Association of 2 Lysyl Oxidase Gene Single Nucleotide Polymorphisms with Keratoconus

Abstract

Purpose

Keratoconus (KC) is the most common primary ectatic corneal disease, characterized by progressive thinning of the cornea, affecting its shape and structure and leading to visual loss. Lysyl oxidase is an important component of the extracellular matrix and contributes to the homeostasis of corneal stromal extracellular matrix via enzymatic reaction. This nationwide registration study aims to examine the association of KC with 2 known single nucleotide polymorphisms, rs2956540 and rs10519694, in a population of Iranian descent.

Design

Case–control.

Participants

One hundred seventy-eight subjects with KC and 180 clinically healthy subjects participated in the study.

Methods

Genomic DNA was extracted from peripheral blood samples, and their genotypes were determined using tetra-primer amplification refractory mutation system–polymerase chain reaction.

Main Outcome Measures

Allele frequency for rs2956540 and rs10519694.

Results

Genotype frequency was significantly different between cases and controls for rs2956540 (P value = 0.019). The rs2956540 C allele carriers were significantly more frequent among KC cases than healthy controls (P value_chi-square = 0.015, P value_{Fisher exact} = 0.017). There was a significant difference in genotype frequency between groups for rs10519694 (P value = 0.001). T allele carriers were significantly more frequent among KC patients (P value_chi-square = 0.002, P value_{Fisher exact} = 0.001). Sex stratification revealed no significant differences in genotype frequency between males and females in cases and controls. Fitting the general linear model showed that rs10519694 could be considered a predictor for the development of KC (P value = 0.001); however, this was not observed for rs2956540 (P value = 0.323).

Conclusions

rs2956540 and rs10519694 are associated with KC in a population of Iranian descent. rs10519694 could potentially be used for KC risk prediction.

Keratoconus (KC), or conical cornea, is characterized by bilateral, progressive, noninflammatory thinning and protrusion of the cornea, which usually leads to visual impairment from myopia and irregular astigmatism.¹ KC is the most common corneal ectatic disorder² and one of the most common reasons for corneal transplantation, especially in younger patients.3, 4, 5 Its estimated prevalence is 0.17 to 47.90 per 1000 in different populations.6, 7, 8 Both genders are equally affected, and the disease is more prevalent among Asian individuals than White individuals.^8,9

Keratoconus is a multifactorial disease with a single etiology remaining unknown. Family-based and twin studies have demonstrated that both genetic and environmental factors impact pathogenesis.10, 11, 12

Histochemical analysis shows disturbances in the homeostasis of the extracellular matrix (ECM) of the keratoconic cornea.^13,14 Lysyl oxidase (LOX),¹⁵ metalloproteinase 9,^16,17 tissue inhibitor of metalloproteinase 1 and 2,^16,17 transforming growth factor beta-induced gene (TGFBI),^16,18,19 glycoprotein fibronectin (FN1),¹⁶ integrin,¹⁶ and thrombospondin 1^16,20 are some examples of ECM-related genes involved in the development of KC.

LOX is a copper-dependent enzyme encoded by the LOX gene (location: 5q23.2) that plays an important role in maintaining the ECM by forming elastin and collagen cross-links using oxidative deamination catalysis of lysine and hydroxylysine residues.²¹ Connections between the LOX chromosomal location in familial KC and altered LOX activity in individuals with KC support the promising role of LOX in the incidence of KC.22, 23, 24 A case–control genome-wide association study in White American individuals revealed that the association of KC with single nucleotide polymorphisms (SNPs) lies within LOX.¹⁵

Several studies have already highlighted the role of LOX variants, especially SNPs, in the risk of KC.^15,23,25, 26, 27 A study in an Iranian population demonstrated that rs1800449 is significantly associated with KC and increasing KC risk, whereas a similar effect was not observed in rs2288393.²¹ Another study revealed that both rs1800449 and rs2956540 are significantly associated with KC,¹⁵ as did a separate study of European descendants examining rs2956540 and rs3735520.²⁸

Among all SNP variants located in LOX, rs2956540 and rs10519694 are 2 intronic variants located in the fourth intron of LOX. According to the Genotype-Tissue Expression (GTEx) database (https://gtexportal.org/home/), rs2956540 and rs10519694 have been known as expression quantitative trait loci in several tissues. Therefore, it is possible that rs2956540 and rs10519694 might be expression quantitative trait loci in corneal tissue, too, and alter the risk of KC. Also, SNP rs2956540 may affect gene expression through transcriptional regulation, as predicted by Genomatrix (http://www.genomatix.de/) that it can alter the binding sites of several transcription factors like PTX1, CMYB, and ISM1.²⁹

Few studies have evaluated the association of these 2 variants with KC among different populations.²⁹ The latest meta-analysis in 2015 showed that rs2956540 and rs10519694 variants have a significant association with KC, whereas it failed to replicate this result in rs1800449 and rs2288393.²⁹ This study explores the association of rs10519694 and rs2956540 with KC in a population of Iranian descent.

Methods

The study was performed in Negah Eye Hospital, Tehran, Iran, between January 2020 and January 2022. It was approved by the institutional review board and ethics committee of the Ophthalmic Research Center, Research Institute for Ophthalmology and Vision Science, Shahid Beheshti University of Medical Sciences, Tehran, Iran (IR.SBMU.ORC.REC.1399.015), and met the Declaration of Helsinki criteria. Keratoconus was primarily diagnosed based on clinical criteria including Vogt’s striae, Fleischer’s ring, corneal stromal thinning and protrusion, and videokeratography findings.30, 31, 32 The Pentacam HR (Oculus) and the Corvis ST (OCULUS Optikgeräte GmbH) were done for all suspicious cases to detect mild forms of KC (Tables 1 and 2).33, 34, 35, 36 All diagnoses were approved by 2 cornea specialists (F.D., A.B.-R.). Complete medical histories were obtained from all patients with KC. No other ocular disorders or risk factors (including contact lens use, eye rubbing, atopy, etc.) or systematic diseases (including connective tissue disorders) were uncovered. All healthy controls had no personal or familial history of eye-related, metabolic, or immune system–related disease. All participants or their legal guardians signed consent forms.

Table 1.

Pentacam Criteria for Risk of Keratectasia

Criteria	Normal	Suspect	Abnormal
Pentacam
K max (D)	< 47.2	47.2–49	> 49
Against the rule astigmatism (D)	< 1	1–2	> 2
Corneal astigmatism (D)	< 6	6–7	> 7
Thinnest point (μm)	> 500	470–500	< 470
Difference between pachy apex and thinnest location (μm)	< 10	10–20	> 20
Difference in central thickness between 2 eyes (μm)	< 10	10–30	> 30
Displacement of the thinnest point from the center (mm)	< 0.5	0.5–1	> 1
Skewed steepest radial axis (SRAX) (degrees)	< 10	10–20	> 21
IS value (inferior–superior difference at the 3 mm) (D)	< 1.4	1.4–1.9	> 1.9
IS value (inferior–superior difference at the 5 mm) (D)	< 1.4	1.4–2.5	> 2.5
Anterior elevation (μm)	< 10	10–12	> 12
Posterior elevation (μm)	< 15	15–17	> 17

D= diopter; IS value = the inferior-superior value.

Table 2.

The Clinically Suggested Cutoff Values for Keratoconus Indices in Screening Clinical and Subclinical Cases

Parameter	Clinical Keratoconus	Subclinical Keratoconus
Tomographic
CKI	∗	∗
KI	1.07	∗
IHA	10.4	∗
IHD	0.017	∗
TKC	1.0	2.0
ISV	36.6	∗
IVA	0.28	0.15
Rmin	7.04	∗
PE	20.5	10.5
IS value	1.1	1.9
KISA	60%	100%
Pachymetric
ART-Min	606	∗
ART-Max	356	368
ART-Avg	444	490
BAD_D	2.02	1.31
CCT	515	518
PPI-Min	0.87	0.80
PPI-Max	1.53	1.40
PPI-Avg	1.18	1.08
TCT	506	502

∗ = although several studies have addressed the values of these parameters, clear cutoffs have not been achieved; ART = Ambrosio's relational thickness indices; BAD_D = Belin/Ambrósio deviation; CCT = central corneal thickness; CKI = central keratoconus index; IHA = index of height asymmetry; IHD = index of highest decentration; IS value = the inferior–superior value; ISV = index surface variance; IVA = index of vertical asymmetry; KI = keratoconus index; KISA = The KISA index was derived from the following 4 indices: central K; I-S; the SRAX index, an expression of irregular astigmatism occurring in keratoconus; and the astigmatism index, which quantifies the degree of the regular corneal astigmatism (simulated K1 − simulated K2); PE = prediction error; PPI = pachymetric progression indices; Rmin = minimum radius of curvature; TCT = thinnest corneal thickness; TKC = topographic keratoconus classification.

Genotyping

Five milliliters of peripheral blood was collected from all participants and stored in an ethylenediaminetetraacetic acid (EDTA) tube at −20° C until DNA was extracted. Total genomic DNA was extracted using a salting-out protocol.³⁷ Agarose 1% gel electrophoresis and ultraviolet (UV) spectroscopy were used to assess the quality and quantity of extracted DNA, respectively. rs2956540 and rs10519694 were genotyped using tetra-primer amplification refractory mutation system–polymerase chain reaction (PCR). The 3’allele-specific primers designed by the Primer1 online tool (http://primer1.soton.ac.uk/primer1.html) were used for amplification. The tetra-primer amplification refractory mutation system–PCR was carried out in 20 μl mixtures containing 100 ng of genomic DNA, 10 μl of Taq DNA Pol 2X Master Mix Red (Amplicon), 0.5 μl of 2 outer primers, and 1 μl of 2 inner primers for each SNP, and the remainder came from double-distilled water. Polymerase chain reactions were performed using the Eppendorf thermocyclers (Eppendorf AG 22331 OMIM 148300) under the following conditions: 94° C for 4 min, followed by 35 cycles including denaturing at 94° C for 30 s; annealing for 30 s at 60° C and 62° C for rs2956540 and rs10519694, respectively; and extension at 72° C for 30 s. A final extension was conducted at 72° C for 5 min. Finally, the PCR product ran on 2% gel electrophoresis. Based on the sizes of the PCR products on the gel electrophoresis, the genotypes were determined (Table 3).

Table 3.

T-ARMS-PCR Primer Sequences, Melting Temperatures (T_m), and the Size of PCR Product for Every Genotype

SNP	Primer Sequence	PCR Product	Genotype (Blood Pressure)
rs2956540	Forward inner primer (C allele): ACTTATTTTTCCTCCATTTGCTAAGCC	209	CC (417, 209)
	Reverse inner primer (G allele): GTTTTATGCTGAAAATAGAATAGTGGTAGC	265	GG (417, 265)
	Forward outer primer (5' - 3'): CTGACATAGATTTTAACTGACACGCATT	417	CG (417, 265, 209)
	Reverse outer primer (5' - 3'): CAGTCCACAATGAAGAACAAAAATTTAC
rs10519694	Forward inner primer (C allele): AAATATTCACATCAATAAGTAAATGAAGGC	252	CC (402, 252)
	Reverse inner primer (T allele): TATTTTTCTCCTCCCAGCCTGTAGACGA	208	TT (402, 208)
	Forward outer primer (5' - 3'): TGGTTTTGAGTTTAGGTAATCAAGGTCC	402	CT (402, 252, 208)
	Reverse outer primer (5' - 3'): TGCTAGAATTGAATGGCAGTATTGAGTT

SNP = single nucleotide polymorphism; PCR = polymerase chain reaction; T-ARMS = tetra-primer amplification refractory mutation system.

Statistical Analysis

All statistical analyses were performed using SPSS Statistics for Windows, version 22.0 (SPSS Inc). Two-tailed Mann–Whitney and Pearson chi-square tests were used to compare age and gender. A Pearson chi-square test was used to compare allele and genotype frequency between cases and controls and to assess divergence from Hardy–Weinberg equilibrium (HWE). The allele risk was evaluated using a 2-tailed Fisher exact test and odds ratio with a 95% confidence interval. Logistic regression was used to fit the general linear model to the data to predict KC risk using rs2956540 and rs10519694. Statistical significance was defined as a P value of 0.05.

Results

Demographic Data

A total of 358 unrelated individuals (178 KC and 180 healthy controls) were enrolled in the study. The mean age of KC patients (92 men and 86 women) and healthy controls (97 men and 83 women) were 31.97 ± 9.37 years and 32.71 ± 6.88 years, respectively (P value = 0.33). Gender was matched in both groups (P = 0.75) (Table S4).

HWE

In association studies, we typically hypothesize that healthy controls remain in HWE, whereas the subject groups deviate from HWE. This means one of the frequencies of the genotypes is higher or lower than what would be expected from mutation, natural selection, and so on. Thus, in both rs2956540 and rs10519694, KC subjects departed from HWE (rs2956540: P value = 0.038/rs10519694: P value = 0.048), whereas healthy controls remained in HWE (rs2956540: P value = 0.93/rs10519694: P value = 0.090) (Table S5).

Allele Frequency

To calculate the frequency of each allele, the homozygotes of that allele were multiplied by 2 and added to the heterozygotes, and the result was divided by twice the population of the group. In this set of data, both rs2956540 and rs10519694 risk alleles showed higher frequencies than those displayed in the dbSNP database (rs2956540 C allele = 0.3, rs10519694 T allele = 0.2). Allele frequency did not significantly differ for rs2956540 (P value = 0.2654) and rs10519694 (P value = 0.08263) between cases and controls (Table 6).

Table 6.

rs2956540 and rs10519694 Allele Frequency

	rs2956540		rs10519694
	C	G	C	T
Case	174 (48%)	182 (52%)	210 (58%)	146 (42%)
Control	160 (44%)	200 (56%)	236 (65%)	124 (35%)
Pearson chi-square	1.2402		3.0124
df	1		1
P value	0.2654		0.0826

C = cytosine; df = degrees of freedom; G = guanine; T = thymine.

Genotype Frequency

To calculate the genotype frequency in each group, the total number of each genotype was divided by the total population of that group. Genotype frequency was significantly different between cases and controls for rs2956540 (P value = 0.019). rs2956540 C allele carriers (CC and CG) were significantly more frequent among KC cases than healthy controls (P value_chi-square = 0.015, P value_{Fisher exact} = 0.017). There was a significant difference in genotype frequency between groups for rs10519694 (P value = 0.001). T allele carriers were significantly more frequent among KC patients (P value_chi-square = 0.002, P value_{Fisher exact} = 0.001) (Tables 7 and 8). Sex stratification revealed no significant differences in genotype frequency between men and women in cases and controls (Table 9). Fitting the general linear model showed that rs10519694 could be considered a predictor for the development of KC (P value = 0.001); however, this was not observed for rs2956540 (P value = 0.323; Table 10).

Table 7.

rs2956540 and rs10519694 Genotype Frequency

	rs2956540			rs10519694
	CC	CG	GG	CC	TC	TT
Case	34	106	38	54	102	22
Control	39	82	59	84	68	28
Pearson chi-square	7.94			14.03
df	2			2
P valuechi-square	0.019			0.001

C = cytosine; G = guanine; T = thymine.

Table 8.

Comparison of C Allele Carriers of rs2956540 and T Allele Carriers of rs10519694 between Cases and Controls

	rs2956540		rs10519694
	CC + CG	GG	CC	TT + TC
Case	140	38	54	124
Control	121	59	84	96
Pearson chi-square	5.91		10.08
df	1		1
P valuechi-square	0.015		0.002
P valueFisher exact	0.017		0.001
OR (CI = 0.95)	1.79		2.01

CI = confidence interval; C = cytosine; G = guanine; OR = odds ratio; T = thymine.

Table 9.

Gender-Specific Genotype Stratification

		rs2956540			rs10519694
		CC	CG	GG	CC	TC	TT
Case	Female	17	53	16	22	49	15
	Male	17	53	22	32	53	7
Pearson chi-square		0.746			4.721
df		2			2
P value		0.689			0.094
Control	Female	19	37	27	42	31	10
	Male	20	45	32	42	37	18
Pearson chi-square		0.142			1.737
df		2			2
P value		0.932			0.420

C = cytosine; G = guanine; T = thymine.

Table 10.

Fitting Logistic Regression Model to Genotype Data

	Beta	Standard Error	P Value	OR
rs2956540	−0.265	0.269	0.323	0.767
rs10519694	−0.725	0.223	0.001	0.484
Intercept	0.670	0.290	0.021	1.954

OR = odds ratio.

Discussion

Keratoconus is a complex, multifactorial, genetic condition that can manifest in isolation or in association with other systemic genetic disorders. Bilaterality, familial aggregation,38, 39, 40 concordant monozygotic twins,⁴¹ correlation with inflammatory bowel disease, Down syndrome,⁴² Leber congenital amaurosis,⁴³ diabetes mellitus,²⁶ and ethnic⁴⁴ differences in prevalence and incidence⁴⁵ all point to a genetic cause. Genetic relationships in KC will facilitate discovering biomarkers for early detection and monitoring progression. A total of 14% of KC patients have a family history.⁴⁰ It remains uncertain how familial and sporadic KC differ genetically. Because family history does not alter genetic severity, genetic research can pool all cases.⁴⁶ Numerous studies have investigated associations between isolated KC and genetic factors.^2,46, 47, 48, 49 Several have suggested LOX to be one of the most promising genes associated with KC.²³ Using an oxidative deamination catalytic reaction, LOX affects ECM homeostasis and maintenance in collagen and elastin-rich tissues, such as the cornea.⁵⁰

Thus far, numerous studies have examined the association of KC with a wide spectrum of LOX single nucleotide variants. Thr392Pro and Pro32Leu substitutions are examples of exonic, nonsynonymous point mutations observed in 2 Brazilian and Chinese KC patients,^8,51 whereas –116C > T and –58C > T are 2 LOX 5’UTR mutations that have been observed in 2 advanced KC patients.⁵¹ rs1800449 and rs2288393 are 2 well-known intergenic SNPs whose relationships to KC have been studied extensively.^15,21,52,53 The rs2956540 and rs10519694 are 2 well-known SNPs found in the fourth intron of the LOX gene and connected to KC in different populations.^8,12,29,54

Alterations in LOX corneal tissue expression in KC patients have been demonstrated. One study showed LOX to be significantly downregulated in KC patients compared with controls.⁵⁵ Dudakova et al observed lower LOX activity in corneal fibroblast cultures from KC individuals.¹⁴

In addition to a significant drop in LOX mRNA levels, Pahuja et al showed a reduction in the ratio of LOX expression with increasing KC grades.⁵⁶ Collectively, the results of these studies suggest that alterations in LOX biological activity through changes in DNA sequences, particularly in noncoding intronic sequences, correlate with the risk of KC development.

We looked at the relationship between 2 known intronic SNPs in the LOX gene, rs2956540 and rs10519694, and KC in a population of Iranian descent. A recent study identified rs2956540-G as a risk allele for KC in a population of Europeans.²⁸ However, we found the rs2956540-C allele was highly prevalent in KC patients, whereas the G-allele was more common in the healthy population. Additionally, we noted no significant difference in rs10519694 allele frequency between cases and controls. Furthermore, we found that being an rs10519694-T carrier was more frequently and significantly observed among Iranian KC patients. Although previous studies did not clearly define which rs10519694 allele is the risk allele, we found rs10519694-T to be a significant risk factor for KC. As in previous studies, we found no significant association between gender and rs2956540 and rs10519694 alleles in both KC subjects and healthy controls.

In this study, we examined gene and allele frequencies in a referral center in a consecutive, nonrandom group of Iranian descent. Further studies on KC patients evaluating more SNPs with larger sample sizes in different populations are needed for definitive results.

Conclusion

We evaluated the association between 2 well-known LOX gene SNPs, rs2956540 and rs10519694, and KC in a population of Iranian descent for the first time. We found the rs10519694 variant is more prevalent in KC patients. Further studies should be conducted integrating other genetic information and questionnaires to assess environmental factors and characterize the severity of allergies and frequency of eye rubbing.