Isoforms of Secretory Group Two Phospholipase A (sPLA2) in Mouse Ocular Surface Epithelia and Lacrimal Glands

Abstract

Purpose.: To compare and contrast the distribution patterns of select secretory group two phospholipase A (sPLA2) isoforms in corneal epithelia (CN), conjunctival epithelia (CNJ), and lacrimal glands (LG) of BALB/c and C57BL/6 mice.

Methods.: Gene expression of select sPLA2 isoforms was quantified via real-time reverse-transcription PCR (qRT²-PCR). Immunofluorescence assay (IFA) of the sPLA2-IIa, -V, and -X isoforms were used to confirm qRT²-PCR results. sPLA2-IIa function was confirmed via in vitro CN and CNJ culturing.

Results.: qRT²-PCR revealed that sPLA2 isoforms (pla2g5, 12a, and 12b), cPLA2 isoform (pla2g4a), iPLA2 isoform (pla2g6), and PLA2-receptor (pla2r1) were present in all tissues of both strains, whereas sPLA2 isoforms (pla2g1b, 2e, and 3) were absent. sPLA2 isoforms (pla2g2a, 2d, 2f, and 10) showed tissue- and strain-specific expression: 2a in BALB/c CNJ only; 2d at higher levels in CNJ than LG; and 2f and 10 in CN and CNJ, but absent in LG. Upon dry eye (DE) induction, pla2g2a, 2d, and 2f were upregulated in BALB/c CNJ, and 10 was absent from CN. Furthermore, BALB/c DE mice showed upregulation of pla2r1 in CN and CNJ and downregulation of 12a and 12b in LG. IFA of sPLA2-IIa, -V, and -X in DE CNJ confirmed the upregulation of pla2g2a, 5, and 10. Last, in vitro CN and CNJ culturing confirmed that sPLA2-IIa amplifies ocular surface inflammation in CNJ but not in CN.

Conclusions.: sPLA2 isoforms exhibit differential expression patterns when comparing BALB/c with C57BL/6 mice; and DE with control BALB/c mice. These findings suggest that at least some sPLA2 isoforms must have significant roles in ocular surface physiology and inflammation.

Introduction

Group two phospholipase A (PLA2) enzymes hydrolyze membrane phospholipids through the sn-2 position to produce lysophospholipids and free fatty acids, including arachidonic acid (AA), which can be further converted into inflammatory mediators such as prostaglandin E₂ (PGE₂), leukotrienes, and eicosanoids. ^{1 ,2} Mammalian genomes code for more than 30 PLA2 enzyme genes. Their protein products can be grouped into six major classes based on structure, functional location, and/or calcium dependence: secreted PLA2s (sPLA2s), Ca²⁺-dependent cytosolic PLA2s (cPLA2s), Ca²⁺-independent PLA2s (iPLA2s), platelet-activating factor acetylhydrolases (PAF-AHs), lysosomal PLA2s (LPLA2), and adipose-specific PLAs (AdPLA2s). It has been suggested that many of these are involved in bioprocesses including phospholipid metabolism, membrane remodeling and homeostasis, inflammation, and cancer development. ³  

To date, more than 11 sPLA2 isoforms have been identified in humans and mice. ³ They share several characteristics, such as relatively low molecular weight, high calcium dependence, and high thermal stability. Besides the group two C gene (PLA2g2C) in humans, which is a nonfunctional pseudogene, ⁴ and the sPLA2-IIa gene in the B6 mouse line (such as C57BL/6 mice), which is naturally mutated, ^5,6 all the other sPLA2 isoform genes code for proteins that catalyze the conversion of membrane phospholipids to mediators of inflammation. Theoretically, these genes could influence the immune response; and thus, any perturbation in sPLA2 homeostasis could potentially result in a defect in immune response and disease process. 

Indeed, a number of sPLA2 isoforms have already been associated with disease. For example, sPLA2-Ib has been suggested to be important in obesity, ^7,8 diabetes, ⁹ psoriasis, ¹⁰ and meconium aspiration syndrome (MAS). ¹¹ sPLA2-III has been associated with atherosclerosis and colon cancer. ^12,13 sPLA2-V and -X have been shown to play important roles in rheumatoid arthritis (RA), ¹⁴ coronary artery disease (CAD), ¹⁵ and atherosclerosis. ^16–19 Furthermore, a number of nonsecretory PLA2 isoforms have also been associated with diseases; for example, cPLA2-4a has been shown to be involved in asthma, ²⁰ adult respiratory distress syndrome (ARDS), ²¹ and pulmonary fibrosis. ^22,23  

sPLA2-IIa has been shown to be associated with a number of inflammatory, autoimmune, and allergic diseases such as RA, asthma, septic shock, Crohn's disease, and atherosclerosis, earning its nickname “the inflammatory sPLA2.” ³ Recently, our lab has shown that sPLA2-IIa plays a significant role in modulating inflammation when the ocular surface is compromised, such as in dry eye (DE) disease. ^24,25 The fact that C57BL/6 mice were able to develop DE disease without a functional sPLA2-IIa protein product suggests that other sPLA2 isoforms must take its place in ocular surface inflammation. ^24,26 Since we have shown that sPLA2-IIa is active in the ocular surface and increases in association with DE disease, ^24,25 it will be more elucidative to compare and contrast the roles of sPLA2 isoforms in ocular inflammation with BALB/c mice that have a functional gene pla2g2a⁺ for sPLA2-IIa to C57BL/6 mice that have nonfunctional pla2g2a⁻ mutants. 

Although a number of sPLA2 isoforms have been well described in both human and animal subjects with respect to nonocular disease, only sPLA2-IIa has been extensively studied in ocular surface diseases. ^24,25 Many questions remain to be answered about the other sPLA2 isoforms with respect to ocular surface diseases, such as (1) Which of the isoform genes are normally expressed in the corneal epithelia (CN), conjunctival epithelia (CNJ), and lacrimal glands (LG) in common laboratory mouse strains, such as the BALB/c and C57BL/6 strains? (2) What adaptive changes in sPLA2 homeostasis and immune function have occurred in C57BL/6 mice lacking a functional sPLA2-IIa gene (pla2g2a)? (3) What shift occurs from normal isoform homeostasis in ocular surface to a diseased one, such as in DE? (4) Which of the sPLA2 isoforms have therapeutic potential in ocular surface disease, such as DE disease? and, finally, (5) What are their mechanisms of action? 

To address some of these questions, we used quantitative real-time reverse-transcription PCR (qRT²-PCR) to first compare sPLA2 isoform gene expression patterns in CN, CNJ, and LG of BALB/c mice with those of C57BL/6 mice. We then compared expression patterns in CN, CNJ, and LG of DE BALB/c mice with those of control BALB/c mice. Furthermore, to confirm our qRT²-PCR results, we assayed for the presence of sPLA2-IIa, -V, and -X in BALB/c CNJ via IFA. Last, we confirmed the role of sPLA2-IIa in amplification of ocular surface inflammation via in vitro CN and CNJ culturing. 

Materials and Methods

DE Mouse Model

A previously described DE C57BL/6 mouse model (scopolamine injection with air ventilation) was adopted for use in BALB/c mice. ^26–31 This DE BALB/c mouse model protocol was approved by the Institutional Animal Care and Use Committee of Mount Sinai School of Medicine in compliance with standards from the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Mice were assayed in four independent experiments using at least 20 BALB/c mice (6–8 weeks old) each. In each experiment, half of the mice (equally split into two groups of five mice per cage) were injected subcutaneously with 200 μL of 2.5 mg/mL scopolamine four times a day (9 AM, 12 PM, 3 PM, and 6 PM) for 5 to 10 days. The scopolamine-injected mice were placed in a specially designed mouse cage equipped with a fan, which was placed under a negative-pressure air-hood. The fan and hood helped minimize the humidity within the cage. The other half of the mice in each experiment (also split equally into two groups) served as the control; so they did not receive scopolamine injections and were kept under normal environmental conditions. DE diagnosis and sample collections were done as previously reported. ²⁴ The establishment of DE and ocular surface inflammation in experimental mice was further confirmed as previously described by molecular biology approaches and histologic analyses. ²⁴ The CN, CNJ, and LG of six mice (C57BL/6 strain) were used to compare their sPLA2 isoform expression pattern with that of the BALB/c strain. In all cases, one eye of each mouse was used for histology while the other for qRT²-PCR. 

Quantitative Real-Time Reverse Transcription PCR (qRT²-PCR)

Mice were euthanatized after sedation. CN cells were obtained by scraping the corneal surfaces with razor blades. CNJ and LG were excised from each eye under a dissecting microscope. Upon collection, individual samples of CN, CNJ, and LG from each mouse were immediately submerged in 100 μL of extraction buffer and kept at −80°C until further processing. 

A total of six normal BALB/c mice, six DE BALB/c mice, and six C57BL/6 mice were used for the qRT²-PCR analysis. Total RNA isolation, genomic DNA digestion, and RNA purification were performed using an Arcturus PicoPure RNA isolation kit (Applied Biosystems, Mountain View, CA). The purified total RNAs were quantified by a Nano-Drop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE), and only the RNA samples with ratios A₂₆₀/A₂₈₀ ≥ 2.0 and A₂₆₀/A₂₃₀ ≥ 1.7 were used for the qRT²-PCR assay. cDNA synthesis was performed using a SuperScript First-Strand Synthesis kit (Invitrogen, Grand Island, NY). The qRT²-PCR assays were performed using a SYBR Green/ROX PCR Master Mix from SABiosciences (Qiagen, Valencia, CA) on an ABI 7900HT high-throughput real-time PCR system as described (Applied Biosystems). All reactions were performed in a total volume of 10 μL, containing 3 ng of reverse transcribed cDNA (measurement based on the initial RNA concentration) and 250 nM of unique primer mixture for each respective gene. These specifications resulted in a threshold cycle (C_t ) of around 20 for the control GAPDH gene. Each reaction was repeated at least three times to get statistical satisfaction. A control with no reverse transcription was also performed for every tissue sample to rule out potential sample contamination with genomic DNA. 

Ten out of the eleven sPLA2 isoforms shared between humans and mice were used in the qRT²-PCR assays (sPLA2-V, sPLA2-XIIa, sPLA2-XIIb, sPLA2-IIa, sPLA2-IId, sPLA2-IIf, sPLA2-X, sPLA2-Ib, sPLA2-IIe, and sPLA2-III; the protein product of pla2g2c was excluded from the study because pla2g2c is a nonfunctional pseudogene in humans). We also included one representative from each of the cPLA2 (cPLA2-4α) and iPLA2 (iPLA2β) groups, and the sPLA2 receptor gene (pla2r1) (see Table 1 for the names of select PLA2 isoforms and their corresponding gene symbols). A panel of established and commercially available primer sets (Qiagen) was used for qRT²-PCR. The primer sets were as follows: pla2g1b (QT00103579), pla2g2a (QT00109977), pla2g2d (QT00120638), pla2g2e (QT01049125), pla2g2f (QT00173838), pla2g3 (QT01758057), pla2g4a (QT00098259), pla2g5 (QT00197806), pla2g6 (QT00136192), pla2g10 (QT00097307), pla2g12a (QT00138208), pla2g12b (QT00118398), and pla2r1 (QT00105518). Glyceraldehyde 3-phosphate dehydrogenase gene (Gapdh; QT01658692) was used as an internal control for normalization. 

Table 1.

 

Expression of Selective PLA2 Isoforms and Receptor in CN, CNJ, and LG of BALB/c and C57BL/6 Mice*

Gene ID	Product	BALB/c (n = 6)			C57BL/6 (n = 6)
		CN	CNJ	LG	CN	CNJ	LG
pla2g4a	cPLA2a	+	+	+	+	+	+
pla2g5	sPLA2-V	+	+	+	+	+	+
pla2g6	iPLA2	+	+	+	+	+	+
pla2g12a	sPLA2-XIIa	+	+	+	+	+	+
pla2g12b	sPLA2-XIIb	+	+	+	+	+	+
pla2r1	Receptor	+	+	+	+	+	+
pla2g2a	sPLA2-IIa	–	+	–	–	–	–
pla2g2d	sPLA2-IId	±	+	+	±	+	+
pla2g2f	sPLA2-IIf	+	+	–	+	+	–
pla2g10	sPLA2-X	±	+	–	±	+	–
pla2g1b	sPLA2-Ib	–	–	–	–	–	–
pla2g2e	sPLA2-IIe	–	–	–	–	–	–
pla2g3	sPLA2-III	–	–	–	–	–	–

 +, detected under the qRT²-PCR conditions (mean C_t < 35, and SD < 0.5 C_t ); –, not detected under the same qRT²-PCR conditions (mean C_t ≥ 35); ±, inconsistent, where the positive rates for pla2g2d in CN of BALB/c control, DE, and C57BL/6 mice group were 67%, 33%, and 20%, respectively; and the positive rates for pla2g10 in CN of BALB/c control, DE, and C56BL/6 were 67%, 0%, and 40%, respectively.

*  The cDNA concentration equivalent to 3 ng of total RNA was used in each reaction. Data represent the mean ± SEM of 3 repeats normalized by Gapdh.

The qRT²-PCR was run as follows: one cycle at 95°C for 10 minutes for initial activation, followed by 40 cycles for amplification (each composed of a denaturing step at 95°C for 15 seconds and an annealing/extending step at 60°C for 1 minute). For all reactions, immediately following the amplification step, a single cycle of the dissociation (melting) curve program was run at 95°C for 15 seconds, then at 60°C for 15 seconds, and last at 95°C for 15 seconds. The arbitrary abundance of mRNA expression of a target gene (n) was calculated as 2^−ΔCt(n), where C_t (the threshold cycle number) is the mean value of at least three replicate reactions of the same sample of n, and ΔC_t = C_t (target gene) – C_t (Gapdh). Any value of C_t equal to or greater than 35 was considered a negative call, or having no expression. The relative expression of a target gene to the reference gene (Gapdh) (fold of change) between groups was calculated as 2^−ΔΔCt , where the ΔΔC_t = ΔC_t (group1) – ΔC_t (group2).^{13,32 ,33} Data from the qRT²-PCR assays were first evaluated by comparing tissue mRNA expressions of our internal control gene, the constitutively expressed GAPDH gene. The average Gapdh C_t values for CN, CNJ, and LG of BALB/c control mice were 20.43 ± 0.09, 19.90 ± 0.16, and 20.33 ± 1.02, respectively (n = 18 per tissue, 6 mice times 3 replicates per mouse). The mean C_t for all three control BALB/c tissue samples was 20.22 ± 0.16. 

The same calculation method was used for DE BALB/c mice and normal C57BL/6 mice. The average Gapdh C_t values for CN, CNJ, and LG of the DE BALB/c mice were 20.57 ± 0.14, 20.18 ± 0.37, and 20.24 ± 0.23, respectively. The mean Gapdh C_t for all three DE BALB/c tissue samples was 20.33 ± 0.12. The average Gapdh C_t values for CN, CNJ, and LG of C57BL/6 mice were 20.25 ± 0.32, 19.62 ± 0.52, and 19.36 ± 0.62, respectively. The mean Gapdh C_t for all three C57BL/6 tissue samples was 19.74 ± 0.46. Calculations were also performed by tissue. The average Gapdh C_t values for CN, CNJ, and LG of all tested mice were 20.42 ± 0.09, 19.90 ± 0.16, and 19.97 ± 0.31, respectively. Thus, the cDNA input and PCR efficiency for all qRT²-PCR reactions were very close to one another. Control reactions for reagents alone, as well as genomic DNA contamination, were negative. Thus, we validated the use of the mean Gapdh C_t for normalization and comparison of our data. 

Immunofluorescence Assay (IFA)

CNJ cryosections (6–8 μm) were stained with polyclonal goat anti-mouse sPLA2-IIa, -V, or -X IgG (Santa Cruz Biotechnology, Santa Cruz, CA), followed by Alexa Fluor-488 or −594-labeled polyclonal donkey anti-goat IgG as indicated (Jackson ImmunoResearch Laboratories, Inc., Invitrogen, CO; Grand Island, NY) and 4′-6-Diamidino-2-phenylindole (DAPI), following our previously published methodology. ²⁴ Fluorescent signals were observed and recorded with a Zeiss Axioplan 2IE fluorescence microscope (Gottingen, Germany) (Mount Sinai School of Medicine Shared Research Facilities). Resulting digital images were analyzed using Photoshop CS4 (Adobe Systems, Inc., San Jose, CA) and Image-J software (NIH; http://rsbweb.nih.gov/ij/docs/index.html) by an unbiased observer. 

In Vitro CN and CNJ Culturing and PGE₂ Analysis

Control and DE CNJ, and normal CN of BALB/c mice were each washed with ice-cold PBS and incubated in 95% air and 5% CO₂ at 37°C for 1 hour in 96-well plates containing 200 μL of SHEM's medium supplemented with 5% fetal bovine serum. Then, control and DE CNJ cultures were treated at 37°C for 12 hours with SHEM's medium with 25 μg/mL of human recombinant sPLA2-IIa or without. 

Normal CN cultures were treated at 37°C for 12 hours with SHEM's medium with no sPLA2-IIa or TNF-α as control, with 25 μg/mL of human recombinant sPLA2-IIa, 10 ng/mL of TNF-α, or TNF-α + sPLA2-IIa, in addition to varying amounts of sPLA2-IIa inhibitor S-3319 (see Fig. 9). In CN and CNJ cultures, PGE₂ production was then measured. 

Statistical Analysis

Data are presented as the mean ± SEM or SD as indicated. Statistical analyses were run in computer-generated 2-tailed bivariant Student's t-tests using Excel or in 1-way ANOVA using SPSS (SPSS Inc., Chicago, IL). Two-tailed significance was established at a confidence level of 0.05 > P > 0.95. 

Results

Overall Evaluation and PLA2 Isoform Patterns

Data from the qRT²-PCR assays were first used to compare and contrast gene expression of the 10 sPLA2 isoforms shared between humans and mice, plus cPLA2-4α, iPLA2β, and the sPLA2 receptor. As shown in Table 1, qRT²-PCR revealed that sPLA2 isoforms pla2g5, 12a, 12b, the cPLA2 isoform pla2g4a, the iPLA2 isoform pla2g6, and the sPLA2 receptor pla2r1 were detected in all three tissues of both strains, whereas sPLA2 isoforms pla2g1b, 2e, and 3 were not detected in any tissue of either strain. Furthermore, sPLA2 isoforms pla2g2a, 2d, 2f, and 10 exhibited tissue- and/or strain-specific expression patterns: pla2g2a was detected in BALB/c CNJ only; in both strains, pla2g2d was detected at consistently higher levels in CNJ than in LG but was variable in CN, the pla2g2d-positive rates in CN for BALB/c control, BALB/c DE, and C57BL/6 strains were 67%, 33%, and 20%, respectively; in both strains, the pla2g2f was detected in CN and CNJ, but not in LG; in both strains, the pla2g10 was detected in all CNJ but not in any LG, and the pla2g10-positive rates in CN for BALB/c control, BALB/c DE, and C57BL/6 strains were 67%, 0%, and 40%, respectively. 

sPLA2 Isoform Expression in CN, CNJ, and LG of Control BALB/c Mice

The relative mRNA abundance values for the selected PLA2 isoforms were then calculated for all three BALB/c tissues via the formula 2^−Ct(n); C_t ≥ 35 signified no expression. As shown in Figure 1, the most significant feature was pla2g2a, which was detected only in CNJ as expected, albeit at much lower levels relative to the other selected sPLA2 isoforms. ³⁴ The pla2g2f and 10 were also shown to be special, as they were not detected at all in LG under our current experimental settings. The observed trend for tissue-specific activity levels of the selected PLA2 isoforms was: CNJ > CN > LG. 

Figure 1.

 

mRNA abundance of select PLA2 isoforms in CN, CNJ, and LG of BALB/c mice. Data were normalized by the average of triplicate measurements of GAPDH gene transcription among the three tissues and displayed as mean 2^−ΔCt ± SEM. ΔCt = Ct _(nGOI) − Ct _{(n Gapdh)}; GOI, gene n of interest; Ct _(nGOI), mean threshold cycle from triplicate reactions for specific gene of interest in individual CN, CNJ, or LG samples; Ct _(nGAPDH), mean threshold cycle from triplicate reactions for GAPDH in the same samples for GOI; n = 6 mice × 3 replicates = 18 reactions.

The pla2g4a, 6, and 12a were the dominant isoforms, constitutively expressed in all three tissues. The pla2g4a was the highest transcribed isoform in CN and CNJ, while pla2g12a was the highest transcribed in LG. The pla2g6 and 12a were the dominant isoforms in LG. The pla2g2f, 6, 12a, and 5 also showed very high levels in CN besides 4a. Although sPLA2 receptor pla2r1 was detected higher in CNJ than in CN, they were at similar levels; whereas in LG, the expression was five times lower compared with that in CNJ or CN. 

sPLA2 Isoforms Expressed in CN, CNJ, and LG of C57BL/6 Mice

Similar analyses were performed for CN, CNJ, and LG of C57BL/6 mice. The relative mRNA expression values between the two murine strains are summarized in Tables 1 and 2. First, as shown in Figure 2, pla2g2a exhibited the most statistically significant difference between the two mouse strains, since it was not detectable in any of the C57BL/6 mouse tissues. Second, as shown in Table 2 and Figure 2, although pla2g4a, 6, and 12a were still the dominant isoforms constitutively expressed in all three tissues, their overall expression levels were much lower in C57BL/6 mouse tissues when compared with those of BALB/c mice. Still, pla2g4a, 6, and 12a were the dominant isoforms among all three tissues of both strains, suggesting their expression is constitutive and their functions are conservative. In general, compared with the isoforms of control BALB/c mice, the isoforms in C57BL/6 tissues (except for pla2g2d in CN and pla2g2f in CNJ) exhibited lower levels of transcription (Table 2). The receptor pla2gr1 expression was significantly lower in C57BL/6 CN and CNJ (less so in LG), further reflecting overall lower levels of PLA2 isoform expression in C57BL/6 tissues. 

Table 2.

 

Relative mRNA Expression Levels in Controls: C57BL/6 (n = 6) versus BALB/c (n = 6) Mice

Gene	CN		CNJ		LG
	Mean ± SEM	P Value	Mean ± SEM	P Value	Mean ± SEM	P Value
Gapdh	1.00 ± 0.00	N/A	1.00 ± 0.00	N/A	1.00 ± 0.00	N/A
pla2g2a	0.00 ± 0.00	N/A	0.00 ± 0.00	N/A	0.00 ± 0.00	N/A
pla2g2d	3.27 ± 1.38	N/A	−5.99 ± 0.07	0.00	−1.39 ± 0.28	0.18
pla2g2f	−1.53 ± 0.32	0.15	11.91 ± 11.53	0.01	0.00 ± 0.00	N/A
pla2g4a	−3.57 ± 0.16	0	−2.49 ± 0.24	0.03	−1.84 ± 0.43	0.24
pla2g5	−1.70 ± 0.22	0.05	−12.18 ± 0.04	0.03	−1.05 ± 0.60	0.47
pla2g6	−3.37 ± 0.13	0	−2.81 ± 0.18	0.01	−3.49 ± 0.15	0.01
pla2g10	0.00 ± 1.17	0.01	−2.43 ± 0.10	0.02	0.00 ± 0.00	N/A
pla2g12a	−5.52 ± 0.05	0.01	−4.86 ± 0.10	0.00	−7.19 ± 0.06	0.00
pla2g12b	−1.30 ± 0.14	0.18	−1.95 ± 0.20	0.04	−5.33 ± 0.07	0.23
pla2r1	−2.72 ± 0.12	0.07	−3.10 ± 0.13	0.00	−1.24 ± 0.55	0.39

 N/A, not applicable.

Figure 2.

 

mRNA abundance of select PLA2 isoforms in CN, CNJ, and LG of C57BL/6 mice. Data were normalized the same way as described in Figure 1; n = 6 mice × 3 replicates = 18 reactions.

Also, there were significant differences noted in the overall isoform distribution patterns in CN and CNJ between the two strains; LG patterns were very similar between the two strains. For example, pla2g2f expression in the C57BL/6 strain was 2-fold lower in CN and 12-fold higher in CNJ; pla2g2d expression in C57BL/6 mice showed differences in the opposite direction: a 3-fold increase in CN and a 6-fold decrease in CNJ. We speculate these differences in distribution may perhaps be part of an adaptive strategy to compensate for the loss of a functional sPLA2-IIa in C57BL/6 mice. 

sPLA2 Isoform Expression Patterns Change Upon Desiccation

Next, we studied changes that occurred in expression patterns of select PLA2 isoforms in the inflamed ocular surface of desiccation-induced DE BALB/c mice. We compared and contrasted isoform expression levels between CN, CNJ, and LG of normal and DE BALB/c mice. A summary of select PLA2 isoform gene expression patterns can be found in Figure 3. 

Figure 3.

 

mRNA abundance of select PLA2 isoforms in CN, CNJ, and LG of DE BALB/c mice. Data were normalized the same way as described in Figure 1; n = 6 mice × 3 replicates = 18 reactions.

As shown in Table 3, besides the expected upregulation of “inflammatory sPLA2-IIa” in DE mice (see Figs. 4, 5, and 8), the other very significant change in response to DE induction in BALB/c mice was the disappearance of pla2g10 in CN and LG. The pla2g2f was shown to be upregulated in DE CNJ, with an experimentally confirmed large intrastrain variance (fold of change is 32 ± 31, P value = 0.17). The pla2g5 was upregulated in all three DE tissues. In fact, all of the isoforms in CNJ and most of them in CN were upregulated, whereas all the tested isoforms except pla2g5 in LG were downregulated in DE mice compared with that of the untreated controls. Consistent with this pattern, the pla2gr1 receptor was upregulated in CN and CNJ but was downregulated in LG of DE mice. These results suggest that multiple PLA2 isoforms participate in DE pathogenesis in BALB/c DE mice. 

Table 3.

 

Relative mRNA Expression Levels in BALB/c Mice: DE (n = 6) versus Control (n = 6)

Gene	CN		CNJ		LG
	Mean ± SEM	P Value	Mean ± SEM	P Value	Mean ± SEM	P Value
Gapdh	1.00 ± 0.00	N/A	1.00 ± 0.00	N/A	1.00 ± 0.00	N/A
pla2g2a	0.00 ± 0.00	N/A	1.66 ± 0.83	0.24	0.00 ± 0.00	N/A
pla2g2d	1.21 ± 0.51	0.34	1.71 ± 0.73	0.18	−1.31 ± 0.12	0.10
pla2g2f	−1.00 ± 0.11	0.49	32.03 ± 31.00	0.17	0.00 ± 0.00	N/A
pla2g4a	1.35 ± 0.20	0.05	1.43 ± 0.65	0.27	−4.05 ± 0.11	0.04
pla2g5	1.05 ± 0.10	0.36	1.42 ± 0.70	0.31	2.05 ± 1.29	0.27
pla2g6	1.26 ± 0.13	0.02	1.82 ± 0.75	0.15	−1.18 ± 0.16	0.24
pla2g10	0.00 ± 0.00	N/A	1.14 ± 0.29	0.35	0.00 ± 0.00	N/A
pla2g12a	−1.57 ± 0.15	0.11	1.20 ± 0.60	0.37	−2.39 ± 0.06	0.00
pla2g12b	1.10 ± 0.20	0.33	1.23 ± 0.49	0.33	−3.92 ± 0.22	0.20
pla2r1	1.60 ± 0.55	0.10	1.56 ± 0.63	0.20	−1.68 ± 0.26	0.18

 N/A, not applicable.

Figure 4.

 

Relative mRNA expression levels of sPLA2-IIa in CN, CNJ, and LG of DE BALB/c mice. Data of DE BALB/c mice or control mice were first normalized with their corresponding levels of Gapdh and then compared with each other, respectively. Data are presented as mean value ± SEM, n = 6 in DE or control groups, with three replicate reactions for each sample of each mouse.

Figure 5.

 

Immunofluorescence assay of sPLA2-IIa in CNJ of normal C57BL/6 mice (a), control BALB/c mice (b, d), and DE BALB/c mice (c, e, f). Fresh frozen samples were mounted on slides, fixed with ice-cold acetone, stained with primary polyclonal goat anti-mouse sPLA2-IIa IgG (Ab1, 1:50 dilution), followed by Alexa_Fluor-594–labeled polyclonal donkey anti-goat IgG (1:200 dilution, shown in red), and counterstained with 4′-6-Diamidino-2-phenylindole (DAPI, shown in blue). The yellow arrows point to some of the sPLA2-IIa–positive stained cells. The magnifications of images a–e are ×100, while f is the higher-magnification image (×400) of the inset in e.

sPLA2-IIa Is Activated in DE Pathogenesis

As demonstrated by the qRT²-PCR results illustrated in Figure 4, pla2g2a mRNA expression for sPLA2-IIa was upregulated only in DE CNJ, but not in either DE CN or DE LG, suggesting that sPLA2-IIa was specifically induced, either directly or indirectly, by desiccation. 

To confirm our qRT²-PCR results on mRNA expression, we chose to assay for the presence and distribution pattern of the protein product sPLA2-IIa via indirect immunofluorescence. Two negative controls were introduced: the first, as shown in Figure 5a, was C57BL/6 tissues that had no sPLA2-IIa expression and therefore showed no staining with primary sPLA2-IIa antibody Ab1; the second control, as shown in Figures 5b and 5c, was BALB/c tissues (both normal and DE) in the absence of primary sPLA2-IIa antibody showing no staining. These controls helped confirm that the cells showing staining were sPLA2-IIa–positive cells. 

With the primary sPLA2-IIa antibody Ab1 added, BALB/c control CNJ showed some scattered positive staining (Fig. 5d), while BALB/c DE CNJ showed heavy staining (Fig. 5e). The insets of Figure 5e were further examined at higher magnification as shown in Figure 5f. However, neither BALB/c control nor DE displayed any sPLA2-IIa–positive staining in CN or LG. In addition, IFA with Ab1 of sPLA2-V and -X showed increased staining in CNJ (Figs. 6, 7). The relative protein expression levels of the three sPLA2 isoforms in CNJ are summarized in Table 4. 

Figure 6.

 

Immunofluorescence assay of sPLA2-V in CNJ of control and DE BALB/c mice. Fresh frozen samples were mounted on slides, fixed with ice-cold acetone, stained with primary polyclonal goat anti-mouse sPLA2-V IgG (Ab1, 1:50 dilution), followed by Alexa-Fluor-488–labeled polyclonal donkey anti-goat IgG (1:200 dilution, shown in green in c and g), and counterstained with DAPI (shown in blue in b and f). The yellow arrows point to some of the sPLA2-V–positive stained cells. The controls were stained without Ab1 (d and h). To help locate the CNJ position, the same samples with hematoxylin and eosin staining were included (a and e). Magnification is ×200.

Figure 7.

 

Immunofluorescence assay of sPLA2-X in CNJ of control and DE BALB/c mice. Fresh frozen samples were mounted on slides, fixed with ice-cold acetone, stained with primary polyclonal goat anti-mouse sPLA2-X IgG (Ab1, 1:25 dilution), followed by Alexa-Fluor-488–labeled polyclonal donkey anti-goat IgG (1:200 dilution, shown in green in a and c), and counterstained with DAPI (shown in blue). The yellow arrows indicate the sPLA2-X–positive stained regions. The controls were stained without Ab1 (b and d). Magnification is ×400.

Table 4.

 

Relative Expression Levels of Selected sPLA2 Isoforms in CNJ

		sPLA2-IIa (Mean ± SD)	sPLA2-V (Mean ± SD)	sPLA2-X (Mean ± SD)
DE/Ctrl (x-fold) (signal)*	CNJ	2.03 ± 2.51	1.99 ± 1.66	3.45 ± 2.37
DE/Ctrl (x-fold) (signal/nuclei)†	CNJ	2.76 ± 4.47	1.99 ± 0.5	1.96 ± 2.47

 Ctrl, control eye.

*  The ratio was calculated based on the positive signals of sPLA2 isoforms in CNJ of DE/Ctrl.

†  The ratio was calculated as follows: (positive signal of DE/DAPI signal of the same area)/(positive signal of Ctrl/DAPI signal of the same area).

Our IFA results are consistent with our qRT²-PCR results for sPLA2-IIa, -V, and -X mRNA expression patterns in BALB/c mice, confirming that sPLA2-IIa, -V, and -X were upregulated in BALB/c mice in response to DE induction. 

sPLA2-IIa Amplifies Inflammation in DE CNJ Cultures

Previous studies have demonstrated a near 2-fold increase in the activity and concentration of sPLA2-IIa in the tears of DE patients when compared with the tears of the age-matched controls. ^25,35 The increase was confirmed to be the result of transcriptional upregulation of pla2g2a. Also, the upregulation of pla2g2a was accompanied by the upregulation of inflammatory cytokine gene transcription. ^24,25 This is true in both humans and mice, indicating that sPLA2-IIa plays an important role in ocular surface inflammation in DE disease. Thus, published data suggest that sPLA2-IIa activity is mainly from CNJ; LG and CN are not the major source of the sPLA2-IIa isoform. 

Since normal tears contain high concentrations of sPLA2-IIa but do not cause ocular surface inflammation, while tears from a traumatized ocular surface have a 2-fold increase in sPLA2-IIa and do cause inflammation, it was of interest to us to further examine the role of sPLA2-IIa in ocular surface inflammation. This was accomplished using our previously established in vitro CNJ culture model. ^24,25  

As shown in Figure 8a, without the addition of sPLA2-IIa, BALB/c CNJ (both DE and control) PGE₂ production was approximately 4 μg/mL; whereas with the addition of sPLA2-IIa, control CNJ PGE₂ production increased 4-fold and DE CNJ PGE₂ production increased 7-fold. These results show that sPLA2-IIa significantly stimulates CNJ inflammation via PGE₂ production, especially in DE CNJ. 

Figure 8.

 

sPLA2-IIa amplifies PGE₂ production and inflammation in CNJ cultures. Conjunctival organs from control and DE BALB/c mice were pre-incubated in fresh SHEM's medium for 1 hour, followed by an additional 12 hours' incubation with fresh medium containing either 25 μg/mL of human recombinant sPLA2-IIa alone (a), or with 25 μg/mL of the sPLA2-IIa and the indicated amount of S-3319 inhibitor together (b). Values are presented as mean ± SEM (μg/mL) of two experiments. Each experiment contained replicate treatments at indicated conditions (n = 4 eyes in each treatment group). The relative inhibition of PGE₂ production is the ratio of PGE₂ produced by CNJ with sPLA2-IIa against CNJ without sPLA2-IIa at an indicated inhibitor concentration. CNJ, control BALB/c CNJ; DE CNJ, dry eye BALB/c CNJ.

Figure 9.

 

The role of sPLA2-IIa in CN cultures. Normal BALB/c CN were nucleated and pre-incubated in SHEM's medium for 1 hour, followed by an additional 12 hours' incubation with SHEM's medium containing 10 ng/mL of TNF-α alone, 25 μg/mL of human recombinant sPLA2-IIa alone, or the two together, either with or without the indicated amount of S-3319 inhibitor. Values are presented as mean ± SEM (ng/mL) of replicate measurements at indicated conditions (n = 4 in TNF-α- or sPLA2-IIa–treatment groups and n = 2 in S-3319 alone groups).

When various amounts of sPLA2-IIa–specific inhibitor (S-3319) were added, PGE₂ production was reduced in an apparently dose-dependent manner (Fig. 8b). DE CNJ was more sensitive to S-3319 inhibition than the healthy CNJ, indicating that the amplification of PGE₂ production and inflammation is likely both sPLA2-IIa– and DE-dependent. 

We further examined the role of sPLA2-IIa on CN culture using an adapted in vitro model as above. In comparison with CNJ cultures, CN cultures showed much less response: the relative PGE₂ production (fold of change ± SEM) was 1 ± 0.22, 1.36 ± 0.46, 1.64 ± 0.56, and 2.85 ± 0.19, respectively, for treatment with no addition control, 10 ng/mL TNF-α, 25 μg/mL sPLA2-IIa, and the combination of TNF-α + sPLA2-IIa (Fig. 9). The addition of S-3319 inhibitor at all tested concentrations did not show any significant inhibition on PGE₂ production of either the TNF-α only or the sPLA2-IIa only treatment. Paradoxically, the addition of S-3319 inhibitor to the TNF-α + sPLA2-IIa treatment showed an increase in PGE₂ production (2.94 ± 0.06–fold at 2 μM of S-3319 and 3.93 ± 0.09–fold at 10 μM of S-3319). Similarly, the addition of 10 μM of S-3319 to the CN control treatment also showed an increase in PGE₂ production (2.63 ± 0.23–fold). We therefore reason that the addition of S-3319 itself or the solvent of S-3319 (DMSO) caused the significant increase in PGE₂ production in CN in the sPLA2-IIa + TNF-α treatment, and not the sPLA2-IIa + TNF-α treatment. So, sPLA2-IIa could not have been the cause of the PGE₂ increase in normal CN. 

Discussion

sPLA2 Isoforms in Ocular Surface Inflammation

The major finding of this study is that a variety of sPLA2 isoforms participate in ocular surface inflammation. The upregulation of sPLA2-IIa, -V, and -X in BALB/c CNJ in response to DE induction was evidenced by RT²-PCR results and confirmed by IFA. The important role of sPLA2-IIa in amplification of ocular surface inflammation was further elucidated using in vitro CN and CNJ organ cultures in the presence or absence of inflammation inducers and the sPLA2-IIa–specific inhibitor S-3319. Our results of these in vitro assays demonstrated that sPLA2-IIa is not only induced specifically in BALB/c CNJ but also likely functions mainly on CNJ. Additionally, in comparing sPLA2 isoform expression patterns between BALB/c and C57BL/6 mice, it became apparent that each strain harbors a different sPLA2 isoform set to modulate the pathophysiologic processes; namely, inflammation at the ocular surface. 

We came to these conclusions by observing the following four points: (1) the pla2g2a is expressed only in BALB/c CNJ; (2) most sPLA2 isoforms in CN/CNJ/LG displayed significant differences in expression between the C57BL/6 and BALB/c strains, possibly as a compensatory mechanism for the loss of a functional pla2g2a gene in C57BL/6 mice; (3) the overall expression levels of select sPLA2 isoforms are higher in the BALB/c strain than they are in the C57BL/6 strain; and (4) the sPLA2 receptor (pla2gr1) expression is higher in BALB/c mice than it is in C57BL/6 mice. 

A number of mouse strains, including the C57BL/6 strain, have been reported to lack a functional sPLA2-IIa gene because of mutations in the pla2g2a gene. The lack of a functional sPLA2-IIa gene does not seem to result in an obvious functional disadvantage in either health or disease. ^6,26,31,36 This phenomenon implies either of two possibilities: (1) sPLA2-IIa is not as active in inflammation as we had thought; or (2) the absence of sPLA2-IIa can be fully or partially compensated for by other sPLA2 isoforms maintaining sPLA2 homeostasis and function. The former possibility can be ruled out because many inflammatory diseases and cancers have been shown to be associated with changes in sPLA2-IIa expression. ^3,37–40 The latter possibility is supported by the existence of more than 10 functionally active sPLA2 isoforms in both humans and rodents. ¹ Research confirms that different isoforms are present in different species. 

Accumulated evidence supports the important pathophysiologic roles of sPLA2-IIa in response to various inflammatory stimulants. Of interest, the differential expression of sPLA2 isoforms between BALB/c and C57BL/6 mice has been shown to affect immune response and inflammation in nonocular tissues under certain conditions. For example, sPLA2-Ib is specifically expressed in murine gastric mucosa in its inactive pro-enzyme form. This pro-enzyme is activated by proteases in situ. Under normal environmental conditions, sPLA2-Ib is sufficient to mediate the immune response in the glandular stomach in both murine strains. However, when infected by Helicobacter felis , glandular pro-enzyme activation is greatly reduced. BALB/c mice compensate via activity of their functional sPLA2-IIa and are thus able to mount an immune attack against the infection. C57BL/6 mice on the other hand cannot compensate and are therefore much more sensitive to the infection. ^41,42  

The differences between BALB/c and C57BL/6 mice are further exhibited in their T-cell–mediated immune response, such as in experimental autoimmune uveitis, ⁴³ tumorigenesis, ⁴⁴ and desiccation-induced ocular surface inflammation. ^36,45 In all instances, the BALB/c strain was shown to have more CD4⁺CD25⁺ Treg cells and was more susceptible to the suppression of CD4⁺CD25⁻ Treg cells than the C56BL/6 strain. ⁴⁶ C57BL/6 mice were more likely to display a Th-1–type response, whereas BALB/c mice were more likely to display a Th-2–type response. ^36,45 BALB/c Treg cells were found to be more efficient at ablating the CD4⁺CD25⁻ effector T-cell response than were C57BL/6 mice. ⁴⁷  

From the studies discussed, it is evident that the differences in sPLA2 isoform expression between the two murine strains are fundamental and go much further beyond the pla2g2a gene mutation. Given that both humans and the BALB/c strain express sPLA2-IIa, and that sPLA2-IIa is actively involved in ocular surface inflammation, the BALB/c strain provides a more helpful model than does the C57BL/6 strain for studying ocular surface inflammation and the phospholipase cascade. 

sPLA2 Isoforms in DE Disease

Our study reveals that multiple sPLA2 isoforms are involved in the DE process, including at least pla2g2a, 5, and 10; and possibly 2f and 12a. Also, many (if not all) of the tissue components of the tear functional unit, such as CN, CNJ, and LG, are affected. This suggests that at least some of the isoforms are more active in ocular surface inflammation than others. ⁴⁸ Previous studies have shown sPLA2-IIa, -V, and -X to be associated with many inflammatory diseases including rheumatoid arthritis, atherosclerosis, and Crohn's disease. ²⁴ Previously, Kolko et al. found that in the human retina, pla2g1b is highly expressed in cells of neurodermal origin and is downregulated in proliferating retinal pigment epithelium and in several diseased corneal endothelial cells. In rat retina, pla2g1b and 5 were expressed more than pla2g2f and 10; and expression of pla2g1b, 5, and 10 was significantly induced via light-induced retinal damage. ^49–51 Landreville and coworkers, using human corneal epithelial cell extracts, detected higher expression of sPLA2-III, -IVa/c, -X, and -XIIa, but not of -Ib, -IIa, -IId, -IIe, -IIf or -IVb.⁵² In agreement with these findings, our results show the absence of both pla2g1b and 2e in CN, CNJ, and LG of both strains. This indicates that these two isoforms are specific to the neuroderm. 

The absence of sPLA2-III (pla2g3) expression in all tested mouse samples was a surprise, however. sPLA2-III is unique among the sPLA2 isoforms in that it possesses unusual N- and C-terminal domains. Its central domain is cleaved out in the matured enzyme, making it more closely related to bee venom than to its mammalian counterparts. ³ sPLA2-III was shown to have important roles, together with sPLA2-IIa, -V, and -X, in LDL/HDL metabolism and pathogenesis of atherosclerosis. ¹² It was even proposed as a candidate biomarker for human colorectal adenocarcinoma. ¹³ A plausible explanation for the absence of sPLA2-III in our samples is that the sPLA2-III prefers to be expressed in neuronal-like cells, promoting their differentiation and apoptosis. ^53,54  

Our findings of the absence of pla2g10 transcription in CN and LG of DE mice accompanied by changes in pla2g2a, 2d, 2f, and 5 expression in all three tested tissues imply the existence of a functional cohort of PLA2 isoforms constellated to keep sPLA2 homeostasis on the ocular surface and mediate DE inflammation. 

Although many studies have shown sPLA2-IIf (pla2g2f) to be present in many disease processes, ^{12–14,17,52,55} its role has never been clearly defined. In our RT²-PCR results, sPLA2-IIf expression showed one of the greatest differences between control and DE BALB/c CNJ (Fig. 5), and between BALB/c and C57BL/6 strains (Fig. 3). Its potential role in the pathophysiology of disease on the ocular surface need to be further explored. 

Finally, sPLA2-IIa (pla2g2a), the unique enzyme that is present only in BALB/c CNJ, functions to amplify inflammation in compromised and damaged CNJ epithelia (as in DE disease), but not in compromised CN. This amplification effect on CNJ can be diminished by sPLA2-IIa–specific inhibition. These results are consistent with our previous findings on humans and mice. ^24,25 In the past, the sPLA2-IIa (pla2g2a) isoform was found to be overexpressed in patients with open angle glaucoma and exfoliation glaucoma, whereas expression levels of sPLA2-V (pla2g5), cPLA2 (pla2g4), and iPLA2 (pla2g6) were found not to be significantly altered. ^56,57  

Future studies to elucidate the roles and regulatory mechanisms of sPLA2 isoforms on the ocular surface will improve our understanding of inflammation, such as in DE disease, and provide novel targets and approaches to treatment. Knockout mouse strains with isoform-specific inhibition will be crucial in our research efforts. 

Acknowledgments

The authors thank Kristina Phu for her technical support with IFA and Richard Deng for his contributions in statistical analysis of data and editing of the manuscript. 

Footnotes

 Supported in part by the Martin and Toni Sosnoff Foundation.

Footnotes

 Disclosure: Y. Wei, None; A. Pinhas, None; Y. Liu, None; S. Epstein, None; J. Wang, None; P. Asbell, None