Abstract

   Breast cancer (BC), characterised by its diverse subtypes and molecular
   heterogeneity, remains a major challenge in oncology. Despite advances
   in chemotherapy, such as doxorubicin (Dox), limitations persist due to
   toxicity and drug resistance. Pigment epithelium-derived factor (PEDF)
   is a multifunctional protein with unique anti-tumour properties. The
   aim here was to elucidate metabolic reprogramming in human BC cell
   lines using a metabolomics approach. Untargeted gas
   chromatography-quadrupole mass spectrometry (GC/Q-MS) was employed to
   identify the metabolic alterations in BC cell lines MCF-7 (ER-positive)
   and MDA-MB-231 (TNBC) following treatment with PEDF, Dox, and their
   combination (Dox+PEDF) in comparison to untreated controls. Statistical
   models were employed using a combination of multivariate and univariate
   analyses, including partial least squares discriminant analysis
   (PLS-DA) and one-way ANOVA, applied by MetaboAnalyst and SIMCA
   software. To address the potential for multiple-testing errors, false
   discovery rate (FDR)-adjusted p-values were calculated to ensure robust
   statistical reliability. The overall analysis revealed significant
   metabolic alterations across the treatment groups, with distinct
   patterns emerging in carbohydrate, lipid, and amino acid metabolisms.
   In MCF-7 cells, PEDF combined with Dox significantly decreased cystine
   levels and modulated aspartic acid and lipid-related metabolites,
   indicating potential shifts in redox homeostasis and membrane
   composition. In MDA-MB-231 cells, the combination treatment
   significantly reduced glucose-6-phosphate and lactate levels,
   suggesting remodeling of glycolytic flux and redox balance.
   Furthermore, the combination of PEDF and Dox influenced amino acid and
   lipid metabolism. Pathway enrichment and correlation analyses revealed
   significant perturbations in glutathione metabolism, energy pathways,
   and lipid signaling, with notable differences between the two cell
   lines. Combining Dox and PEDF induced coordinated changes in metabolic
   networks, suggesting synergistic and antagonistic mechanisms that
   impact multiple biochemical pathways. These findings underline the
   importance of combining PEDF with chemotherapy to improve treatment
   outcomes in BC.

   See also the graphical abstract[28](Fig. 1).

   Keywords: breast cancer, PEDF, doxorubicin, GC/MS, metabolomics,
   metabolic profiling

1. Introduction

   Breast cancer (BC) remains a critical challenge in oncology due to its
   biological complexity and diverse subtypes (Zubair et al.,
   2021[[29]47]). The classification of BC is largely based on the
   presence or absence of specific receptors, including those for estrogen
   (ER), progesterone (PR), and human epidermal growth factor (HER2).
   Subtypes like luminal (ER/PR positive), HER2-positive, and
   triple-negative breast cancer (TNBC; which lacks all three receptors)
   exhibit distinct clinical behaviours and treatment responses (Dent et
   al., 2007[[30]17], Cancer Genome Atlas Network, 2012[[31]13]). Despite
   recent advancements in targeted therapies, effective treatment of BC
   remains challenging due to its propensity for treatment resistance and
   the risk of recurrence (Abooshahab et al., 2024[[32]1]). This
   highlights the urgent need for continuous innovation in treatment
   strategies to combat this disease.

   Current treatment strategies often rely on chemotherapy, such as
   doxorubicin (Dox), a well-known anthracycline valued for its potent
   anti-tumour activity through DNA intercalation and inhibition of
   topoisomerase II, ultimately leading to apoptosis in unbridled cells
   (Tacar et al., 2013[[33]41]). However, resistance to chemotherapy and
   the adverse side effects associated with Dox pose substantial obstacles
   in clinical practice. Consequently, exploring combination therapies
   that can enhance Dox efficacy while minimising toxicity has become a
   focal point in cancer research. Pigment epithelium-derived factor
   (PEDF), a multifunctional protein with anti-angiogenic,
   anti-inflammatory, and anti-tumour properties (Abooshahab et al.,
   2021[[34]2]), has shown promise in enhancing the anti-cancer effects of
   Dox while potentially reducing its toxicity. Our group has previously
   found that PEDF may help alleviate some of Dox's side effects on vital
   organs, including the heart, small intestine, and testes, indicating a
   protective effect that could support combination therapy (Tacar et al.,
   2015[[35]40]). In addition, our recent study further explored the
   combined effects of Dox and PEDF on human BC cells, revealing that the
   Dox+PEDF treatment significantly impacts signaling pathways and
   metastatic markers expression across different BC types (Abooshahab et
   al., 2024[[36]1]). However, the precise biochemical mechanisms
   underlying the combined action of Dox and PEDF remain largely
   unexplored.

   Building on those findings, this study employed metabolomics profiling
   to characterise the effects of Dox and PEDF combination on BC cells.
   Metabolomics, the comprehensive analysis of small molecules, provides a
   unique window into the biochemical landscape of cancer cells, revealing
   insights into the metabolic adaptations that drive cancer progression
   and treatment response (Beger, 2013[[37]10], Abooshahab et al.,
   2024[[38]5]). Unlike other omics approaches, metabolomics provides
   direct insights into cellular function and biochemical activity,
   capturing real-time physiological responses to treatment (Beger,
   2013[[39]10], Schmidt et al., 2021[[40]35]). Metabolomics studies on
   Dox-treated cancer cells have revealed significant alterations in
   pathways related to amino acid metabolism, energy production, and
   oxidative stress (Alsherbiny et al., 2021[[41]7], Alarcon-Zapata et
   al., 2023[[42]6], Govindasamy et al., 2023[[43]22], Liu et al.,
   2023[[44]28], Rushing et al., 2023[[45]34]). More recently, studies by
   our team that utilised metabolomics to investigate PEDF's effects on BC
   cells have reported alterations in pathways associated with glucose
   metabolism, lipid processing, and amino acid synthesis (Abooshahab et
   al., 2023[[46]3], Abooshahab et al., 2023[[47]4]).

   In this study, we employed gas chromatography-mass spectrometry
   (GC/MS)-based metabolomics to investigate the metabolic alterations in
   two human BC cell lines, MCF-7, an ER-positive luminal subtype, and
   MDA-MB-231, a TNBC subtype, following treatment with PEDF, Dox, and
   their combination. These cell lines are widely used in vitro studies
   and offer robust models for exploring key aspects of BC biology due to
   their contrasting phenotypic and genotypic profiles. By elucidating the
   unique and shared metabolic pathways affected by these treatments, we
   aimed to find the molecular mechanisms driving the potential synergy or
   antagonism between PEDF and Dox at biochemical levels. Such insights
   could pave the way for more effective and less toxic treatment
   strategies for BC, ultimately improving clinical outcomes for patients
   with different BC subtypes.

2. Materials and Methods

2.1. Materials

   Recombinant PEDF was purchased from MD Bioproducts (Bethesda, MD, USA).
   The MCF-7 and MDA-MB-231 cell lines were obtained from the American
   Tissue Culture Collection, ATCC (Manassas, VA, USA). Dulbecco's
   modified Eagle medium (DMEM), foetal bovine serum (FBS), and
   antibiotic/antimycotic were all purchased from Sigma-Aldrich, Saint
   Louis, MO, USA. HPLC-grade Doxorubicin (Dox), isopropanol (IPA),
   methanol (MeOH), 4,4′-dibromooctafluorobiphenyl, and water (H[2]O) were
   obtained from Sigma-Aldrich (Saint Louis, MO, USA). Derivatisation
   reagents, including methoxyamine (MOX), trifluoroacetamide (MSTFA), and
   1% trimethylsilyl chloride (TMCS), were purchased from Thermo Fisher
   Scientific, Waltham, MA, USA.

2.2. Cell Cultivation and Harvesting for Metabolomics Analysis

   The human luminal-A (ER+/PR+/−/HER2−) breast adenocarcinoma cell line
   (MCF-7) and the human triple-negative breast cancer cell line
   (MDA-MB-231) were cultured in DMEM with 10% foetal bovine serum (FBS)
   and 1% antibiotics and antimycotics under normal glucose conditions (5
   mM). The cultures were grown to 80% confluence under controlled
   conditions at 37 °C with 5% CO2.

   For metabolomics studies, MCF-7 and MDA-MB-231 cells were seeded at
   densities of 4 × 10⁵ cells/well and 35 × 10⁵ cells/well, respectively,
   in 24-well plates. After 24 h of culture, cells were exposed to one of
   the following treatments: PEDF (100 nM) (Filiz et al., 2012[[48]19]),
   Dox (0.5 µM) (Aniogo et al., 2017[[49]9]), or a combination of PEDF and
   Dox, with untreated cells as the control group, and incubated
   overnight. After 24 h, the cells underwent trypsinisation and were
   centrifuged at 700 g for 5 min. The culture medium was discarded, and
   the cell pellets were rapidly frozen in liquid nitrogen to quench
   metabolism and stored at −80 °C until sample preparation. The
   experiment was carried out with five replicates per treatment group and
   two technical replicates.

2.3. Sample Processing for Metabolomics Profiling

   Metabolite extraction and derivatisation were performed following a
   previously described method with a few modifications (Abooshahab et
   al., 2023[[50]3], Abooshahab et al., 2023[[51]4]). Briefly, 500 µL of a
   chilled protein precipitation solution (a mixture of methanol,
   isopropanol, and water) 2:2:1 (v/v/v) was added to each sample. Then
   the samples were thoroughly mixed and kept at 4 °C for 20 minutes to
   facilitate precipitation. Subsequently, the mixtures were centrifuged
   at 21,952 g for 15 minutes at 4 °C. The resulting supernatants,
   containing cellular metabolites, were collected and dried under a
   nitrogen stream using a 24-position MICROVAP evaporator. The dried
   samples underwent a two-step derivatisation process. First,
   methoxyamine hydrochloride (40 μL) was added, and the mixture was
   incubated at 60 °C for 1 hour to achieve methoxyamination. In the
   second step, MSTFA, containing 1% TMCS as a trimethylsilylation agent,
   was introduced, and the reaction was completed by incubating the
   mixture at 45 °C for 20 minutes. Subsequently, each sample was
   dissolved in 20 µL of an injection standard consisting of
   4,4′-dibromooctafluorobiphenyl at a concentration of 10 mg/L in hexane.
   The supernatant from each sample was promptly transferred into glass
   vials compatible with the autosampler for GC/MS analysis.

2.4. Metabolomics Data Acquisitions Using GC/MS

   An Agilent 5977B MSD/Agilent 8860 GC system fitted with a Restek
   Rxi-5-ms column (30 m length × 0.25 mm internal diameter (id); 0.25 μm
   film) was used to analyse the derivatised samples. Approximately 1 µL
   of each sample was injected at a split ratio of 1:1 in random order.
   The chromatographic method was conducted at a constant flow rate of
   helium, 1 mL/min, as the carrier gas. The initial oven temperature was
   held at 50 °C for 1 min, ramping from 20 °C/min to 320 °C, then held at
   320 °C for 5 min. The MS source, transfer line, and quadrupole
   temperature were set at 150 °C, 290 °C, and 250 °C, respectively,
   operating in electron ionisation mode at −70 eV. After a 5.2 min
   solvent delay, mass spectrometry data were collected at a scan rate of
   20 spectra/sec within the m/z 50-600 range.

2.5. Data Processing and Statistical Analysis

   The raw data files were converted to ABF format using a free conversion
   tool (Lai et al., 2018[[52]27]). The data were processed with MS-DIAL
   (version 4.9) to generate a comprehensive data matrix, including
   InChIKey identifiers, peak intensities, and the average retention time
   (RT) from the original dataset. The parameters used in MS-DIAL are
   provided in Table S1. Metabolite names were assigned to the GC/MS
   spectra based on mass spectral similarity and verified against multiple
   spectral libraries, such as the Fiehn library, MassBank with a ≥70%
   similarity threshold, and the Human Metabolome Database (HMDB) for
   precise metabolite identification. To ensure data accuracy and quality,
   contaminant ions derived from derivatisation reagents were excluded
   from the dataset. Peak intensities of duplicate features were summed to
   avoid redundancy. Blank reduction was performed to eliminate noise and
   ensure robust metabolite detection.

   To normalise peak intensities, sum normalisation and autoscaling were
   applied on both cell lines using MetaboAnalyst (v6.0), ensuring a
   near-normal distribution of values. Multivariate statistical analyses,
   including partial least squares-discriminant analysis (PLS-DA), were
   performed in SIMCA-P 14.0 (Umetrics, Umeå, Sweden) to distinguish
   metabolic variations between experimental groups. Metabolites with
   variable importance in projection (VIP) minimum scores above 0.8 were
   deemed significant for group discrimination (Cox et al., 2013[[53]16]).
   Model reliability was assessed through cross-validated predictive
   residuals, CV-ANOVA, and permutation tests. To identify metabolites
   significantly influenced by experimental factors, one-way ANOVA was
   conducted on the annotated GC/MS dataset, with p-values adjusted for
   multiple testing using false discovery rate (FDR) correction.
   Metabolite intensities were visualised as box plots using the R
   packages “ggpubr” and “tidyverse.” Additionally, Pearson correlation
   coefficients were computed to explore pairwise correlations among all
   metabolites, displayed as a heatmap via the “corrplot” R package.
   Finally, primary biological pathways were examined through enrichment
   pathway analysis in MetaboAnalyst (v6.0), with significance thresholds
   set at p-value < 0.05.

3. Results

3.1. Metabolic Profiles Among the Groups

   The metabolomics data from the MCF-7 and MDA-MB-231 cell lines were
   processed separately using MS-DIAL, identifying a total of 368
   compounds for MCF-7 and 380 compounds for MDA-MB-231. From these, 43
   metabolites for MCF-7 and 50 metabolites for MDA-MB-231 were considered
   reliable for further analysis. The identified metabolites for each cell
   line were then categorised using the ClassyFire classification system
   (Djoumbou Feunang et al., 2016[[54]18]). At the superclass level, the
   metabolic profiles of MCF-7 and MDA-MB-231 cells revealed a
   predominance of organic acids and derivatives, accounting for the
   majority of metabolites, 69.76% in MCF-7 and 68% in MDA-MB-231. This
   was followed by lipids and lipid-like molecules, comprising 13.95% in
   MCF-7 and 12% in MDA-MB-231. Organic oxygen compounds comprised 11.62%
   in MCF-7 and 10% in MDA-MB-231 (Figure 2A, 2B[55](Fig. 2)). At the
   subclass level, amino acids, peptides, and their analogs dominated both
   cell lines, representing 53.5% of the metabolites in MCF-7 and 47.9% in
   MDA-MB-231. Tables S2 and S3 provide a detailed list of the identified
   metabolites, including their respective classes.

Figure 2. Figure 2. Metabolic features of breast cancer cell lines. A, B)
Stacked bar charts representing the distribution of metabolite subclasses by
percentage across different metabolite categories. The PLS-DA score plots
visualise group separation among control, PEDF, Dox, and Dox+PEDF groups in
MCF-7 (C) and MDA-MB-231 (E). Each point represents a sample, colour-coded by
treatment group. The R² and Q² values indicate the model's explained variance
and predictability, respectively, with significant group clustering (p <
0.05). D, F) Permutation tests as a validation plot for the models displaying
R² and Q² values for 100 permutations. The values demonstrate the model's
robustness and predictive accuracy.

   [56]Figure 2
   [57]Open in a new tab

   We employed supervised PLS-DA models to identify the metabolites that
   distinguished the groups in both cell lines. The PLS-DA score plots
   exhibited distinct and well-separated clusters for the four groups,
   with acceptable levels of explained variance for MCF-7 and MDA-MB-231,
   highlighting significant metabolic alterations across the samples
   (Figure 2C, 2E[58](Fig. 2)). The models were statistically significant,
   as evidenced by CV-ANOVA (p < 0.05). Additionally, the predictive
   performance of the models was validated through a permutation test (n =
   100), where the original R² and Q² values exceeded those of the
   permuted models, confirming the models' reliability and predictive
   strength (Figure 2D, 2F[59](Fig. 2)).

3.2. Response of MCF-7 Metabolomes to PEDF, Dox, and Dox+PEDF

   A one-way ANOVA (p < 0.05, FDR q < 0.05) followed by multivariate
   analysis (VIP score > 0.8) identified 24 significantly altered
   metabolites in MCF-7 from a total of 43 metabolites (Table 1[60](Tab.
   1)). These metabolites demonstrated marked differences across the four
   experimental groups, particularly impacting glycolysis, amino acid
   metabolism, and lipid biosynthesis. Box plots were generated for the
   most statistically significant metabolites to provide a clear
   visualisation of the group-specific variations in metabolite abundance,
   as shown in Figure 3[61](Fig. 3).

Table 1. Significant alterations in MCF-7 metabolites.

   [62]Table 1
   [63]Open in a new tab

Figure 3. Figure 3. Box plots for significantly altered metabolites in MCF-7
cells. The boxplots display the distribution of 24 metabolites in MCF-7 cells
that exhibited the highest significance (p-values < 0.05 and VIP scores ≥
0.8) in the analysis of variance. These boxplots enable comparison of the
four groups: Control, PEDF, Dox, and Dox+PEDF. On the x-axis, each group is
represented by individual metabolites, while the y-axis indicates the
normalised peak intensity. Metabolites showing significant differences were
calculated using Tukey's Honestly Significant Difference (Tukey HSD) test and
indicated as (*) p ≤ 0.05, (**) p ≤ 0.01, and (***) p ≤ 0.001.

   [64]Figure 3
   [65]Open in a new tab

   Cystine showed the most notable variation among the treatment groups (q
   = 1.23E-09, VIP = 1.100). Compared to the control group, a significant
   decrease and increase were observed in the PEDF and Dox groups. The
   combination of Dox and PEDF reduced cystine level compared to Dox
   treatment alone, although it remained elevated relative to the control.
   This suggests a potential antagonist interaction between Dox and PEDF
   in lowering cystine levels, reflecting alterations in redox
   homeostasis. Aspartic acid was significantly reduced in the PEDF, Dox,
   and Dox+PEDF groups (q = 1.46E-06, VIP = 0.986) compared to the control
   group. Interestingly, in the combination group (Dox+PEDF), a slight
   increase in aspartic acid levels was observed compared to the Dox
   treatment alone, suggesting a potential modulatory effect of the
   combined treatment on aspartic acid metabolism. Additionally, amino
   acids such as glutamic acid and proline showed significant reductions,
   particularly in the Dox+PEDF group, indicating that amino acid
   metabolism is extensively affected, possibly influencing cell growth
   and apoptosis. Myo-inositol levels decreased across all three treatment
   groups (PEDF, Dox, and Dox+PEDF) compared to the control (q = 2.30E-06,
   VIP = 0.985), with a further reduction observed in the combination
   group. Lipid-related metabolites, including phosphoethanolamine,
   palmitic acid, and myristic acid, were significantly altered and
   responded differently to the treatment groups. On the other hand, the
   treatments significantly impacted cholesterol levels (q = 0.0003, VIP =
   1.005), particularly in the Dox and Dox+PEDF groups.

3.3. Response of MDA-MB-231 Metabolomes to PEDF, Dox, and Dox+PEDF

   The same analysis was performed for MDA-MB-231 to find substantially
   altered metabolites among groups. Of 50 metabolites, 23 were
   significant in MDA-MB-231 (Table 2[66](Tab. 2)). Among all
   significantly altered metabolites in MDA-MB-231 cells,
   glucose-6-phosphate (G6P) exhibited the most considerable variation
   across treatments. There was a marked reduction in G6P levels in the
   PEDF group compared to control, Dox, and Dox+PEDF treatments (q =
   1.22E-09, VIP = 1.379). Dox exposure resulted in a slight increase in
   G6P levels, whereas the combination of Dox+PEDF led to a significantly
   greater reduction in G6P levels than Dox alone. Lactate levels were
   also significantly altered across the treatment groups (q = 8.42E-05,
   VIP = 1.378). PEDF treatment resulted in the lowest lactate levels,
   while Dox treatment significantly increased lactate levels compared to
   the control. Interestingly, the combination of Dox + PEDF caused a
   reduction in lactate levels compared to Dox alone; however, this
   reduction did not reach statistical significance. For inosine and
   ribulose-5-phosphate, the most pronounced increases were observed in
   the Dox+PEDF group. PEDF treatment alone also significantly elevated
   these metabolites compared to the control. Still, the combination
   treatment exerted a more robust effect compared to PEDF alone,
   suggesting a synergistic interaction between Dox and PEDF on nucleotide
   metabolism and oxidative stress pathways. Similar trends were observed
   for the amino acids glycine and lysine, which significantly increased
   following PEDF treatment compared to the control. However, Dox
   treatment reduced the levels of these amino acids. Importantly, the
   combination of Dox+PEDF led to significantly higher levels of glycine
   and lysine than Dox alone, suggesting that the combination therapy
   exerts an additive effect on amino acid metabolism. Glutamic acid
   levels were significantly elevated across the PEDF, Dox, and Dox+PEDF
   groups compared to the control (q = 0.0006, VIP = 1.018). The Dox+PEDF
   combination induced the highest glutamic acid levels, which were
   significantly greater than those observed with Dox alone. This
   highlights the synergistic effect of the combination treatment on
   glutamic acid accumulation.

Table 2. Significant alterations in MDA-MB-231 metabolites.

   [67]Table 2
   [68]Open in a new tab

   Both stearic acid (q = 0.0007, VIP = 1.021) and palmitic acid (q =
   0.0140, VIP = 0.963) were significantly elevated in all treatment
   groups compared to control, with the Dox+PEDF combination showing the
   most evident effect. The levels of these fatty acids were significantly
   higher in the combination group compared to either Dox or PEDF alone.
   On the other hand, cholesterol levels were significantly reduced in the
   Dox group compared to the control. However, the Dox+PEDF combination
   partially restored cholesterol levels, although they remained below
   those of the control group. This partial recovery significantly
   differed from the Dox group alone, suggesting that PEDF may mitigate
   Dox-induced lipid dysregulation. Box plots were created for the most
   statistically significant metabolites to illustrate the group-specific
   variations in metabolite abundance, as depicted in Figure 4[69](Fig.
   4).

Figure 4. Box plots for significantly altered metabolites in MDA-MB-231. The
boxplots display the distribution of 23 metabolites in MDA-MB-231 cells that
exhibited the highest significance (p-values < 0.05 and VIP scores > 0.8) in
the analysis of variance. These boxplots allow for comparing the four groups:
Control, PEDF, Dox, and Dox+PEDF. On the x-axis, each group is represented by
individual metabolites, while the y-axis indicates the normalised peak
intensity. Metabolites showing significant differences were calculated using
Tukey's Honestly Significant Difference (Tukey HSD) test and indicated as (*)
p ≤ 0.05, (**) p ≤ 0.01, and (***) p ≤ 0.001.

   [70]Figure 4
   [71]Open in a new tab

3.4. Significant Metabolite Correlations in MCF-7 and MDA-MB-231 Cells
Following Treatments

   The correlation analysis of MCF-7 and MDA-MB-231 cells treated with
   PEDF, Dox, and the combination of Dox+PEDF revealed a complex metabolic
   interplay characterised by significant positive and negative
   correlations. In MCF-7 cells, 861 metabolic correlations were found
   among treatment comparisons, revealing distinct interaction patterns.
   For the control vs. PEDF group (Figure 5A[72](Fig. 5)), 389 positive
   and 10 negative correlations were identified. The control vs. Dox
   comparison (Figure 5B[73](Fig. 5)) showed 404 positive and 102 negative
   correlations, while the control vs. Dox+PEDF group (Figure 5C[74](Fig.
   5)) exhibited 318 positive and 70 negative correlations.

Figure 5. Figure 5. Metabolite-metabolite correlations in MCF-7 and
MDA-MB-231 cells. A, B, C) Correlation matrices for MCF-7 cells and D, E, F)
correlation matrices for MDA-MB-231 cells comparing Control vs. PEDF, control
vs. Dox, and control vs. Dox+PEDF, wherein non-significant correlations (p >
0.05) are left uncoloured. Significant interactions (*p < 0.05, **p < 0.01,
***p < 0.001) are colour-coded, with positive correlations in red and
negative correlations in blue.

   [75]Figure 5
   [76]Open in a new tab

   In control vs. PEDF analysis, strong positive correlations were
   observed, including cholesterol with aminomalonate (r = 0.743),
   cholesterol with myo-inositol (r = 0.703), and fumaric acid with
   myo-inositol (r = 0.853). Notable negative correlations included
   1-monostearin with cysteine (r = −0.723) and glycerol with glycine (r =
   −0.817). For the control vs. Dox group, prominent positive correlations
   were identified, such as cholesterol with aminomalonate (r = 0.743),
   cholesterol with oleic acid (r = 0.759), and fumaric acid with succinic
   acid (r = 0.837). Significant negative correlations included isoleucine
   with glycerol (r = −0.667) and glycine with glycerol (r = −0.817). In
   the control vs. Dox+PEDF comparison, key positive correlations were
   found between cholesterol and beta-alanine (r = 0.671) and cholesterol
   with myo-inositol (r = 0.703). Strong negative correlations were
   observed between glycine and glycerol (r = −0.817) and between uracil
   and 1-monostearin (r = −0.689).

   In MDA-MB-231 cells, 1,225 metabolic correlations were identified
   across treatment groups. The control vs. PEDF group revealed 359
   positive and 84 negative correlations (Figure 5D[77](Fig. 5)).
   Noteworthy positive correlations included those between cholesterol and
   cysteine (r = 0.802), cholesterol and pyruvic acid (r = 0.846), and
   succinic acid and phosphoethanolamine (r = 0.817). Prominent negative
   correlations were observed between cholesterol and leucine (r =
   −0.844), 3-hydroxybutyric acid and oleic acid (r = −0.849), and cystine
   and glycine (r = −0.884). The control vs. Dox comparison showed 184
   positive and 86 negative correlations (Figure 5E[78](Fig. 5)).
   Significant positive correlations included oxoproline with glutamate (r
   = 0.961), oxoproline with glycerol (r = 0.919), and 1-monopalmitin with
   arachidonic acid (r = 0.776). Strong negative correlations featured
   glucose-6-phosphate with proline (r = −0.793), glycine with glycerol (r
   = −0.750), and glutamic acid with lactic acid (r = −0.777). For the
   Control vs. Dox+PEDF group, 348 positive and 208 negative correlations
   were identified (Figure 5F[79](Fig. 5)). Key positive correlations
   included 1-monopalmitin with ribose (r = 0.756), cholesterol with
   glycine (r = 0.638), and oxoproline with proline (r = 0.884). Notable
   negative correlations were cystine with glycine (r = −0.884),
   3-hydroxybutyric acid with 1-monopalmitin (r = −0.722), and succinic
   acid with leucine (r = −0.750).

3.5. Pathway Analysis for MCF-7 and MDA-MB-231 Cells Following Treatments

   Pathway analyses of significant metabolites across treatment groups in
   each cell line were conducted using MetaboAnalyst (v.6.0). All enriched
   pathways can be found in Tables S4 to S9. The pathway analysis revealed
   significant metabolic alterations in MCF-7 and MDA-MB-231 cell lines
   across various treatment conditions. In the MCF-7 cell line, treatment
   with PEDF (Figure 6A[80](Fig. 6)) primarily influenced
   phosphatidylethanolamine and phosphatidylcholine biosynthesis pathways.
   In contrast, Dox treatment (Figure 6B[81](Fig. 6)) elicited marked
   enrichment in pathways related to amino acid metabolism, including the
   urea cycle, beta-alanine metabolism, and arginine and proline
   metabolism, emphasising significant metabolic perturbations. The
   combination of PEDF and Dox (Figure 6C[82](Fig. 6)) demonstrated an
   even broader range of enriched pathways, with notable involvement of
   arginine and proline metabolism, the urea cycle, and glutamate
   metabolism, suggesting a potential synergistic or additive effect that
   extends the metabolic impact observed with individual treatments.

Figure 6. Figure 6. Metabolic Pathway Enrichment Analysis in MCF-7 and
MDA-MB-231 Cells. The figure shows metabolic pathway enrichment analysis for
MCF-7 (A, B, C) and MDA-MB-231 (C, E, F) cells under different treatments
(Control vs. PEDF, Dox, and Dox+PEDF). Bubble size reflects the enrichment
ratio, and bubble colour represents p-values, ranging from yellow (less
significant) to red (more significant). The largest circles indicate the
highest enrichment ratios, highlighting the key metabolic pathways that are
significantly impacted by each treatment.

   [83]Figure 6
   [84]Open in a new tab

   In the MDA-MB-231 cell line, PEDF treatment (Figure 6D[85](Fig. 6))
   prominently enriched pathways such as glutathione metabolism, the
   Warburg effect, and the malate-aspartate shuttle, underscoring
   significant metabolic reprogramming in this more aggressive phenotype.
   Dox treatment (Figure 6E[86](Fig. 6)) showed pronounced enrichment in
   pathways, including carnitine metabolism, glutamate metabolism, ammonia
   recycling, arginine and proline metabolism, and bile acid biosynthesis.
   Notably, the combination of Dox and PEDF (Figure 5F[87](Fig. 5))
   yielded the most extensive enrichment, particularly in glutamate
   metabolism, glutathione metabolism, and glycine and serine metabolism,
   reflecting the greater impact on these pathways under combination
   therapy.

4. Discussion

   Breast cancer treatment is often constrained by the side effects
   associated with Dox, a widely used chemotherapeutic agent, which can
   compromise therapeutic efficacy (Belger et al., 2024[[88]11]). To
   address this limitation, combination therapies are being explored to
   enhance treatment outcomes while mitigating adverse effects (Jiang et
   al., 2017[[89]23], Abooshahab et al., 2024[[90]1]). PEDF, a serine
   protease inhibitor, has demonstrated potential for inhibiting tumour
   growth and metastasis (Abooshahab et al., 2021[[91]2]). Recent studies
   have indicated that PEDF can enhance the efficacy of Dox in BC cells by
   hindering tumour cell migration and impacting signaling pathways
   regarding tumour progression, suggesting a synergistic interaction
   between PEDF and Dox (Jones et al., 2023[[92]24], Abooshahab et al.,
   2024[[93]1]). However, the precise metabolic mechanisms underlying this
   modulation by PEDF remain inadequately understood. Given this,
   GC/MS-based metabolomics was employed to determine the combined effects
   of PEDF and Dox on BC cellular metabolic alterations. Our data revealed
   that alterations in metabolite levels distinctly exhibited unique
   response patterns in one cell line but not in the other. These
   differences underscore the metabolic divergence between the two BC cell
   lines and provide insights into potential metabolic vulnerabilities
   specific to each. Combining Dox and PEDF impacts energy production,
   amino acid metabolism, and antioxidant defences, suggesting a
   multifaceted approach to overcoming cancer cell metabolic resilience.

   Glycolysis is one of the central metabolic pathways in cancer, where
   many aggressive tumours exhibit the "Warburg effect," relying on
   glycolysis for energy, even in the presence of oxygen (Vander Heiden et
   al., 2009[[94]43], Potter et al., 2016[[95]33]). This metabolic shift
   facilitates rapid ATP production and supplies intermediates for
   anabolic pathways crucial for cell growth and proliferation (Potter et
   al., 2016[[96]33]). Lactate, the end product of glycolysis, is also
   generated to create an acidic tumour microenvironment, promoting
   invasion, immune evasion, and metastasis (Sharma et al., 2022[[97]36]).
   Our study showed that Dox treatment led to opposing effects on lactate
   levels in the two cell lines, directly reflecting differences in
   glycolytic activity. In MDA-MB-231 cells, Dox increased lactate
   production, indicating an upregulation of glycolysis. This is a common
   adaptive response in aggressive cancer cells, where glycolysis provides
   not only energy but also a continuous supply of NAD+ through lactate
   dehydrogenase, necessary for maintaining glycolytic flux. This result
   reflects those of Maria et al. (2017[[98]29]), who found that Dox
   increased the single pulse of lactic acid using High-Resolution Magic
   Angle Spinning (^1H HR-MAS) NMR (Maria et al., 2017[[99]29]). In
   contrast, MCF-7 cells exhibited a decrease in lactate production upon
   Dox treatment, indicating a shift away from glycolysis. This finding is
   consistent with Karim et al., who showed that Dox reduced intracellular
   lactate levels (Karim et al., 2024[[100]25]). MCF-7 cells are generally
   less glycolytically active and more oxidative in nature, which may
   render them less capable of switching to glycolysis in response to Dox.
   When PEDF is combined with Dox, both cell lines exhibit reduced lactate
   levels. This also accords with our earlier observations, which showed
   that PEDF reduced lactate levels in BC cells (Abooshahab et al.,
   2023[[101]3][[102]4]). This suggests that PEDF may help normalise
   cancer cell metabolism and potentiate the cytotoxic effects of Dox by
   limiting reliance on glycolysis, a crucial pathway for energy and
   biosynthesis in cancer cells.

   The Pentose Phosphate Pathway (PPP), which branches from glycolysis at
   G6P, is crucial in managing oxidative stress by producing NADPH, a
   reducing equivalent necessary for regenerating antioxidants like
   glutathione (Perl et al., 2011[[103]32], Ge et al., 2020[[104]21]). In
   MDA-MB-231 cells, Dox increased G6P levels, indicating an activation of
   the PPP to counteract the oxidative stress induced by Dox (Songbo et
   al., 2019[[105]39], Shi et al., 2023[[106]37]). The increased PPP
   activity would provide NADPH, which is essential for neutralising
   reactive oxygen species (ROS) and maintaining cellular redox balance
   (Patra et al., 2014[[107]31]). This adaptation highlights the metabolic
   flexibility of MDA-MB-231 cells, enabling them to redirect glucose
   metabolism into the PPP for antioxidant defense rather than energy
   production. Interestingly, when Dox is combined with PEDF, G6P
   intensity was decreased in MDA-MB-231 cells, which supports our
   previous research regarding the effect of PEDF on the levels of G6P in
   this cell line (Abooshahab et al., 2023[[108]3]). This reduction
   suggests that PEDF may inhibit PPP activity, thereby limiting NADPH
   production and, consequently, the cell's capacity to counter oxidative
   stress, which can support the idea that PEDF can inhibit the generation
   of ROS (Yamagishi et al., 2005[[109]44]). G6P was not identified in
   MCF-7 cells, potentially due to reduced dependency on the PPP for redox
   balance.

   Amino acids are building blocks for protein synthesis and contribute to
   biosynthetic and antioxidant pathways in cancer cells (Chen et al.,
   2024[[110]14]). Cysteine is essential for glutathione synthesis, a
   major cellular antioxidant (Chiang et al., 2022[[111]15]). A study by
   Tacar et al. highlighted Dox's role in oxidative stress and its impact
   on cellular components, including thiol-containing molecules like
   cysteine (Tacar et al., 2013[[112]41]). Additionally, research by Mehra
   et al. indicates that Dox-induced cardiomyopathy is associated with
   increased activity of cysteine proteases, suggesting alterations in
   cysteine metabolism (Mehra et al., 2017[[113]30]). Our findings align
   with those observations, showing that in MDA-MB-231, Dox decreases
   cysteine intensity while increasing cystine in MCF-7, reflecting its
   consumption to combat ROS. The further depletion of cysteine with
   Dox+PEDF suggests PEDF may impair cysteine uptake or recycling,
   potentially limiting glutathione synthesis and weakening cellular
   antioxidant defences, thereby increasing susceptibility to apoptosis.
   The decrease in cysteine levels in MDA-MB-231 and the slight decrease
   in cystine levels in MCF-7 after combination exposure further
   corroborate the idea that a strong link exists between the perturbation
   in GSH metabolism and PEDF functions that may influence Dox activity in
   this area as well.

   Other amino acids, including asparagine, aspartic acid, glutamic acid,
   glycine, lysine, and proline, display differential responses to the
   treatments, suggesting disruptions in amino acid metabolism and its
   link to biosynthetic and redox homeostasis. Glutamic acid decreases
   significantly with Dox in both cell lines, especially in MDA-MB-231
   cells. Glutamic acid is a critical amino acid for maintaining cellular
   redox balance, as it serves as a precursor for glutathione synthesis
   (Altman et al., 2016[[114]8], Yoo et al., 2020[[115]46]). The reduction
   of glutamic acid suggests an increased demand for glutathione synthesis
   to combat Dox-induced oxidative stress. However, with PEDF's presence,
   glutamic acid levels increased in MDA-MB-231 cells but remained low in
   MCF-7. PEDF's contrasting effects on glutamic acid levels between the
   two cell lines may indicate differences in how each cell line
   metabolises glutamate or regulates oxidative stress. In MDA-MB-231
   cells, PEDF might protect against oxidative stress differently, perhaps
   by directly modulating ROS production, thereby lessening the need for
   glutathione synthesis. However, in MCF-7 cells, PEDF's effect could be
   less supportive in this way.

   The observed decrease in aspartic acid and uracil in MDA-MB-231 cells
   under Dox+PEDF treatment suggests a significant compromise in
   nucleotide synthesis, which is crucial for DNA repair and replication.
   Given Dox's mechanism of inducing DNA damage (Tacar et al.,
   2013[[116]41]), disruption in nucleotide availability likely
   exacerbates the cytotoxic effects as cells struggle to repair damaged
   DNA effectively. In contrast, MCF-7 cells showed a reduction in
   asparagine, along with increased levels of aspartic acid and uracil,
   indicating a distinct impact on nucleotide synthesis in this cell line.
   However, since MCF-7 cells have a slower proliferation rate than
   MDA-MB-231 cells (Theodossiou et al., 2019[[117]42]), this disruption
   in nucleotide synthesis may have less immediate consequences.
   Notwithstanding, limited nucleotide availability under Dox+PEDF
   treatment restricts MCF-7 cells' capacity for DNA repair, thereby
   enhancing Dox's cytotoxicity by further weakening cellular defences
   against DNA damage.

   The regulation of lipid metabolism observed in both cell lines is
   noteworthy. Lipid metabolism, particularly fatty acid synthesis and
   cholesterol metabolism, is critical for rapidly-dividing cancer cells
   (Snaebjornsson et al., 2020[[118]38], Fu et al., 2021[[119]20]). Fatty
   acids are essential for membrane synthesis, energy storage, and
   signaling molecule production (Snaebjornsson et al., 2020[[120]38]).
   Cholesterol plays a crucial role in maintaining membrane fluidity and
   regulating cellular signaling (Fu et al., 2021[[121]20]). In our study,
   Dox+PEDF significantly impacts cholesterol metabolism in both cell
   lines, but with notable differences in scope and effect. Studies have
   demonstrated that Dox can suppress the expression of HMG-CoA reductase
   (HMGCR), a pivotal enzyme in cholesterol biosynthesis. This suppression
   results in reduced cholesterol levels, which is linked to the
   inhibition of the epidermal growth factor receptor (EGFR)-Src signaling
   pathway (Yan et al., 2020[[122]45]). Research indicates that PEDF
   influences lipid metabolism, particularly by promoting lipolysis, the
   breakdown of fats into free fatty acids. This effect is mediated
   through its interaction with adipose triglyceride lipase (ATGL), an
   enzyme crucial for lipid metabolism (Borg et al., 2011[[123]12]).
   However, its direct effect on cholesterol is not fully understood. Both
   PEDF and Dox independently reduced cholesterol levels in both cell
   lines, consistent with previous studies demonstrating their
   cholesterol-lowering effects (Yan et al., 2020[[124]45], Abooshahab et
   al., 2023[[125]3][[126]4]). However, the combination led to a slight
   increase and decrease in MDA-MB-231 and MCF-7 cells, respectively,
   indicating cell line-specific responses to the combined treatment.

5. Conclusions

   This study is a primary step towards establishing a robust platform for
   using in vitro metabolomics to gain a better picture of the biochemical
   alterations resulting from combined PEDF and Dox treatment. The
   combined effects of Dox and PEDF lead to a multifaceted disruption of
   metabolic pathways, synergistically and antagonistically targeting the
   cancer cells' metabolic adaptability. In summary, Dox and PEDF disrupt
   key metabolic pathways essential for cancer cell survival, including
   glycolysis, the PPP, amino acid, and lipid metabolism, underscoring
   their ability to hinder cancer cell energetics, antioxidant defenses,
   and biosynthetic capacity. PEDF was shown to normalise metabolic
   activity by reducing glycolysis reliance in both MDA-MB-231 and MCF-7
   cell lines, thereby limiting lactate production and its associated
   tumour-promoting effects, either alone or in combination with Dox.
   Additionally, PEDF suppressed PPP activity by reducing G6P levels,
   impairing NADPH production and cellular antioxidant defenses,
   especially in MDA-MB-231 cells. This metabolic alteration enhances
   oxidative stress and reinforces Dox's cytotoxic effects.

   The combination treatment also impacted amino acid metabolism, with
   significant depletion of cysteine, which weakens glutathione synthesis
   and antioxidant capacity, and may lead to heightened susceptibility to
   apoptosis. Furthermore, disruptions in glutamic acid, aspartic acid,
   and nucleotide synthesis pathways suggest impaired DNA repair and redox
   balance, contributing to the synergistic cytotoxicity of PEDF and Dox.
   The modulation of lipid metabolism, particularly cholesterol synthesis,
   by combination therapy further emphasises its potential to target key
   biosynthetic pathways critical for cancer cell proliferation and
   survival. Overall, the synergistic interaction between PEDF and Dox
   offers a promising therapeutic strategy that targets multiple metabolic
   vulnerabilities in BC cells. However, the distinct responses in
   different cell lines underscore the importance of tailoring treatments
   based on tumour-specific metabolic profiles for optimal efficacy.
   Further studies are warranted to investigate the translational
   potential of these findings, particularly in vivo settings and targeted
   analysis, to confirm the therapeutic relevance of PEDF and Dox
   combination therapy in BC treatment.

Declaration

Author Contributions:

   Conceptualisation, CRD.; methodology, RA.; formal analysis, RA.;
   writing-original draft preparation, RA.; writing-review and editing,
   CRD, HAS.; supervision, CRD.; project administration CRD. All authors
   have read and agreed to the published version of the manuscript.

Funding:

   This research received no external funding.

Informed Consent Statement:

   Not applicable.

Acknowledgments: