Organic acidemias (OAs) are inherited metabolic disorders caused by deficiency of enzymatic activities in the catabolism of amino acids, carbohydrates, or lipids. These disorders result in the accumulation of mono-, di-, or tricarboxylic acids, generally referred to as organic acids. The OA outcomes can involve different organs and/or systems. Some OA disorders are easily managed if promptly diagnosed and treated, whereas, in others cases, such as propionate metabolism-related OAs (propionic acidemia, PA; methylmalonic acidemia, MMA), neither diet, vitamin therapy, nor liver transplantation appears to prevent multiorgan impairment. Here, we review the recent developments in dissecting molecular bases of OAs by using integration of mass spectrometry- (MS-) based metabolomic and proteomic strategies. MS-based techniques have facilitated the rapid and economical evaluation of a broad spectrum of metabolites in various body fluids, also collected in small samples, like dried blood spots. This approach has enabled the timely diagnosis of OAs, thereby facilitating early therapeutic intervention. Besides providing an overview of MS-based approaches most frequently used to study the molecular mechanisms underlying OA pathophysiology, we discuss the principal challenges of metabolomic and proteomic applications to OAs.
The term “inborn errors of metabolism” (IEMs) was coined by Garrod in 1908 to describe genetically determined conditions, such as alkaptonuria, albinism, pentosuria, and cystinuria [
One of the primary challenges presented by IEMs is their extreme diversity, which has always made them difficult to classify. Currently, IEMs are categorized according to the affected organ (as in “neurological” or “hepatic” diseases) or to the affected organelle (e.g., “mitochondrial,” “peroxisomal,” or “lysosomal” disorders) or to the age of presentation (neonatal or adult-onset IEM). Because each of these approaches is informative, no single, universal classification system exists [
The genetic basis of IEMs is extremely heterogeneous and can involve any type of genetic defect: one or more point mutations, deletions or insertions, or genomic rearrangements (see Supplementary Table
The biological effects of IEM mutations can be mediated by four main processes: (a) direct toxicity of accumulating upstream metabolites; (b) deficiency of downstream metabolites; (c) feedback inhibition or activation by the metabolite on the same or different pathway; and (d) diversion of metabolic flux to secondary pathways [
Organic acidemias or organic acidurias (OAs) are inherited metabolic disorders caused by deficiency of enzymatic activities in the catabolism of amino acids, carbohydrates, or lipids [
Multisystem involvement in OAs.
Hearth | Skeletal muscle | Liver | Pancreas | Kidney | Brain | |
---|---|---|---|---|---|---|
MSUD | + | + | + | + | ++ | |
PA | + | + | + | + | ++ | |
MMA | + | + | + | + | + | ++ |
IVA | + | + | + | + | + | |
GA I | + | + | + | + | + | |
|
+ | + | + | + | + | |
HMG-CoA lyase D | + | + | + | + | + |
Some OA disorders are easily managed if promptly diagnosed and treated, but, in others cases, functional deterioration of brain and other organs with high energy demands are quite common [
PA is caused by mutations in either the propionyl-CoA carboxylase alpha (PCCA) or beta (PCCB) genes, which encode for the mitochondrial enzyme PCC. Deficiency of PCC activity leads to accumulation and excretion of propionate, 3-hydroxypropionate, methylcitrate, and propionylglycine, as well as ammonia and lactate, especially during metabolic crises [
Two enzymatic phenotypes of apoenzyme deficiency are recognized for MMA, both involving the methylmalonyl-CoA mutase (MUT): fibroblasts from
MMA can also be triggered by an aberrant intracellular metabolism of MUT cofactor, vitamin B12 (also referred to as cobalamin, Cbl). In particular seven enzymes are responsible for the transport, processing, and delivery of the appropriate Cbl form via a mitochondrion-targeted route to MUT [
Here, we review the latest results in dissecting molecular bases of OAs by using integration of mass spectrometry-based metabolomic and proteomic strategies. Generally, measurements of metabolites in various body fluids are the current tools for diagnosis. Mass spectrometry (MS) has facilitated the rapid and economical evaluation of a broad spectrum of metabolites from small samples, including dried blood spots. This approach has enabled the timely diagnosis of OAs, thereby facilitating early institution of therapy.
The term “metabolome” was introduced by Oliver et al. to indicate “the holistic quantitative set of low molecular-weight compounds (<1000 Da)” [
An appropriate study design is crucial to ensure the correct data interpretation of metabolic experiments. The inclusion of a sufficient number of subjects in each group, the sample collection, the choice of technological platforms, the processing of generated data, and the application of various bioinformatic methods are all factors that maximize analytical power of metabolomic approach aimed at identifying compounds and pathways of interest [
Two different strategies, named targeted and untargeted metabolomics, can be adopted, for this purpose (Figure
Schematic view of metabolomic methods. Samples deriving from body fluids (i.e., urine, plasma, and blood) are source for metabolomics. Two different strategies can be adopted. Targeted metabolomics allows the quantitation of a limited number of metabolites based on an a priori hypothesis. Untargeted metabolomics allows the determination of all the metabolites detectable in biofluids, without an a priori hypothesis. Biological interpretation of qualitative and quantitative alterations of metabolomics dataset correlates the metabolite patterns to biological pathways and cellular processes.
On the other hand, untargeted metabolomics is a nonspecific approach whose main aim is to determinate the whole set of metabolites detectable in a fluid or tissue, thus providing a functional fingerprint of the pathophysiological state of the body.
Whereas metabolic targeted and targeted procedures indicate what is happening at a biochemical level, the success of metabolomics in biomarker translation, with respect to other omics techniques, resides in the robustness of the adopted protocols and instrumentations, in the highly quantitative aspect, easily adapted to new assays and already located in many clinical testing laboratories.
Many successful studies have been conducted using urine, plasma, or serum samples. Urine metabolome better reflects kidney pathophysiological changes, while metabolome in whole blood, plasma, and serum is more related to systemic changes. The advantage of metabolomic analysis is to use noninvasive or minimally invasive sample collection procedure. Urine, in fact, easily and noninvasively collected, represents an “open system” that includes the intermediate metabolites, thus reflecting specific metabolic processes [
Due to the huge diversity of chemical structures and the large differences in abundance, there is no single technology available to analyze the entire metabolome. The most appropriate methodology may be selected as a compromise between the chemical selectivity, sensitivity, and speed of the different techniques. In addition to MS, also nuclear magnetic resonance (NMR) is used in order to analyze a large number of metabolites (up to 20–50) simultaneously [
As for processing of generated data, statistical analysis has a great impact on metabolite identification and quantitation and on the resulting biological interpretation. Two common statistical approaches may be adopted: the unsupervised method such as principal components analysis (PCA) and the supervised method, including partial least squares (PLS). In particular, PCA involves the transformation of the variables into a set of unrelated orthogonal components, with the first and a subsequent component explaining the largest and smaller amount of variance in the data, respectively. PCA helps to identify outliers in metabolomic experiments and also identify other technical issues that could produce variability in the data. PLS models correlate a feature of interest with the entire metabolomic dataset, and hence the components of a PLS model indicate how much the particular metabolite contributes to statistical significance of a specific dataset [
In order to obtain biological interpretation of whole metabolomic dataset, it is important to correlate metabolites belonging to the same metabolic pathway or chairing common quantitative changes. Hence, metabolomic data can be clustered into different “network” of metabolic pathways, in which nodes represent experimental and known metabolites [
Moreover, a proper validation of the obtained results represents an important challenge of metabolomic investigation. Hence, it is pivotal to compare the analytical data between different studies and/or laboratories.
In the last decade metabolomic approaches were applied in the investigation of IEMs to better understand the molecular processes underlying their development. The recent literature showed the big potential and impact of these approaches especially for diseases characterized by a-symptomatic development.
An example of diagnosis of pre- and a-symptomatic diseases is represented by the targeted metabolomic approach applied to newborn screening for inherited metabolic disorders. The IEMs including OAs were diagnosed by using tandem mass spectrometry since the 1990s [
Biomarker in OAs identified in DBS by LC-MS/MS analysis.
MSUD | Val ↑ | ||
Ile/leu ↑ | |||
|
|||
PA | C3 ↑ | Gly ↑ | C3/C0; C3/C4; C3/C16 |
|
|||
MMA mut | C3 ↑ | Gly ↑ | C3/C0; C3/C4; C3/C16 |
|
|||
MMA CblA and B | C3 ↑ | Gly ↑ | C3/C0; C3/C4; C3/C16 |
|
|||
MMA CblC and D | C3 ↑ | Gly ↑; Met ↑; C16:1OH ↑ | C3/C0; C3/C4; C3/C16 |
|
|||
IVA | C5 ↑ | ||
|
|||
GA I | C5DC ↑ | C5DC/C4; C5DC/C8; C5DC/C12; C5DC/C3DC | |
|
|||
|
C5:1 ↑ | C5OHn/↑ | |
|
|||
HMG-CoA lyase D | C5OH ↑ | C6DCn/↑ |
n means normal level,
Organic acids and acylglycines in OAs.
Urinary organic acids detected by GC-MS | Urinary acylglycines detected by LC-MS/MS | |
---|---|---|
MSUD | 2-Keto-isocaproic acid ↑ | |
2-OH-isovaleric acid ↑ | ||
2-Keto-isovaleric acid ↑ | ||
2-Keto-3-methylvaleric acid ↑ | ||
2-OH-3-methylvaleric acid ↑ | ||
|
||
PA | 3-OH-propionic acid ↑ | Tiglylglycine ↑ |
Methylcitric acid ↑ | Propionylglycine ↑ | |
|
||
MMA | Methylmalonic acid ↑ | |
Methylcitric acid ↑ | ||
|
||
IVA | 3-Hydroxyisovaleric acid ↑ | Isovalerylglycine ↑ |
|
||
GA I | Glutaric acid ↑ | Glutarylglycine ↑ |
3-OH glutaric acid ↑ | ||
|
||
|
2-Methyl-3-OH butyric acid ↑ | Tiglylglycine ↑ |
2-Methyl-acetoacetic acid ↑ | ||
|
||
HMG-CoA lyase D | 3-Methyl-glutaconic acid ↑ |
↑ means increase.
Dénes et al. [
Wojtowicz et al. [
Moving to untargeted examples, Wikoff et al. used untargeted LC-MS analysis to simultaneously profile thousands of metabolites on plasma of MMA and PA patients in order not only to characterize the metabolomic pattern of the these two diseases but also to define the specific difference between them [
Finally, an untargeted metabolomic approach was tested by Miller et al. [
In this regard, the recent technological advancements in MS may certainly promote the automation of the MS-based metabolomics analysis, thus allowing (a) reducing costs, (b) increasing throughput, (c) ensuring greater reproducibility, substantially cutting down on sample-handling errors, and (d) encouraging a greater focus on the absolute quantification. So the automation of technologies represents a great improvement especially when a high number of metabolomic analyses is required to reduce the number of false positives (normal sample reported as abnormal) and false negative (abnormal sample reported as normal). To date, the most successful example of metabolomic application to OAs is represented by metabolic targeted methods utilized in the newborn screening. Not only does MS-based newborn screening help in diagnosing or even predicting disease, but also the same techniques can also be used to determine the optimal therapy and monitor or customize the therapeutic dose. One of the main reasons for the success and widespread adoption of high-throughput MS-based screening is the very low sample costs.
However, a major limit of the metabolomic strategies is actually the limited number of identified metabolites due to the small metabolite coverage obtained by the so far developed MS profiling methods. It is well known that biological interpretation has to be performed on a high number of metabolites. It is challenging to get a good biological interpretation based on only fragments of the overall picture.
Proteomics has the potential to complement metabolomics and contribute to a better understanding of disease processes. The term “proteome” was first used by Wilkins et al. in 1996 to indicate snapshots of protein composition from a particular tissue or organism, at defined time points and under given physiological (or pathological) conditions [
In most quantitative proteomic workflows, MS-based procedures can be grouped in two major approaches: labeling and label-free methods [
Schematic view of quantitative proteomic methods. Samples deriving from patients (i.e., tissues, cells, and body fluids) are the sources for clinical proteomics. Label-free and labeling proteomic approaches are the two main groups of MS-based strategies aimed at identifying and quantifying differentially expressed proteins between two different samples A and B (i.e., cells or tissues from OA patients versus healthy controls). Label-free methods include SpC and MS/MS TIC approaches. On the other hand, the other methods are based on metabolic labeling, such as SILAC, or chemical such as DIGE, ICAT, TMT, and iTRAQ.
The classical quantitative proteomic methods utilize dyes coupled to a high-resolution protein separation technique, such as 2D electrophoresis. In particular, the use of fluorescent dyes in 2D-DIGE protocols increases sensitivity, offers a linear dynamic range, and allows both the quantitative comparison of gel-based protein patterns and their identification by MALDI-TOF or by LC-MS/MS techniques [
The other accurate quantitative approaches are based on stable isotope labeling; in this case, quantitation is achieved by comparing mass spectrometric signal intensities between corresponding labeled and unlabeled peptides. Isotope labels can be introduced chemically (ICAT, iTRAQ, and TMT) or metabolically (SILAC) into amino acids [
On the other hand, label-free approaches enable relative protein quantitation in complex mixture by (a) measuring the number of acquired MS/MS spectra for all peptides assigned to a given protein or (b) directly comparing the mass spectrometric signal intensity, namely, TIC of MS/MS spectra, assigned to all peptides for a given protein [
In clinical proteomic applications, the complex nature of human proteome represents a major challenge: in fact, the protein large dynamic range goes from 1–105 or 106 in cells up to 109–1010 in plasma [
Whatever quantitative proteomic approach is chosen, after protein separation and/or enzymatic digestion, the peptide mixture is injected into a mass spectrometer, usually coupled online with a HPLC. In particular, LC-MS/MS technologies are routinely used for protein/peptide identification in human complex samples [
As for quantitative proteomics, studies comparing protein levels between two different samples are aimed at detecting differential proteins whose expression significantly changes between conditions (Figure
Despite the biochemical characterization of OAs, the molecular mechanisms underlying the pathophysiology of these diseases remain poorly understood. However, changes that occur at protein level are now beginning to be explored by using clinical proteomic approaches. The proteome of MMA is so far the most explored thus representing the starting point for proteomic studies applied to OA.
In this context, much of our knowledge arise from (a) ex vivo studies with fibroblasts from MMA patients [
Summary of proteomic results from specimens of MMACHC patients.
Cellular system | Main results | MS-based proteomic technology | References |
---|---|---|---|
Fibroblasts | Underexpression of proteins related to apoptosis and metabolism. Overexpression of oxidative stress proteins | 2D-DIGE/MALDI-TOF and MALDI-TOF/TOF | Ebhardt et al., 2006 [ |
|
|||
Fibroblasts | Differentially expressed proteins related to cellular metabolism and regulation, cytoskeleton assembly, neurological system, cell signaling, and detoxification | 2D-DIGE/LC-MS | Richard et al., 2011 [ |
|
|||
Lymphocytes | Deregulation of proteins involved in oxidative stress and cellular detoxification, energy metabolism, cytoskeleton organization, and assembly | 2D-DIGE/LC-MS/MS |
Hannibal et al., 2015 [ |
|
|||
Liver | Differentially expressed proteins involved in energy and carbohydrate metabolism | 2D-DIGE/LC-MS/MS |
Caterino et al., 2015 [ |
To date, few efforts were dedicated to the finding of molecular signatures and altered cellular pathways, and, as a consequence, to the identification of useful protein targets for designing alternative therapies and/or predicting therapeutic outcomes. A major challenge of clinical proteomic studies related to rare metabolic diseases is often the small sample size due to the unavailability of cells or tissues from patients as well as from age-matched healthy subjects.
To our knowledge, all published studies, related to the proteome of different MMA forms, apply the labeling approach 2D-DIGE coupled to MS/MS techniques.
In this context, Richard et al. reported the first proteomic analysis of patients with isolated MMA by using 2D-DIGE/MS approach, by using MALDI-TOF and MALDI-TOF/TOF for protein identification [
To partially overcome the small sample size drawback, Hannibal et al. [
Interestingly, in a recent study conducted by using 2D-DIGE and LC-MS/MS or MALDI-TOF/TOF, some of the results observed in
To the best of our knowledge, the reported studies clearly show that the proteomic approach is useful for understanding cellular and metabolic processes underlying OA defect. Among MS-based techniques, only the 2D-DIGE analysis platform was utilized to study global protein expression, thus showing its feasibility also to other metabolic diseases. Although it is well known that the 2D-DIGE technique does not detect the whole proteome from a given source in comparison to LC-MS/MS-based methods, the papers here reviewed have successfully used this approach as a starting point to perform differential proteomics on samples derived from patients with IEMs.
The metabolomic data in OAs, collected to data, derive from extended newborn screening performed using LC-MS/MS platform [
The success of metabolomics in biomarker translation, with respect to other omics techniques, resides in the robustness of the adopted protocols and instrumentations, in the highly quantitative aspect, easily adapted to new assays and already located in many clinical testing laboratories. High automation of technologies represents a great improvement especially when a high number of metabolomic analyses are required. Not only does MS-based newborn screening help in diagnosing or even predicting disease, but also the same techniques can also be used to determine the optimal therapy and monitor or customize the therapeutic dose. However metabolomic strategies need to improve its analytical protocols; indeed they lack a standard operating procedure to analyze biofluids and a proper validation of the obtained results that allow comparing the analytical data between different studies and/or laboratories. A major limit of the metabolomic strategies with respect to other omics techniques, as proteomics, resides in the reduced number of identified metabolites. Compared to proteins, metabolites are a very heterogeneous molecular class due to their different physicochemical properties; so the simultaneous extraction by biofluids is difficult. This reason prevents a good biological interpretation of partial obtained data. The above limit may be overcome by utilizing complementary MS-based technologies. The future challenge in the study of OAs metabolomics is to enable simultaneous targeted and untargeted methods aimed at obtaining sensitive and accurate detection of predetermined metabolites, while allowing detection and identification of still unknown metabolites.
Much of our knowledge of protein changes in OAs arose from proteomic analysis with fibroblasts, lymphocytes, and liver from MMA patients by using 2D-DIGE technology coupled to MS. These clinical proteomic studies are challenged by specimen availability from OA patients as well as healthy subjects. Further investigations, including label-free proteomic approaches, could be employed for relative protein quantitation between specimen from OA patients and healthy controls for their versatility and the required small amount of biological samples.
By a global view of protein variations associated with MMA defect, most of the identified differentially expressed proteins are involved in energy metabolism, cellular detoxification, oxidative stress, and cytoskeleton assembly (Table
Two-Dimensional Differential In-Gel Electrophoresis
Cobalamin
Free Carnitine
Propionylcarnitine
Malonyl-L-carnitine
Butyryl-L-carnitine
Valeryl-L-carnitine
Tiglyl-L-carnitine
Glutaryl-L-carnitine
Hydroxyglutaryl-L-carnitine
Octanoyl-L-carnitine
Dodecanoyl-L-carnitine
Hexadecanoyl-L-carnitine
Dried blood spot
Data Dependent Acquisition
Data Independent Acquisition
Glutaric acidemia type I
Gas chromatography
3-Hydroxy-3-methylglutaryl-CoA
High-resolution mass spectrometry
Isotope Coded Affinity Tags
Inborn errors of metabolism
Isobaric Tag for Relative and Absolute Quantitation
Isovaleric acidemia
Liquid chromatography
Methylmalonic acidemia
Mass spectrometry
Tandem mass spectrometry
Maple syrup urine disease
Methylmalonyl-CoA mutase
Nuclear magnetic resonance
Organic acidurias
Propionic acidemia
Principal components analysis
Propionyl-CoA carboxylase alpha genes
Propionyl-CoA carboxylase beta genes
Partial least squares
Posttranslational modifications
Stable Isotope Labeling by Amino acids in Cell Culture
Spectral Counting
Total Ion Current
Tandem Mass Tags
Beta ketothiolase deficiency.
The authors declare that there are no competing interests regarding the publication of this paper.
This work was supported by Italian Ministry of Health (GR-2010-2317596), the IRCCS-SDN Foundation, and Associazione Culturale DiSciMuS RCF.