-Azuara, 1990; Ruiz-Azuara, 1996; Ruiz-Azuara, 1997).

The cells were cultured in 96-well plates with 10,000 cells/well with 100 μL of medium in a humidified incubator at 37°C, in the presence of 5% CO₂ for 24 h. Subsequently, the culture medium was withdrawn and 90 μL of medium and 10 μL of compound solution was added. Cells were exposed to treatment for 24 h and after this were retired. Then, an MTT assay was conducted according to the procedure reported by Mosmann (Mosmann, 1983).

A total of 1,000,000 cells were cultured in Petri dishes (10 cm) in a humidified incubator at 37°C, in the presence of 5% CO₂, and 5.4 ml of medium was added with 600 μL of compound solution in triplicate during the evaluation times (20 and 40 min, without treatment, cisplatin, and CasIIgly). After the treatment time, the cells were separated and again their viability was checked by a Trypan blue assay. For metabolite extraction, the cell suspension was centrifuged at 3,500 × g for 5 min at 4°C, the supernatant was discarded, and the cell button was maintained for 5 min in ice. Subsequently, 1 ml of acetonitrile/water mixture (50%) was added and the suspension was sonicated for 20 min. Finally, the cells were centrifuged at 14,500 × g for 15 min at 4°C. The supernatant was recovered and the acetonitrile was evaporated (Beckonert et al., 2007).

The dry samples of metabolites was resuspended in 700 μL of PBS-TSP solution [NaH₂PO4 (1 mM)/Na₂HPO₄ (1 mM)/TSP(0.02 mM), pH = 7.4]. The samples were homogenized and centrifuged at 12,000 × g for 5 min. The supernatant of each sample was transferred to a 5-mm NMR tube for analysis.

¹H-NMR analysis was conducted at 298 K on an Avance III HD spectrometer operating at a ¹H frequency of 699.95 MHz, 16.4T (Bruker, Billerica, MA, United States) equipped with a 5-mm z-axis gradient TCI cryoprobe and SampleJet automatic sample changer. ¹H-NMR spectrum was recorded using the CPMG (Carr-Purcell-Meiboom-Gill) pulse sequence with residual water signal suppression using presaturation (cpmgpr1d). A total of 256 scans were collected into 64 k datapoints over a spectral width of 8,403.361 Hz, and a relaxation delay of 4 s. Free induction decays (FIDs) were multiplied by an exponential function with a line-broadening factor of 0.3 Hz before Fourier transformation. The ¹H-NMR spectra were manually corrected for phase and base line distortion using TopSpin 3.5 pl 6 (Bruker, Billerica, MA, United States). ¹H NMR chemical shifts were referenced to the TSP signal at 0.00 ppm.

Each ¹H-NMR spectrum was subdivided into 0.02 ppm buckets and were normalized to the total area. The residual signal of water was excluded. The resulting data matrix were subjected to chemometric analysis in SIMCA 17.0.1 software (Umetrics, Sartorius Stedim Biotech). Principal Component Analysis (PCA) was performed, one of the most useful exploratory and unsupervised techniques to exclude outliers, and to identify the dominant variant in the dataset not associated with the biological effect. Subsequently, Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) was used for identifying differences among groups, which uses class memberships to maximize the discrimination between groups for a particular biological effect. The quality of each model was determined by the goodness of fit parameter (R ²) and the goodness of prediction parameter and the fraction of the total variation predicted by a component (Q²). The determination of the metabolites responsible for the separation between control and BD groups was done using the loadings plot and the variable importance in the projection (VIP) value of each bin in the model to identify biomarkers.

The identification and quantification of metabolites were realized with Chenomx NMR Suite v. 8.6 (Chenomx Inc., Edmonton, Canada), based on the location of individual resonances on the spectra using the TSP as internal reference and the area under the curve.

The metabolite concentrations were adjusted for each sample to the cellular volume obtained in their recuperation at the end of the treatment with the compounds for their comparison and analysis.

Quantitative Enrichment Analysis (QEA) was realized with MetaboAnalyst 5.0 software using the metabolites concentration for the identification of the relevant metabolic pathways in each compound analyzing both treatment times together.

The activity of the compounds over the MDA-MB-231 cell line was determined as IC₅₀. This concentration for each compound was used for the subsequent experiments. Table 1 shows the CI₅₀ for cisplatin and CasIIgly.

The mean inhibitory concentration of the compounds (Cisplatin, 21.71 μM and CasIIgly, 1.55 μM) was administered to the MDA-MB-231 cells for 0, 20, 40, 60, 120 min, and 4 and 6 h, in order to make sure that the effects observed in the metabolomic assay were a result of the treatment and not due to cell death. All cells were maintained above 90% viability even within the compounds, the data are presented in the Supplementary Material.

Once the viability of the cells was verified, we employed two treatment times: 20 and 40 min for metabolomic experiments as a first approximation for the behavior of these compounds.

Typical 700 MHz ¹H NMR spectra and representative metabolites assignments of the samples are shown in Figure 1.

The ¹H-NMR spectra of the samples were subjected to multivariate analysis to assess grouping and to identify variations for cells without treatment and with both compounds. Firstly, a supervised method was employed to explore and visualize variations and possible patterns in the dataset (Figure 2A). The resultant model had two latent variables which were able to account for 82% of spectral variation (R²X), 97.8% of group variation (R²Y), and the prediction accuracy was 72.8% (Q²). OPLS-DA showed differences between samples due to the different treatment behaviors. Subsequently, the matrix was divided in two according to the treatment time: one for 20 min and another one for 40 min. For the 20 min treatment, the OPLS-DA had two latent variables which were able to account for 100% of spectral variation (R²X), 100% of group variation (R²Y), and the prediction accuracy was 100% (Q²) (Figure 2B). For the 40 min treatment, the OPLS-DA had two latent variables that were able to account for 80.8% of spectral variation (R²X), 89.6% of group variation (R²Y), and the prediction accuracy was 57.3% (Q²) (Figure 2C). These results suggested that cisplatin and CasIIgly treatments had different and important effects over cell metabolism as they were differently grouped in comparison with no treatment cells even at the same time.

Variable importance in projection (VIP > 1.5) was considered to establish the bins responsible for the separation between samples. Some metabolites assigned to those bins were glucose, choline, alanine, lactate, pyruvate, glutamate, valine, glutamine, leucine, and succinate (Supplementary Tables S1, 2). These metabolites were the cause of the difference between the samples, so they were quantified at both times for each treatment, and each compound treatment was subsequently subjected to QEA, where we considered pathways with a p-value <0.05 as statistically significant.

In Figure 3A (and Supplementary Table S6), we present the altered metabolic routes for no treatment cells. For the discussion section, we only considered metabolic pathways that were statistically significant, and that also had an important effect over cancer progression. With this criterion, phospholipid biosynthesis, betaine metabolism, phosphatidylcholine and phosphatidylethanolamine biosynthesis, and methionine metabolism were relevant.

For cisplatin, the altered metabolic pathways in the samples are presented in Figure 3B (and Supplementary Table S7). For the discussion section, we only considered metabolic pathways that were statistically significant, and that also had an important effect over cancer progression, with this criterion, phosphatidylcholine, phosphatidylethanolamine, phospholipid biosynthesis and methionine metabolism were relevant.

The altered metabolic pathways in the samples due to CasIIgly treatment are shown in Figure 3C (and Supplementary Table S8). For the discussion section, we only considered metabolic pathways that were statistically significant, and that also had an important effect over cancer progression, with this criterion, the Warburg effect, pyruvate metabolism, gluconeogenesis, glycolysis, electron transport chain, β-oxidation, and the pentose pathway were relevant.