Cell biologists generally consider that microtubules and actin play complementary roles in long-and short-distance transport in animal cells. On the contrary, using melanosomes of melanocytes as a model, we recently discovered that the motor protein myosin-Va works with dynamic actin tracks to drive long-range organelle dispersion in opposition to microtubules. This suggests that in animals, as in yeast and plants, myosin/actin can drive longrange transport. Here, we show that the SPIRE-type actin nucleators (predominantly SPIRE1) are Rab27a effectors that cooperate with formin-1 to generate actin tracks required for myosin-Va-dependent transport in melanocytes. Thus, in addition to melanophilin/myosin-Va, Rab27a can recruit SPIREs to melanosomes, thereby integrating motor and track assembly activity at the organelle membrane. Based on this, we suggest a model in which organelles and force generators (motors and track assemblers) are linked, forming an organelle-based, cell-wide network that allows their collective activity to rapidly disperse the population of organelles long-distance throughout the cytoplasm.

This 'highways and local roads' model suggests that MTs are tracks for long-range transport (highways) between the cell centre and periphery, driven by kinesin and dynein motors. Meanwhile AFs (local roads) and myosin motors work down-stream picking up cargo at the periphery and transporting it for the 'last m' to its final destination. This model makes intuitive sense as MTs in animal cells in culture typically form a polarised radial network of tracks spanning >10 m from the centrally located centrosome to the periphery and appear ideally distributed for long-distance transport. Meanwhile, with some exceptions in which AFs form uniformly polarised arrays, e.g. lamellipodia, filopodia and dendritic spines, AF architecture appears much more complex. In many fixed cells AF appear to comprise populations of short (1-2 m length), with random or anti-parallel filament polarity, and not an obvious system of tracks for directed transport 5,6 . This view is exemplified by the co-operative capture (CC) model of melanosome transport in melanocytes 7,8 . Skin melanocytes make pigmented melanosomes and then distribute them, via dendrites, to adjacent keratinocytes, thus providing pigmentation and photo-protection (reviewed in 9 ). The CC model proposes that transport of melanosomes into dendrites occurs by sequential longdistance transport from the cell body into dendrites along MTs (propelled by kinesin/dynein motors), followed by AF/myosin-Va dependent tethering in the dendrites. Consistent with this, in myosin-Va-null cells melanosomes move bi-directionally along MTs into dendrites, but do not accumulate therein, and instead cluster in the cell body 7,10 . This defect results in partial albinism in mammals due to uneven pigment transfer from melanocytes to keratinocytes (e.g. dilute mutant mouse and human Griscelli syndrome (GS) type I patients; Figure 1A) 11,12 . Subsequent studies revealed similar defects in mutant mice (and human GS types II and III patients) lacking the small

Exocytic transport of transmembrane receptors and secreted ligands provides the basis for cellular communication in animals. The RAB8/RAB3/RAB27 trafficking regulators function in transport processes towards the cell membrane. The small G-proteins recruit a diversity of effectors that mediate transport along microtubule and actin tracks, as well as membrane tethering and fusion. SPIRE actin nucleators organise local actin networks at exocytic vesicle membranes. By complex formation with class-5 myosins, vesicle transport track generation and motor protein activation are coordinated. Our phylogenetic analysis traced the onset of SPIRE function back to the origin of the Holozoa. We have identified SPIRE in the closest unicellular relatives of animals, the choanoflagellates, and the more distantly related ichthyosporeans. The discovery of a SPIRE-like protein encoding a KIND and tandem-WH2 domains in the amoebozoan Physarum polycephalum suggests that the SPIRE-type actin nucleation mechanism originated even earlier. Choanoflagellate SPIRE interacts with RAB8, the sole choanoflagellate representative of the metazoan RAB8/RAB3/RAB27 family. Major interactions including MYO5, FMN-subgroup formins and vesicle membranes are conserved between the choanoflagellate and mammalian SPIRE proteins and the choanoflagellate Monosiga brevicollis SPIRE protein can rescue mouse SPIRE1/2 function in melanosome transport. Genome duplications generated two mammalian SPIRE genes (SPIRE1 and SPIRE2) and allowed for the separation of SPIRE protein function in terms of tissue expression and RAB GTPase binding. SPIRE1 is highest expressed in the nervous system and interacts with RAB27 and RAB8. SPIRE2 shows high expression in the digestive tract and specifically interacts with RAB8. We propose that at the dawn of the animal kingdom a new transport mechanism came into existence, which bridges microtubule tracks, detached vesicles and the cellular actin cytoskeleton by organising actin/myosin forces directly at exocytic vesicle membranes.The new degree of freedom in transport may reflect the increased demands of the sophisticated cellular communications in animals.