The identification of novel disease genes sometimes overshadows another crucial form of genomic discovery: expanding the phenotype associated with known disease genes. This is especially important in the era of pervasive clinical genetic testing.
Exome sequencing, which interrogates all ~20,000 protein-coding genes simultaneously, is rapidly becoming a frontline diagnostic test for patients with rare genetic conditions. The obvious advantage of exome sequencing is its comprehensiveness: the coding regions of virtually all genes are sequenced. From a clinician’s point of view, I imagine that has tremendous appeal. Yet the cost of this comprehensive approach is that it uncovers thousands of protein-coding variants in the patient.
Automated filtering can reduce the list to ~100-300 variants that are somewhat rare and predicted to affect a gene associated with at least one of the patient’s phenotypes. That’s still far too many to select for full ACMG assessment. Thus, in our laboratory and many others, the decision of whether or not to assess a variant gives considerable weight to the overlap between the patient’s clinical features and the phenotype set of a genetic disorder. The accuracy of that process depends on:
- Thorough and precise phenotyping of the patient by expert clinicians
- The expertise of clinical directors and variant scientists
- Current knowledge of genotype-phenotype relationships for disease genes
Dependency #3 is the one that’s most likely to change over time. This is supported by recently published studies of clinical WES reanalysis, which have consistently found that new knowledge (i.e. a newly identified disease gene) is the most common source of positive findings among previously negative WES cases.
A Patient with Severe Muscular Atrophy
The rare disease genomics study at our institution has been running for about four years. One of the first cases we enrolled was a child with arthrogryposis multiplex congenita, i.e. joint contractures affecting multiple parts of the body. This condition is believed to be the result of reduced intrauterine movement. In the case of our patient, the reason for that reduced movement was a striking lack of skeletal muscle.
We enrolled the patient and his parents on our research protocol and performed whole-genome sequencing. This uncovered a de novo variant in a gene called BICD cargo adapter 2 (BICD2). Back in 2013, three studies of large family kindreds with dominant muscular atrophy affecting predominantly the lower limbs had linked the disorder to missense variants in BICD2. The Online Mendelian Inheritance in Man (OMIM) database, the definitive resource for gene-disease associations, described the condition as spinal muscular atrophy, lower extremity dominant, type 2 (SMALED2).
A dominant form of muscular atrophy could in theory describe our patient, but there were two problems:
- All reported disease-causing variants to date were missense changes, whereas our patient had an inframe deletion of a single amino acid.
- Our patient had severe atrophy throughout the body at birth, whereas the patients with SMALED2 generally presented later in life and usually with only lower limbs affected.
As a result of the inconsistencies, we chased other leads in this case, none of which panned out. BICD2 remained our top candidate, but not everyone was convinced that it was the answer. Myself included.
Serendipitous Meeting at ASHG
In October 2017, I went to the annual meeting of the American Society of Human Genetics (ASHG). On the way to the meeting, I perused the program through the mobile app and searched the submitted abstracts for some of my topics of interest. One of those was the BICD2, and it came back with a hit: a poster from a researcher at Mount Sinai that described a patient with a de novo inframe indel in BICD2.
And it was the same indel we’d found in our patient, a deletion of a single amino acid: p.(Asn546del).
I met the poster’s author, who was an Ob/Gyn doing a fellowship in genetics. My main concern was that we might have the same patient. We quickly determined that this was not the case — her patient was several years older and female. Then we discussed the clinical features, and determined that the overlap was significant. Same gene, same variant, similar clinical presentation in unrelated patients. That’s the holy grail of rare disease research. It was enough to publish, and (importantly) enough to convince me and my collaborators that BICD2 was the answer after all.
BICD2 Structure and Function
BICD2 itself is a fascinating gene. It’s one of two human homologs of the fly gene Bicaudal D (bicD), so named because mutating it in Drosophila produced flies with two trunk segments, rather than a head and a trunk. As a component of the dynein molecular motor complex, human BICD2 plays a pivotal role in intracellular transport. The protein’s three coiled-coil domains have specific binding partners:
BICD2’s N-terminal stabilizes the dynein-dynactin complex. Direct visualization in live cells indicates that the complex alone is unable to interact with microtubules, but tethering of BICD2’s C-terminus to different membrane cargoes induces their movement toward microtubule minus ends. As a cell prepares to divide, BICD2 switches its binding preference from RAB6A to nucleoporin RANBP2, recruiting it to the nuclear pore context and regulating dynein and kinesin to keep the centrosomes closely tethered to the nucleus prior entering mitosis.
Many of the consequences of a pathogenic BICD2 mutations in muscle cells are striking enough to be seen under a microscope. The most common mutation in SMALED2 (p.S107L) occurs in the first coiled-coil domain and appears to increase BICD2’s binding affinity for dynein. This causes the normally compact Golgi apparatus to disperse throughout the cell, a phenomenon of impaired dynein function called Golgi fragmentation.
New Information on BICD2 and Disease
Some new information emerged as I worked on the case report. First, the study coordinator discovered that our patient had recently passed away at the age of six. This was saddening, even though our finding would not have changed it.
Second, I learned that there had been new reports on BICD2 in the intervening years, some of which described a more severe phenotype with onset in utero associated with de novo mutations in BICD2. At around the time I’d gone to ASHG in 2017, a group at the University of Cologne (Storbeck et al) published a study emphasizing the phentoypic extremes of BICD2 mutation carriers. Four of their five reported patients had also passed away at a young age. The last line of their abstract read:
Our data define an additional severe disease type caused by BICD2 and emphasize a possibly variable etiology of BICD2-opathies with regard to primary muscle and neuronal involvement.
Around the time we submitted our manuscript, a group in France published a case report of an inframe indel in BICD2 segregating in a family with non-progressive SMA. Now there was precedent for the variant type as well.
Our report in Molecular Case Studies added to the growing number of reports of patients with BICD2 variants who manifested a disease that was far more severe than the later-onset, lower-extremity-predominant SMA on record for BICD2. This was brought to the attention of the curators of the OMIM database, who revised their entry for BICD2 to recognize two associated disorders:
- SMALED 2A (MIM #615290), the classic presentation of later-onset affecting mainly lower limbs
- SMALED 2B (MIM #618291), the severe systemic disease with prenatal onset
Both conditions are caused by dominant missense or inframe variants in BICD2. The large family kindreds published in 2013 fall under type 2A. Our patient and others with de novo mutations generally fall under type 2B.
A Comprehensive Look at BICD2 Variants and Disease
New information continues to emerge as more patients are described in the literature, or their variants are submitted to the ClinVar database by clinical laboratories. As best I could tell, there were close to a hundred patients with pathogenic BICD2 variants described across a dozen publications and/or the ClinVar database.
I spent much of 2019 collating all of these reports and organizing them into the most comprehensive review to date of the genetic basis and phenotypic spectrum of BICD2 disease in humans. It was just published in Annals of Neurology and I hope you’ll give it a read:
Koboldt DC, Waldrop MA, Wilson RK, and Flanigan KM. The Genotypic and Phenotypic Spectrum of BICD2 Variants in Spinal Muscular Atrophy. Ann Neurol. 2020 Apr;87(4):487-496. doi: 10.1002/ana.25704. PubMed: 32057122