If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
The objective of the study was to clarify the interplay between metabolites and microRNAs (miRs) in the aqueous humor (AqH) of bullous keratopathy (BK) patients to retain human corneal endothelium (HCE) integrity.
Design
Prospective, comparative, observational study.
Participants
55 patients with BK and 31 patients with cataract (Cat) as control.
Methods
A biostatic analysis of miRs and metabolites in the aqueous humor (AqH), hierarchical clustering and a least absolute shrinkage and selection operator (Lasso) analysis were employed. The miR levels in AqH of bullous keratoplasty (BK, n=18) and cataract (Cat, n=8) patients were determined using 3D-Gene human miR chips. Hierarchical clusters of metabolites detected by LC/MS or GC/MS in AqH specimens from two disease groups, BK (total n=55) and Cat (total n=31), were analyzed twice to confirm the reproducibility. The analytical procedure applied for investigating the association between metabolites and miRs in AqH was the exploratory data analysis of biostatistics to avoid any kind of prejudice. This research procedure includes a heat-map, cluster analysis, feature extraction techniques by principal component analysis (PCA) and a regression analysis method by Lasso. The cellular and released miR levels were validated using RT-PCR and mitochondria membrane potential were assessed to determine the functional features of the released miRs.
Main Outcome Measures
Identification of interacting metabolites and miRs in AqH attenuating HCE degeneration.
Results
The metabolites that decreased in the AqH of BK patients revealed that 3-hydroxyisobutyric acid (HIB), 2-aminobutyric acid (AB) and branched-chain amino acids, and serine were categorised into the same cluster by hierarchical clustering of metabolites. The positive association of HIB with miR-34a-5p was confirmed (p=0.018), and the Lasso analysis identified the interplay between miR-34a-5p and HIB, between miR-24-3p and AB and between miR-34c-5p and serine (p=0.041, 0.027, 0.009, respectively). HIB upregulated the cellular miR-34a expression, mitochondrial membrane potential and release of miR-184 in de-differentiated cultured human corneal endothelial cells.
Conclusions
Metabolites and miRs in AqH may synchronize in ensuring the integrity of the HCE to maintain efficient dehydration from the stroma.
Writing Committee for the Cornea Donor Study Research Group Donor age and factors related to endothelial cell loss 10 years after penetrating keratoplasty: Specular Microscopy Ancillary Study.
Five-year postoperative results revealed that this cultured HCE cell injection therapy has an effective and long-lasting stable outcome against diverse HCE failures, including Fuchs’ endothelial corneal dystrophy (FECD). Recently, we revealed the new mechanisms regulating the pathological progression of HCE failures.
The upregulated expression of CD44 through the repression of micro-RNA (miR)-34a by reactive oxygen species (ROS) and elevation of c-Myc by oxidative stress, may impair mitochondrial bioenergetics, resulting in HCE failures, accompanied by severe stromal edema.
HCECs sharing a CD44-/dull differentiated phenotype may be discriminated by the level of cellular miRs or miRs released included in the extracellular vesicle (EVs).
In succession, we described recently that failure in the cell competition for space occupancy, between degenerated and non-degenerated cells in a single layer tissue, through EV miRs released by HCE cells is a partial cause of the pathogenesis of BK, including FECD.
The elevated release of miR-184 and -24-3p into aqueous humor (AqH) may be involved in a cellular interplay to dampen the vicious cycle of HCE degeneration induced by endoplasmic reticulum (ER) stress.
miRs are potential biomarkers for the diagnosis of a wide range of tissue disorders.
On the other hand, many metabolites in AqH might also participate in the epigenetic regulation of HCE cell fate decisions. Metabolomics refers to the detection of metabolites to assign disease-specific metabolic signatures.
By analysing the plasma metabolomic profile, Barca et al. clarified the mitochondrial energetic impairments involved in the pathogenesis of age-related macular degeneration (AMD).
Further, Yamaguchi et al. explored the usefulness of the multi-omics landscape of AqH in patients with BK and found that mitochondrial energy-producing proteins were significantly decreased.
Aqueous Humor Analysis Identifies Higher Branched Chain Amino Acid Metabolism as a Marker for Human Leukocyte Antigen-B27 Acute Anterior Uveitis and Disease Activity.
Intracellular miR-34a is expressed selectively and exclusively in differentiated cultured HCE cells, which exert a long-term sustained clinical efficacy in the cell injection regenerative medicine developed by us.
The differentiated mature cultured HCE cells are disposed to mitochondria-dependent oxidative phosphorylation (OXPHOS), whereas de-differentiated cultured HCE cells are inclined to a glycolytic metabotype.
In the current study, we had tried to verify our hypothesis that the synchronized collaboration between extracellularly released miRs and a branched chain amino acid (BCAA) metabolite in AqH may function concertedly in maintaining cellular mitochondria phenotypes to protect the homeostasis and integrity of HCE tissues.
MATERIALS AND METHODS
Patients and Approval of Acquisition of Aqueous Humors
The acquisition of AqH and the experimental study protocols described were approved by the Institutional Review Board of Kyoto Prefectural University of Medicine, Kyoto, Japan (Approval No. ERB-C-245-8). The human tissue used in this study was handled in accordance with the tenets set forth in the Declaration of Helsinki. HCE cells were obtained, including informed written consent for eye donation for research, from human donor corneas supplied by CorneaGen (Seattle, WA, USA) Eye Bank. All procedures were conducted in accordance with the ARVO Statement for the Use of Human Materials in Ophthalmic and Vision Research, and all the details of the current experimental study protocols were approved by the Institutional Ethical Committee of Kyoto Prefectural University of Medicine, Kyoto, Japan. All patients were recruited from the University Hospital Kyoto Prefectural University of Medicine or the Baptist Eye Institute between February 2014 and October 2018. All the patients received routine clinical diagnosis, including biochemical blood test, and no serious metabolic diseases, such as diabetes, was found.
To determine metabolite levels by LC/MS and GC/MS-based metabolomic analysis, BK patients I (n=21, Table-S1), BK patients II (n=34, Table-S2-1) and corneal dysfunctional failure (CDF) patients (n=7, Table-S2-2) and control cataract (Cat) patients without CDFs (n=31 Table-S3) were enrolled. CDF patients were composed of 3 keratoconus, 2 corneal opacity, 1 corneal scarring and 1 corneal dystrophy, who received keratoplasty. For reference, AqH specimens of patients with pseudoexfoliation glaucoma (PEG) were included (n=29, Table-S4). To determine the expression levels of miRs in the DNA microarray of 3D-Gene® (Toray Industries, Inc., Tokyo, Japan), 18 BK patients (Table-S5) and eight Cat patients without CDFs (Table-S6) were additionally enrolled. For the Lasso analysis between metabolites and miRs, AqH from 29 patients (27 BK and 2 CDF, Table-S7) from the enrolled patients was arbitrary collected and analysed. The study was conducted in accordance with the tenets set forth in the Declaration of Helsinki. Written informed consent was obtained from all participants following the provision of a detailed explanation of the study protocol, including AqH collection. All experiments were performed in accordance with the institutional guidelines.
Human aqueous humor
AqH was obtained at the beginning of surgery without blood contamination using a specially designed 30-gauge needle integrated with a disposable pipette (Nipro, Osaka, Japan), as described previously,
or a disposable 1-mL syringe with a 30-gauge needle. Approximately 150 μl of AqH was collected, immediately frozen and stored at −80°C until analysis. The turnover of AqH is several minutes, therefore, we controlled the medication strictly from 24 h before the surgery for BK patients (refer to ST-1 and 2 for medication).
HCE donors, cell cultures of HCE cells and reagents
The human tissue used in this study was handled and cultured as detailed in previous publications.
Aqueous Humor Analysis Identifies Higher Branched Chain Amino Acid Metabolism as a Marker for Human Leukocyte Antigen-B27 Acute Anterior Uveitis and Disease Activity.
The former SPs elicited a proportion of CD44-/dull greater than 95%, while the latter SPs elicited a proportion less than 70%. The cultured HCE cells at passages 2 to 5 were used for all experiments. All the cultured HCE cells were primary cells prepared from the imported HCE tissues from different donors. To gain the cultured HCE cell SPs distinct in the levels of cell surface CD44 expression, we have used the different time periods or culture passages to gain the sufficient number of cells for the experiments. Corneal endothelial cell (CEC) culture models had been carried out under almost confluent culture conditions (>90 %) to analyze the role of HIB on miR expression and secretion. L-AB was bought from Sigma-Aldrich (St. Louis, MO, USA) and Dl-3-hydroxyisobutyric acid from Chemodex (St. Gallen, Switzerland).
RNA extraction and miRNA profiling 3D-Gene® microarray analysis
For miR expression profiling, 3D-Gene® Human miRNA Oligo Chips (miRBase version 17-19; Toray Industries Inc.) were used and analysed as described previously.
All the data were globally normalized per a microarray, such that the median of the signal intensity was adjusted to 25.
Metabolomic Analysis
For metabolomics, 30 μl of an AqH sample was mixed with 154 μl of methanol containing 2.34 μg/ml of 2-isopropylmalic acid (Sigma-Aldrich Japan, Tokyo, Japan), which was utilised as an internal standard. The obtained mixture was shaken at 1,200 rpm for 10 min at 37°C (Maximizer MBR-022UP, Taitec, Koshigaya, Japan). After centrifugation at 16,000 × g for 20 min at 25°C, 120 μl of the supernatant was mixed with 72 μl of 1% acetic acid in water and 96 μl of chloroform, followed by vortex mixing for 15 s. After centrifugation at 2,000 × g for 10 min at 25°C, the upper layer was divided into two aliquots (each 60 μl) and dried in a centrifugal evaporator (CVE-3100, Tokyo Rikakikai Co. Ltd., Tokyo, Japan). One of the dried aliquots was dissolved in 40 μl of a methoxamine solution (20 mg/mL in pyridine, Sigma-Aldrich Japan, Tokyo, Japan) and shaken at 1,200 rpm for 30 min at 37°C. Twenty microliters of N-methyl-N-(trimethylsilyl) trifluoroacetamide (GL Science, Tokyo, Japan) were added for trimethylsilyl derivatisation, followed by agitation at 1,200 rpm for 30 min at 37°C. After centrifugation, 50 μl of the supernatant was transferred to a glass vial and subjected to GC/MS measurement. For LC/MS analysis, the other dried aliquot was solubilised in 25 μl of 0.1% formic acid in water and then subjected to LC/MS analysis.
GC/MS analysis was performed with a GCMS-QP2010 Ultra (Shimadzu, Kyoto, Japan). The derivatised metabolites were separated on a DB-5 column (30 m × 0.25 mm id, film thickness 1.0 μm, Agilent Technologies, Santa Clara, CA, USA). The helium carrier gas was set at a flow rate of 39 cm/s. The inlet temperature was 280°C and the column temperature was first held at 80°C for 2 min, then raised at a rate of 15°C/min to 330°C and held for 6 min. One microliter of the sample was injected into the GC/MS in the split mode (split ratio 1:3). The mass spectra were obtained under the following conditions: electron ionisation (ionisation voltage 70 eV), ion source temperature 200°C, interface temperature 250°C, full scan mode in the range of m/z 85–500 and scan rate 0.3 s/scan. Chromatographic peak identification was performed using the NIST library or the Shimadzu GC/MS database and further confirmed with authentic commercial standards. LC separation was conducted on a Shim-pack GIST C18-AQ column (3 μm, 150 mm × 2.1 mm id, Shimadzu GLC, Kyoto, Japan) with a Nexera UHPLC system (Shimadzu). The mobile phase consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B). The gradient programme was as follows: 0–3 min, 0% B; 3–15 min, linear gradient to 60% B; 15–17.5 min, 95% B; 17.5–20.0 min, linear gradient to 0% B; hold for 4 min; flow rate, 0.2 ml/min. The column oven temperature was maintained at 40°C, and the LC system was coupled with a triple-quadruple mass spectrometer, the LCMS-8060 (Shimadzu), which was operated in the electrospray ionisation and multiple reaction monitoring mode. All ion transitions and collision energies were optimized experimentally using authentic standards of each metabolite. For semi-quantitative analysis, the area of each metabolite peak was calculated and divided by the area of the internal standard peak. 96 and 114 metabolites were successfully measured in GC/MS and LC/MS, respectively, whereas the amounts of 0 and 26 metabolites, respectively, were lower than the detection threshold.
Biostatic Analysis of miRNAs and Metabolites in AqH
This study was an exploratory data analysis with unknown effect sizes and confidence intervals for the hypotheses to be tested, and no statistical sample size calculations were performed. On the other hand, the statistical comparison of 29 samples is considered a medium sample size
as the research of miR biomarkers for eye disease, and it is presumed to reveal an association between miR levels and metabolites in AqH of BK patients.
Procedures to investigate the association between metabolites and miRs
In this research, the analytical procedure for investigating the association between metabolites and miRs in AqH applied was the exploratory data analysis (EDA) of biostatistics to avoid any kind of prejudice. This research procedure includes a heat-map, cluster analysis, feature extraction techniques by PCA and a regression analysis method by Lasso. The flow chart of this study is diagrammed in Fig. 1.
Fig. 1Flow chart of the exploratory data analysis of metabolites and miRs in the AqH of patients. AqH: aqueous humor, Lasso: least absolute shrinkage and selection operator, miR: miRNA, PCA: principal component analysis, ROC: receiver operating characteristic.
The HCA was used to simplify the classification of metabolites or miRs in AqH. Dendrograms supplemented were used to show the hierarchical relationships between samples, and HCA was performed using a Euclidean distance matrix and a criterion of object similarity applied using Ward’s linkage method.
Principal component analysis
The new orthogonal variables calculated by PCA are explained by a dimensionality reduction set of uncorrelated data, called principal components (PCs). The interpretation of the PCA results was given by the projection of an individual sample onto the axes defined by PCs, termed the ‘score’. The PCA score plot indicates the sample similarity, depending on the distance, that is whether plotted close to or further apart from each other. In addition, the loading plot indicated in the PCA scatter plots was commonly used to examine relations between variables. The magnitude of the vectors shows the strength of their contribution to each component, and the closeness to each other indicates the high correlation between them. In total, 20 metabolites were extracted based on the extent of the amount changed from the control group. The PCA biplot shows the loading vectors of the metabolite in an arrow and the PCA score of the samples. Prior to PCA, all metabolite levels were normalized using the standardised data by subtracting the mean and dividing by the standard deviation.
The least absolute shrinkage and selection operator
The Lasso logistic regression analysis was applied to select metabolites with a close association to miRs or inversely to select miRs with a close association with metabolites. Prior to an analysis by Lasso, all these metabolites and miR levels were classified into two categories (binarisation; larger or smaller than the median values in all BK patients). We used univariate analysis to validate the association of b variables selected by the Lasso regression analysis. In addition, the ROC curve was investigated to evaluate the association of multiple metabolites to a selected miR or of multiple miRs to a selected metabolite.
Statistics
Metabolite levels in AqH were analysed using the Wilcoxon rank sum test, and the prognostic validity of the metabolite or miR levels was evaluated by analysing the ROC curve for the median of the values observed for all data samples, as measured using the AUC. The statistical analyses above were performed using R v.4.1.2 (The R Foundation for Statistical Computing, Vienna, Austria) with the R package exactRankTests, and heat-map images were generated using the R package gplots. A PCA was executed with the R package stats, and a Lasso regression model analysis was executed with the R package glmnet. Analysis of the ROC curves and display was executed with the R package pROC, and differences were considered significant at p<0.05.
Quantitative Real-Time Polymerase Chain Reaction and MicroRNA Expression Profiling
Polymerase chain reaction (PCR) was performed under the previously described conditions.
The levels of miR-34a-5p were normalised to that of glyderaldehyde-3-phosphate dehydrogenase (GAPDH), and the results were presented as 2-ΔCt (relative units of expression). The primer used is the miR-34a assay ID 000426 (Taqman micro assays, Thermo Fisher Scientific, Waltham, MA, USA).
Mitochondrial Respiration Assay
A real-time metabolic analysis of live CECs was performed using the Seahorse XFe24 extracellular flux analyser (Agilent Technologies, Santa Clara, CA, USA). Cultured HCE cells were seeded on an XF24 flux analyser plate. The Mito Stress test was performed according to the manufacturer’s protocol, where a cell culture medium was replaced 1 hour before the assay with a minimal XF DMEM medium supplemented with 2 mM glutamine, 10 mM glucose and 1mM sodium pyruvate (pH 7.4). The OCR and extracellular acidification rate (ECAR) were analysed at basal conditions and after sequential injections of 1 μM oligomycin, 1 μM FCCP and 0.5 μM rotenone and antimycin A. The assay results were normalised based on the viable cell number, counted by Cell Insight NXT (Thermo Fisher Scientific).
Mitochondrial membrane potential
A change in the mitochondrial membrane potential (MMP) was detected using the JC-1 MitoMP Detection Kit (Dojindo Laboratories, Kumamoto, Japan). After the treatment, cells were harvested by the TrypLE Select (Thermo Fisher Scientific) treatment and suspended at 106 cells/mL in medium. Collected cells were incubated with 2 μM JC-1 for 30 min at 37°C. After washing with Hanks balanced salt solution, the cells were analysed using the BD FACSCanto II Flow Cytometry System (BD Biosciences, Franklin Lakes, NJ, USA). For the fluorescence imaging analysis, cells were incubated with 2 μM JC-1 for 30 min at 37°C and analysed using the BZ X-700 Microscope System (Keyence Corporation, Osaka, Japan).
Statistical Analysis of Functional Analysis
Data are presented as the means ± SEM, and a statistical analysis of differences was performed using the student’s t-test (comparison between two groups) or analysis of variance followed by Tukey’s or Dunnett's test (comparison among two groups). Values shown on the Figure graphs represent the mean ± SE.
DATA AVAILABILITY
All data generated or analysed during this study are included in this published article and its supplementary information files. Further data are available from the corresponding author upon reasonable request.
RESULTS
The metabolites in the AqH of patients who received corneal transplantation or cell injection therapy
The hierarchical clustering method of metabolites in AqH was performed using a criterion of object similarity. Dendrograms were used to represent the results of hierarchical clustering, and they clarified the hierarchical relationships between metabolites. The horizontal and vertical dimensions correspond to patients and metabolite levels, respectively (Supplementary Fig. S2). Hierarchical clustering was constructed to visualize the differences in metabolite levels in the AqH of BK patient group I (BK I), group II (BK II) versus control Cat patients (Fig. S2 a, b). A clinical summary of these patients is listed in Table S1-S3. In BK patients (I and II), four (Fig. S2a metabolites detected by liquid chromatography [LC]/mass spectrometry [MS]) and three (Fig. S2b, those by gas chromatography [GC]/MS) metabolite clusters were generated, respectively. The metabolite levels were increased in cluster MLC2 and moderately in MGC3 in a BK patient group, whereas they decreased in MLC3 and moderately in MGC2 (Fig. S2a, b). Carnitine and acylcarnitines were included in MLC2, while cystine (Cys2), cysteine (Cys), β-alanine (Ala), ornithine (Orn), a metabolite of valine (Val), 3-hydroxyisobutyric acid (HIB), 2-aminobutyric acid (AB) and citric acid were included in MGC3 cluster. Further, spermine, spermidine, phosphocholine, uridine, nicotinamide and putrescine were identified in cluster MLC3, while the branched chain amino acids (BCAAs) leucine (Leu), isoleucine (Ile), Val and Ser were identified in MGC2.
The clusters of metabolites in AqH are BK specific
To clarify that the observed profiles of metabolites in AqH are dependent on the disease phenotypes, but not non-specific, we compared the hierarchical clustering constructed for metabolites in the AqH of seven non-BK CDF versus BK patients, where both groups received keratoplasty (Fig. S3a, b, Table-S2). The metabolite levels increased more in the AqH of BK patients in cluster MLC2/3 than in that of non-BK patients, whereas they slightly decreased in MLC4/5. Most of the metabolites extracted from these clusters showed a clear distinction between the AqH of BK and non-BK CDF patients, indicating clearly the presence of BK specific clusters.
The clustering of patient diseases with regard to metabolites in AqH
The metabolite levels in AqH were summarized in an expression cluster heat-map with a cluster of BK (n = 21), non-BK CDF (n = 7) and Cat (n = 31) patients. The horizontal and vertical dimensions correspond to patients and metabolite levels, respectively. Dendrograms showed the hierarchical relationships between metabolites and patients, where the metabolite levels detected by LC/MS and GC/MS in the three target patient groups - BK, non-BK-CDF and Cat - are summarized in the expression cluster heat-maps (Fig. S4a, b). The clinical summary of the patients is presented in Table-S1, -S2 and -S3. At first glance, it is clear that the clusters formed in the vertical direction did not classify the patients into disease categories. BK (red coloured in circle diagram) and Cat (blue coloured) patients were both scattered among three vertical clusters in (a) LC/MS- and (b) GC/MS-detected metabolites. Therefore, we decided to construct in the following analysis the hierarchical clustering to visualize differences in only the metabolite levels, instead of the hierarchical clustering of both metabolites and patients.
The usefulness of heat map analysis of metabolites in the AqH of BK patients
Next, we compared the metabolites of BK patients with control Cat patients. Figure 1, a diagram of the exploratory data analysis of the metabolites and miRs in AqH is depicted as a flow chart. The metabolite levels in two target patient groups (BK I [n = 21] and Cat patients [n = 31]) are summarized in the expression cluster heat-maps (Fig.5.a, LC/MS, b, GC/MS). Spermine, spermidine, uridine, nicotinamide, putrescine, Ser and BCAAs were included in the MLC3 cluster and succinic acid, scyllo-Inositol, putrescine, uric acid, lactic acid and ascorbic acid in the MGC2 cluster. To solidify the observed results, the metabolite levels of other BK patients II (n = 34) and Cat patients (n = 31) were similarly analysed (Fig. 5.c, d). The metabolite levels were increased in the BK group in clusters MLC3 and MGC2, and they decreased in clusters MLC4 and MGC5/6. The metabolites that increased or decreased in BK patients (I and II) are summarised in Tables 8a and 8b, showing the similar skewing of metabolite profiles. The results indicated these heat-maps exhibited useful metabolite cluster patterns to distinguish the metabolites profiles in the AqH of patients with BK from those in the Cat patients.
Fig. 5Differential expressions of metabolites with regard to disease categories. (a) Heat-map cluster of 114 metabolites detected by LC/MS and 52 AqH specimens from two disease groups (BK I, n = 21 and Cat n = 31). (b) Heat-map cluster of 96 metabolites detected by GC/MS and the same 52 AqH specimens described in (a). (c) Heat-map cluster of 114 metabolites detected by LC/MS and 65 AqH specimens from two disease groups (BK II, n = 34 and Cat n = 31). (d) Heat-map cluster of 96 metabolites detected by GC/MS and the same 65 AqH specimens described in (c). AqH: aqueous humor, BK: bullous keratopathy, Cat: cataract, GC/MS: gas chromatography/mass spectrometry, LC/MS: liquid chromatography/mass spectrometry.
Table 8The identification of metabolites in the AqH of BK—including FECD—patients. (a) BK Ⅰ (21) versus Cataract (31)
LC/MS
GC/MS
Cluster MLC3
Cluster MLC4
Cluster MGC5
Cluster MGC2
Spermine
Propionylcarnitine
Glycine
Succinic acid
Spermidine
2-Methylbutyrylcarnitine(C5)
Hypotaurine
Glycolic acid
Uridine
Butyrylcarnitine(C4)
Citric acid
Suberic acid
Phosphocholine
Acetylcarnitine
Ornithine
Pipecolinic acid
Nicotinamide
Isobutyrylcarnitine(C4)
Proline
scyllo-Inositol
Putrescine
Cystine
Aconitic acid
Pantothenic acid
Tyrosine
Isovalerylcarnitine(C5)
Cysteine
Hypoxanthine
Tryptophan
Proline
Ribose
Acetoacetic acid
Phenylalanine
Carnitine
Fucose
Putrescine
Histidine
Acetylcholine
1,5-Anhydro-D-sorbitol
Serine
Hexanoylcarnitine(C6)
Uric acid
Glutamic acid
Uric acid
Tryptophan
Threonine
N-Acetylarginine
2-Oxoisocaproic acid
2-Aminoadipic acid
Trimethyllysine
Linoleic acid
Isoleucine
Quinolinic acid
Oleic acid
Leucine
N6-Acetyllysine
3-Methyl-2-oxobutyric acid
Valine
Epinephrine
Lactic acid
trans-urocanic acid
2-Hydroxyisobutyric acid
Adenosine
Caproic acid
4-Hydroxybenzoic acid
Glucuronic acid
Inosine
Ascorbic acid
Guanosine
Riboflavin
Urea
2-Hydroxybutyric acid
2-Aminobutyric acid
Creatine
Methionine
Asparagine
Alanine
asy-Dimethylarginine
Lysine
Glutamine
Arginine
(b) BK Ⅱ (34) versus Cataract (31)
LC/MS
GC/MS
Cluster MLC4
Cluster MLC3
Cluster MGC2
Cluster MGC5
Nicotinamide
4-Hydroxyhippuric acid
Cysteine
myo-Inositol
Taurine
Ornithine
Cystine
Arabinonic acid
Uridine
Citrulline
Gluconic acid
Fructose
2-Oxoglutaric acid
Cystine
Ribose
Sorbitol
Spermine
Acetylcholine
Maltose
Arabinose
Spermidine
Proline
Sucrose
Adenine
Putrescine
4-Hydroxyproline
2-Deoxytetronic acid
Glucuronic acid
Phosphocholine
3-Aminoisobutyric acid
Glycine
Ascorbic acid
Lactic acid
Betaine
Proline
Tryptophan
Lysine
Pipecolinic acid
4-Hydroxyproline
Caproic acid
2-Aminobutyric acid
Guanidinoacetic acid
Ornithine
Glucose
Creatine
asy-Dimethylarginine
Isocitric acid
Pyridoxine
Asparagine
4-Cresol sulfate
Citric acid
Mannose
Serine
Indoxyl sulfate
Aconitic acid
Glutamine
Phenylacetylglutamine
Succinic acid
Cluster MGC6
Arginine
Hippuric acid
Suberic acid
Leucine
2-Aminoadipic acid
Phenol sulfate
Hypotaurine
Valine
Alanine
Acetylglycine
Indolelactic acid
2-Aminobutyric acid
Isoleucine
Butyrylcarnitine(C4)
2-Hydroxyglutaric acid
Threonine
Leucine
Acetylcarnitine
Isoleucine
Valine
Hexanoylcarnitine(C6)
Methionine
Threonine
Propionylcarnitine
Phenylalanine
Methionine
Isobutyrylcarnitine(C4)
Serine
Histidine
2-Methylbutyrylcarnitine(C5)
Tyrosine
Tryptophan
Isovalerylcarnitine(C5)
Tyrosine
Carnitine
Phenylalanine
Kynurenic acid
N-Acetylarginine
Trimethyllysine
N6-Acetyllysine
Uric acid
(a)The metabolites in two target patient groups (BK patients received CIT [n = 21] and Cat [n = 31]) in the expression cluster heat-maps (Fig. 2. a, LC/MS, b, GC/MS) are listed. (b) Similarly, the metabolites in two target patient groups (BK patients who received keratoplasty [n = 34] and Cat [n = 31]; Fig. 2c, d) are listed. The metabolites that increased or decreased in BK patients are summarised in Table 1a (received CIT) and 1b (received keratoplasty), showing the similar skewing of metabolite profiles.
Principal component analysis (PCA) was performed using the data from 20 metabolites extracted based on the statistical test of the BK group compared to the control Cat group. The distribution across PC1 and PC2 reveals three or two distinct metabolite profiles for the BK I (n = 21) and Cat (n = 31) groups. A comparison of the two groups is summarised in PCA biplots (Fig. 6a, b). In the first two principal components, the Cat group allowed sufficient discrimination power over the BK I group. Positive correlations were detected by LC/MS among the metabolites, including carnitine, acylcarnitine, trimethyllysine and acetylcholine in PC1 (Fig. 6a). Similarly, in metabolites detected via GC/MS, positive correlations were mainly found among Pro, Gly, citric acid, aconitic acid, Cys2, hypo-taurine, Cys, fucose, gluconic acid, ribose, sucrose and isocitric acid (Fig. 6b). Tricarboxylic acid (TCA) cycle-related metabolic intermediates and urea cycle-related metabolites, as well as the metabolites involved in clusters MLC2/3 in Fig. 5, primarily contribute to the sufficient discrimination power of the Cat group in the first two principal components (Fig. 6a, b).
Fig. 6PCA of the metabolites in AqH. The first two PCs (PC1 and PC2) are plotted and coloured according to the disease category (control Cat, blue and BK, red). (a) PCA score plots of LC/MS-based metabolite characteristics across BK and Cat (upper panel). PC1 and PC2 explain 72.0% of observed variability. PCA biplot of 20 metabolites and 52 samples (BK I, n = 21, Cat, n = 31) (lower panel). Symbols represent the strength of the contribution to each PC as vectors of metabolites (red arrows) and samples (symbol 1 for BK and symbol 2 for Cat). (d) PCA score plots of GC/MS-based metabolite characteristics across BK and Cat (upper panel). PC1 and PC2 explain 42.5% of the observed variability. PCA biplot of 20 metabolites and 52 samples (lower panel). Others are the same as in (a). PCA: principal component analysis, AqH: aqueous humor, PCs: principal components, BK: bullous keraptopathy, Cat: cataract, GC/MS: gas chromatography/mass spectrometry, LC/MS: liquid chromatography/mass spectrometry.
The identification of metabolites in the AqH of BK patients
Next, we tried to determine the differential expressions of metabolites in the AqH of BK patients (I, II) and PEG patients (ST-4). Of the 122 selected metabolites with levels significantly distinct from those in the Cat group, 19 were unique to I, 11 to II and 29 to the PEG group (p<0.05; Wilcoxon rank sum test). The 19 metabolites specific to the BK patients include HIB and AB, 2-oxoglutaric acid, β-Ala, glutamine (Gln), 2-hydroxybutyric acid, 2-hydroxyisobutyric acid, nicotinamide and glycolic acid. On the other hand, 43 metabolites were shared among all three disease categories, indicating these metabolites are related to disease phenotypes common in these diseases (Table 9). These metabolites included carnitine, acylcarnitine, citric acid, isocitric acid, aconitic acid, uric acid, uridine, spermidine and 2-oxoglutaric acid, which are mostly involved in the metabolite clusters MLC4 and MLC3 in Fig. 5a, c and MGC3 and MGC5 in Fig. 5b, d.
Table 9Candidate metabolites in the AqH of BK patients
Differential expressions of metabolites in the AqH of BK patients received cell injection therapy, corneal dysfunctional failure (CDF) patients received keratoplasty and pseudoexfoliation glaucoma (PEG) patients were explored. Of the 122 selected metabolites with levels significantly distinct from those in the Cat group, 19 were unique to the BK group (listed on the left side, p<0.05; Wilcoxon rank sum test). On the other hand, 43 metabolites are shared commonly among all three groups (listed on the right side, p<0.05), indicating these metabolites are related to disease phenotypes common in CDFs.
AqH; aqueous humors, BK; bullous keratopathy, Cat; cataract PEG: pseudo-exfoliation glaucoma, CDF: corneal dysfunctional failures composed of three keratoconus, three corneal opacity, one corneal scarring and one corneal dystrophy, who received keratoplasty.
Next, the hierarchical clustering method of miRs in AqH was performed using a criterion of object similarity. A clinical summary of the patients BK (n = 18) and Cat (n = 8), selected arbitrary for the analysis, is provided in Table-S5 and -S6. We constructed the hierarchical clustering to visualize differences in only miR levels, instead of both miRs and patients, and we compared miR levels in AqH between the BK and control Cat groups (Fig. 7a). The miR levels were higher in the BK group in cluster MmiR7 and lower in clusters MmiR1/3/4/6. Further, miR-29a-5p, -29b-1-5p, -29b-3p, -29c-3p, -34a-3p, -34a-5p, -34c-3p, -34c-5p, -378c, -378d and -378j were identified in cluster MmiR1. Of note, miR-34a-5p, 34c-5p and isoforms of the miR-378 and miR-29 family were present in the same cluster. Intriguingly, EV miR-184, recently found to be released from cultured HCE cells in a greater amount,
was classified into the cluster MmiR3 and miR-24-3p into the cluster MmiR6, both of which were identified as miRs that decreased in BK patients (Fig. 7b).
Fig. 7Differential expressions of miRs with regard to disease categories. (a) Heat-map cluster of detected miRs and 26 AqH specimens (BK, n = 18, Cat, n = 8). (b) Darker green indicates the level of each metabolite is lower than the mean, and darker red suggests a higher metabolite level than the mean (b) Main miRs extracted from four clusters that showed the decreased amount in the AqH of BK patients compared with Cat patients. MiRs discussed in the text are listed. MiRs indicated with a larger font size means those correlated clearly with metabolites in AqH miR; miRNA, AqH: aqueous humor, BK: bullous keratopathy, Cat: cataract.
Extraction of miRs closely associated with metabolites
The Lasso regression analysis identified miRs closely associated with the selected metabolites in the AqH of BK patients (n = 29, patients’ metabolites and miRs were measured in their AqH). After the Lasso regression analysis, candidate miRs associated with metabolites with nonzero coefficients out of 2,217 miRs were selected as critical and validated with a univariate analysis and receiver operating characteristic (ROC) analysis. At first, we investigated the association between selected individual metabolites and miR levels in the AqH of BK patients. The 15 metabolites showed significantly distinct expression levels between the higher and lower miR expression groups, in terms of the median values of the corresponding miRs (Fig. 8a). The positive associations were evident between 4-hydroxybenzoic acid and miR-378d, between xanthosine and miR-378i, between HIB and miR-34a-5p, between scyllo-Inositol and miR-378a-5p, between decanoic acid and miR-378h and between Thr and miR-302c-3p (p<0.05; Wilcoxon rank sum test). The negative associations were clarified between 3-hydroxybutyric acid and miR-92a-2-5p, between citric acid and miR-24-2-5p, between Ser and miR-34c-5p, between Thr and miR-29c-5p, between Tyr and miR-29c-5p, between xanthosine and miR-29b-1-5p, between GSSG and miR-378g, between scyllo-inositol and miR-378i and between ascorbic acid and miR-378j (p<0.05; Wilcoxon rank sum test, Fig. 8a). Further, we constructed an ROC curve to evaluate the ability of multiple candidate miRs associated with the selected metabolites in AqH. Among them, only seven metabolites are shown in Fig. 8b-1 and b-2, and six metabolites showed significant area under the ROC curve (AUC) values ≥ 0.83. These results suggest that multiple miR combinations correlate far better with the expression levels of specified metabolites in the AqH of BK patients than a single miR molecule.
Fig. 8Extraction of miRs associated with metabolites by Lasso analysis. (a) The Lasso regression analysis identified miRs closely associated with the selected metabolites in the AqH of BKF patients (n = 29). After the Lasso regression analysis, candidate miRs associated with metabolites with nonzero coefficients out of 2,217 miRs were selected. The 15 metabolites showed significantly distinct expression levels between the higher and lower miR expression groups, in terms of the median values of the corresponding miRs. Relative metabolite levels are presented with statistical differences (Wilcoxon rank sum test). Values are mean ± standard error. BKF: patients composed of 27 BK including FECD + one corneal opacity + one keratoconus, whose miRs and metabolites were both provided for Lasso analysis.(b-1) ROC curve analysis of associations between selected miRs from Lasso variable selection method and metabolite levels measured by LC/MS.(b-2) ROC curve analysis of associations between selected miRs from Lasso variable selection method and metabolite levels measured by GC/MS. The optimal λ-value for Lasso modelling a subset of miR was 0.227 with a transformed log (λ) of −1.482.AUC: area under the ROC curve, CI: confidence interval, Lasso: least absolute shrinkage and selection operator, miR; miRNA, ROC; receiver operating characteristic, AqH: aqueous humor, BK: bullous keratopathy, Cat: cataract control group, GC/MS: gas chromatography/mass spectrometry, LC/MS: liquid chromatography/mass spectrometry.
Extraction of metabolites in AqH closely associated with miRs
After the Lasso regression analysis, candidate metabolites with nonzero coefficients and associated with miR of 114 metabolites detected by LC/MS or 96 metabolites detected by GC/MS were selected as critical to dictating miR levels in AqH. At first, we investigated the association between selected individual miRs and metabolite levels in the AqH of BK patients. The four miRs were depicted to show the significantly distinct expression levels between the higher and lower metabolite expression groups, in terms of the median values of the corresponding metabolites (Fig. 9a). Positive associations were evident between miR-34a-5p and HIB, between miR-34a-5p and 2-oxobutyric acid, between miR-24-3p and AB and between miRNA-92a-2-5p and hypo-taurine (p<0.05; Wilcoxon rank sum test). On the other hand, negative associations were evident between miR-34a-5p and Orn, between miR-24-3p and Orn, between miR-92a-2-5p and 3-hydroxybutyric acid and between miR-92a-2-5p and succinic acid (p<0.05; Wilcoxon rank sum test; Fig. 9a). Next, we constructed an ROC curve to evaluate the ability of multiple candidate metabolites to associate with the selected miRs in AqH (Fig. 9b). Among five miRs, four showed significant AUC values ≥ 0.83. HIB was associated with miR-34-a together with only one other miR, showing AUC values=0.83 (Fig. 9a, thee second row). In the inversely directed Lasso analysis, positive associations were also evident between miR-34a-5p and HIB and between miR-24-3p and AB, while negative associations were identified between miR-34a-5p or -24-3p and Orn (Fig. 9a). The ROC curve identified multiple candidate metabolites associated with the specified miR in AqH (Fig. 9b). MiR-34a was associated with six metabolites, including HIB, with AUC 0.98, and with two metabolites: HIB and urea, with AUC 0.80. MiR-24-3p was associated with eight metabolites, including AB, with AUC 0.96, and with two metabolites, including AB and oleamide, with AUC 0.80. Further, miR-184 was associated with two metabolites, putrescine and 1,5-anhydro-D-sorbitol, with AUC 0.71. These results suggest that multiple metabolite combinations correlate with the expression levels of miRs in the AqH of BK patients.
Fig. 9Extraction of metabolites associates with miRs by Lasso analysis.After the Lasso regression analysis, candidate metabolites associated with miRs with nonzero coefficients out of 114 metabolites detected by LC/MS or 96 metabolites detected by GC/MS were selected as critical to dictating the miR levels in AqH. (a) The association between miR levels and metabolites in the AqH of BK patients. The four miRs were depicted with regard to the significantly distinct expression levels between the higher and lower metabolite expression groups, in terms of the median values of the corresponding metabolites. A univariate analysis of metabolites (GC/MS) associated with miR levels. Relative miR levels are presented with statistical differences (Wilcoxon rank sum test). Values are mean ± standard error. (b) ROC curve analysis of associations between selected metabolites (GC-MS) from the Lasso variable selection method and miR levels. The optimal λ-values for Lasso modelling a subset of metabolites by LC/MS and GC/MS were 0.213 and 0.318 with transformed logs (λ) of −1.546 and −1.546, respectively. AUC: area under the ROC curve. CI: Confidence interval, Lasso: least absolute shrinkage and selection operator, miR: miRNA, ROC: receiver operating characteristic, AqH: aqueous humor, BK: bullous keratopathy, Cat: cataract control group, GC/MS: gas chromatography/mass spectrometry, LC/MS: liquid chromatography/mass spectrometry.
As mentioned above, HIB showed a close correlation with miR-34a in two ways: a metabolite association with miRs (Fig. 8) and a miR association with metabolites (Fig. 9), whereas AB showed a correlation with miR-24-3p in only one way: an miR association with metabolites (Fig. 9). Therefore, we investigated whether HIB would modulate mitochondrial respiration in differentiated mature and de-differentiated immature cultured HCE cells. Of note is the 0.2 mM HIB elevated oxygen consumption rate (OCR) of mitochondria in differentiated, but not immature de-differentiated cultured HCE cells (Fig. 10.a). In addition, HIB induced the elevation of MMPs, detected by JC-1 staining, only in the latter de-differentiated cultured HCE cells, whereas no elevation was confirmed in the former differentiated cultured HCE cells (Fig. 10.b). This indicates that HIB restores mitochondrial dysfunction only in term of its polarization in de-differentiated HCE cells, but not in polarized functional mitochondria.
Fig. 10HIB and AB modulate mitochondrial functions (a) A real-time metabolic analysis of cultured HCE cells was performed using the Seahorse XFe24 extracellular flux analyser (Agilent Technologies, Santa Clara, CA, USA). The cultured HCE cells (lot #210512, passage 2, culture day 58, left upper) were tested as the responder cultured HCE cells with a high quality for HIB. Similarly, cultured HCE cells (lot #210608, passage 5, culture day 34) were tested as those with lower quality for HIB. For the Mito Stress test, a cell culture medium was replaced 1 hour before the assay with a minimal XF DMEM medium supplemented with 2 mmol/L glutamine, 10 mmol/L glucose and 1mmol/L sodium pyruvate (pH 7.4). The OCR was analysed at basal conditions and after sequential injections of 1 μM oligomycin, 1 μM FCCP, 0.5 μM rotenone and antimycin A. The assay results were normalised based on the viable cell number counted using the Cell Insight NXT (Thermo Fisher Scientific). The experiments were repeated five times. (b) A change in the MMP was detected using the JC-1 MitoMP Detection Kit (Dojindo Laboratories, Kumamoto, Japan). Collected cells were incubated with 2 μM JC-1 for 30 min at 37°C and analysed by the BZ X-700 Microscope System. The cultured HCE cells (lot #210608, passage 4, culture day 101, upper and lot #210610B, passage 5, culture day 35, lower) were tested as the responder cultured HCE cells with low quality for HIB (b-1). HIB showed no effect on cultured HCE cells with a higher quality (lot #210114, passage 2, culture day 52). The distinct proportions of mature CD44-/dull and de-differentiated immature CD44++/+++ cultured HCE cell SP were detected by a fluorescence cell sorter. The proportions were indicated in circles. hCECs: cultured human corneal endothelial cells, HIB: 3-hydroxyisobutyric acid, OCR: oxygen consumption rate, MMP: mitochondrial membrane potential, SP: subpopulation.
There was an abundant presence of miR-184 in EV released by differentiated mature cultured HCE cells expressing higher cellular miR-34a, leading to a sufficient tissue regenerative capacity in CIT.
It is noteworthy that the expression levels of miR-34a only in immature de-differentiated cultured HCE cells was elevated significantly by HIB (Fig. 11). Quite similarly, HIB upregulated the release of miR-184 only in immature dedifferentiated cultured HCE cells, indicating the participation of HIB in protective cell competition through miR-184 in AqH (Fig. 12).
Fig. 11HIB regulates the expression of cellular miR-34a.The regulation of the miR-34a expression by a metabolite, HIB, in both mature CD44-/dull and de-differentiated immature CD44++/+++ cultured HCE cells was investigated. RNA extraction and the miR-34a expression microarray analysis followed the procedures mentioned preiously.
The expression levels of miR-34a in the latter SPs was found elevated significantly by HIB, while in the former SPs, HIB showed no significant modulation at the concentration of 0.2 mM. HCE cells: human corneal endothelial cells, HIB: 3-hydroxyisobutyric acid, MMP: mitochondrial membrane potential, SP: subpopulation.
Fig. 12HIB elevated the amount of miR-184 released extracellularly.For miR expression profiling, 3D-Gene® Human miRNA Oligo Chips (miRBase version 17-19; Toray Industries) were used and analysed, as described previously.
All the data were globally normalised per microarray, such that the median of the signal intensity was adjusted to 25. (a) In mature CD44-/dull cultured HCE cell metabolites, HIB elicited the repressed extracellular release of miR-184 at a concentration of 0.2 mM. (b) In immature CD44++/+++ cultured HCE cells, HIB inversely showed an increase in the extracellular release of miR-184 at a concentration of 0.2 mM. All experiments were repeated in triplicate and biological replicates were n = 3. HCE cells: human corneal endothelial cells, HIB: 3-hydroxyisobutyric acid.
Writing Committee for the Cornea Donor Study Research Group Donor age and factors related to endothelial cell loss 10 years after penetrating keratoplasty: Specular Microscopy Ancillary Study.
The current study indicates that non-degenerated HCE cells may compete with degenerated cells to alleviate the mitochondrial dysfunctions of HCE tissues through the finely tuned molecular communication between miRs released into AqH and metabolites, secreted into AqH.
MiR-34a represses CD44 expression and downregulates intracellular pH, which is responsible for the skewing of mitochondrial respiration to OXPHOS.
The expression levels of miR-34a in immature de-differentiated cultured HCE cells SPs were elevated significantly by HIB, potentiating mitochondrial polarization (Fig. 10, 11) and indicating the possible perturbation of mitochondrial bioenergetics by metabolites in AqH produced by the heterogeneous cells in HCE tissues, either directly or indirectly, via the production of miR-184 (and miR-24-3p).
Hierarchical clustering analysis (HCA) of metabolite profiles in the culture supernatant (CS) identified subsets of metabolites that correlated with the cultured HCE cell phenotypes,
and we confirmed variations in extracellularly secreted metabolites by HCA among cultured HCE cell SPs, distinct in their clinical efficacy in cell injection regenerative medicine.
It is of note that this profile of metabolites detected is quite similar to that clarified here as metabolites in AqH of BK patients (Fig.5, Table 8).
Considering the continuous exposure of HCE tissues to AqH, the sharing of similar metabolite profiles, albeit partially, between the CS of cultured HCE cell SPs and the AqH of patients in specific disease category would imply the presence of heterogeneous cells (degenerated and non-degenerated) as constituents in a single-layered HCE tissue with distinct metabolic signatures. Of note, the previous study revealed that differentiated cultured HCE cells catabolise BCAAs and Ser more actively than immature de-differentiated cultured HCE cells.
In differentiated cells, OXPHOS was activated, accompanied by the active catabolism of BCAAs, whereas in immature transformed leukaemia cells, these phenotypes were inversely disposed to a glycolytic type.
In accordance with these findings, BCAT2, a mitochondrial aminotransferase for BCAAs, and the branched-chain alpha-ketoacid dehydrogenase complex (BCKDC), were both upregulated selectively in mature cultured HCE cells.
In the same clusters with BCAAs, Ser was also found decreased in the AqH of BK patients (Fig. 5a, cluster MLC3,Fig. 5b, cluster MLC4), which might be ascribed to the fact that more glucose-derived carbon is channelled into Ser biosynthesis in HCE cells of BK patients to support the cell proliferation.
MiR-34a-5p with a close correlation to HIB, as well as miR-34c-5p to Ser (Fig. 8a) were classified into the same miR clusters (Fig. 7b); that is, both metabolites were decreased in the AqH of BK patients, although the former correlated with HIB positively and the latter with Ser negatively (Fig. 8a). The clarification of the network among the metabolites HIB/Ser and miR-184 in AqH and cellular miR-34a-5p/miR34c-5p in HCE cells will be critical going forward.
HIB belongs to the hydroxyl-carboxylic acid family, and they are key metabolic intermediates of energy metabolism. HIB is produced during the catabolism of Val as an energy source, and Bjune et al. reported that plasma HIB is a marker of hepatic mitochondrial fatty acid oxidation.
Our study provides insight into TCA cycle-related metabolites in AqH associated with fatty acid flux in HCE cells, reflective of mitochondrial β-oxidation performed by the carnitine family. These metabolites were commonly increased in the AqH of BK, FECD, keratoconus and corneal opacity patients (Cluster MLC4, Fig. 5a, MLC3, Fig. 5b, Table 8). However, the role of HIB in lipid catabolism in a single-layered HCE tissue remains elusive.
Jun Group reported the widespread downregulation of intracellular miR levels in the HCE tissues of patients with late-onset FECD, and it indicated significant downregulation of the miR-29 family.
In addition, Iliff et al. reported a single-base-pair substitution in miR-184 with regard to the disease phenotype of EDICT, a syndrome characterised by endothelial dystrophy, iris hypoplasia, congenital cataract and stromal thinning.
It is of interest that miR-184 and -24-3p, both released extracellularly, were clustered as miRs that decreased in the AqH of BK patients (Fig. 7b), whereas the miR-34a and -29 families, highly repressed intracellularly in degenerated HCE tissue, were combined into the same cluster (MmiR1,Fig. 7a). MiR-34a-5p, as a critical cellular constituent of the HCE, and the cellular miR-29 family were both at relatively lower levels in AqH.
Instead, relatively large amounts of miR-184, -24-3p, -92b-5p and -23b-3p were detected in the AqH of BK patients. The activation of p53 by stress signals results in enhanced EV production by cultured HCE cell SPs.
and were involved in senescence targeting of the p53 pathway. Oxidative stress also plays a major role in the chronic degenerative process of HCE in FECD.
Reduced GSH plays an essential role in the maintenance of the intracellular redox state, and the dysregulation of its homeostasis is implicated in the pathophysiology of a number of tissues.
Oxidative stress activates the GSH biosynthetic pathway to compensate for increased GSH consumption. In this context, it will be also relevant to clarify the role of AB in AqH to maintain the integrity of the HCE tissue exposed to ROS. Putrescene, spermidine and spermine were all decreased more in the AqH of BK patients than in that of the control (Fig. 5a, cluster MLC3,Fig. 5b, cluster MLC4). Spermidine is an aliphatic polyamine and a precursor to spermine. The functional role of putrescene/spermidine/spermine, namely the polyamine axis, may be critical to maintaining HCE tissue integrity, considering its role in cell proliferation.
The Lasso regression analysis identified miRs closely associated with the selected metabolites in the AqH of BK patients. Among the six positive associations evident between metabolites and miRs, we selected the combination of HIB and miR-34a-5p for further wet experiments, considering the particularly relevant role of cellular miR-34a in the HCE cell fate decision and its fragile plasticity in the expression levels between degenerated and non-degenerated HCE tissues.
Clarifying the negative associations between metabolites and miR-92a-2-5p, -24-2-5p, -378g and -34c-5p remains a task for future works. In the inversely directed Lasso analysis, positive associations were also evident between miR-34a-5p and HIB and between miR-24-3p and AB, while negative associations were identified between miR-34a-5p or -24-3p and ornithine (Fig. 9a).
we reported that large amounts of miR-184 and -24-3p were detected in the pre-surgical AqH of BK patients; in addition, extracellularly released miR-184 and -24-3p were more abundant in CS derived from differentiated mature cultured HCE cells. The study provoked the hypothesis that the cellular interplay among heterogeneous HCE cells, through EV miR-184, may compensate the exacerbated degeneration in a single-layer HCE tissue. Zhao et al. identified different expressions of miRs in mouse CE during aging,
and more recently, Buono et al. analysed the effects of mesenchymal stem cell-derived EVs in an in vitro endoplasmic reticulum (ER) stress model of corneal dystrophy, having found that its effects were correlated to the transfer of ER stress targeting miRs to the CEC.
The expression levels of the cellular miR-34a and -29 families in the human neonatal corneal tissues were significantly downregulated, whereas no downregulation of cellular miR-184, -24-3p, -23b-3p, -23a-3p or -92b-5p were observed in HCE tissues.
The presence of these miRs in AqH may be indicative of their unknown paracrine function to regulate the integrity of HCE tissues.
Considering the association of miR-184 mutation with BK, FECD, EDICT and corneal dystrophy, the loss of the molecular interplay between miR-184 and metabolites, leading to dysfunctional mitochondria in a single-layer HCE tissue, may be a potential causal event in the pathogenesis of these diseases. Defects in HCE give rise to posterior endothelial corneal dystrophies (ECDs), leading to compromising visual acuity through edematous stroma. Dominantly inherited FECD is most common among corneal dysfunctional failures, and its lifetime incidence is around 4% over age 40.
(these findings were reproducibly confirmed in this study, Fig. 7). In addition, we confirmed that the cellular miR-184 expression level in miR-34a transfected cells was only slightly elevated, but the extracellular miR-184 and -24-3p levels were markedly upregulated, while the extracellular miR-23-3p and -92b-5p levels showed no significant difference as a result of the transfection.
HIB functioned to restore the repressed expression of cellular miR-34a concomitantly with the restoration of the depressed extracellular release of miR-184 (Figs. 11, 12). EV miR-184 and HIB play a key role in coping with the ER stress responsible for HCE degeneration through up-regulations of the mitochondria membrane potential and biogenesis (Figs. 10, 13).
Fig.13MiRs and metabolites in AqH synchronize in maintaining cellular phenotypes in corneal endothelium homeostasis.This figure summarizes the hypothesis of the synchronized role of miR-184 and a metabolite, HIB, in AqH to attenuate the aggravated degenerative invasion into the surrounding single HCE cell layer by dampening the vicious cycle. HIB induces the production of miR-184 through the regulation of the cellular mir-34a expression followed by the restoration of impaired mitochondrial functions, thereby maintaining the osmotic gradient in the HCE layer as indispensable for the efficient efflux of water from the stroma to the apical side, as described.
In addition, AB may contribute to this interplay through the anti-oxidative effect to attenuate the repressed expression of miR-34a by ROS by inducing increased intracellular glutathione levels.
Our findings offer the new idea that non-degenerated HCE cells may compete with degenerated HCE cells through the alleviation of HCE degeneration caused by the ROS-induced repression of miR-34a, resulting in the downregulation of CD44.
The integrity of HCE tissues relies on a combination of paracrine signals provided by the microenvironments of their ambient neighbours or niches, such as metabolites and miRs in AqH, thereby restraining the expansion of degenerated cells. The current study provides insights into the fine-tuned paracrine cellular interplay between heterogeneous cells not only in HCE tissue, but also in other single-layer tissues, under homeostasis and during perturbations.
Limitation and future prospects
It should be noted that this study did have some limitations. First, the number of BK patients enrolled was small. Thus, further refinement may be necessary to yield the more solid conclusion. BK is designated on the basis of clinically accumulated observations, the severity of the diseases is diversified and heterogeneous. This study might be the first challenge to pioneer the way to classify molecularly the pathogenic heterogeneity of BK patients to develop the molecular diagnosis. To validate the fascinating hypothesis here presented, a cohort study with a large number of BK patients heterogeneous including medication are necessary, together with a statistically sufficient power. 24 hours medication washout is not sufficient to reset the biochemical signatures relating to gene and protein expression, and thus aqueous findings, secondary to drug effect. Therefore, there cannot be definitive conclusion of nil effect of preoperative medications on outcomes. The international collaborative studies with artificial intelligence (AI) may also open a new avenue to provide an innovative diagnosis for Bk patients including FECD (Fig. S14). In addition, the experimental verification of the capability of metabolites and/or miRs to increase or decrease the proportion of cultured CD44-/dull SPs may surmount the weakness of this study
Aqueous Humor Analysis Identifies Higher Branched Chain Amino Acid Metabolism as a Marker for Human Leukocyte Antigen-B27 Acute Anterior Uveitis and Disease Activity.
*These authors equally contributed to this article：Morio Ueno, Kengo Yoshii
Financial Support
Supported by the Projects for Technological Development from the Japan Agency for Medical Research and Development, AMED (Tokyo, Japan) 19bm0404033h0002, and JSPS KAKENHI Grant Number JP26293376. The sponsor or funding organisation had no role in the design or conducting of this research.
Conflict of Interest
No conflicting relationship exists for any author in the subjects herein described.
Acknowledgements
The authors wish to thank Takaaki Sato deeply for his continuous encouragement during this study, Asako Hiraga for her technical assistance and Kojiro Imai for his assistance in collecting the clinical specimens throughout the study. The authors are also grateful to CorneaGen Inc. for the generous gift of corneal tissues.
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:
Précis
The synchronised collaboration exists in AqH between miR-184 and a metabolite, hydroxyisobutyric acid, in ensuring the HCE homeostasis to retain efficient dehydration from the corneal stroma by upregulating cellular miR-34a.
00031-3&id=gr1.jpg){kind=link}
00031-3&id=gr2.jpg){kind=link}
00031-3&id=gr3.jpg){kind=link}
00031-3&id=gr4.jpg){kind=link}
00031-3&id=gr5.jpg){kind=link}
00031-3&id=gr6.jpg){kind=link}
00031-3&id=gr7.jpg){kind=link}
00031-3&id=gr8.jpg){kind=link}
00031-3&id=gr9.jpg){kind=link}
00031-3&id=gr10.jpg){kind=link}