The role of SPIRE actin nucleators in cellular transport processes

Interconnected cells with diverse structures and functions are the underlying basis of animal life (Brunet and King, 2017). In order to generate, organise and dynamically maintain distinct cell types, animal cells have developed a sophisticated intracellular cargo transport system (Zhen and Stenmark, 2015), where forces are generated by cytoskeletal filaments and associated motor proteins (Dogterom and Koenderink, 2019). In this Review, we specifically focus on actomyosin force generation by SPIRE actin nucleators and the interacting class 5 myosin (MYO5) motor proteins in intracellular membrane transport processes (Alzahofi et al., 2020; Pylypenko et al., 2016; Schuh, 2011). The directed transport of organelles, proteins, RNAs and small molecules enables not only polarised cell structures, but also specific cell–cell communication processes required to orchestrate functional variety. RABs – small GTPases of the RAS superfamily – function as molecular switches as they regulate the anchoring of protein complexes at vesicle membrane transport carriers, ultimately defining their nature, transport routes and destinations (Zhen and Stenmark, 2015). A comprehensive genomic analysis has subdivided the eukaryotic RAB GTPases into six supergroups, which represent the diversity of transport routes (Klopper et al., 2012). Exocytic transport towards the plasma membrane is of specific importance for cell polarity and cell communication. Among the RABs related to exocytic transport, RAB3, RAB8 and RAB27 proteins, which are members of the secretory group (Fukuda, 2008, 2013), and RAB11 proteins of the recycling group (Welz et al., 2014) have been well studied.

Active intracellular membrane transport is mediated by motor proteins, which slide along cytoskeletal filaments. Motor proteins of the kinesin and dynein family mediate microtubule-based transport, and myosin motor proteins transport cargo along actin tracks (Langford, 1995; Ross et al., 2008). Considering the enrichment of actin filaments at the animal cell cortex and in cell protrusions such as filopodia, microvilli and dendritic spines, it is generally accepted that long-range transport is mainly facilitated by trafficking along microtubule tracks, whereas actin–myosin (actomyosin) transport rather mediates the local delivery of cargo in cortical regions (referred to as the highways and local roads model; Hammer and Sellers, 2011; Hume and Seabra, 2011). However, there are some exceptions to the principle of highways and local roads, where actin polymerisation-driven motility or actomyosin mediate long-range transport processes beyond the tracks of the cortical actin cytoskeleton (Rottner et al., 2017). For instance, actin comet tails, which are regulated by the Arp2/3 actin nucleator complex, have been shown to facilitate transport independent of the cortical actin cytoskeleton by using the pushing forces of actin polymerisation against the outer surfaces of the cargoes (Loisel et al., 1999; May et al., 1999; Welch et al., 1998). Here, the Arp2/3 complex generates a network of branched actin filaments (Mullins et al., 1998), which assemble at the cargo surface and disassemble at the tail in a treadmilling mechanism (Loisel et al., 1999). Intracellular motility driven by actin comet tails includes the propulsion of intracellular bacteria (including Listeria and Shigella) (Loisel et al., 1999; May et al., 1999; Welch et al., 1998), intracellular vesicles (Benesch et al., 2002) and autophagosomes (Kast et al., 2015). In terms of actomyosin-driven long-range transport, meshworks of linear actin filaments have been discovered in mouse oocytes and melanocytes, where they are generated by vesicle-bound SPIRE actin nucleators in cooperation with FMN-subgroup formins (also referred to herein as FMN) (Alzahofi et al., 2020; Pfender et al., 2011). The actin meshworks generated by SPIRE proteins and FMN-subgroup formins in mouse oocytes and melanocytes serve as tracks for class 5 myosin motor-dependent transport (Alzahofi et al., 2020; Pfender et al., 2011; Schuh, 2011). In addition to cytoplasmic actin meshworks, the SPIRE actin nucleators have been found to organise a variety of different actin structures, which include structures originating at the cell cortex and at mitochondria–endoplasmic reticulum (ER) contact sites (Holthenrich et al., 2022; Manor et al., 2015; Pfender et al., 2011; Scheffler et al., 2021). Mammalian genomes have two genes encoding SPIRE proteins (SPIRE1 and SPIRE2) and two genes encoding FMN-subgroup formins (FMN1 and FMN2) (Schumacher et al., 2004). The mammalian SPIRE genes translate into four proteins: three SPIRE1 isoforms and a single SPIRE2 protein (Kollmar et al., 2019 preprint) (Fig. 1A). A third vertebrate SPIRE gene (SPIRE3) is found in frogs, birds, fish and reptiles, but is absent in mammals, and the protein product is less conserved (Kollmar et al., 2019 preprint) (Fig. 1A).

Here, we summarise the current knowledge of SPIRE proteins in terms of structure, protein interactions, actin filament assembly activity and cell biological functions in order to establish a mechanistic framework underlying the multiple SPIRE-regulated actin structures. This Review will provide an in-depth account of SPIRE-mediated force generation in cellular membrane transport processes and mitochondria dynamics and describe the functions of SPIRE proteins in mammalian reproduction, skin pigmentation and wound healing, as well as the role of SPIRE in host–pathogen interactions.

SPIRE research began with the description of a Drosophila melanogaster (fruit fly) mutant in 1989, without knowing exactly which gene was affected and what the protein sequence and function might be (Manseau and Schupbach, 1989). During fly oocyte development, spire function is required for the correct localisation of determinants to the posterior pole and to the dorsal anterior corner (Wellington et al., 1999; Manseau and Schupbach, 1989). The spire-mutant oocytes display premature microtubule-dependent cytoplasmic streaming during mid-oogenesis, which is suppressed by a SPIRE-organised cytoplasmic actin meshwork in wild-type oocytes (see below) (Dahlgaard et al., 2007; Theurkauf, 1994; Wellington et al., 1999). It took a further ten years from the description of the mutant flies for the D. melanogaster SPIRE (Dm-SPIRE) protein sequence to be deduced (Otto et al., 2000; Wellington et al., 1999). However, the first SPIRE protein sequence discovered was that of the sea squirt Ciona savignyi (posterior end mark 5, PEM-5) (Satou and Satoh, 1997). No SPIRE proteins have been found in plants and fungi (Kollmar et al., 2019 preprint). All known SPIRE proteins share a common array of defined structural motifs, which mediate interactions with FMN-subgroup formins, actin, class 5 myosin motor proteins, RAB GTPases and phospholipid bilayers (Welz and Kerkhoff, 2019) (Fig. 1A,B). By considering their structural motifs, it can be concluded that SPIRE proteins function within a membrane-associated multiprotein complex that coordinates the generation of actin structures and myosin motor protein activity (Pylypenko et al., 2016) (Fig. 2).

The N-terminal SPIRE kinase non-catalytic C-lobe domain (KIND) evolved from the catalytic protein kinase fold into a protein-interaction module through the loss of the kinase N-lobe, with only the kinase C-lobe structure remaining (Ciccarelli et al., 2003) (Fig. 1B,C). The KIND domain was originally discovered based on its high similarity to p21-activated kinase 1 (PAK1) (Ciccarelli et al., 2003), and it structurally aligns with the PAK1 kinase C-lobe fold (Vizcarra et al., 2011; Zeth et al., 2011) (Fig. 1C). The KIND domain mediates the interaction with FMN-subgroup formins by binding the formin–SPIRE interaction (FSI) sequence motif, which is located at the very C-terminal end of FMN-subgroup formins (Pechlivanis et al., 2009). The structure of the SPIRE1 KIND domain in complex with the FMN2-FSI peptide has been solved by X-ray crystallography (Vizcarra et al., 2011; Zeth et al., 2011) (Fig. 1C).

Besides the interaction with FMN-subgroup formins, the SPIRE KIND domain also mediates an intramolecular interaction with the C-terminal FYVE_2 zinc finger domain [formerly designated as modified Fab1, YOTB/ZK632.12, VAC1, EEA1 (FYVE), mFYVE] (Tittel et al., 2015) (Figs 1B and 2). The intramolecular KIND–FYVE_2 interaction competes with the intermolecular KIND–FMN-FSI interaction and is not detected when SPIRE is bound to membranes (Tittel et al., 2015). A model has been proposed in which SPIRE proteins form a cytoplasmic backfolded, autoinhibited conformation that is opened up for interactions with formins upon SPIRE membrane targeting (Tittel et al., 2015) (Fig. 2).

The SPIRE KIND domain is followed by a tandem array of multiple Wiskott-Aldrich syndrome protein (WASP) homology 2 (WH2) domains, which mediate the interaction with monomeric globular actin (G-actin), as well as with actin filaments (Bosch et al., 2007; Bradley et al., 2020; Ducka et al., 2010; Montaville et al., 2014; Otto et al., 2000; Quinlan et al., 2005; Sitar et al., 2011) (Figs 1A,B and 2). Crystallographic analysis of a single Dm-SPIRE WH2 domain in complex with actin shows that the WH2 domain consists of a N-terminal α-helix that binds to the hydrophobic cleft between subdomains 1 and 3 at the barbed end of the actin monomer (Ducka et al., 2010) (Fig. 1D). The C-terminal tail of the WH2 domain covers the surface of actin between subdomains 2 and 4 (Ducka et al., 2010) (Fig. 1D). Electron microscopy data of Dm-SPIRE–actin complexes and small-angle X-ray scattering (SAXS) data of Dm-SPIRE–actin complexes in solution indicate that the four WH2 domains of Dm-SPIRE align actins along the long axis of the core of the Dm-SPIRE–actin particle (Quinlan et al., 2005; Sitar et al., 2011). Based on the X-ray crystallography and SAXS data, a model has been proposed in which the actin monomers bound by the N-terminal helix and C-terminal tail of the WH2 domain build rigid units that are linked by unstructured flexible linkers (Sitar et al., 2011).

The central region of the mammalian SPIRE proteins encodes a short, highly conserved myosin-5-interaction motif, which has been shown to interact with the globular tail domain (GTD) of all three mammalian class 5 myosin motors (MYO5A, MYO5B, MYO5C) (Pylypenko et al., 2016) (Fig. 1A,B). A peptide encoding the GTD-binding motif (GTBM) of human SPIRE2 has been co-crystallised with the human MYO5A GTD (Pylypenko et al., 2016) (Fig. 1E). A superimposition of the crystal structures of MYO5A-GTD in complex with RAB11A and the SPIRE2-GTBM, respectively, shows that SPIRE and RAB11 contact different surfaces of the MYO5 GTD (Pylypenko et al., 2016) (Fig. 1E). Indeed, the existence of a tripartite complex consisting of MYO5A, RAB11A and SPIRE2 has been confirmed by protein interaction studies and colocalisation analyses (Pylypenko et al., 2016). The interaction of SPIRE2 and RAB11A with MYO5A has been shown to contribute to the membrane targeting of the MYO5 motor protein (Pylypenko et al., 2016).

Melanophilin (MLPH) and SPIRE proteins synergise in melanosome transport in mouse melanocytes (see below) (Alzahofi et al., 2020), and they share a similar direct interaction mode with MYO5A (Pylypenko et al., 2013, 2016) as well as the ability to interact with RAB GTPases (Alzahofi et al., 2020; Fukuda et al., 2002; Nagashima et al., 2002; Strom et al., 2002; Wu et al., 2002). MLPH and SPIRE proteins have a structurally related zinc finger domain (the FYVE_2 domain) that interacts with RAB GTPases (Kerkhoff et al., 2001; Matesic et al., 2001), which mediates the interaction of SPIRE1 and MLPH with RAB27A (Alzahofi et al., 2020). The canonical FYVE zinc finger domain comprises eight cysteine residues, which complex two Zn²⁺ ions (Misra and Hurley, 1999; Stenmark et al., 2002). Interaction with phosphatidylinositol-3-phosphate (PI3P) is a hallmark of the canonical FYVE domain, and a typical representative of the canonical FYVE family is the early endosomal antigen 1 (EEA1) protein, which regulates endocytic vesicle trafficking (Langemeyer et al., 2018). The FYVE_2 domain and canonical FYVE domains share the arrangement of eight cysteine residues and two Zn²⁺ ions, but FYVE_2 lacks PI3P-binding motifs (Kukimoto-Niino et al., 2008; Misra and Hurley, 1999; Ostermeier and Brunger, 1999). Instead, the FYVE_2 domain is characterised by conserved N- and C-terminal flanking sequences, which mediate interactions with RAB3, RAB8 and RAB27 proteins of the secretory group (Fukuda, 2008, 2013) (Fig. 1F). Proteins containing the FYVE_2 domain are involved in different steps of exocytic and/or secretory vesicle transport, such as synaptic vesicle release, which is regulated by the FYVE_2-containing RIMS1 and RIMS2 proteins (Fukuda, 2003; Han et al., 2015; Wang et al., 1997), or the cortical transport of melanosomes, which depends on MLPH (Fukuda, 2008; Hume and Seabra, 2011; Matesic et al., 2001). The FYVE_2 domain-containing protein rabphilin-3A is an effector of the secretory RAB3A GTPase and is targeted to synaptic vesicle membranes (Li et al., 1994; Yamaguchi et al., 1993). Crystal structures of the FYVE_2 domains of MLPH and rabphilin-3A have been solved in complex with their interacting RAB GTPases, RAB27A and RAB3A, respectively (Kukimoto-Niino et al., 2008; Ostermeier and Brunger, 1999) (Fig. 1F). The crystal structures show that an α-helix, N-terminal to the zinc finger motif, provides a major contribution to the interaction, and this helix corresponds to the SPIRE box (SB), a highly conserved motif of the SPIRE family (Alzahofi et al., 2020; Kerkhoff et al., 2001). Consequently, C-terminal sequences of mammalian SPIRE1 have also been shown to interact with RAB27A and RAB3A (Alzahofi et al., 2020; Lagal et al., 2014). Direct interactions between SPIRE1 and RAB GTPases of the exocytic and/or secretory group suggest that SPIRE proteins might have a more general role in cell signalling within the endocrine or nervous systems. The recently discovered function of SPIRE1 in the RAB27A-regulated secretion of von Willebrand factor from human endothelial cells (Holthenrich et al., 2022) is a first confirmation of this hypothesis (see below).

Biomimetic membrane-binding studies addressing the interaction of the SPIRE FYVE_2 domain with lipid bilayers have revealed that the interaction involves both the presence of negatively charged lipids and hydrophobic contributions from the turret loop of the FYVE_2 domain that intrudes the lipid bilayer (Tittel et al., 2015). A potential regulation of SPIRE function by negatively charged membrane lipids such as phosphatidylserine has been discussed (Tittel et al., 2015). These in vitro membrane assays have also revealed that when the FYVE_2 domain of SPIRE2 is bound to membranes, it cannot interact with its KIND domain (Tittel et al., 2015), suggesting that the interaction of SPIRE with membranes releases its autoinhibited backfolded conformation (Fig. 2).

The presence of a WH2 actin-binding domain places the SPIRE proteins in the family of actin filament organisers (Otto et al., 2000; Wellington et al., 1999) (Fig. 1A,B). Consistently, actin filament structures in oocytes, zygotes and melanocytes have been found to be regulated by cooperation between SPIRE and FMN proteins (Alzahofi et al., 2020; Dahlgaard et al., 2007; Pfender et al., 2011; Scheffler et al., 2021). The cooperative activity of SPIRE and FMN proteins in the assembly of actin filaments has been studied in detail using in vitro actin polymerisation assays (Bradley et al., 2020; Montaville et al., 2014; Quinlan et al., 2007).

Owing to the relative instability of actin dimers and trimers and the presence of actin-binding proteins, such as profilin and β-thymosin, de novo generation of actin filaments requires actin nucleation factors, which help to overcome the kinetic barriers (Pollard, 2016).

In vitro actin polymerisation assays have revealed that a SPIRE protein tandem array containing four WH2 actin-monomer-binding domains nucleates actin filament assembly from free actin monomers by bringing together multiple G-actin proteins (Quinlan et al., 2005; Sitar et al., 2011).

In analogy, other proteins with tandem arrays of multiple WH2 domains have been identified as nucleators of actin polymerisation, such as the mammalian COBL protein (Ahuja et al., 2007) and the bacterial virulence factors VopL (Vibrio parahaemolyticus) (Burke et al., 2017; Liverman et al., 2007; Namgoong et al., 2011), VopF (Vibrio cholerae) (Burke et al., 2017; Tam et al., 2007) and Sca2 (Rickettsia rickettsii) (Kleba et al., 2010; Madasu et al., 2013). These proteins comprise the group of tandem WH2 domain-based nucleators (Dominguez, 2016).

The almost identical phenotypes of Drosophila spire-mutant flies and mutants of the Drosophila FMN-subgroup formin gene cappuccino (capu) (Emmons et al., 1995; Manseau and Schupbach, 1989; Wellington et al., 1999), as well as a strong interaction between the two proteins (Quinlan et al., 2007), provided early indications of a cooperative mechanism of action for SPIRE proteins and FMN-subgroup formins in actin filament assembly. Formins dimerise via their formin homology 2 (FH2) domains, forming a torus-shaped structure that interacts with G-actin (Otomo et al., 2005). In addition to nucleating actin filament polymerisation by binding of multiple actin monomers to the dimeric FH2 domains, formins can also accelerate the elongation of actin polymerisation (Goode and Eck, 2007; Pollard, 2016). The formin dimer stays processively attached to the fast-growing barbed end of the actin filament via its FH2 domains, and binding of profilin–actin to multiple profilin-binding-sites within its flexible proline-rich formin homology 1 (FH1) domain enhances the elongation rate by rapidly delivering actin to the fast-growing barbed end (Paul and Pollard, 2008; Pruyne et al., 2002; Romero et al., 2004).

The KIND domain of SPIRE proteins interacts with the C-terminal FSI motif of FMN-subgroup formins (Pechlivanis et al., 2009; Vizcarra et al., 2011; Zeth et al., 2011), and it has been shown that each Capu-FH1-FH2-FSI dimer can bind to two Dm-SPIRE KIND domains (Quinlan et al., 2007), indicating that SPIRE proteins and FMN-subgroup formins assemble into a heterotetrameric complex of two SPIRE and two formin proteins (Dietrich et al., 2013; Quinlan et al., 2007; Quinlan and Kerkhoff, 2008). A dimeric, nucleation-incompetent Capu-FH1-FH2-FSI mutant (Capu-I706A) enhances actin nucleation by Dm-SPIRE-KIND-WH2 (Quinlan et al., 2007), indicating that the arrangement of the SPIRE tandem WH2 domains into a dimeric structure enhances SPIRE actin nucleation activity (Dietrich et al., 2013; Quinlan et al., 2007; Quinlan and Kerkhoff, 2008). Consistent with this, forced dimerisation of the Dm-SPIRE WH2 domain cluster by fusion of the Dm-SPIRE-WH2 sequences with the VopL C-terminal domain (VCD) causes a dramatic increase in nucleation activity (Dominguez, 2016; Namgoong et al., 2011). Dimerisation through the VCD is also essential for the actin filament nucleation activity of VopL, which contains three WH2 domains (Dominguez, 2016; Namgoong et al., 2011).

In cells, most of the unpolymerised actin is bound either to profilin or β-thymosin, with profilin-bound actin being the main source of monomers for actin polymerisation under physiological conditions (Montaville et al., 2014; Pollard, 2016). SPIRE1-KIND-WH2 alone is unable to efficiently stimulate actin filament assembly from profilin-bound actin (Montaville et al., 2014). In contrast, the FMN2-FH1-FH2-FSI protein efficiently accelerates the assembly rate of profilin-bound actin (Montaville et al., 2014). However, in cooperation with the N-terminal region of the SPIRE1 protein (SPIRE1-KIND-WH2) the FMN2-FH1-FH2-FSI filament assembly rate was strongly enhanced (Montaville et al., 2014), showing a synergy of SPIRE1 and FMN2 in actin filament assembly. A model has been proposed in which the formin dimerises SPIRE proteins and the two strands of the helical actin filament can be nucleated by the two SPIRE-WH2 domain clusters (Dietrich et al., 2013; Quinlan and Kerkhoff, 2008). Subsequently, the SPIRE and formin proteins dissociate and the formin drives filament elongation through a processive association with the fast-growing barbed end and the attraction of profilin-bound actin through its FH1 domain (Dietrich et al., 2013; Quinlan and Kerkhoff, 2008) (Fig. 2).

A recent study employed a biomimetic approach to mimic actin filament assembly by vesicle-localised SPIRE proteins in order to experimentally address the mechanism underlying the synergy between SPIRE proteins and FMN-subgroup formins (Bradley et al., 2020). In the study, recombinant Dm-SPIRE-KIND-WH2 proteins were coupled to microspheres, and cooperation with the Drosophila formin Capu (Capu-FH1-FH2-FSI) in actin filament assembly was studied using total internal reflection fluorescence (TIRF) microscopy (Bradley et al., 2020). The following conclusions, reflecting our current knowledge of the underlying biochemistry, can be drawn. Dm-SPIRE proteins bind to and attract the formin Capu to the beads, and Dm-SPIRE and Capu synergise in actin filament assembly. The actin filament barbed ends project away from the beads, whereas Dm-SPIRE retains the pointed ends at the beads through its WH2 domain cluster. It has been suggested that Capu binds to the fast-growing barbed end where it mediates further filament elongation (Bradley et al., 2020) (Fig. 2).

Besides SPIRE-mediated actin filament assembly from actin monomers, the interaction of the SPIRE proteins with filamentous actin has been studied. At the barbed ends of actin filaments, alternate binding of FMN2 and SPIRE1 proteins has been described, with SPIRE1 and FMN2 competing for binding (Montaville et al., 2014) (Fig. 2). The displacement of FMN2 by SPIRE1 inhibits formin-driven filament elongation (Montaville et al., 2014), and the first WH2 domain of the Dm-SPIRE protein has been shown to mediate interaction with the barbed end (Bradley et al., 2020). However, it still needs to be determined whether this mechanism also applies for the first WH2 domain of the mammalian homologues.

In addition to their role in filament assembly, SPIRE proteins can also disassemble actin filaments (Bosch et al., 2007; Sitar et al., 2011). A single Dm-SPIRE WH2 domain has been found to be sufficient for sequestering actin monomers and severing of actin filaments (Sitar et al., 2011). Monitoring of actin filament severing with TIRF microscopy suggests that, owing to its high actin-binding affinity, the WH2 domain is able to displace an actin monomer from an existing filament, resulting in a fast disruption of filamentous actin (Sitar et al., 2011). Actin monomer sequestering has also been described for WH2 domains of COBL and Sca2 proteins (Dominguez, 2016; Husson et al., 2011; Madasu et al., 2013). A cellular function of such SPIRE-mediated actin filament severing, however, has not yet been described.

In summary, it is well established that SPIRE and FMN-subgroup formins cooperate in the assembly of actin filaments at membranes and that the formed filaments project away from the membrane.

Given their direct interaction with class 5 myosins and RAB GTPases, as well as their biochemical effect on actin dynamics, SPIRE proteins can be considered bona fide regulators of actomyosin-driven vesicle transport (Alzahofi et al., 2020; Pylypenko et al., 2016). Such a role was assumed in an initial molecular cell biological study, which showed that mammalian SPIRE1 proteins colocalise with RAB11 on cytoplasmic vesicles in mammalian cells, although specific actin structures at those vesicles were not detected, presumably because of the extensive staining of the bulk cellular actin cytoskeleton (Kerkhoff et al., 2001). The first SPIRE-regulated actin structure was discovered later through an analysis of D. melanogaster spire mutants (Dahlgaard et al., 2007). Here, high intensity actin staining of fly oocytes led to the discovery of an isotropic mesh of actin filaments in the oocyte cytoplasm that was absent from Drosophila spire- and capu-mutant oocytes (Dahlgaard et al., 2007) (Fig. 3A). Dm-SPIRE and Capu cooperate in the generation of the cytoplasmic actin meshwork, which suppresses microtubule–kinesin-driven cytoplasmic streaming during fly oogenesis (Dahlgaard et al., 2007; Emmons et al., 1995; Theurkauf, 1994; Wellington et al., 1999). A detailed mechanistic analysis of the origin of the Drosophila cytoplasmic actin filament mesh has not been performed, and it remains to be elucidated whether the mesh originates from vesicles and whether myosin proteins are associated.

Interestingly, a knockout of FMN2 also affects mouse oocyte maturation. In FMN2-knockout oocytes, the metaphase spindle is not correctly positioned during meiosis I and polar body extrusion is largely impaired; this causes a polyploidy and hypofertility phenotype (Leader et al., 2002). In analogy to the cytoplasmic actin filament mesh in flies, it was reasonable to search for an equivalent actin meshwork in mouse oocytes. Indeed, the subsequent discovery of an actin filament mesh regulated by SPIRE1 and SPIRE2 in cooperation with FMN2 in mouse oocytes (Pfender et al., 2011) (Fig. 3B) triggered an entirely new area of research and largely contributed to our current mechanistic understanding of mammalian oocyte maturation and fertilisation. The cooperation between mouse SPIRE proteins and FMN2 has been shown to be crucial for the regulation of the two essential steps of asymmetric oocyte division: the asymmetric positioning of the metaphase spindle and polar body extrusion (Pfender et al., 2011) (Figs 3B and 4B). In addition to organising the cytoplasmic actin filament mesh, SPIRE1, SPIRE2 and FMN2 have been found to be strongly enriched in the cleavage furrow during polar body extrusion, and SPIRE1 and SPIRE2 co-depleted oocytes fail to assemble a cleavage furrow upon anaphase onset (Pfender et al., 2011) (Fig. 4B).

It is important to note that the mouse oocyte actin filament mesh originates from cytoplasmic RAB11A vesicles (Pfender et al., 2011; Schuh, 2011). At the surface of these vesicles, SPIRE1 and SPIRE2 in cooperation with FMN2 generate actin filaments, which serve as tracks for the long-range transport of RAB11A vesicles towards the oocyte cortex (Pfender et al., 2011; Schuh, 2011) (Fig. 3B). Here, the processive actin filament sliding activity of MYO5B generates the forces for motility of the RAB11A vesicles. As described above, distinct surfaces of the MYO5A GTD directly interact with RAB11A and SPIRE2 (Pylypenko et al., 2013, 2016). Therefore, the related MYO5B might coordinate targeting of the actin nucleation activity of SPIRE1 and SPIRE2 proteins to RAB11A vesicles by forming a tripartite protein complex at the membrane surface (Figs 1E and 3B).

More recently, a very similar vesicle-originated actin filament mesh has been discovered that mediates the centrifugal dispersion of melanosome dense-core vesicles towards the plasma membrane in mouse melanocytes (Alzahofi et al., 2020). Melanosome transport is regulated by RAB27A, which in this case does not directly interact with the processive MYO5A motor, but instead interacts with the C-terminal FYVE_2 domain of SPIRE1 (Fig. 3C); actin filaments are thus generated owing to a cooperation between SPIRE1 and FMN1 (Alzahofi et al., 2020). As discussed above, MLPH is also essential for the actomyosin-dependent long-range transport of melanosomes and has a FYVE_2 domain RAB27A-interaction module in common with SPIRE proteins, as well as a closely related MYO5A-GTD-binding motif (Alzahofi et al., 2020; Pylypenko et al., 2013, 2016) (Fig. 3C), suggesting that both MLPH and SPIRE proteins contribute to the targeting and activation of MYO5A. In addition, MLPH contains an actin filament-binding motif at its C-terminal end, which might be capable of facilitating the anchoring of actin filaments at melanosome membranes (Hume et al., 2006; Kuroda et al., 2003).

Computer simulations of melanosome centrifugal dispersion employing Cytosim (https://gitlab.com/f-nedelec/cytosim; Alzahofi et al., 2020), performed based on the assumption of a melanosome-associated processive motor protein, suggest that the pointed ends of actin filaments stay attached to the melanosome membrane and that actin filaments project peripherally into the cytoplasm (Fig. 3D), perfectly recapitulating the cytoplasmic dispersion of melanosomes (Alzahofi et al., 2020). In addition to directly anchoring actin filaments by retaining binding of the pointed ends via the SPIRE-WH2 cluster as described above, MLPH or even MYO5A might function as filament anchors at melanosome membranes.

However, it remains speculative whether the exocytic RAB27- or RAB11-regulated and FMN–SPIRE–MYO5-dependent long-range transport mechanism can be generalised to other cellular transport processes. Detailed mechanistic analyses of SPIRE and FMN function in exocytic, recycling and secretory transport processes, especially in cells of the endocrine and nervous systems, may reveal interesting new findings regarding SPIRE protein function in this context.

Evidence for a SPIRE1 function in secretory transport processes has been provided by a recent study showing that SPIRE1 promotes the externalisation of von Willebrand factor from endothelial cells (Holthenrich et al., 2022). Here, SPIRE1 and MYO5C have been found to accumulate at actin ring structures at the release sites of Weibel–Palade bodies (Fig. 4A). Weibel–Palade bodies are elongated secretory granules that store two major components: von Willebrand factor, which functions in haemostasis, and P-selectin, which has a role in recruiting leukocytes during inflammation (McCormack et al., 2017). A mechanistic understanding of SPIRE1 protein function in the von Willebrand factor externalisation process is still missing. In this respect, it is important to note that the human umbilical vein endothelial cells (HUVECs) employed in the study do not express FMN1 and FMN2 (Holthenrich et al., 2022). This suggests that SPIRE1 either cooperates with a different formin or even acts independently.

Actin filaments can contribute to cellular force generation in different ways. Besides actomyosin-driven forces, such as the vesicle-originated actomyosin meshes described above, the polymerisation of actin filaments itself can exert pushing forces (Rottner et al., 2017). This is best studied for cell edge protrusions of lamellipodia or filopodia, in which membrane-bound actin filament elongation factors of the formin and Ena/VASP families facilitate actin subunit incorporation at the tip of actin filaments that point into the plasma membrane with their fast-growing ends (Rottner et al., 2017). A similar mechanism of pushing forces has recently been discovered to initiate the inward movement of male and female pronuclei in mouse zygotes (Scheffler et al., 2021) (Fig. 4C). In this process, SPIRE2–FMN2-regulated actin filaments accelerate a fast, inward-directed movement of the pronuclei away from the plasma membrane as the initial step of pronuclear fusion, which occurs in the centre of the zygote. The movement of the male pronucleus has been studied in detail (Scheffler et al., 2021). During fertilisation, the male pronucleus assembles within the fertilisation cone at the sperm entry site (Scheffler et al., 2021). RAB11A has been shown to enrich at the cell surface of the zygote, directly behind the forming male pronucleus, and to recruit SPIRE2 and FMN2 proteins, which subsequently assemble actin filaments that launch a fast inward movement of the pronucleus (Fig. 4C, right panel) (Scheffler et al., 2021). FMN2 is highly concentrated at a ring structure behind the male pronucleus (Scheffler et al., 2021), indicating that it drives actin filament elongation at the pronucleus membrane, whereas SPIRE2 is concentrated at the cell membrane (Scheffler et al., 2021) where they might contribute to anchoring of the pointed ends of the actin filaments. This motility mechanism, therefore, seems to be different from actin comet tail propulsion mediated by Arp2/3 (Loisel et al., 1999), which is characterised by treadmilling of the actin filaments. In parallel to the fast inward movement of the pronucleus, a dynamic network of microtubules assembles and finally slowly moves the pronucleus further towards the zygote centre (Scheffler et al., 2021). Since RAB11A does not directly interact with SPIRE2 (Pylypenko et al., 2016), it remains to be determined how the recruitment of the actin nucleator complex to the fertilisation cone is mediated.

Another example of force generation by elongation of SPIRE1-assembled actin filaments is in the regulation of mitochondrial division. A short sequence motif of 58 amino acids, which is encoded by the alternatively spliced SPIRE1 exon 13, specifically targets the mammalian SPIRE1 protein to mitochondria, where it is integrated into the outer mitochondrial membrane (Manor et al., 2015) (Fig. 5). Mitochondrial SPIRE1 (mitoSPIRE1, also known as Spire1C) cooperates with inverted formin-2 (INF2-CAAX isoform), which is anchored to the ER by a C-terminal farnesyl group (Korobova et al., 2013; Ramabhadran et al., 2011), in assembling actin filaments at ER–mitochondria contact sites (Manor et al., 2015). The results of simulations suggest a model in which the actin structures organised by mitoSPIRE1–INF2-CAAX can induce mitochondrial constriction, thereby contributing to mitochondrial division at ER–mitochondria contact sites (Manor et al., 2015). The mitochondrial SPIRE1 isoform contains all structural and functional motifs of the vesicular SPIRE isoforms (Figs 1A and 5A) and has the capacity to bind class 5 myosins (Fig. 1A,B). It is therefore tempting to assume that mitoSPIRE1 could organise actomyosin structures at mitochondria, which might regulate mitochondrial motility, since RNA interference experiments in cultured Drosophila neurons have shown that MYO5 activity opposes microtuble-based axonal transport of mitochondria (Pathak et al., 2010).

Biochemical and cell biological studies have provided detailed insights into the mechanisms of SPIRE-regulated actin filament organisation and force generation in intracellular membrane transport processes. The role of SPIRE in forming vesicle-originated actomyosin networks provides an additional means of intracellular transport beyond the cortical actin filament tracks and transport along microtubules (Alzahofi et al., 2020; Pfender et al., 2011; Schuh, 2011) (Fig. 3B,C). However, this specific transport mechanism has so far only been found in mammalian oocytes and melanocytes, but it is likely, although still speculative at this point, that cytoplasmic SPIRE-regulated actomyosin networks have a more general role in vesicle and organelle transport in animal cells. Given that SPIRE1 expression has been found to be highest in neurons (Schumacher et al., 2004) and that SPIRE2 is expressed predominantly in epithelial cells of the digestive tract (Pleiser et al., 2010), future studies should focus on analysing the as-yet-unknown roles of SPIRE proteins in the nervous system and digestive tract. In this regard, it is interesting to note that the SPIRE1 and FMN2 genes have a nearly identical expression pattern in the mouse nervous system, with both genes being highly expressed in hippocampal neurons (Schumacher et al., 2004). A SPIRE1-mutant mouse model has been generated by gene trap insertion (Texas A&M Institute for Genomic Medicine, clone OST416113; https://www.tigm.org/repository/), which lacks any functional vesicular or mitochondrial SPIRE1 (Pleiser et al., 2014). Phenotyping of the SPIRE1 mutants has revealed that the mice exhibit increased fear-related behaviour in fear-conditioning experiments (Pleiser et al., 2014). It is interesting to note that the FMN2-knockout mouse also has a fear-related phenotype (Agis-Balboa et al., 2017), and importantly, in humans, the loss of FMN2 function causes intellectual disability (Law et al., 2014). However, the neuronal functions of SPIRE1 and FMN2 and their potential cooperation in this system are unknown. Considering the direct interaction between SPIRE1 and the secretory RAB27A GTPase (Alzahofi et al., 2020), one could speculate that neuronal SPIRE1 might function in the neuroendocrine system or even in synaptic transmission. Here, the role of mitoSPIRE1 in the regulation of mitochondrial dynamics might also be important for neuronal signalling (Manor et al., 2015; Rangaraju et al., 2019). We thus anticipate that future efforts aimed at elucidating SPIRE function in the nervous system may turn out to be as important and rewarding as the comprehensive studies of SPIRE in mammalian reproduction have been in the past.

Considering the tremendous impact infectious diseases have for our societies, it will also be important to follow up two recent studies showing that SPIRE proteins influence the life cycles of bacterial and viral pathogens (Andritschke et al., 2016; Torres et al., 2022) (Box 1). A mechanistic understanding of SPIRE protein function in these processes is still missing. A detailed analysis, however, might pave the way for novel antiviral and antibacterial drugs.