Cilia play important signaling or motile functions in various organisms. In Human, cilia dysfunctions are responsible for a wide range of diseases, called ciliopathies. Cilia assembly is a tightly controlled process, which starts with the conversion of the centriole into a basal body, leading to the formation of the ciliary bud that protrudes inside a ciliary vesicle and/or ultimately at the cell surface. Ciliary bud formation is associated with the assembly of the transition zone (TZ), a complex architecture of proteins of the ciliary base which plays critical functions in gating proteins in and out of the ciliary compartment. Many proteins are involved in the assembly of the TZ, which shows structural and functional variations in different cell types or organisms. In this review, we discuss how a particular complex, composed of members of the DZIP1, CBY and FAM92 families of proteins, is required for the initial stages of cilia assembly leading to ciliary bud formation and how their functional hierarchy contributes to TZ assembly. Moreover, we summarize how evidences in Drosophila reveal functional differences of the DZIP1–CBY–FAM92 complex in the different ciliated tissues of this organism. Whereas it is essential for proper TZ assembly in the two types of ciliated tissues, it is involved in stable anchoring of basal bodies to the plasma membrane in male germ cells. Overall, the DZIP1–CBY–FAM92 complex reveals a molecular assembly pathway required for the initial stages of ciliary bud formation and that is conserved from Drosophila to Human.
Cilia are highly conserved organelles present at the surface of eukaryotic cells where they act as cell antenna by capturing and transmitting signals, or as propellers providing cell motility and/or generating fluid flow. During the past few years, their importance has been highlighted because defects in cilia assembly or function are involved in a wide range of pathologies of varying severity, all classified as ciliopathies . From a structural point of view, cilia are composed of four major parts: (i) the axoneme, the backbone of cilia; (ii) the ciliary gate that selects and sorts the protein content in and out of cilia [2,3] and comprises the transition zone (TZ) and the transition fibers (TF). They connect, respectively, the axoneme and the basal body to the plasma membrane; (iii) the basal body, derived from the mother centriole from which the axoneme is templated and (iv) the ciliary membrane which concentrates multiple receptors sensing extracellular signals (Figure 1A).
Organization and assembly of the ciliary base.
Ciliogenesis starts with the conversion of the mother centriole into a basal body. Several routes leading to the effective docking of the centriole to the plasma membrane have been described  (Figure 1B). In mammalian mesenchymal cells, cilia assembly relies on the recruitment of vesicles to the distal end of the mother centriole by the centriolar distal appendages. The resulting ciliary vesicle caps the distal end of the basal body, underneath which the TZ starts to assemble leading to the formation of the ciliary bud. The ciliary vesicle eventually fuses to the plasma membrane. Completion of cilia elongation last relies on effective intraflagellar transport (IFT) inside the ciliary compartment in most organisms, the notable exceptions being Plasmodium and Drosophila sperm flagella [5–7]. Several molecular components have been involved in the recruitment of the ciliary vesicles to the centriolar distal appendages, in particular EHD proteins in association with PACSIN [8,9], LRRC45 , components of the RAB8 and RAB11 vesicular transport machinery and myosin VI motor  (for review see also [12,13]). A second route has also been described in polarized cells, where the basal body directly docks to the plasma membrane without apparent recruitment of ciliary vesicles inside the cell . How basal bodies precisely dock to the plasma membrane is much less described in several organisms like Drosophila or Caenorhabditis elegans and the mechanisms by which centrioles dock to the plasma membrane remain incompletely described. As well, what governs the selection of a specific route to build a cilium in one type of cell is not understood.
An accepted view is that the TZ is involved in sorting proteins in and out of the cilium, whereas TF proteins are also required for ciliary vesicle recruitment and basal body docking to the plasma membrane. However, recent observations indicate that the role of each module of TF or TZ proteins could be more complex than initially expected and vary between tissue and organisms [14–17]. In this review, we aim to illustrate how recent studies on a particular TZ complex, DZIP1–CBY–FAM92, bring novel understandings on the mechanisms that mediate ciliary budding in several organisms from Drosophila to Human.
Organism and cell-specific variations in ciliary gate architecture and composition
The ciliary gate is composed of the TF that emanate from the basal bodies, and the TZ that decorates the base of the axoneme, both showing characteristic ultrastructures by electron microscopy (EM). In particular, the TZ is characterized by electron-dense Y-shaped linkers that connect the axoneme to the ciliary membrane in most studied organisms [18,19]. However, these ultrastructural observations also show striking variations in the organization of the ciliary base between organisms, but also between cells in a same organism. For example in protozoa, specific structures observed by EM, called the basal and terminal plates, encase the TZ, but are not formally described for mammalian primary cilia [18,19]. In Drosophila, centrioles lack distal appendages and the TZ shows variations in both its organization and composition between cell types from different tissues [15,20,21]. In C. elegans, cilia do not show TF at the ciliary base and even lose their basal body . More, controlled remodeling of the TZ is observed in at least one type of ciliated sensory neurons in worms . This illustrates that the ciliary gate is not fixed and can be actively remodeled after its assembly and during the cilium life cycle. The most striking example of this remodeling, is the migration of the ring centriole in Drosophila spermatids . The ring centriole derives from the primary cilium TZ and migrates along the growing axoneme, thus maintaining a ciliary cap during spermatogenesis.
The molecular architecture of the ciliary gate has concentrated many studies in the past years, in part because of the highly deleterious consequences in humans of defects in ciliary gate proteins. At the basal body, a hierarchy of five proteins, namely CEP83, CEP89, SCLT1, CEP164 and FBF1, conserved among many organisms, is involved in building the TF and recruiting the ciliary vesicles [3,25,26]. Genetic analysis of human patients, biochemical screens in cells or genetic screens in model organisms have led to the characterization of several functional modules required for proper assembly and function of the TZ: three core complexes, the MKS (Meckel syndrome), NPHP (Nephronophthisis) and CEP290 modules play critical roles in organizing the TZ in many organisms (for review see ) (Figure 1C). A subset of all the TZ proteins from the MKS complex identified so far are conserved among most ciliated organisms . Impairment of MKS proteins impacts the transport of proteins inside the cilium and leads to alterations of the axonemal ultrastructure. Strikingly, in Drosophila, mutants of different MKS components only show very transient alterations of sensory cilia assembly, axonemes being still formed essentially as in wild type animal . This illustrates strong variations in the need for these different core TZ proteins between organisms. While recent studies indicate that TZ proteins are differentially required to build the TZ in different tissues of one organism [15,16], all studies indicate that they control ciliary composition and organization, but apparently not the initial attachment of centrioles to cytoplasmic vesicles or to the membrane leading to ciliary budding.
The DZIP1–CBY–FAM92 TZ assembly complex is required for ciliary bud formation
Recently, a complex of TZ proteins composed of DZIP1, CBY and FAM92, was identified in various organisms and seems to play also a function during initiation of cilia assembly, in addition to gating ciliary entry. DZIP1 (DAZ interacting zinc finger protein 1 ) and DZIP1L define a class of proteins conserved in animals, from Drosophila to mammals, but not in C. elegans. They are recruited to basal bodies and play a role in ciliogenesis in zebrafish, mouse, Human and Drosophila [30–37]. In agreement with a gating function of TZ proteins, Iguana/Dzip1 was first identified as a regulator of Sonic Hedgehog (SHH) signaling in zebrafish [31–34,38]. Its paralog, DZIP1L was also shown to play a similar role at the TZ by modulating SHH signaling in mouse and Human [35,36]. Mutations in DZIP1L and DZIP1 in Human are, respectively, associated with ARPKD (autosomal recessive polycystic kidney disease)  and asthenoteratospermia, one of the most common causes for male infertility .
DZIP1/L were further shown to strongly regulate the initiation of cilium assembly. The precise step at which DZIP1 or DZIP1L acts in the initiation of cilium assembly comes from observations by EM of the respective mutants. In iguana/dzip1 mutant fishes, centrioles apparently contact the plasma membrane, but the elongation of the ciliary axoneme does not take place at all in neural tube primary cilia . This indicates that Iguana/Dzip1 fosters axonemal growth while basal bodies contact the plasma membrane. Two different Dzip1l mutant mouse models show altered SHH signaling associated with altered cilia signaling and formation [35,36]. In the most severe mouse model, reduced DZIP1L function leads to defects in basal body anchoring and impairs ciliary bud formation. In this study, a significant number of basal bodies with docked vesicles but no ciliary buds were observed (11.7% versus 2.7% in wild type) and the number of undocked mother centriole inside the cell is higher in the absence of DZIP1L compared with control (67.1% versus 46% in wild type) . These indicate that Iguana in zebrafish and DZIP1L in the mouse are required to initiate ciliary budding, at least in a specific set of ciliated cells. In Drosophila, only one ortholog of DZIP1 and DZIP1L is present. Drosophila dzip1 hypomorphic mutants show aborted elongation of the TZ in all ciliated tissues and additionally improper docking of the basal body to the plasma membrane in spermatocytes .
Thus, conversely to most other described TZ components, members of the DZIP1 family in mouse, zebrafish and Drosophila are critical for membrane attachment to the basal body and to turn on ciliary bud formation and early axonemal growth. A comparable role in basal body anchoring to the ciliary membrane was also described for another group of TZ components. Double mutants of Mks5 (Rpgrip1l) or Mks6 (CC2D2A) and Nphp4 in C. elegans lead to aberrant positioning of the basal bodies at the tip of the dendrites in sensory neurons, thus indicating that MKS and NPHP components functionally co-operate in the basal body to membrane attachment . Such cooperative role has not been described in other organisms yet. Note that whereas DZIP1 and FAM92 are apparently not present in C. elegans, Mks5 and Nphp4 are not found in Drosophila melanogaster, all four are found in mammals. Thus, different complexes could be involved in this critical step in different organisms and/or tissues.
However, in Drosophila spermatocytes, impaired membrane attachment in the absence of DZIP1 leads, in contrast with what is observed in other organisms or tissues, to excessive microtubule axonemal elongation, indicating that microtubule elongation per se does not strictly require DZIP1 in this cell type. As discussed below, this observation likely reveals tissue-specific regulation of TZ assembly.
Several possible mechanisms by which DZIP1 and DZIP1L affect the initiation of TZ assembly and ciliary budding can be proposed: the first one is to mediate CP110 removal, as DZIP1L mutant cells fail to remove CP110 from the basal body distal end in mouse cells . However, data in Drosophila are not sufficient to claim that CP110 needs to be removed before axonemal elongation. Because DZIP1 is shown to interact with BBS4 and regulates BBsome trafficking to the primary cilium it could also promote ciliogenesis through IFT transport . However, this mechanism is likely not operating in Drosophila as IFT is not involved in cilia assembly in spermatocytes where DZIP1 defective centrioles fail to dock [6,7]. Moreover, no functional data on BBS proteins are available in this organism to address a more specific function of the BBsome in basal body docking. Last, DZIP1L has been shown to regulate RPGRIP1L recruitment at the TZ in mammals , but RPGRIP1L is not conserved in Drosophila, indicating that the molecular pathway by which DZIP1 regulates TZ assembly is likely not restricted to RPGRIP1L recruitment.
To explain how DZIP1/L mediate ciliary bud formation, another strong hypothesis comes from the observation, shared by recent publications, that DZIP1 and DZIP1L proteins interact with the TZ protein Chibby (CBY, [36,37,42,43]). This small protein, initially identified in Drosophila , is shown to be required in mouse cells to foster basal body docking in multi-ciliated epithelial cells [45,46]. In Drosophila, CBY delineates the ciliary TZ and is required for cilium assembly in all ciliated tissues . In all organisms, DZIP1/L appear to be required to recruit CBY at the TZ [36,37,42,43]. DZIP1/L-CBY complex may promote basal body anchoring by interacting with proteins belonging to the FAM92 family (Family with sequence similarity 92, [37,42]). Members of this family share a BAR-domain that is predicted to bind and shape membranes . There are two paralogs of FAM92 in vertebrates FAM92A and B, whereas only one paralog is found in Drosophila. Mammalian FAM92A/B or Drosophila FAM92 interacts with CBY and plays a function in cilium assembly . In Human, FAM92A was recently linked to autosomal recessive postaxial polydactyly, a congenital ciliopathy [49–51]. In Drosophila, depletion of FAM92 leads to TZ defects in all ciliated tissues and defective docking of the basal bodies to the plasma membrane in spermatocytes as observed by EM (Figure 2A,B). FAM92 protein could mediate, either the remodeling of the plasma membrane in contact of the basal body to foster ciliary growth or, stabilize basal body-membrane interactions. But observations from Drosophila do not strictly permit to distinguish these two possibilities. In agreement with a role in membrane organization, the mammalian FAM92A protein is necessary for the curvature of the mitochondrial membrane and its absence causes severe disruption of mitochondrial morphology and ultrastructure, which subsequently impairs organelle bioenergetics . As well, the overexpression of both FAM92 and CBY1 result in the formation of RAB8 positive vesicles in mammalian cells . In Drosophila, supernumerary vesicles positive for FAM92 and CBY could also be visualized in the cytoplasm when FAM92 and CBY are co-overexpressed (Lapart, unpublished observation) indicating the cooperation of CBY and FAM92 in remodeling vesicles. FAM92A was demonstrated to show positive membrane curvature activity in mammalian cells . Thus, FAM92 could play a role similar to PACSIN during the initial stages of ciliogenesis .
The DZIP1–CBY–FAM92 complex is required for ciliary budding and TZ assembly.
Altogether these recent data indicate that DZIP1–CBY–FAM92 form a functional TZ complex involved in basal body and membrane initial interaction, which triggers the budding of the axoneme. Whereas in mammals, TF are also proposed to be involved in the initial stages of ciliogenesis [3,17], in Drosophila, TF have not been observed at the distal end of the centrioles, even though some TF-like structures are detectable at the ciliary base in sensory neurons [15,20,53–55]. Thus, DZIP1–CBY–FAM92 represents a prominent complex mediating centriole to basal-body conversion independently of the presence of TF.
Note that hypomorphic mutants of DZIP1 or DZIP1L have more severe phenotypes than CBY or FAM92 null mutants in mouse or Drosophila [36,37]. Hence, DZIP1/L functions cannot be solely explained by their interactions with CBY and FAM92 and other interactors, still to be identified, are likely also required.
A regulatory loop is operating between CEP290 and DZIP1–CBY–FAM92 at the ciliary base
From studies in mouse and flies, a functional hierarchy can be established between DZIP1, CBY and FAM92. DZIP1 is required to recruit CBY and FAM92 and these latter are mutually interdependent for their localization at the ciliary base. In flies, removal of CBY or FAM92 also slightly affects the recruitment of DZIP1. From evidences in Drosophila, the pathway by which DZIP1–CBY–FAM92 are recruited to the basal body relies at least on CEP290 . CEP290 is important for TZ assembly in many ciliated organisms including mammals, Drosophila, C. elegans and protozoa. Human CEP290 mutations are responsible for different types of ciliopathies, indicating a central role of this protein in cilium assembly and homeostasis (for review see ). In cep290 Drosophila mutants, all members of the DZIP1–CBY–FAM92 are strongly reduced or lost at the TZ in all ciliated cells. More strikingly, observations in flies also indicate that DZIP1 and FAM92 exert a function in restricting CEP290 at the TZ. In their absence, CEP290 domain is expanded at the TZ (Figure 2C). Such interaction has not been formally addressed in other systems, but in mouse, reduction in DZIP1L alone is not sufficient to affect CEP290 localization at the base of primary cilia in mouse embryonic fibroblasts . From existing data, we can draw the hierarchical organization and functional interaction of CEP290 and DZIP1–CBY–FAM92, where CEP290 lies upstream of DZIP1, which in turn recruits CBY and FAM92. In Drosophila, a feedback regulation mediated by DZIP1 restricts CEP290 distribution at the proximal TZ (Figure 2D). However, CEP290 has also been shown to interact with different classes of proteins involved in trafficking pathways [56–63], indicating that CEP290 ciliary functions also take place independently of the DZIP1–CBY–FAM92 complex.
Molecular pathways required for basal body docking differ between tissues as revealed in
Drosophila defective in DZIP1–CBY–FAM92
Work in Drosophila also reveals striking differences in the behavior of the basal bodies in the absence of the DZIP1–CBY–FAM92 complex and supports the conclusion that ciliary gate assembly differs between tissues and cells.
In sensory neurons, basal bodies are still apposed to the membrane in the absence of DZIP1 or FAM92. On the contrary, the basal bodies from mutant spermatocytes are found partially or totally undocked at the cell membrane  (Figure 3). This indicates intrinsic differences in the mechanisms linking basal bodies to membranes between the two tissues. Such differences can also be observed in other model organisms for members of this complex. In mouse, CBY1 fosters basal body docking to the apical cell membrane during airway cell differentiation, but no such severe phenotypes are observed in other ciliated mouse tissues. However, because there are at least two paralogs for each members of this complex in mammals, possible redundancy in this molecular pathway in some tissues could mask the requirement for this complex in ciliary budding. Thus, only work in Drosophila is yet conclusive regarding this tissue-specific behavior.
Specific behavior of the basal bodies in
Drosophila spermatocytes are revealed in the absence of DZIP1–CBY–FAM92.
The work from Drosophila also reveals astonishing aspects of basal body maturation and axonemal elongation. First, in the absence of DZIP1 in spermatocytes, basal body attachment is impaired and this is associated with aberrant axonemal microtubule elongations from the distal tips of the centrioles. This observation highlights the crucial role of the ciliary membrane in controlling the growth of the axoneme during spermatogenesis. How this is performed is not understood at this point and remains hypothetical. Recruitment during this initial basal body–membrane interactions of proteins that regulate microtubule elongation have been proposed, among which members of the depolymerizing kinesin family Klp59D and Klp10 in Drosophila [54,64], both involved in regulating microtubule growth at the TZ or basal bodies in spermatocytes. Second, Drosophila spermatocytes have two pairs of centrioles, each composed of a mother and daughter centriole. All of the centrioles are able to dock and form short cilia. In the absence of DZIP1 or FAM92, the daughter centriole is more often found undocked compared with the mother one (in 79% and 75% of affected centriolar pair in fam92 and dzip1 mutants, respectively) . These results indicate that a functional asymmetry exists between mother and daughter centriole even though both are able to build a primary cilium. It is unclear why daughter centrioles are more prone to docking defects than the mother centrioles in spermatocytes. Until very recently, no molecular determinants were described to be enriched in one centriole versus the other in spermatocytes. However, the discovery that Sas4 and Sas6 are enriched at the base of the daughter centrioles  could open new avenues to understand the intrinsic differences between the two centrioles in these cells. These differences between mother and daughter centrioles are also associated with differences in the timing of basal body docking, as mother centrioles dock before daughter centrioles . This timing could also better stabilize the TZ of the mother centriole and so increase its anchoring properties.
In summary, the identification of the DZIP1–CBY–FAM92 pathway reveals an evolutionarily conserved role of these TZ proteins, not only restricted to gating proteins in and out of the cilium, but also involved in initiating and fostering axonemal elongation to trigger TZ assembly and ciliary budding. These proteins also reveal:
Functional differences in the importance of this TZ complex in different tissues or cell types. Understanding how the evolution of different TZ complexes in different organisms contributes to specific aspects of ciliary function is an important future perspective.
During Drosophila male meiosis, basal body to membrane attachment is involved in regulating the growth of the axoneme revealing a regulatory role of the membrane compartment in axonemal microtubule assembly. How these membrane-basal body communications regulate ciliary budding is still an important challenge in the field of ciliogenesis.
The authors declare that there are no competing interests associated with the manuscript.
Work was funded by a grant from the Agence Nationale de la Recherche, ANR-17-CE13-0023-01 DIVERCIL.
These authors contributed equally to this work.