DISCUSSION
Precision medicine utilizing genomic methods implies that medical
geneticists can effectively navigate a series complex interactions that
ultimately induce malformation conditions such as HPE. The challenges to
this effort are many. These can include, in part, the variable quality
of the data, the impulse to over-filter datasets in an effort to
simplify the genetic interpretation, the lack of clear criteria for
driver mutations as either sole or multifactorial factors (e.g. digenic
inheritance), the paucity of models for determining modifier effects,
the frequent lack of clinical details of the family structure and
affection status, any teratogen/environmental exposure considerations,
the incomplete understanding of potential gene:gene and/or
gene:environment interactions, the lack of effective guidelines for
non-coding variation, the tendency to ignore the key role of the timing
of these interacting factors, as well as the continuously evolving
considerations attempting to foster an integrated interpretation into
medical practice. Extrapolating from lessons gleaned from the example of
carcinogenesis clearly supports the role of considering both genes and
the environment. This is almost certainly the case with the pathogenesis
of HPE. Furthermore, the genetic variation of human SMO is now
clearly implicated in both rare developmental malformations, as well as
somatic variants in human cancer. We now argue that the knowledge of the
Hh pathway components with respect to their fundamental sequence
conservation and their gene regulatory relationships provides insight
into both.
Herein, we investigated the role of the SMO gene in craniofacial
and forebrain development using zebrafish as a model system. This study
is intended as a comprehensive analysis of variants detected in a
malformation (HPE) known to be caused by disturbed function of the
hedgehog pathway. In doing so, our study directly addresses the apparent
contradiction that although impaired Hh signaling is the most common
pathological mechanism for HPE, the hypo-function of SMO is
indeed uncommon. And yet, gain-of-function mutations in SMO are a
frequent cause of both basal cell carcinoma and many of the most common
pediatric brain cancers.
Many of the complicating factors in precision medicine are evident in
the present report on these six cases. Current guidelines (Richards et
al., 2015) appropriately rank the relative importance of the novelty of
a variant, the de novo status, the co-inheritance of relevant
variants, instances of similar variants with proven pathogenicity, the
family history, similarly affected relatives, the bioinformatic
pathogenicity scores, etc., into an integrated determination. Of the
proband-specific findings in this report (Suppl.TableS7), only theFGF8 and SHH mutations meet these criteria for
pathogenicity, while the SMO variants remained unclassified.
Thus, functional studies are often the determinative factor in assigning
clarity to the interpretation. Nevertheless, functional studies are
frequently difficult to perform in vivo with developmental
programs with an extensive self-regulating GRN.
To date, most published assays of Hh activity have depended on
reporter-fusions (e.g. 8X-GLIBS-luciferase) in readily transfectable
cells or similar GLI-responsive markers (Myers et al., 2013). While
these methods are readily quantifiable, they greatly oversimplify the
known tissue interactions of developmental processes that operatein vivo . These cell-based assays often require the absence of
Ptch and an intact GRN present in the living animal. A promising advance
is the recent development of rapid assays of SMO activity that
can determine biological activities in minutes or hours (Myers et al.,
2017).
In all of our over-expression assays, we empirically determined that
additional interventions were required to obtain interpretable data.
Furthermore, the activation, or suppression, of the GRN obscured any
differences between the variants themselves, or with the WT control.
Indeed, the marker and morphological criteria we used as a biological
response required many hours or days to become measurable. We observed
that many of these over-expression measures were highly time-dependent
(data not shown). By contrast, the ability to generate CRISPER/Cas9
mutant test lines has proven to be an immensely powerful tool. It avoids
the requirement for co-interventions that can sometimes obscure the
results.
Finally, recent advances have independently supported our identification
human p.V404 as an important residue in the 7TM domain of SMO .
While formally no longer considered novel (PM2), we observe that the
single whole genome detection of p.V404M occurred in a distinct ethnic
group for our proband. The analogous murine residue p.V408 is implicated
in cholesterol agonist binding (PM1). An engineered variant p.V408F in
this site sterically blocks cholesterol and abrogates GLI-luciferase
reporter activity (Deshpande et al., 2019). V404, the binding pocket
residues for p.V408, was reported the mutation have the potential
function as alter sterol affinity or directly stabilize the active SMO
conformation (Sharpe et al., 2015). While our human variant p.V404M
lacks the same bulky Phe side chain, the structural similarities are
consistent with the requirements of this residue being essential inSMO function.
In conclusion, the value of Targeted Sequencing of the Hh pathway has
established a framework for integrated genetics interpretations of the
potential for gene:gene interactions that can serve as a basis for more
comprehensive and accurate genetic counseling in HPE. Strict adherence
to existing guidelines indicate that the driver mutations described in
this report, such as FGF8 and SHH , rarely require
co-morbid modifiers. As whole exome and whole genome data become more
readily available for HPE, we anticipate that this autosomal dominant
with modifier model will continue to be subject to experimental
validation and result in further insight.