Metal Species in Biology: Bottom-Up and Top-Down LC Approaches in Applied Toxicological Research

ion of Cd from plasma proteins by chelating agents [36].Thus, the application of “bottom-up” and/or “top-down” LC approaches each offers unique, albeit different insights into themetabolism of toxicmetals (Figure 4).The combined application of these techniques in conjunction with other techniques can contribute to better understand the interplay of toxic metals with endogenous ligands in mammals. Considering that the exposure of various populations to toxic metals is of increasing societal concern [206–214], LC approaches can be applied to develop practical solutions to address toxic metal-related problems in the near future. Another societal issue that is indirectly related tometals is the significant decline in the number of therapeuticmedicinal drugs that are being approved. This undesirable trend may be reversed if metal-based drugs can be developed that offer improved efficacy and reduced side effects compared to those that are in use today. To this end “bottom-up” LC approaches can provide insight into the binding of a metal-based drug and/or its hydrolysis products to mixtures of key plasma proteins (e.g., HSA, Tf). Results from such studies are useful to predict the absorption of a metal-based drug into the target organ cells via established uptake mechanisms. On the other hand, the addition of a metal-based drug to blood plasma followed by the LC analysis of the obtained mixture allows to simultaneously observe the hydrolysis of the parent drug and the plasma protein binding of the in situ generated hydrolysis products. This approach has considerable potential to advance our understanding of the biochemical fate of established or synthetic metal-based drug candidates in plasma which will contribute to potentially unravel the molecular basis of their side effects. In addition, “top-down” LC approaches are also ideally suited to gain fundamentally new insight into mechanism of action by which ameliorating agents (e.g., NAC) modulate the metabolism of metal-based drugs (e.g., cis-platin) in plasma. LC can therefore play an important role with regard to the development of novel strategies to decrease the adverse health effects ofmetal-based drugs while leaving their antitumor effect largely intact [37].


Introduction
Based on his work on plant pigments, M.S. Tswett serendipitously invented adsorption chromatography as a universal analytical method to separate chemical compounds in 1906 [1]. After an interregnum of ∼40 years, the discovery of liquidliquid partition chromatography by A.J.P. Martin and R.L.M. Synge breathed new life into this ground-breaking separation approach [2] and was-together with the development of the first general theory of chromatography-awarded the Nobel Prize in Chemistry in 1952 [3]. Over the intervening years, liquid chromatography (LC) has matured into one of the most enabling and widely used separation technologies in modern analytical chemistry [4]. This evolution has involved the development of stationary phases with progressively smaller particle diameters [5], the miniaturization of column dimensions [6,7] and resulted in the beginning of high performance liquid chromatography (HPLC) in the early 1970s [8]. The need to further increase the column efficiency required the utilization of even smaller particle diameters and culminated in the introduction of ultra performance liquid chromatography (UPLC) in 2004 [9]. In the context of separation science these LC techniques allowed, for the first time, to analyze complicated biological fluids, such as blood plasma for proteins [10][11][12] or small molecular weight metabolites [13,14] and put analytical chemists in the unique position to tackle biological complexity. Advances in LC technology have therefore played a pivotal role in setting the stage for modern proteomics [15], metabonomics/metabolomics [16][17][18], and systems biology research [19]. Concomitant with these technological developments, a variety of separation mechanisms were established which allowed LC to become an indispensable tool for the purification of active pharmaceutical ingredients (APIs), such as blood coagulation factors from plasma and recombinant proteins from bacterial lysates [20]. Reflecting on this brief historic trajectory of LC and considering the crucial role that this versatile separation technique has served in the context of food analysis and safety [21,22], it is tempting to equate its significance in contemporary life science research with the role that the microscope plays in biology, the spectroscope has in chemistry, and the Hubble telescope holds in astronomy.
Against this background, it is interesting to note that within the specific research area of "metal species in biology"-loosely defined as the elucidation of health relevant interactions between environmentally abundant toxic metals or metal-based drugs with ligands within organismsapproaches which involve the hyphenation of LC separations with inductively coupled plasma-(ICP-) based detectors [23] have also become more frequently applied over the past 10 years [24]. To understand this trend, one may identify three contributing reasons. The first one is the capability of LC-based approaches to provide fundamentally new insight into the complex interactions between metal species and endogenous ligands in a variety of ways [25][26][27][28]. This notion constitutes the main focus of this paper and will be elaborated on in detail later. The second reason is the realization that these interactions provide an important starting point to potentially unravel the mechanisms which link the exposure to certain metal species-unwitting in the context of environmentally abundant toxic metals [29][30][31] and deliberate with regard to intravenously administered metal-based drugswith adverse human health effects [32,33]. Thus, the application of LC-based approaches has potential to provide insight into the etiology of hitherto elusive metal species related disease processes [34,35]. The third reason is the inherent ability that LC-based approaches offer in terms of developing more effective antidotes against specific metal species to ultimately advance human health [36,37]. As a result, the application of LC-based approaches will likely continue to play an important role in "metal species in biology" research in the years to come. Given that several excellent reviews have detailed either their historic development [38] or various recent methodological advances [24,27,28,[39][40][41], the focus of the present one is to illustrate the different ways in which LC has been employed as a tool to gain insight into the complex interactions that unfold after metal species enter mammalian organism. Furthermore, particular emphasis will be given to studies that have attempted to specifically probe certain aspects of these interactions under near physiological conditions. It is hoped that the presented synopsis may contribute to foster progress toward tackling the by far biggest adversary in the context of revealing the bioinorganic processes which link the exposure of mammals to certain metal species with disease processes: biological complexity [19].
Within the scope of elucidating the role that metals play in health and disease, one may differentiate between three distinguished categories: essential metals/metalloids, toxic metals/metalloids, and pharmacologically active metal compounds/drugs. Owing to the central role that essential metals play in health [42], extensive research has been and continues to be directed toward the elucidation of the exceedingly complex homeostatic regulation mechanisms which maintain their tissue concentrations within a narrow range throughout life [43][44][45][46][47][48][49]. Comparatively less, however, is known about the role that environmentally abundant toxic metals may play in the etiology of human disease processes [34,35,50,51], and progress towards the development of more effective chelating agents to efflux these from intoxicated organisms (without harming them) is slow [52]. In a similar vein, the biomolecular basis of the side effects that are associated with clinically used metal-based drugs (and often limit the treatment of patients) remains fragmentary [53][54][55], and surprisingly little is known about the biochemical mechanism(s) by which so-called "ameliorating agents" mitigate the side-effects of metal-based drugs in patients [37,56,57].
From a conceptual point of view, the effect that any metal species exerts in an organism above a minimum dose and length of exposure-this includes acute toxicity and chronic toxicity-may be investigated by two principle approaches. The traditional "bottom-up" or "reductionistic" approach is based on the isolation of the toxicological problem from the organism followed by in depth studies of the presumed underlying biochemical mechanism(s) using appropriate techniques. For example, to explain why the ingestion of an improperly prepared puffer fish is fatal in ∼60% of the reported poisoning cases (lethal dose in adults: 10 g/kg body wt), one may want to uncover the molecular mechanism of action of the responsible toxin, tetrodotoxin. To do so, however, one needs to identify first a putative biological target molecule. Since early studies indicated that tetrodotoxin may adversely affect voltage-sensitive Na + channels [58], definitive proof of this hypothesis required detailed studies into the binding of tetrodotoxin to this putative Na + channel protein target [59]. Importantly, this approach can only be employed if some a priori information about a likely molecular mechanism of action (e.g., inhibition of voltagesensitive Na + channels) is available. In contrast, the "topdown" or "systems biology" approach attempts to directly analyze complex biological samples with an appropriate analytical technique to unravel the molecular mechanism of action of a particular toxin therein [19,60]. This alternative approach hinges on identifying an analytical technique that is capable of detecting a toxin-induced change in a likely biological compartment (e.g., blood plasma [26,61]) or to infer that a relevant biotransformation of the toxin of interest has occurred in a particular organ (e.g., the liver) by analyzing an appropriate effluent stream (e.g., bile [62][63][64][65]). Importantly, the "top-down" approach offers the advantage that no a priori information about a likely mechanism of toxicity is required and that fundamentally new insights into hitherto unknown mechanisms of action may be obtained. Even though the "bottom-up" approach has been in use longer than the "top-down" approach, the latter is receiving progressively more attention lately. This may be attributed to the fact that organisms are comprised of diverse anatomical structures which maintain intricately intertwined metabolic networks ("the sum is more than its parts") [66], whose perturbation by a toxin requires appropriate visualization tools [67,68]. Despite this trend, it needs to be stressed that it does not matter in principle whether one starts "bottom-up" or "topdown" to solve a particular toxicological problem at hand, since the overall goal is to obtain new insight by whatever means [69]. Ultimately, the information that is obtained by either approach should corroborate that of the other to validate a novel toxicological mechanism of action. Based on the overall focus of this paper to stress the unique niche that LC-based methodologies occupy in "metal species in biology" research, it is divided into four parts. The first one provides an overview of "bottom-up" LC approaches which have provided insight into the reaction of toxic metals with relevant endogenous ligands. Part two summarizes the results of "bottom-up" LC approaches which have contributed to better understand the interaction of metal-based drugs with individual ligands. Part three reviews "top-down" LC approaches which have been applied to study the interaction of toxic metals with blood plasma and their abstraction from proteins by chelating agents. The last part aims to highlight investigations that have employed "topdown" LC approaches to investigate the metabolism of metalbased drugs in blood plasma and/or cells. Owing to this rather broad overall scope, a few select examples which illustrate in what way each LC approach has allowed insight will be presented. To further stress the role that LC played in these studies, the element-specific detector(s) that were employed will only be identified if it was deemed to be useful for the reader.

Interaction of Absorbed/Injected Metal Species with Endogenous Ligands
In principle, the toxicity of any metal species may be attributed to interactions with endogenous target ligands that occur either in the bloodstream [70] {in case of a gastrointestinally absorbed metal, one must also consider metal-amino acid complexes, since these may be the species that are absorbed into the bloodstream [71]; with regard to the metal-based drugs cis-platin (Figure 1(j)) or carboplatin (Figure 1(k)), only their hydrolysis product(s) may appreciably interact with endogenous ligands [53]} and/or within organs [72,73]. It is also important to note that the biological response of an organism to any given metal species is dose-dependent ("the dose makes the poison") and that a different set of endogenous ligands appear to be targeted in acute [74] as opposed to chronic As III toxicity [64]. Blood plasma alone may contain up to 10,000 potential protein ligands [75]. Given the large number of possible metal speciesligand interactions and considering that these occur in water [76]-which is considerably more complicated to predict from a thermodynamic point of view than interactions that take place in aprotic solvents [77]-it is unsurprising that they remain rather poorly understood with regard to environmentally abundant toxic metals [35,78] as well as metal-based drugs [79][80][81].
In the bloodstream the ligand(s) which a particular metal species will bind to is determined by (a) its affinity for each available binding site (e.g., on plasma proteins, external membrane receptors, and/or on cell membranes [85]) and (b) the relative concentration of all binding sites [86]. This binding event often involves the coordination of the metal species to nitrogen, oxygen, and sulfur donor atoms on peptides and proteins [87]. In principle, a metal speciesligand interaction may be either reversible (noncovalent bonding), irreversible (covalent bonding [82]) or involve an actual chemical reaction (e.g., cis-platin reduces S-S bonds in human serum albumin, HSA [88]). Reversible interactions between a metal species and endogenous ligands in the bloodstream play a central role in the distribution of the former within the organism, as they will-together with the relative availability of uptake mechanisms into organs [89] -determine the toxicological target organ(s) [35,90]. Irreversible interactions between a metal species and a particular ligand in the bloodstream-which are the result of a chemical reaction [91]-are of particular relevance [32,84,91] as they can exert far-reaching systemic adverse health effects at the organ level if the ligand is an essential ultratrace element, such as Se ( Figure 2, red arrows). Once the metal species and/or its metabolites enter(s) a particular organ cell, processes (a) and (b) will unfold anew within the cytosol, albeit with a different set of proteins and at reducing conditions, since all mammalian cells contain between 0.1-10 mM glutathione (GSH, Figure 1(a)) [92]. The intracellularly formed metalligand complex may then either disrupt vital biochemical processes in the cytosol [93,94] or enter subcellular compartments (e.g., mitochondria, lysosomes, and nuclei [95]) where (a) and (b) will unfold and may result in cell death [96][97][98] ( Figure 2, dashed red arrows).
Although interactions of metal species with endogenous ligands that occur in the bloodstream are in principle as toxicologically relevant as those that unfold within organs [32], the former may be more important than we currently think. This is because irreversible interactions in this central biological compartment may result in systemic toxicity ( Figure 2, red arrows), whereas reversible interactions will be involved in determining the target organ ( Figure 2, dashed red arrows). In human blood plasma, the 20 most abundant plasma proteins-they comprise ∼97% of the total protein [99] and Higher-level triggers of cell signalling Figure 2: Conceptual depiction of the reductionist causal chain to construct an organism (black arrows) as well as the systems biology view of downward causation (blue arrows) adapted from [69]. There are two principle ways by which a toxic metal species that has entered the bloodstream may exert adverse health effects. Red arrows: a metal species interacts with an endogenous target ligand in the bloodstream (e.g., formation of compounds with As-Se and Hg-Se bonds [32,82]; binding of cis-platin derived hydrolysis products with HSA block the Zn 2+ binding site [83] or the displacement of Zn from the extracellular glycoprotein Cu, Zn superoxide dismutase [26]) which may affect cell signalling, gene expression, and/or the distribution of an essential ultratrace element to organs and therefore affect the whole organism [84]. Dashed red arrows: a metal species enters a specific cell type where it interacts with intracellular target ligands (e.g., binding of cis-platin derived hydrolysis products to DNA [55]) to induce apoptosis. Note that the red arrows pertain to all organs within an organism (systemic toxicity), whereas the dashed red arrows correspond to the target organ only (selective toxicity). In the context of developing more targeted metal-based anticancer drugs, it is therefore critical not only to increase their selective toxicity (i.e., their preferential delivery toward cancer cells), but also to minimize/avoid systemic toxicity caused by unintended interactions with endogenous ligands in the bloodstream [83].
almost half of them are Cu, Fe, and Zn-metalloproteins [100] -have so far received most attention as target ligands for metal species. By far the most abundant plasma protein human serum albumin (HSA), for example, has binding sites for Cd 2+ [101], cis-platin derived hydrolysis products [80], and numerous other metal species [102]. Despite the fact that extensive research has been directed to investigate the interaction of individual metal species with individual plasma proteins ("bottom-up" approach), comparatively few studies have attempted to probe the interaction of metal species with more than one ligand (which is the norm in biological systems), with whole plasma/blood in vitro, let alone in vivo ("top-down" approach). A comprehensive understanding of all interactions that unfold in vivo is particularly challenging, since the biotransformation/metabolism of a metal species either in the bloodstream [32,34,35,53,61,84,91] or in ISRN Chromatography 5 the liver (from which metabolites may be expelled back into the bloodstream for urinary excretion) may result in species that could also bind to particular plasma proteins over time. Considering the logistical challenges that are generally associated with the execution of in vivo studies, it is not surprising that our understanding of the metabolism of metal species is incomplete. Our lack of understanding the biomolecular basis of the side-effects of intravenously administered metal-based drugs may be similarly attributed to the comparatively few studies that have been conducted to probe their interactions and that of their hydrolysis products with extracellular (plasma proteins) and intracellular (cytosolic) proteins [103,104]. Ultimately, the outcome of any metal species that has entered a mammalian organism and interacted with a variety of endogenous ligands will be that it is either deposited as an insoluble and possibly nontoxic complex (e.g., in case of mercury or arsenic [105][106][107]) or that the parent metal species together with its metabolites is excreted in urine as has been reported for Pt-based anticancer drugs [108] or in feces as is the case for Hg 2+ , CH 3 Hg + , Cd 2+ , As III and Pb 2+ [109].

Why Use LC to Probe Interactions of Metal Species with Endogenous Ligands?
A large variety of instrumental techniques have been applied to study interactions between individual metal species and endogenous ligands, including heteronuclear NMR spectroscopy [112], electrospray-ionization mass spectrometry (ESI-MS) [113], extended X-ray absorption fine structure spectroscopy (EXAFS) [113][114][115], and electron paramagnetic resonance spectroscopy (EPR) [116]. Since none of the aforementioned techniques can probe the platinum speciation at micromolar concentrations inside cells [117], however, there is a need for methods that are capable of monitoring the reaction of Pt compounds in biological fluids [118]. In this context, one may ask what specific information LC can provide in terms of probing the interaction of a metal species with endogenous ligands? This question will be answered and discussed later (see . In addition, what practical advantages are there in using an LC approach? One advantage of LC approaches is the fact that the required equipment, such as HPLC pumps as well as the ICP-based detectors (needed to monitor metal species in the column effluent) are available in many analytical laboratories. In addition, near physiological conditions during the LC separation can be maintained by using phosphate-buffered saline (PBS)-buffer (note that certain seemingly useful biological buffers must be strictly avoided to preclude the generation of artifacts when blood plasma is analyzed [119]) and thermostating the LC column. With regard to "top-down" LC approacheswhich generally involve the analysis of a complex biological sample (e.g., blood plasma) with an LC separation system that is coupled online to a ICP-detector-the analytical result will be inherently less complex than that of conventional proteomic techniques [120]. This, in turn, greatly facilitates the interpretation of the obtained results [100]. Another advantage of "top-down" LC approaches is that they allow one to rapidly analyze a biological fluid. The plasma metalloproteome, which is defined as all Cu, Fe and Zn metalloproteins that are present in plasma, for example, can be obtained in <25 min [100]. Importantly, this capability opens the door to probe dynamic processes that involve metal species in biological fluids in vitro in almost real time. Given that the analysis of any biological fluid by "top down" LC approaches essentially results in the detection of multiple metal peaks, it is important to point out that the analysis of collected metal-containing fractions with some of the aforementioned techniques (e.g., ESI-MS [119], EXAFS [121], or immunoassays [100]) can provide vital information to structurally characterize a detected novel metal species in order to gain insight into a bioinorganic mechanism of action.

''Bottom-Up'' LC Approaches to Probe the Interaction of Toxic Metals with Endogenous Ligands
The interaction of immobilized metal ions for protein ligands is specifically exploited in the LC-based methodology referred to as immobilized metal affinity chromatography (IMAC) [122]. Since IMAC has been successfully applied to gain insight into the binding of Co 2+ , Ni 2+ , Cu 2+ , and Zn 2+ to serum albumins from different mammalian organisms [123], one may anticipate that IMAC has also been applied to study the interaction of toxic metals with biological ligands. Few studies have been applied to probe the interaction of toxic metals (e.g., Hg 2+ ) with endogenous ligands by this approach [124]. LC-based approaches, however, have been applied to investigate the interaction of toxic metals with relevant endogenous ligands although in a different manner (see below). In terms of choosing relevant endogenous ligands, L-glutathione (GSH) must be considered, as this tripeptide is found in a large variety of mammalian cells (including erythrocytes) at intracellular concentrations up to 10 mM [125]. The vital role that GSH plays in critical biochemical detoxification reactions (e.g., the degradation of H 2 O 2 [126]) and the affinity that it exhibits for several toxic metals based on Pearson's HSAB principle [127] makes it a toxicologically highly relevant target. GSH has, in fact, been referred to as the first line of defense against the acute toxicity of Cd 2+ and Hg 2+ [128,129], and its reaction with numerous toxic metals has therefore been extensively investigated [130]. The interaction of GSH with toxic metals/metalloid compounds, however, is not only relevant with regard to their acute toxicity, but also their chronic toxicity, since the formation of complexes can result in the activation of the metal/metalloid [131]. Since Lcysteine (Cys, Figure 1(b)) is present in hepatocyte cytosol at 0.2-0.5 mM [126] and at 10 M in blood plasma [132], it represents another toxicologically relevant endogenous ligand. Because of its structural analogy to Cys, N-acetyl-Lcysteine (NAC, Figure 1(c)) is another potentially interesting ligand.

Interaction of a Toxic Metal with a Single Ligand.
The interaction of a single toxic metal with GSH has been studied using size-exclusion chromatography (SEC) employing the "retention analysis method" (Figure 3(a)). The latter was • SEC: competitive binding of As(OH) 3 to GSH and DMPS [135] • RP-HPLC: coordination behavior of different species • RP-HPLC: discovery of novel metal-ligand complexes (GS-Hg-Cys) [121] • SEC: confirmation of spectroscopically  originally conceived to study the binding of L-tryptophan to HSA [133,134], which involved the measurement of the retention volume of the former as a function of the HSA concentration in the mobile phase. A progressive decrease in the retention time of L-tryptophan with an increase in the HSA concentration indicated a binding event, since the on-column formed L-tryptophan-HSA complex was excluded from the pores of the stationary phase. Hence, this LC approach was deemed to be potentially useful to investigate the interaction between a toxic metal with an endogenous ligand at near physiological conditions. Conceptually, the dynamic equilibrium between the metal-species (injected) and the endogenous ligand (in the mobile phase) is kept constant throughout the entire chromatographic process (Figure 3(a)), which is closely related to what happens when a toxic metal enters a ligand-laden biological compartment, such as blood plasma or a cell ( Figure 4). Using the aforementioned approach and employing a Sephadex G-10 SEC column and PBS-buffer (pH 7.4) as the mobile phase, the reaction between arsenite [As(OH) 3 ] and GSH (1) was investigated [135]: The injection of 30 g As in form of As(OH) 3 onto this system resulted in the elution of an As peak after the inclusion volume which was indicative of an unknown interaction with the stationary phase backbone. The addition of 0.5 mM GSH to the mobile phase decreased the retention time of the injected As(OH) 3 and was rationalized in terms of the formation of (GS) As complexes ( = 1-3). Applying this LC approach, the effects of pH (range 2-10), column temperature (4, 25 and 37 ∘ C), and the GSH concentration (0.5-7.5 mM) on the formation of (GS) As complexes were investigated. It was found that their formation was facilitated by low temperatures (4 ∘ C), pH 6.0-8.0, and high GSH concentrations (7.5 mM). Simulating the physicochemical conditions of erythrocyte cytosol (∼ 3.0 mM GSH) and hepatocytes (∼7.5 mM GSH) by adding the corresponding GSH concentrations to the mobile phase (pH 7.4) and maintaining the column to 37 ∘ C provided evidence for the formation of (GS) As complexes. To ascertain that the GSH-induced retention shift of the injected As(OH) 3 to shorter retention times was truly caused by the formation of (GS) As complexes, the chelating agent 2,3-dimercapto-1-propanesulfonate (DMPS, Figure 1(h)) was added (1.0 and 2.0 mM) to a 7.5 mM GSH containing mobile phase. As expected, the injection of As(OH) 3  ? Figure 4: Unraveling the metabolism of metal species is a prerequisite to gain deeper insight into the molecular basis of their adverse health effects in mammalian organisms. Toxicologically relevant interactions between metal species and endogenous ligands unfold in blood plasma, erythrocytes and hepatocytes and can be investigated by applying "bottom-up" or "top-down" techniques. Interactions between toxic metals (Hg 2+ , As III , Se IV , CH 3 Hg + ) and small molecular weight ligands (GSH, Cys, NAC) that unfold within erythrocytes and/or hepatocyte cytosol and have been studied with "bottom up" LC approaches ( Figure 3). Interactions between metal species (Cd 2+ , cisPt) and cytosolic and plasma proteins have been studied using "top-down" LC-based approaches. (a) Detection of Cd-metallothionein complexes in liver cytosol [110]. (b) Displacement of Zn 2+ from a plasma metalloprotein by Cd 2+ [26]. (c) Binding of Cd 2+ to plasma proteins [26,36,111]. (d) Abstraction of Cd 2+ from plasma proteins by chelating agents (CA) [36]. (e) Comparative binding of cis-platin/carboplatin to plasma proteins [53]. (f) Modulation of the interaction between cis-platin and plasma proteins by sodium thiosulfate (STS) [37]. Note that the combined application of "bottomup" and "top-down" approaches will be required to better understand the interplay between metal species, small molecular weight ligands and proteins to ultimately unravel their adverse health effects and to develop better mitigation strategies by modulating their metabolism by the administration of chelating agents/ameliorating agents.
the As peak toward the inclusion volume which was rationalized in terms of the stronger binding of As(OH) 3 to the vicinal dithiol groups of DMPS (chelate effect) as compared to GSH (monothiol, Figure 1(a)). Thus, this principle LC approach allowed insight into the dynamic on-column formation of (GS) As species under near physiological conditions. To obtain the same information with other techniques, such as 1 H-NMR can be challenging because of line-broadening phenomena at higher temperatures and a comparative lack of sensitivity [135]. In addition, the chosen LC approach can be ideally employed to study competitive interactions between a toxic metal/metalloid compound and two ligands by dissolving them in the mobile phase (i.e., GSH and DMPS). The disadvantages of this approach are that no complex formation constant can be extracted and that the structural characterization of an on-column formed metal-peptidecomplex by ESI-MS is next to impossible due to the utilization of PBS-buffer and the excess GSH therein. In addition, no information about the specific binding site on the employed ligand can be obtained (this information, however, can be easily obtained by 1 H-NMR spectroscopy). Overall, this LC approach nevertheless provided unique insight into the physicochemical conditions under which the highly toxic metalloid species As(OH) 3 will react with GSH to form labile (GS) As complexes.
Encouraged by this outcome, this LC approach was also applied to study the interaction of other As III compounds,  [136]. Both of these methylated As III compounds are of toxicological significance because their GS-complexes methylarsonous diglutathione [(GS) 2 AsMe] and dimethylarsinous glutathione [GSAsMe 2 ] are believed to play a pivotal role during the methylation of As(OH) 3 in the mammalian liver [137][138][139]. To probe these interactions, a Sephadex G-15 SEC column was used. In order to mimic the physico-chemical conditions that are prevalent in protein-free hepatocyte cytosol [136], the mobile phase was comprised of 10.0 mM GSH in PBSbuffer. Varying the column temperature between 4, 25, and 37 ∘ C, 10 g of As in form of As(OH) 3 , methylarsonous acid [MeAs(OH) 2 ], dimethylarsinous acid [Me 2 AsOH], or arsenobetaine (AsB, Figure 1(i)) were chromatographed separately. AsB eluted in the inclusion volume irrespective of the column temperature, which indicated that it did not form a complex with GSH. At 4 ∘ C, the injection of As(OH) 3 resulted in the elution of (GS) 3 As which was inferred from its elution before a ∼MW 600 GSSG standard. Interestingly, an increase of the column temperature from 4 ∘ C to 37 ∘ C progressively shifted the elution of the injected As(OH) 3 toward the inclusion volume, whereas the retention times of the injected MeAs(OH) 2 and Me 2 AsOH remained essentially unchanged [136]. These findings implied that the As-S bonds in MeAs(GS) 2 and Me 2 AsGS are comparatively more stable than those in (GS) 3 As. This is toxicologically relevant because all of these GS-complexes are substrates for ATP-driven hepatocyte export pumps which are located at the apical and the basolateral side of hepatocytes [138][139][140]. The efflux of GS-conjugates from hepatocytes to the bloodstream, for example, can be mediated by multidrug resistant protein 1 (MRP1) [140,141]. It is therefore conceivable that the more stable species MeAs(GS) 2 and Me 2 AsGS are more efficiently expelled from hepatocytes to the bloodstream via a MRP than the comparatively unstable (GS) 3 As ( Figure 5). If this more efficient efflux of MeAs(GS) 2 and Me 2 AsGS compared to (GS) 3 As by a MRP is experimentally confirmed, it would provide a potential explanation for the evolution of hepatic enzymes that methylate As(OH) 3 . This is because organisms that had developed the capability to methylate As(OH) 3 to MeAs(OH) 2 and Me 2 AsOH in the liver would more efficiently efflux As III via the kidneys [136] and therefore have an evolutionary advantage.
More recently, this LC approach was applied to systematically develop a RP-HPLC separation of mercuric mercury (Hg 2+ ) and methylmercury (CH 3 Hg + ) based on using aqueous mobile phases and GSH or Cys as mobile phase modifiers (Figure 3(a)). Optimal separation was achieved with 10 mM Cys in 50 mM phosphate-buffer (pH 7.5) and using a Gemini RP-HPLC column [143]. During the optimization process, the pH dependent retention behavior of Hg 2+ and CH 3 Hg + was investigated by separately injecting these mercurials using a GSH or Cys-containing mobile phase. Between pH 5 and 8, the formation of complexes between CH 3 Hg + and each thiol was essentially independent of the mobile phase, whereas the formation of complexes between Hg 2+ and GSH was unaffected in the pH range 6.0-8.0, but strikingly As(OH) 3 As(OH) 3 + 3GSH As(GS) 3 CH 3 As(GS) 2
Transport to kidneys and excretion of methylated As V species in urine * * Figure 5: Proposed model which can provide a feasible explanation for the evolution of enzymes which methylate As III in hepatocytes: As(OH) 3 is enzymatically converted ( * ) to the trivalent As compounds MeAs(OH) 2 and Me 2 AsOH, which readily react with cytosolic glutathione (GSH) to MeAs(GS) 2 and Me 2 AsGS. The increased stability of As-S bonds in MeAs(GS) 2 and Me 2 AsGS compared to the much less stable (GS) 3 As will facilitate the efflux of the former species to the bloodstream by MRP export pumps [136]. The translocation of these species to the kidneys and subsequent oxidation will result in the excretion of methylated As V species in urine [142]. different at pH 5.0 and 5.5. The stronger retention of the complexes that were formed between each mercurial and GSH (GS-Hg-SG and GS-Hg-CH 3 ) compared to the structurally related Cys (Cys-Hg-Cys and Cys-Hg-CH 3 ) was attributed to the comparatively greater hydrophobicity of GSH. The observed differences in the hydrophobicity of these metal complexes may be of toxicological significance, as this physicochemical characteristic is likely to play an important role during the intracellular disposition of Hg 2+ and CH 3 Hg + [144].

Interaction of a Toxic Metal with Two Endogenous Ligands.
Most recently, the aforementioned RP-HPLC separation of Hg 2+ and CH 3 Hg + was improved because the on-column formed Hg 2+ complex essentially eluted in the void volume. This improvement was achieved by partially replacing Cys of the mobile phase by the comparatively more hydrophobic NAC (Figures 1(c) and 3(b), respectively). A baseline separation of Hg 2+ and CH 3 Hg + with the former species being retained was achieved with 7.5 mM NAC and 2.5 mM Cys in 50 mM phosphate buffer (pH 7.4) [145]. The separation was attributed to the formation of comparatively more hydrophobic mercurial-complexes with NAC compared to Cys. The observed formation of hydrophobic Hg 2+ -NAC complexes (at physiological pH) is of potential relevance to explain why the administration of rats with NAC effectively protected them against the nephrotoxicity of HgCl 2 [146]. In addition, the formation of a CH 3 Hg-NAC complex at pH 7.4 may be involved in the profound NAC-induced mobilization of CH 3 Hg + from tissue compartments (including the brain!) to urine which was observed in mice [147] and possibly involves the renal organic anion transporter-1 (OAT-1) [148]. Considering that the clearance of Hg 2+ from organs was not accelerated by NAC in mice [147] and taking into account that NAC did not alter the plasma concentration and the urinary excretion of the essential metals Ca, Mg, Fe, Zn, and Cu in humans after the administration of an oral dose of 600 mg/day for 2 weeks [149], this thiol ligand appears to be rather selective for the mobilization of CH 3 Hg + . Since NAC is inexpensive, well tolerated by patients (it is used to treat acetaminophen toxicity), and was orally effective to remove nearly 90% of a CH 3 Hg + dose in mice within 2 days, further investigations should explore the utility of NAC as an antidote against CH 3 Hg + in clinical settings [147,150].
The aforementioned results indicated that this LC approach is amenable to study the interaction between a toxic metal with two ligands. Although this very idea may at first seem irrelevant considering the underlying biological complexity, one must not forget that the individual concentrations of GSH and Cys in hepatocyte cytosol may greatly exceed those of every single cytosolic protein [121]. Hence, interactions of a toxic metal with these thiols may play an important role in their intracellular distribution [72,73]. Such competitive interactions can be investigated with LC by leaving the concentration of ligand A in the mobile phase constant and observing the retention behavior of the metal species of interest as a function of increasing concentrations of ligand B in the mobile phase (Figure 3(b)) [121]. Using a Gemini RP-HPLC column (37 ∘ C) and a mobile phase that was comprised of 5.0 mM GSH in PBS-buffer (pH 7.4), 5 g of Hg 2+ , CH 3 Hg + , or Cd 2+ were separately injected ( Figure 6). A single Cd-species eluted essentially in the void volume ( = 0.1), which was structurally characterized as tetrahedral (GS) 4 Cd by X-ray absorption spectroscopy (XAS). In contrast, a Hg 2+ species-presumably the linear GS-Hg-SGwas somewhat retained on this column ( = 0.9), whereas a CH 3 Hg + species was retained considerably ( = 3.8) (Figure 6). When Cys was incrementally added to the 5.0 mM GSH containing mobile phase (0.5 mM/1.0 mM increments up to 10 mM) and each metal species was injected, the retention time of Cd 2+ remained unchanged. The retention time of the injected CH 3 Hg + continuously decreased and was rationalized in terms of the rapid displacement of GSmoieties from CH 3 Hg + by more hydrophilic Cys-moieties. In contrast, the retention behavior of Hg 2+ displayed a characteristic stepwise decrease of its retention time with gradually increasing Cys concentrations. This behavior was explained in terms of the consecutive displacement of two GS-moieties that were coordinated to Hg 2+ by Cys-moieties. These results revealed fundamental differences with regard to the Hg-S bonding in these mercury complexes and provided direct experimental evidence for the formation of the proposed species GS-Hg-Cys [151] under near physiological conditions.

Interaction of Two or Three Toxic Metals with One Endogenous Ligand.
To elucidate in what way the interplay of toxic metals with endogenous ligands may contribute to the etiology of disease processes [34], one must be aware of the fact that in reality human populations are rarely exposed to a single toxic metal [152]. Furthermore, toxic metals are always ingested together with essential elements and the metabolism of certain combinations of toxic metals and essential elements may therefore be biochemically intertwined and of toxicological significance. Direct experimental evidence in support of this possibility was reported in 1941, when the simultaneous administration of rats with As(OH) 3 and selenite (HSeO 3 − )an essential ultratrace element-resulted in the mutual detoxification of these individually highly toxic metalloid species [153]. Considering that free selenite has been detected in human blood plasma [154], this intriguing trace element antagonism was hypothesized to be implicated in the chronic toxicity of As(OH) 3 . Investigations aimed at uncovering its biomolecular basis revealed that it appeared to be based on the rapid in vivo assembly of the seleno-bis(S-glutathionyl) arsinium ion, (GS) 2 AsSe − (Figure 1(d)) in the bloodstream (in erythrocytes [91]) followed by its translocation to the plasma, the liver and its eventual excretion in bile [62,64]. Since the structure of this species was entirely derived from EXAFS and Raman spectroscopy, its unequivocal confirmation required an alternative technique. Surprisingly, this at first difficult task could be accomplished using an LC-based approach. To synthesize (GS) 2 AsSe − , equimolar As(OH) 3 and HSeO 3 − are reacted at physiological pH with ≥8 moleequivalents GSH as follows: Given that the obtained reaction mixture was supposed to contain (GS) 2 AsSe − and 3 mole equivalents of GSSG ( Figure 1(e)), the only notable difference between these species is their net charge: −3 on (GS) 2 AsSe − and −2 on GSSG (see Figures 1(d) and 1(e)). Thus, a LC-based separation approach that could exploit this charge difference would indirectly confirm the spectroscopically derived structure. It was surmised that the separation of these species might be possible by employing a micellar chromatography approach based on SEC (Figure 3(c)) and the detection of the corresponding species by ICP-AES [155]. A Sephadex G-25 SEC column served as the stationary phase, and 0.1 M Trisbuffer with 50 mM hexadecyltrimethylammonium bromide (HDTAB; critical micellar concentration 0.92 mM, pH 8.0) served as the mobile phase. The analysis of the reaction mixture revealed the elution of a species that contained As, Se, and S, which was reasonably well separated from GSSG. The integration of the obtained S-peaks confirmed that indeed two S-atoms (derived from two GS-entities) coeluted with an As and Se-containing species. Remarkably, the elution of (GS) 2 AsSe − followed by GSSG implied additional negative charge on the former species and thus corroborated the spectroscopically derived solution structure [155]. To gain insight into the biochemical mechanism of formation of (GS) 2 AsSe − , equimolar bis-glutathionylarsenous acid [(GS) 2 AsOH] was reacted with HSe − in aqueous solution.
Analysis of the reaction mixture on a Sephadex G-10 SEC column using 0.1 M Tris-buffer (pH 8.0) as the mobile phase clearly demonstrated that (GS) 2 AsSe − had formed (this was confirmed by XAS). Thus, the investigated reaction mechanism can explain how (GS) 2 AsSe − might be formed in vivo [131]. A related trace element antagonism was discovered in 1967 when the coadministration of rats with HgCl 2 and HSeO 3 − resulted in negligible toxicity, whereas essentially all animals in the HgCl 2 group died [156]. Although this antagonism was later demonstrated to involve the formation of a nontoxic Hg/Se species in the bloodstream [157], the structural characterization of this species remained elusive. To synthesize a model compound that was intended to uncover the molecular basis of this HgCl 2 /HSeO 3 − antagonism, equimolar HgCl 2 /HSeO 3 − was reacted with 5 mole equivalents of GSH in PBS-buffer [61]. The analysis of the obtained black solution by 199 Hg-NMR spectroscopy, 77 Se-NMR spectroscopy, and a variety of mass spectroscopic techniques (e.g., FAB-MS and ESI-MS), however, did not provide any useful information toward the structural characterization of the contained black solution species. It was eventually decided to directly analyze this mixture on a Sephadex G-25 SEC column and detecting Hg, Se, and S by means of an ICP-AES. The analysis of the black solution revealed a single Hg, Se, and S-containing entity which eluted in the void volume followed by GSSG (this indicated that a redox reaction had occurred). Detailed analysis of the obtained S-specific chromatogram implied that 5 GS-moieties were associated with the detected Hg-Se species and EXAFS analysis of the isolated Hg/Se/S-species finally revealed this species as (Hg-Se) 100 (GS) 5 , a mercuric selenide core with ligation to 5 GS-moieties via S-Se bonds. The analysis of blood plasma obtained from New Zealand white rabbits 25 min after they had been injected with HgCl 2 and HSeO 3 − revealed that the Se and Hg near edge XAS spectra most closely resembled those of the synthetic (Hg-Se) 100 (GS) 5 . Thus, conventional SEC (without adding any ligand to the mobile phase buffer) coupled with an ICP-AES played a pivotal role in the elucidation of the structural basis of this mineral antagonism.
Evidence in favor of the existence of an even more fascinating "three-way" antagonism between As III , Se IV , and CH 3 Hg + -three individually highly toxic metal/metalloid species-was first reported based on feeding experiments with Japanese quail in 1982 [158]. In particular, the survival rates of quail which were fed CH 3 Hg + (10 mg Hg kg −1 ) increased from 5 to 75% (after 16 weeks) when selenite (6 mg Se kg −1 ) was included into the CH 3 Hg + laced diet. Quite counterintuitively, the survival rate increased to 100% when As III (15 mg As kg −1 ) was fed together with Se IV and CH 3 Hg + . Taking into consideration that (GS) 2 AsSe − is assembled in erythrocyte lysate [91] and that CH 3 Hg + is known to cross the erythrocyte membrane [159] and given the high affinity of CH 3 Hg + for selenide [160], it was deemed possible that this antagonism might involve the reaction of (GS) 2 AsSe − with CH 3 Hg + within erythrocytes to yield the complex [(GS) 2 AsSeHgCH 3 ]. The addition of CH 3 HgOH to (GS) 2 AsSe − and the analysis of the obtained aqueous solution by EXAFS revealed that the proposed complex had indeed formed [161]. To elucidate if this reaction might also unfold in erythrocytes, rabbit erythrocyte lysate was supplemented with synthetic (GS) 2 AsSe − and equimolar CH 3 HgOH. XAS analysis revealed that 90% [(GS) 2 AsSeHgCH 3 ] was present. To investigate the stability of this toxicologically relevant species, [(GS) 2 AsSeHgCH 3 ] was chromatographed on a Sephadex G-15 SEC column using PBS-buffer as the mobile phase and an ICP-AES as the simultaneous As, Se, S, and Hgspecific detector (Figure 3(c)). The coelution of As, Se and S was indicative of (GS) 2 AsSe − and was followed by a Hgspecies. These findings indicated that [(GS) 2 AsSeHgCH 3 ] had dissociated during the chromatographic process and that the Se-Hg bond was labile when PBS-buffer was used as the mobile phase (this, however, does not necessarily imply that this complex is unstable in vivo). Although the biochemical fate of [(GS) 2 AsSeHgCH 3 ] is currently unknown, these results firmly established the plausibility of its formation under near physiological conditions to explain the remarkable "three-way" antagonism.

''Bottom-Up'' LC Approaches to Probe the Interaction of Metal-Based Drugs with Endogenous Ligands
Although metal-based drugs constitute a minority among the plethora of medicinal drugs that are in use today [162], their fraction appears to be slowly increasing [163,164]. The serendipitously discovered cis-platin, for example, is still one of the world's best selling anticancer drug [165] which is extraordinarily effective to treat certain types of cancers [166,167]. Since only two other Pt-based anticancer drugs, carboplatin (Figure 1(k)) and oxaliplatin (Figure 1(l)), have received worldwide approval for their treatment of human cancer patients [168], only these three Pt-based drugs will be discussed. For a historical perspective of cis-platin and its proposed mode of action the reader is referred to a recent review [165]. In general, comparatively few studies have been reported in which LC has been employed to study the interaction of these Pt-based drugs with either individual plasma proteins, small molecular peptides (GSH), and/or amino acids (Cys) [169,170]. This must be attributed to the fact that this interaction is complicated by the fact that these Pt-drugs produce multiple Pt-containing hydrolysis products [171] which will then bind to plasma proteins [53], thiol-containing biomolecules [170], and/or the pharmacological target DNA [172,173]. Thus, the few studies that have employed LC to gain insight into the interaction of Pt-based anticancer drugs with plasma proteins effectively studied primarily the protein binding of the in situ generated hydrolysis products. These studies involved the incubation of individual Pt-based drugs with individual plasma proteins (e.g., HSA) at physiologically relevant conditions (0.1 M NaCl in pH 7.4 buffer) followed by the subsequent LC-analysis of the reaction mixture at different time intervals. In this manner, the binding of cis-platin to the key plasma protein HSA was investigated [174]. Subsequent studies revealed that both isomers induced the formation of the albumin dimer, with the transisomer exhibiting a much greater tendency to induce protein dimerization [175]. Furthermore, transplatin cleaved the S-S bonds in HSA more readily than cis-platin [175,176]. More recent studies investigated the comparative binding of cis-platin to transferrin (Tf) or HSA. After the incubation of 1 M cis-platin with either apotransferrin (Tf, 3 mg/mL) or HSA (30 mg/mL) in 0.9% NaCl and 10 mM Tris-acetic acid (pH 7.4), subsequent analysis by anion exchange HPLC-ICP-MS (Mono Q) revealed complete binding of cis-platin derived species to HSA in 24 h (Pt/HSA ratio = 4 : 1), but only 50% in case of apo-Tf after 96 h (Pt/apo-Tf = 1 : 1) [177]. The fact that the binding of cis-platin derived hydrolysis products to Tf does not involve the displacement of Fe [178] is of pharmacological relevance, since only the fully Fe 3+ -loaded Tf is capable of binding to the Tf receptor on the surface of cells. These findings strongly suggest that cis-platin derived species may enter cancer cells via receptor-mediated endocytosis of Tf [179]. In addition, the comparative binding of cis-platin and a ruthenium(II)arene complex (RAPTA-T) to HSA and Tf (both holo-Tf and apo-Tf) at different molar ratios has been investigated by SEC-ICP-MS [180]. Importantly, these investigations revealed that cis-platin bound to HSA and Tf to the same extent, whereas RAPTA-T showed a marked preference to bind to Tf. In addition, RAPTA-T showed a markedly higher preference for holo-Tf than apo-Tf. Thus, "bottom up" LC approaches allow to assess the comparative binding of metal-based drugs and/or their hydrolysis products to key plasma proteins under near physiological conditions. The results that are derived from such studies allow to predict whether it is likely that an investigated metal-based drug and or its hydrolysis products may enter cells via established pathways.

''Top-Down'' LC Approaches to Probe the Interaction of Toxic Metals with Endogenous Ligands
Blood represents the principal medium for the transport of gastrointestinally absorbed nutrients including essential metals as well as dietary toxic metals to organs. Interactions between toxic metals and constituents of the bloodstream may adversely affect the transport of sufficient amounts of essential elements to their intended destinations within an organism [26,181]. To this end, the interactions of As(OH) 3 and Hg 2+ with the essential ultratrace element Se represent the first examples for this toxicological mechanism of action ( Figure 2, red arrows) [32,84]. Thus, the bloodstream in itself represents a biological compartment where relevant mechanisms of action may unfold. Accordingly, the detection of inorganic pollutant-induced biochemical changes in plasma or erythrocyte lysate represents a feasible approach to possibly gain insight into the etiology of pollutant induced adverse health effects in mammals. In this context it is important to stress that everyone who embarks on the quest to probe these interactions should be aware of the difficulty that is associated with sampling a living organism without changing the composition of any given sample before or during the analysis [26,182]. In principle, one has to distinguish between perturbations that may be detected within minutes after an appropriate dose of a particular toxin has been added to blood plasma in vitro [26] or administered to animals [183] and perturbations that are only detectable in blood plasma or erythrocyte lysate after a given model organism has been exposed to a particular toxin for weeks or months [184,185]. With regard to the nature of these perturbations, one must distinguish between those that are displayed at the mRNAlevel [186], the small molecule level [13,17], the protein level [12], and those at the metalloprotein level [26,75]. At present, numerous proteomic approaches are available to visualize changes of any given proteome over time, such as increased or decreased concentrations of specific proteins (e.g., in erythrocyte lysate [75]), but many of these techniques are time consuming, expensive, and incapable of detecting changes at the metalloprotein level. Conversely, there are comparatively few LC-based approaches that can be employed to determine the "metalloproteome" of any biological fluid. Fortuitously, however, all of these can accomplish this task in a comparatively rapid manner.
The first notable LC-based approaches that were applied to determine the metalloproteome of biological fluids in a truly "top-down" manner are the pioneering studies of K.T. Suzuki who analyzed rat liver cytosol (SEC-AAS) [110] and blood plasma (SEC-ICP-AES) [111]. After the injection of rats with Cd 2+ (0.4 mg Cd/kg body wt), blood plasma was analyzed (after 1, 2, 3, 5, 10, 20, and 30 min) using a system that was comprised of an Asahipak GS-520 SEC column, 10 mM Tris-HCl buffer in 0.9% NaCl (pH 7.4) as the mobile phase and an ICP-AES to monitor the emission lines of S, Cd, Zn, Cu, Fe, and P [111]. At the 3 min time point, the coelution of a single Cd-peak with Zn and S implied the binding of Cd 2+ to the most abundant plasma protein rat serum albumin (RSA). In addition, these authors detected a comparatively broad Zn-peak (which eluted before Zn-RSA) along with a single Cu-peak (ceruloplasmin), and three unidentified and nonbaseline separated Fe-peaks, as well as two unidentified P-entities (a small one in the high molecular weight and an intense one in the <60 kDa elution range). Interestingly, the analysis of plasma 20 min after the injection revealed that the Cd-peak that was detected after 3 min was bearly detectable, which implied rapid translocation of Cd 2+ from the bloodstream to organs. Concomitantly, the intensity of the broad Zn-peak had increased in intensity compared to that at the 3 min time point. Collectively, these results demonstrated that the analysis of blood plasma by SEC-ICP-AES can provide valuable insight not only into the direct interaction of a toxic metal ion with biological ligands in blood plasma (i.e., the rapid binding of Cd 2+ to RSA and the subsequent translocation of Cd 2+ to organs), but also into changes of the metalloproteome (e.g., the observed changes of the Zn-metalloproteome in response to Cd 2+ ). More recently, the application of the same analytical approach (stationary phase: Superdex 200 SEC column; mobile phase: PBS-buffer [100]) was applied to analyze rabbit plasma to which increasing doses of Cd 2+ (0.5, 2, and 200 g Cd 2+ /mL) had been added in vitro [26]. The results revealed at least two Cd 2+ binding proteins after the addition of the low Cd 2+ doses. At the high Cd 2+ dose, however, an altered Znmetalloproteome pattern was detected which was attributed to a Cd 2+ -induced displacement of Zn 2+ ions from its binding sites on a Zn-metalloprotein-presumably Cu, Zn superoxide dismutase-which has been previously observed [187].
LC-based approaches may not only be applied to detect metalloproteins, but can also be employed to detect toxicologically relevant species in complex biological fluids. To this end, it has been reported that (GS) 2 AsSe − is excreted in rabbit bile after their injection with As(OH) 3 and HSeO 3 − [62]. Even though the presence of this species was inferred from the analysis of bile by XAS, this technique can inherently not conclusively identify the sulfur donor in this species. The analysis of bile from similarly treated rabbits by SEC-ICP-AES revealed a single As, Se, and S-containing entity and S-containing bile components. The addition of synthetic (GS) 2 AsSe − to this bile and reanalysis revealed an increased intensity of the As and the Se peak and therefore unequivocally identified GS-moieties as the ligands in the species that is excreted from the liver [63].
In the context of toxic metals interacting with endogenous ligands within organisms, however, their abstraction from proteins by a chelating agent is also eminently health relevant, since a better understanding of these events will directly contribute to the development of safer chelating agents for their effective tissue mobilization. Since the manipulation of the distribution of metal ions in biological organisms is a very complicated process (i.e., not very amenable to theoretical modeling [188]), one relies on empirical evidence from biological experiments to gain insight. This notion is exemplified by the recent application of SEC to systematically assess the efficiency of three chelating agents to abstract Cd 2+ from rabbit plasma proteins in vitro [36]. Using a Superdex 200 SEC column, PBS-buffer as the mobile phase and an ICP-AES, the analysis of rabbit plasma that had been spiked with 2.0 g Cd 2+ /mL revealed two distinct Cd-peaks, which were tentatively identified as 2 -macroglobulin and rabbit serum albumin. The dose-dependent addition of meso-2,3-dimercaptosuccinic acid (DMSA, Figure 1(f)), diethylenetriamine pentaacetic acid (DTPA, Figure 1(g)), or 2,3dimercapto-1-propanesulfonic acid (DMPS) to the Cd-spiked plasma (0.33-0.99 mM chelating agent) revealed DTPA to be most effective in abstracting Cd 2+ from plasma proteins (100% removal at 0.33 mM) to a Cd 2+ -chelate complex which eluted in the 5 kDa elution range. The simultaneous detection of the emission lines of Ca, Cu, Fe, and Zn, however, provided arguably the most valuable insight into unintended consequences of the investigated chelating agents. Whereas none of the chelating agents adversely effected the plasma distribution of Fe at all investigated doses, all of them mobilized Zn 2+ from plasma proteins to ∼5 kDa Zn-species (at 0.33 mM DTPA removed 80%, DMPS 63%, and DMSA 29%). Furthermore, DTPA resulted in a dose-dependent elution of a [Ca-DTPA] 3− complex, which is undesirable from a clinical point of view, since a decrease in the concentration of free plasma Ca 2+ is associated with twitching and titanic convulsions in humans [189]. These unintended chelating agent-induced perturbations of the plasma distribution of Ca 2+ and Zn 2+ could therefore result in the dyshomeostasis of these essential elements in vivo and-if they were repeatedly administered intravenously to patients-potentially result in pathologies [48]. The obtained results were therefore assessed in terms of finding a chelating agent and dose which offer a compromise between maximizing the abstraction of Cd 2+ from plasma proteins and minimizing that of Zn 2+ and ideally avoiding the complexation of free Ca 2+ . This compromise was a dose of 0.33 mM DMSA. Based on these in vitro results the employed LC-approach is destined to play an increasingly important role in the context of developing safer chelating agents for the treatment of patients that suffer from metal intoxications [190][191][192].

''Top-Down'' LC Approaches to Probe the Interaction of Metal-Based Drugs with Endogenous Ligands
In the context of the significant recent decline in the number of therapeutic medicinal drugs that are being approved [54,193], metal-based drugs offer several conceptual advantages compared to molecularly targeted ones that are constructed from organic fragments [54,162], but their side-effects remain a major problem [166]. Owing to the rising costs that are required to bring a new drug to the market [194], the resulting greater scrutiny to the return on R&D investment [195] translates to an increased need to be able to predict the efficacy and potential side-effects of promising drug candidates before expensive clinical studies are initiated [196]. Therefore, a better understanding of the interaction of promising metal-based drug candidates with endogenous ligands is becoming more important, and in vitro studies can offer valuable insight in this regard. In terms of ISRN Chromatography 13 developing better metal-based anticancer drugs, for example, their efficacy can be assessed in cell culture experiments using appropriate cancer cell lines (Figure 2, dashed red arrows), but one needs to be aware that this may not be sufficient to predict in vivo activity [197]. The prediction of their systemic side-effects in model organisms or patients (Figure 2, red arrows), however, remains challenging [83]. Conceptually, the side effects that intravenously administered metal-based anticancer drugs, such as cis-platin exert in patients must be attributed to their limited selectivity toward the tumor tissue compared to healthy tissues (they are therefore also referred to as shotgun cytotoxics [162]) and the complex interactions that unfold between the parent drug and its hydrolysis products with plasma proteins [198], erythrocytes [199], or cytosolic ligands [200]. Since interactions between clinically used metal-based drugs and proteins are not well understood [201], in vitro studies with blood plasma may provide insight into the biomolecular basis of their side effects. To this end, comparatively few studies have been conducted which employed "top-down" LC-approaches to probe such interactions [202]. One of the first studies which utilized SEC-ICP-MS to analyze blood serum which had been spiked with cis-platin over a 24 h period used a Supelco Progel TSK SEC column and 30 mM Tris-buffer (pH 7.2) [198]. An alternative approach for the analysis of rabbit blood plasma was developed by our group using a Superdex 200 SEC column and PBS-buffer as the mobile phase [100]. The developed method was first applied to study the binding of the surrogate metal-based drug arsenobetaine (Figure 1(i)) to human and rabbit blood plasma to which it had been added in vitro [203].
The results revealed that arsenobetaine-which is present in many marine organisms that are frequently used for human consumption-did not bind to plasma proteins >300 Da over a 6 h period, which provided a feasible explanation for the rapid excretion of this compound in urine. The first clinically used metal-based drugs that were investigated employing this "top down" LC approach were the anticancer drugs cis-platin and carboplatin [53]. The addition of pharmacologically relevant doses of each Pt-drug to human plasma in vitro and the subsequent analysis of plasma by SEC-ICP-AES over a 24 h period revealed a considerably faster hydrolysis of cisplatin than carboplatin. In addition, this study revealed that the hydrolysis products of both Pt-compounds appeared to bind to the same plasma proteins. Even though results that are obtained with blood plasma in vitro are in itself insufficient to predict if the same dynamic changes will occur in vivo [204], the fact that cis-platin and carboplatin are known to exhibit entirely different side effects in patients [53] strongly suggests that their interaction with plasma constituents could be involved in mediating their side effects. Another study applied RP-HPLC using a ODS-3 column to analyze rat blood to which oxaliplatin had been added [70]. The results revealed that 4 h after incubation 53% of oxaliplatin was associated with erythrocytes and that 58% of that was bound to cytosolic proteins (the rest was bound to membrane or in the ultrafiltrate). Interestingly, the analysis of erythrocyte lysate revealed several chemically unreactive Pt-containing species (oxaliplatin metabolites which contained GSH or Cys) and the parent oxaliplatin. The analysis of plasma revealed that the major biotransformation products were the main breakdown product of oxaliplatin, Pt(dach)Cl 2 . In addition, Cys and methionine adducts of Pt(dach)Cl 2 were detected. In view of the severe side effects of Pt-based anticancer drugs, the intravenous administration of model organisms with so-called ameliorating agents (either shortly before or after the metal-based drug) has been demonstrated to be effective [37]. Sodium thiosulfate (STS), for example, has been demonstrated to effectively ameliorate the side-effects of cis-platin. In order to gain insight into the biomolecular mechanism by which STS may exert this desirable effect, a pharmacologically relevant dose of STS was added to human plasma in vitro before or after a pharmacologically relevant dose of cis-platin. The subsequent determination of the Ptdistribution in plasma by SEC-ICP-AES revealed that a Pt-STS complex was formed. Although the structure of this complex has not been elucidated, its formation in blood plasma provides a feasible explanation for the STS-mediated amelioration of the side-effects of cis-platin in mammalian model organisms. Importantly, these results represent a starting point for the systematic exploration of other ameliorating agents in in vitro studies.

Outlook
The increased anthropogenic emission of toxic metals into the environment [34] will be inevitably associated with an increased influx of inorganic pollutants into the human bloodstream of various populations, including children. An improved understanding of the biochemistry of toxic metals therein as well as in target organs is therefore urgently needed not only to better understand the biomolecular mechanisms of their chronic toxicity, but also to develop inexpensive, efficient, and safe therapies to mitigate toxic metal-induced adverse health effects in affected populations. To date, the application of "bottom-up" LC approaches has provided valuable insight into the formation of toxic metalligand complexes at near physiological conditions {(GS) 3 As Perhaps most importantly, LC has been shown to be capable of probing the dynamic and competitive binding of a toxic metal to two ligands {As(OH) 3 to GSH/DMPS [135], Hg 2+ to Cys/GSH, CH 3 Hg + to Cys/GSH [121]}. In contrast, "topdown" LC approaches have allowed to investigate the binding of the toxic metal Cd 2+ to proteins in plasma in vitro [36] and to visualize a toxic metal induced perturbation (Cd 2+ displaced Zn 2+ from a plasma metalloprotein [26]), which constitutes a relevant mechanism of toxicity [205]. Furthermore, the analysis of bile by SEC-ICP-AES allowed to identify the thiol ligands of the detected As-Se species as GS which established (GS) 2 AsSe − as the molecular species that is excreted from the liver [63]. Last but not least "top-down" LC approaches can be employed to study the dose-dependent abstraction of Cd 2+ from plasma proteins by chelating agents [36]. Thus, the application of "bottom-up" and/or "top-down" LC approaches each offers unique, albeit different insights into the metabolism of toxic metals (Figure 4). The combined application of these techniques in conjunction with other techniques can contribute to better understand the interplay of toxic metals with endogenous ligands in mammals. Considering that the exposure of various populations to toxic metals is of increasing societal concern [206][207][208][209][210][211][212][213][214], LC approaches can be applied to develop practical solutions to address toxic metal-related problems in the near future.
Another societal issue that is indirectly related to metals is the significant decline in the number of therapeutic medicinal drugs that are being approved. This undesirable trend may be reversed if metal-based drugs can be developed that offer improved efficacy and reduced side effects compared to those that are in use today. To this end "bottom-up" LC approaches can provide insight into the binding of a metal-based drug and/or its hydrolysis products to mixtures of key plasma proteins (e.g., HSA, Tf). Results from such studies are useful to predict the absorption of a metal-based drug into the target organ cells via established uptake mechanisms. On the other hand, the addition of a metal-based drug to blood plasma followed by the LC analysis of the obtained mixture allows to simultaneously observe the hydrolysis of the parent drug and the plasma protein binding of the in situ generated hydrolysis products. This approach has considerable potential to advance our understanding of the biochemical fate of established or synthetic metal-based drug candidates in plasma which will contribute to potentially unravel the molecular basis of their side effects. In addition, "top-down" LC approaches are also ideally suited to gain fundamentally new insight into mechanism of action by which ameliorating agents (e.g., NAC) modulate the metabolism of metal-based drugs (e.g., cis-platin) in plasma. LC can therefore play an important role with regard to the development of novel strategies to decrease the adverse health effects of metal-based drugs while leaving their antitumor effect largely intact [37].

Conclusion
In terms of defining the bioinorganic mechanisms by which metal species exert toxicity which may eventually result in pathologies at the organ level, we are confronted with considerable knowledge gaps. The biomolecular mechanisms which determine the distribution of particular toxic metals to target organs, for example, are largely unknown even though their binding sites on key plasma proteins have been structurally characterized [181]. Similarly, the biomolecular basis of the severe side-effects that clinically used metal-based drugs exert in patients (e.g., the intravenously administered cis-platin) remains elusive, and the mechanisms by which ameliorating agents modulate the metabolism of cis-platin to mitigate its side effects essentially remain a terra incognita. Collectively, this undesirable situation must be attributed to the lack of a comprehensive understanding of the reversible and irreversible interactions that unfold between metal species (and their metabolites) with endogenous ligands in various biological compartments in vivo. Since it is precisely these interactions which define-in their entirety! -the toxicity that any given metal species (or mixtures thereof) will exert at the organ level (Figure 2), the application of "bottom-up" and "top-down" LC-based approaches has the potential to provide much needed insight into toxicologically relevant aspects of metal-ligand interactions under near physiological conditions. The concerted application of LCbased methodologies in conjunction with other advanced spectroscopic techniques will play an important role in the quest to better understand the disposition, the metabolism, the detoxification, the toxicology, the abstraction of toxic metals/metal-based drugs from proteins, and the modulation of the metabolism of metal-based drugs by ameliorating agents. Importantly, the information that can be obtained from such studies must be integrated into the wider biochemical context to establish progressively more refined models of the metabolism of the metal species of interest. Overall, the application of LC to "metal species in biology" related problems has the potential to establish maximum tolerable doses of a given metal species that may be ingested or injected before a particular biological network/system is perturbed, which would facilitate the interpretation of biomonitoring studies [75]. "Top-down" LC approaches are particularly suited to play a leading role in the context of developing better chelating agents to mobilize toxic metals from organisms, to possibly gain insight into the side effects of metal-based drugs, and to identify "ameliorating agents" which may be coadministered along with metal-based drugs to mitigate their often severe side-effects in patients. The interdisciplinary nature of this research endeavor-at the interface of physiology, biology, inorganic chemistry, biochemistry, analytical chemistry, toxicology and medicineis very much in the spirit of the father of LC. In fact, the inscription on M.S. Tswett's gravestone reads "He invented chromatography, separating molecules but uniting peoples. " May this notion continue to serve as a guiding principle as LC-based approaches will continue to serve humanity well into the 21st century.