Our search for maternally-mediated genetic risk of orofacial clefts in offspring was motivated by evidence from animal studies demonstrating an ability of the products of maternal genes to directly intervene and protect the fetus. For example, Letterio et al. 
showed that maternal Tgfb1 was able to cross the placenta and rescue Tgfb1−/−
mice. Similar observations were made in an earlier experiment that tested whether maternal epidermal growth factor (Egf) could be transported to the fetus via the placenta 
. In humans, however, evidence for maternal gene effects on the risk of clefting is less direct, and the few published studies are primarily single-gene association analyses that provide only a partial assessment of risk. To address this limitation, we focused here on broader gene categories and pathways, including the cholesterol pathway, the folate/homocysteine pathway, and genes involved in the detoxification of xenobiotic compounds.
Of the large number of candidate genes tested in this study, only FLNB
was detected by both methods and across both populations in the iCP cleft category. This gene belongs to a family of actin-binding proteins that are highly conserved and widely expressed during development 
. Filamins were discovered as the first family of non-muscle actin-binding proteins 
. They stabilize the cytoskeletal network by cross-linking actin, and thus linking the cell membrane to the cytoskeleton, and by providing scaffolds on which intracellular signaling and protein trafficking pathways are organized 
. In humans, mutations in FLNB
disrupt vertebral segmentation, joint formation, and skeletogenesis 
. Examples of skeletal disorders include boomerang dysplasia, autosomal-recessive spondylocarpotarsal syndrome, autosomal-dominant Larsen syndrome, and the perinatal lethal atelosteogenesis phenotypes I and III 
. Interestingly, many of the filaminopathies manifest cleft palate
as part of the overall phenotype 
, which is consistent with our findings of an association of FLNB
in iCP alone and not in the larger sample of iCL/P.
did not fully meet our stringency criterion for replication, despite being the top two genes associated with iCP in TRIMM analyses of the Norwegian triads. This may be due to the small size of the Danish iCP triads (69 iCP case-parent triads), which may have limited the power to detect an association. Nonetheless, both of these genes are plausible candidates for orofacial clefting. HIC1
encodes a zinc-finger transcription factor and maps to chromosome 17p13.3, within a 350 kb region found to be deleted in most patients with Miller-Dieker lissencephaly syndrome (MDLS) 
. Patients with MDLS exhibit a range of developmental anomalies, including omphalocele, limb and digit defects, and craniofacial dismorphology. Further, mice lacking Hic1
die perinatally and have cleft palate among a range of developmental defects 
. The fact that Hic1−/−
mice exhibit cleft palate is noteworthy, given that the association of HIC1
was confined to iCP in our data (with no association seen in the larger iCL/P sample). HIC1
is also a potential tumor suppressor gene; it is frequently hypermethylated and its expression is downregulated in several types of cancer 
. The link to cancer is noteworthy given the higher risk of cancer reported among parents whose first liveborn child had cleft lip/palate 
maps to chromosome 9q22–q31 and encodes a Krüppel-like zinc finger protein. Recent genome-wide linkage analyses of a large number of multiplex families from diverse populations uncovered a highly significant linkage signal to the 9q22–q33 region encompassing ZNF189 
. Although several important candidate genes for clefts have been characterized in this region (e.g. human homolog of patched
), receptor tyrosine kinase-like orphan receptor 2 (ROR2 
), transforming growth factor beta receptor type 1 (TGFBR1 
), and forkhead box E1 (FOXE1 
has not previously been linked with clefting.
Our study was based on two national cleft cohorts of similar ancestry, two complementary and robust statistical methods, and a large panel of SNPs in one of the largest available collections of cleft candidate genes. Despite the breadth of our approach, there was little evidence to suggest that maternal genes influenced the risk of iCL/P or iCP in our data. This apparent lack of maternal gene effects is consistent with recent epidemiological data on familial patterns of recurrence of orofacial clefts. If maternal genes had an impact on the risk of clefting through effects on the uterine environment, mother-to-offspring recurrence would be higher than father-to-offspring recurrence. However, mother-to-offspring recurrence of clefts was not statistically different from father-to-offspring recurrence in either Norway or Denmark 
. There was no statistically significant difference either between parent-to-offspring and sibling-to-sibling recurrence, suggesting that fetal genes alone are more likely to explain the majority of genetic risk for orofacial clefts.
Our results are also consistent with those of a larger study that investigated whether half sibships ascertained through an affected proband had a higher risk of clefts when the mother was the common parent 
. A higher occurrence of clefting would be expected if a major maternal effect exists, but no such evidence was found in that study. Finally, in our recent pathway-wide analysis of maternal genes and the risk of CL/P and CP among 29 genes involved in folate/one-carbon metabolism, we found no convincing indication that genetic variants in these folate metabolism genes play an etiologic role in orofacial clefting 
It is also possible that maternal genes alone do not confer risk of clefts to the newborn unless specific environmental exposures are also present. For example, a reduced capacity of mothers to biotransform toxins due to a genetic susceptibility has been proposed as a plausible explanation for the adverse effects of smoking and alcohol consumption on pregnancy outcomes 
. A non-additive interaction may be triggered only when the mothers are exposed to smoking or alcohol during the first trimester of pregnancy. In addition, interactions between the maternal and fetal genotype cannot be ruled out. Such interactions should be evident as ‘main effects’ of the participating genotypes, but the magnitude of the apparent effect would be blunted without accounting for the etiologic cofactor, making these effects difficult to detect.
In conclusion, with the possible exception of FLNB, HIC1 and ZNF189, our data suggest that maternal genes do not contribute significantly to orofacial clefting in the Norwegian and Danish samples. This is consistent with recent reports on familial patterns of recurrence of orofacial clefts. It is likely that fetal genes explain the majority of genetic risk for orofacial clefts in these two populations. However, our study does not rule out the possibility that maternal genes may affect risk through interactions with specific teratogens or fetal genes.