Randomization of Left-right Asymmetry and Congenital Heart Defects: The Role of DNAH5 in Humans and Mice

Abstract

Background

Nearly one in 100 live births presents with congenital heart defects (CHD). CHD are frequently associated with laterality defects, such as situs inversus totalis (SIT), a mirrored positioning of internal organs. Body laterality is established by a complex process: monocilia at the embryonic left-right organizer (LRO) facilitate both the generation and sensing of a leftward fluid flow. This induces the conserved left-sided Nodal signaling cascade to initiate asymmetric organogenesis. Primary ciliary dyskinesia (PCD) originates from dysfunction of motile cilia, causing symptoms such as chronic sinusitis, bronchiectasis and frequently SIT. The most frequently mutated gene in PCD, DNAH5 is associated with randomization of body asymmetry resulting in SIT in half of the patients; however, its relation to CHD occurrence in humans has not been investigated in detail so far.

Methods

We performed genotype / phenotype correlations in 132 PCD patients carrying disease-causing DNAH5 mutations, focusing on situs defects and CHD. Using high speed video microscopy-, immunofluorescence-, and in situ hybridization analyses, we investigated the initial steps of left-right axis establishment in embryos of a Dnah5 mutant mouse model.

Results

65.9% (87 / 132) of the PCD patients carrying disease-causing DNAH5 mutations had laterality defects: 88.5% (77 / 87) presented with SIT, 11.5% (10 / 87) presented with situs ambiguus; and 6.1% (8 / 132) presented with CHD. In Dnah5^mut/mut mice, embryonic LRO monocilia lack outer dynein arms resulting in immotile cilia, impaired flow at the LRO, and randomization of Nodal signaling with normal, reversed or bilateral expression of key molecules.

Conclusions

For the first time, we directly demonstrate the disease-mechanism of laterality defects linked to DNAH5 deficiency at the molecular level during embryogenesis. We highlight that mutations in DNAH5 are not only associated with classical randomization of left-right body asymmetry but also with severe laterality defects including CHD.

Introduction

Congenital heart defects (CHD) represent about 30% of major congenital anomalies in humans and account for nearly one-third of fetal deaths.

Cardiac development in vertebrates is strongly associated with the formation of body axes, especially the left-right body axis (Figure 1). Regular asymmetric organogenesis results in the characteristic arrangement of visceral organs, a situation called situs solitus (SS). Failure to establish the normal positioning of organs manifests in laterality defects, such as completely reversed arrangement of organs, called situs inversus totalis (SIT). In addition, the term situs ambiguus (SA) summarizes all other kinds of laterality defects such as left-, or right- isomerisms or situs inversus of only one body cavity (thoracalis versus abdominalis), and is also often associated with CHD. Heterotaxy is another term for SA but in accordance with Shapiro et al.⁵ we hereby refer to any laterality defect other than SIT as SA.

Figure 1: Cardiac development.

Initial steps of laterality and cardiac development take place during early embryonic stages. (A) A schematic of a mouse embryo at Theiler Stage (TS) 11 (embryonic day 7.5–8.5) demonstrates important structures for left-right body asymmetry development: mesoderm derived progenitor cells of the primary heart field are found posteriorly to the developing head folds. By fusion, this heart field forms the cardiac crescent¹ (indicated in dark red). Posteriorly and adjacent to the cardiac crescent, the secondary heart field is found (indicated in dark orange). Concurrently, the Nodal signaling cascade (indicated by dark blue shadow) transduces initial signals of asymmetry established at the left-right organizer (LRO; indicated in green) throughout the lateral plate mesoderm (LPM). (B) Schematic transverse section of mouse LRO indicates that every cell of the LRO carries a single monocilium. By rotational movement the motile monocilia of pit cells (colored in blue) generate the leftward nodal flow (arrow). This leftward flow is sensed by monociliated LRO crown cells (colored in green) and is required for induction of (C) Nodal signaling; asymmetric Dand5 and Nodal expression at the LRO-surrounding region influences asymmetric expression of factors of the Nodal signaling cascade (Nodal, Lefty2, Pitx2) that propagates through the left but not the right LPM and facilitates asymmetric organogenesis. (D) The asymmetric cardiac development begins at TS12 and in contrast to transient Nodal expression², the expression of Pitx2 continues to and beyond this stage and is limited to cells derived from left-sided progenitor cells³. The primary heart tube is formed at the midline⁴ by fusion of the cardiac crescent. (E) The heart then begins to beat and undergoes a rightward looping, the first obvious sign of lateral body asymmetry⁴. (F) During further development the inflow tract gives rise to the early common atrium (indicated in light blue and pink). The common atrium moves cranially and differentiates into the right (RA) and left atrium (LA). The outflow tract forms the efferent vessels, and by further folding of the heart tube and formation of the septum, the left (LV) and the right ventricle (RV) separate. (G) Finally, this process results in the fully developed four-chambered heart. In the course of organogenesis, various additional known asymmetries arise, such as those of the lungs, liver and spleen.

Interestingly, CHD and other laterality defects are described in different ciliopathies such as oral-facial-digital syndrome (MIM311200), nephronophthisis (MIM604387), Meckel syndrome (MIM249000), or Senior-Loken-syndrome (MIM266900)^6,7. Primary ciliary dyskinesia (PCD) (MIM608644) is another classical example of genetically inherited disorders linking both cilia and laterality defects.

In the healthy body, multiple motile cilia line the airways and facilitate mucociliary clearance. In PCD, this important function is dysfunctional due to impaired ciliary beating. Thus, common symptoms in PCD comprise infant acute respiratory distress syndrome, chronic otitis media, recurrent and / or chronic bronchitis, sinusitis and chronic wet cough that often cause bronchiectasis. Of note, PCD patients also manifest dysfunction of embryonic left-right organizer (LRO) cilia function that leads to randomization of left-right body asymmetry. Thus, approximately half of these PCD patients present with SIT⁸. In addition to SIT, the prevalence of CHD in PCD is 6–16%^5,8,9.

To date, more than 40 genes are known to be associated with PCD⁹ and related motile ciliopathies. The most frequently mutated gene in PCD is DNAH5^9–11, which encodes a component of the ciliary outer dynein arms (ODAs) and is essential for force generation during ciliary beating¹⁰. Classically, PCD caused by DNAH5 mutations is associated either with SS or SIT¹⁰ but SA and CHD have also been reported¹². The prevalence of CHD has been investigated thus far in general PCD collectives irrespective of the individually mutated gene^5,8,9 and DNAH5 mutations in only a small number of related individuals¹². Studies on the prevalence of laterality defects in larger cohorts of patients with specifically defined DNAH5 mutations are currently lacking. It is assumed that laterality defects in DNAH5-mutant patients result from immotility of motile monocilia at the LRO¹⁰, a transient embryonic structure essential for proper left-right patterning (Figure 1). However, this hypothesis has not been directly proven, leaving the exact patho-mechanism by which DNAH5 mutations affect the establishment of body laterality unresolved. Here, we performed genotype / phenotype correlations to determine the prevalence of SI, SA and CHD associated with DNAH5 mutations in a cohort of 132 patients with a defined PCD diagnosis. We further analyzed structure and function of LRO motile monocilia and the Nodal signaling cascade in Dnah5-mutant mice to determine the mechanism underlying formation of CHD associated with DNAH5 deficiency. This provides further understanding regarding the disease mechanism underlying laterality defects and CHD that is likely applicable also to human DNAH5-mutant patients.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request. Full material and methods are available in the supplemental data.

Human subjects

Written informed consent was obtained from all participants using protocols approved by the Institutional Ethics Review board of the University of Muenster (Muenster, Germany) and collaborating institutions.

Animal experiments

Animal experiments complied with ethical regulations and were approved by local government authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, Germany (AZ 84–02.05.20.12.164, AZ 84–02.05.20.12.163, and AZ 84–02.05. 5.15.012)).

Results

Study cohort and genotype / phenotype correlations

The final study cohort of 132 PCD patients representing 109 families (13.4% of the initial collective) consisted of 58.3% (77 / 132) female and 41.7% (55 / 132) male patients. 44.7% (59 / 132) of the patients carried homozygous DNAH5 mutations and 55.3% (73 / 132) carried putative compound heterozygous DNAH5 mutations (for selection criteria please refer to supplemental figure 1). Due to limited availability of parental DNA, we performed segregation analyses and confirmed homozygosity or compound heterozygosity of mutations in 44.1% (26 / 59) and 46.6% (34 / 73) of patients, respectively. Overall, we identified 101 distinct mutations in DNAH5. 11.9% (12 / 101) affected splice regions, of which eleven were obligatory and one was a facultative splice site mutation (c.13338+5G>A) previously reported to be disease-causing¹¹. Furthermore, 63.4% (64/101) of mutations were predicted to generate an early stop codon either as nonsense (48.4% [31/64]) or frameshift mutations (51.6% [33/64]). Missense mutations were identified in 22.8% (23/101). The remaining 2.0% (2 / 101) were in-frame deletions of a single amino acid, affecting highly conserved amino acid residues and predicted to be disease-causing or deleterious by MutationTaster2¹³ and PROVEAN¹⁴, respectively.

Mutations in DNAH5 have been previously reported to cluster within some exons¹¹. In our cohort, the location of these mutations was distributed over the entire DNAH5 gene (supplemental table 1, supplemental figure 2). Of note, we identified five exons that were affected in at least eight patients, namely exons 34, 49, 63, 76 and 77 (summarized in supplemental figure 2); exon 63 was the most frequently mutated exon in 28.0% (37 / 132) of patients.

Unrelated families shared common mutations. For example, we identified the exon 63 variant c.10815delT (p.Pro3606Hisfs*23) in 26 families, which has been reported as an ancient founder mutation¹¹. In seven families we identified the mutation c.5563dupA (p.Ile1855Asnfs*6). Five families were associated with the variants c.10615C>T (p.Arg3539Cys), c.13194_13197delCAGA (p.Asp4398Glufs*16) and c.13486C>T (p.Arg4496*). Four families each carried the mutations c.8440_8447delGAACCAAA (p.Glu2814*) and c.13458_13459insT (p.Asn4487*) (further mutations were identified in more than one family; supplemental table 1 summarizes all mutations and the frequency of their occurrence).

Diagnostic findings in the 132 DNAH5-mutant patients were as follows (see also supplemental figure 3): Transmission electron microscopy (TEM) analyses were performed for 57.6% (76 / 132) and confirmed ODA defects in 93.4% (71 / 76) of patients. We performed immunofluorescence (IF) analyses to investigate DNAH5 localization in 76.5% (101 / 132) and confirmed abnormal DNAH5 protein distribution in 92.1% (93 / 101) of these patients. 6.6% (5 / 76) of TEM results and 7.9% (8 / 101) of IF results were inconclusive due to sample quality. Overall, in 84.8% (112 / 132) of patients, the absence of ODAs was proven by TEM (19 / 132), IF (41 / 132), or both (52 / 132), in accordance with the identified mutations in DNAH5. Additionally, respiratory ciliary beating was evaluated by light microscopy in 56.8% (75 / 132) of patients and all (75 / 75) showed an abnormal ciliary beating pattern, as expected for an ODA defect. Typically, compared to healthy control (supplemental movie S1) cilia in DNAH5-mutant patients showed only residual flickery movement or were immotile (supplemental movie S2).

In this well characterized cohort, we investigated the prevalence of laterality defects and CHD (Figure 2A-C). 34.1% (45 / 132) of the patients presented with SS, and 65.9% (87 / 132) presented with a laterality defect. Among the latter, 88.5% (77 / 87) presented SIT and 11.5% (10 / 87) with SA, including eight patients displaying a variety of congenial cardiac malformations (Table 1). CHD were associated with other laterality defects in 62.5% (5 / 8) of patients. Taken together, SA was observed in 7.6% (10 / 132) of patients; 80.0% (8 / 10) of these SA patients suffered from different kinds of congenital cardiac malformations (Table 1). Of note, only 73.5% (97 / 132) of the cohort were assessed previously for cardiac malformations. This indicates that surveys of the heart phenotype are not included in routine PCD diagnostic setups and CHD may be an overlooked or underreported feature of PCD.

Figure 2:

(A) Distribution of situs manifestations in a cohort of 132 PCD patients carrying disease-causing DNAH5 mutations. (B) Chest- X-ray of siblings OP-1501 II1 and II2 carrying disease-causing DNAH5 mutations (homozygous c.2710G>T, p.Glu904*) demonstrate randomization of left-right body axis indicated by presence of (top) situs solitus as well as (bottom) situs inversus, respectively. White arrows point towards the direction of the heart apex. R and L indicate the right and left body side. (C) Positions of mutations in DNAH5 and predicted changes of the DNAH5 protein in PCD patients with CHD (for more details on all mutations in this study please refer to supplemental table 1). (D) DNAH5 deficiency results in loss of ciliary outer dynein arms (ODAs). (I) A schematic cross section of a cilium. ODAs are attached to the peripheral microtubules. The large multiprotein complexes are preassembled in the cytoplasm and transported to the ciliary axoneme. Lack or deficiency of one of the components such as DNAH5 or DNAI1 results in the loss of ODAs from the ciliary axoneme. (II) Transmission electron microscopy analyses show that in human heathy control cells, ODAs are clearly visible (exemplarily marked by arrowhead), whereas DNAH5 deficient cilia of OP-1501 II1 completely lack ODAs. (III-IV) Immunofluorescence analyses using an antibody against acetylated α-tubulin as a ciliary marker (indicated in green) (III) shows a panaxonemal DNAH5 localization (red, colocalization with tubulin indicated by yellow color in merged image) in a healthy control (IV) but the absence of panaxonemal DNAH5 in a patient carrying biallelic DNAH5 mutations. (V) Likewise, DNAI1 (red) is shown to localize to cilia in healthy control cells (indicated by yellow color in merged image), (VI) but is absent from cilia in human DNAH5-mutant patients, as represented by OP-1501 II1.

Table 1:

Situs and cardiac features of PCD patients carrying DNAH5 mutations suffering from SA

ID	Age (years)	Gender	overall situs*	congenital cardiac defects
				septal defects	valve defects	vessel defects
CHD-1	16	f	SS	ASD	-	-
CHD-2	3	m	SS	ASD	-	-
CHD-3	22	f	SS	ASD	-	-
CHD-4	26	m	SI	VSD	subpulmonary stenosis	DORV
CHD-5	12	f	SI	VSD	-	-
CHD-6	n/a	m	SI	VSD	-	-
CHD-7	17	m	SI	ASD, VSD	pulmonary atresia	transposition of great vessels
CHD-8	35	f	SI thoracalis	AVSD (TOF)	subpulmonary stenosis (TOF), insufficiency of the shared AV valve	overriding aorta (TOF)
PS-1	n/a	f	SS + isolated polysplenia	-	-	-
PS-2	n/a	f	SI + isolated polysplenia	-	-	-

ASD: atrial septal defect; AV: atrial ventricular; AVSD: atrial ventricular septal defect; DORV: double outlet right ventricle; f: female; m: male; n/a: data not available; PS: polysplenia SI: situs inversus; SS: situs solitus; TOF: tetralogy of Fallot; VSD: ventricular septal defect

^*In general, we define situs ambiguus as any laterality defect other that SIT. In this specific table under the term “overall situs” we indicate the positioning of the majority of organs.

Interestingly, in some families there were reports of other family members suffering from heart disease. Four individuals from three families died of heart defects as children or young adults, one of those in connection with SIT and another one in connection with respiratory symptoms. Unfortunately, retrospectively we could not analyze these persons for suspected DNAH5 mutations due to lack of DNA samples. However, they likely carried biallelic DNAH5 mutations.

We investigated whether there are correlations between the genotype and particular phenotypes such as SIT or SA. The occurrence of SIT and SA was independent of the type (stop / missense mutation) or position of the mutation in the DNAH5 gene (Figure 2C). We did not observe any hereditary accumulation of CHD, and all CHD patients belong to unrelated families. Taken these findings together, we conclude that there is no association between genotype and a particular situs phenotype in DNAH5-mutant patients. Just by chance, the affected individuals present with SS, SIT or SA.

Left-right organizer pit monocilia are immotile in Dnah5^mut/mut mouse embryos

CHD, SA and other situs defects are well known to be associated with defective determination of the left-right body axis, a process that involves motile LRO pit monocilia. Because these processes cannot be investigated in human embryos directly, we utilized mouse embryos, which represent the best characterized mammalian animal model for laterality development; mutations in orthologous genes often have a similar outcome regarding situs defects.

The protein DNAH5 is an axonemal component of respiratory cilia¹⁵, and lack of DNAH5 results in their dysmotility¹⁰. Because human DNAH5¹⁰ and mouse Dnah5¹⁶ mutants both display laterality defects, we hypothesized that DNAH5 is also a component of motile LRO pit monocilia and is essential for their beating. By high speed video microscopy, we observed beating of pit monocilia in a large portion of cells in the LRO cavity in control mouse embryos. By contrast, we did not observe any properly moving motile cilia in Dnah5^mut/mut embryos (supplemental movies S3-S6, Figure 3, supplemental table 2 and supplemental figure 4). The mean active area was 0.59±0.48% (wildtype) and 0.69±0.52% (heterozygous mutant) of the field of view compared to 0.08±0.14% in the homozygous mutant mouse embryos. Cilia in control mouse embryos exhibited the typical rotatory beating pattern with a median beat frequency of 5.3Hz in wildtype and 6.5Hz in Dnah5^wt/mut embryos (Figure 3I, supplemental movies S3-S6, supplemental table 2 and supplemental figure 4). By contrast, the median beat frequency was highly significantly (p>0,001; Kruskal-Wallis test) reduced to 0Hz in Dnah5^mut/mut embryos (Figure 3I). We have also found a significant difference in the beat frequency between wildtype and Dnah5^wt/mut embryos. However, we assume that this is possibly an effect of the small number of wildtype embryos analyzed. These findings show that DNAH5 is indeed essential for beating of LRO monocilia and thus for generation of leftward directed flow.

Figure 3: Monocilia of the left-right organizer pit are immotile in Dnah5^mut/mut mouse embryos.

(See also supplemental movies S3-S6) (A) Light microscopic image of the LRO cavity of an 8.25 dpc Dnah5^wt/mut control embryo. (B) The activity map shows ciliary movement colored in dark grey. (C) In the magnification, some LRO monocilia are marked exemplarily with colors to be better visualized. Rotatory movement is indicated by black arrows. (D) Time series of frames indicates fast clockwise rotation of LRO pit monocilia in the control embryo. (E) Light microscopic image of a Dnah5^mut/mut embryo’s LRO cavity. (F) No ciliary movement of LRO monocilia is detected as shown in the activity map or (G) in the magnification of the Dnah5^mut/mut LRO, where some cilia are exemplarily colored. (H) Time series frames of the homozygous mutant embryo show that LRO monocilia do not move. (I) Ciliary beat frequencies in all analyzed wildtype (n=2), heterozygous (n=7) and homozygous (n=5) Dnah5 mutant embryos. Boxes in diagram represent the median, the 25%, and the 75% percentile. Outliners are marked by circles or diamonds. Statistical significance was verified using the Kruskal-Wallis test. ROI: region of interest, CBF: ciliary beat frequency, ***: p-value lower than 0.001

Left-right organizer pit monocilia in Dnah5^mut/mut mice have a defective outer dynein arm composition

Based on the fact that DNAH5-deficient multiple motile cilia largely lack ODAs¹⁰ (Figure 2D) and that mouse LRO pit monocilia lacking DNAH5 protein are immotile, we inferred that monocilia of Dnah5^mut/mut mouse LROs lack proper ODA composition as well.

Unfortunately, direct analyses of DNAH5 localization in LRO pit monocilia of mouse embryos were not possible due to lack of cross-species specificity of the available anti-DNAH5 antibody directed against human DNAH5 protein.

To overcome this technical limitation, we used an alternative strategy. We have reported that in multiple motile cilia, the loss of one ODA component results in the lack also of other ODA components^10,15 (Figure 4D). We selected an antibody detecting the ODA intermediate chain DNAI1 to investigate the ODA status in mouse LRO monocilia. Specificity of this antibody was shown by the fact that it reacts with axonemes in wildtype human and mouse respiratory cells but not axonemes in respiratory cells of DNAH5-mutant human individuals (Figure 2D) or Dnah5^mut/mut mice (Figure 4A).

Figure 4: Absence of outer dynein arm component DNAI1 from left-right organizer monocilia in Dnah5^mut/mut mice.

(A) (upper panel) In respiratory cells of control mice the outer dynein arms (ODA) protein DNAI1 (shown in red) co-localizes with the ciliary marker acetylated α-tubulin (marked in green), as indicated by the yellow color in the merged image. (lower panel) Consistent with defective ODA composition, the cilia in Dnah5^mut/mut mice lack DNAI1. (B) (upper panel) In control mouse embryos at 8.25 dpc, DNAI1 localizes to LRO monocilia (lower panel) but it is undetectable in Dnah5^mut/mut embryos LRO monocilia. Scale bars represent 5µm. a: anterior, p: posterior, l: left, r: right

In 94% (17 / 18) of control mouse embryos, the anti-DNAI1 antibody specifically recognized LRO pit monocilia (Figure 4B, supplemental figure 5) showing panaxonemal localization of DNAI1. In only one embryo (6% (1 / 18) no signal for DNAI1 was observed, very likely because of the inadequate developmental stage for this analysis (7 somites, several hours after the ciliary action generates leftward flow). In 100% (8 / 8) of Dnah5^mut/mut mouse embryos, DNAI1 was not detected in LRO monocilia (Figure 4B, supplemental figure 5), consistent with a defective ODA composition. In addition, we have reported that Dnah5 is expressed at the LRO in mouse embryos¹⁰. Hence we conclude that DNAH5, similar to respiratory and other cells with multiple motile cilia, is a component of ciliary ODAs in LRO cells and that its deficiency results in an absence of ODAs and loss of ciliary motility in Dnah5^mut/mut mouse embryos.

Nodal signaling is altered in Dnah5^mut/mut mouse embryos

The Nodal signaling cascade (Figure 1) is crucial for the proper development of the left-right body axis. Asymmetric, sinistral expression of Nodal and other components of the cascade represent early signs of lateralization.

To gain further insights into the effects of DNAH5 deficiency on this key process of laterality establishment, we performed whole mount in situ hybridization (ISH) analyses of mouse embryos during this critical period of time (Theiler Stage 11–12). We investigated the following four key components of this signaling pathway: Nodal, showing an early expression at the LRO-surrounding region that becomes enhanced specifically at the left side and later propagates through the left lateral plate mesoderm (LPM); Lefty1 and Lefty2, expressed at the midline and within the left LPM, respectively, that both maintain a feedback suppression of the Nodal signaling; and Pitx2, first expressed in the left LPM, but during further development its expression persists and gives raise to asymmetric organogenesis¹⁷ (processes of left-right axis development are shown in Figure 1).

Nodal expression in crown cells of the LRO was present in the majority of mouse embryos analyzed (91% (60 / 66)) (Figure 5A). In 14 wildtype and 30 Dnah5^wt/mut controls, Nodal expression was either equally distributed to both sides (21% (3 / 14) and 30% (9 / 30), respectively) or enriched at the left side (71% (10 / 14) and 53% (16 / 30), respectively). Only in rare cases (7.1% (1 / 14) and 6.7% (2 / 30), respectively) was Nodal signal stronger on the right rather than on the left side. Consistent with a defective left-right axis development, the Dnah5^mut/mut embryos clearly showed randomized enrichment of Nodal expression (Figure 5A): 18% (4 / 22) showed bilaterally equal amount, 32% (7 / 22) showed enriched expression at the left side, 36% (8 / 22) showed enriched expression at the right side of the LRO, and 13.6% (3 / 22) showed no expression of Nodal at all.

Figure 5: Randomized expression pattern of genes of the Nodal signaling pathway in Dnah5^mut/mut mice analyzed by whole mount in situ hybridization.

(A) In 8.25 dpc control mouse embryos, Nodal expression was restricted to two regions: 1. to crown cells of the left-right organizer (LRO) and 2. to the left lateral plate mesoderm (LPM). Nodal expression level on the left side of the LRO was mostly equal (27% (12 / 44)) or higher than at the right side (59% (26 / 44)). In Dnah5^mut/mut mice, the expression pattern of Nodal was randomized in LRO crown cells; abnormal right (4% (3 / 22)) or bilateral expression (0.5% (1 / 22)) within the lateral mesoderm was observed. (B) Expression pattern of Lefty genes in 8.25 dpc embryos: in wildtype and heterozygous embryos, Lefty1 expression was detected at the midline and Lefty2 expression was observed in the left LPM (61% (11 / 18)). In Dnah5^mut/mut embryos, the expression of Lefty2 was randomized, as we observed expression in the right LPM (16% (3 / 19)) as well as bilateral expression (16% (3 / 19)). (C) Pitx2 expression was observed within the left LPM in wildtype and heterozygous 8.25 dpc embryos (100% (23 / 23)). In Dnah5^mut/mut embryos, the expression pattern of Pitx2 was randomized. We counted embryos with expression in the left (25% (4 / 16)) as well as in the right plate mesoderm (6% (1 / 16)); half of the embryos presented bilateral expression (50% 8 / 16)) and some lacked Pitx2 expression in the lateral mesoderm (19% (3 / 16)). Ctrl: control, n: number of analyzed embryos for the respective genotype, wt: wildtype (Dnah5^wt/wt), het: heterozygous (Dnah5^wt/mut), Dnah5^mut/mut: homozygous Dnah5 mutant

As the initial Nodal signal from the LRO is subsequently transduced to the LPM, we evaluated Nodal expression also in this critical region. In control mouse embryos, Nodal expression in the LPM was either still absent (55% (24 / 44)) or exclusively at the left side of the embryo (45% (20 / 44)) (Figure 5A). In Dnah5^mut/mut embryos, Nodal expression was observed in the left (4.5% (1 / 22)) or the right LPM (13.6% (3 / 22)) as well as bilaterally (4.5% (1 / 22)). 77% (17 / 22) of Dnah5^mut/mut embryos showed no Nodal LPM expression (Figure 5A). However, at adequate developmental stages for Nodal LPM activity (four to six somite stage), 83.3% (15 / 18) of control embryos showed this expression, in contrast to only 45% (5 / 11) of Dnah5^mut/mut mutants. This finding may indicate a delayed induction of the signaling cascade (supplemental figure 6).

Next we investigated the expression of Lefty1, a repressor of Nodal maintaining a midline barrier for the signaling cascade, combined with the expression of Lefty2, likewise a Nodal repressor but expressed in the LPM. We observed the proposed expression pattern of Lefty1 at the midline (78% (14 / 18)), and Lefty2 exclusively in the left LPM (61% (11 / 18)) in 18 wildtype and heterozygous control mouse embryos (Figure 5B). In 19 analyzed Dnah5^mut/mut mouse embryos, the expression pattern of Lefty1 was comparable to control embryos, showing Lefty1 midline expression (68% (13 / 19)) (Figure 5B and supplemental figure 7). However, the expression pattern of Lefty2 in the LPM indicated randomized induction of the Nodal cascade in Dnah5^mut/mut embryos: Lefty2 expression was detected in 42% (8 / 19) at the left, 16% (3 / 19) at the right side, and 16% (3 / 19) bilaterally (Figure 5B). The latter two expression patterns were not observed for any of the 18 control embryos. No expression of Lefty2 in the LPM was observed in 39% (7 / 18) of control and 26% (5 / 19) of Dnah5^wt/mut mouse embryos.

Another key player of the Nodal signaling cascade is the transcription factor PITX2. Pitx2 expression is induced by NODAL and modified by LEFTY2, and ultimately shapes asymmetric organogenesis¹⁷. Similar to Nodal and Lefty2, Pitx2 expression was restricted exclusively (100%) to the left LPM in 23 wildtype and heterozygous control mouse embryos (Figure 5C). By contrast, Pitx2 expression was randomized in Dnah5^mut/mut mouse embryos: in the left (25% (4 / 16)), or in the right (6% (1 / 16)) LMP, but also bilateral (50% (8 / 16) or absent (19% (3 / 16)) (Figure 5C).

These results provide evidence that the proper expression of Nodal signaling factors in mouse embryos depends on the function of DNAH5, proving that DNAH5 works upstream of this cascade. Remarkably, early steps of this signaling cascade were laterally randomized but mostly unilateral in Dnah5^mut/mut embryos, whereas the later developmental steps, represented by Pitx2 expression, frequently occurred in a bilateral fashion.

Discussion

Since its identification almost twenty years ago as a cause of PCD and SIT¹⁸, DNAH5 mutations have been linked to other laterality defects such as CHD. Although previous studies have revealed a prevalence of 6–16% for SA and / or CHD in general PCD cohorts^5,8,9, thus far no large-scale studies have reported the occurrence of such defects, in particular related to DNAH5 mutations.

To determine the prevalence of laterality defects, we analyzed a cohort of 132 DNAH5-mutant PCD patients and identified 58.3% with SIT and 34.1% with SS. The remaining 7.6% presented with SA, with 6.1% of these showing CHD; this agrees with the reported prevalence of 6–16% for SA / CHD in general PCD cohorts^5,8,9.

Although DNAH5 mutations have been linked to laterality defects¹⁸, the underlying disease-mechanism has not been investigated directly. Instead, an unproven hypothesis has been widely accepted for years: Based on the observations that a) DNAH5-deficient multiple motile cilia are dysmotile due to defective ODA composition^10,16,19 and b) Dnah5 is expressed exclusively at the mouse LRO during early embryogenesis¹⁰, it has been speculated that laterality defects linked to DNAH5 mutations result from ODA-defective and consequently immotile LRO monocilia^10,19.

Rotatory beating (motility) of LRO monocilia is needed to generate the leftward nodal flow²⁰, which in turn is essential for the proper induction of the Nodal signaling cascade (Figure 1).

Indeed, by high speed video microscopy, we demonstrate here that LRO pit monocilia in Dnah5^mut/mut mouse embryos are immotile as previously hypothesized. In addition, by IF analyses of DNAI1 localization, a critical interaction partner of DNAH5, we also showed that LRO cilia in Dnah5^mut/mut mouse embryos lack normal ODAs. This directly proves that DNAH5 is essential for the correct ODA composition and motility of LRO monocilia. Because a) mice are a well-established and -characterized animal model for the development of the left-right body axis and b) the phenotypic outcome of mutations in laterality associated genes are comparable between mouse and man, we assume that the same pathomechanism underlies laterality defects in human DNAH5-deficency: absence of DNAH5 results in a lack of ODAs in LRO monocilia, causing immotility in this special type of motile cilia.

Immotility of LRO cilia results in the absence of leftward nodal flow²⁰. Consequently, the evolutionary conserved Nodal signaling cascade is induced only by chance on one side of the LRO or the other. By its self-reinforcing and –modulating characteristics²¹, the signaling cascade propagates through this particular side of the embryo. If the initial asymmetric expression pattern manifests sufficiently early and stable, it ultimately drives development of either SS or SIT²¹.

In our study, we demonstrate the general disease mechanism by which laterality defects occur due to DNAH5 deficiency. Because the signaling factor gene Pitx2 is expressed later during cardiogenesis³, other disease mechanisms independent from initial determination of left-right asymmetry might also play a role in the occurrence of CHD in PCD patients.

Similar to other SIT-associated mutations, DNAH5 mutations in humans are classically described as causing randomization of body laterality, with approximately half of the affected presenting SIT and the other half with SS¹⁸. In agreement, we showed that in 62.5% of Dnah5^mut/mut mouse embryos, which generally express key factors of the Nodal signaling cascade in the LMP, this expression occurs randomly either in the left or right LPM, resulting in either SS or SIT. Consistent with this observation, the vast majority (92.4%) of DNAH5-mutant patients presented with either SS or SIT.

Randomization of body laterality in DNAH5-deficiency can also be attributed to the lack of nodal flow, similar to the Dnah11-mutant iv/iv-mouse model (Dnah11 encodes another ODA heavy chain) displaying laterality defects, immotile LRO monocilia, and lack of nodal flow²⁰. We assume that immotile LRO monocilia lacking DNAH5 do not generate this flow, since it has been shown that a weak flow generated by as little as two motile cilia is sufficient to induce normal left-right signaling at the LRO²², which we did not observe in Dnah5^mut/mut mouse embryos. In addition, our results substantiate the role of DNAH5 specifically for LRO pit monocilia motility rather than for sensory cilia function (e.g. in LRO crown cells) or other elements of the complex processes at the LRO: If sensory cilia function or signal transduction were impaired, components of the Nodal signaling cascade would most probably be expressed always bilaterally, delayed, or not at all²³, resulting in a high prevalence of isomeric defects²⁴.

We observed SA in 7.6% of DNAH5-mutant patients, including 6.1% with CHD and 1.5% presenting with polysplenia. These phenotypes may arise from a delayed or reduced initial strength of the first Nodal cascade signal, due to lack of nodal flow. Hence, as also observed in DNAH11-deficient iv/iv mice²⁰, the signaling cascade may be delayed and / or its self-restricting function via Lefty1/Lefty2 (on the opposite side of the midline) could be reduced²¹. This would result in bilateral Nodal signaling and ultimately the occurrence of SA. Indeed, bilateral LPM expression pattern was observed occasionally for Nodal and Lefty2 and frequently for Pitx2, explaining the prevalence of 7.6% SA in patients carrying balletic DNAH5 mutations.

In mice, Dnah5 deficiency has been associated with SA and CHD as well: prenatally (16.5–18.5 dpc), 40% of Dnahc5^{del593/del593} offspring display SA and / or heart defects¹⁹. Interestingly, cardiac defects in these mice comprise a spectrum of defects including double outlet right ventricle malformations, tetralogy of Fallot anomalies, and septal defects (ventricular septal defect and / or atrial septal defect¹⁹), phenotypes that we also report here in human DNAH5-mutant PCD patients. Therefore, the outcome of DNAH5 deficiency appears comparable between human and mouse.

However, the prevalence of SA in DNAH5-mutant patients (7.6%) is much lower than the prevalence in prenatal mice (40%)¹⁹, raising the hypothesis that the prevalence of SA / CHD is probably higher in human but this value is obscured by the increased risk of pre- or perinatal mortality. Indeed, only Dnahc5^{del593/del593} mouse offspring with SS or SIT but not SA / CHD survive postnatally¹⁹. However, it is still possible that the huge discrepancy between the prevalence of SA in our human cohort and the Dnahc5^{del593/del593} mouse offspring is a result of the specific mouse mutation causing a more severe cardiac phenotype. There is evidence for lethality of cardiac malformations in families associated with PCD: a study reported that two of seven (28.6%) family members with PCD died due to CHD¹². In our study cohort, 2.3% (3 / 132) of families reported additional family members who were deceased due to cardiac disease, in effect increasing the proportion of CHD-related families to 10.1% (11 / 109). Taken together, these data support our hypothesis that DNAH5 mutations are associated with lethal CHD in humans. In summary, our findings confirm that human DNAH5 mutations are not only associated with classical randomization of left-right body asymmetry (SS or SIT) but also SA / CHD. Therefore, DNAH5 should be considered as a heterotaxy gene in humans. Of note, PCD has symptoms like hypoxemia that overlap with CHD, and there is evidence for lethality of cardiac malformations in families associated with PCD¹². Thus, we highly recommend a thorough analysis of family history and cardiac examinations as part of routine diagnostic workup in PCD patients.

Vice versa, the prevalence of ciliary dysfunction / PCD is well above average (41.8%) in human SA and CHD study cohorts²⁵, and mouse models for CHD are strongly associated with defective ciliary function (56%)²⁶. Our data confirm and extend previous findings that a genetic ciliopathy may be the underlying cause for SA or CHD. CHD are at the forefront of clinical symptoms and therapeutic interventions; therefore we assume that in a proportion of patients with CHD (especially those with general SS), PCD and lung involvement are overlooked. We therefore highly recommend paying special attention to examination of ciliopathy-related symptoms when diagnosing CHD with or without laterality defects.

Non-standard Abbreviations and Acronyms

CHD

Congenital Heart Defects

dpc

days post coitum

IF

Immunofluorescence

ISH

In situ Hybridization

LPM

Lateral Plate Mesoderm

LRO

Left-Right Organizer

nt

nucleotide

ODA

Outer Dynein Arm

PCD

Primary Ciliary Dyskinesia

ROI

Region Of Interest

SA

Situs Ambiguus

SIT

Situs Inversus Totalis

SS

Situs Solitus

TEM

Transmission Electron Microscopy

TS

Theiler Stage

WES

Whole Exome Sequencing

*PCD study group

Israel Amirav² (MD), Luisa Biebach¹ (MD), Dorit Fabricius³ (MD), Matthias Griese⁴ (MD), Jörg Große-Onnebrink¹ (MD), Karsten Häffner⁵ (MD), Andreas Hector⁶ (MD), Andreas Jung⁷ (MD), Petra Kaiser-Labusch⁸ (MD), Thomas Kaiser¹ (MD), Christina Keßler¹ (MD), Richard Kitz⁹ (MD), Michael R. Knowles¹⁰ (MD), Cordula Koerner-Rettberg¹¹ (MD), Ulf Kristoffersson¹² (MD, PhD), Margaret W. Leigh¹³ (MD), Pontus Mertsch¹⁴ (MD), Bernhard Mischo¹⁵ (MD), Kim G. Nielsen¹⁶ (MD), Marco Poeta¹⁷ (MD), Ernst Rietschel¹⁸ (MD), Samra Roth¹⁹ (MD), Francesca Santamaria¹⁷ (MD), Christian Schmalstieg¹ (MD), Miriam Schmidts²⁰ (MD), Carsten Schwarz²¹ (MD), Nicolaus Schwerk²² (MD), Horst Seithe²³ (MD), Johannes Tebbe¹ (MD), Claudius Werner¹ (MD), Maimoona A. Zariwala²⁴ (PhD).

¹Department of General Pediatrics, University Children’s Hospital Muenster, Muenster, Germany

²Pediatric Pulmonology Unit, Dana-Dwek Children’s Hospital, Tel-Aviv, Israel, for the Israeli PCD Consortium

³Department of Pediatrics, University Hospital Ulm, Ulm, Germany

⁴Department of Pediatrics, Ludwig Maximilian University Munich, Dr. von Hauner Children’s Hospital, Munich, Germany

⁵Department of General Pediatrics, Adolescent Medicine and Neonatology, University of Freiburg, Faculty of Medicine Medical Center - University Freiburg, University of Freiburg, Freiburg, Germany

⁶Children’s Hospital of the University of Tübingen, Pediatric Infectiology, Immunology & Cystic Fibrosis, Hoppe-Seyler-Str. 1, Tübingen, Germany

⁷Division of Respiratory Medicine, University Children’s Hospital, 8032 Zürich, Switzerland

⁸Professor-Hess-Kinderklinik, Bremen, Germany

⁹Department of Pediatric Pneumology, Clementinen Children’s Hospital, Frankfurt am Main, Germany

¹⁰Department of Medicine and Marsico Lung Institute, University of North Carolina School of Medicine, Chapel Hill, North Carolina

¹¹Department of Paediatric Pneumology, University Children’s Hospital of Ruhr University, Bochum, Germany

¹²Department of Clinical Genetics and Pathology, Laboratory Medicine, Region Skåne and Lund University, Lund, Sweden

¹³Department of Pediatrics, University of North Carolina, Marsico Lung Institute, Chapel Hill, North Carolina

¹⁴Comprehensive Pneumology Center (CPC-M), Member of the German Center for Lung Research (DZL), Munich, Germany

¹⁵Department of Pediatrics, Marienkrankenhaus St. Josef Kohlhof, Neunkirchen, Germany

¹⁶Danish PCD & Child Centre, CF Centre Copenhagen, Paediatric Pulmonary Service, ERN Accredited, Department of Paediatrics and Adolescent Medicine, Copenhagen University Hospital, Rigshospitalet, Denmark

¹⁷Department of Pediatrics, Federico II University, Naples, Italy

¹⁸CF Center, Children’s Hospital, Faculty of Medicine, University of Cologne, Germany

¹⁹KUNO University Children’s Hospital, Regensburg, Germany

²⁰Center for Pediatrics and Adolescent Medicine, University Hospital Freiburg, Freiburg University Faculty of Medicine, Freiburg, Germany

²¹Department of Pediatric Pulmonology, Immunology and Intensive Care Medicine, Cystic Fibrosis Centre Berlin, Charité-Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Berlin, Germany

²²Department for Pediatric Pneumology, Allergologssy and Neonatology, Hannover Medical School, Hannover, Germany

²³Formerly: Department of Pediatrics, Klinikum Nürnberg Süd, Nürnberg, Germany; now retired

²⁴Department of Pathology/Lab Medicine and Marsico Lung Institute, University of North Carolina School of Medicine, Chapel Hill, North Carolina