Metabolomics Analysis of Human Vitreous in Diabetic Retinopathy

Vitreous samples from patients with rhegmatogenous retinal detachment (n = 25) and proliferative diabetic retinopathy (n = 9) were profiled along with...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/jpr

Cite This: J. Proteome Res. 2018, 17, 2421−2427

Metabolomics Analysis of Human Vitreous in Diabetic Retinopathy and Rhegmatogenous Retinal Detachment Nathan R. Haines,†,§ Niranjan Manoharan,† Jeffrey L. Olson,† Angelo D’Alessandro,‡ and Julie A. Reisz*,‡ †

Department of Ophthalmology, University of Colorado, Anschutz Medical Campus, Aurora, Colorado 80045, United States Department of Biochemistry and Molecular Genetics, University of Colorado, Anschutz Medical Campus, Aurora, Colorado 80045, United States

J. Proteome Res. 2018.17:2421-2427. Downloaded from pubs.acs.org by UNIV OF SUNDERLAND on 09/27/18. For personal use only.



S Supporting Information *

ABSTRACT: The vitreous humor is a highly aqueous eye fluid interfacing with the retina and lens and providing shape. Its molecular composition provides a readout for the eye’s physiological status. Changes in cellular metabolism underlie vitreoretinal pathologies, but despite routine surgical collection of vitreous, only limited reports of metabolism in the vitreous of human patients have been described. Vitreous samples from patients with rhegmatogenous retinal detachment (n = 25) and proliferative diabetic retinopathy (n = 9) were profiled along with control human vitreous samples (n = 8) by untargeted mass-spectrometry-based metabolomics. Profound changes were observed in diabetic retinopathy vitreous, including altered glucose metabolism and activation of the pentose phosphate pathway, which provides reducing equivalents to counter oxidative stress. In addition, purine metabolism was altered in diabetic retinopathy, with decreased xanthine and elevated levels of related purines (inosine, hypoxanthine, urate, allantoate) generated in oxidant-producing reactions. In contrast, the vitreous metabolite profiles of retinal detachment patients were similar to controls. In total, our results suggest a rewiring of vitreous metabolism in diabetic retinopathy that underlies disease features such as oxidative stress and furthermore illustrates how the vitreous metabolic profile may be impacted by disease. KEYWORDS: untargeted metabolomics, absolute quantification, vitreous, vitreoretinal disease, diabetic retinopathy, mass spectrometry, retinal detachment



aggressive medical and surgical therapy.3 Early and minimally invasive methods for observing systemic changes in the eye may provide robust and facile tools to inform management of higher risk patients and further develop the understanding of how cellular metabolism is altered in vitreoretinal diseases. Retinal and vitreous metabolism vary depending on underlying retinal pathology, and, to date, a lack of full metabolic profiling of the vitreous in healthy and disease states hinders the use of this powerful analytical approach to distinguish characteristics of disease states and patient outcomes. Obtaining vitreous fluid for analysis is fairly straightforward in patients undergoing vitrectomy surgery and has significantly lower risk than obtaining retinal tissue. To date, however, there exist only limited descriptions of vitreous metabolism in human patients focusing on a small subset of metabolites.4 Here we examined the metabolic profiles of human vitreous fluid from patients with retinal pathologies (RD or DR) and patients without significant retinal disease. High-throughput5 generation of a comprehensive profile of the vitreous metabolome in the context of healthy and disease states provides the means to identify key metabolic axes that

INTRODUCTION The vitreoretinal interface plays a significant role in surgical retinal diseases ranging from symptomatic vitreous floaters to more serious problems of macular holes, rhegmatogenous retinal detachments (RDs), ocular traumas, and diabetic eye disease. Severity of disease and other factors, such as age and ethnicity, affect but do not fully explain disparities that exist in surgical outcomes. Underlying these pathologies are cellular changes that may be causative or correlative factors. For example, an epithelial to mesenchymal transition (EMT) of retinal pigmented epithelial (RPE) cells following rhegmatogenous RD allows for their migration into the vitreous and is associated with extracellular matrix remodeling, leading to proliferative vitreoretinopathy (PVR).1 Although the development of PVR affects a minority of RD patients, a subsequent surgery is required, and ultimate visual outcomes are often poor. In diabetic patients, there is significant variability in the severity of diabetic retinopathy, which is not fully explained by the patient’s glucose control.2 Diabetic retinopathy (DR), the leading cause of blindness among working age adults in the U.S. and developed world (Centers for Disease Control), requires intravitreal injection, laser therapy, or surgical intervention simply to halt disease progression. Certain patients have refractory DR that fails to improve despite © 2018 American Chemical Society

Received: March 14, 2018 Published: June 7, 2018 2421

DOI: 10.1021/acs.jproteome.8b00169 J. Proteome Res. 2018, 17, 2421−2427

Article

Journal of Proteome Research

Figure 1. (A) Initial study of human vitreous metabolism. (B) Receiver operating characteristic (ROC) curves for top metabolites distinguishing control and diabetic retinopathy vitreous. (C) ROC curves for top metabolites distinguishing retinal detachment and diabetic retinopathy vitreous.

4 °C for 30 min then spun 10 min at 10 000g and 4 °C. Protein and lipid pellets were discarded, and supernatants were analyzed by ultra-high-pressure liquid chromatography−mass spectrometry (UHPLC−MS) on a Thermo Vanquish UHPLC (San Jose, CA) coupled to a Thermo Q Exactive mass spectrometer (Bremen, Germany) in positive and negative ion modes (separate runs). Solvents were water and acetonitrile supplemented with formic acid (0.1% - positive mode) or ammonium acetate (5 mM - negative mode). Initial analysis utilized a 3 min isocratic run as previously described;5,6 subsequent extensive analysis (including absolute quantification) was performed using a 9 min gradient from 5 to 95% acetonitrile organic phase, as described.7 Samples were randomized, and a quality-control sample was injected every 10 runs. Data analysis was performed using Maven (Princeton University) following file conversion by MassMatrix (Case Western Reserve University). Absolute concentrations were obtained using the following equation

are disrupted in retinal damage or disease. Moving forward, robust and rapid global profiling may aid in stratification of disease states or prediction of adverse postoperative outcomes such as PVR.



EXPERIMENTAL SECTION

Vitreous Sample Collection

Colorado Multiple Institutional Review Board (COMIRB) approval was obtained for retrieval of human vitreous at the time of routinely scheduled vitrectomy on patients at the University of Colorado Health Eye Center (Aurora, CO). After anesthesia, the eye was prepped in usual sterile ophthalmic fashion. Trocars were inserted in standard fashion for 3-port pars plana vitrectomy. Prior to any surgical maneuvers and with infusion off, the vitrector was used to manually aspirate samples into a tuberculin syringe (∼0.3 mL) while cutting at 7500 cuts per minute. Samples were deidentified, and patient age, gender, diagnosis, ocular history, medical history, preoperative visual acuity, and date of sample collection were documented. Vitreous samples were classified as control (epiretinal membrane), retinal detachment, and diabetic retinopathy and were immediately cooled and stored at −80 °C.

[light] = (peak area light /peak area heavy)[heavy]*DF

where DF is the dilution factor, in this case, 3 (i.e., 25 μL of vitreous in a total 75 μL volume). Absolute concentrations for additional acylcarnitines (Table S3) were estimated using the labeled acylcarnitine with closest structural similarity (i.e., similar fatty acyl moiety carbon backbone length). Relative quantification data were normalized to median and autoscaled within the MetaboAnalyst 3.0 platform prior to visualization and statistical analysis. Hierarchical clustering analysis was performed using GENEE (Broad Institute). Bar graphs were prepared using GraphPad Prism 5.03. Receiver operating characteristic curves, partial least-squares-discriminant analysis, and statistical analysis (ANOVA) for heat maps prepared using MetaboAnalyst 3.0.

UHPLC−MS Metabolomics

All solvents were Optima grade (Fisher Scientific). Vitreous samples were thawed on ice, then 25 μL of vitreous was mixed with 50 μL of ice cold extraction buffer (5:3:2 MeOH/ACN/ H2O). For absolute quantification, all stable isotope-labeled standards were purchased from Cambridge Isotope Laboratories. Where applicable, the extraction buffer contained 3.75 μM of an amino acid mixture (MSK-A2-1.2), an acylcarnitine mixture (NSK-B) diluted 1:200 according to the manufacturer’s instructions (final concentrations: free carnitine D9, 0.76 nM; acetylcarnitine D3, 0.19 nM; propanoyl D3, 0.038 nM; butyryl D3, 0.038 nM; isovaleryl D9, 0.038 nM; octanoyl D3, 0.038 nM; myristoyl D9, 0.038 nM; palmitoyl D3, 0.076 nM), [2,2,4,4-D4]citrate (1.5 μM), [U-13C]α-ketoglutarate (1.5 μM), [U-13C]succinate (1.5 μM), [1,4-13C2]fumarate (1.5 μM), and [1-13C]pyruvate (1.5 μM). Samples were vortexed at



RESULTS AND DISCUSSION Patient vitreous samples were obtained during the course of standard vitrectomy. As an aqueous medium interfacing with the retina, lens, and numerous cell types, the biomolecular composition of the vitreous humor provides a systemic overview of eye physiology. To assess the global metabolism 2422

DOI: 10.1021/acs.jproteome.8b00169 J. Proteome Res. 2018, 17, 2421−2427

Article

Journal of Proteome Research

Figure 2. (A) Expanded study of human vitreous metabolism. (B) Partial least-squares-discriminant analysis (PLS-DA) illustrates partial overlap between the control and RD groups and distinction of the DR cohort. (B) Hierarchical clustering analysis of significantly altered metabolites (p < 0.05 ANOVA). Values are normalized across each row, where red and blue represent up- and down-regulation, respectively. Subclusterings within the retinal detachment (RD) and diabetic retinopathy (DR) cohorts are evident and suggest phenotypic differences.

metabolites, and 8 acylcarnitines (see the Experimental Section for a full list of metabolites and isotope labeling) in patient vitreous fluid. In a manner analogous to the workflow described above, metabolites were identified from UHPLC−MS data, and statistical significance was determined using ANOVA (MetaboAnalyst). A partial least squares discriminant analysis (PLSDA, Figure 2B) prepared using all identified metabolites reveals a distinct clustering of the diabetic cohort with respect to RD and control vitreous. A hierarchical clustering analysis (Figure 2C) illustrates the overall findings revealed by relative peak area quantification of 150+ metabolites, with a focus here on those with p < 0.05. As expected, major findings from the initial experiment are confirmed in this second cohort, namely, decreases in ascorbate, 5-oxoproline, and fumarate in the diabetic vitreous along with increased levels of proline, citrulline, and aspartate in this group. The expanded metabolome profile utilized here allowed for a closer investigation of glucose metabolism (elaboration follows) and the measurement of less polar compounds like fatty acids and acylcarnitines, which were differentially present in the diabetic group (decreased and increased, respectively) compared with epiretinal membrane controls and patients with retinal detachments. Acylcarnitine elevation has been noted in the urine of diabetic mice in a recent study of diabetic nephropathy.10 Interestingly, the hierarchical clustering in Figure 2C reveals disparities or subgroupings within the RD and DR groups, although in separate areas of the metabolome. RD patients cluster into two subgroups when examining levels of several fatty acids along with metabolites downstream of glucose (phosphoglycerate, glyceraldehyde 3-phosphate, hexose phosphate, and lactate). Similarly, the DR patients can be roughly discriminated by levels of proline, citrulline, glucose downstream product ribose phosphate, creatinine, and several others. We questioned whether an obvious demographic or clinical parameter was contributing to the delineation along metabolic phenotypes in the RD and DR groups, but close examinations of gender, age, ethnicity, day one postoperative visual acuity, final visual acuity, and multivariate analysis using PLS-DA did not reveal a significant correlation with or

of vitreous samples from patients with detached retina or DR versus control, batched analysis was performed on a preliminary set of 34 patient samples (epiretinal membrane: n = 9, age 68 ± 6 years; rhegmatogenous RD: n = 17, age 62 ± 10 years; proliferative DR: n = 8, age 41 ± 10 years) via metabolite extraction, followed by analysis by untargeted UHPLC−MS metabolomics (Figure 1A). This analysis utilized only 25 μL of vitreous per patient and resulted in the identification of >100 named metabolites (KEGG, Table S1), with 17 metabolites demonstrating statistically significant differences among cohorts (p < 0.05 by ANOVA, Figure S1). The DR cohort emerged as metabolically distinct from both the epiretinal membrane control and the RD cohorts, as judged by Student’s t test, with the latter two being more similar to each other. To identify the metabolites best distinguishing DR vitreous from the vitreous of control and RD patients, receiver operating characteristic (ROC) curves were prepared (Figure 1B,C). Purine metabolite xanthine was the leading biomarker for DR versus control, and proline and citrulline, both key components of nitrogen metabolism, along with aerobic glycolysis end-product pyruvate, were crucial for distinguishing DR vitreous both from control and RD. An upregulation of nitrogen metabolites such as proline, Namidinoaspartate, creatinine, and citrulline was observed in the diabetic retinopathy cohort along with a decrease in this group of several antioxidant system metabolites (5-oxoproline, ascorbate) and mitochondrial metabolites involved in the TCA cycle (Figure S1). Reduced levels of ascorbate have been noted in DR vitreous versus macular hole controls,8 and a previous comparison of DR versus control patient vitreous noted increased the activity of the arginine to proline pathway.9 A subsequent expanded cohort of patient samples (epiretinal membrane: n = 9, age 67 ± 6 years; rhegmatogenous RD: n = 25, age 63 ± 10 years; proliferative DR: n = 9, age 45 ± 13 years) was next utilized to prospectively validate preliminary findings and more deeply investigate energy and redox metabolic pathways (Figure 2A). In addition to the global, untargeted metabolomics analysis performed initially, we also utilized stable isotope-labeled metabolite standards for the absolute quantification of 17 amino acids, 5 mitochondrial 2423

DOI: 10.1021/acs.jproteome.8b00169 J. Proteome Res. 2018, 17, 2421−2427

Article

Journal of Proteome Research

Figure 3. Glucose metabolism in the vitreous of control, retinal detachment, and diabetic retinopathy patients. Levels of metabolites downstream to glucose suggest a downregulation of glycolysis accompanied by activation of the pentose phosphate pathway (PPP) in diabetic retinopathy vitreous (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 by ANOVA).

Figure 4. Purine metabolism in the vitreous of control, retinal detachment, and diabetic retinopathy patients. Decreases in xanthine emerged as a strong biomarker for diabetic retinopathy. Increases in this cohort of metabolites upstream to xanthine suggest increased xanthine oxidase activity in DR. Critically, several of these reactions generate reactive oxygen species (red) (* p < 0.05, ** p < 0.01, *** p < 0.001 by ANOVA).

diagnoses of nephropathy, hypertension, and hyperlipidemia, presence of proteinuria, blood glucose level at time of surgery, hemoglobin A1C level, hematocrit, total hemoglobin beta, the presence of fibrosis or tractional retinal detachment, cataract status, and eye drops taken preoperatively. None of these additional variables (reported in Table S2) considered provided strong evidence to underlie the observation of subcohorts in the RD and DR groups. Other factors to consider are the potential for a varied length of time between

predictive measure for the clustering of the subgroups. To further investigate, in the RD cohort, we considered patient BMI, smoking status, the presence of diabetes, the presence of proliferative vitreoretinopathy (PVR) both pre- and postoperatively, number of retinal breaks, number of days detached, location of break(s), cataract status, macula status, and eye drops taken preoperatively. Similarly, in the DR population studied, we considered diabetes type, insulin dependency of diabetes, duration of diabetes, previous 2424

DOI: 10.1021/acs.jproteome.8b00169 J. Proteome Res. 2018, 17, 2421−2427

Article

Journal of Proteome Research

which notably produces the potent reactive oxygen species (ROS) hydrogen peroxide along with urate. H2O2 is also generated in the subsequent metabolism of urate by urate oxidase. Alternatively, because the vitreous is a systemic fluid whose composition reflects the cumulative effects of metabolism in the retina and lens, it is possible that increased purine levels occurring in DR are a result of increased production and release in this tissue or decreased uptake from the vitreous. Overall, increased activity in purine metabolism leads to ROS formation and thus the potential for oxidative stress. A study of plasma from type 2 diabetics with DR also identified elevated levels of many purines compared with the plasma of diabetics without DR, but a corresponding increase in XO was not observed.21 In contrast, a metabolomics study of vitreous from a mouse model of DR reported decreases in AMP, hypoxanthine, xanthine, and inosine.9 Human DR vitreous samples profiled in the same study revealed a 1.4fold increase in hypoxanthine, but accompanying information on related purines was not obtained.9 Returning to central energy metabolism, we next utilized absolute quantification to closely examine trends within mitochondrial metabolites. Four TCA cycle metabolites (citrate, α-ketoglutarate, succinate, and fumarate) were quantified using spiked-in SIL standards (Figure 5, Table

sample collection in the OR and freezing and the fact that the separation of patients into two subcohorts for analysis generates a small n when considering the impacts of clinical, demographic, and social variables to differential metabolic profiles. Levels of aspartate and linoleic acid have been noted to stratify the stages of DR, with aspartate increased and linoleic acid decreased during progression to proliferative DR,11 although all DR patients analyzed here possess the proliferative form. A deeper examination within energy metabolic pathways reveals that DR is associated with a vitreous glucose metabolism altered in comparison with the vitreous of nondiabetic patients. Despite unchanged total glucose levels, downstream glycolytic metabolites glyceraldehyde 3-phosphate, 2/3-phosphoglycerate, and the lactate to pyruvate ratio are decreased in DR vitreous (Figure 3). Additionally, we observed in DR vitreous a 10.7-fold (DR vs control, p = 0.00076) increased level of pentose phosphate pathway (PPP) product ribose phosphate, suggesting that glucose in these patients is routed away toward the PPP. Late glycolytic intermediates, such as final product lactate, may be formed via PPP-based glucose oxidation, feeding back into glycolysis in the stage of glyceraldehyde 3-phosphate. Although the present analysis is merely based on steady-state measurements for obvious logistic reasons (i.e., stable isotope tracing with appropriately labeled glucose can discern lactate production from glycolysis vs the PPP12), it is interesting to note that activation of the PPP has similarly been noted in the plasma of DR patients13 and in the retina of a rat model of DR upon glucose exposure.14 This metabolic response is observed under conditions of oxidative stress because the PPP is the main NADPH-generating pathway utilized to maintain glutathione homeostasis and fuel antioxidant enzymes. Indeed, a recent study revealed that glucose 6-phosphate dehydrogenase deficiency, affecting the rate-limiting enzyme of the PPP, exacerbates DR in humans.15 The healthy retina is a highly glycolytic tissue with a Warburg-like phenotype.16−18 Although mitochondrial metabolism provides the higher output of ATP per unit of glucose-derived pyruvate, the kinetics of glycolysis are sufficiently faster and provide a robust ATP output despite generating only 2 equiv per molecule of glucose. In addition, we were curious about whether levels of serine and glycine, amino acids whose biological pools are at least partially derived from glycolysis, were altered in DR. Rhodopsin, a key photoreceptor protein rich in serine and glycine residues, experiences rapid light damage requiring continuous regeneration. Rhodopsin levels are decreased in a rat model of DR19 and are closely tied to the phosphorylation of pyruvate kinase M2 (PKM2) that stabilizes the Warburg-like phenotype.20 Interestingly, we did not observe a decrease in serine levels in DR vitreous (glycine was not observed), suggesting that rhodopsin deficits are a result of limiting factors aside from amino acid building blocks, such as altered retinoid metabolism.19 Along with the observation of increased ribose phosphate in the DR vitreous, a biomarker analysis using ROC curves (Figure 1) revealed xanthine (decreased levels) to be the strongest predictor of DR in this sample set. An expanded look at purine metabolism (Figure 4) illustrates several metabolites upstream of xanthineguanine, inosine, and hypoxanthine as increased in the DR group. These findings coupled to decreased levels of xanthine suggest, as one possible explanation, an upregulation of xanthine oxidase (XO),

Figure 5. Absolute quantification of TCA cycle metabolites in human vitreous samples. Molar values were obtained using stable isotopelabeled standards spiked into each sample. Median values are reported in Table S3 (* p < 0.05, ** p < 0.01 by ANOVA).

S3). These metabolites were measured in the micromolar range with the DR group often on the lowest end of the range for the three groups. Long-chain acylcarnitines, derived from fatty acid β-oxidation and fueling carbons to the TCA cycle via acetyl-CoA, were elevated in both the RD and DR groups compared with control (absolute quantification measurements found in Table S1). A previous report using relative (fold change) quantification noted carnitine increases in DR vitreous versus control.9 Examining specifically C8 and C10 acylcarnitines and their monounsaturated counterparts, we observe a sharp and statistically significant increase in the vitreous of DR patients: octanoylcarnitine (3.3-fold increase, p = 2.25 × 10−4), octenoylcarntine (5.8-fold, p = 6.48 × 10−4), decanoylcarnitine (4.1-fold, p = 8.65 × 10−5), and decenoylcarnitine (2.2-fold, p = 1.15 × 10−3). This trend has been similarly noted in the urine and retina of mouse models of diabetes.10 The final area of focus for absolute concentration measurements was amino acids. These metabolites were present in the vitreous at the tens of micromoles level and tended to be highest in the DR group, with only a few exceptions, aspartate, threonine, and tyrosine (median values reported in Table S3). The vitreous of RD patients tended to have ∼1.5-fold higher levels of amino acids relative to control patients, whereas the amino acid levels in DR patients were approximately two-fold elevated compared with control. Trends of such similar magnitude within a metabolite class imply connectivity in the molecular mechanism; the possibility exists that increased free amino acid content in the RD patients and even more so 2425

DOI: 10.1021/acs.jproteome.8b00169 J. Proteome Res. 2018, 17, 2421−2427

Article

Journal of Proteome Research Present Address

in the DR cohort could be the result of imbalanced protein homeostasis in the eye or cell death and subsequent release of intracellular metabolites into the vitreous. The sharpest increase within the amino acids was for proline in the DR cohort, matching previous observations by us and others.9 A crucial consideration when examining the full scope of the vitreous metabolome of DR in comparison with the vitreous of nondiabetic patients presented here is the relative age of the cohorts profiled. Because of the presence of type 1 diabetes or >10 years since diagnosis of diabetes, the DR patients have an average age of ∼20 years less than the other two groups (see Table S2 for a full list of patient ages), whose eye diseases occur later in life.22−26 Age is a well-established factor impacting the metabolome of systemic fluids like plasma and urine,27,28 and thus an area of high interest for subsequent studies involves disentangling the factors of age and diabetes (to the extent possible) and their relative contributions to the steady-state profile of DR vitreous.

§

N.R.H.: Piedmont Retina Specialists, Greensboro, North Carolina, United States. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Support for this work was provided by the National Institutes of Health S10 (OD021641) and the Boettcher Webb-Waring Biomedical Research Award, both to A.D. We are grateful to Marc Mathias, M.D. and Frank Siringo, M.D., O.D. for their participation in vitreous sample collection.



(1) Tosi, G. M.; Marigliani, D.; Romeo, N.; Toti, P. Disease pathways in proliferative vitreoretinopathy: an ongoing challenge. J. Cell. Physiol. 2014, 229, 1577−1583. (2) Agarwal, A.; Soliman, M. K.; Sepah, Y. J.; Do, D. V.; Nguyen, Q. D. Diabetic retinopathy: variations in patient therapeutic outcomes and pharmacogenomics. Pharmacogenomics Pers. Med. 2014, 7, 399− 409. (3) Bolinger, M. T.; Antonetti, D. A. Moving past anti-VEGF: Novel therapies for treating diabetic retinopathy. Int. J. Mol. Sci. 2016, 17, 1498. (4) Lauwen, S.; de Jong, E. K.; Lefeber, D. J.; den Hollander, A. Omics biomarkers in ophthalmology. Invest. Ophthalmol. Visual Sci. 2017, 58, Bio88−bio98. (5) Nemkov, T.; Hansen, K. C.; D’Alessandro, A. A three-minute method for high-throughput quantitative metabolomics and quantitative tracing experiments of central carbon and nitrogen pathways. Rapid Commun. Mass Spectrom. 2017, 31, 663−673. (6) Nemkov, T.; D’Alessandro, A.; Hansen, K. C. Three-minute method for amino acid analysis by UHPLC and high-resolution quadrupole orbitrap mass spectrometry. Amino Acids 2015, 47, 2345− 2357. (7) McCurdy, C. E.; Schenk, S.; Hetrick, B.; Houck, J.; Drew, B. G.; Kaye, S.; Lashbrook, M.; Bergman, B. C.; Takahashi, D. L.; Dean, T. A.; Nemkov, T.; Gertsman, I.; Hansen, K. C.; Philp, A.; Hevener, A. L.; Chicco, A. J.; Aagaard, K. M.; Grove, K. L.; Friedman, J. E. Maternal obesity reduces oxidative capacity in fetal skeletal muscle of Japanese macaques. JCI Insight 2016, 1, e86612. (8) Barba, I.; Garcia-Ramirez, M.; Hernandez, C.; Alonso, M. A.; Masmiquel, L.; Garcia-Dorado, D.; Simo, R. Metabolic fingerprints of proliferative diabetic retinopathy: a 1H-NMR-based metabonomic approach using vitreous humor. Invest. Ophthalmol. Visual Sci. 2010, 51, 4416−4421. (9) Paris, L. P.; Johnson, C. H.; Aguilar, E.; Usui, Y.; Cho, K.; Hoang, L. T.; Feitelberg, D.; Benton, H. P.; Westenskow, P. D.; Kurihara, T.; Trombley, J.; Tsubota, K.; Ueda, S.; Wakabayashi, Y.; Patti, G. J.; Ivanisevic, J.; Siuzdak, G.; Friedlander, M. Global metabolomics reveals metabolic dysregulation in ischemic retinopathy. Metabolomics 2016, 12, 15. (10) Mirzoyan, K.; Klavins, K.; Koal, T.; Gillet, M.; Marsal, D.; Denis, C.; Klein, J.; Bascands, J. L.; Schanstra, J. P.; Saulnier-Blache, J. S. Increased urine acylcarnitines in diabetic ApoE−/− mice: Hydroxytetradecadienoylcarnitine (C14:2-OH) reflects diabetic nephropathy in a context of hyperlipidemia. Biochem. Biophys. Res. Commun. 2017, 487, 109−115. (11) Li, X.; Luo, X.; Lu, X.; Duan, J.; Xu, G. Metabolomics study of diabetic retinopathy using gas chromatography-mass spectrometry: a comparison of stages and subtypes diagnosed by Western and Chinese medicine. Mol. BioSyst. 2011, 7, 2228−2237. (12) Reisz, J. A.; Wither, M. J.; Dzieciatkowska, M.; Nemkov, T.; Issaian, A.; Yoshida, T.; Dunham, A. J.; Hill, R. C.; Hansen, K. C.; D’Alessandro, A. Oxidative modifications of glyceraldehyde 3phosphate dehydrogenase regulate metabolic reprogramming of stored red blood cells. Blood 2016, 128, e32−42.



CONCLUSIONS We provide a global analysis of central carbon and nitrogen metabolism of vitreous from two clinical cohorts of patients suffering from either DR or RD. Although preliminary, the study paves the way for future follow-up investigations to further our understanding of the molecular drivers of these diseases. Here candidate markers of retinal pathologies were identified, verified with absolute quantitation, and validated in an independent prospective cohort. Because vitreous collection is a routine medical procedure and quantitative targeted massspectrometry-based small-molecule analysis is readily available, this study will inform future clinical trials to validate the validity of the present findings, with the ultimate goal of providing new avenues for the clinical treatment of retinal pathologies.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jproteome.8b00169.



REFERENCES

Table S1. Identified metabolites in human vitreous samples with retention times on 9 min gradient C18 method. (XLSX) Table S2. Demographics and clinical characteristics of participating patients. (XLSX) Table S3. Absolute quantification values for TCA cycle intermediates, amino acids, and acylcarnitines in human vitreous samples. Figure S1. Hierarchical clustering analysis of significantly altered metabolites in human vitreous. Figure S2. ROC curves for the validation cohort. (PDF)

AUTHOR INFORMATION

Corresponding Author

*Tel: 303-724-3339. E-mail: [email protected]. ORCID

Angelo D’Alessandro: 0000-0002-2258-6490 Julie A. Reisz: 0000-0002-7296-4963 2426

DOI: 10.1021/acs.jproteome.8b00169 J. Proteome Res. 2018, 17, 2421−2427

Article

Journal of Proteome Research (13) Chen, L.; Cheng, C. Y.; Choi, H.; Ikram, M. K.; Sabanayagam, C.; Tan, G. S.; Tian, D.; Zhang, L.; Venkatesan, G.; Tai, E. S.; Wang, J. J.; Mitchell, P.; Cheung, C. M.; Beuerman, R. W.; Zhou, L.; Chan, E. C.; Wong, T. Y. Plasma metabonomic profiling of diabetic retinopathy. Diabetes 2016, 65, 1099−1108. (14) Coffe, V.; Carbajal, R. C.; Salceda, R. Glucose metabolism in rat retinal pigment epithelium. Neurochem. Res. 2006, 31, 103−108. (15) Cappai, G.; Songini, M.; Doria, A.; Cavallerano, J. D.; Lorenzi, M. Increased prevalence of proliferative retinopathy in patients with type 1 diabetes who are deficient in glucose-6-phosphate dehydrogenase. Diabetologia 2011, 54, 1539−1542. (16) Atsuzawa, K.; Nakazawa, A.; Mizutani, K.; Fukasawa, M.; Yamamoto, N.; Hashimoto, T.; Usuda, N. Immunohistochemical localization of mitochondrial fatty acid beta-oxidation enzymes in Muller cells of the retina. Histochem. Cell Biol. 2010, 134, 565−579. (17) Ng, S. K.; Wood, J. P.; Chidlow, G.; Han, G.; Kittipassorn, T.; Peet, D. J.; Casson, R. J. Cancer-like metabolism of the mammalian retina. Clin. Exp. Ophthalmol. 2015, 43, 367−376. (18) Sas, K. M.; Kayampilly, P.; Byun, J.; Nair, V.; Hinder, L. M.; Hur, J.; Zhang, H.; Lin, C.; Qi, N. R.; Michailidis, G.; Groop, P. H.; Nelson, R. G.; Darshi, M.; Sharma, K.; Schelling, J. R.; Sedor, J. R.; Pop-Busui, R.; Weinberg, J. M.; Soleimanpour, S. A.; Abcouwer, S. F.; Gardner, T. W.; Burant, C. F.; Feldman, E. L.; Kretzler, M.; Brosius, F. C., 3rd; Pennathur, S. Tissue-specific metabolic reprogramming drives nutrient flux in diabetic complications. JCI Insight 2016, 1, e86976. (19) Malechka, V. V.; Moiseyev, G.; Takahashi, Y.; Shin, Y.; Ma, J. X. Impaired rhodopsin generation in the rat model of diabetic retinopathy. Am. J. Pathol. 2017, 187, 2222−2231. (20) Rajala, R. V.; Rajala, A.; Kooker, C.; Wang, Y.; Anderson, R. E. The Warburg effect mediator pyruvate kinase M2 expression and regulation in the retina. Sci. Rep. 2016, 6, 37727. (21) Xia, J.; Wang, Z.; Zhang, F. Association between related purine metabolites and diabetic retinopathy in type 2 diabetic patients. Int. J. Endocrinol. 2014, 2014, 1. (22) Hletala, K.; Harjutsalo, V.; Forsblom, C.; Summanen, P.; Groop, P.-H. on behalf of the FinnDiane Study, G. Age at Onset and the Risk of Proliferative Retinopathy in Type 1 Diabetes. Diabetes Care 2010, 33, 1315−1319. (23) Feltgen, N.; Walter, P. Rhegmatogenous Retinal Detachment an Ophthalmologic Emergency. Deutsches Ä rzteblatt International 2014, 111, 12−22. (24) Potic, J.; Bergin, C.; Giacuzzo, C.; Daruich, A.; Konstantinidis, L.; Wolfensberger, T. J. Primary rhegmatogenous retinal detachment: risk factors for macular involvement. Graefe's Arch. Clin. Exp. Ophthalmol. 2018, 256, 489−494. (25) Cheung, N.; Tan, S.-P.; Lee, S. Y.; Cheung, G. C. M.; Tan, G.; Kumar, N.; Cheng, C.-Y.; Wong, T. Y. Prevalence and risk factors for epiretinal membrane: the Singapore Epidemiology of Eye Disease study. Br. J. Ophthalmol. 2016, 101, 371−376. (26) Fraser-Bell, S.; Guzowski, M.; Rochtchina, E.; Wang, J. J.; Mitchell, P. Five-year cumulative incidence and progression of epiretinal membranes: the Blue Mountains Eye Study. Ophthalmology 2003, 110, 34−40. (27) Rist, M. J.; Roth, A.; Frommherz, L.; Weinert, C. H.; Kruger, R.; Merz, B.; Bunzel, D.; Mack, C.; Egert, B.; Bub, A.; Gorling, B.; Tzvetkova, P.; Luy, B.; Hoffmann, I.; Kulling, S. E.; Watzl, B. Metabolite patterns predicting sex and age in participants of the Karlsruhe Metabolomics and Nutrition (KarMeN) study. PLoS One 2017, 12, e0183228. (28) Trabado, S.; Al-Salameh, A.; Croixmarie, V.; Masson, P.; Corruble, E.; Feve, B.; Colle, R.; Ripoll, L.; Walther, B.; BoursierNeyret, C.; Werner, E.; Becquemont, L.; Chanson, P. Chanson, P. The human plasma-metabolome: Reference values in 800 French healthy volunteers; impact of cholesterol, gender and age. PLoS One 2017, 12, e0173615.

2427

DOI: 10.1021/acs.jproteome.8b00169 J. Proteome Res. 2018, 17, 2421−2427