Point mutations in the candidate genes FOXE1, GLI2, MSX2, SKI, SATB2, and SPRY2 appear in aggregate to contribute to as much as 6% of isolated cleft lip and palate cases, enriched for cases with bilateral cleft of the lip with cleft palate and a positive family history. The mutations found in this study are conserved in other mammals, may disrupt exonic splicing enhancer sequences, and were not found in between 400 to 2,000 control chromosomes. The JAG2 M597I and A657H mutations, although they appear to disrupt exonic splicing enhancer sequences and possibly damage the JAG2 protein, according to the PolyPhen prediction, are not conserved in other species and may be rare polymorphic sites.
Testing a larger number of control samples proved to be a useful way to differentiate rare polymorphisms from etiologic mutations. The LHX8
D20A, and TBX10
R354Q variants initially were not observed in approximately 200 matched controls. This number of controls is commonly used to assume that if a variant is not present, it is likely causal despite models that have shown that a larger number of controls is useful in eliminating rare variants [8
]. However, when we tested these variants in the extended set of 1,064 controls, we found the LHX8
E221A variant in 17 individuals, the SKI
A388V mutation in nine, the SPRY2
D20A mutation in 60 individuals, and the TBX10
R354Q in six individuals. Although the presence of the amino acid changes in unaffected controls does not exclude them from playing a role in CL/P, it does place them in a lower priority group for additional functional analysis and makes them difficult to use in any applied genetic counseling setting as they may be, at best, modifiers with low penetrance contributory alleles.
Some variants, such as the SKI A388V mutation, which is conserved back to Xenopus and Tilapia but is found in controls, demonstrate that species conservation alone maybe not enough to argue for an etiologic role of a given variant.
Our study illustrates the difficulties of defining as causal a mutation rarely seen in the population. Many of the missense mutations found in the cases studied were seen only when we extended our screen to a larger control group comprised of samples from ethnically diverse groups from almost all parts of the world. In addition, none of the mutations segregated from an affected parent. Incomplete penetrance is likely the explanation for the mutations that may be causal as has been clearly shown for other genes that contribute to clefting such as MSX1
] and FGFR1
] mutations. It is likely that we found these mutations only because we tested cases more likely to present a stronger genetic contribution (cases with positive family history and bilateral cleft of the lip with cleft palate). Mutations like the MSX1
P147Q and others that appear to show incomplete penetrance are comparable to autosomal dominant disorders resulting from mutations in SCN5A
], or NKX2.5
P147Q mutation was seen in two cleft cases from the Philippines, but in none of the over 1,600 controls. It appears that this specific mutation underlies approximately 0.15% cases of apparent isolated CL/P. As shown previously this variant results in variable expression and decreased penetrance that make prospective studies of its phenotypic outcome necessary before accurate genetic counseling risk can be measured [5
Rigorous demonstration that a mutation disrupts a genuine exonic splicing enhancer requires that the sequence autonomously promote splicing and that enhancement be absent in the mutant. An advantage of the score matrix approach [14
] is that it allows direct testing of predicted effects on individual putative enhancer sites, rather than having to characterize exonic splicing enhancers by testing multiple random mutations and/or deletions along an exon. All 15 mutations on Table S2 appear to inactivate and/or create a predicted exonic splicing enhancer of at least one of the four serine/arginine-rich (SR) proteins. However, the presence of a high-score motif in a sequence does not necessarily identify that sequence as an exonic splicing enhancer in its native context.
In TBX1, we found one rare intronic variant in homozygous form in an Iowa cleft case, which could indicate that clefts arise from recessive functional intronic mutations in TBX1, or microdeletions that cannot be visualized by direct sequencing. This case does not have a 22q deletion involving the UFD1L gene. Detecting this rare homozygote in the absence of this variant in any other of the 400 people tested suggests these individuals may be identical by descent at this locus and gene. This variant itself, or others in linkage disequilibrium in TBX1, might be a hypomorphic allele whose joint presence results in enough change in gene expression or function to trigger a phenotype. Therefore, other alleles in regulatory regions of TBX1 should be a priority for identification.
Previous work from our group has screened FGFR1, IRF6, MSX1, TGFA,
for mutations on cleft cases. Point mutations in MSX1
appear to contribute approximately to 2% of all CL/P cases [4
point mutations also appear to contribute to CL/P. In addition, FGFR1
loss-of-function mutations can cause forms of Kallmann syndrome that mimic isolated CL/P at birth and during childhood. Mutations in IRF6,
which cause the syndromic forms of clefts, the Van der Woude and popliteal pterygium syndromes [12
], were not found in the same collection of cleft cases as the present study, although rare non-coding variants in conserved regions were disclosed. However, IRF6
is strongly associated with CL/P, and it is likely a genetic modifier for clefts [15
]. Previously for TGFA,
five variants in conserved non-coding segments were found in individual cases but not seen in 278 controls [17
]. In the present study we found another nine non-coding rare variants in single individuals, but we did not find the original five reported rare variants or any coding mutation that could be etiologic. If these variants are disease-causing mutations, they could explain, not only the conflicting results from association studies of isolated orofacial clefts and TGFA
], but also the linkage studies that suggest a cleft susceptibility loci in 2p13, the TGFA
]. For TGFB3,
we previously reported one missense mutation (K130R) in a cleft palate–only case not seen in 350 controls [19
]. We did not find this or any other mutations in TGFB3
in our current study population.
Loss-of-function mutations in GLI2
are associated with pituitary anomalies and holoprosencephaly-like features [20
]. In this report, the three pedigrees segregating GLI2
loss-of-function mutations with complete clinical information presented orofacial clefts and polydactyly. We found four missense mutations in GLI2
in highly conserved amino acids. One of the cases also presented with polydactyly ().
We performed linkage disequilibrium studies in the genes that we found potentially disease-causing missense mutations. We found association between a marker (rs2843159) and a haplotype in SKI
in the Filipino population. This association was also found in an independent population dataset from South America. The SKI
locus, 1p36.3, was previously suggested as a cleft susceptibility loci in Caucasians [21
null mice present with clefts involving the lip [22
], and the association we found appears to be stronger when cases with the involvement of the lip are included.
The possible trend for an association between clefts with a palate phenotype and SATB2
are in agreement with the cytogenetic evidence. Deletions and balanced translocations point to the existence of a locus on 2q32-q33, for which haploinsufficiency results in isolated cleft palate. A mutation analysis of SATB2
(located at 2q32-q33) in 70 unrelated patients with isolated cleft palate only did not reveal any coding region variants [23
]. However, a meta-analysis of 13 genome scans for clefts indicated 2q32-q35 as a clefting susceptibility locus [7
]. We studied 184 cleft lip and palate cases and found one missense mutation in SATB2
(T190A) that was not seen in approximately 1,200 controls. Based on the linkage disequilibrium and mutation analysis results of our study, we believe a regulatory element outside SATB2
coding regions may be implicated in clefting.
In summary, point mutations in six of the 20 candidate genes selected from expression, animal, and human data may to contribute to about 5% of isolated clefts, more likely those with more-severe phenotypes and/or a positive family history. Etiologic variants in regulatory elements of SKI, JAG2,
may contribute to isolated clefts as well. Predictions by ESEfinder (http://rulai.cshl.edu/tools/ESE/
) and PolyPhen regarding the function of the missense mutations found in this study, as well as exonic splicing enhancement and protein damaging, are challenging to interpret.
A major challenge in these studies was the frequent absence of a cleft phenotype in near relatives of an affected proband with a cleft and a rare missense mutation. In some cases these variants may not be etiologic, but in others, reduced penetrance for the cleft may be an active force as has been seen commonly in clefts [9
] and other birth defects such as congenital hearth disease. Similarly these mutations may only be modifiers of the phenotype.
Cases due to microdeletions or isodisomy may contribute to clefts as well. This study illustrates the validity of testing greater numbers of controls to determine rare polymorphic variants and prioritize functional studies for rare point mutations. Given other recent data on the roles of FGFR1, IRF6, and MSX1 in isolated CL/P, one can begin to consider sequencing of a panel of high-probability candidate genes for genetic counseling indication. Although issues of penetrance and even etiology for any given mutation are not yet resolved, progress in this direction is now measurable.