Non-Syndromic Cleft Palate and Transverse Limb Deciency Segregating in a Family of Dogs

Oro-facial clefts are one of the most common birth defects in humans, most are non-syndromic, and few have established molecular diagnoses. Here we report the morphology and genetic transmission of isolated cleft palate in a naturally occurring dog model. Palate morphology was evaluated grossly, by microcomputed tomography, and by histologic examination of serial coronal sections. In repeated matings of a clinically normal sire/dam pair, 18% (12/68) of live-born pups had full-length cleft of the secondary palate with no other abnormalities. At the gestational stage of normal palate fusion, palate shelves of affected fetuses were above the tongue but did not meet at midline. Mandibles were normal, and oral epithelium and periderm were intact. Genetic transmission was determined in experimental backcross matings of surgically repaired affected dogs with a normal parent, which produced 20 cleft, 11 male and 9 female, and 24 normal-palate pups. Furthermore, all offspring of matings between affected dogs had cleft palate. These data were as expected under the hypothesis of autosomal recessive transmission of the cleft palate trait ([1 df, N = 44] Χ 2 = 0.36, p = 0.55). About half of cleft offspring produced in backcross matings of which the dam had cleft palate, also exhibited various transverse limb deciencies. No limb deciencies occurred in backcross offspring of a dam with normal palate, suggesting a possible maternal effect. This dog family constitutes a large animal model of nonsyndromic isolated cleft palate coincident with developmental limb deciency. = 6] 4.05 ± 0.53 mm, p > 0.25) nor between the right and left sides by either measurement method. Clinical CT of normal and affected littermates at 6 weeks of age showed the hard palate was continuously ossied across the roof of the oral cavity in normal pups, whereas the cleft palate pups had the same longitudinal gap between bones of the hard palate and tissue of the soft palate (not shown). These scans revealed no other abnormalities in the cleft palate pups.


Introduction
Orofacial clefts (OFC) are a group of craniofacial developmental defects that are common in humans and cause signi cant medical, psychosocial, and nancial burdens (Wehby and Cassell 2010). OFC are subdivided into cleft lip only (CLO), cleft lip with or without cleft palate (CL/P) and cleft palate only (CPO).
OFC occur either alone or as a feature of various syndromes involving other organ systems.
Nonsyndromic CL/P occurs in 1 in ~ 500 to 1 in ~ 2,500 live births, depending on the geographical, racial, and ethnic distribution of the study population (Dixon et al. 2011;Leslie and Marazita 2013). Defects leading to CL/P originate in embryonic development of the upper lip primordia and primary palate (Jiang et al. 2006). Nonsyndromic CPO occurs in 1 in ~ 1,500 live births, and defects leading to CPO originate in failed emergence or fusion of palate shelves to form the secondary palate. There is some overlap of risk factors for each, but CL/P and CPO seldom occur in the same family lines (Dixon et  OFC and LRD can be co-morbidities in humans and animals, particularly when caused by teratogen exposure (Kochhar et al. 1984;Hurst et al. 1995 Here we present characterization of the pathology and results of experimental matings consistent with simple autosomal recessive inheritance of CPO in this family. During breeding experiments some offspring from cleft-affected dams born with CPO also exhibited TLD. While unexpected, the type of limb de ciency and pattern of occurrence suggest various models of causation of both phenotypes.

Methods
Dogs were maintained by personnel of Campus Animal Resources of Michigan State University in a breeding colony for genetic disease research. The average vivarium census over the course of this study included ~ 30 breeding females, of which 6 were in the CPO line. The CPO line were mixed-breed dogs with genetic contributions from curly coated retrievers, giant schnauzers, beagles, and mongrels.
Protocols for animal use were approved by the MSU Institutional Animal Care and Use Committee. Dog matings were either brother/sister intercrosses or backcrosses (affected dog X normal parent) performed either by natural or arti cial insemination. Palate fusion of newborn pups was determined by physical examination of the oral cavity. Affected pups and littermate controls euthanized on the rst postnatal day were processed variously for histologic analysis or X-ray computed microtomography (µCT) of the head. Processing for light microscopic and immunohistochemical evaluation was as previously described (Freiberger et al. 2021). In some matings, day 0 of gestation was determined by serial measurements of pre-ovulatory serum progesterone concentration, as previously described (Freiberger et al. 2021), and fetal pups were removed by cesarean section on day 39 (d39). Immediately upon removal from the uterus, fetal pups were placed in ice-cold sterile saline, examined under a dissection microscope, and decapitated. Liver was snap frozen for DNA isolation and PCR-based sex determination, as described previously (Meyers-Wallen et al. 2017).
For histology, d39 fetal heads from 3 cleft-affected and 3 normal littermate pups were xed in formalin and processed for para n embedding. Serial coronal sections (7 µm) were collected from the rostral tip of the planum nasale through the back of the eyes. Hematoxylin and eosin (H&E) and immunostaining were performed as described previously (Freiberger et al. 2021). For µCT, heads from 3 cleft-affected and 3 normal newborn littermates were xed overnight in formalin, skinned, and transferred to 50% ethanol. Micro-CT was performed on a Perkin Elmer QuantumGX system (PerkinElmer, Inc, Waltham, MA, USA) using the following image acquisition parameters: high resolution scan mode, 57-min gantry rotation time, 90kVp/88uA power, 72 mm eld of view, 144 um voxel size, 512 slices at 144 um slice thickness, and 144 um 3 voxel resolution. We standardized all images for HU intensity from 0-5000 HU. Image rendering and analysis was performed using Analyze 14.0 (Biomedical Imaging Resource, Mayo Clinic, Rochester, MN, USA). To allow an unobstructed 3D image view of the ventral aspect of the palate, some 3D images were rendered after removing the mandible using the visualization software. We used 3D images to measure the length and width of mandibles, compiling measurements for the right and left mandible as separate areas. To obtain the total length of the mandible, we added measurements from i) the most caudal point of the condylar process to a point ventral to the 4th premolar on the ventral surface of the mandibular body, and ii) from that point to the most rostral extent of alveolar bone, excluding protruding teeth. Additionally, we developed a "curved" measurement function by free drawing several point-to-point measurements along the curved ventral surface of the mandible. Curved lengths were ~ 4% longer than the simple 3-point measurements. The width of each mandible was measured on a ventraldorsal view perpendicular to the long axis at the level of the suture between the palatine processes of the maxillae and palatine bones.
A male and 3 female affected pups were raised to six weeks of age for surgical closure of the palate cleft.
Each received 2 mL of the dam's serum subcutaneously on day 1, for passive transfer of immunity, and intermittent feedings of increasing amounts of bitch milk replacer (Esbilac®, PetAg, Hampshire, IL, USA) by stomach tube 8-9 times/day. Hydration was assessed continuously by monitoring skin turgor and fecal consistency and maintained as needed by subcutaneous or enteral administration of warm lactated Ringer's solution. Growth was monitored by daily weighing. At 3 weeks of age, feedings were supplemented by digital gavage with moistened kibble. Growth was normal throughout the rst 6 weeks (exponential, according to the equation y = 246.34e 0.0545x with R² = 0.998) (Alves 2020). At 6 weeks of age and ≥ 2 kg body weight, the affected pups underwent general anesthesia, computed tomography of the head, and 2-layer closure of the cleft, restoring both nasal and oral mucosae of the soft and hard palate. Recovery was uneventful; routine feedings of soft food began the day after surgery, evolving to regular kibble and return to group housing over the next 3 weeks.
For statistical analyses of breeding experiments, we used i) the Chi-Square goodness-of-t test to compare observed versus expected Mendelian ratios of affected/normal pups; ii) Fishers exact test when experimental values were less than 5; and iii) Odds ratio to assess association of phenotypes with potential maternal effects. Micro-CT measurements of mandibles of cleft and normal littermates (n = 3 in each group) were compared by two-tailed Student's t-test both separately for right and left mandibles and with the sides combined.

Results
Cleft phenotype: Isolated cleft palate occurred infrequently (~ 5%) in a breeding colony of mix-bred dogs maintained since 1991, until we began intercrosses between littermates of a pup born with cleft palate. In repeated matings (dog 170 X dog 173, Fig. 1), 12 affected pups (18%), 5 male and 7 females, were born among 68 total offspring. The cleft phenotype was remarkably consistent; at birth there was a 1.5-2 mm wide, full-length midline cleft of the secondary palate, extending caudally from the incisive papilla through the hard and soft palates (Fig. 2b). The sides of the cleft were essentially parallel. There were no clefts of lip or primary palate. No other external morphological abnormalities were apparent in pups born to this dam.
Micro-CT of normal newborn pups showed an ~ 0.6 mm longitudinal midline gap between mineralized bones of the hard palate (Fig. 3a-c), but the gap was lled entirely by soft tissue fused at the midline ( Fig. 3c). In cleft littermates, µCT demonstrated an ~ 3.2 mm wide longitudinal midline gap between boney edges of the bilateral palatine processes of the maxillae and horizontal laminae of palatine bones ( Fig. 3d-f). Soft tissue covering the boney tissues reduced the gap to 1.7-2.0 mm (Fig. 3d). Deciduous dentition was complete, and the vomer bone was intact. In 3 normal and 3 cleft littermates, there was no signi cant difference of the mandibular length or width (curved length: normal [n = 6] 36.58 ± 1.77 mm vs. With intensive nursing care we rescued 1 male and 3 females born with cleft palate and raised them to 6weeks of age, at which time they each underwent surgical reconstruction of the palate. After recovery from surgery they each went back into the general population and grew normally into healthy adults. At appropriate ages, they each developed full adult dentition and gained reproductive competency. The male routinely exhibited good libido and 500-800 million sperm/ejaculate when collected for arti cial insemination. The sperm were ~ 95% progressively motile with few morphologic abnormalities. The females each had their rst estrous cycle between 12 and 14 months of age, and each became pregnant when bred in their second cycles. None have had issues of poor health and are 4.5-7 years of age, at present. To explore potential pathophysiological mechanisms, we performed histologic analyses on serial coronal sections of heads from fetuses with and without cleft palate. By cesarean section we interrupted backcross matings of each of the 3 surgically repaired affected females with their common sire on estimated d39, the previously de ned day of normal palate fusion (Freiberger et al. 2021). Cleft palate was readily identi ed in 9 of 19 fetuses, 5 males and 4 females (Fig. 1). As expected, palate shelves in normal littermates had begun to fuse along the length of the secondary palate (Fig. 4a-c), as evidenced by dissolution of the epithelia in the medial edge seam. In affected fetuses, however, although the palate shelves were elevated above the tongue, a gap remained between them along the entire length of the palate shelves (Fig. 4d-f). We also tracked palate fusion by immuno-staining with antibodies directed against K17, a cytosolic protein marker for periderm, and p63, a nuclear protein marker for basal epithelial cells. Again, as expected in normal fetuses at this timepoint, the signals for both K17 (red) and p63 (green) were mostly gone along the medial edge seam, except at the epithelial triangles located at extreme nasal and oral ends of the medial edge seam (Fig. 5d, h). For cleft-affected littermates, the K17 and p63 signals on the oral and nasal sides of the palate shelves did not appear different from normal littermates. However, unlike the palate shelves from normal fetuses, the K17 and p63 signals remained robust and continuous along the medial ends of each unfused palate shelf and the vomer (Fig. 5l, p).

Clinical genetics:
Twelve pups, 5 male and 7 female, among 68 total offspring (18%) were born with cleft palate, occurring in 7 matings of clinically normal littermate parents (Fig. 1). While this result was not statistically different from 25% expected for an autosomal recessive trait ([1 df, N = 68] Χ 2 = 1.96, p = 0.16), the percentage of affected pups accumulating with each new litter had hovered at 18% since the third litter (5/30), suggesting that the de cit was real. Breeding surgically repaired affected dogs provided opportunity for prospective experiments. In backcross experiments, three matings of the affected male (248) to his dam (173) produced 3 cleft-affected pups, 2 male and 1 female, among 11 total offspring (27%). Although again not statistically different from the expected 50% for an autosomal recessive trait in backcross matings (p = 0.39), these results and those of the 170 X 173 matings, indicated above, suggested that the dam (173) might be transmitting a factor that suppressed penetrance of the cleft trait. To test that hypothesis, we performed backcross matings of the surgically repaired affected females. Collectively, matings of affected females 286, 329, and 335 ( Fig. 1) to their common sire (170) produced 17 cleftaffected pups, 9 male and 8 female, among 33 total offspring, exactly as expected for a simple autosomal recessive trait ([1 df, N = 33] Χ 2 = 0.30, p = 0.86). Furthermore, 2 matings of the affected male (248) with his affected full sib (286) produced 8 offspring, 3 male and 5 female, all of which had cleft palate, again as expected for a simple recessive trait. Thus, it appeared that penetrance of the cleft trait was reduced only when 173 was the dam.
As often occurs with inbreeding, we exposed a second phenotype among some offspring of matings that did not include dam 173. TLD appeared in offspring of backcrosses to the family sire (170) and intercrosses between cleft-affected dogs (Fig. 1). These matings produced 41 offspring, of which 16 had normal palate and 25 had cleft palate. Two male normal-palate offspring and 14 of the cleft-palate pups, 4 male and 10 female, had various TLD. An invariable abnormality was absent toenails of digits 3 and 4 on both hindlimbs (Fig. 6d). More severe abnormalities were only in the forelimbs, including various degrees of unilateral or bilateral TLD that were often asymmetric (Fig. 6c, e). The most severe de ciency was bilateral absence, amelia, of forelimbs below the scapulohumeral joint in 6 pups. Abnormalities in the two normal-palate pups were the absent rear toenails (Fig. 6h, i) and unilateral fusion of forepaw digits 4 and 5 with absence of the digit 5 toenail (right forepaw of one and left forepaw of the other) (Fig. 6g). As noted, the limb defects were highly associated with cleft palate (OR = 10.2, 95% CI: 1.9-54). Moreover, we produced 14 pups with limb defects among 25 CPO offspring from matings when the dams had cleft palate, but no limb defects among the 3 CPO offspring from backcross matings when the dam had normal palate (p = 0.002).

Discussion
Sporadic occurrence of cleft palate has been reported in many mammalian species (Łobodzińska, 2014), and animal models have been major contributors to understanding normal and abnormal palatogenesis. Mouse models have been particularly useful for dissecting the molecular genetics and mechanisms of craniofacial development, but they are rarely useful as clinical models for post-natal or in utero cleft palate repair because mice are small, and the gestational window for post-operative healing without scarring is short (Lorenz and Longaker 2003). The molecular genetic etiology of CPO in the dog family reported here is not yet known. However, the morphological and histological analyses may provide clues. The immuno-staining for K17 and p63 appeared normal, showing that periderm and basal epithelial formation and function were not grossly affected. At gestational d39, the palate shelves are fusing in normal dog fetuses (Freiberger et al. 2021), but while the palate shelves of affected pups had elevated over the tongue, they had failed to fuse. This could be failure of the palate shelves to grow and elongate horizontally so that the epithelia did not meet at midline. Alternatively, it could be failure of the opposing shelves to fuse at the midline when the leading-edge epithelia met, with subsequent growth of the head causing a relative retraction of the shelves that left the gap. Analyses of mutant mouse models indicate that developmental regulation of mesenchymal cell proliferation, palate shelf elongation, and fusion at the medial epithelial seam involves multiple signaling pathways, including sonic hedgehog (SHH), wingless-related integration site (WNT), broblast growth factor (FGF), bone morphogenic protein (BMP), and transforming growth factor β (TGFβ), and that there is extensive cross-regulation between pathways (Gritli- Even with a shared-etiology hypothesis, the large number of genes involved in palate and limb development and the pleiotropic effects of variants in those genes, argue against a candidate gene approach to determining the etiology of CPO in this family. Rather, we propose that a whole genome approach is warranted. Results of our breeding experiments are not consistent with a simple inheritance model that accounts for both phenotypes. The pedigree suggests that CPO is inherited as an autosomal recessive trait, but possibly with the added complexity of a suppressor allele. If there was an allele suppressing CPO, it appeared to be restricted to dog 173 because the frequency of CPO was as expected of simple recessive inheritance, except when 173 was the dam. One possible model is that a suppressor allele was inherited from 173 as a dominant allele with incomplete penetrance, and the same allele suppressing CPO also suppressed limb defects. However, this model fails to explain the complete absence of limb defects in CPO offspring from crosses with 173. The attractive hypothesis that limb defects were caused by the same recessive allele as caused CPO also requires modi cation because 11 of 25 offspring with CPO when 173 was not the dam did not exhibit limb defects. An alternative hypothesis is that the family sire (170) harbored a dominant TLD-causing allele that was only fully expressed in CPO offspring. And yet another possibility compatible with the data is that TLD is caused by a recessive allele that segregates independent of the CPO allele but, again, is fully expressed only in CPO offspring.
An explanation of our observations could also be that while CPO is indeed an autosomal recessive trait in this family, the limb defects were not an inherited trait, but instead resulted from insults to developing fetuses created in utero by the maternal genome. This model is fully consistent with the pedigree and is also suggested by the type of limb defects observed, notably variably expressed TLD. TLD are unique among LRD in that many are attributed to vascular disruption from placental or fetal blood clots during limb development, such as may occur in cases of maternal thrombophilia (Sadler and Rasmussen 2010; Ordal et al. 2016;Hunter 2000). This model lends itself to the observed maternal effect.
In summary, we describe a dog model for non-syndromic CPO. Backcross matings led to limb defects as a second phenotype in this family. While the two phenotypes may be related genetically, the current pedigree suggests that breeding strategies can be designed to minimize the occurrence of limb defects in these dogs. Thus, this line of dogs represents a well-characterized and highly reproducible large animal model for non-syndromic CPO. Finally, the pedigree suggests models of inheritance for the CPO and TLD phenotypes that provide a genetic framework to analyze future whole genome data.

Declarations
Funding There was no extramural support for this research.
Con icts of Interest The authors declare no con ict of interest.
Availability of data and material Not applicable  Pedigree of a dog family with high frequency cleft palate and transverse limb de ciency. The ancestry of dogs 170 and 173 is curly coated retriever from the sire side and mixed lineage from the dam side, including giant schnauzer, beagle and mongrel. Phenotype labeling is shown in the key. The numbers inside the symbols represent the number of individuals for that phenotypic class. The number above a symbol is the identi cation number for that dog, as appear in the text. An asterisk is placed next to the identi cation numbers of the four affected dogs whose cleft palate was surgically repaired, including one male (248) and three females (286, 329, and 335).     Variable expressivity of limb defects. Newborn dog with typical forelimbs (a) and hindlimbs (b). Newborn dogs with variable transverse limb de ciencies in the forelimbs (c, e). All dogs with forelimb de ciencies were also missing the toenails for digits 3 and 4 in the hindlimbs (d). Images of forepaws (f and g) and hind paws (h and i) of a dog born without cleft palate, but with limb abnormalities. The right forepaw is normal (f), but in the left forepaw (g), digits 4 and 5 are fused and digit 5 is missing the toenail. Like all dogs with cleft palate and limb defects, digits 3 and 4 on both hindlimbs are missing the toenails (h, i).