Metabolomics: A New Tool to Reveal the Nature of Diabetic Kidney Disease

Abstract

Metabolomics is a field of systems biology that draws on the scientific methods of other groups to qualitatively or quantitatively characterize small molecule metabolites in organisms, revealing their interconnections with the state of the organism at an overall relative macroscopic level. Diabetic kidney disease (DKD) is well known as a chronic metabolic disease, and metabolomics provides an excellent platform for its clinical study. A growing number of metabolomic analyses have revealed that individuals with DKD have metabolic disturbances of multiple substances in their bodies. With the continuous development and improvement of metabolomic analysis technology, the application of metabolomics in the clinical research of DKD is also expanding. This review discusses the recent progress of metabolomics in the early diagnosis, disease prognosis, and pathogenesis of DKD at the level of small molecule metabolites in vivo.

Diabetes mellitus (DM) is one of the chronic noncommunicable diseases that seriously endangers human life and health, affects the quality of life of patients, and increases the economic burden of countries. According to the latest data from the International Diabetes Federation, there were about 463 million people with DM worldwide in 2019, and it was expected to reach 700 million by 2045.¹ Diabetic kidney disease (DKD) is not only one of the most common chronic microvascular complications of diabetes but is also an important cause of end-stage renal disease (ESRD). Some research shows that about 30% to 50% of ESRD worldwide is due to DKD.² Alicic et al³ revealed that the prevalence of DKD in diabetic patients to be 40%. Current clinical indicators used to diagnose DKD include estimation of glomerular filtration rate (eGFR) and urine albumin creatinine ratio (UACR), although they have limited high sensitivity and specificity. The eGFR decreases significantly only when severe renal damage occurs in patients with DKD because the blood creatinine, which is used to calculate the eGFR, drops significantly only when the eGFR is below 50% of normal. Increased UACR is considered a characteristic manifestation of renal damage, but some studies have shown that a significant proportion of DKD patients do not have albuminuria.^4,5 In terms of treatment, despite the fact that the number of patients with DM and DKD is increasing annually, there are very limited therapies available to slow down or reverse the progression. The main therapeutic strategies for DKD include strict blood pressure and glycemic control, antidiabetic drugs, etc; however, these measures only slow down the progression of DKD, and there is no way to reverse DKD so far. This requires us to examine DKD from a new perspective, deepen the mechanism research, discover more biomarkers that can help early diagnosis, and develop new treatment strategies.

DKD is a chronic metabolic disease that often causes metabolic disorders of water, electrolytes, proteins, and lipids in the body. The dynamics of small molecule metabolites (relative molecular mass <1000) in an organism reflect the physiological or pathological state of the organism in real time. Metabolomics gives us a new perspective on DKD, which would help us to deepen our mechanistic studies, discover more biological markers for early diagnosis, and develop new therapeutic strategies.

Metabolomics

Metabolomics originated in the 1970s,⁶ when Professor Nicholson⁷ proposed the concept of “metabolomics” and drew up a metabolic fingerprint by scanning body fluids through nuclear magnetic resonance (NMR) techniques. With the completion of the sequencing of the Human Genome Project marking the advent of the postgenomic era, more and more research is focusing on the functional analysis of genomes. The emergence of metabolomics bridges the gap between genes and disease phenotypes and provides a possibility for the functional analysis of genomes. Metabolomics mainly focuses on the quantitative analysis of small molecule metabolites in organisms by following the research ideas of genomics to find the relationship between metabolites and physiological/pathological changes in the organism. Currently, metabolomics, genomics, transcriptomics, and proteomics have become important components of systems biology. The core concept of metabolomics is to use the metabolic state of the organism to reflect its overall functional status. It allows a comprehensive analysis of the dynamics of endogenous and exogenous compounds in the body and systematically reflects the expression levels of genes and proteins in the body in response to various stimuli. Many metabolic substances in the body have been maintained in a relatively balanced and stable state for a long time, which is regulated and influenced by genes, diet, environmental factors, and intestinal microorganisms; once this balance is disrupted, it would herald the onset or development of disease.⁸ Metabolomics, as an emerging omics discipline, has a wide range of applications in the fields of plants, food, disease diagnosis, and microbiology.⁹ Meanwhile, it has a multidisciplinary intersection and is closely related to organic chemistry, analytical chemistry, chemometrics, informatics, and biology.^10,11 Metabolomics is positioned downstream of proteomics and belongs to the continuation of the central dogma, which allows the mapping of body fluids (plasma, urine, tissue fluids, etc) and metabolic profiles to dynamically reflect the function and state of the organism. Compared with transcriptomics and proteomics, metabolomics has its own unique advantages: (1) as a continuation of the central dogma, it can amplify small changes at the gene level; (2) as the end product of gene expression and body metabolism, the number of metabolites in the body is much smaller than the number of genes and proteins; (3) as a relatively macroscopic representation of the body’s state, metabolomics is closer to the phenotype of disease and allows us to visualize the current physiological or pathological state of the organism; (4) the chemical composition of metabolites is similar across species, and the metabolomic analysis techniques used are more generalizable.¹²

In general, the main processes of metabolomics research include sample collection and preparation, sample testing, metabolic data collection, and data analysis. Currently, the samples used for metabolomics are mainly from blood, urine, and tissue fluids, and a sufficient number of representative samples are usually required to reduce individual differences. Different analysis platforms require different sample preparation processes; for example, nuclear magnetic resonance (NMR) does not require much processing of the sample and is noninvasive and nondestructive,¹³ while liquid chromatography/gas chromatography (LC/GC) coupled with mass spectrometry (MS) may require derivatization to increase the volatility of the sample.¹⁴ Data analysis is the most critical part of performing metabolomics studies. The sample testing process would generate a large amount of fragmented data. In order to extract potentially valuable information from these data, it is necessary to reduce the dimensionality of the data with the help of some multivariate mathematical statistical analysis methods. Partial least squares-discriminant analysis (PLS-DA), principal component analysis (PCA), neural networks, cluster analysis, and support vector machines are the most commonly used data analysis methods in metabolomics.^12,15 The appropriate data analysis method is selected according to the specific experimental design scheme as well as the experimental results. For example, PCA can be chosen when the results are more different between groups, while PLS-DA is preferred when the results are less different between groups and the sample size is more different between groups.¹⁵ On the basis of PLS-DA, characteristic variables of diseases is then screened out by the variable importance of projection (VIP), with the screening criterion of VIP >1.¹⁶ In turn, this would then be used to construct machine learning models and regression models for early identification and diagnosis of diseases. In addition, metabolic pathway enrichment analysis based on characteristic variables can find the specific metabolic pathways associated with diseases, which would contribute to the study of pathogenesis. MetaboAnalyst (4.0), the R-code-based online analysis tool for metabolomics, provides a convenient one-stop shop for metabolomics data analysis (screening of characteristic markers, cluster analysis, and enrichment of metabolic pathways), interpretation, and integration with other omics data.¹⁷ The basic flow chart of metabolomics research is shown (FIGURE 1).

Metabolomics Analysis Platform

Given that metabolomics requires the detection of large samples, high-throughput analytical platforms play a crucial role in this. Currently, metabolomics typically employs NMR, MS, LC/GC, and coupling of various analytical platforms to analyze and map metabolites.¹⁸ The coupling of these techniques can improve the resolution, sensitivity, and selectivity during sample analysis and help to map more metabolic profiles. Different metabolomics analysis platforms have their own unique features^14,19–22; the characteristics of the main metabolomics analysis platforms are shown in TABLE 1.

Early Diagnosis of DKD Based on Metabolomics

Although DKD still faces great challenges in terms of treatment, early diagnosis of DKD has become a promising research direction. The gold standard for the diagnosis of DKD is renal puncture biopsy, but this invasive test is difficult for patients to accept in the clinical setting. At present, DKD is mostly diagnosed clinically, and the main diagnostic criteria are eGFR and UACR. As mentioned above, eGFR and UACR still have some limitations as diagnostic criteria for DKD, which forces us to search for biological markers with higher sensitivity and specificity. Metabolomics, with its high-throughput detection technology and unique data processing methods, is uniquely positioned to find specific biological markers (TABLE 2).23–36

Currently, most of the biological markers identified based on metabolomics that contribute to the early diagnosis of DKD are focused on fatty acids, amino acids, lipids, and nucleotide metabolites. Fatty acids have an important impact on the metabolism of the body, and fat mobilization can increase the level of nitric oxide (NO) accumulation in the matrix, ultimately leading to increased glomerulosclerosis.³⁷ Some studies have shown that pathologically elevated saturated fatty acids have a strong toxic effect on cells,³⁴ which in turn promotes the development and progression of diabetes.²⁸ The role of unsaturated fatty acids in DKD is still a controversial topic. Some studies have shown that oleic acid ameliorates palmitic acid-induced endoplasmic reticulum stress, inflammation, and insulin resistance^38,39; however, other researchers suggested that high levels of oleic acid cause endoplasmic reticulum stress and apoptosis, causing proliferation or damage to glomerular thylakoid cells.^40,41 Phospholipids are key components of the phospholipid bilayer of biological membranes, which includes glycerophospholipids and sphingolipids. Peng et al⁴² showed that the expression levels of phosphatidylethanolamine (m/z = 750), phosphatidylglycerol (m/z = 747), and phosphatidylcholine (m/z = 802) were lower in DKD patients than in healthy controls. In addition, Zhu et al²⁹ identified plasma phospholipids in T2DM and DKD patients based on an HPLC-ESI/MS platform and screened 18 phospholipids (VIP >1, P < .05) as potential biomarkers by PLS-DA data analysis, including 2 novel biomarkers, phosphatidylinositol (m/z = 909) and sphingomyelin (m/z = 801). Several amino acid metabolic abnormalities were also observed in DKD individuals. Glutamate has an important role in regulating insulin secretion and maintaining glucose homeostasis in the body,⁴³ which has been suggested as a potential biological marker for early diagnosis of DKD.³⁴ In addition, plasma tryptophan levels were found to be positively correlated with estimated glomerular filtration rate in studies,³¹ while its metabolite, kynurenine, was negatively correlated with eGFR.³⁰

Prognostic Risk Assessment for DKD Based on Metabolomics

Metabolites in the organism are constantly changing dynamically in response to internal and external factors, and fluctuations in the metabolic profile suggest changes in the physiological/pathological state of the organism. By mapping and analyzing the metabolic profile of the organism, the disease progression and prognosis would be better assessed. Tavares et al⁴⁴ investigated the role of plasma metabolites in the assessment of DKD prognosis by an untargeted metabolomics approach. The results of the research showed that an adverse event (death, doubling of blood creatinine, or dialysis treatment) occurred in 30.3% of DKD patients during 2.5 years of follow-up; univariate Cox regression analysis showed that 1,5-anhydroglucitol (hazard ratio (HR): 0.10, 95% CI: 0.02–0.63, P =.01), norvaline (HR: 0.01, 95% CI: 0.001–0.4, P =.01), and aspartic acid (HR: 0.12, 95% CI: 0.02–0.64, P =.01) were negatively associated with the occurrence of DKD adverse events. Niewczas et al⁴⁵ conducted a prospective study with 158 patients with type 1 diabetic kidney disease followed up for a long period of time (median follow-up, 11.5 years). During the follow-up period, 99 patients (63%) progressed to ESRD; for the shorter the duration of diabetes, the younger the age, and the poorer the glycemic control, the faster the decline in renal function.⁴⁵ To further find the independent risk factors for the progression of DKD, they examined the serum of patients by high-throughput metabolomics technology, and the results revealed that 7 modified metabolites, including C-glycosyltryptophan, pseudouridine, O-sulfotyrosine, N-acetylthreonine, N-acetylserine, N6-carbamoylthreonyladenosine, and N6-acetyllysine, were independent risk factors for the progression of DKD to ERSD.⁴⁵ Another comprehensive study of plasma metabolic profiles in patients with type 2 diabetes mellitus showed that certain metabolites (amino acids and their derivatives) predicted the onset of ESRD approximately 10 years earlier and independent of baseline albuminuria and renal function.⁴⁶ Tofte et al⁴⁷ illustrated that alterations in plasma metabolites (particularly polyols and branched-chain amino acids) were associated with future renal impairment in the type 1 diabetes mellitus (T1DM) population; ribonucleic acid was associated with a higher risk, while the amino acids isoleucine, leucine, and valine were associated with a lower risk of combined renal endpoint (≥30% decline in eGFR, ESRD, and all-cause mortality). Acylcarnitine is essential for intracellular energy metabolism and plays an essential role in the β-oxidation metabolism of fatty acids. Research has shown impaired fatty acid oxidation in patients with DKD; the abnormal plasma acylcarnitine levels could be a reflection of the degree of disease progression.^48,49 It has also been shown that metabolites such as uremic toxin and carnitine were strongly associated with the progression of microproteinuria in patients with T1DM.⁵⁰ Looker et al⁵¹ showed that 14 markers, such as symmetrical dimethylarginine, uracil, acylcarnitine, and hydroxyproline, were associated with a rapid decline in renal function in patients with T2DM, suggesting that their combination might greatly improve the prediction of renal function in the future (TABLE 3).44–47^,49–51

Study of the Pathogenesis of DKD on Metabolomics

Metabolic profiles represent the complete collection of metabolites in an organism, and perturbations in metabolic profiles may help to reveal the pathogenesis of the disease early. The convergence of multiple omics plays an important role in the study of DKD pathogenesis. Recent studies have shown that a novel oral hypoglycemic agent (SGLT-1) could act as a renoprotective agent by inhibiting glucose reabsorption and reducing renal tubular energy expenditure,^52–54 which laterally suggested an energy imbalance in the renal tissues of DKD patients. Sharma et al,⁵⁵ through urinary metabolomics studies in DKD patients, identified 13 metabolites whose expression was significantly reduced, and the reduction in organic anion expression was associated with the downregulation of the expression of the organic anion transporter (OAT) gene in DKD patients. Bioinformatics analysis revealed that 12 of the 13 differentially expressed metabolites were associated with mitochondrial metabolism and that the expression of PGC1α, a major regulator of mitochondrial activity, was significantly lower in the renal tissue of DKD patients (overall inhibition of mitochondrial activity).⁵⁵ Multiple metabolic pathways (pentose phosphate, purine and pyrimidine metabolism, hexosamine biosynthesis, and tricarboxylic acid cycle) were overactivated, and mitochondrial complex 1 activity was enhanced in DKD mouse models (streptozotocin-induced), while specific mitochondrial complex 1 inhibitors were effective in reducing glomerular and tubular injury through modulation of the body’s metabolic state and oxidative stress.⁵⁶ Some scholars have reported abnormalities in fatty acid metabolism in DKD.³³ Saulnier et al⁵⁷ showed a correlation between urinary metabolites and renal structure in patients with DKD, where short-chain fatty acids in urine were associated with structural destruction of endothelial cells in the early stages of DKD. In addition, other studies have shown that defects in fatty acid oxidation play a key role in the development of renal fibrosis.⁴⁸ Multi-omics studies in DKD have revealed that accumulation of extracellular matrix, abnormal activation of the inflammatory microenvironment, and metabolic disturbances contributed to the development of glomerulosclerosis and tubulointerstitial fibrosis. Further integrative analysis revealed that linoleic acid metabolism and fatty acid β-oxidation are significantly inhibited in the pathogenesis and progression of DKD.⁵⁸ Carnitine is an important substance in the oxidative metabolism of fatty acids and protects against cellular oxidative stress from various sources⁵⁹; indeed, Devi et al²⁴ showed abnormal carnitine metabolism in patients with DKD. The role of NADPH oxidase 4 in the development and progression of DKD is poorly understood. You et al⁶⁰ demonstrated the presence of abnormal tricarboxylic acid cycle-related metabolites in DKD mice and that NADPH oxidase 4 could cause an increase in ferredoxin levels. This in turn led to endoplasmic reticulum oxidative stress and increased expression of HIF-1α and TGF-β. Chinese medicines are increasingly used in the treatment of DKD, but the mechanism of action is not well understood. Studies in network pharmacology combined with metabolomics have shown that herbal medicines (shenyankangfu tablets) may improve insulin resistance by regulating the biosynthesis of unsaturated fatty acids and the metabolism of starch and sucrose.⁶¹ Epigenetics has become more widely accepted in the pathogenesis of DKD.⁶² High glucose conditions may alter the histone amino acid modification status (methylation, acetylation), causing changes in chromatin structure, facilitating the binding of transcription factors, and ultimately leading to the expression of genes associated with inflammation and fibrosis.⁶³ Epigenetic markers could not only lead to abnormal expression of metabolism-related genes⁶⁴ but may also predispose offspring to metabolic abnormalities and shortened lifespan.⁶⁵ A prospective cohort study by Niewczas et al⁴⁵ demonstrated that circulating levels of 7 modified metabolites (C-glycosyltryptophan, pseudouridine, O-sulfotyrosine, N-acetylthreonine, N-acetylserine, N6-carbamoylthreonyladenosine, and N6-acetyllysine) were associated with the rate of decline in renal function and the risk of ESRD. It has also been observed that elevated levels of serum modified amino acid metabolites (O-sulfotyrosine) were highly correlated with chronic decompensation of renal function.⁶⁶

Conclusion

Metabolomics, as an important component of systems biology, is expanding its application in DKD clinical research. Numerous small molecule metabolites (such as amino acids and lipids) in the body have been screened and are expected to become biological markers for the early diagnosis of DKD. The evolution of the metabolic profile with the disease course gives us enough opportunities to evaluate the prognosis of DKD, enabling us to make early interventions. Metabolomics as a new tool gives us a new perspective through which to look at the pathogenesis of DKD. Powered by information biology techniques, we have understood that there are disorders of mitochondrial energy metabolism within individuals with DKD, which severely compromise the structure and function of the kidney. Epigenetics plays an important role in the pathogenesis of DKD, and an increasing number of researches show the presence of abnormal modified metabolites in DKD patients. In addition, metabolomics also has given herbal medicine an arena to demonstrate its efficacy. In the future, with the innovation of high-throughput analysis platforms, cross-fertilization of multi-omics, and comprehensive infiltration of information biotechnology, we will be further exposed to the nature of DKD, which is necessary if we are to overcome it.

Abbreviations

Abbreviations
 
• DKD

  diabetic kidney disease

 
• DM

  diabetes mellitus

 
• ESRD

  end-stage renal disease

 
• eGFR

  estimation of glomerular filtration rate

 
• UACR

  urine albumin creatinine ratio

 
• NMR

  nuclear magnetic resonance

 
• LC/GC

  liquid chromatography/gas chromatography

 
• MS

  mass spectrometry

 
• PLS-DA

  partial least squares-discriminant analysis

 
• PCA

  principal component analysis

 
• VIP

  variable importance of projection

 
• NO

  nitric oxide

 
• T1DM

  type 1 diabetes mellitus

 
• OAT

  organic anion transporter

 
• HR

  hazard ratio

Funding

Grant of Zhejiang Medicine and Health Technology Project, China (No. 2018ZH029, 2020KY871), the Major Project for Science & Technology Innovation 2025 in Ningbo, China (No. 2019B10035), and Grant of Ningbo Social Development, China (No. 2019C50080).

REFERENCES

Author notes