Genetic interaction between central pair apparatus genes CFAP221, CFAP54, and SPEF2 in mouse models of primary ciliary dyskinesia

Primary ciliary dyskinesia (PCD) is a genetically heterogeneous syndrome that results from defects in motile cilia. The ciliary axoneme has a 9 + 2 microtubule structure consisting of nine peripheral doublets surrounding a central pair apparatus (CPA), which plays a critical role in regulating proper ciliary function. We have previously shown that mouse models with mutations in CPA genes CFAP221, CFAP54, and SPEF2 have a PCD phenotype with defects in ciliary motility. In this study, we investigated potential genetic interaction between these CPA genes by generating each combination of double heterozygous and double homozygous mutants. No detectable cilia-related phenotypes were observed in double heterozygotes, but all three double homozygous mutant lines exhibit early mortality and typically develop severe PCD-associated phenotypes of hydrocephalus, mucociliary clearance defects in the upper airway, and abnormal spermatogenesis. Double homozygous cilia are generally intact and display a normal morphology and distribution. Spermiogenesis is aborted in double homozygotes, with an absence of mature flagella on elongating spermatids and epididymal sperm. These findings identify genetic interactions between CPA genes and genetic mechanisms regulating the CPA and motile cilia function.


Results
Double homozygosity results in early lethality. Mice homozygous for the nm1054, Cfap54 gt/gt , and bgh mutations lack CPA projection proteins CFAP221, CFAP54, and SPEF2, respectively (Fig. 1a). Each single mutant was shown to exhibit early mortality on the C57BL/6J (B6) background due to severe hydrocephalus but typically live a normal life span on the 129S6/SvEvTac (129) background or a mixed background without gross hydrocephalus 26,27,29 . In contrast, early mortality was common for double homozygotes on a mixed background (Fig. 1b). The most severe mortality phenotype was observed in nm1054/nm1054;bgh/bgh double mutants, with all mice dying or requiring euthanasia due to severe hydrocephalus by weaning age (3 weeks). The bgh/ bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygotes also demonstrated substantial mortality, with less than half surviving to adulthood (8 weeks). Survival was best for nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt double mutants, with most mice surviving to adulthood. Gross hydrocephalus was common in mice that died prior to 8 weeks, regardless of genotype. All double heterozygotes (nm1054/+ ;bgh/+ , nm1054/+ ;Cfap54 gt/gt /+ , and bgh/+ ;Cfap54 gt/gt /+) www.nature.com/scientificreports/ survived to adulthood and were analyzed at 8 weeks or older, along with wild type (WT) controls (Supplementary Table S1).
Double homozygosity results in severe hydrocephalus. To assess the extent of damage in the brains of double mutant mice on a mixed background, we performed histological analyses on coronal sections through the lateral ventricles, through which cerebrospinal fluid (CSF) flows and accumulates under hydrocephalic conditions. The lateral ventricle is appropriately narrow in WT and double heterozygous mice with no evidence of tissue damage ( Fig. 2a-d). However, a severe hydrocephalic phenotype was observed in all three double homozygotes ( Supplementary Fig. S1). The nm1054/nm1054;bgh/bgh brain shows enlarged ventricles but not extensive tissue damage, and ependymal cilia lining the ventricular wall are intact (Fig. 2e,h). It is important to note, however, that only one mouse survived long enough for analysis (Fig. 1b), so the phenotype may have been more severe in those animals that died earlier. The nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt double homozygotes typically showed mild ventricular dilatation without extensive tissue damage, and ependymal cilia remained intact (Fig. 2f,i). The most severe phenotype was observed in bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt brains, which commonly exhibited severe ventricular dilatation, as well as extensive damage to the white matter and cerebral cortex that was often accompanied by intraventricular hemorrhaging with a dramatic influx of red blood cells (Fig. 2g,j). While nm1054, bgh, and Cfap54 gt/gt single mutants were each shown to exhibit severe hydrocephalus, this phenotype was only observed on the more susceptible B6 background and not on a mixed background 26,27,29 . Double homozygosity results in severe airway abnormalities. Since motile ciliary dysfunction typically results in airway abnormalities, we assessed the pathology of the maxillary sinus cavity. WT and double heterozygous mice have a clear sinus cavity with little to no evidence of mucus ( Fig. 3a-d). In contrast, all three double homozygotes exhibit a more severe defect in mucociliary clearance and airway pathology (Supplementary Fig. S1). The one surviving nm1054/nm1054;bgh/bgh mouse shows severe accumulation of mucus and extensive infiltration of neutrophils, which is common in sinusitis pathology and is indicative of the tissue repair process (Fig. 3e,h). Despite the mucus and white blood cell accumulation, some motile cilia remain intact on the epithelial surface (Fig. 3h). The nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt double homozygotes typically showed some areas of mild mucus accumulation but no evidence of extensive neutrophil infiltration, and epithelial cilia remain intact (Fig. 3f,i). Like the nm1054/nm1054;bgh/bgh mouse, the bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt sinus cavity typically shows severe accumulation of mucus and areas with mild infiltration of neutrophils, sometimes accompanied by red blood cells (Fig. 3g,j). Motile cilia remain largely intact on bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt sinus epithelia (Fig. 3j). The sinus phenotype in double homozygous mice is consistent with mucus accumulation in the single nm1054, bgh, and Cfap54 gt/gt mutants on a mixed background, as well as neutrophil infiltration in bgh mice, but the presence of red blood cells was never observed in single mutants 26,27,29 .
The morphology and distribution of motile cilia lining the sinus cavity was further assessed by immunohistochemistry (IHC). Ciliary marker acetylated tubulin is expressed normally in WT cilia (Fig. 4a,a′), while basal body marker γ-tubulin localizes to the ciliary base (Fig. 4f,f′). Staining of acetylated tubulin confirms that the intact cilia on all three double homozygotes have a generally typical morphology and distribution consistent with WT mice, and there is no statistically significant difference in acetylated tubulin levels ( Fig. 4b-e,b′-d′). In addition, γ-tubulin is normally expressed and properly localized to the ciliary base in all three double homozygotes, suggesting that ciliogenesis is not perturbed (Fig. 4g-j,g′-i′). Known CPA protein SPAG6 is expressed throughout the axoneme of WT cilia (Fig. 4k,k′), and relatively weak SPAG6 staining in cilia from double homozygous sinus epithelia suggests that defects in proper CPA assembly are possible, although there are no statistically significant differences ( Fig. 4l-o,l′-n′). Analysis of known radial spoke marker RSPH4A shows little difference in axonemal expression or protein level between WT and double homozygous mice ( Fig. 4p-t,p′-s′), further confirming a largely normal ciliary assembly. Consistent with the normal pathology of the sinus cavity in double heterozygous mice, the morphology and distribution of motile cilia appears normal in all three double heterozygotes, also with no statistically significant difference in acetylated tubulin levels ( Supplementary Fig. S2). In addition, tracheal CBF was measured for double heterozygotes using a high speed video microscopy approach, and levels were not statistically different from WT for any of the double heterozygous mice, confirming that the motile cilia are functionally normal (Supplementary Fig. S3). CBF analysis could not be performed efficiently or accurately on double homozygotes due to the low number of mice surviving to adulthood and technical challenges associated with measurement using the small neonatal or young postnatal tracheae.
Double homozygosity impairs spermatogenesis. Although male mice lacking CFAP221, SPEF2, or CFAP54 are infertile 26,27,29 , fertility tests were not performed for double homozygotes due to the low number of mice surviving to sexual maturity. Double heterozygotes are fertile and were used routinely for breeding to obtain double homozygotes. Effect of double homozygosity on spermatogenesis was investigated using a histopathological approach. As elongating spermatids develop into mature spermatozoa during spermiogenesis, the final step of spermatogenesis, WT flagella extend into the lumen of the seminiferous tubule of the testis (Fig. 5a). In contrast, nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygotes both show an absence of flagella in the lumen of the seminiferous tubule (Fig. 5b,c), indicating that flagellar formation is perturbed. While this is different from the presence of motile cilia on the sinus epithelia from these mice, it is not dissimilar to the spermatogenesis phenotype observed in single nm1054, bgh, or Cfap54 gt/gt mutants 26,27,29 . No testis data was obtained for nm1054/nm1054;bgh/bgh double homozygotes, as none survived to sexual maturity, and the only one that survived long enough to collect tissues was a female. Normal spermiogenesis was observed in the testes from nm1054/+ ;bgh/+ , nm1054/+ ;Cfap54 gt/gt /+ , and bgh/+ ;Cfap54 gt/gt /+ double heterozygous mice (Fig. 5d-f), which is consistent with their fertility. www.nature.com/scientificreports/ To further investigate the ability of double homozygous testes to assemble flagella, we performed IHC analysis. An antibody to ciliary and flagellar marker acetylated tubulin effectively detects the elongating flagella in the lumen of the WT seminiferous tubule (Fig. 6a,a′). While no flagella are present in the lumen of nm1054/ nm1054;Cfap54 gt/gt /Cfap54 gt/gt or bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygous testes, there are defined structures detected by the acetylated tubulin antibody in the developing spermatids from both mutants (Fig. 6b,c,b′,c′), suggesting that some rudimentary axonemal structures possessing acetylated tubulin may be assembled in those cells. The antibody to basal body marker γ-tubulin primarily stains immature spermatogonia and developing www.nature.com/scientificreports/ spermatocytes within the WT seminiferous tubule and both double homozygotes ( Fig. 6d-f,d′-f '′), suggesting that early stages of spermatogenesis are unaffected. However, while CPA protein SPAG6 is detected throughout the flagellar axoneme in WT seminiferous tubules (Fig. 6g,g′), low SPAG6 levels without a clear staining pattern in double homozygotes suggests that CPA assembly may be impaired (Fig. 6h,h′,i,i′). The defined axonemal structures detected by the acetylated tubulin antibody in nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt spermatids are not detected by the SPAG6 antibody, with only occasional sperm heads observed. An antibody to known dynein marker DNAI1 also stains the length of the flagellar axoneme in WT testis (Fig. 6j,j′). In contrast to SPAG6, DNAI1 shows strong expression in the developing spermatids surrounding the lumen of the nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt seminiferous tubules (Fig. 6k,k′,l,l′), indicating that dynein pre-assembly may occur in the spermatid cytoplasm even in the absence of mature flagella. Acetylated tubulin staining in nm1054/+ ;bgh/+ , nm1054/+ ;Cfap54 gt/gt /+ , and bgh/+ ;Cfap54 gt/gt /+ double heterozygotes highlights their morphologically normal elongating spermatid flagella ( Supplementary Fig. S4).
In addition to testis pathology, the morphology of mature sperm from the epididymis was also assessed using light microscopy. WT sperm have a hook-shaped head and a long flagellum (Fig. 7a). In contrast, sperm were rarely present in the nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt or bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygous epididymis, and those that were present have normal heads but severely shortened flagellar stubs that would likely impair proper sperm motility (Fig. 7b,c). As described above, sperm data could not be obtained for nm1054/nm1054;bgh/bgh double homozygotes, as only one female mutant survived long enough to collect www.nature.com/scientificreports/ tissues. Consistent with the testis data and normal fertility, all double heterozygous mice have epididymal sperm that are indistinguishable from WT sperm (Fig. 7d-f).

Discussion
In this study, we investigated potential genetic interaction between CPA genes CFAP221, SPEF2, and CFAP54 by crossing the nm1054, bgh, and Cfap54 gt/gt mouse models with mutations in those genes. We identified an absence of ciliary phenotypes in double heterozygotes, but nm1054/nm1054;bgh/bgh, nm1054/nm1054;Cfap54 gt/gt / Cfap54 gt/gt , and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygotes all exhibit PCD phenotypes of hydrocephalus,  www.nature.com/scientificreports/ mucociliary clearance defects, and spermatogenesis abnormalities. Double mutants also exhibit severe early mortality on a mixed background that is not observed in single mutants. Double homozygous cilia appear intact and normally distributed, with the exception of the ependymal cilia, where the severe ventricular dilatation observed in bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double mutants can cause extensive tissue damage. In addition, spermiogenesis is aborted in double homozygous males, and there is an absence of mature flagella on elongating spermatids and epididymal sperm. The absence of histopathological or ciliary phenotypes in any of the double heterozygous mice (nm1054/+ ;bgh/+ , nm1054/+ ;Cfap54 gt/gt /+ , and bgh/+ ;Cfap54 gt/gt /+) indicates that haploinsufficiency for two different CPA genes does not result in a ciliary phenotype. While this is expected given the autosomal recessive inheritance associated with these mutations, it demonstrates that a dose-dependent reduction in two different CPA proteins does not put a significant enough strain on the CPA to noticeably affect its function or the function of the motile cilium. The PCD phenotypes associated with the double homozygous mutants, however, are striking and generally more severe than single mutants. Most notable is the early mortality associated with double homozygotes on a mixed genetic background (Fig. 1). Homozygotes with either the nm1054, bgh, or Cfap54 gt/gt mutation exhibit early mortality on the B6 background due to severe hydrocephalus, presumably due to genetic modifiers segregating in that strain that influence susceptibility to severe hydrocephalus 26,27,29,30,38 . On the 129 background or a mixed background, single homozygotes do not develop severe hydrocephalus and live a normal life span, despite exhibiting airway ciliary phenotypes and male infertility. All three double homozygotes (nm1054/nm1054;bgh/bgh, nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt , and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt ) develop hydrocephalus on a mixed background (Fig. 2), suggesting that the defect in CSF flow in double homozygotes may be primarily due to severe ciliary dysfunction and highlighting genetic interactions. Given that several double homozygotes had to be euthanized in the early post-natal period due to severe hydrocephalus, the hydrocephalus is the most likely cause of the early mortality.
Despite the presence of motile cilia in the brain and airway of double homozygous mutants, there is an absence of mature flagella on elongating spermatids and epididymal spermatozoa from nm1054/ nm1054;Cfap54 gt/gt /Cfap54 gt/gt and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygous mutants (Figs. 5, 7). Reproductive data was not obtained for nm1054/nm1054;bgh/bgh double homozygotes due to mortality prior to sexual maturity. The difference between motile cilia and sperm flagella is consistent with single nm1054, bgh, and Cfap54 gt/gt homozygous mutants, as well as mutations in other axonemal genes that prevent proper flagellar formation without affecting ciliogenesis 4,26,27,29,39 . Presence of acetylated tubulin staining in developing spermatids www.nature.com/scientificreports/ from nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygotes suggests that the spermatids start to make axonemes, but the process aborts early in spermiogenesis (Fig. 6). In addition, DNAI1 staining in spermatids also indicates that cytoplasmic dynein pre-assembly is occurring in the double homozygotes. However, low staining without a clear staining pattern for CPA marker SPAG6, along with absence of detectable axonemal structures, suggest that CPA assembly may be perturbed. These data underscore fundamental differences in the mechanisms that drive and regulate motile ciliogenesis and sperm flagellar formation. Interestingly, the nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt double homozygous phenotype is noticeably less severe than the phenotypes of nm1054/nm1054;bgh/bgh and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygotes. The nm1054/nm1054;Cfap54 gt/gt /Cfap54 gt/gt double mutants exhibit a much higher survival rate (Fig. 1), show mild ventricular enlargement in the brain (Fig. 2, Supplementary Fig. S1), and display only modest amounts of mucus accumulation in the sinus cavity without extensive white blood cell infiltration (Fig. 3, Supplementary  Fig. S1). This mild phenotype may be due to CFAP221 and CFAP54 both associating with the C1d projection of the CPA (Fig. 1). If disrupting both proteins still largely only perturbs the single complex, the phenotype may not be substantially more severe than single nm1054 or Cfap54 gt/gt mutants. SPEF2, however, localizes to the C1b projection, so loss of SPEF2 and either CFAP221 or CFAP54 would likely disrupt two different protein complexes, put additional strain on the CPA, and have a more substantial effect on ciliary function. As a result, nm1054/nm1054;bgh/bgh and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt double homozygotes have more severe brain and airway phenotypes, further highlighting genetic interactions between SPEF2 and CFAP221 or CFAP54. These findings are consistent with previous studies identifying genetic interactions between RPGR and CEP290 16 and between SPAG6 and SPAG16L 21 , both due to particularly severe ciliary phenotypes.
Because of the early mortality associated with double homozygosity, combined with the low rate of double homozygote birth (1/16 of offspring from two double heterozygotes), mouse numbers for phenotypic analysis were limited and prevented detailed physiological analysis of ciliary motility. However, histopathological analyses demonstrate severe PCD-associated phenotypes and genetic interactions. Additional cell biological and biochemical studies are required to determine how these protein complexes regulate mammalian ciliary motility, as well as how the mechanisms regulating spermiogenesis differ from those driving motile ciliogenesis. As a result, these studies will enable a better understanding of the genetic and molecular mechanisms underlying PCD inheritance and pathogenesis.

Methods
Mice. The nm1054, bgh, and Cfap54 gt/gt mouse lines were maintained on the C57BL/6J (B6) and 129S6/ SvEvTac (129) backgrounds as previously described 26,27,29 . Each mutation results in loss of transcript or protein expression, and each line was genotyped as previously described 26,27,29 . Since previous studies of the single mutations on B6, 129, and mixed backgrounds indicated that the only phenotype influenced by background strain was hydrocephalus, which was more severe and resulted in early mortality on the B6 background 26,27,29 , all   www.nature.com/scientificreports/ nm1054/+ ;Cfap54 gt/gt /+ , and bgh/+ ;Cfap54 gt/gt /+) and double homozygous (nm1054/nm1054;bgh/bgh, nm1054/ nm1054;Cfap54 gt/gt /Cfap54 gt/gt , and bgh/bgh;Cfap54 gt/gt /Cfap54 gt/gt ) animals surviving to 8 weeks of age were used for all analyses at that age or later. Double homozygous animals dying prior to 8 weeks due to phenotype severity were used for histological and immunohistochemical analysis at the age of death when available. Wild type (WT) animals from the same generations were used as controls. Mouse numbers are detailed in Supplementary  Table S1. Due to early mortality associated with double homozygotes, it was often difficult to collect or analyze samples, but the pathological phenotype was still assessable in the available samples. Other than male infertility, no sex-specific differences have ever been observed for the phenotypes associated with nm1054, bgh, or Cfap54 gt/gt PCD [26][27][28][29][30] . Therefore, male and female mice were pooled for analyses in this study. All experiments involving animals were performed in accordance with the Animal Welfare Act and National Institutes of Health (NIH) policies and were approved by the Sanford Research Institutional Animal Care and Use Committee.
Histology. Heads were immersion fixed in Bouin's fixative until the bones were decalcified. Coronal slices were then cut through the maxillary sinuses and the lateral ventricles of the brain. Testes were immersion fixed in 10% buffered formalin. Fixed testes, brain slices, and sinus slices were embedded in paraffin, sectioned, and stained with Hematoxylin and Eosin (H&E) as previously described [28][29][30] . Stained tissue sections were analyzed by light microscopy using an upright Leica DM6000B microscope and Leica Aperio VERSA slide scanner. Spermatozoa were collected from the epididymis and diluted in phosphate buffered saline (PBS), following which they were spread onto slides, dried, fixed in methanol, and stained with the Camco differential stain kit (Cambridge Diagnostic Products, Inc.) as previously described 29 . Stained spermatozoa were analyzed by light microscopy on an Olympus IX71 inverted microscope. Because some double homozygous mice died or required euthanasia at earlier ages, thereby resulting in different tissue and organ sizes, severity of brain and sinus phenotypes were quantitatively compared through a scoring system. Severity of the brain phenotype was scored based on the following criteria: 1-no ventricular dilatation; 2-slight dilatation or opening of the ventricle with no visible damage to adjacent tissue; 3-extensive dilatation into the center of the hemisphere with no visible damage to adjacent tissue; 4-extensive dilatation into the center of the hemisphere with damage to adjacent tissue (loss of ependyma, loss of white matter, cortical thinning, and/or hemorrhaging). Severity of the sinus phenotype was scored based on the following criteria: 1-no visible mucus in the maxillary sinus cavity; 2-traces of mucus accumulating in the maxillary sinus cavity; 3extensive mucus accumulation filling large portions of the sinus cavity; 4-extensive mucus accumulation filling large portions of the sinus cavity with infiltration of neutrophils and/or red blood cells. Statistical significance was determined by one-way ANOVA.
immunohistochemistry. Testes, brain slices, and sinus slices were fixed and prepared as described above for histology. The fixed tissues were embedded in paraffin, sectioned, and stained using the BenchMark XT automated slide staining system (Ventana Medical Systems, Inc.) as previously described [28][29][30] . Primary antibodies included a mouse acetylated tubulin antibody (Sigma Aldrich T6793) at a 1:10,000 dilution, a mouse γ-tubulin antibody (Sigma Aldrich T6557) at a 1:500 dilution, a rabbit SPAG6 antibody (Invitrogen PA5-58389) at a 1:500 dilution, a rabbit RSPH4A antibody (Sigma Aldrich HPA031198) at a 1:100 dilution, and a rabbit DNAI1 antibody (Sigma Aldrich HPA021649) at a 1:300 dilution. Biotin SP-conjugated AffiniPure donkey antimouse (Jackson ImmunoResearch 715-065-151) and goat anti-rabbit (Jackson ImmunoResearch 111-065-144) secondary antibodies were both used at a 1:1,000 dilution. The slides were visualized by light microscopy using the upright Leica DM6000B microscope and Leica Aperio VERSA slide scanner, and high magnification images were acquired using the upright Nikon Eclipse Ni-E microscope with a 100 × objective and a Nikon DS-2MV camera.
The Aperio ImageScope software (v12.3.3.7014) was used for quantification of staining intensity on slidescanned images of sinus sections. Regions of interest were drawn around cilia in fields where the ciliated epithelial cells were clearly in focus. Positivity, which is determined by the number of positive pixels divided by the number of total pixels (positive staining plus negative counter-stain pixels) in a selected region, represents staining intensity and was calculated for each image using the Positive Pixel Count v9 analysis algorithm with default manufacturer input settings. Statistical analysis was performed using GraphPad Prism 6 (v6.02). Outliers were identified using the ROUT analysis (Q = 1%), and any rare data point that was an order of magnitude different from all other values within a given cohort was removed from analysis. Statistical significance was determined by one-way ANOVA. ciliary beat frequency (cBf) analysis. Tracheae were dissected into Medium 199, Hank's Balanced Salts and equilibrated at room temperature for 1 h, after which they were cut into rings using a Nikon SMZ1000 stereomicroscope. The rings were placed into fresh media and maintained at 28 °C using a thermal plate (Takai Hit TP-110RS05) for consistency throughout the experiment. CBF was analyzed on an Olympus IX71 inverted microscope using the Sisson-Ammons Video Analysis system as previously described 26,27,29,40 . Statistical significance was determined by one-way ANOVA.

Data availability
No large datasets were generated or analyzed during the current study. Representative data are included in this published article (and its Supplementary Information files), and all raw data is available from the corresponding author upon reasonable request. www.nature.com/scientificreports/