Centriolar satellites are non-membranous cytoplasmic granules that concentrate in the vicinity of the centrosome, the major microtubule-organizing centre (MTOC) in animal cells. Originally assigned as conduits for the transport of proteins towards the centrosome and primary cilium, the complexity of satellites is starting to become apparent. Recent studies defined the satellite proteome and interactomes, placing hundreds of proteins from diverse pathways in association with satellites. In addition, studies on cells lacking satellites have revealed that the centrosome can assemble in their absence, whereas studies on acentriolar cells have demonstrated that satellite assembly is independent from an intact MTOC. A role for satellites in ciliogenesis is well established; however, their contribution to other cellular functions is poorly understood. In this Review, we discuss the developments in our understanding of centriolar satellite assembly and function, and why satellites are rapidly becoming established as governors of multiple cellular processes. We highlight the composition and biogenesis of satellites and what is known about the regulation of these aspects. Furthermore, we discuss the evolution from thinking of satellites as mere facilitators of protein trafficking to the centrosome to thinking of them being key regulators of protein localization and cellular proteostasis for a diverse set of pathways, making them of broader interest to fields beyond those focused on centrosomes and ciliogenesis.
The centrosome is composed of a pair of centrioles surrounded by pericentriolar material (PCM) and serves as the primary microtubule-organizing centre (MTOC) in animal cells. Centriolar satellites are an array of non-membranous granules that surround the centrosome in most vertebrate cells (Fig. 1A). Satellites were first observed 60 years ago as electron-dense particles of 70–100 nm in diameter in electron microscopy sections of centrosomes (Bernhard and De Harven, 1960; Bessis and Breton-Gorius, 1958; de-Thé, 1964).
However, molecular insight into the properties of satellites only began when the satellite protein PCM1 was identified in human cells using human autoimmune antiserum (Balczon et al., 1994) (Fig. 1B). Subsequent cloning and characterization of PCM1 demonstrated that satellites move along microtubules in a dynein-dependent manner towards the centrosome, are required for the assembly of centrosomal proteins, and undergo cell cycle-dependent assembly and disassembly (Dammermann and Merdes, 2002; Kubo and Tsukita, 2003; Kubo et al., 1999). Additionally, immunogold electron microscopy showed PCM1 localizing to the electron-dense particles found around centrosomes and also to the fibrous granules formed during ciliogenesis in differentiated multiciliated cells (Kubo et al., 1999). Consequently, PCM1 staining has been used to define satellites in cells. Since then, additional proteins have been shown to localize to satellites through colocalization with PCM1, including Bardet–Biedl syndrome 4 protein (BBS4), centrosomal protein of 131 kDa (CEP131), CEP290, and oral-facial-digital syndrome 1 protein (OFD1) (Bärenz et al., 2011; Kim et al., 2004, 2008; Lopes et al., 2011; Staples et al., 2012) (see Table 1).
The primary cilium (see Box 1, Fig. 1A) is central to a number of signalling pathways, including those mediated by Hedgehog, Wnt, Notch, Hippo, G-protein coupled receptors (GPCRs), platelet-derived growth factor (PDGF), mammalian target of rapamycin (mTOR) and TGF-β (Wheway et al., 2018), with ciliary dysfunction manifesting as a range of conditions termed ciliopathies (for a review, see Reiter and Leroux, 2017). Primary cilia are microtubule-based organelles that are nucleated from one of the centrioles that constitute the centrosome, thereby placing satellites in close association with cilia (for a review, see Malicki and Johnson, 2017; Fig. 1A). Indeed, mutations in genes encoding satellite components have been identified as causing ciliopathies such as Bardet–Biedl, Joubert, Meckel Gruber, and oral-facial-digital syndromes (Chevrier et al., 2016; Coene et al., 2009; Kim et al., 2004, 2008; Lopes et al., 2011; Romio et al., 2003; Stephen et al., 2017; Valente et al., 2006; Reiter and Leroux, 2017). In addition, satellite-resident proteins have been linked to other human disorders, including microcephaly (MCPH), psychiatric illnesses and Huntington's disease (Datta et al., 2010; Kamiya et al., 2008; Keryer et al., 2011; Kodani et al., 2015). Therefore, satellite integrity is necessary for human development and the physiology of several organs in the adult. This might reflect the requirement for satellites in the regulation of various fundamental aspects of both centrosome and cilia assembly and function (Tollenaere et al., 2015a). Their exact function in these processes, however, remains poorly understood, whereas the cellular function of satellites has broadened to roles in autophagy and actin filament nucleation and organization, and potentially other pathways (Farina et al., 2016; Obino et al., 2016; Pampliega et al., 2013; Tang et al., 2013).
The primary cilium is a microtubule-based organelle extending from the apical surface of most animal cells, where it acts as a cellular ‘antenna’ to receive signals from the extracellular environment (Malicki and Johnson, 2017). Primary cilia are nucleated from mother centriole-derived basal bodies during G0 or quiescence and consist of a microtubule-based axoneme surrounded by a specialized membrane that harbours signal receptors (Kim and Dynlacht, 2013). Assembly of a primary cilium, termed ciliogenesis, is a tightly regulated multistep process entirely dependent on the presence of a mature mother centriole decorated with appendages (Joukov and De Nicolo, 2019). The distal appendages promote membrane docking (Tanos et al., 2013), whereas the sub-distal appendages are thought to aid in cilium positioning (Uzbekov and Alieva, 2018). Early during ciliogenesis, pre-ciliary vesicles are transported to the basal body, associating with the distal appendages and fusing to form a larger ciliary vesicle (Lu et al., 2015; Wu et al., 2018). Ciliary membrane proteins, including ADP-ribosylation factor-like protein 13B (ARL13B) and components of the RAB8A–RAB11A GTPase cascade become enriched within the ciliary vesicle. The centriolar protein TALPID3 interacts with RAB8A and is required for the efficient recruitment of ciliary vesicles to the basal body (Kobayashi et al., 2014; Wang et al., 2016). RAB11A recruits RABIN8 [also known as Rab-3A-interacting protein (RAB3IP)], which in turn activates RAB8A (Knödler et al., 2010). Together, RAB8A and ARL13B promote ciliary membrane expansion and selective trafficking of proteins to the cilium. Following removal of CP110 (also known as CCP110) from the distal end of the basal body, the microtubules of the centriole can extend to form the axoneme of the cilium (Spektor et al., 2007). Subsequently, the transition zone that partitions the cilium from the cell body forms. The BBSome and intra-flagellar transport (IFT) machinery is recruited to the distal appendages and mediates trafficking of ciliary axonemal proteins (Mukhopadhyay et al., 2017). Within the cilium, the BBSome acts as an adaptor between ciliary membrane proteins and IFT particles that are cycling through the cilium via IFT (Wingfield et al., 2018). BBSome ciliary function is mediated through RAB8A, while the protein complex is imported into the cilium in an ARL6-dependent manner (Mourão et al., 2014; Nachury et al., 2007). The cilium then grows to a steady-state length through ongoing trafficking of components to and from the cilium (Breslow and Holland, 2019). Ciliation failure or dysfunction of primary cilia is the root cause for a range of disorders called ciliopathies.
Recently, we have gained a more comprehensive understanding of satellite composition through proteomic and interactome analysis of these enigmatic organelles (Gheiratmand et al., 2019; Gupta et al., 2015; Quarantotti et al., 2019), placing hundreds more proteins in close association with satellites beyond the already established ∼30 factors (Tollenaere et al., 2015a) (Fig. 2A). Additionally, detailed analysis of the cellular proteome in cells devoid of satellites has revealed new pathways in which satellites might function (Odabasi et al., 2019). In this Review, we describe centriolar satellite biogenesis and the known regulatory pathways that influence the satellite proteome and protein activity or abundance. Afterwards, we discuss the relationship of satellites with centrosomes and primary cilia before bringing together the findings of recent studies that are driving the emerging concept that satellites are key regulators of multiple processes and proteostasis for a diverse set of pathways.
Centriolar satellite composition, biogenesis and regulation
The satellite–centrosome relationship and emergence of satellite protein subgroups
Centriolar satellites are more complex than originally anticipated, with over 65 proteins now assigned to these structures by cellular localization (see Table 1). This number is likely to increase substantially in the future if candidates that were identified in the satellite proteome or interactome are validated as bona fide satellite components (Gheiratmand et al., 2019; Quarantotti et al., 2019). PCM1 has long been considered a satellite scaffold protein, interacting with at least 20 other satellite proteins (Hori and Toda, 2017). Recently, proximity interaction mapping of 22 satellite bait proteins has identified 84 proteins that, alongside PCM1, form a highly connected network of proteins that has ∼40% overlap with the satellite proteome (Gheiratmand et al., 2019; Quarantotti et al., 2019) (Fig. 2A). Although these results expose extensive interactions between satellite-associated proteins, how and where satellites assemble remains poorly understood.
Satellites are closely associated with microtubules and the centrosome, with disruption of the microtubule network resulting in satellite dispersal away from the centrosome (Dammermann and Merdes, 2002; Kubo and Tsukita, 2003; Kubo et al., 1999) and depletion of components of the microtubule machinery reducing satellite intensity in the vicinity of centrosomes (Gupta et al., 2015). In addition to mediating minus-end-directed transport of satellites to the vicinity of centrosomes, it is plausible to consider that microtubules may also play a role in satellite assembly. Strikingly though, in cells treated with the microtubule-depolymerizing drug nocodazole, PCM1 is still able to interact and colocalize with several satellite components, including CEP290, CEP131, OFD1, coiled-coil domain-containing protein 14 (CCDC14), progesterone-induced-blocking factor 1 (PIBF1) and CEP72 (Gheiratmand et al., 2019). This suggests that an intact microtubule network is not necessary for satellite assembly (Fig. 2B). But what about centrosomes? Their requirement for satellite formation was investigated in acentriolar cells that were generated by either genetic knockout of key centriole duplication factors in chicken cells (Quarantotti et al., 2019), or prolonged treatment of human cells with centrinone (Gheiratmand et al., 2019), which inhibits polo-like kinase 4 (PLK4), the master regulator of centriole duplication (Wong et al., 2015). Both approaches prevent new centriole formation, thereby devoiding cells of a functional MTOC. In the absence of an MTOC, microtubules fail to focus at a single point and satellites become dispersed (Khodjakov et al., 2000). However, in acentriolar cells, the satellite proteome was largely unaltered (Quarantotti et al., 2019), and most interactions between satellite and centriole components were maintained (Gheiratmand et al., 2019). Together, these results indicate that satellite integrity and assembly are largely independent of functional centrosomes (Fig. 2B).
Given the growing number of satellite constituents and cellular processes that they might have a role in, it is possible that satellites are composed of different subgroups of proteins to serve different functions in the cell. For example, centriolar satellites localize MCPH-associated proteins to the centrosome; however, distinct satellite components interact with and are required for the centrosomal localization of different MCPH-associated proteins (Kodani et al., 2015). Additionally, CEP89 colocalizes with all PCM1-containing granules in cells, but fewer OFD1 and CEP290-containing satellites colocalize with CEP89 (Sillibourne et al., 2013), suggesting the protein content of individual satellites is not always equivalent. Recent proteomic and super-resolution microscopy spatial profiling provided further support for this idea by revealing discernible sub-populations of satellites and diversity in satellite composition (Gheiratmand et al., 2019), and the possibility that PCM1 is only present on a subset of satellites or associates more transiently with some populations of satellites than others (Gheiratmand et al., 2019).
Despite satellites appearing to contain different amounts of PCM1, they all seem to be sensitive to PCM1 depletion from cells: removal of PCM1 leaves cells devoid of satellites and restricts satellite proteins to the centrosome (Gheiratmand et al., 2019; Hoang-Minh et al., 2016; Lopes et al., 2011; Odabasi et al., 2019; Quarantotti et al., 2019; Wang et al., 2013). Therefore, PCM1 might be needed to assemble satellites, but not to maintain them; PCM1 might be lost over time, or it might associate with satellites in a dynamic manner that has yet to be captured experimentally. Strikingly, despite disruption of satellites upon knockout of PCM1, a large number of the interactions between satellite proteins persist in the absence of PCM1 (Gheiratmand et al., 2019). This suggests that these interactions can occur at the centrosome in cells lacking satellites or that complexes of satellite proteins can assemble independently of PCM1, possibly as precursors to which PCM1 would then be recruited to in order to form functional satellites.
The regulation of the abundance and activity of satellite proteins
Satellites disappear during mitosis, even though PCM1 levels remain constant throughout the cell cycle (Kubo and Tsukita, 2003). Therefore, post-translational modifications of centriolar satellite proteins likely plays a fundamental role in the regulation of satellite integrity. Indeed, phosphorylation of PCM1 at serine 372 by PLK4 during G1 mediates its dimerization and interaction with BBS4 and CEP290 (Hori et al., 2016) (Fig. 2C). Consequently, knockdown of PLK4 causes the dispersal of satellites. PCM1 is also phosphorylated by other kinases, including cyclin-dependent kinase 1 (CDK1) and PLK1 (Hori and Toda, 2017; Olsen et al., 2010; Santamaria et al., 2011; Wang et al., 2013). Thus, phosphorylation of PCM1 by these cell cycle-regulated kinases might contribute to the cell cycle-dependent remodelling of satellites and, in particular, their dispersal in mitosis (Hori and Toda, 2017; Kubo et al., 1999; Zhang et al., 2017). Dispersal of several membrane-less organelles, including satellites, has also been shown to be triggered by dual-specificity tyrosine-phosphorylation-regulated kinase 3 (DYRK3) (Rai et al., 2018) (Fig. 2C).
Phosphorylation of satellite components further plays a role in the cellular stress response. Centriolar satellites are reported to undergo a striking reorganization upon stress with the loss of PCM1, CEP131 and CEP290, but not OFD1, from satellites; this is triggered in a p38α (also known as mitogen-activated protein kinase 14, MAPK14)-dependent manner (Villumsen et al., 2013) (Fig. 2C). Specifically, phosphorylation of CEP131 at serine 47 and 78 by the p38-activated kinase MK2 (also known as MAPKAPK2) creates a dual binding site for 14-3-3 proteins when cells are exposed to ultraviolet light-induced stress (Tollenaere et al., 2015b). The interaction with 14-3-3 mediates cytoplasmic sequestration of CEP131, thereby blocking de novo satellite formation.
Aside from phosphorylation, ubiquitylation also plays a vital role in the regulation of satellite integrity. The E3 ligase Mind-bomb1 (MIB1) is a bona fide centriolar satellite protein that interacts with known satellite components and has been demonstrated to decorate a subset of satellite proteins with ubiquitin (Čajánek et al., 2015; Dho et al., 2019; Firat-Karalar et al., 2014; Gupta et al., 2015; Jakobsen et al., 2011; Villumsen et al., 2013; Wang et al., 2016). MIB1 promotes the poly-ubiquitylation of PCM1 and CEP131, leading to their destabilization (Wang et al., 2016) (Fig. 2D). In line with this, removal of MIB1 caused the stabilization of PCM1 and CEP131, thereby enhancing the stability of centriolar satellites. MIB1 has also been reported to mediate the mono-ubiquitylation of PCM1, CEP131 and CEP290, which has been proposed to maintain satellite integrity under normal conditions through an unknown mechanism (Villumsen et al., 2013). In support of this, MIB1 becomes inactivated in response to cellular stress, leading to the loss of PCM1, CEP131 and CEP290 ubiquitylation, which might contribute to satellite reorganization following cellular insult (Villumsen et al., 2013). Additionally, MIB1 can promote the ubiquitylation of the satellite components CCDC14, KIAA0753 (also known as protein moonraker) and OFD1; whereas these modifications did not lead to the degradation of these proteins, they could potentially also contribute to the maintenance of satellite integrity (Dho et al., 2019) (Fig. 2D).
Deubiquitylating enzymes (DUBs) counteract the action of ubiquitin ligases to promote stability of their target proteins. Two DUBs, cylindromatosis (CYLD) and probable ubiquitin carboxyl-terminal hydrolase FAF-X (USP9X), have recently been identified as regulators of centriolar satellite proteostasis (Douanne et al., 2019; Han et al., 2019; Li et al., 2017; Wang et al., 2017). Knockdown of CYLD significantly reduced the cellular abundance of PCM1 and CEP131, which was partly reversed by treatment with a proteasome inhibitor (Douanne et al., 2019). This suggests that CYLD acts to prevent the proteasomal degradation of PCM1 and CEP131 (Fig. 2D). MIB1 undergoes auto-ubiquitylation, with increased ubiquitylation of MIB1 being observed when CYLD is depleted (Douanne et al., 2019; Wang et al., 2016). Importantly, co-depletion of MIB1 in CYLD-silenced cells prevented PCM1 disappearance and rescued satellite organization (Douanne et al., 2019). Thus, CYLD-mediated deubiquitylation of MIB1 prevents the E3 ligase from promoting the proteasomal degradation of PCM1. In contrast, USP9X binds to, and directly acts upon, PCM1 and CEP131 to deubiquitylate them and antagonize their degradation (Han et al., 2019; Li et al., 2017) (Fig. 2D). Silencing of USP9X reduces the cellular levels of both PCM1 and CEP131, disrupting satellites and leading to the localization of CEP290 to the centrosome (Han et al., 2019; Li et al., 2017; Wang et al., 2017). Co-depletion of MIB1 in USP9X-silenced cells largely rescued the effects of USP9X loss (Han et al., 2019). Together, this provides strong support for ubiquitylation of satellite proteins playing a crucial role in satellite stability and turnover.
Additional DUBs and ubiquitin ligases might also participate in the maintenance of centriolar satellite integrity. In support of this, the E3 ubiquitin ligase UBR5 was found to ubiquitylate the satellite component CSPP1 (centrosome and spindle pole associate protein 1), promoting both its localization to satellites and satellite organization (Shearer et al., 2018), and a number of conserved E3 ubiquitin ligases were identified in the satellite proteome (Quarantotti et al., 2019). The hexameric AAA ATPase vacuolar protein sorting-associated protein 4A (VPS4) has also been shown to regulate PCM1, with expression of an inactive form leading to the loss of satellites around the centrosome and reduction in cellular PCM1 levels (Ott et al., 2018). Conversely, autophagy selectively removes OFD1 from satellites to promote ciliogenesis (Tang et al., 2013) and regulates satellite turnover through PCM1 degradation (Holdgaard et al., 2019).
Given the number of proteins localizing to satellites, it is unsurprising that the regulation of satellite integrity is emerging to be a multifaceted and complex process. The use of distinct, but sometimes overlapping, pathways to regulate satellite integrity allows them to undergo remodelling in response to both the cell cycle and cellular insults. This in turn is likely to regulate their function at those times.
Centriolar satellites as conduits for centrosomal components
The centrosome is a complex molecular assembly that, as the major MTOC in cells, plays an instrumental role in a plethora of cellular processes, including cell motility, intracellular transport, mitotic spindle assembly and cell division (Conduit et al., 2015). As a single-copy organelle, the centrosome and its centriole pair undergo a single round of duplication once per cell cycle (see Box 2). Since the discovery that satellites move along microtubules towards the centrosome (Kim et al., 2008; Kubo et al., 1999), and cluster in its vicinity, research has focused on the role satellites play in centrosome assembly and/or function. Consequently, satellites have been shown to mediate the function of centrosomal proteins through either targeting proteins to the centrosome or sequestering them to limit their centrosomal recruitment. For example, the amount of centrin, pericentrin and ninein is reduced when satellites are perturbed through knockdown of PCM1 (Dammermann and Merdes, 2002). Additionally, disruption to the microtubule network was observed when PCM1, BBS4 or CEP290 were silenced in cells (Dammermann and Merdes, 2002; Kim et al., 2004, 2008). In the case of BBS4, cell cycle progression was also hindered (Kim et al., 2008). Centrosome number is tightly controlled, with dysregulation being linked to a number of disease states (Fujita et al., 2016) (see Box 2). Satellites have also been implicated in centrosome number control through regulating centriole number: depletion of the satellite components SSX2-interacting protein (SSX2IP, also known as ADIP), CEP72, sperm-associated antigen 5 (SPAG5), CEP131, KIAA0753 and CEP90 perturbs centriole duplication (Hori et al., 2015; Kodani et al., 2015). Conversely, depletion of CCDC14 leads to an increase in the number of centrioles (Kodani et al., 2015). Together, this demonstrates that satellites can exert both positive and negative effects on centriole number.
Centrosomes are composed of a pair of microtubule-based barrel-shaped centrioles surrounded by PCM. The older of the two centrioles, termed the mother, is distinguished by the presence of two sets of appendages at its distal end, whereas the two centrioles are linked by a tether at their proximal ends (Nigg and Holland, 2018). Centrosomes duplicate once and only once per cell cycle, ensuring that two centrosomes are present during mitosis (Conduit et al., 2015). Centrioles define the site where PCM is recruited and, thus, the number of centrosomes in a cell (Gönczy, 2012). Failure to regulate centrosome number is linked to aneuploidy and cancer formation (Nigg and Holland, 2018). Centrioles duplicate in a highly regulated process that initiates at G1/S with the formation of a procentriole at the proximal end of each of the pre-existing centrioles. Procentriole formation is driven by localized recruitment and concentration of PLK4 (Bettencourt-Dias et al., 2005; Habedanck et al., 2005; Yamamoto and Kitagawa, 2019), which is mediated by CEP192, CEP152 and CEP63 (Brown et al., 2013; Cizmecioglu et al., 2010; Kim et al., 2013; Sonnen et al., 2013). Phosphorylation of STIL by PLK4 drives SAS6 to form a cartwheel structure that defines the nine-fold symmetrical structure of the centriole (Guichard et al., 2017; Leda et al., 2018). Centrosomal P4.1-associated protein (CPAP) and CEP135 then initiate centriole elongation, which continues through S and G2 (Gönczy and Hatzopoulos, 2019; Loncarek and Bettencourt-Dias, 2018). In G2 phase, the centrosome grows in size and microtubule nucleation capacity, a process coined centrosome maturation, through the recruitment of additional PCM (Palazzo et al., 2000). Centrosome separation occurs concomitantly with maturation and is driven by dissolution of the tether linking the two parental centrioles. Separation allows the centrosomes to move apart to form the spindle poles and, thereby, ensures each daughter cell inherits a single centrosome (Agircan et al., 2014). When a cell enters G0, the centrosome undergoes morphological and functional alterations, leading to the formation of a primary cilium. Ciliogenesis is part of the canonical centrosome cycle in most proliferating cells (Joukov and De Nicolo, 2019).
In chicken cells, nearly half of the known centrosome proteome can be found in centriolar satellites (Quarantotti et al., 2019), whereas ∼40% of the satellite interactome overlaps with that of the centrosome in human cells (Gheiratmand et al., 2019). Significantly, the abundance of the centrosome proteome does not change upon the loss of satellites and centriole duplication, cell proliferation, and cell cycle progression are all unaffected in cells lacking satellites (Odabasi et al., 2019) (Fig. 2B). Thus, the primary function of satellites is likely not to be directly related to centrosome biogenesis. Indeed, when satellites were disrupted through loss of PCM1, satellite proteins were still able to localize at the centrosome (Gheiratmand et al., 2019; Hoang-Minh et al., 2016; Lopes et al., 2011; Odabasi et al., 2019; Quarantotti et al., 2019; Wang et al., 2013). Additionally, BBS4, OFD1 and CEP290 have been shown to persist at the centrosome during mitosis when PCM1 disperses (Kim et al., 2004; Lopes et al., 2011). This suggests that proteins are capable of associating with the centrosome in a PCM1- and, therefore, satellite-independent manner. However, despite normal expression levels, the centriolar proteins CEP41, CEP112 and CEP135 were reduced in the satellite proteome of acentriolar cells, indicating a centriole-dependent association with PCM1 (Quarantotti et al., 2019).
Considering the significant overlap between the centrosome and satellite proteomes, the abundance of a protein at one organelle does not necessarily reflect its abundance in the other. For example, the centriole-assembly factor CEP152 is one of the most abundant proteins at the centrosome (Bauer et al., 2016), but is not found in the satellite proteome. By contrast, CEP290 is one of the least abundant centrosomal proteins, but is a well-established satellite component (Quarantotti et al., 2019). This implies that regulatory mechanisms exist to account for the enrichment of proteins in each compartment. In addition to CEP152, two other proteins, spindle assembly abnormal protein 6 homolog (SAS6, also known as SASS6 in mammals) and SCL-interrupting locus protein (STIL), which together are critical for driving centriole assembly, are absent from centriolar satellites (Quarantotti et al., 2019). This may simply reflect the low cellular abundance of these proteins or hint at the exclusion of centriole assembly initiation factors from satellites to prevent ectopic centriole formation at these sites.
Thus, the role of satellites is emerging to be far more nuanced than being simple facilitators of protein trafficking. However, cell or tissue specificities are possible, especially given the discovery of the hierarchical recruitment of MCHP proteins to the centrosome by satellites to drive centriole duplication (Kodani et al., 2015). Despite this, one thing that cell lines lacking satellites have resoundingly informed us on is the role of satellites in cilia assembly and function.
The role of centriolar satellites in ciliation and ciliary functions
The aforementioned relationship between mutations in a number of genes that encode satellite proteins and ciliopathies (Mykytyn et al., 2001; Romio et al., 2003; Valente et al., 2006), and the ciliopathy-like phenotypes seen upon loss of centriolar satellites in zebrafish (Stowe et al., 2012), strongly suggest a central role for satellites in either ciliation or ciliary-related functions. Indeed, knockdown of various satellite proteins revealed that ciliogenesis is perturbed in the absence of satellite components (Conkar et al., 2017; Kim et al., 2008; Klinger et al., 2014; Lee and Stearns, 2013; Mikule et al., 2007; Staples et al., 2014).
Ciliogenesis is a complex, multistep process (see Box 1) and it is emerging that satellites contribute to numerous points in the pathway (Fig. 3). The BBSome is a ciliary trafficking complex of seven proteins found mutated in the ciliopathy Bardet–Biedl syndrome, and BBSome-interacting protein 10 (BBIP10), which together are required for ciliation (Loktev et al., 2008; Nachury et al., 2007). Intriguingly, five members of the BBSome have now been identified in the satellite proteome (Quarantotti et al., 2019). BBS4 localizes to both satellites and the BBSome, suggesting that it might provide a molecular link between the two protein assemblies. Knockdown of PCM1, CEP131, CCDC13 and CCDC66 reduced recruitment of BBS4 to the cilium (Conkar et al., 2017; Staples et al., 2014; Stowe et al., 2012), whereas SSX2IP mediates RAB8-enabled BBSome targeting (Klinger et al., 2014). Significantly, acute (siRNA-mediated) loss of the satellite protein CEP131 reduced the rate of ciliation whereas chronic loss (gene deletion) did not (Hall et al., 2013), suggesting a non-essential role in ciliogenesis. Indeed, depletion of CEP131 increased the amount of BBSome proteins, other than BBS4, localizing to cilia (Chamling et al., 2014), demonstrating that satellites can function as both positive and negative regulators of ciliogenesis through the control of recruitment of the BBSome to the primary cilium.
Aside from driving the recruitment of proteins required for ciliogenesis, satellites also function to sequester negative regulators away from the basal body during cilia formation. TALPID3 (also known as KIAA0586) is a centriolar protein that is essential for ciliogenesis; through RAB8, TALPID3 ensures the efficient recruitment of ciliary vesicles to the basal body (Kobayashi et al., 2014). In agreement with knockdown phenotypes, PCM1−/− human cells are defective in ciliogenesis (Wang et al., 2016). PCM1 tethers MIB1 to satellites; however, in absence of PCM1, MIB1 is able to poly-ubiquitylate and destabilize TALPID3, thereby abrogating ciliary vesicle recruitment. In fact, PCM1 loss affects the abundance of only two satellite components: MIB1 levels are elevated, whereas SSX2IP levels drop (Quarantotti et al., 2019). By impacting a suppressor and an activator of ciliogenesis, respectively, these changes would collude to block ciliogenesis in the absence of PCM1. Intriguingly, TALPID3 was detected in the satellite proteome, suggesting MIB1 is unlikely to target TALPID3 at this location if they occupy the same satellite populations (Quarantotti et al., 2019).
In mouse inner medullary collecting duct (IMCD3) Pcm1-knockout cells, the centrosomal levels of MIB1 were comparable to control cells, whereas the amount of TALPID3 at the centrosome was increased (Odabasi et al., 2019). This potentially explains why TALPID3 degradation, as directed by MIB1 at the centrosome, was not induced in these cells lacking satellites. As human retinal pigment epithelial (RPE1) PCM1-knockout cells show a more complete block to ciliogenesis than IMCD3 cells, this difference suggests a cell type-dependent variability of the centrosomal targeting of key ciliogenesis factors by centriolar satellites to control ciliation (Odabasi et al., 2019; Wang et al., 2016). IMCD3 cells lacking satellites that were able to ciliate had cilia of equivalent length to control cells, supporting the idea that satellites are required to initiate assembly of the ciliary axoneme. Indeed, defective recruitment of intra-flagellar transport protein 88 (IFT88) in Pcm1-knockout IMCD3 cells places satellites upstream of this step of ciliogenesis (Odabasi et al., 2019). A functional relationship between satellites and the IFT machinery is further supported by the identification of putative interactions between satellites and IFT components (Gupta et al., 2015; Huttlin et al., 2017, 2015). The fraction of Pcm1−/− IMCD3 cells within the population that were able to form cilia also informs us that the recruitment of ciliary membrane and shaft proteins is altered in cells lacking satellites (Odabasi et al., 2019). This supports a role for PCM1 in the ciliary recruitment of proteins that are required for different stages of ciliogenesis and suggests that defects in their recruitment might be the underlying cause of satellite-related ciliogenesis defects in disease.
Beyond regulating protein recruitment for ciliogenesis, satellites have also been implicated in ciliary function (Fig. 3). Functional primary cilia are crucial for the transduction of the Hedgehog signalling pathway (Goetz et al., 2009; Wheway et al., 2018). IMCD3 cells lacking satellites that could form cilia had a reduced ability to accumulate the Hedgehog receptor Smoothened into cilia in response to a Hedgehog pathway agonist and, downstream of this, failed to activate expression of the Hedgehog target gene Gli1 (Odabasi et al., 2019). This suggests that – in addition to efficient ciliation – satellites are also required for activation of the Hedgehog pathway. In support of this, 30 regulators of Hedgehog signalling were found to overlap with the satellite proteome, intricately linking satellites to this key function of cilia (Breslow et al., 2018; Quarantotti et al., 2019). Primary cilia are also required for the establishment of the highly organized architecture and apicobasal polarity of epithelial cells in 3D (Delous et al., 2009; Mahjoub and Stearns, 2012; Otto et al., 2010). This function can be assayed in vitro by growing epithelial cells in a 3D gel matrix, whereupon the cells will organize into polarized, spheroid structures that mimic the in vivo organization of epithelial tissues (Torras et al., 2018). Cells lacking satellites were found to have a reduced ability to form properly organized spheroids as compared to control cells (Odabasi et al., 2019). This correlated with a ciliogenesis defect, supporting a vital role for satellites in proper cilia formation and signalling, which impacts on epithelial organization in 3D.
Taken together, satellites have critical roles at numerous points during primary cilium assembly, and also contribute to proper ciliary function. The importance of satellites in these roles is reflected in the number of human ciliopathies that satellite genes are mutated in and can help us better understand the pathology of ciliopathy diseases.
Centriolar satellite function in proteostasis and beyond
The significant rearrangement of the global proteome in cells lacking satellites indicates that satellites regulate cellular proteostasis (Odabasi et al., 2019). Specifically, proteins related to the actin cytoskeleton were enriched, in line with previous reports that satellites target actin assembly factors to the centrosome to allow it to act as an actin-nucleating centre (Farina et al., 2016; Obino et al., 2016). This also raises the possibility that satellites function in actin-regulated ciliary process, including pre-ciliary vesicle transport to the basal body and cilia length regulation (Kim et al., 2015, 2010; Kohli et al., 2017; Wu et al., 2018). There was also an increase in the level of proteins linked to neurogenesis pathways, including neuronal body and extensions, postsynaptic membrane, and axon guidance. For example, PCM1, HOOK3, disrupted in schizophrenia 1 protein (DISC1), and serologically defined colon cancer antigen 8 (SDCCAG8) have been implicated in neuronal progenitor cell maintenance and neuronal migration during cortical development (Insolera et al., 2014). Furthermore, mutations in PCM1 have been associated with schizophrenia (Datta et al., 2010; Gurling et al., 2006).
Regulated protein degradation, by means of autophagy and the ubiquitin proteasome system, is pivotal to development and the maintenance of cellular homeostasis (for a review, see Dikic, 2017). Previous studies have demonstrated that satellites regulate the stability of proteins (Joachim et al., 2017; Wang et al., 2016), suggesting that satellites function as bona fide regulators of protein degradation (for a review, see Lecland and Merdes, 2018) (Fig. 3). Indeed, MIB1 was found to be upregulated in cells lacking satellites, therefore corroborating the crosstalk between PCM1 and MIB1 stability (Akimov et al., 2011; Joachim et al., 2017; Odabasi et al., 2019; Villumsen et al., 2013; Wang et al., 2016). Furthermore, depletion of OFD1 or BBS4 led to the accumulation of Wnt signalling mediators such as β-catenin and Dishevelled (Gerdes et al., 2007; Liu et al., 2014). OFD1 and BBS proteins interact with proteasomal subunits, with loss of OFD1 or BBS4 leading to depletion of multiple subunits from the centrosomal proteasome. Moreover, the signalling defects in OFD1- or BBS4-depleted cells could be restored through overexpression of proteasomal subunits or chemical activation of the proteasome (Liu et al., 2014). Importantly, proteasome-dependent pathways that are not associated with ciliopathies were also defective in the absence of OFD1 and BBS4; for example, nuclear factor (NF)-κB activity was decreased, which correlated with a concomitant increase in levels of the cytoplasmic inhibitor of NF-κB, IκBβ (also known as NFKBIB) (Liu et al., 2014). As autophagy has been reported to regulate OFD1 levels (Tang et al., 2013), these results place OFD1 at the crosstalk between the two protein degradation systems. Furthermore, given the link between PCM1 and Huntington's disease related proteins Huntingtin (HTT) and Huntingtin-associated protein 1 (HAP1) (Keryer et al., 2011), it is tantalizing to speculate that satellites play a role in proteostasis- and aggregation-related diseases.
Analysis of the satellite interactome also places tubulin modifiers, components of P-bodies and members of the microtubule-based microtubule-nucleation-mediating HAUS complex in these organelles (Gheiratmand et al., 2019). Together, these results link satellites to a diverse range of cellular process (Fig. 3). It will be intriguing to follow future studies as the molecular function of satellites in these pathways become unravelled.
Conclusions and future directions
Here, we discussed recent studies that have revealed a much more complex molecular and functional relationship between satellites and centrosomes and cilia than previously thought. Indeed, contrary to early reports, satellites appear to be dispensable for the assembly of a functional centrosome. Despite this, specific roles for satellites in building functional cilia have been revealed, including requirements for efficient ciliogenesis, regulation of ciliary content, and response to the Hedgehog pathway. In addition, roles for satellites in more diverse pathways have been implicated, placing satellites as regulators of protein targeting and cellular proteostasis beyond the centrosome (Fig. 3).
Establishment of the satellite proteome and interactome (Gheiratmand et al., 2019; Quarantotti et al., 2019) is paving the way for elucidating the functional relevance of satellites. These studies have revealed that satellites and centrosomes form independently, but share a substantial fraction of proteins and interactions. This implies that satellites could provide a reservoir of centrosomal proteins when transcription is restricted or in response to cellular stress. Indeed, current evidence suggests that satellites are capable of modulating the centrosomal recruitment of proteins. Related to this is the establishment of different subpopulations of satellites, although the functional relevance of these is still unclear at present. Future studies on satellites in different contexts will be important for identifying the full range of satellite functions. For example, satellite composition and interactors may vary during different phases of the cell cycle, in myoblasts, where PCM1 is perinuclear (Srsen et al., 2009), and in differentiated multiciliated cells, where satellites are related to the fibrous granules assembled during ciliogenesis. This has the potential to reveal the specific functions of satellites in different cell and tissue types, thereby defining how satellites contribute to the prevention of disease states. Identification of changes to the cilium proteome and interactome in cells lacking satellites will thus be required to elucidate the complete list of cilium proteins regulated by satellites.
Despite increased understanding of satellite composition, it is still unclear as to how and where satellites assemble. Key to solving this will be establishing whether there is hierarchical recruitment of satellite components and what shapes the morphology and size of satellites. In addition, future studies may consider the potential conservation of satellite-like structures in basal metazoans, which would help in the understanding of their complexity in vertebrates.
Satellite composition is sensitive to a range of different stresses, but satellite dynamics under various conditions has yet to be addressed. The report that satellites are sensitive to the DYRK3 kinase (Rai et al., 2018), which has dissolvase activity, opens the intriguing possibility that satellite assembly and disassembly is regulated by phase separation. Little is known about how or where satellites assembly; this, and how it relates to satellite function in a range of processes, will therefore be an interesting avenue for future research.
Work in the Pelletier lab is funded by the Canadian Institute for Health Research, The Canadian Cancer Society, Genome Canada, the Natural Sciences and Engineering Research Council of Canada, the Ontario Ministry for Research and Innovation, the Canada First Research Excellence Fund Medicine by Design (MbD) programme and the Krembil Foundation. L.P. holds a Tier 1 Canada Research Chair in Centrosome Biogenesis and Function.
The authors declare no competing or financial interests.