Cell Metabolomics to Guide the Design of Metal-Based Compounds

Abstract

Despite the increasing interest in the development of novel metal-based compounds for cancer treatment, these molecules are currently poorly characterized in mechanistic terms, due to their multiple macromolecular targets inside the cells. In this review, we show how ¹H NMR metabolomics provides a powerful tool to investigate the metabolic perturbations induced by metal-compounds in cells. The chemical identity and concentration of metabolites detected in cell lysates and their respective growth media by NMR can be viewed as a global fingerprint that describes the response to drug treatment. In this framework, the applications of NMR-based metabolomics to study cellular effects induced by the treatment of cells with anticancer metal-based compounds are comprehensively reviewed.

1. NMR-Based Cell Metabolomics

Metal-based compounds constitute a variegate family of very promising drugs for cancer treatment [1,2]. Despite the huge interest, their intracellular modes of action and targets remain to be elucidated to a large extent. In fact, the nature of the metal ion, its oxidation state, along with the nature of its ligands, all contribute to the cellular reactivity of the complex, modulating the affinity for different intracellular targets as well as its ability to cross biological membranes. Thus, it is not easy to predict a priori the cellular behavior of these complexes.

In this context, NMR-based metabolomics of cell cultures constitutes an ideal tool to gain information about the cellular modes of action of the different metal complexes in an accessible, manageable, and practical biological setup [3,4]. Metabolomics is a methodology focused on the analysis of all the metabolites (i.e., small molecules with a molecular weight <2 kDa, according to the definition of the Human Metabolome Database [5]) that are present in a biological sample. This technique has the ability to bring to light variations in the levels of intra- and extra-cellular metabolites induced by drug treatments and these changes can be directly related to the up- or down-regulation of specific pathways involved in the drug’s modes of action [3,6,7,8,9]. This methodological approach also offers the possibility to identify specific markers of tumor responsiveness or development of resistance to anticancer metal complexes.

Among all the NMR active nuclei, ¹H is doubtless the most used for metabolomic applications. The overall ¹H NMR spectral profile of cells constitutes an untargeted fingerprint of the NMR-detectable portion of their whole metabolome. Each type of cell line is characterized by a typical NMR fingerprint that can be considered as a dynamic mirror of its metabolic profile. The metabolomic fingerprint contains information on the number and relative concentration of the most abundant (>1 µM) metabolites that take part in the main metabolic pathways of the cell, i.e., a pool of 30–40 metabolites including amino acids, nucleotides, sugars, organic acids, etc. It is important to note that with NMR, the only metabolites that are detectable are (i) the metabolites that do not interact with macromolecules present in solution (such as proteins) or (ii) the fractions of the interacting metabolites that are free in solution, in exchange with the bound fractions; the bound fraction is, instead, invisible [10,11].

In Figure 1 the metabolomic fingerprints of the A2780 ovarian cancer cell line, GL261 glioblastoma cell line, and DAOY medulloblastoma cell line are reported as an example.

Since a substantial fraction of metabolites of the central metabolism are phosphorylated (nucleotides, intermediates of glycolysis, the pentose phosphate pathway, etc.), ³¹P NMR can be used to expand the coverage of the metabolome but also to monitor lipids (phospholipids, phosphocholines, etc.) [12]. ³¹P NMR has also proven to be a powerful, non-perturbing probe of membranes whose composition influences their biophysical properties that, in turn, are important for metal-based drug uptake and resistance development [13].

For the analysis of the cell metabolome (endo-metabolome), a fast quench of the cellular metabolism is required to obtain reliable fingerprints. One of the most common quenching methods consists of the use of cold methanol (approx. −40 to −50 °C). Subsequently, chloroform is added to separate the two phases, i.e., polar and non-polar extracts [6,7,14].

Our lab developed a fast and highly reproducible protocol for the analysis of the endo-metabolome that avoids extraction [3]. The protocol is based on procedures usually adopted in western blot proteomics, where the cells are scraped into phosphate buffer and supplemented with a protease and phosphatase inhibitor cocktail diluted in dimethyl sulfoxide (DMSO) that is used to quench enzymatic reactions and stabilize the cellular metabolism. The cells are then lysed by sonication in ice, centrifuged, and finally the supernatant is recovered for the analysis. This procedure allows the obtainment of cell lysates that are stable on the time scale of the NMR sample acquisition [15]. As a drawback, the components of the cocktail of inhibitors present as free molecules in solution give rise to NMR resonances that may overlap with the NMR signals of the metabolites (see Figure 1).

Additionally, high-resolution magic angle spinning (HR-MAS) NMR offers the possibility of obtaining the metabolomic fingerprint of intact cells, avoiding the step of cell breaking [16]. This allows the analysis of the real-time metabolome of viable cells until degradation processes occur. In Figure 2, the metabolomic fingerprint of intact human endometriotic epithelial 12Z cells is reported as an example. The method has been used in several studies for the characterization of cultured 2D cells [17], and more recently in 3D cultures [18], as they represent in vivo features more accurately than standard 2D systems (see

Table 1. NMR-based metabolomic studies on the cellular effects of metal-based compounds. In the case of comparisons of different compounds, only the structures of metal compounds are reported. m/c = methanol/chloroform; m/c/w = methanol/chloroform/water.

Ref	Compound(s)	Structure(s)	Cell Line(s)	Type of Experiments	Sample Type(s)	Drug Treatment (Concentration)	Exposure Time
Platinum
[19]	Cisplatin		Human liver L02 cells	1H NMR	Cell extracts (m/c/w)	1 nM, 1 µM and 1 mM	24 h
[20]	Cisplatin		Human lung cancer A549 cells	1H HR-MAS	Cell lysates	IC50 at 24 h	12, 18, 24, 36, 48 h
[21]	Cisplatin		Human medulloblastoma DAOY cells Human supratentorial primitive neuroectodermal tumor PFSK-1 cells	1H HR-MAS 1H NMR	Intact cells Lipid droplet extracts (m/c)	10 µM	12, 24, 48 h
[22]	Cisplatin		Human osteosarcoma MG-63 cells	1H HR-MAS	Cell lysates	IC50 at 24 h (30 µM) and 50 µM	12, 18, 24, 48 h
[23]	Cisplatin		Rat glioblastoma BT4C cells Human glioblastoma U87-MG cells Human medulloblastoma DAOY cells Human supratentorial primitive neuroectodermal tumor PFSK-1 cells	1H HR-MAS 1H NMR	Intact cells Cell extracts (m/c)	10 µM	12, 24, 48 h
[24]	Cisplatin		Human ovarian cancer A2780 cells Human cisplatin-resistant ovarian cancer A2780cis cells	1H NMR	Cell lysates Growth media	IC50 at 72 h	24, 48 h
	Carboplatin
	Oxaliplatin
[25]	Cisplatin Biphenyl nicotinamide compound Vinblastine		Human breast cancer MCF-7 cells	1H NMR	Cell lysates	IC50 time not specified	72 h
[26]	Cisplatin Doxorubicin Methotrexate		Human osteosarcoma MG-63 cells	1H HR-MAS	Cell lysates	IC50 at 24 h	12, 24, 36, 48 h
[27]	Doxorubicin Tamoxifen Cisplatin		Human breast cancer MDA-MB-231 cells	1H HR-MAS	Intact cells	Not specified	24 h
[28]	Oxaliplatin Oxaliplatin + vitamin C		Human hepatocellular carcinoma HCC SMMC-7721 cells	1H NMR	Cell extracts (m/c/w)	50 µM oxaliplatin	48 h
[29]	[Pt(O,O′-acac) (γacac)(DMS)]		Human cisplatin-resistant ovarian cancer Skov-3 cells	1H NMR	Cell extracts (m/c/w) Growth media	IC50 at 24 h	6, 12, 24 h
	Cisplatin
[30]	[Pt(ƞ1C2H4OMe) (DMSO) (phen)]+		Human neuroblastoma cancer SH-SY5Y cells	1H NMR	Cell extracts (m/c/w) Growth media	IC50 at 24 h	6, 12, 24 h
	Cisplatin
Gold
[31]	Auranofin		Human ovarian cancer A2780 cells	1H NMR	Cell lysates Growth media	IC50 at 72 h	3, 6, 12, 24 h
[32]	Auranofin conjugated with the nanocarrier ferritin		Human ovarian cancer A2780 cells	1H NMR	Cell lysates Growth media	IC50 at 72 h	24 h
[33]	Au(NHC)Cl		Human ovarian cancer A2780 cells	1H NMR	Cell lysates Growth media	IC50 at 72 h	6, 12, 24 h
	[Au(NHC)2]PF6
Ruthenium
[34]	[(η6-p-MeC6H4iPr)2Ru2(μ-S-p-C6H4tBu)3] Cl(DiRu-1)		Human ovarian cancer A2780 cells Human cisplatin-resistant ovarian cancer A2780cis cells Human embryotic kidney HEK-293 cells	1H HR-MAS	Cell lysates	IC50 time not specified	24 h
[35]	(ŋ6-p-cymene) Ru(Cl2) (PPh2(p-MeBenzoate) NCH2CH2N (pMeBenzoate) (PPh2)(Cl2)Ru(ŋ6-p-cymene) (GA113)		Human melanoma A375 cells	1H NMR	Cell extracts (m/c)	IC50 at 24 h	24 h
[17]	[(p-cymene)6Ru6 (tpt)2(dhnq)3]6+		Human ovarian cancer A2780 cells Human cisplatin-resistant ovarian cancer A2780cis cells Human embryotic kidney HEK-293 cells	1H HR-MAS	Intact cells	IC50 at 96 h	24, 72 h
Other metal-based compounds in comparison to cisplatin
[36]	Cu(II)- Casiopeina Igly		Human breast cancer MDA-MB-231 cells	1H NMR	Cell extracts (acetonitrile/water)	IC50 at 24 h	20 min and 40 min
	Cisplatin
[37]	Pd(II)Spermine		Human osteosarcoma MG-63 cells Human osteoblastic HOb cells	1H HR-MAS	Cell lysates	IC50 at 24 h	12, 24, 36, 48, 72 h
	Cisplatin
[38]	Pd(II)Spermine		Human osteosarcoma MG-63 cells	1H NMR	Cell extracts (m/c/w)	IC50 at 24 h	24, 48, 72 h
	Pt(II)Spermine
	Cisplatin
	Oxaliplatin
[39]	Pd(II)Spermine		Human osteosarcoma MG-63 cells	1H NMR	Lipidic extracts (m/c/w)	IC50 at 24 h	48 h
	Pt(II)Spermine
	Cisplatin
	Oxaliplatin
[40]	Pd(II)Spermine		Breast cancer MDA-MB-231 cell-derived xenograft mouse model	1H NMR	Cell extracts (m/c/w)	Spermine: 5 mg/kg/day Cisplatin: 2 mg/kg/day	5 days post-injection
	Cisplatin
[41]	Pd(II)Spermine		Six-week-old female BALB/cByJ mice	1H NMR	Brain tissue extracts (m/c/w)	Spermine: 3 mg/kg Cisplatin: 3.5 mg/kg	1, 12, 48 h post-injection
	Cisplatin

).

The analysis of the endo-metabolome via cell lysates can be nicely complemented by the analysis of the exo-metabolome, i.e., the pool of metabolites presents in the extracellular medium. As the exo-metabolome is the result of an interchange of metabolites between cells and the culture medium, two groups of molecules can be identified, i.e., the taken-up substrates and the released metabolic products. The identification of the two groups of molecules can be easily achieved by the comparison of the cellular exo-metabolome with the profile of the “blank” growth medium.

The collection and analysis of the exo-metabolome samples is technically simple because it only requires centrifugation to separate culture media and cells [3].

Figure 3 reports the exo-metabolome of the A2780 ovarian cancer cell line growth in standard RPMI160 medium, GL261 glioblastoma cell line growth in DMEM medium, and DAOY medulloblastoma cell line growth in MEM medium.

Highly complementary information can be obtained by a separate analysis of both the exo- and the endo-metabolome, and their integrated analysis provides a very accurate picture of the cell’s metabolic behavior. Figure 4 reports the complete list of detected metabolites, in both cell lysates and growth media of A2780 cell line and their classification based on their chemical structures.

Pathway analysis can be performed using the list of metabolites altered by treatment with drugs, with the final aim of obtaining a mechanistic explanation of the changes observed, thus strengthening the information generated by the NMR analysis [43]. Different cell lines share the same main metabolic pathways (although with different weight of each of them on the overall metabolome), but the treatment with different compounds with different mechanisms of action will affect different biochemical processes. Online biological databases, such as KEGG (Kyoto Encyclopedia of Genes and Genomes) Pathway Database [44] or SMPD (Small Molecules Pathway Database) [45], provide information on a large number of metabolic pathways and can be easily consulted to identify and visualize how each metabolite is involved in the different biochemical pathways. Figure 5 reports the list of the metabolic pathways in which the metabolites identified in the NMR spectra of A2780 cells are involved.

In this framework, the selection of appropriate cell models to obtain biological pictures as similar as possible to the in vivo ones is of extreme importance. The comparison of the metabolomic profiles of patient-derived primary cells with those of established cell lines allows highlighting of the similarity and differences among the different cell types [15,46].

2. NMR Metabolomics and Metal-Based Compounds: State-of-the-Art

The study of the mechanisms of action of metallodrugs is fundamental for the development of more effective therapeutic strategies. As previously discussed, the detection of metabolic pathways altered upon treatment can be achieved by monitoring the changes over time. Table 1 reports the studies in the literature that have exploited NMR metabolomics to evaluate the metabolic responses of cancer cells to different metal-based drugs.

As expected, Pt(II)-based drugs are the most studied. The effects of cisplatin were characterized, alone or in comparison with other drugs (or drug candidates), in a variety of cancer cells by analyzing whole or lysed cells, respectively, via ¹H HR-MAS or using a solution of ¹H NMR for the analysis of cell lysates, extracts, and their cultured media. Despite the great variability of the observed cellular responses, mainly due to different cancer cell lines used and the largely different conditions of the various experiments, these studies revealed some common trends in the induced metabolic changes such as strong effects on lipid metabolism, membrane composition, a decrease in the overall cellular metabolic activities, and the alteration in the levels of some other diagnostic metabolites like the UDP-N-acetyl-glucosamine.

To our knowledge, only two metabolomic studies compared the metabolic effects induced by the three clinically approved Pt(II)-based drugs. The group of Ana Gill compared the effects of cisplatin and oxaliplatin in human osteosarcoma cells [39], while our research group compared the effects of cisplatin, carboplatin, and oxaliplatin in human ovarian cancer cells [24]. Interestingly, the presence of different Pt(II)-based ligands in the square planar geometry of these molecules does not significantly affect the patterns of metabolomic alterations, but only seems to modulate the extent of the changes. In both cases, cisplatin produced the strongest effects.

The numerous side effects and the emergence of resistance to the clinically approved Pt(II)-based drugs have encouraged further development of platinum compounds as well as of other metal-based compounds as anticancer drug candidates. The cellular effects of two novel Pt(II)-based complexes were evaluated using ¹H NMR metabolomic profiling of cell extracts and cultured media. The [Pt(O,O′-acac)(γ-acac)(DMS)] complex showed very different metabolic changes with respect to cisplatin in the SKOV-3 cell line; in particular, higher levels of pyruvate were observed in addition to a very different lipid expression [29]. The mechanism of action of [Pt(ƞ¹-C₂H₄OMe)(DMSO)(phen)]⁺ was instead investigated on SH-SY5Y cells at different time points. The NMR data revealed a faster action of the new complex compared with cisplatin, with a response already observed after six hours of exposure, suggesting a cytosolic target with peculiar alteration of the glutathione metabolism pathway and of the diacylglycerol expression [30].

Due to the high similarity between Pt(II) and Pd(II), palladium chelates are among the molecules that have drawn increasing interest. In particular, the cellular effects of Pd(II)–spermine and Pt(II)–spermine complexes have been extensively characterized via ¹H NMR metabolomics in osteosarcoma cells, also in comparison to cisplatin and oxaliplatin [37,38,39,41]. In this case, it is the different metal center that modulates the extent of the typical changes in the cell metabolome. Treatment with Pd(II)–spermine showed weaker changes at the level of the metabolites but higher capability to modulate lipids and membrane compositions, in comparison to its Pt analogue.

To our knowledge, no NMR metabolomics studies were published on the cellular effects of the Ru-compounds that have been entered into the clinical trials (such as NAMI-A and NKP1339). Julier Furrer and collaborators tested two novel Ru-complexes (i.e., p-cymene)₆Ru₆(tpt)₂(dhnq)₃]⁶⁺ and DiRu-1) on both cisplatin-sensitive and -resistant A2780 ovarian cancer cells and on human embryotic kidney HEK-293 cells via HR-MAS NMR analysis. While different metabolic responses were detected depending on the cell type and incubation time for (p-cymene)₆Ru₆(tpt)₂(dhnq)₃]⁶⁺ prims [17], the DiRu-1 complex showed an overall action on the redox homeostasis, the Warburg effect, and on lipid metabolism [34]. The third and latter tested Ru-compound via NMR metabolomics is GA113. The complex was tested for its anti-cancerous potential against a human malignant melanoma A375 cell line showing a high cytotoxicity. The six metabolites that were found to be altered after treatment are involved in (i) alanine, aspartate, and glutamate metabolic pathway, and (ii) glycine, serine, and threonine pathway [35].

Gold-based compounds represent another very interesting class of promising anticancer compounds [47]. Unlike Pt(II) compounds, gold(I) complexes have very low affinity for DNA, but they have very high affinity for cellular proteins with solvent-exposed cysteines or selenocysteines. Thus, due to the intrinsic nature of these metal centers, these molecules are supposed to give rise to multiple intracellular interactions with many functional proteins, rather than with a single enzyme or protein. In this context, our research group is carrying out a comparative NMR metabolomic study to analyze the responses of A2780 human ovarian cancer cells to a panel of selected gold(I) compounds, previously characterized with good anticancer properties. The obtained results for auranofin [31] and two carbene molecules [33] were published in 2020 and 2023, respectively. Interestingly, the treatment with the three compounds affects the metabolome of A2780 cells in different ways. Thus, contrary to what was observed for Pt(II)-based drugs, in this case it is the nature of the different gold(I) ligands that can modulate the reactivity of the compounds and also their intracellular targets resulting in substantially different modes of actions. For auranofin, both when administered alone [31] or conjugated with ferritin (to improve its targeted delivery towards cancer cells) [32], the most striking intracellular alteration was a very large and early increase of glutathione. The gold monocarbene and dicarbene complexes behave differently from auranofin but display many alterations in common between the two of them; however, the latter amplifies the effects. Glycolysis is significantly affected and both drugs also caused a reduced uptake of several amino acids [33].

NMR metabolomics can also be applied in the study of drug treatment resistance. Again, almost all the examples present in the literature deal with cisplatin resistance. These works compared the metabolic profiles of the parental cisplatin-sensitive (i) ovarian cancer A2780 [15,24], (ii) breast cancer MDA-MB-231 [48], and (iii) lung adenocarcinoma A24 cell lines [49] with those of the respective derived cisplatin-resistant lines. The paired sensitive and resistant cell lines show significantly different profiles in terms of the concentrations of many metabolites, suggesting that the acquired cisplatin resistance is associated with multiple metabolic alterations. Interestingly, the different resistant cell lines share common changes for taurine, creatine, glutathione, myo-inositol and choline-containing compounds that can be considered putative markers of resistance insurgence. The relationship between the changes in membrane lipids and the cisplatin-resistance has also been studied via ¹H and ³¹P NMR metabolomics in lung adenocarcinoma A549 cells [50]. The results indicated that components and properties of membrane phospholipids of cisplatin-sensitive and -resistant A549 cells (both before and after cisplatin treatment) were significantly different, highlighting the strong correlation between the biophysical properties of the cellular membranes and the development of resistance.

3. Conclusions

In this review, the application of NMR-based metabolomics of cell culture in metal-based drug research have been described.

NMR metabolomics can significantly contribute to the study of the mechanisms of action of drugs as it is an excellent investigative tool to characterize treatment-induced metabolic alterations at the cellular level according to an untargeted approach. This type of approach results are particularly suitable in the case of metallodrugs, which possess multiple and often unidentified intracellular targets.

¹H NMR spectral acquisition is relatively fast. Each cell lysate spectrum requires 30–60 min on average depending on the quantity of cells; only 5–15 minutes are instead required for the analysis of growth media. This permits to test multiple treatment conditions. In particular, studying different treatment times can allow us to detect both early and late cellular drug-induced events. Nevertheless, the reliability and reproducibility of the obtained results strongly depends upon the use of established procedures that avoid undesired changes in the metabolome induced by post-culture procedures.