Most of the good and lasting things in my life began for me at Duke University in Durham, North Carolina. While at Duke, I met my wife, my two daughters were born, and I encountered my first patient with North Carolina macular dystrophy (NCMD). As a resident on a rotation at the Oteen Veterans Administration Hospital in Asheville, North Carolina, a patient presented to the clinic for routine follow-up. He had a large defect involving most of the maculae in both eyes but had surprisingly good vision (20/50).

Having worked with Margaret Pericak-Vance, PhD, in molecular genetics at Duke, the words of the geneticist William Bateson entered my mind: “treasure your exceptions.” It was clear that this patient’s disease was an exception. While taking a family history, the patient reminisced that “Ole Hamp Lefler examined me and all of my kin 20 years ago.” I quickly realized he was the proband of the original Lefler Wadsworth Sidbury syndrome family and I immediately knew the size of the pedigree and their potential importance in better understanding macular diseases (Figure 1).¹ During my Asheville VA rotation on nights and weekends, I would pack up an old handheld Kowa camera (which used film in those days), drops, and portable examination equipment. I would drive to pick up the proband, and together we would stop at various houses, farms, and so on to study as many family members as possible. By the end of the 3-month rotation, I had blood samples and examination data on 200 family members and genealogy data on more than 1,000.

I spent the first few years cleaning up the clinical data and refining the phenotype. This led to my first and only solo-author paper.² In that publication, I defined NCMD as an autosomal dominant, congenital, completely penetrant macular degeneration. Most importantly, I found the disease not to be progressive despite an earlier paper calling it a “dominant progressive foveal dystrophy.”³ Because of the huge phenotypic variability, there are many names given to this one disease in this one family. Some of the names (none of which are good) are dominant macular degeneration and aminoaciduria, dominant progressive foveal dystrophy, central areolar pigment epithelial dystrophy, central pigment epithelial and choroidal degeneration, Caldera maculopathy, and NCMD. The many and diverse names have contributed to confusion surrounding NCMD. In his textbook Stereoscopic Atlas of Macular Diseases: Diagnosis and Treatment, J. Donald M. Gass, MD, named NCMD for the founder effect of this single family.⁴ It is also a suboptimal name, reflecting Dr. Gass’s frustration in devising a nomenclature that truly incorporated the entire phenotype.

The phenotype is highly variable, which can cause confusion when examining only 1 or 2 family members. I defined 3 grades of severity. Earlier researchers had called them “stages,” implying that there was a progression from one to the next. This was not the case, however. The grade does not change at any time during a patient’s life. Grade 1 is merely small drusen in the fovea and the patients are asymptomatic. Grade 2 is confluent drusen (Figure 2). Grade 3 is a large excavated lesion in the central macula typically with sharp shelving edges with fibrosis (Figure 3). The center of fixation is along the nasal edge of the lesion. Choroidal neovascularizations can occur, and only in these cases do patients experience progressive vision loss. The surrounding fibrosis, I believe, is due to choroidal neovascular membranes (CNVMs). However, because they usually grow along the temporal edge, vision is not affected. The OCTs of the lesions are quite dramatic. Many patients with these lesions were misdiagnosed with toxoplasmosis. Because of the huge phenotypic variability, the phenocopies of NCMD are many and include toxoplasmosis, drusen of all causes including AMD, Best disease, and CNVMs of all causes.

My initial goal in studying the original NCMD family was to map the disease-causing gene by using linkage analysis. When I began, restriction fragment length polymorphisms (RFLPs) were the state-of-the-art genetic markers. This was labor intensive, and at top speed in Drs. Alan Rose and Pericak-Vance’s lab at Duke, I could do 4 Southern blots with 2 to 3 markers a week. After slogging through RFLPs for a year, I learned of a new polymerase chain reaction (PCR)-based genetic marker (“CA repeats,” microsatellites) developed by James Weber, PhD, founder of PreventionGenetics in Marshfield, Wisconsin. He allowed me to come into his lab at the Marshfield Clinic to utilize his markers. I was the first person in ophthalmology to use them. Within 2 weeks in his lab, I performed this procedure with more than 200 markers. I brought my gels back to Duke for Dr. Pericak-Vance’s group to analyze. This took time because the new highly informative markers did not lend themselves to binary coding and analyses.

After several months of analysis, I still did not have a positive logarithm of odds score. By then, I had a position at Medical University of South Carolina in Charleston. My clinical practice was slow, and I had an NIH grant on NCMD to maintain. Also, I had almost excluded the entire genome according to an analysis program called Exclude. I began thinking about non-Mendelian genetic issues that might cause me to miss linkage. I looked for evidence of genetic anticipation suggesting trinucleotide repeat expansions and found none. I eventually did find evidence of a type of segregation disorder known as “parent of origin.” I went back to Dr. Weber’s lab for another 2 weeks and ran another 200 markers. When doing a quick hand analysis of the data, it was apparent I had finally gotten linkage to chromosome 6q16. Dr. Pericak-Vance’s more formal analysis later confirmed this. The Human Genome Organization (HUGO) Gene Nomenclature Committee (HGNC) named it MCDR1 (MC = macula, D = dystrophy, R = retina, 1 = first mapped in the human genome) and I found myself on the Human Chromosome 6 committee along with now NIH director Francis S. Collins, MD.

I took a position at the University of Florida, where I established my lab and had technicians for the first time; I was later recruited to the University of California-Los Angeles. I continued to learn of new families with NCMD and began to use a classical positional cloning strategy with the newest techniques coming out of the Human Genome Project. Jeffrey M. Vance, MD, PhD, researcher, molecular geneticist, and clinician (also spouse and colleague of Dr. Pericak-Vance), had previously warned me to stick with gene mapping and linkage analysis and not get into positional cloning because “it will kill you.” This proved almost to be prophetic. Despite this warning, my lab became a miniature version of the The Human Genome Project, focused only on an 880-kilobase region, using yeast artificial chromosomes (YACs), bacterial artificial chromosomes (BACs), P1 artificial chromosomes (PACs), and even the Fugu cDNA library.

By 2001, my lab had sequenced all known expressed sequences (genes) and some of the promoters, including an unnamed zinc finger. Because zinc finger genes are typically involved in development and this one was expressed in the embryonic retina, it was a particularly attractive candidate. However, our sequencing of the exons and promoter found nothing significant. By then I knew the mutations causing MCDR1 would be strange and unusual. I also knew that studies like exome sequencing would be worthless. Shortly thereafter, despite methodical progress, I lost my NIH grant to a competitor who claimed to have extensive families with NCMD. That study ultimately did not further the research on finding the mutations.

I went into solo private practice and helped a research colleague secure a grant on NCMD that we would work on together. From the beginning, I knew we needed to do whole genomic sequencing of my targeted region. My research partner and others wanted to perform exome sequencing only, which I knew would be fruitless. This project, and offers of assistance from other researchers, also did not pan out. The idea of racing alone against some rather intimidating national and international labs was daunting. But although NCMD has been a career-long commitment and passion of mine, for most of my competitors, it was not much more than a curious side project. And feeling alone was no longer strange to me; in studying NCMD, an “orphan disease,” I often felt like I was the orphan. However, I knew I could still find the mutations.

I stayed engaged in the genomics research field by consistently attending the American Society of Human Genetics and The Association for Research in Vision and Ophthalmology (ARVO) meetings. Eventually, a new, more efficient, and cheaper sequencing technology became commercially available, called next-generation sequencing (NGS). I provided the precise interval I wanted sequenced based on my genetic crossover data (880 kilobases total) to a sequencing company along with genomic DNA from 6 of my NCMD families (2 different haplotpyes). I also provided the company with my American Express Plum Card number.

Two months later, I received my raw data and the fun began. The volume of data generated was huge. Sorting through this was tedious and time consuming. For instance, 1 of the 6 affected subjects had more than 2,100 variants to sort through to decide which was a possible disease-causing mutation and which was a normal variant. My approach to “bioinformatics” was mostly an Excel spreadsheet. Initially, I still had not found a variant that segregated. Frustrated, I thought about genetic variants/mutations that could be missed by NGS. One possibility could be copy number variants. I found another genomics company to outsource these evaluations to; they used a process of hybridizing 60-mer oligonulceotides. This study also came up empty handed (for me, anyway, considering the heavy use of my American Express Plum Card).

After these failures, I began to doubt myself and my data. Were my genomic intervals correct? Were my crossovers correct? I decided to do whole-genome sequencing, not just exon sequencing, to begin searching for the mutations outside of my intervals. As expected, the amount of data this generated dwarfed the amount of data I had from my targeted sequencing. I calculated that at the rate I was going through the sequencing data, I would find the mutation by the time I was 125 years old.

Then, at ARVO in 2015, Ed Stone, MD, PhD, graciously offered his assistance, which included the resources of the Stephen Wynn Eye Research Institute. He has superb bioinformatics programs and personnel, including Adam Deluca, PhD. Similar discussions and offers from other groups and institutions had not worked out. But I knew that if Dr. Stone committed his resources to this problem, there would be significant progress. I needed this, professionally and personally. I returned home from ARVO and sent Dr. Stone my data for the 6 subjects representing 2 different haplotypes.

Three weeks later, he called me to say he had found something interesting. I sent him another 50 DNA samples to see if the variants segregated. Another 3 weeks later, I received a call with even better news – the data were looking good. I then sent him 250 DNA samples, which were as many samples of the 35 NCMD families I had collected over the past 2 decades. We selected the 11 best families of my data set for our first publication, which was featured on the cover of Ophthalmology.⁵

Richard G. Weleber, MD, FACMG, wrote an editorial in Ophthalmology about our study, calling it “one of the most important studies in our field in the past several decades.”⁶ What made our findings most significant was there are 3 point mutations all within 100 base pairs in a deoxyribonuclease I (DNase I) hypersensitivity binding site, which acts as a regulator of the retinal transcription factor positive regulatory domain containing 13 (PRDM13), 12 kilobases away. These point mutations are therefore in a noncoding region, which added to the complexity of finding them. Additionally, one of the NCMD families (from Belize) had a large duplication involving PRDM13 (Figure 4). This, along with the induced pluripotent stem cell (iPSC) and reverse transcription polymerase chain reaction (rtPCR) work from Dr. Stone’s lab, showed the gene’s importance in the embryogenesis of the retina, and specifically the macula. PRDM13 is the zinc finger I had studied 15 years earlier.

It appears that overexpression of PRDM13 is the cause of the NCMD phenotype. This is a new pathway to explore in macular diseases, and it appears that PRDM13 is important in the embryogenesis of the macula. PRDM13 is expressed in many lower animals (vertebrate and invertebrate) without maculae. This is a new gene and new pathway to study in macular disease. Understanding and manipulating this gene could facilitate controlling or growing new maculae.

Several lessons learned while studying this disease may be useful to those on a similar path:

1. Be a “lumper,” not a “splitter.” Much of the early confusion about this disease was because of the tendency of clinicians to be splitters. Without being too cynical, splitting generally results in publication opportunities. But once one sees multiple family members and appreciates the huge phenotypic variability within a single family, lumping unequivocably becomes the correct approach. This variability is observed with a single gene and a single mutation. The splitting, dividing, and subdividing of drusen in studying AMD, for example, do not seem to alter clinical outcome or prognosis.
2. Challenge the literature. If it does not make sense, there is likely an opportunity to correct it. This is a specialty of intelligent physicians; be creative and challenge the status quo.
3. Make friends and collaborate. I have met many fascinating clinicians, researchers, and patients. While there are too many collaborators to mention here, one very rewarding experience was a field trip through the back country of Belize with Maurice Rabb, MD, who passed away in 2005. Our success on NCMD was in large part due to the many national and international collaborations I forged. In addition to facilitating the research, collaborating also makes the work much more fun. Some of my darkest moments in this research were during times I had no collaborators available.
4. Persevere. The English author Samuel Johnson once said, “Great works are performed not by strength, but perseverance.” Whether my work represents perserverance, passion, or hardheadedness remains unclear. Studying NCMD became much more than a research project; for me it became, and still is, personal. But perhaps that commitment was critical to success.

I would not have pursued this research if not for the intellectual curiosity inspired by my mentors at Duke. The study now includes 40 families worldwide with the NCMD phenotype. Many of these have been ascertained through national and international collaborations. Our success in finding these mutations would not have been possible without having multiple large families from different ethnicities. It would have been even more difficult proving the single point mutation in the DNase I hypersensitivity binding site in the single North Carolina family as the causative mutation. We continue to study the disease, looking for more mutations and mechanisms. RP

REFERENCES

1. Lefler WH, Wadsworth JAC, Sidbury JB Jr. Hereditary macular dystrophy and amino-aciduria. Am J Ophthalmol. 1971;71(Suppl.):224-230.
2. Small KW. North Carolina macular dystrophy, revisited. Ophthalmology. 1989;96(12):1747-1754.
3. Frank HR, Landers MB III, Williams RJ, Sidbury JB. Dominant progressive foveal dystrophy. Am J Ophthalmol. 1974;78:903-916.
4. Gass JDM. Stereoscopic Atlas of Macular Diseases: Diagnosis and Treatment, 4th ed. St. Louis, MO: Mosby; 1997.
5. Small KW, DeLuca AP, Whitmore SS, et al. North Carolina macular dystrophy is caused by dysregulation of the retinal transcription factor PRDM13. Ophthalmology. 2016 Jan;123(1):9-18.
6. Weleber RG. Dysregulation of retinal transcription factor PRDM13 and North Carolina macular dystrophy. Ophthalmology. 2016;123(1):2-4.