Metabolomics: The Principles and Potential Applications to Transplantation

Authors


Corresponding author: David Wishart, david.wishart@ualberta.ca

Abstract

This review provides a summary of the applications and potential applications of metabolite profiling (i.e. metabolomics) in monitoring organ transplants. While the concept of metabolomics is relatively new to organ transplantation, the idea of measuring metabolites as a quick, noninvasive probe of organ function is not. Indeed, metabolite measurements of serum creatinine have long been used to assess pre- and post-operative organ function. Over the past 10 years, a number of lesser-known, organ-specific metabolites have also been shown to be good diagnostic indicators of both organ function and viability. In general, metabolomics offers a complementary picture to what can be revealed via techniques based on genomics, proteomics or histology. Because metabolic changes typically happen within seconds or minutes after an ‘event’, whereas some transcript, protein abundance or tissue changes may take place over days or weeks, metabolomic measurements may offer a particularly useful and inexpensive diagnostic tool to monitor donor organ viability or to detect organ rejection. The excitement associated with metabolomics, however, must be tempered by the fact that the technology for rapid metabolite identification is still in its infancy, and that metabolites are but one part of a very complex picture pertaining to organ function.

Introduction

What is metabolomics? The short answer is that metabolomics is the lesser-known cousin to genomics and proteomics. Just as genomics is concerned with the high-throughput, global measurement of all the genes in the genome, metabolomics is concerned with the high-throughput, global measurement of all the small-molecule metabolites in the metabolome. The metabolome is formally defined as the collection of all small-molecule metabolites (endogenous or exogenous) that can be found in a living cell or living organism. Metabolomics is a relatively new term, having been coined less than 5 years ago (1). Metabolomics is also known as metabonomics (2) or metabolic profiling (3). To some, metabolomics may seem like yet another unpronounceable entry into the 'omics' mania that is sweeping through most fields of biomedical research. However, it is actually a legitimate field, attracting substantial resources and attention from the NIH, Genome Canada, the European Union and many multi-national pharmaceutical companies. As a result, the field of metabolomics has grown quite rapidly, with more than 200 papers published on the subject in 2005.

Just like genomics and proteomics, metabolomics only become possible with the advent of recent technological breakthroughs in small-molecule separation and identification. These include the development of capillary electrophoresis systems and nano-HPLC systems for rapid compound separation, the invention of robust fourier transform ion cyclotron resonance (FTICR) mass spectrometry (MS) for exact mass determination, the development of high-resolution, high-throughput nuclear magnetic resonance (NMR) spectrometers and the creation of new chemometric software tools to rapidly process spectral or chromatographic patterns (4). With these hardware and software innovations, it is now possible for both scientists and clinicians to identify and quantify not just 1 or 2 small molecules at a time, but literally hundreds or even thousands of small-molecule metabolites in as little as a few minutes (4–7).

This newfound capacity to measure hundreds of important metabolites, quickly, easily and inexpensively has obviously opened the door to many potential applications in a large number of areas. These include such diverse fields as beverage testing, general plant biology, functional genomics, drug toxicology and testing, genetic disease (inborn errors of metabolism or IEMs) testing and even organ transplant monitoring (4–7). The purpose of this minireview is to provide the reader with a brief overview of what metabolomics is, what it is not and how this newly emerging discipline is already having an impact in the clinic and the field of organ transplant biology.

Metabolomics Versus Clinical Chemistry

The measurement of small-molecule metabolites has been an integral part of clinical practice and clinical chemistry for more than 100 years. Indeed, the monitoring of blood gases, the measurement of urinary glucose, or the quantitation of serum creatinine continue to be an integral part of the diagnostic arsenal for today's physician. These one- or two-component chemical tests provide a quick, inexpensive, quantitative and relatively noninvasive way of monitoring human physiology. What distinguishes metabolomics from clinical chemistry is the fact that in metabolomics one is measuring not just one or two compounds at a time but literally hundreds. In other words, metabolomics is 'ultrahigh-throughput' clinical chemistry.

Many of the metabolites that might be measured in a standard metabolomics run include the familiar clinical standards (amino acids, glucose, creatinine, urea). But it is also possible to simultaneously identify and quantify dozens of lesser-known compounds such as lactate, pyruvate, taurine, glycerol, thromboxane, citrulline, methylmalonate and citrate (3–5). By being able to measure all the metabolites at once, it is possible to get a far more comprehensive picture of what is happening to a patient's physiology or metabolism. This is why metabolomics is sometimes called metabolic profiling. Indeed, metabolomics provides a metabolic profile or 'signature' that is potentially as informative as the genetic signature from a gene chip. What is more, because metabolic responses are often measured in seconds or minutes (whereas other types of physiological responses are typically measured in days or weeks), metabolomic measurements can potentially yield important physiological information that is not normally accessible with gene chips, 2D gels or tissue biopsies.

In clinical chemistry, most metabolites are typically identified and quantified using colorimetric chemical assays. In metabolomics, a large number of metabolites are measured using nonchemical, noncolorimetric methods such as GC-MS (gas chromatography–mass spectrometry), LC-MS (liquid chromatography–mass spectrometry) or NMR spectroscopy. Interestingly, in some versions of metabolomic analysis, the compounds are actually not identified—only their spectral patterns and intensities are recorded (5,6). In other versions of metabolomic analysis, all (or most) of the compounds are identified and quantified (3,4,7). The former approach is based strongly on computer-aided pattern recognition and sophisticated statistical techniques like principal components analysis (PCA). The latter approach relies on spectral curve-fitting and prior chemical or spectral knowledge. Both methods have their advantages and disadvantages, although there is a growing trend toward the latter approach of absolute compound identification and quantification.

As with most clinical chemistry tests, the majority of metabolomic measurements are performed on biofluids, not tissues (3–7). This is done with the assumption that the chemicals found in these fluids are largely reflective of the physiological state of the organ that produces, or is bathed in, that fluid. Hence urine reflects processes going on the kidney; bile, the liver, cerebrospinal fluid (CSF), the brain and so on. The blood is a special biofluid as it potentially reflects all processes going on in all organs. This can be both a blessing and a curse because metabolite perturbations in the blood, while easily detectable, cannot be easily traced to a specific organ or a specific cause. In metabolomics, the choice of biofluids over tissues is also dictated by the fact that fluids are far easier to process and analyze with today's NMR, MS or HPLC instruments. Likewise, the collection of biofluids is generally much less invasive than the collection of tissues. Consequently, this review will focus primarily on metabolomic data collected on biofluids.

Metabolomics in Organ Transplantation

Metabolite measurements have been part of organ transplant monitoring for more than 40 years (8). While most metabolite measurements have been restricted to just a few well-known compounds (creatinine, glucose), there is a surprisingly large body of literature describing injury-dependent changes for a large number of lesser-known metabolites. Until recently, most of these measurements have been done using 'classical' clinical chemistry methods such as GC-MS (3), but since 1999 a growing number of reports have described the use of true metabolomic methods (NMR, LC-MS, spectral pattern analysis, etc.) to monitor organ function (9–15). Regardless of the technique or speed by which these compounds were measured, it is still quite instructive to explore what these metabolite measurements have revealed and what compounds are proving to be particularly good diagnostic or prognostic biomarkers.

In general, metabolite measurements have been performed to monitor two key aspects of organ physiology: (i) organ reperfusion injury and (ii) organ function (or dysfunction). Both types of measurements have been performed and reported on almost all solid organs that can be transplanted including kidneys, liver, lung and heart. The frequency of these reports for specific organs closely reflects the reported frequency of the corresponding organ transplant, with kidney (60%) leading the way, followed by liver (21%), heart (10%), pancreas (5%) and lung (4%). Most metabolite measurements associated with organ transplant analysis have been performed ex vivo, using biofluids such as urine, serum or bile (9–24). More recently, a few measurements have been performed in vivo using NMR chemical shift imaging techniques. These primarily measure inorganic phosphate or phosphorylated metabolites (ATP, ADP and phosphocreatine).

A summary of the types of metabolites and the observed changes associated with specific kinds of organ transplantation or organ dysfunction is given in Table 1. More detailed explanations of these observations will follow later in this review. As might be expected, most of the small molecules measured by metabolomic methods are associated with generic metabolic processes (glycolysis, gluconogenesis, lipid metabolism) found in all living cells. Changes to these 'universal' metabolites such as glucose, citrate, lactate, 2-oxoglutarate, ATP and ADP reflect changes in cell viability (apoptosis), levels of oxygenation (anoxia, blood flow), local pH, general homeostasis and so on (5). These molecules obviously can be quite informative of cell function or cell stress—and therefore, organ function. Other kinds of metabolites are specifically associated with tissue remodeling, muscle atrophy and myofibrillar breakdown (methyl-histidine, creatine, tuarine, glycine). These metabolites may provide important information about the extent of tissue repair or tissue damage (2,5). Some compounds, such as trimethylamine-N-oxide (TMAO), are actually used as buffers to stabilize serum proteins from the effects of accumulated waste products (9). Still other, less abundant metabolites such as thromboxane, histamine or chlorotyrosine may reflect changes in immune function or inflammation (10–12). As a general rule, metabolomic methods do not yet have the sensitivity to routinely detect or quantify low abundance second-messenger molecules (diacylglycerol, cyclic nucleotides), certain immune indicators (neopterin, prostaglandins) and many hormones. However, these too may soon be more accessible with continuing technical advances.

Table 1.  A summary of statistically significant metabolomic or metabolite measurements relevant to organ transplantation or organ dysfunction
OrganConditionMetabolite(s) increasedMetabolite(s) decreasedRef
Kidney (human)Chronic renal failureTMAO, dimethylamine, urea, creatinine (serum) 9
Kidney (rat)Renal damage (chemical)Acetone, lactate, ethanol, succinate, TMAO, dimethylamine, taurine (urine & serum)Citrate, glucose, urea allantoin (urine & serum)13
Kidney (human)Graft dysfunctionTMAO, dimetheylamine lactate, acetate, succinate, glycine, alanine, (urine) 14
Kidney (rat)Graft dysfunction reperfusion injuryTMAO, citrate, lactate, dimetheylamine, acetate (urine) 15
Kidney (rat)Reperfusion injury (ischemia)TMAO, allantoin (serum) 16
Kidney (human)Graft dysfunction CsA toxicityTMAO, alanine, lactate, citrate (urine & serum) 17
Kidney (mouse)NephrectomyMethionine, citrulline, arginine, alanine (urine & serum)Serine (serum)18
Kidney (mouse)NephrectomyGuanidinosuccinate, guanidine, creatinine, guanidinovalearate, (urine & serum)Guanidinoacetate (urine)19
Kidney (human)Acute rejection Nitrates, nitrites, nitric oxide metabolites (urine)20
Liver (human)Reperfusion injuryLactate, pyruvate, glycerol, alanine, glutamate, GABA, taurine, arginine (>19 h) (liver catheter)Arginine (<19 h) (liver catheter)22
Liver (Rat)Reperfusion InjuryCitrate, Succinate, Ketone bodies (good function)Citrate, succinate, ketone bodies (poor function)23
Liver (human)IschemiaMethylarginine dimethylarginine (liver catheter) 24
Liver (human)Graft dysfunctionGlutamine (serum & urine)Urea (urine)25
Liver (human)Post-transplantPhosphatidylcholine (bile) 26
Heart (human)RejectionNitrate (urine) 27
Heart (human)RejectionGeneral lipids, lipoproteins, VLDL, LDL, phosphatidylcholine (serum) 28
Heart (mouse)Acute rejection Phosphocreatine, PO4 (in vivo)30
Heart (human)Ischemia Phosphocreatine, PO4 (in vivo)31
Heart (human)Congestive heart failure N-acetylaspartate, creatine, choline myo-inositol (in vivo)34
Heart (human)RejectionThromboxane A2 (urine) 12

Because small molecules have the ability to diffuse almost anywhere in the body, a continuing challenge in metabolomics is to be able to rationalize the source and cause for these small-molecule signals. Sometimes excessive levels of certain metabolites may arise simply from exercise, a stubbed toe (distal injury), reduced fluid intake or changes in diet. In addition, the influence of immunosuppressive drugs or chronic infections on metabolite levels can also lead to confusing metabolite readings (17). Sorting these inconsequential changes from the more consequential changes is still a key challenge for metabolomics. Another challenge faced by both researchers and physicians alike is knowing what levels of certain metabolites are 'normal' and what levels are not. Certainly normal ranges are well known for the classic compounds of clinical chemistry (creatinine, urea, glucose, many amino acids), but once the list goes much beyond these compounds, many scientists and most clinicians are just guessing. Efforts have recently been undertaken by several groups across North America to obtain a more comprehensive listing of compounds in the human metabolome and a more complete picture of normal physiological concentrations of most readily detectable metabolites (http://www.hmdb.ca). This work is expected to be completed in late 2007, and it should provide an important reference set for future metabolomic studies.

Metabolomics in Kidney Transplantation

Because the kidney is particularly specialized in producing a metabolically rich and plentiful biofluid (i.e. urine), it is not surprising to find that most metabolomic studies related to organ transplants and organ function have focused on the kidney (see Table 1). In fact, over the past 15 years, more than 30 papers have been published describing or assessing urinary (and serum) biomarkers associated with donor organ injury, post-transplantation function, renal dysfunction, acute rejection and subclinical rejection (13–21). One common feature to almost all of these studies is the substantial (3–4-fold) increase seen in both urine and serum concentrations of TMAO (13–17). This compound is believed to be a homeostatic 'rescue' compound that allows blood proteins to cope with the increased concentrations of urea, guanidine and guanidino derivatives (all strong protein denaturants) that arise during renal failure or renal stress (9,19). TMAO is also known to be produced by the renal medulla, suggesting that graft dysfunction may be initially manifested by damage to renal medullary cells (14). While it is possible to identify the sources of tissue damage or impaired function through metabolomics, it is not always possible to identify the direct causes (rejection vs. infection). Perhaps with more studies and the measurement of more metabolites, it may eventually be possible to identify metabolite combinations that are specific to certain causes of impaired function.

In addition to reports of elevated levels of TMAO and other organic amines, metabolomic studies of transplanted, dysfunctional or rejected kidneys have also been used to detect the byproducts of inflammatory cells or localized inflammation such as nitrates, nitrites and histamines (10,20). Further inspection of Table 1 reveals that damaged kidneys also appear to rapidly elevate serum and urinary levels of lactate, acetate, succinate, ethanol and urea, which are all markers of Kreb's cycle (i.e. metabolic) distress. Some of these metabolite changes may also be a partial consequence of cellular stress responses to immunosuppressive drugs or low-grade infection (17).

While most of the studies summarized here describe changes to specific metabolites in serum, plasma or urine, there are certain kinds of metabolomics studies that do not identify specific compounds per se. Rather, they identify characteristic spectral (NMR, IR or MS) patterns that are reflective of changes to unknown or unidentified metabolites. Therefore, specific patterns, rather than specific compounds, are actually used to diagnose medical conditions (5). One report from 5 years ago demonstrated that episodes of subclinical rejection could be consistently detected by NMR or IR spectral patterns found in urine (6,21). Certainly, given the appeal of being able to noninvasively detect subacute or early-stage renal rejection, it seems a little odd that this technique is not yet widely used. One reason may be that this pattern-based approach does not explicitly identify a specific marker or compound. Without a tangible marker, often the FDA, physicians and clinical testing labs are reluctant to adopt such a diagnostic technique.

Metabolomics in Liver Transplantation

Just like the kidney, the liver is a key organ in general metabolism and waste handling. Therefore, as one might expect, small-molecule metabolites have proven to be excellent measures of liver function. Indeed, metabolomic studies using NMR, HPLC and MS methods have been used for a number of years to measure the extent of organ injury due to storage (22,23), the rate of organ recovery post-transplantation (22) and to identify prognostic and diagnostic markers of organ rejection and dysfunction (24–26). Because the liver is the organ primarily responsible for urea cycle metabolism, many of the key biomarkers are actually metabolites of urea, glutamine or arginine. For instance, the abundance of methylated arginine derivatives in donor livers is strongly prognostic of eventual organ rejection (24). Similarly, the quantity and rate of recovery of extracellular arginine levels appears to be a good indicator of organ function in the first 24 h post-transplant (22). Finally, abnormal levels of urea (reduced) and glutamine (increased) in both serum and urine may serve as potential, noninvasive diagnostic indicators of acute organ rejection (25). Certainly, the small number of organ-specific markers for liver function suggests that metabolomics may have much to offer the liver transplant community.

Metabolomics in Heart Transplantation

Considerable interest also persists in developing rapid, noninvasive methods to measure cardiac function and to detect cardiac rejection. Several promising methods have been developed that measure small-molecule biomarkers in either plasma, urine or in the heart directly. In metabolomic studies of urine, most work has focused on identifying metabolites that are indicative of inflammation or byproducts of the inflammatory process. These include nitrates (byproducts of the NO released by inflammatory cells), thromboxane A2 or B2 and neopterin (12,27). These studies found that these immuno-related metabolites were significantly elevated in patients experiencing rejection. Another metabolomic approach to detect or monitor cardiac rejection involves studying serum or plasma (28,29). One report, using 1H NMR methods, described the detection of acute cardiac rejection with a sensitivity and specificity >90% (28). This very simple approach involved analyzing the spectral width of selected lipid and lipoprotein peaks. It appears that this pattern recognition method indirectly measures the abundance of LDL and VLDL particles—with higher than normal levels generally being prognostic of rejection.

The most common metabolomic approach to monitoring cardiac function or cardiac rejection has been through in vivo31P NMR. These studies consistently find significantly reduced phosphocreatine/phosphate ratios as well as reduced ATP/phosphate and ADP/phosphate ratios (30,31). Not surprisingly, similar in vivo findings have also been seen in renal rejection and reduced renal function (32,33). Other in vivo1H NMR studies of cardiac dysfunction have noted an increase in N-acetylaspartate, myo-inositol and creatine in brain tissue (34). Given the importance of the heart in maintaining the body's aerobic metabolism, it is somewhat surprising that more work has not been done in monitoring plasma or even urinary levels of 2-oxoglutarate, lactate or acetate—all of which should be good indicators of the level of aerobic respiration and blood oxygenation.

Conclusion

The application of metabolomics to solid organ transplant monitoring, while still in its infancy, is providing yet another indication of how metabolomics is starting to change many medical diagnostic practices. As outlined in this review, there are now a number of metabolites that can be easily measured in urine, serum and bile—both in vivo and ex vivo—that can provide reliable indications of organ function, organ injury and early-stage organ rejection. As the field advances, it is likely that more chemical markers or more specific chemical profiles will be discovered that will allow even more precise diagnostic determinations. However, it is important to remember that metabolilte measurements will not be a panacea to solve all (or even most) diagnostic problems. Indeed, it may be many years before metabolomic measurements become a routine part of medical practice. As with any new technology, there is a certain amount of hype that often creates an unreasonable expectation about its potential or promise. While metabolomics clearly offers a number of exciting prospects, one must always remember that metabolites are only a small part of the pyramid of life (Figure 1), reflecting changes happening at many levels (organ, tissue, cell and molecule) arising from innumerable interactions involving the genome, the proteome, the immune system, exogenous chemicals and environmental variables. Understanding organ rejection, detecting certain kinds of organ injury or predicting the outcome of an organ transplant will always require the input from many disciplines and many technologies. What metabolomics potentially offers is 'top of the pyramid' perspective that may one day allow a more complete picture to be drawn of this very complex process.

Figure 1.

The pyramid of life, illustrating the relationship between the human genome, the proteome and the metabolome. The estimated number of human genes, enzymes and metabolites (>1 μM concentration) is indicated in each coloured tier of the pyramid.

Acknowledgments

The author wishes to thank Genome Prairie (a division of Genome Canada) and the Canada Foundation for Innovation (CFI) for financial support.

Ancillary