DNA samples obtained from new patients with ARS and anomaly were screened for PITX2 gene mutations in exons and at least 100-bp into adjacent intron regions as previously described 9. DNA was isolated from blood spots using the QIAGEN QIamp^® DNA mini kit following the dried blood spot protocol. PCR was conducted in a GeneAmp^® PCR system 9700 in 30 μl reactions containing 1.5 mmol/l Mg 2+, 40–100 ng DNA, Biolase Reaction Buffer (Bioline), 0.25 mM each dNTP, 1.5 units Biolase DNA polymerase (Bioline) and 0.2 μM of each oligodeoxynucleotide primer. Cycling profile was one cycle of 94°C for five minutes, 30 cycles of 94°C for 45 seconds, 56°C for 45 seconds, and 72°C for 45 seconds, and one cycle of 72°C for ten minutes. Products were either purified using Millipore MultiScreen^® Separations System or sequenced directly. Products were sequenced bidirectionally with an ABI Prism^® 3700 DNA sequencer using ABI Prism^® BigDye™ Terminator Cycle Sequencing Ready Reaction Kits. The 20 μl reactions contained 5% DMSO, 1 μl of PCR product DNA, and 0.32 μM oligodeoxynucleotide primer. Cycling profile was conducted in a GeneAmp^® PCR System 9700. Cycling profile was one cycle of 96°C for 30 seconds, 35 cycles of 96°C for ten seconds, 50°C for 5 seconds and 60°C for four minutes. Extension products were purified using Agencourt CleanSEQ^® Reaction Clean up. PCR products were then sent to Agencourt, Inc to be sequenced. Sequences were manually examined for mutations and confirmed by additional independent amplification and sequencing.

A primer positioned at g.16944 of the PITX2 genomic sequence (GenBank accession number AF238048) and a downstream primer at g.17393 were used to PCR-amplify a 473 bp fragment that included part of exon 1b and the downstream intron using human genomic DNA as a PCR template; EcoRI and BamHI sites were appended to the 5' and 3' ends, respectively. Primers are listed in Table 1. A 653 bp fragment containing exon 4 and surrounding intron was amplified using an upstream primer at position g.17812 (with a 5' BamHI site appended) and a downstream primer at g.18439 containing a 3' KpnI site. A 713 bp fragment containing a portion of exon 5 and the upstream intron was generated using an upstream primer positioned at g.20001 and a downstream primer at g.21147; 5' KpnI and 3' HindIII sites were added to primers. Cycling profile was one cycle of 94°C for 45 s, 30 cycles at 94°C for 45 s, 60°C (55°C for exon 1b) for 45 s, and 72°C for 45 s, and one cycle of 72°C for 10 min. Fragments were inserted individually into pGEM-3Z (Promega) using the same restriction sites, and plasmids were sequenced using an ABI PRISM BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems). Using the appended restriction sites, exon/intron fragments were excised from pGEM-3Z and shuttled sequentially into the same sites of pcDNA3.1/myc-His(-)A (Invitrogen) to create the PITX2c minigene (Figure 2B). Mutant minigenes were constructed by overlap PCR using the above pGEM-3Z constructs as template 33. Sense and antisense primers containing the desired mutations were used with the appropriate outside primer for PCR to synthesize half-substrates. Mutant exon fragments were inserted into pGEM-3Z and mutations were verified by DNA sequencing. The mutant fragments were then used to replace the WT fragment in the minigene construct as described above. N-terminal FLAG-epitope tagged minigenes were constructed by excising the entire minigene from the pcDNA constructs using SalI and a partial EcoRI digestion, and shuttling the fragments into the same sites of p3X-FLAG CMV 7.1 (Sigma).

HEK293 and HeLa cells were grown in minimal essential growth medium supplemented with 10% fetal calf serum. Human cornea stromal cells were grown in Dulbecco's modified Eagle medium supplemented with 10% fetal calf serum. Cells were grown to about 60% confluence in 6 cm dishes (HEK293 and HeLa) or 6 well plates (cornea stromal cells) and transfected with 2 μg of minigene DNA and 0.5 μg GFP using the calcium phosphate method (Amersham Biosciences) for HEK293 and HeLa cells, or Lipofectamine (Invitrogen) for cornea stromal cells. Total RNA was harvested 48 h later using the Qiagen RNeasy kit. For RT-PCR, reverse transcription was performed on 1 μg of total RNA using oligo dT in a volume of 25 μl. For PCR, 2.5 μl of the reverse transcription mixture was used in a 25 μl PCR reaction using the T7 primer to pcDNA3.1/myc-His(-)A vector and to exon 5 (for minigene mRNA) or to exon 1b and exon 5 (for endogenous plus minigene mRNA), and primers to GFP or cellular GAPDH. Cycling conditions were one cycle of 94°C for 45 seconds, 30 cycles of 94°C for 45 seconds, 60°C for 45 seconds, and 72°C for 45 seconds, and one cycle of 72°C for 45 seconds. To evaluate the accuracy of splicing of WT and mutant mRNA, RT-PCR products were cloned and sequenced. RT-PCR products were blunted with Klenow fragment and then digested with EcoRI, which cuts at a site located upstream of exon 1b. The products were inserted into the EcoRI and SmaI sites of pGEM-3Z and individual clones were sequenced. To assess minigene splicing by RNase protection assay, a 700 bp riboprobe was made by in vitro transcription using T7 polymerase and HindIII-linearized pGEM-3Z containing the exon/intron 4 fragment. This includes 264 nt and 183 nts of upstream and downstream intron, respectively, that flank the 206 nt exon 4. A 371 nt GFP riboprobe was made using SP6 polymerase and EcoRI-linearized pGEM-3Z containing a GFP fragment. RNase protection assays were carried out as described previously 34 using 2.5 μg RNA. Correct splicing generates a ~206 nt protected band and intron retention yields a ~386 nt band. The GFP probe generates a ~309 nt protected band. Results were visualized by autoradiography and quantitated with a Storm 820 PhosphorImager (Amersham Biosciences).

HEK 293 cells were transfected with 2 μg p3X-FLAG minigene DNA and 1 μg pEGFP as described. Cells were washed with phosphate buffered saline, harvested by scraping, the cells were resuspended in water, an equal volume of 2× solubilizing buffer (2% SDS, 10% β-mercaptoethanol, 5% glycerol, 50 μM Tris, pH 7, 0.005% bromphenol blue) was added, and samples were sonicated for five seconds. Samples were subjected to 10% SDS-PAGE and transferred to nitrocellulose membrane by electrotransfer. Immunoblotting was done as described 33, blots were blocked with nonfat dry milk, incubated with anti-FLAG M2 monoclonal primary antibody (Sigma) or anti-GFP monoclonal primary antibody (Covance), and after washing, goat anti-mouse IgMμ HRP-conjugate secondary antibody (Upstate) was applied. Blots were developed using Supersignal West Pico Chemiluminescent Substrate (Pierce). Results were visualized by autoradiography or images were obtained using a Fluorchem IS-8800 (Alpha Innotech).

DNA samples obtained from new patients with Axenfeld-Rieger syndrome and anomaly were screened for mutations in exons and at least 100-bp in the adjacent intron regions. Two new families with intronic PITX2 mutations have been identified. The first mutation was identified in a patient (0175) affected with ARS (Figure 1A). The patient displayed all classic features of this condition and some additional defects: Rieger anomaly of the eye, wide-spaced eyes, thin upper lip, hypoplastic maxilla, dental defects, redundant periumbilical skin, and an anteriorly placed anus. The mutation identified in this patient is a G>T change in the -1 position of the 3' ss associated with exon 4 (IVS4-1G>T) (g.18072G>T as seen in PITX2 sequence GenBank # AF238048; Figures 1A and 2C). Both parents display a normal phenotype; testing of the patient's mother confirmed normal PITX2 sequence (data not shown) and the father was not available for testing. This mutation was reported previously in an ARS patient with severe ocular malformations, underdeveloped maxilla and redundant umbilical skin 25. Perveen et al. 18 reported a G>C mutation at the same position and affected individuals from this family demonstrated severe ARS phenotypes: severe iris hypoplasia emulating aniridia in one patient and joint hypermobility and an anteriorly placed anus in another individual were reported.

The second mutation was found in patient 689, who displayed Axenfeld-Rieger anomaly in one eye only and no craniofacial, dental or umbilical abnormalities. The patient had normal ocular pressure and vision, the right eye had a posterior embryotoxon visible on slit lamp exam with iris sheets sweeping up to the embryotoxon inferiorly, and no iris hypoplasia was noted. Slit lamp exam of the left eye demonstrated normal development. Clinical family history of this patient contains records of developmental ocular defects but dental or umbilical anomalies were never noted (Figure 1B). The mutation was identified as a G>C substitution at the +5 position of the 5' ss of PITX2 exon 4 (IVS4+5G>C)(g.18283G>C as seen in PITX2 sequence GenBank # AF238048; Figures 1B and 2C). The same change was previously reported in a patient affected with the complete ARS 9. Collection and examination of clinical information from this family indicated a substantial family history of complete ARS (pedigree and clinical diagnoses are shown in Figure 1C).

Most Rieger syndrome patients have point mutations in the coding region of PITX2 9 and most of these affect the DNA binding homeodomain 35. Previous work and the data above brings to eight the number of families with mutations that do not affect the protein coding potential of the gene, and the number of different mutations is now five 917182125. We set out to examine three recurrent intronic mutations, each of which was described in two unrelated families of Axenfeld-Rieger spectrum, and to correlate the phenotypes with potential molecular defects in pre-mRNA splicing. The IVS4-1G>T and IVS4+5G>C mutations were described above. A third mutation studied was an A>G change 11 nt upstream of the 3'ss of exon 5 (IVS5-11A>G) that was originally described in family 4 by Semina et al. 9 and later by Borges et al. 21 who reported the same change in another family (family 5); both families displayed classic features of ARS.

The PITX2 gene is located at 4q25 and numerous isoforms are produced through the use of alternative promoters and extensive alternative splicing of the six exons (Figure 2A) 28. PITX2 is expressed developmentally and in some adult tissues but not blood (EVS, unpublished) and thus patient samples to analyze splicing directly are not readily available. To overcome this limitation, a minigene for expression in tissue culture cells was designed that included the exons that comprise one of the common PITX2 isoforms, PITX2c, (Figure 2B) 36. Since the signals required for constitutive splicing are usually within ~100 nucleotides of splice junctions 3237, the introns were truncated to facilitate DNA cloning. Because there is no evidence that the extensive 5' and 3' untranslated regions are needed for constitutive splicing, exons 1b and 5 were also truncated; exon 1b starts 40 nt upstream of the translation start codon and the exon 5 fragment ends with the last amino acid codon. Minigene splicing was analyzed in HeLa and HEK293 cells since they have been used extensively to study constitutive RNA splicing. Additionally, cornea stromal cells (generously provided by Dr. Watsky, University of Tennessee) 38 were used since PITX2 is expressed during corneal development and this cell line endogenously expresses PITX2, and thus may more faithfully reproduce PITX2 splicing that occurs in patients.

HeLa and HEK293 cells were mock transfected or transfected with 2 μg of minigene DNA and/or a GFP-expressing transfection control plasmid and RT-PCR was performed on RNA isolated 40–48 h later. While PITX2c is not expressed in these cell types (data not shown), to ensure that only minigene RNA was detected, an upstream primer complementary to vector sequence was used in combination with a primer specific to PITX2 exon 5. Correct splicing would result in a 585 bp RT-PCR product whereas unspliced RNA would produce a 1564 bp product. The 585 bp band was observed only from transfected HEK293 cells and analysis of GFP expression demonstrated equivalent transfection efficiency and gel loading (Figure 3A, lanes 2 and 3). Similar results were obtained with HeLa cell RNA (data not shown). These results suggest that the HeLa and HEK293 cells faithfully express correctly spliced minigene RNA. As an additional test to verify the utility of the minigene, human cornea stromal cells were similarly examined. The primer pair specific for the minigene resulted in the 585 bp spliced product only in transfected cells (Figure 3B, lanes 2 and 3, top) whereas a primer pair that does not distinguish between transfected and endogenous PITX2c mRNA produced a 404 bp product that was present in both but more abundant in transfected cells, consistent with correct splicing of endogenous and transfected PITX2 mRNA in corneal cells (lanes 2 and 3, middle). Analysis of cellular GAPDH showed equal sample loading (Figure 3B, bottom).

To further assess the accuracy of minigene mRNA splicing in HEK293 and human cornea stromal cells, the RT-PCR products were cloned and three and four independent clones were sequenced, respectively. Each sequence showed correct minigene splicing (Figure 3C). These data indicate that the PITX2c minigene was efficiently and accurately spliced and therefore validated the system for determining the effects of patient mutations on RNA splicing. This system would not address potential tissue-specific splicing defects that may be important for ARS.

Patient 0175 reported here and patient 2 previously described by Lines et al. 25 display severe features of ARS (see above) and harbor a G>T mutation in the -1 position of the 3' ss of exon 4 mutation (IVS4-1G>T) (Figures 1A and 2C). This changes the canonical "AG" acceptor splice site to an "AT". In addition, a G>C substitution at the same site was previously reported in a patient that displayed classic ARS 1825. To test the effect of the G>T mutation on splicing, the three cell lines were transfected with WT or IVS4-1G>T minigenes and the splicing status of the mRNA was analyzed by RT-PCR. A single RT-PCR product of the appropriate size for correctly spliced RNA was observed with HEK293 (Figure 4A, top, lane 3), cornea stromal (bottom, lane 7), and HeLa cell RNA (not shown). A GFP-expressing plasmid used as a transfection and RNA loading control showed equal loading in each case (Figure 4A). To examine the possibility that a subtle change in splicing occurred that was not obvious from the size of the RT-PCR product, the PCR products were cloned and 13 HEK293 and three cornea cell independent clones were sequenced. All sequences showed that splicing was shifted 2 nt downstream to the next available "AG" dinucleotide (Figure 4B). Based on the results with the minigene in cultured cells, it is likely that all PITX2 mRNA from this allele is incorrectly spliced in these patients.

The G>C mutation at the +5 position of the 5'ss of exon 4, IVS4+5G>C, was originally described in a family with ARS 9 and above we described the same mutation in a patient 689 with an isolated Axenfeld-Rieger anomaly. To assess the effect of this mutation on splicing, HeLa, HEK293, and human cornea stromal cells were transfected with wild type (WT) or mutant minigenes and GFP, and the isolated RNA was analyzed by RT-PCR. As shown in Figure 5A for HEK293 and cornea stromal cells, a band of the size expected for correctly spliced RNA was again observed from the WT minigene (lanes 2 and 5) but two RT-PCR products were seen with IVS4+5G>C (lanes 3 and 6). The lower band size was consistent with correctly spliced minigene RNA whereas the second product was larger but not the expected size for completely unspliced RNA. Analysis of GFP showed equal loading for each sample (Figure 5A). Similar results were seen with HeLa cell RNA (data not shown).

To determine the nature of the aberrant PCR product observed from the IVS4+5G>C minigene, both RT-PCR products from HEK293 RNA were cloned and independent clones of six large products and eleven 585 bp products were sequenced. The sequences showed that the lower band of IVS4+5G>C RNA was derived from correctly spliced mRNA and that the upper band retained the intron between exons 4 and 5 (Figure 5B).

To accurately quantitate the frequency of intron retention in IVS4+5G>C RNA, an RNase protection assay (RPA) was performed using a 700-nt riboprobe that spanned the 206 nt exon 4 and included upstream and downstream intron sequences. Correctly spliced RNA results in a ~206 nt protected band while retention of the intron would yield a 386 nt band (Figure 5C). RNA from the WT minigene was virtually completely spliced (Figure 5D, lane 4) whereas approximately 71% of the RNA from IVS4+5G>C retained the second intron (Figure 5D, lane 5). Thus, mutation IVS4+5G>C causes significant disruption in PITX2c minigene splicing.

The previously reported patients from family 4 9 and family 5 21 have classic ARS and harbor an A>G mutation 11 nt upstream of the 3' ss associated with PITX2 exon 5 (mutation IVS5-11A>G) (g.20745A>G as seen in PITX2 sequence GenBank # AF238048; Figure 2C). To determine if this mutation affects splicing, HeLa, HEK293 and human cornea stromal cells were transfected with the WT or IVS5-11A>G minigenes and the isolated RNA was analyzed by RT-PCR. The RT-PCR product from IVS5-11A>G RNA was the same size as correctly spliced PITX2c mRNA (Figure 4A, lanes 4 and 8). PCR of GFP served as a loading control. To determine if the mutation caused an aberrant splice that did not alter the RT-PCR product size appreciably, the RT-PCR product was cloned and 11 independent HEK293 clones were sequenced. All of the sequences showed that splicing was shifted exclusively to the newly created "AG" dinucleotide acceptor site 11 nt upstream of the authentic 3' ss (Figure 4C). Based on these results, it is likely that PITX2 mRNA from this allele is always incorrectly spliced in these patients.

Aberrantly spliced mRNA derived from the mutant patient PITX2 genes is expected to result in truncated PITX2c protein since frameshifts occur in each case (Figure 6A). PITX2 protein levels were determined in transfected cells by western blot (Figure 6B) after normalizing protein levels to GFP expression and RNA expression as determined by RPA. RPA analysis demonstrated that mRNA expression was roughly similar for each minigene (Figure 6C) and that, except for IVS4+5G>C, each minigene mRNA was doubly spliced (although aberrantly).

The WT minigene produced a protein of the expected size, ~36 kDa (Figure 6B, lane 1). The IVS4-1G>T minigene produced a much smaller protein (~12 kDa, lane 4), but this was expected since this protein would terminate at a single frame-shifted amino acid after PITX2 position 70. It appears that the IVS4-1G>T protein is unstable since it was present at only ~5% of WT levels, despite its mRNA being at least as abundant as WT. The IVS4+5G>C mRNA generated a normal sized protein and a second product of the size (~29 kDa) appropriate for truncated protein derived from intron-containing mRNA (lane 2); when corrected for mRNA levels, the truncated protein was about one tenth the abundance of normal protein. It was also evident that the level of normal protein expressed from the IVS4+5G>C minigene was roughly proportional to correctly spliced mRNA levels, which was about 30% of that produced from the WT minigene. Conversely, the level of truncated protein generated from the intron-retaining mRNA (~71% of minigene RNA produced) was considerably less than expected if protein expression was proportional to mRNA. These data are consistent with inefficient export of the intron-containing IVS4+5G>C mRNA from the nucleus. However, the possibility that the mRNA is exported but not translated, or exported but produces an unstable protein, cannot be excluded. Finally, the IVS5-11A>G mRNA would direct synthesis of a protein containing 138 residues of PITX2 followed by 117 frame-shifted amino acids and again, an expected ~33 kDa protein was expressed (lane 3) at a level similar to WT.