Metabolomic Profile in the Aqueous Humor of Congenital Ectopia Lentis

Abstract Purpose To explore the metabolic profiles in the aqueous humor (AH) of patients with congenital ectopia lentis (CEL). Methods We conducted a comprehensive analysis of the metabolites of AH samples of patients with CEL (n = 22) and age-matched patients (n = 22) with congenital cataract by ultra-high performance liquid chromatography tandem-mass spectrometry. The metabolomic characteristics were visualized by principal component analysis, orthogonal partial least squares discriminant analysis and heat map. The levels of the differential metabolites were also compared between CEL patients with and without FBN1 mutations. Pathway enrichment analysis was performed by using Kyoto Encyclopedia of Genes and Genomes. Receiver operating characteristic analysis was performed to select potential biomarkers. Results There were 175 differential metabolites identified between the two groups. Eight metabolites were found to be potential biomarkers in AH of CEL patients. The CEL group showed a significant increase in α-ketoglutarate and decrease in citrate, suggesting that the tricarboxylic acid (TCA) cycle was disturbed. l-proline, prolyl-hydroxyproline, and l-histidine were reduced, which prompted enhanced degradation of microfibrils and collagen. Insidious retinal nerve damage was implied because N-Acetyl-aspartylglutamic acid and N-Acetyl-l-aspartic acid were found to be significantly increased. Pathway enrichment analysis indicated that disturbances in amino acid metabolism and carbohydrate metabolism were the key processes in the pathogenesis of CEL and that TCA cycle disorder may be the driving force behind disease occurrence. Conclusion These data reveal the characteristics in the metabolomic profiles of the AH of CEL patients, which help provide insights into the pathogenesis of this rare disease.


Introduction
Congenital ectopia lentis (CEL), which is often diagnosed as an early ocular manifestation of Marfan syndrome (MFS), is a heritable connective tissue disease in which the lens deviates from the normal physiological position because of lens zonule laxity or rupture, hence causing vision loss, refractive errors, strabismus, secondary glaucoma, and even retinal detachment and blindness. 1,2 CEL mostly occurs in children and adolescents who are in a critical period of eyeball development, which seriously endangers their vision and visual development. 3 Previous studies have shown that the TGFb signaling pathway is activated in patients with MFS, [4][5][6] leading to the increased expression of matrix metalloproteinases (MMPs), which can break the fibrillin-rich microfibrils 7,8 and induce the degradation of the extracellular matrix and smooth muscle cell apoptosis in aortic tissue. 9 However, research on the pathogenic mechanism of CEL is hard to conduct due to the technical challenges in visualizing the zonules and the difficulty in obtaining impaired zonules. Our previous work has found that the level of TGFb2 in the aqueous humor (AH) of patients with CEL was significantly increased, 10 so understanding the changes in the contents of the AH may be of great help in comprehending the pathogenesis of CEL.
As the final product of the enzymatic process, metabolites reflect physiological and pathological functions and activities from cells to tissues. Metabolomic analysis explains the pathological process of diseases by identifying metabolic spectrum networks. Benke et al. found that severe cardiovascular involvement in Marfan patients was associated with higher homocysteine concentration in plasma and that the impairment of the folic acid metabolic pathway played an important role in the development and remodeling of the aortic extracellular matrix. 11 In addition, a study of the pattern of collagen metabolism in nine fibroblast cultures from Marfan patients showed that the metabolisms of some kinds of collagens might represent the molecular basis of Marfan syndrome. 12 However, research on the metabolic mechanism of CEL is still lacking.
Recent studies have revealed AH changes in metabolites in several eye diseases, such as primary congenital glaucoma, myopia, diabetic retinopathy, central retinal vein occlusion, and age-related macular degeneration. [13][14][15][16][17] AH is an important intraocular fluid, and lens zonules, ciliary body, and lens, all are immersed in the AH, in which the metabolites of the tissue pathological process diffuse. Consequently, metabolite analysis of AH is an important part of elucidating the underlying mechanism of CEL. Therefore, we collected the AH of CEL patients, and hoped to provide the metabolic spectrum changes of CEL disease and the pathogenesis that might be involved in it through metabolomics.

Human subject
A total of 44 patients (44 eyes) were recruited for the current study, including 22 patients with CEL and 22 patients with congenital cataract as the controls. The study was approved by the Ethics Committee of Zhongshan Ophthalmic Center (2020KYPJ146) and followed the principles of the Declaration of Helsinki. All included patients were from Zhongshan Ophthalmic Center. The inclusion criteria of the CEL group were as follows: diagnosed with CEL; without ocular inflammation, ocular trauma, history of intraocular surgery, or other ocular diseases such as congenital cataract, congenital glaucoma; and without other systemic diseases such as gigantism or homocystinuria. The exclusion criteria were the presence of ocular inflammation, long-term oral medication. The inclusion criteria for congenital cataract patients in the control group were as follows: mild to moderate congenital cataract, axis length < 26 mm, without a history of intraocular surgery, ocular trauma, other eye diseases, and systemic diseases such as hypertension, diabetes, without long-term administration of medications. All patients underwent a comprehensive eye examination before surgery, and CEL patients performed next-generation sequencing.

AH collection
Before all the surgical procedures, a small incision was made along the limbus, and a 1 mL sterile syringe was punctured into the anterior chamber. Then, 100-150 mL of AH was extracted and quickly transferred to a 2 mL Cryogenic Vials (Corning) in liquid nitrogen, which was then stored at -80 C. The AH samples were collected from July 2020 to April 2021.

UHPLC-MS analysis
AH samples were thawed and subjected to ultra-high performance liquid chromatography tandem-mass spectrometry (UHPLC-MS). The liquid chromatography tandem mass spectrometry (LC/MS) portion of the platform was based on a Dionex Ultimate 3000 RS UHPLC fitted with a Q-Exactive plus quadrupole-Orbitrap mass spectrometer equipped with a heated electrospray ionization (ESI) source (Thermo Fisher Scientific, Waltham, MA, USA). AH samples were thawed at room temperature, and 50 mL of sample were added to a 1.5 mL Eppendorf tube with 10 mL of 2-chlorolphenylalanine (0.3 mg/mL) dissolved in methanol as the internal standard, and the tube was vortexed for 10 s. Subsequently, 150 mL of ice-cold mixture of methanol and acetonitrile (2/1, v/v) was added, and the mixtures were vortexed for 1 min, ultrasonicated in an ice-water bath for 10 min, and stored at -20 C for 30 min. The extract was centrifuged at 13 000 rpm, 4 C for 15 min. Then, 150 mL of supernatant was drawn in a brown and glass vial and dried. A 200 mL mixture of methanol and water (1/4, v/v) was added to each sample, and the samples were vortexed for 30 s and ultrasonicated for 3 min before being stored at -20 C for 2 h. The samples were thawed and centrifuged at 13 000 rpm at 4 C for 10 min. The supernatants (150 mL) from each tube were collected using crystal syringes, filtered through 0.22 mm microfilters, and transferred to LC vials. The vials were stored at -80 C until LC-MS analysis. Quality control (QC) samples were prepared by mixing the aliquots of all the samples to form a pooled sample. Metabolites were analyzed in both ESI positive and negative ion modes by employing ACQUITY UPLC HSS T3 column (1.8 mm, 2.1 Â 100 mm). The binary gradient elution system consisted of (A) water (containing 0.1% formic acid, v/v) and (B) acetonitrile (containing 0.1% formic acid, v/v), and separation was achieved using the following gradient: 0 min, 5% B; 2 min, 5% B; 4 min, 25% B; 8 min, 50% B; 10 min, 80% B; 14 min, 100% B; 15 min, 100% B; 15.1 min, 5% B; and 16 min, 5% B. The flow rate was 0.35 mL/min, and the column temperature was 45 C. The mass range was from massto-charge ratio (m/z) 100 to 1200. The resolution was set at 70 000 for the full MS scans and 17 500 for the HCD MS/MS scans. The collision energy was set at 10, 20, and 40 eV. The mass spectrometer operated as follows: spray voltage, 3800 V (þ) and 3000 V (À); sheath gas flow rate, 35 arbitrary units; auxiliary gas flow rate, eight arbitrary units; capillary temperature, 320 C; Aux gas heater temperature, 350 C; S-lens radio frequency level, 50. The QCs were injected at regular intervals throughout the analytical run to provide a set of data from which repeatability could be assessed.

Data preprocessing and statistical analysis
The original LC-MS data were processed by Progenesis QI V2.3 software (Nonlinear, Dynamics, Newcastle, UK) for baseline filtering, peak identification, integral, retention time correction, peak alignment, and normalization. Compound identification was based on precise m/z, secondary fragments, and isotopic distribution using the Human Metabolome Database, Lipidmaps (V2.3), Metlin, EMDB, PMDB to perform qualitative analyses. A data matrix was combined from the positive and negative ion data. The matrix was imported in R to carry out principle component analysis (PCA) to observe the overall distribution among the samples and the stability of the whole analysis process. Orthogonal partial least squares discriminant analysis (OPLS-DA) and partial least squares discriminant analysis (PLS-DA) were utilized to distinguish the metabolites that differed between the groups. To prevent overfitting, sevenfold cross-validation and 200 Response Permutation Testing were used to evaluate the quality of the model. Variable importance of projection (VIP) values obtained from the OPLS-DA model were used to rank the overall contribution of each variable to group discrimination. A two-tailed Student's t-test was further used to verify whether the differential metabolites between groups were significant. Differential metabolites were selected with VIP > 1.0 and p < 0.05. Pathway enrichment analysis was performed by using Kyoto Encyclopedia of Genes and Genomes (KEGG) library. Receiver operating characteristic (ROC) analysis were performed using Stata/SE software version 16.0 (StataCorp, USA) for discrimination of CEL from controls.

Results
There was no significant difference between the CEL group and control group regarding age, gender, or surgical eyes, as shown in Table 1. Thirteen CEL patients were tested and found to have FBN1 mutations.

Metabolic profiling of human AH
In the 44 AH samples, a total of 2803 metabolites were identified (Supplemental Table S1), including 700 metabolites according to the database from the KEGG, which covered most of the central metabolism pathways. As shown in Figure 1(A), samples from patients with CEL and samples from control were separated, and the CEL samples were clustered together. The results indicate that patients with CEL had an identical and distinct metabolomics profile. OPLS-DA demonstrated a clear separation between the two groups ( Figure 1(B)). The cross-validation and permutation tests show that the cumulative R 2 Y was 0.976 and Q 2 was 0.612, suggesting that the model fitted well. The validation plots (Figure 1(C)) from permutation test supported the validity of the established OPLS-DA model as all permuted R 2 and Q 2 values on the left were lower than the original point on the right, and that the Q 2 regression line in green had a negative intercept.

Metabolic variation between patients with CEL and controls
There were 175 differential metabolites between the two groups, of which 82 differential metabolites were upregulated and 93 were downregulated compared with the control  group (Supplemental Figure S1). Citrate, L-proline, N-Acetyl-L-Histidine (NAH), prolyl-hydroxyproline (Pro-Hyp), L-histidine, pantothenic acid, nicotinamide, and phthalic acid were significantly reduced, while a-Ketoglutarate, N-Acetyl-aspartylglutamic acid (NAAG), N-Acetyl-L-aspartic acid (NAA), arginine, phosphoglycolic acid, and S-lactylglutathione were also significantly increased in the CEL group (Figures 2 and 3). In addition, the a-Ketoglutarate/citrate (K/C) ratio in the AH of the CEL group was significantly higher than that of the control group (p < 0.001).
The CEL group was divided into two subgroups-with FBN1 mutations or not-and a dependent t-test analysis of several significantly different metabolites showed that there was no statistical difference between the subgroups (Supplemental Figure S2).

Metabolic enrichment pathway analysis
A further metabolic pathway analysis of the 51 significantly different metabolites (according to KEGG) revealed that the 43 most relevant pathways were enriched, among which alanine, aspartate, and glutamate metabolism; central carbon metabolism in cancer; and the tricarboxylic acid (TCA) cycle were the most significantly influenced pathways (Figure 4).

Potential biomarkers analysis for discrimination
We performed ROC analysis to select the potential biomarkers for discrimination and assessed the diagnostic performances. ROC analysis showed that Alanyl-Arginine  Figure 5).

Discussion
The study aimed to identify changes in the metabolic distribution of the AH of CEL patients by applying an untargeted metabolomics analysis. Significant differences were detected between CEL group and controls involving organic acids, amino acids, and their derivatives, such as a-ketoglutarate, citrate, L-proline, L-arginine, L-histidine, coumaric acid, and pantothenic acid. The metabolic differences suggested that the disorders of carbohydrate metabolism and amino acid metabolism might be of great significance in the process of zonule lesions.
The TCA cycle might be the driving force of the pathogenesis of CEL. As the result showed, citrate was downregulated when compared with the control group. Citrate is not just a process product of energy production: it is a key mediator of metabolic reprogramming. 18 Citric acid is produced by acetyl-CoA in the TCA cycle and converted into a-ketoglutarate or activated acetyl-CoA carboxylase for fatty acid biosynthesis. Citric acid is the starting material for acetyl-CoA to enter the TCA cycle. Its concentration reflects the level of metabolism in the TCA cycle. In the current study, the average citrate level of the CEL group was lower than that of the control group, while the average a-Ketoglutarate level was relatively higher; hence, the relative Figure 2. Heatmap analysis of discriminative metabolites. The blue color represents low relative level of each metabolite, and the red color represents high relative level of each metabolite. Figure 3. Relative levels of a-Ketoglutarate and citrate in AH and the K/C ratio between control and CEL groups. Ã , p < 0.05; ÃÃÃ , p < 0.001. K/C ratio was found to be significantly higher. Research has shown that cellular conditions that increase the K/C ratio will initiate reductive glutamine metabolism. 19 It has also been found that the K/C ratio is positively correlated with glutamine levels in the AH of patients with central retinal vein occlusion. 20 However, glutamine levels were not detected and compared in the current study. Our results indicate that the disturbance of the TCA cycle and carbon mechanism may initiate some cellular conditions of the ciliary epithelial cell, but its underlying mechanism needs to be further explored. Similarly, Fraenkl et al. revealed that plasma citrate levels in glaucoma patients were significantly reduced; they suggested that citrate was expected to be a biomarker for the diagnosis of glaucoma. 21 The decrease of citric acid in CEL patients might give a hint as to why a partial portion of CEL patients are prone to glaucoma.
The decreased levels of L-Proline and Pro-Hyp implied increased degradation of zonule microfibrils and collagen. L-Proline is essential for producing collagen, reducing collagen loss, and preventing the aging process. 22,23 Collagen is the most abundant protein in animals, accounting for around 30% of all proteins in humans, 24 and it exists as the main extracellular matrix component in connective tissues such as skin, muscle, cartilage, bone, tendon, ciliary epithelium, and lens zonules. 25,26 Studies have reported that the cross-linking of proline and glycine-rich regions mediate the formation of extracellular dimers and might be a crucial early step in the extracellular assembly of fibrillin into microfibrils. 27 The level of L-proline in the AH of CEL patients was significantly decreased compared with that of the control group, suggesting that the decreased synthesis of structural protein fibrillin and disorder of microfibril assembly were important underlying mechanisms that led to thinner and slacker suspensory zonules. Collagen is mainly composed of three amino acids: glycine, proline, and hydroxyproline (Hyp). 28 Pro-Hyp is the main derivative produced through the interaction of proline and ascorbic acid and is exclusively present in collagen. 29 Pro-Hyp is a dipeptide derived from collagen and generated endogenously when collagen turnover is enhanced. 30 Many studies have shown that Pro-Hyp has beneficial effects on joints, skin, and other connective  tissues. 28,31,32 Zonular fibers contained no hydroxyproline, but when examined by polyacrylamide gel electrophoresis, a band migrating in the alpha-position of collagen was observed, 33 and the mammalian lens zonules are rich in collagen, such as Type II, IV, VII, and IX collagen. 25 Defects in the metabolism of ascorbic acid, glutamine, proline, and cysteine can cause impaired collagen biosynthesis. 34 In the current study, it was found that the level of Pro-Hyp significantly decreased, along with the level of L-proline and L-histidine in the AH of CEL patients, indicating increased degradation of zonule microfibrils and collagen, which might result in progressive impairment of lens zonules.
Surprisingly, the present study has found a higher K/C ratio and higher average levels of NAAG and NAA in AH of the CEL group, suggesting the potential risk of neural retinal impairment in CEL patients. In the AH of patients with retinal artery occlusion, the concentration of glutamate was found to be significantly higher, suggesting the important role of glutamate in retinal injury. 20 NAAG is an important peptide neurotransmitter that is synthesized by NAA and glutamate under the catalysis of NAAG synthase, 35 and it has a strong inhibitory effect on the release of excitatory transmitter glutamate after brain injury. 36 Dreyer et al. showed that the level of glutamate in the vitreous of patients with glaucoma was increased compared with the control group, exerting a damaging effect on retinal ganglion cells. 37 In addition, it has been shown that a-Ketotrexate and glutamate can be converted to each other by transamine and deamine. In the AH of the CEL group, a higher K/C ratio and higher average levels of NAAG and NAA were observed, so it is reasonable to assume that there was an overload of glutamate inside the ocular tissue. An inappropriate concentration in the posterior chamber may be toxic to the retina, which may indicate a potential state of retinal damage in CEL disease. In addition, our previous work discovered that CEL patients had altered microvascular flow of the retina and significantly thinner retinal nerve fiber layers, which were also associated with poorer visual acuity. 38 There were no vitreous or retinal samples providing evidence; however, this study aroused our concerns about retinal impairment in CEL patients.
Our study has shown that NAH and L-histidine were significantly lower in the AH of patients in the CEL group than in the congenital cataract group. NAH is an important biomolecule in the retina and lens of poikilothermic vertebrates, 39 and it is synthesized from L-histidine and acetylcoenzyme A by histidine N-acetyltransferase. 40 Early studies found that the L-histidine is transported into the lens from the surrounding fluid, and then, NAH is synthesized and maintained in a high concentration to form a concentration gradient to sustain lens water homeostasis. 41 In addition, researchers have observed that lens cells could not hydrolyze NAH, so NAH is then released into ocular fluid and decomposed into L-histidine under the action of NAH deacetylase. 42 Further studies have claimed that NAH functions as a molecular water pump to maintain a highly dehydrated lens and avoid cataract formation. 43 In the AH of patients in the CEL group, NAH and L-histidine were significantly lower than those in the congenital cataract group; however, this relative difference might be because of the increased levels of NAH and L-histidine in patients with congenital cataract. Studies have found that the lens NAH concentration is inversely related to the development of cataracts. 44,45 When the absorption of L-histidine by the lens decreases, the synthesis of NAH and its protective effect on the water balance of the lens fiber reduce. Therefore, this result should be explained as a disorder of the metabolism of L-histidine and NAH in the AH of patients with congenital cataract in the control group.
Although lens dislocation developed for different reasons, the final metabolites and pathways involved in metabolism were not found to be different. There was no significant difference in the metabolites of the AH between the FBN1mutated patients and non-FBN1-mutated patients in CEL group. Based on the results, it is not yet believed that FBN1 mutation will affect the metabolic process of the disease. Besides, except for FBN1 mutations, mutations in LTBP2, ADAMTSL4, ADAMTS10, and ADAMTS17 have also been reported to cause congenital lens dislocation. [46][47][48][49][50][51] It is important to note that all the above genes mutations may have impact on fibrillin-1 lost function or imbalanced production. 48,52 As previous studies reported, fibrillin-1 has been identified as a potential key substrate of downstream pathogenic mechanism of TGFb and MMPs in connective tissue disease such as MFS. 7,9 Considering fibrillin-1 may serve as the final joint pathogenetic molecule, the ultimate metabolic pathway of ocular zonules relaxation or rupture may hence be similar. Further studies are needed to investigate the relations between phenotype, genotype and metabolic characteristics in CEL patients.
Besides, we assessed the impact of 175 metabolites by ROC analysis. The results indicated that eight metabolites could serve as potential biomarkers in AH for a good discrimination between CEL and controls. Pelargonidin is a phenolic substance with antioxidant activity that could reduce intraocular pressure and oxidative damage, and prevent the development of glaucoma by maintaining antioxidant enzyme levels. 53 Petunidin, an anthocyanin, also has antioxidant activity and was reported to inhibit lens opacity. 54 Mimosine is a nonprotein amino acid, tyrosine analogue, that can chelate iron to block the normal mammalian cell cycle, 55 activate apoptosis, and induce reactive oxygen species production. 56 These results may be related to the final phenotype of the disease. However, since very scant information is available about the role of these substances in ocular diseases, it is difficult to determine the deeper internal association.
The present study has several limitations, including a relatively small sample size since CEL is a relatively rare disease. Considering the liquid characteristics of AH, the current study only used LC-MS/MS, which could not analyze volatile metabolites. In addition, owing to ethical and agematching restrictions, the control group comprised congenital cataract patients, which may have led to bias to some extent. However, we comprehensively referenced several metabolomics databases and identified the metabolites as much as possible, which uncovered the most extensive human AH metabolome to date. Most importantly, to the best of our knowledge, the current study was the first to report the metabolic profile in the AH of CEL and provide greater insights into our understanding of the pathogenesis.
In conclusion, our study provided an overview of the metabolites in the AH of CEL patients and elaborated on the underlying mechanisms involved in the disease. A distinctive metabolomic profile was described for the first time with amino acid-related and organic acid-related metabolites and their metabolic pathways. Further studies and validation are still needed to uncover the role of differentially expressed metabolites in the occurrence of CEL.

Author contributions
DYZ and GMJ designed the study. LYL and YQL wrote and revised the manuscript. LYL, DWG, HWY and HTQ collected the aqueous humor samples. LYL and BZ analyzed the data and visualized. YQL, DYZ and GMJ provided critical review for the manuscript.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Data availability statement
The data that support the findings of this study are available from the corresponding author, DY Zheng or GM Jin, upon reasonable request.