Introduction

Cilia are microtubule-based structures projected from the surface of almost all differentiated cells. These highly conserved organelles can be classified as motile and non-motile (also known as primary cilia). Motile cilia are involved in cell motility, mainly moving extracellular fluids1,2, while primary cilia function as sensory cellular antennae for a wide range of signalling pathways such as Sonic hedgehog (Shh), Wnt, planar cell polarity (PCP) and/or Notch, which are essential during embryonic development and also in adult physiology3,4. Given the great variety of functions performed by these organelles as well as their presence in nearly all mammalian cells, defects in ciliary axonemal structure and/or cilia function can result in a widespread range of phenotypes leading to human genetic disorders known as ciliopathies5,6,7.

Bardet-Biedl syndrome (BBS; MIM #209900) is an autosomal recessive ciliopathy associated with defects in primary cilia8. A wide spectrum of clinical symptoms are observed in BBS patients, among which retinal dystrophy, obesity, limb abnormalities, cognitive impairment, renal and urogenital anomalies have been classified as primary features. A number of secondary features, such as diabetes mellitus type 2, developmental and psychomotor delay, cardiovascular anomalies and craniofacial defects, are frequently observed. Pathogenic variants in 21 genes (BBS1-21) have been related to this syndrome, which account for nearly 80% of BBS patients9,10.

These genes encode proteins mainly localized to basal bodies and centrosomes, but can also be found in the ciliary axoneme and the transition zone (reviewed in Novas et al.11). BBS proteins are necessary to maintain primary cilia structure and function12, some of them through multiprotein complexes. Such is the case of the BBSome, composed by eight BBS proteins (BBS1, BBS2, BBS4, BBS5, BBS7-BBS9 and BBIP10/BBS18), which participates in the vesicle trafficking of membrane proteins to and inside the primary cilium to promote ciliogenesis13,14. The BBSome assembly is mediated by the BBS-chaperonin complex, formed by BBS6, BBS10 and BBS12 proteins15,16, together with CCT/TRiC chaperonins. Therefore, altered or non-functional BBS proteins often entail consequences in cilia formation and/or maintenance.

The high genetic and phenotypic heterogeneity typically associated with this syndrome, together with the overlapping phenotypes among ciliopathies which hinder clinical assessment, have made molecular diagnosis a difficult task in many cases. Because of that, researchers are taking advantage of the great variety of current molecular approaches, attempting to achieve a rapid and accurate diagnosis for these patients. In this sense, it is clear that the NGS technology has revolutionised the field of molecular diagnosis and is increasingly being implemented in biomedical research and clinical practice. However, these methods generate a large number of genomic variations whose interpretation represents a great challenge.

This fact points out the importance of performing in vivo assays, which are crucial to gain more knowledge about the mechanisms underlying human ciliopathies and to functionally evaluate genetic variants. Several model organisms have been extensively used to study the complex genetic basis of this group of disorders17. Although each model has strengths and limitations, vertebrate models have been shown to be more advantageous, mainly to investigate the abnormal organogenesis associated to human ciliopathies17,18.

During the last decade, zebrafish (Danio rerio) has emerged as a powerful vertebrate model organism for biomedical research, allowing us to interpret the pathogenicity of genetic alterations identified in ciliopathy patients1,18. This is because zebrafish offers numerous advantages compared with other animal models such as low cost, rapid development, embryonic transparency and easy manipulation, as well as large fecundity rates with clutches of several hundred eggs. In addition, the availability and relatively easy-handling of antisense KD techniques contribute to the understanding of gene function. Particularly, morpholino antisense oligonucleotides (MOs), when properly controlled for off-target effects and toxicity, have been shown to be a powerful tool to study gene function1,19,20.

The zebrafish model has several ciliated organs containing both primary and motile cilia1. KD of BBS genes in zebrafish are reported to cause early developmental phenotypes typically associated with PCP pathway defects21,22. These are usually initiated along with KV disruption, a transient ciliated organ that, when affected, leads to defects in left-right asymmetry establishment, the first embryonic process related to cilia function2.

Here we report the functional characterization of several new BBS variants identified in five unrelated patients clinically diagnosed with BBS. In vivo assays were performed in zebrafish by combining antisense MO gene KD approach and human mRNA for rescue experiments to assess developmental defects during gastrulation, particularly in KV.

Results

Molecular genetic diagnosis

The use of different genetic tools (genotyping microarray, direct sequencing, homozygosity mapping, and whole exome sequencing –WES-) led us to identify seven candidate variants in three BBS genes in this group of patients clinically diagnosed with BBS (shown in Table 1). Three of the BBS1 (MIM #209900) variations (except p.Met390Arg) were previously reported as novel by our group23 and included in this study for functional characterization. The missense change in this gene (p.(Val366Asp)) has been predicted to be pathogenic by four bioinformatics tools (Table 2). The nonsense and deletion variants have been assumed as pathogenic.

Table 1 Genetic and phenotypic data of the patients under study.
Full size table
Table 2 Pathogenicity prediction at protein level of all variants by several in silico tools.
Full size table

Homozygosity mapping, performed in the only patient belonging to a consanguineous family (RP1573), combined with direct sequencing revealed a novel homozygous missense change (p.(Gly411Ala)) in MKKS/BBS6 (MIM #604896) gene (Table 1), which has been predicted to be pathogenic by three out of four bioinformatics tools (Table 2), localizes in a highly conserved region of the encoded protein (Fig. 1) and segregates from both parents (Fig. 2). On the other hand, two variants in heterozygous state (p.(Arg138Cys) and p.(Phe180Phefs*6)) were identified by WES in BBS5 (MIM #603650) gene in patient RTP23. All pathogenicity tools provided a damaging prediction for the missense change (Table 2). The novel deletion was assumed to be pathogenic. Both have been validated by direct sequencing and segregate within the family (Fig. 2). We also analysed in silico their potential effect on splicing, finding that all of them have a positive prediction with at least two out of four tools, either modifying or eliminating a donor or acceptor splice site (Table 3). Novel variants were absent in 100 control alleles of Galician origin, and their frequency was checked in several public databases.

Figure 1
figure 1

Alignment of a fragment of BBS6/MKKS protein showing complete conservation of residue 411 across species. Homo sapiens (human), Pan troglodytes (chimpanzee), Mus musculus (mouse), Rattus norvegicus (rat), Canis lupus familiaris (dog), Xenopus tropicalis (frog), Danio rerio (zebrafish).

Full size image
Figure 2
figure 2

Segregation of the variants identified in BBS5 and BBS6 genes.

Full size image
Table 3 Impact prediction of BBS variants on splice sites, an indicative of possible splicing defects.
Full size table

Whole-mount in situ hybridization reveals early developmental defects in zebrafish

According to the previous evidences of the potential pathogenicity of the identified variants, their functional effect was evaluated in vivo. To do this, zebrafish embryos (n = 180–300 embryos/MO) were first injected with specific MOs against each bbs gene to assess KD phenotypes at 8–12 somite stage. The specificity and efficacy of the MOs used in this work have already been established in a previous study24. Thus, consistent with published data24, our results also show that among the MOs-injected animals, 97% showed several gastrulation defects commonly associated with BBS phenotypes, including shortened body axis/length, kinked and wide notochords, and thinner somites (Fig. 4).

Co-injection of each bbs-MO with the corresponding WT human BBS capped-mRNA efficiently rescued those morphant phenotypes, which confirms the specificity of the MOs targeting. Thus, the defects associated with planar cell polarity (PCP), as indicated by various gene markers, were almost completely rescued by wildtype human mRNA injection.

Then, we carried out the individual rescue experiments by injecting embryos (n = 180–300 embryos/variant) with the bbs-MO plus mutant human BBS capped-mRNA to assess their ability of BBS variants to rescue the MO-associated phenotypes. Expression and translation stability of mutated human BBS proteins (from RNA transcripts) from their DNA constructs used in the individual rescue experiments was assessed in an in vitro transcription and translation-coupled system (TNT® Coupled Reticulocyte Lysate Systems; Promega). Figure 3 illustrates the in vitro transcription/translation assay using the WT and mutant human cDNA constructs showing that each pCS + -cDNA construct adequately translated and labelled into a biotinylated protein. The kDa size of the each in vitro translated protein accurately correlates with the expected ORF (Fig. 3), therefore, indicating the transcriptional stability of the different mutated human DNA constructs generated.

Figure 3
figure 3

Expression and translation stability of WT and mutated human BBS proteins (from RNA transcripts) from their DNA constructs in an in vitro transcription and translation assay (TNT). Protein products were analyzed using western blot analysis. Resulting proteins were characterized by size and correlated to the size and sequence of a given gene or ORF. The kDa size of the each in vitro generated protein correlates exactly with the expected ORF (arrow heads), therefore, indicating the transcriptional stability of the different mutated DNA constructs generated. Non-mutated proteins were used as internal controls (Lanes 2, 6 and 10). Due to the low molecular weight of the expected mutated protein (2 kDa), no band were detected on lane 3. Note: Lane M: standard protein marker; Lane 1: negative control; Lane 2: WT BBS1; Lane 3: BBS1:p.(W23*) mutation; Lane 4: BBS1 p.(V366D) mutation; Lane 5: BBS1 p.(H504fs*48) mutation; Lane 6: WT BBS5; Lane 7: BBS5 p.(R138C) mutation; Lane 8: BBS5:p.(F180fs*6) mutation; Lane 9: BBS5(N184S) mutation; Lane 10: WT BBS6; Lane 11: BBS6:p.(G411A) mutation.

Full size image

To validate the defects observed in live embryos and classify them, we performed in situ hybridization with a cocktail of krox20, myoD and pax2 riboprobes and examined embryonic defects on whole embryos flat-mounted at 8–12 ss (n = 60–75) (Fig. 4 and Supplementary Fig. S2). Each coinjection of bbs-MO along with mutant human capped-mRNA was compared with the MO alone and/or with the respective WT rescue. The percentage of rescued embryos was 95%, 93% and 92% for bbs-MO plus BBS1, BBS5 or BB6 WT mRNA coinjections, respectively (Supplementary Fig. S2).

Figure 4
figure 4

Phenotypes of zebrafish embryos at 8–12 ss, after whole mount in situ hybridization. Knockdown of zebrafish bbs1 (A–F), bbs5 (G–L) and bbs6 (M–P) genes affects body axis/length, notochord and somite morphology. Morphology of the controls (A,G,M; dorsal view anterior to the top), morpholino (B,H,M; dorsal view anterior to the top), morpholino plus WT human BBS capped-mRNA (C,I,O), and morpholino plus sense capped-mRNA of different human BSS variants (D–F,J–L,P; dorsal view anterior to the top) zebrafish at 12 somites (ss). The embryos were fixed for in situ hybridization of myoD/krox20/pax2. Injection of WT human BSS1, BSS5 and BSS6 capped-mRNA rescues the phenotype of bbs1, bbs5 and bbs6 MO-injected embryos, respectively. However, the mutated capped-mRNA of the BBS particular changes of each loci (W23*, V366D, H504fs, R138C, F180fs, N184S, G411A – abbreviated nomenclature) were not able to rescue the respective phenotype of BBS1, BBS5 and BBS6 MO-treated embryos. Scales bars: 100 µm.

Full size image

We found that none of the human BBS variants were able to entirely recover the zebrafish knockdown bbs phenotypes, thereby ruling out the presence of benign variations among the seven BBS gene variants studied. Despite of that, some new identified BBS variants (e.g., p.(Arg138Cys)) were able to partially rescue the different zebrafish MO-phenotypes features (see Fig. 4). Then, we classified as potentially hypomorphic those changes which rescued some of the different MO phenotype features but showed a more severe phenotype than WT rescue fish, as functional null when similar to MO, and dominant negative when worse than MO. The displayed phenotypes of embryos hybridized in situ (shown in Fig. 4) are characterized by several gastrulation defects: shortened body length/axis (especially in bbs1 and bbs6 morphants), smaller head (e.g., BBS1:p.(Trp23*)), wide and kinked notochord (in nearly all morphants), longer and thinner somites (e.g., BBS1:p.(Val366Asp), BBS5:p.(Arg138Cys)), partial loss of somite definition (e.g., BBS1:p.(His504Hisfs*48), BBS6:p.(Gly411Ala)) and reduced eye labelling, suggesting smaller eye size (e.g., BBS1:p.(Trp23*) and p.(Val366Asp)). The most efficient mutant rescue, but considerably different from WT rescue, was associated to BBS5:p.(Arg138Cys), while BBS1 and BBS6 changes seemed to produce the more severe phenotypes. The classification of all variants based on in situ results is showed in Table 4.

Table 4 Classification of variants based on WISH results.
Full size table

Direct assessment of cilia located in Kupffer’s vesicle (KV) confirms the deleterious effect of new BBS variants

We analysed zebrafish Kupffer’s vesicle (KV) cilia of embryos at 8–10 (±1) somite stage. About 40–45 cilia except in cases of reduced number of cilia, from 5 KVs were measured for each condition to compare with controls. We calculated the average cilia length with associated standard error (standard deviation) and p-value (Welch’s t- test). As previously reported for other bbs genes25, KD of bbs1, bbs5 and bbs6 resulted in a significant reduction in cilia length, from 5.10 + 0.42 µm in controls to 3.25 + 0.22 µm, 3.43 + 0.24 µm and 3.33 + 0.11 µm in bbs1, bbs5 and bbs6 MO-injected embryos, respectively (p < 0.01; Figs 5C, 6C and 7C, respectively). bbs-MOs and WT human BBS capped mRNA co-injected embryos showed KV cilia length similar to controls (4.97 + 0.31 µm, 4.96 + 0.20 µm and 4.83 + 0.13 µm in bbs1, bbs5 and bbs6 genes; p > 0.05; Figs 5B, 6B and 7B), confirming the rescue of the morphant phenotype in all cases. Then, we compared bss morphants to controls and found that all injected capped mRNA mutant variants resulted in shorter KV cilia (values shown in Figs 5G, 6G and 7E), with the exception of BBS5:p.(Phe180Phefs*6), which in contrast fully rescued the morphant phenotype, being not statistically different from controls (Fig. 6G). BBS1:p.(Trp23*) change was associated with the most significant shortening of KV cilia, averaging 2.36 + 0.07 µm, even almost 1 µm shorter than bbs1MO condition (Fig. 5D,G).

Figure 5
figure 5

Representative images of KVs and comparison of average cilia length corresponding to BBS1 conditions. Projections of confocal z-stacks of Alexa fluor 594 and Dapi stained cilia located in Kupffer’s vesicle (KV) from control (A) and bbs MO-injected embryos (B), bbs MO-injected embryos plus WT capped-mRNA (C) and bbs MO-injected embryos plus BBS1 new changes (W23*, V366D, H504fs) capped-mRNA. (D–F) Note the reduction of the cilia in bbs1-MO- and BBS1 variants capped-mRNA co-injected embryos. (G) Comparison of average cilia length. *p < 0.01, **p < 0.001. Scales bars: 10 µm.

Full size image
Figure 6
figure 6

Representative images of KVs and comparison of average cilia length corresponding to BBS5 conditions. Projections of confocal z-stacks of Alexa fluor 594 and Dapi stained cilia located in Kupffer’s vesicle (KV) from control (A) and bbs5 MO-injected embryos (B), bbs5 MO-injected embryos plus WT capped-mRNA (C) and bbs5 MO-injected embryos plus BBS5 new changes (R138C, F180fs, N184S) capped-mRNA. (D–F) Note the reduction of the cilia in bbs5 MO and BBS5 variants R138C and N184S capped-mRNA co-injected embryos. (G) Comparison of average cilia length. *p < 0.01, **p < 0.001. Scales bars: 10 µm.

Full size image
Figure 7
figure 7

Representative images of KVs and comparison of average cilia length corresponding to BBS6 conditions. Projections of confocal z-stacks of Alexa fluor 594 and Dapi stained cilia located in Kupffer’s vesicle (KV) from control (A) and bbs6 MO-injected embryos (B), bbs6 MO-injected embryos plus WT capped- mRNA (C) and bbs6 MO-injected embryos plus BBS6 new changes (G411A) capped-mRNA. (D) Note the reduction of the cilia in bbs6 MO-injected and BBS6 variant capped-mRNA co-injected embryos. (E) Comparison of average cilia length. *p < 0.01, **p < 0.001. Scales bars: 10 µm.

Full size image

KV cilia are normally organised in a spherical pattern, which can be clearly observed in both controls and WT injected embryos. However, this distribution was disrupted in most morphants, in addition to differences in cilia number. MOs against bbs1, bbs5 and bbs6 genes led to a significant reduction in cilia number (Supplementary Fig. S1) together with a loss of the spherical pattern (Figs 5C, 6C and 7C, respectively). Overall, bbs1 morphants showed the most significant KV defects (Fig. 5D–F), with cilia number markedly reduced in all conditions (Supplementary Fig. S1), in addition to the greatest shortening mentioned above. It is noteworthy that BBS1:p.(Trp23*) morphant cilia appeared to aggregate into a smaller domain (Fig. 5D). Although no significant differences in cilia length of BBS5:p.(Phe180Phefs*6) rescued morphants were found, KV cilia appear to be dispersed, covering a slightly broader area compared to controls (Fig. 6E), as well as a significant reduced number of KV cilia (Supplementary Fig. S1). Interestingly, BBS6:p.(Gly411Ala) non-rescued morphant fish also showed a portion of cilia with a curvy shape, not as frequently seen in other morphants (Fig. 7D), also accompanied with a significant lower number of cilia (Supplementary Fig. S1).

Discussion

The molecular diagnosis of ciliopathies, such as BBS, can be difficult due to the genetic heterogeneity associated with these disorders. The combination of several genetic approaches allowed us to confirm the molecular diagnosis of five BBS patients whereby seven predicted pathogenic variants, some of them novel, were found in three BBS genes: BBS1, BBS5 and BBS6. While BBS1 is the most frequently mutated gene in BBS, accounting for 23% of cases and 15% of the total mutational load10,26, BBS5 represents a minor contributor, with only 2% of cases.

The most common disease-causing mutation in BBS1, p.Met390Arg, was first identified by the genotyping microarray in heterozygous state in patient GBB27, for whom a second heterozygous change in this gene was found by direct sequencing, p.(His504Hisfs*48), a small deletion of 11 nucleotides assumed to be pathogenic. This genotype is consistent with other cases in which the presence of heterozygous p.Met390Arg variant was generally associated with another mutated allele in the same gene27. The other two BBS1 homozygous changes were: p.(Trp23*), a nonsense mutation located in the first residue of the N-terminal domain of the encoded protein, and p.(Val366Asp), which lies in a WD40 repeat domain, usually required for the assembly of multiprotein complexes28. Regarding the three BBS5 variants, one of them was novel, p.(Phe180Phefs*8), and found in the same patient (RTP23) in compound heterozygosity with the missense change p.(Arg138Cys), which was classified as novel at the moment of identification since it was not found in public databases. A third BBS5 mutated allele, p.(Asn184Ser), was found in heterozygous state in a patient (RP1573) who in turn resulted homozygous for the novel BBS6:p.(Gly411Ala) variant. Other families with heterozygosity for p.(Asn184Ser) have been reported29,30, one of them also homozygous for BBS1:p.Met390Arg mutation29. The Asn184 residue, like Phe180, localizes in a putative domain called DM16, a predicted repeat domain evolutionary conserved across species. Therefore, mutations in this domain are thought to be pathogenic30. Many researchers have proposed a complex mode of inheritance in BBS involving the presence of three mutated alleles in two different BBS genes, in which the third allele would also be necessary for the development of the disease (triallelism), or would exert a modifying or epistatic but not causal effect, leading to more severe phenotypes29,31. In this case, there were no siblings available to evaluate the specific effect of this allele, but according to reported cases, it seems to be a genetic modifier.

Although several evidences suggest that the identified variants are likely to be pathogenic, and considering that BBS proteins, as many others related to cilia, are highly conserved across ciliated organisms and share a high protein identity with zebrafish orthologs32, we performed functional analysis in zebrafish model to assess their effects and validate their pathogenicity. Zebrafish model has been widely used to analyse ciliary dysfunction32, with several reported models of BBS, but also of other ciliopathies like Meckel-Gruber syndrome33,34, Joubert syndrome and nephronophthisis, predominantly showing defective melanosome transport, retinal defects, defective cilia in KVs and kidney anomalies, among others25,32,35.

Thus, the injection of MOs against bbs1, bbs5 and bbs6 genes in zebrafish embryos confirmed the previously reported phenotypes, including several developmental defects such as curved and/or shortened body axis, together with notochord and somite anomalies36, which are also commonly associated with other ciliopathies. Thus, Bbs disruption causes phenotypes associated with planar cell polarity (PCP) defect. Using the same MOs, Zaghloul et al. (2010) also reported that the BBS disruption causes phenotypes associated with planar cell polarity (PCP) defects in zebrafish; likewise, there are other studies that also showed evidence for similar axial phenotypes by in vivo loss of function methods in zebrafish21,22. However, other recent studies37 do not report early developmental phenotypes typically associated with PCP pathway defects. These possible discrepancies could be attributed to behavior/mutagenic efficiency of the different MOs used. Our bbs phenotypes were efficiently rescued by the corresponding WT human BBS capped-mRNAs, which confirms the specificity of the observed defects, however new mutant BBS variants mRNAs led to a high number of embryos with an abnormal phenotype, providing clear evidences of their pathogenicity. In addition, morphant embryos also displayed KV abnormalities, mainly concerning cilia length and number. KV has been proposed as a transient ciliated organ of asymmetry with a key role in left-right (LR) patterning by the establishment of a directional fluid flow26,38,39,40, a function equivalent to the mouse node40,41. KV originates during gastrulation and acquires a spherical pattern visible in the tail-bud during the somite stages32. It is known that the disruption of KV represents the first morphological abnormal phenotype associated with bbs gene KD2,26,32, that is why KV has been very useful in functional assessment of ciliopathies.

We focused on KV cilia length since it has been demonstrated that it is less variable than KV cilia number and lumen size40, and an optimal cilia length is required for proper KV functional flows39,40. We found that all genetic changes, with the exception of p.(Phe180Phefs*6), failed to rescue the cilia shortening caused by their respective gene conditional KD, therefore confirming that the tested bbs gene variants are deleterious for protein function, as reported by other authors for bbs gene KD32,36,42,43. In particular, the only nonsense change included in this study, BBS1:p.(Trp23*), resulted in the most reduced cilia length, less than half of the cilia length of control embryos and even shorter than in bbs1-MO injected embryos, which was also consistent with the severity of the phenotype observed by WISH. Despite the BBS6 missense variant (p.Gly411Ala)) resulted in a smaller reduction of cilia length compared to controls, it was also statistically significant, relatively similar to bbs6-MO length and consistent with the corresponding morphant phenotype, classifying it as null. This variant also showed a significant reduction in the number of KV cilia. The p.(Asn184Ser) allele was already analysed by in situ in zebrafish embryos and classified as null in a previous study24, but it was included here to evaluate its effect on KV cilia length, resulting in a significant shortening. This, together with the observed gastrulation defects, also made us classify it as null. Thus, the results of BBS5:p.(Asn184Ser) and BBS6:p.(Gly411Ala) changes, found in the same patient (RP1573) in heterozygous and homozygous state, respectively, could justify the development of a more severe clinical spectrum of all the patients included in this study, which represents a clear example of complex inheritance.

In view of these results, bbs1 morphants seem to display the most severe early developmental phenotypes with clear somitic and notochordal defects, which are also reflected in the most significant KV cilia shortening. Given that mutations in BBS1 are one of the most common causes of BBS syndrome, some authors had already suggested that BBS1 should play an important role43. Thus, it has been demonstrated that BBS1 is a key regulator of BBSome entry into cilia8.

In general, the three mutated genes analysed in our study, as other BBS genes, have been well characterized not only in zebrafish, but also in mice and in vitro ciliated cells. Bbs1- and Bbs6-knockout in mice led to phenotypes similar to those observed in Wnt zebrafish mutants44. The importance of the chaperonin complex in the assembly of the BBSome has been demonstrated in mice, pointing out that Bbs6 is necessary for the stability of Bbs2 and Bbs745. In fact, a missense variant in this gene resulted in a markedly decreased interaction of BBS6 with BBS2, one of the BBSome components, which in turn represents the second subunit being incorporated into the complex15. Mutations in BBS6 are also associated with McKusick-Kaufman syndrome46, another ciliopathy characterized by congenital heart defects, hydrometrocolpos and postaxial polydactyly, one of the primary features of BBS. Moreover, many researchers found that the suppression of BBS1 and BBS5 dramatically affects to the ability of retinal pigment epithelium (RPE) cells to ciliate13,14, consistent with the studies showing KV abnormalities in bbs1, bbs5 and bbs6 KD embryos, suggesting a key role in the formation of KV32,42.

Over the last decade, many efforts have been made to gain insight into the ciliary functions. Thus, it is known that cilia are involved in several developmental pathways, such as Shh, key for tissue morphogenesis, and Wnt signalling, whose proper functioning is necessary for organ growth and differentiation47. In this sense, zebrafish has been a useful tool in the study of the connection between ciliary dysfunction and ciliopathies. The possibility to carry out genetic interaction analyses in zebrafish has revealed the role of the BBSome in the PCP, also named non-canonical Wnt, pathway21. BBS genes also seem to be involved in the regulation of the canonical Wnt pathway22. Additionally, zebrafish model has also revealed the involvement of BBS genes in Shh pathway and, therefore, its usefulness in assaying human polydactyly, one of the most frequent clinical features in BBS patients and typically associated with Shh pathway defects17,42, which could not be directly assayed in mice48. All this reflects the great importance of BBS genes in development and the need of an unaffected BBS function for an appropriate maintenance of cilia and ciliated cells32.

In summary, our results highlight the efficiency of the genetic strategy followed here for the molecular diagnosis of five families with clinical spectrum of BBS, allowing us to identify candidate pathogenic changes in known BBS genes. Functional assays in zebrafish provide clear evidences of the pathogenic effect of the analysed variants. Although certain in silico deletions and nonsense changes are usually considered as pathogenic without performing further functional analysis, this in vivo study confirms their harmful effects. Likewise, in the case of the missense variants the present experiments provided evidence for their implication in the phenotype. Despite the major advances in genomic engineering have provided new approaches to generate genetic mutants for experimental studies, such as zinc-finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and CRISPR/Cas-based genome editing systems, we show that the use of MOs to transiently KD BBS genes in zebrafish is a useful, efficient and relatively rapid approach to assay variant pathogenicity and model human diseases like BBS and other ciliopathies.

Methods

Study cohort

Five patients from unrelated families clinically diagnosed with BBS were included in this study. All of them are of Spanish origin, except one patient from India. A fifth degree of consanguinity was referred for one of the Spanish families (RP1573). This study was approved by the Galician Ethical Committee for Clinical Research (Spain - no.2006/08) and adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from all patients or their guardians.

Mutational analysis

Peripheral blood from all participants and available family members was collected for DNA extraction using the Flexigene DNA kit 250 (Qiagen), following the manufacturer’s protocol. Mutations in three of these families were described in previous work23, while those in the remaining two families were identified by homozygosity mapping (RP1573) and WES (RTP23) (unpublished results). The functional consequences of missense changes were predicted using PolyPhen249, SIFT50, Pmut51 and Mutation Taster52 tools. Nomenclature of novel variants was done according to the guidelines of Human Genome Variation Society (http://www.hgvs.org/).

Sanger sequencing was performed in probands to validate the selected changes from WES data and to analyse coding regions of the candidate gene identified by homozygosity mapping. Then, segregation analysis was carried out in available family members. We also analysed 100 control chromosomes of unrelated Galician individuals, to assess the allele frequency of novel variants, which were also were checked against the following databases: Exome Aggregation Consortium (ExAC, http://exac.broadinstitute.org/), dbSNP (https://www.ncbi.nlm.nih.gov/projects/SNP/) and HGMD (Human Gene Mutation Database, http://www.hgmd.cf.ac.uk/ac/index.php).

Site-directed mutagenesis

cDNAs encoding full length human BBS genes cloned into pCS2 + vector, which were provided by Dr. Norann Zaghloul (University of Maryland School of Medicine), were used to synthesize capped mRNAs for zebrafish rescue experiments. Site-directed mutagenesis was performed using the QuickChange® II XL Site-Directed Mutagenesis kit (Agilent Technologies) to introduce sequence variants of unknown functional significance into the corresponding WT cDNAs, and confirmed by direct sequencing. The primers used are shown in Supplementary Table S1.

In vitro transcription-translation assay

Expression and translation stability of mutated human BBCs proteins (from RNA transcripts) from their DNA constructs was assessed in an in vitro transcription and translation-coupled system (TNT® Coupled Reticulocyte Lysate Systems; Promega). The amplified full length mutated and non-mutated BBS cDNAs were subcloned into the pCS2 + vector. The pCS2 + vector utilizes the SP6 RNA polymerase binding site that can be used directly for in vitro transcription–translation assays. The assay was performed according to the instructions provided with the Promega TNT Coupled Reticulocyte Lysate Systems using TranscendTM non-radioactive translation detection system, (Promega). The synthesized proteins were analyzed by SDS gel electrophoresis on a 12% acrylamide gel. To determine whether all mutant proteins are specifically recognized, western blot analysis and detection of the corresponding biotinylated proteins from the TNT® assay was performed using Streptavidin-HRP Conjugate and TranscendTM Chemiluminescent Substrate as described in the manufacturer protocol.

Zebrafish husbandry

Fish (Zebrafish, Danio rerio) embryos were obtained as previously described53 and staged according to Kimmel et al.54. Experiments were performed with a TU-WT strain (Tuebingen (TU), Nüsslein-Volhard Lab). Dechorionated embryos were collected for in vivo imaging or fixed overnight at 4 °C in 4% paraformaldehyde (PFA) in 1X PBS, washed in 1X PBS, and dehydrated through methanol series and stored at −20 °C in 100% methanol for in situ hybridization. Ethical approval (Ref: CSIC/OH-150/2014) was obtained from the Institutional Animal Care and Use Committee of the IIM-CSIC Institute in accordance with the National Advisory Committee for Laboratory Animal Research Guidelines licensed by the Spanish Authority (RD53/2013) and conformed to European animal directive (2010/63/UE) for the protection of experimental animals.

Morpholino knockdown and mRNA rescue

Translational start blocking MOs (GeneTools, LLC) were used to specifically knockdown bbs1, bbs5 and bbs6 endogenous zebrafish genes (Supplementary Table S2)24. The MOs were resuspended in water to a final concentration of 20 mM (stock solution). For rescue assays, capped mRNAs were transcribed in vitro using the mMessage mMachine SP6 transcription kit (Ambion) and each human BBS1, BB5 or BBS6 cDNA (WT or mutated) cloned into pCS2 + vector were used as templates. For zebrafish embryo injections, 2 nl of each specific bbs-MO and capped human mRNA solution (Supplementary Table S2) were injected into one- or two-cell stage embryos with 1% of phenol red as a tracer, based on amounts previously reported in the literature24. About 180–300 embryos per condition were injected. Microinjection was performed under a dissection microscope (MZ8, Leica) fitted with a MPPI-2 pressure injector (ASI Systems).

Whole-mount in situ hybridization (WISH)

Whole-mount in situ hybridization was carried out on zebrafish embryos previously fixed at 8–12 somite stage (ss) in 4% paraformaldehyde in 1X PBS, using a cocktail of three digoxigenin-labeled antisense RNA probes for pax2, myoD and krox20, according to previously described protocols55,56,57. Images of flat-mounted and cleared embryos were taken using a Leica M165FC stereomicroscope with DFC 310 camera at 10X magnification, in order to analyse the morphological defects in early-development of injected embryos.

Whole-mount Immunofluorescence and cilia analysis

Zebrafish control and injected embryos were fixed at 8–12 ss, for KV analysis, in 4% PFA (Sigma) for 24 h at 4 °C, then washed in 1X PBS- 1% Triton X-100 (PBST; Sigma) and dehydrated gradually through a methanol series and stored in 100% methanol (Sigma) at −20 °C. Whole-mount immunofluorescence experiments were performed as follows: embryos were first washed with PBST several times, blocked with goat serum 5% for 1 h before incubation with primary and secondary antibodies both in PBS-Triton (1%) plus 2% goat serum overnight at 4 °C. We stained KV cilia using a mouse anti-acetylated α-tubulin (1:750; T7451, Sigma) monoclonal antibody and, subsequently, with Alexa Fluor 594-conjugated goat anti-mouse secondary antibody (1:1000, ThermoFisher). Nuclei were stained with DAPI.

Finally, embryos were mounted in 75% glycerol and confocal images were acquired on a Leica SP5 microscope and analysed using ImageJ software. Identical settings (gain, offset, laser power) were applied to all z-stack acquisitions, and all fluorescence measurements were performed on maximum intensity projection, also calculated with ImageJ.

Statistical analyses

Statistics for KV cilia length measures were calculated using one-way ANOVA paired with the Welch significant difference test in SPSS v22.0 statistics software (IBM). To compare cilia number, we performed a Mann-Whitney test for non-parametric data. A p-value of <0.05 was consider as statistically significant.