|Home | About | Journals | Submit | Contact Us | Français|
Mutations of the PITX2 gene cause Axenfeld-Rieger syndrome (ARS) and glaucoma. In this study, the authors investigated genes directly regulated by the PITX2 transcription factor to gain insight into the mechanisms underlying these disorders.
RNA from nonpigmented ciliary epithelium cells transfected with hormone-inducible PITX2 and activated by mifepristone was subjected to microarray analyses. Data were analyzed using dCHIP algorithms to detect significant differences in expression. Genes with significantly altered expression in multiple microarray experiments in the presence of activated PITX2 were subjected to in silico and biochemical analyses to validate them as direct regulatory targets. One target gene was further characterized by studying the effect of its knockdown in a cell model of oxidative stress, and its expression in zebrafish embryos was analyzed by in situ hybridization.
Solute carrier family 13 sodium-dependent dicarboxylate transporter member 3 (SLC13A3) was identified as 1 of 47 potential PITX2 target genes in ocular cells. PITX2 directly regulates SLC13A3 expression, as demonstrated by luciferase reporter and chromatin immunoprecipitation assays. Reduction of PITX2 or SLC13A3 levels by small interfering RNA (siRNA)-mediated knockdown augmented the death of transformed human trabecular meshwork cells exposed to hydrogen peroxide. Zebrafish slc13a3 is expressed in anterior ocular regions in a pattern similar to that of pitx2.
The results indicate that SLC13A3 is a direct downstream target of PITX2 transcriptional regulation and that levels of PITX2 and SLC13A3 modulate responses to oxidative stress in ocular cells.
The PITX2 transcription factor was identified as a causative gene for Axenfeld-Rieger syndrome (ARS; Mendelian Inheritance in Man no. 180500) by Semina et al.1 ARS is a rare autosomal dominant disorder that affects anterior eye development2 and can occur with or without systemic findings. The ocular findings include structural anomalies of the anterior chamber angle and aqueous drainage structures. Nonocular features typically include facial, dental, and umbilical defects. Our recent study3 showed that ARS is associated with an approximately 75% risk for glaucoma in patients with FOXC1 or PITX2 defects. Glaucoma, the second leading cause of blindness throughout the world,4–6 is the most important clinical consequence of ARS.7,8 Glaucoma is often associated with elevated intraocular pressure (IOP), as regulated by the ciliary body (the site of aqueous humor production) and the limbal region (which includes the trabecular meshwork), the principal site of aqueous humor outflow.4,5 The disease can continue to progress in patients even when IOP is significantly reduced. Research conducted in the past two decades has revealed that oxidative stress is involved in glaucoma pathogenesis.9–16 Identifying genes involved in adult anterior segment function or responses to oxidative stress will greatly improve our understanding of glaucoma pathology and will result in finding better treatment options.
The PITX2 protein contains a conserved 60 amino acid homeodomain of the paired-bicoid class that is responsible for DNA binding, localization to the nucleus, and protein-protein interaction.17 Point mutations in PITX2 have been demonstrated to be loss-of-function mutations affecting DNA binding and transcriptional transactivation.1,18–27 Tight control of PITX2 expression is required for normal development of the eye because too much or too little activity of this transcription factor results in anterior segment defects and glaucoma.24,28 We hypothesize that altered regulation and expression of PITX2 target genes in adult tissues may have a significant impact on the development and progression of ARS-associated glaucoma and may reveal the cellular pathways affected in primary glaucoma phenotypes.
We previously designed a hormone-inducible transcription factor expression system to discover target genes for FOXC129 and have now adapted this system to identify PITX2 target genes in ocular cells. We report here the identification and characterization of the Solute carrier family 13 sodium-dependent dicarboxylate transporter member 3 (SLC13A3) as a gene directly regulated by PITX2. Its protein product, NaDC3, is implicated as an important transporter of neuronal metabolites30 and may regulate the uptake of glutathione (GSH) to protect against ocular oxidative stress.31 We demonstrate for the first time that PITX2 and SLC13A3 are necessary for cellular responses to oxidative insult from hydrogen peroxide.
PCR was used to amplify the homeodomain (HD) of wild-type recombinant PITX2A.28 The forward and reverse primers used were 5′-GAT CAT GGG CCC CAA AGG CAG CGG-3′ and 5′-ATG ATC CGG ACC GGG CTC CCT CTT TCT CCA TTT G-3′. The HD fragment was cloned into the pGEM-T vector (Promega, Madison, WI) using the manufacturer's protocol and then subcloned in-frame into ApaI/RsrII sites of pcDNAFOXC1-HR29 (N-terminal Xpresstag; Invitrogen, Burlington, ON, Canada) replacing the forkhead domain.
A BstBI site was inserted between amino acids 138 and 139 of recombinant Xpress-PITX2A28 by site-directed mutagenesis. The hormone-responsive (HR) progesterone receptor binding domain was PCR cloned from the pSWITCH vector (Invitrogen) and then cloned in-frame into the BstBI site.
PITX2C cDNA was amplified by RT-PCR of human trabecular meshwork RNA (forward, 5′-GAA TTC ATG AAC TGC ATG AAA GGC CCG C-3′; reverse, 5′-CAG GCT CAA GTT ACA CAT AT-3′) and subcloned in-frame into EcoRI sites of pCI-HA-PITX2A32 to replace the isoform A-specific region. The PITX2C insert from pCI-HA-PITX2C was subcloned into the XbaI/EcoRI sites of pcDNA4HisMaxA. Xpress-PITX2C mutants T30P, V45L, and R52C were produced by subcloning the XhoI/NarI fragment from pcDNA4HisMax-PITX2A mutant constructs.27
Different lengths of the region immediately upstream of the 5′UTR of human SLC13A3 transcript variant 1 (GenBank accession NM_022829) were PCR amplified from genomic DNA. The primers used for these constructs were −403/+165 (forward, 5′-GCT AGC TCC ACT TTC AGG CTC GAG G-3′; reverse, 5′-CTC GAG GGA GGA GGC GTT ACC TTG G-3′) and −163/+165 (forward, 5′-GCT AGC AGG GAG AGG GCG GGT TCC-3′; reverse, 5′-CTC GAG GGA GGA GGC GTT ACC TTG G-3′). The SLC13A3 upstream elements were subcloned into vector (pGL3Basic; Promega). Deletion of PITX2 binding site A to generate the −163/+165Δ construct was accomplished using a site-directed mutagenesis kit (QuikChange Lightning; Agilent Technologies, La Jolla, CA) by following the manufacturer's instructions. The forward mutagenesis primer of the complementary pair was 5′-ACC CTC CCC CAG GCT GCG CCG GG-3′.
Plasmid DNA was isolated (Maxiprep or QIAprep Spin Miniprep kits; Qiagen Inc., Mississauga, ON, Canada), and open reading frames were fully sequenced to ensure their integrity.
Human nonpigmented ciliary epithelial (NPCE) and human trabecular meshwork (HTM) cell lines were cultured as described previously.29 Cells were transfected with plasmid DNA according to the reagent manufacturer's protocols (FuGENE 6 [Roche, Laval, QC, Canada]; or TransIT-LT1 [Mirus Bio LLC, Madison, WI]).
Small interfering RNA (siRNA) specific to the shared region of all isoforms of human PITX2 was custom ordered from Ambion (Austin, TX) (sense, 5′-CAG CCU GAA UAA CUU GAA Ctt-3′; antisense, 5′-GUU CAA GUU AUU CAG GCU Gtt-3′). siRNA specific to human SLC13A3 was custom ordered from Thermo Fisher Scientific/Dharmacon (Lafayette, CO) (sense, 5′-GGA GGA AGA AUA AAU CUG AUU-3′; antisense, 5′-UCA GAU UUA UUC UUC CUC CUU-3′). Nontargeting control siRNA (AM4611) was purchased from Ambion. siRNA transfection was performed as described previously29 according to the reagent manufacturer's protocol (Lipofectamine 2000; Invitrogen).
PITX2-HD-HR and PITX2-HR constructs were transfected into NPCE cells as 2 μg plasmid DNA + 6 μL transfection reagent for 5 × 105 cells/60-mm plate (or 4 μg DNA + 12 μL reagent for 106 cells/100-mm plate). Whole-cell protein extracts were derived, and immunoblotting was performed as described previously27,33 using antibody (anti-Xpress; Invitrogen).
Hormone induction of recombinant PITX2 in NPCE cells with subsequent RNA extraction and microarray analysis was performed as described previously.29 Statistically significant differences (P < 0.05) in the average normalized fluorescence intensity of each transcript between the hormone-inducible constructs and the empty vector samples were determined. Expression data from Affymetrix (Santa Clara, CA) microarrays were first log2-transformed and then analyzed with dCHIP software.34,35 Changes in transcript levels were compared among three matched sets of RNA from transfected (empty vector vs. PITX2-HD-HR or PITX2-HR) NPCE cells treated with 100 μg/mL cycloheximide and 10−8 M mifepristone. A change in expression was determined to have occurred when at least a 1.5-fold change in expression was detected along with P <0.05 and a present call in 2 of 3 comparisons using Affymetrix software (MAS5).
The in silico tissue expression profile of each gene was determined with an electronic search of the SOURCE database (http://genome-www5.stanford.edu/cgi-bin/source/sourceSearch). The PITX2 DNA binding motif search on the upstream region of genes was performed using RepeatMasker (http://www.repeatmasker.org), Possum (http://zlab.bu.edu/~mfrith/possum), and Ensembl (http://www.ensembl.org/index.html) Web sites and the bicoid binding site matrix (http://www.gene-regulation.com/cgi-bin/pub/databases/transfac/getTF.cgi?AC=M00140). The gene functions were determined using GoSurfer (http://bioinformatics.bioen.uiuc.edu/gosurfer).
Total RNA was extracted from siRNA-transfected HTM cells (TRIzol Reagent; Invitrogen). A standard 20 μL reverse transcription reaction contained the following components: 1 μg DNaseI-treated total RNA, 500 ng oligo d(T), 0.5 mM (each) dNTPs, 1× first-strand buffer, 0.01 M dithiothreitol, 40 U RNase inhibitor (RNaseOUT; Invitrogen), and 200 U reverse transcriptase (M-MLV; Invitrogen). Replicate reactions without reverse transcriptase were used as negative controls; 0.5 μL each reaction was used as a template for 10 μL qPCR reactions using a PCR kit (QuantiTect SYBR Green; Qiagen) and the following primer sets: PITX2 (forward, 5′-AGG CCA CTT TCC AGA GGA AC-3′; reverse, 5′-CGC TCC CTC TTT CTC CAT TT-3′; SLC13A3 (forward, 5′-GGA CAC TTG CTG GTC AAA GAC-3′; reverse, 5′-ATG GCT CAT GGA CTT GAG TGG-3′); HPRT1 (forward, 5′-GCC AGA CTT TGT TGG ATT TGA-3′; reverse, 5′-GGC TTT GTA TTT TGC TTT TCC AG-3′). Thermal cycling was performed on a real-time PCR system (7900HT Fast; Applied Biosystems, Carlsbad, CA) with the following conditions: 95°C, 15 minutes; 40 cycles of 94°C, 15 seconds; 60°C, 30 seconds; 72°C, 30 seconds. The relative quantity of gene transcripts (PITX2 siRNA vs. nontargeting siRNA samples), averaged from quadruplicate reactions, was calculated from the equation 2-ΔΔCt, with HPRT1 used as the endogenous reference.
Chromatin immunoprecipitation (ChIP) was performed using a ChIP kit (Active Motif ChIP-IT Express; Cedarlane Laboratories Ltd., Burlington, ON, Canada) according to the manufacturer's protocol. NPCE cells were grown to 70% to 80% confluence (~6 × 107 cells) on 100-mm plates. The ChIP reactions were set up by adding protein G magnetic beads, ChIP buffer 1, 6.3 μg sheared chromatin (~300–1000 bp), protease inhibitor cocktail (P8340; Sigma-Aldrich Canada Ltd., Oakville, ON, Canada), and 2 μg of anti-PITX2 antibody (H00005308-M01; Abnova Corporation, Taipei, Taiwan), normal rabbit IgG antibody (10500C; Invitrogen), or anti-acetyl-histone H3 antibody (Lys9; 9671; Cell Signaling Technology, Inc., Danvers, MA). One ChIP reaction did not contain any antibody. A 166-bp fragment in the SLC13A3 upstream region containing PITX2 binding site B was amplified with the primers 5′-TCC ACT TTC AGG CTC AGA GG-3′ and 5′-GCA GCT GGC TTA TCC TCT CC-3′. A 243-bp fragment containing PITX2 binding site A was amplified with the primers 5′-AGT GCC TAA CTT TCT CCA ACG-3′ and 5′-CGC TCC ACA CCT TCT TGG-3′. A 277-bp control amplicon from an upstream region lacking predicted PITX2 binding sites was amplified with the primers 5′-TTT GTC TTG CCT GAA TGC TG-3′ and 5′-TGC ACA CTC GTC TCC AAG TC-3′; this fragment was amplified only from genomic DNA and the ChIP input fraction but not from any immunopurified samples.
Transactivation assays were performed as described previously.27 Transfections were performed in triplicate in NPCE or HTM cells in 24-well plates (4 × 104 cells/15-mm well) using 160 ng empty (pcDNA4HisMax) or Xpress-PITX2C vector, 60 ng pGL3Basic-SLC13A3 luciferase reporter plasmid, 20 ng pCMVβ (Promega) transfection control, and 0.72 μL transfection reagent. Cells were harvested with 100 μL passive lysis buffer (Promega) 48 hours after transfection. Firefly luciferase activity was measured by luminometry (Turner Designs, Sunnyvale, CA) from 10 μL protein lysate mixed with 100 μL luciferase assay reagent (Promega) and was standardized to the β-galactosidase (internal control) activity quantitated by the β-galactosidase enzyme assay system (Promega) from 75 μL protein lysate.
Cell viability assays were performed with HTM cells (2 × 105 cells/35-mm well) using 2 μg/mL propidium iodide (P4170; Sigma-Aldrich) as described previously.36 Propidium iodide is excluded by viable cells but penetrates cell membranes of dying or dead cells.
Zebrafish (Danio rerio) were housed in system water on a 14-hour light/10-hour dark cycle; the embryos were acquired by natural spawning and maintained at 28.5°C. The developmental stage was determined by time (hours postfertilization [hpf] or days postfertilization) and morphologic criteria.37 All experiments were conducted in accordance with the guidelines established by the Institutional Animal Care and Use Committee at the Medical College of Wisconsin, adhering to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.
RNA was extracted from zebrafish embryos harvested in reagent (TRIzol; Invitrogen) using standard procedures. cDNA was synthesized from 1 μg total RNA with reverse transcriptase (SuperScript III; Invitrogen) and random hexamers. For RT-PCR analysis of expression, the following slc13a3 primers were used: forward, 5′-TGG AAA GAC AAG GGA AGG AA-3′; reverse, 5′-ATG GTT GGA AAG AGG CAC AC-3′ (254 bp, encompassing exons 1 and 2). For β-actin control, the following primers were used: forward, 5′-GAG AAG ATC TGG CAT CAC AC-3′; reverse, 5′-ATC AGG TAG TCT GTC AGG TC-3′ (329 bp). For in situ hybridization, a zebrafish slc13a3 cDNA fragment was PCR amplified using the following primers: forward, 5′-TGC AGC AGA AGC TAA AGC AA-3′; reverse, 5′-CTA AGC CGG ACT CCT CAC AG-3′ (495 bp, encompassing exons 7–11). The resultant fragment was cloned into vector (pCRII-TOPO; Invitrogen), and the insert sequence was verified by direct automated DNA sequencing. Digoxigenin-labeled RNA probes were synthesized by in vitro transcription, and in situ hybridization was performed as previously described.38
Two hormone-inducible constructs of PITX2 were used in our experiments. In one construct, PITX2-HR, the ligand binding domain (LBD) of the human progesterone receptor was inserted 40 amino acids C-terminal to the homeobox of Xpress-PITX2A (Fig. 1). Another construct, PITX2-HD-HR, replaced the forkhead domain of the FOXC1-HR construct29 with the PITX2 homeodomain (Fig. 1), thus altering the DNA-binding specificity of the modular protein; in this construct, the LBD replaces the inhibitory domain of the FOXC1 backbone. Both hormone-inducible constructs produced proteins of the appropriate molecular weight in transfected NPCE cells (Supplementary Data, http://www.iovs.org/lookup/suppl/doi:10.1167/iovs.10-6967/-/DCSupplemental) and, after administration of the progesterone analog mifepristone (RU-486), transactivated the luciferase reporter gene fourfold more than transfection with the empty vector (Supplementary Data, http://www.iovs.org/lookup/suppl/doi:10.1167/iovs.10-6967/-/DCSupplemental). The hormone-inducible expression system was previously used to identify FOXC1 target genes.29 Mifepristone activated the hormone-inducible constructs, allowing production of mRNA expressed directly in response to PITX2 transcriptional regulation. The addition of cycloheximide eliminated secondary induction events by preventing any new protein from being translated. For controls, NPCE cells were transfected with an empty expression vector and treated in an identical manner with cycloheximide and mifepristone. Total RNA samples collected from three independent experiments were subjected to microarray analysis. The gene expression profile of NPCE cells with induced PITX2 activation was directly compared with that of NPCE cells transfected with the empty vector. The genes differentially expressed by at least 1.5-fold were further analyzed. Analysis of the 45,038 probe set identified 448 genes differentially regulated in response to PITX2-HD-HR and 97 genes differentially regulated in response to PITX2-HR. Forty-seven genes overlapped between the data sets for the two hormone-inducible constructs of PITX2 (Supplementary Data, http://www.iovs.org/lookup/suppl/doi:10.1167/iovs.10-6967/-/DCSupplemental).
The 47 genes with altered expression in the presence of induced PITX2 in multiple microarray experiments were selected for further analysis. The genes were investigated through in silico analyses for expression within the eye and consensus PITX2 DNA-binding sequences (TAATCC, GGATTA, GGCTTAG or CTAAGCC) in the 5′ upstream region of the gene. Fifteen genes with known function, ocular expression, and PITX2 DNA binding sites in the promoter region (Table 1) were subjected to Northern blot analysis or semiquantitative RT-PCR (Supplementary Data, http://www.iovs.org/lookup/suppl/doi:10.1167/iovs.10-6967/-/DCSupplemental). Of these, we chose to further characterize SLC13A3, whose mRNA transcript was confirmed by semiquantitative RT-PCR to increase in response to expression of hormone-inducible PITX2A (data not shown). Real-time two-step RT-PCR using HTM cells that were transfected with siRNA showed that a reduction in PITX2 mRNA to 34% of control levels (all isoforms) led to a reduction of SLC13A3 transcripts (all isoforms) to 42% of control levels (Table 2). Therefore, PITX2 regulated the expression of SLC13A3 in both ocular cell lines examined.
The genomic sequence of SLC13A3 was investigated for putative PITX2 DNA-binding sites using the POSSUM Web site interface. Twelve binding motifs were found in the first 3000 bp upstream of the first exon of SLC13A3 transcript variant 1. The putative PITX2 DNA-binding site at position −26, site A (Fig. 2), is highly conserved in Pan troglodytes (chimpanzee), Macaca mulatta (rhesus monkey), and Canis familiaris (dog) promoter regions (data not shown). The putative PITX2 DNA binding motif at position −311, site B, is highly conserved only in P. troglodytes and M. mulatta promoter regions (data not shown). ChIP analysis of NPCE cells was performed to test the direct binding of PITX2 to the SLC13A3 upstream region. Upstream DNA sequences of SLC13A3 containing PITX2 binding site A and PITX2 binding site B were amplified by PCR in ChIP samples treated with PITX2 or acetyl-histone H3 (Lys9) antibodies (Fig. 2), indicating the fragments were recovered from PITX2-bound, transcriptionally active chromatin. A negative control region (~20 kb upstream, with no Possum-predicted PITX2 binding sites) was not amplified in any ChIP sample, attesting to the specificity of the antibodies (data not shown).
Activation of the SLC13A3 upstream region by PITX2 was tested by luciferase reporter assays in NPCE and HTM cells (Fig. 3). The longest construct, −403/+165, included the two putative PITX2 DNA-binding sites, A (5′-TTAAGCCG-3′) and B (5′-TTAATCCT-3′). The second construct, −163/+165, contained only binding site A, whereas −163/+165Δ was deleted for binding site A. Wild-type recombinant PITX2C caused approximately 7- to 12-fold activation of the reporters containing binding site A but was significantly reduced to approximately 4.5-fold when site A was deleted. Activation of the PITX2-responsive reporters was greatly diminished when disease-causing mutant constructs of PITX2C (T30P, V45L, and R52C) were used.
Recent data indicate that local oxidative stress is a determining factor in the pathology of glaucoma.13–15 We therefore investigated whether PITX2 and SLC13A3 were involved in ocular oxidative stress response pathways. We first tested for cytotoxic concentrations of hydrogen peroxide (200, 400, 600, and 800 μM H2O2) on HTM cells; 600 μM H2O2 had a significant effect on cell viability (Supplementary Data, http://www.iovs.org/lookup/suppl/doi:10.1167/iovs.10-6967/-/DCSupplemental). Next, we performed cell viability assays to test whether reduction of the PITX2 levels influenced the survival and response to oxidative stress. As shown in Figure 4, when HTM cells were transfected with 100 nM PITX2 siRNA and exposed to H2O2, the number of nonviable HTM cells was significantly increased by at least four times compared with cells treated with nontargeting control siRNA and H2O2. We also investigated whether the reduction of SLC13A3 levels influenced the survival and response to oxidative stress. Transfection of HTM cells with 100 nM SLC13A3 siRNA caused a 72% reduction in endogenous RNA levels and an 84% reduction in recombinant protein levels (Supplementary Data, http://www.iovs.org/lookup/suppl/doi:10.1167/iovs.10-6967/-/DCSupplemental). We observed a significant twofold reduction in cell viability in H2O2-treated HTM cells when they were transfected with SLC13A3 siRNA compared with nontargeting control siRNA (Fig. 4).
We characterized expression of the slc13a3 homolog in zebrafish (Fig. 5A) to gain a better understanding of its physiological role in relation to the expression of pitx2. RT-PCR analysis detected slc13a3 transcripts in zebrafish embryos starting from 12 hpf (Fig. 5B). Whole mount in situ hybridization detected enrichment for slc13a3 transcripts in the hatching gland, fins, pharyngeal arches, and periocular mesenchyme at 27 hpf and in the pharynx, kidney, gut, and anterior segment of the eye region in 120 hpf embryos (Fig. 5C). Expression in the developing kidney and gut appeared to be the strongest, particularly at later stages of embryonic development. However, localization of slc13a3 in anterior ocular regions coincides remarkably with pitx2 expression during the same developmental stages.38,39
Mutations in PITX2 cause ARS, a human autosomal dominantly inherited maldevelopment disorder of the eye associated with glaucoma. How defects in PITX2 disrupt the developing and mature eye leading to human disease remains unknown. PITX2 has key roles in the developing eye as Pitx2−/− knockout mice present with severe ocular defects, including agenesis of the cornea, abnormal blood vessels, dysgenic optic nerve, and coloboma.40 However, PITX2 expression is maintained in the adult eye, and recent experiments in the extraocular muscles of mice showed that Pitx2 is also important after birth.41 Loss of Pitx2 postnatally affected extraocular muscle function, making the extraocular muscles more fatigable than in wild-type mice. Furthermore, glaucoma usually develops in patients with PITX2 defects in their early 20s or 30s (early onset) rather than at birth (congenital). These results indicate that PITX2 is not only important during embryogenesis but is also important for the normal function of the adult eye. We have now identified PITX2 target genes in adult eye cells through the use of a hormone-inducible expression system to elucidate the gene regulatory pathways necessary for normal anterior segment function.
Target genes of PITX2 regulation in different cell types have been identified previously through microarray analyses. Although we did not detect any overlap between muscle-specific genes whose expression correlated with Pitx2 gene dosage42 and our data set, two genes from our list of 47 potential direct targets (i.e., PCDP1/MGC33657 and TLL2) were in common with those whose expression in HEK293 cells was altered because of recombinant PITX2 expression.43 Additional direct target genes of PITX2 regulation have previously been identified, including PLOD-1,44 Lef-1,45 FoxJ1,46 and Dkk2,47 but these were not detected in our microarray experiments. We have extensively characterized physically associated transcription factors in a complex pitx2-foxc1-foxc2-pawr regulatory network involved in zebrafish ocular development.48 The identification here of SLC13A3 as an ocular target gene of PITX2 provides us with an important tool to monitor the output of modifications to this regulatory network.
Some of the PITX2 target genes identified here (Table 1) have functions relevant to glaucoma pathology. For example, levels of pyruvate dehydrogenase phosphatase (PDP2) are decreased in response to the stresses of starvation and diabetes resulting in hyperphosphorylation and inactivation of the pyruvate dehydrogenase complex.49,50 Our results also show that SLC13A3 is a direct target of PITX2 regulation. SLC13A3 encodes a transmembrane protein, NaDC3, that acts as a Na+/dicarboxylate cotransporter. Mammalian NaDC3 is expressed in the liver, kidney,51 brain,52 eye, and optic nerve,30,53 and here we demonstrate the localization of zebrafish slc13a3 in similar tissues. Most notably, the pattern of its expression in the periocular mesenchyme and anterior ocular segment at different stages of development overlaps remarkably well with that of zebrafish pitx2.38,39 In situ hybridization of adult mouse eye showed that NaDC3 mRNA is present in anterior and posterior segments. A postulated role of NaDC3 in kidney cells is to transport GSH,31 one of the main defense mechanisms of the human body against oxidative stress. Aqueous humor of glaucoma patients presents with reduced total antioxidant potential54 and decreased plasma GSH levels.55 SLC13A3 is also hypothesized to contribute to the maintenance of visual function because of its role in transporting the neuronal metabolite N-acetylaspartic acid (NAA) across the plasma membrane in ocular tissues.30 Mutations in the gene encoding aspartoacylase, which hydrolyses intracellular NAA, cause Canavan disease, an autosomal recessive disorder associated with optic neuropathy, mental retardation, and brain degeneration.56 The potential roles of SLC13A3 in oxidative stress and in the optic neuropathy caused by Canavan disease indicate SLC13A3 as a good candidate for involvement in glaucoma.
We observed decreased SLC13A3 expression when endogenous PITX2 levels were reduced and increased SLC13A3 expression when recombinant PITX2 activity was increased. Molecular analyses indicated that PITX2 (only isoform C was tested) significantly activated transcription from reporter plasmids containing SLC13A3 upstream elements, whereas PITX2 mutants did not. We also detected binding of endogenous PITX2 to SLC13A3 promoter region elements through ChIP. SLC13A3 mRNA is expressed in the ciliary body, the site of aqueous humor production; therefore, the protein encoded by SLC13A3 might have important functions in the production of aqueous humor. Mutations or downregulation of SLC13A3 may result in increased IOP and glaucoma.
Our analyses indicate that PITX2 activates genes in the anterior segment of the eye involved in a variety of key processes, including response to cellular stress. Cellular stress can be induced by the generation of reactive oxygen species (ROS), and many experimental studies have shown that local oxidative stress induced by ROS is a determining factor in the pathology of glaucoma. In vitro studies have shown that TM cell morphology is altered on exposure to hydrogen peroxide by compromising cellular integrity and cell adhesion.14 In addition, aqueous humor drainage from the anterior chamber of the calf eye is affected when TM cells are perfused with hydrogen peroxide.15 Moreover, it has been shown in humans that TM cells from glaucoma patients presented more oxidative DNA damage than unaffected controls, and this damage is significantly correlated with the elevation of IOP and visual field damage.13,16 Our studies indicate that reduction in PITX2 protein levels produced an increase in cultured trabecular meshwork cell death when the siRNA-transfected cells were exposed to hydrogen peroxide. Moreover, a reduction in SLC13A3 also produced an increase in HTM cell death when the cells were exposed to hydrogen peroxide. These results support the hypothesis that PITX2 acts through SLC13A3 to modulate responses to cellular stress. We speculate that in healthy adults, PITX2 activates SLC13A3 whose encoded protein, NaDC3, transports GSH from the aqueous humor into trabecular meshwork cells. GSH would protect the cells against the accumulation of ROS, maintaining normal trabecular meshwork function with normal drainage of aqueous humor and normal IOP. In certain ARS patients we speculate that, in addition to causing the structural defects of anterior segment development, mutations of PITX2 also cause persistently reduced SLC13A3 expression resulting in decreased GSH transport into trabecular meshwork cells. As a consequence, a defense mechanism against ROS accumulation is compromised, and, when ROS generation exceeds the cell's antioxidant capacity, ROS cause cell dysfunction and cell death. The number of dead cells will be increased over time, resulting in compromised aqueous humor drainage, elevated IOP, and glaucoma. This biochemical model of glaucoma progression challenges the notion that ARS be classified solely as a secondary form of glaucoma.
In summary, we have identified PITX2 target genes in an NPCE cell line through the use of a hormone-inducible expression system and microarray analysis. This method has now been successfully used to identify genes that are direct targets of transcriptional regulation by either FOXC129 or PITX2. As a result we propose that this method can be generally applicable to isolate target genes of transcription factors, depending on the DNA-binding specificity of the construct. Importantly, we have shown that SLC13A3 is a direct target gene of PITX2 and that PITX2 is involved in ocular oxidative cellular stress responses through SLC13A3. We have also analyzed the expression of zebrafish slc13a3 and will be able to monitor its response due to modulation of levels of pitx2 and other members of its regulatory gene network.
The authors thank May Yu for tissue culture expertise.
Supported by the Canadian Institute for Health Research (MAW), Fonds de la Recherche en Santé du Québec (PB, VR), and National Institutes of Health Grant EY015518 (EVS).
Disclosure: M.H. Strungaru, None; T. Footz, None; Y. Liu, None; F.B. Berry, None; P. Belleau, None; E.V. Semina, None; V. Raymond, None; M.A. Walter, None