PERSPECTIVES

action. Science 177, 401–408 (1972). 40. Bonnet, C. Considérations sur les Corps Organisés (Rey, transport represent the movement of 35. Udrisar, D. & Rodbell, M. Microsomal and cytosolic Amsterdam, 1762). fractions of guinea pig hepatocytes contain 100-kilodalton 41. Boyer, P. D. The ATP synthase — a splendid molecular cytoskeletal and cytosolic at much GTP-binding proteins reactive with antisera against alpha machine. Annu. Rev. Biochem. 66, 717–749 (1997). slower rates, and the nature of the carrier subunits of stimulatory and inhibitory heterotrimeric GTP- 42. Haller, A. Elementa Physiologiae Corporis Humani binding proteins. Proc. Natl Acad. Sci. USA 87, 6321–6325 (Bousquet, Lausanne, 1757). structures for these proteins is not known. (1990). 43. Rastogi, V. K. & Girvin, M. E. Structural changes linked to Proteins that associate with 36. Alberts, B. The as a collection of machines: proton translocation by subunit c of the ATP synthase. preparing the next generation of molecular biologists. Cell Nature 402, 263–268 (1999). and move in slow component ‘a’ 92, 291–294 (1998). at average rates of roughly 0.3–3 mm day–1 37. Mitchell, P. & Moyle, J. Chemiosmotic hypothesis of Acknowledgements µ –1 oxidative phosphorylation. Nature 213, 137–139 (1967). This article has benefited from discussions with A. Cattaneo of (~0.004-0.04 m s ), and proteins that associ- 38. Malpighi, M. The Viscerum Structura (Montii, Bologna, the International School for Advanced Studies (S.I.S.S.A.) of ate with microfilaments, as well as many other 1666). Trieste, and has been made possible by bibliographical help from 39. Malpighi, M. Dissertatio Epistolica de Formatione Pulli in L. LIannucci of the University of Pisa. I also thank L. Galli-Resta, cytosolic proteins, are transported in slow Ovo (Martyn, London, 1673). A. Pignatelli and B. Pelucchi for critically reading the manuscript. component ‘b’ at average rates of roughly 2–8 mm day–1 (~0.02–0.09 µm s–1) (TABLE 1).

No movement en masse OPINION In radioisotopic pulse-labelling experiments, slow components ‘a’ and ‘b’ form unimodal asymmetrical waves, often loosely described Slow axonal transport: stop and go as ‘bell-shaped’,which spread as they move along the towards the axon tip (FIG. 1). traffic in the axon Each wave represents the concerted move- ment of many distinct proteins whose indi- vidual waveforms coincide. Early studies on Anthony Brown slow axonal transport stressed the coherence of these transport waves but not the spread- Efforts to observe the slow axonal transport the movement of a unique type of macromol- ing, and this gave rise to the idea that of cytoskeletal polymers during the past ecular structure (TABLE 1). cytoskeletal and cytosolic proteins move decade have yielded conflicting results, and The fast components of axonal transport along the axon en masse, that is, in a slow and this has generated considerable controversy. are now known to represent the anterograde synchronous manner1. The movement of neurofilaments has now and retrograde movement of distinct types of The expectation of a slow and synchronous been seen, and it is rapid, infrequent and membranous along microtubules at movement has had a profound influence on highly asynchronous. This motile behaviour average rates of roughly 50–400 mm day–1 the design of experiments aimed at detecting could explain why slow axonal transport has (~0.5–5 µm s–1), propelled by the action of slow axonal transport. For example, many eluded observation for so long. proteins2. Membranous studies have used fluorescence photobleaching organelles can therefore be considered to be or photoactivation strategies in which fluores- communicate with other cells by the carrier structures for fast axonal transport. cent or caged fluorescent cytoskeletal proteins extending cytoplasmic processes called In contrast, the slow components of axonal are injected into nerve cells and then a popula- and dendrites. Remarkably, axons can attain lengths of one metre or more, although they Table 1 | The moving structures of axonal transport* lack ribosomes and Golgi complexes. Axonal Rate class Average rate Moving structures Composition proteins and Golgi-derived vesicles are formed (selected examples) in the neuronal cell body and are shipped Fast components along the axon by a process called axonal transport. This movement is essential for the Fast anterograde 200–400 Golgi-derived vesicles proteins, mm day–1 and tubules , enzymes of growth and survival of axons, and continues (≈2–5 µm s–1) (secretory pathway) metabolism throughout the life of the nerve cell. Studies on axonal transport in laboratory Bi-directional 50–100 Mitochondria Cytochromes, enzymes of –1 animals with radioisotopic pulse labelling mm day oxidative phosphorylation (≈0.5–1 µms–1) have shown that there are hundreds of axonal- ly transported proteins, but that these proteins Fast retrograde 200–400 Endosomes, Internalized membrane move at a small number of discrete rates, mm day–1 (endocytic pathway) receptors, neurotrophins, ≈ µ –1 which can be categorized as either fast or slow. ( 2–5 m s ) active lysosomal hydrolases Each discrete rate component represents the Slow components movement of a largely distinct subset of pro- Slow component ‘a’ 0.3–3 Neurofilaments, proteins, teins that are transported together throughout mm day–1 microtubules‡ , spectrin, tau proteins their journey along the axon. To explain these observations, Lasek and colleagues proposed Slow component ‘b’ 2–8 Microfilaments, , clathrin, , the structural hypothesis of axonal transport, mm day–1 supramolecular dynactin, glycolytic which postulates that all axonal proteins move (≈0.02–0.09 µm s–1) complexes of the enzymes by association with, or as integral parts of, sub- cytosolic matrix 1 cellular carrier structures . According to this *Data compiled from REFS 1,41,44. ‡ In some neurons, proteins are transported in slow hypothesis, each rate component represents component ‘b’ as well as slow component ‘a’.

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 1 | NOVEMBER 2000 | 153 PERSPECTIVES tion of these proteins is marked by bleaching in axons has been a vexing problem, but the that the extent of bleaching in the photo- or activating the fluorescence in a narrow band most likely explanation is that cytoskeletal pro- bleaching studies on neurofilament proteins across the axon [see supplementary figure teins do not move en masse in axons after all. was only partial, reducing the fluorescence online]. In these experiments, a slow and syn- intensity in the axon to 20–50% of its initial chronous movement should be manifested as Neurofilaments move in fits and starts value6. If the residual unbleached fluores- a slow translocation of the marked zone A recent breakthrough in the study of slow cence in the bleached region exceeded the flu- towards the axon tip. However, most studies axonal transport has come from observations orescence intensity of a single neurofilament, on tubulin, actin and neurofilament proteins on neurofilament proteins, tagged with green then it is likely that the movement of neuro- using one or both of these techniques showed fluorescent protein (GFP), in cultured rat filaments across the bleached zone could have that the marked zone does not move3–9. sympathetic neurons14,15. These cultured neu- gone unnoticed. This could also apply to the Although gradual recovery of the fluorescence rons contain relatively few neurofilaments and photobleaching studies on actin and tubu- was observed after photobleaching, it had no frequently show discontinuities in their axonal lin3–5,9,12,13. For example, in the study of Okabe obvious directionality and was therefore neurofilament array, resulting in short seg- and Hirokawa12 on tubulin, photobleaching attributed to exchange between the bleached ments of axon that lack neurofilaments14. reduced the fluorescence intensity in the axon polymers and diffusible fluorescent subunits. Time-lapse imaging of these naturally occur- to 10–40% of its initial value and, in other Directional movement of the photobleached ring gaps in the axonal neurofilament array similar studies, researchers have estimated or photoactivated zone was observed in cul- has enabled the observation of axonal trans- that as much as 10–20% of the total tubulin tured frog neurons10–12, but it probably resulted port without the need for photobleaching or in the axon could have moved through the from stretching of the axon owing to the rapid photoactivation approaches. Contrary to photobleached gaps without detection4. growth and poor adhesion of these neurons expectations, neurofilaments move rapidly, Similar detection limits have also been esti- on laminin substrates11,13. The repeated failure with peak rates as high as 3 µm s–1, and these mated for the fluorescence photoactivation of so many efforts to demonstrate slow syn- movements are frequently interrupted by pro- technique8. The ability of the photobleaching chronous movement of cytoskeletal proteins longed pauses14,15 (FIG. 2). The average velocity experiments to detect the movement of excluding the pauses is about 0.2–0.3 µm s–1. cytoskeletal polymers may also have been Assuming an average transport rate in the hampered by the short length of the bleached –1 µ a range of 0.3–3 mm day (TABLE 1), we can esti- regions (3–5 m), and the relatively long mate that individual neurofilaments spend time-lapse intervals (typically five minutes or 83–99% of their time pausing during their more). These considerations suggest that the journey down the axon. photobleaching and photoactivation strate- Radioisotopic pulse labelling studies in the gies should be capable of detecting the slow mouse optic nerve led Nixon and axonal transport of cytoskeletal proteins if Logvinenko17 to propose almost 15 years ago they could be optimized to enable the detec- that there are two kinetically distinct popula- tion of single rapidly moving polymers. b tions of neurofilament proteins in axons, one that moves and one that is stationary. Anterograde According to this model, neurofilament poly- mers or oligomers exchange between the mov- 0 ing and stationary phase as they move along the axon18. On the other hand, Lasek and col- leagues16,19 have challenged this hypothesis, 5 arguing that there is a single population of

c neurofilaments in axons that all move relent- 10 lessly, but at a broad range of rates. In princi- ple, the alternating movements and pauses observed for GFP-tagged neurofilaments in 15 cultured neurons14,15 could be regarded as Time (s) transitions between two distinct moving and 20 stationary phases, or simply as the intermittent movements of a single population of neurofil- 25 Figure 1 | Kinetics of slow axonal transport. aments that move at a broad and continuous Diagram illustrating the kinetics of slow axonal range of rates. Further studies will be required transport, as revealed by radioisotopic pulse 30 labelling. a | Radioactive amino acids injected into to distinguish between these two possibilities. the vicinity of the neuronal cell body produce a transient pulse of newly synthesized radioactive Re-evaluating previous approaches Figure 2 | A neurofilament on the move. Time- proteins, which b,c | move together along the axon Why have previous attempts6,7 to observe the lapse images of a neurofilament moving through a by axonal transport. After a time interval ranging axonal transport of neurofilament proteins naturally occurring gap in the axonal neurofilament from hours to months, the animal is killed, the using fluorescence photobleaching failed to array of a cultured nerve cell. The neurofilaments nerve is excised and sliced into segments, and were visualized using green fluorescent protein each segment is analysed biochemically to identify reveal movement? One possible explanation is (GFP)-tagged neurofilament protein M. The the radioactive proteins. The pulse-labelled that those studies were not capable of detect- fluorescence images are shown in inverted proteins form an asymmetrical wave (red) that ing the rapid movement of single cytoskeletal contrast for greater clarity. Scale bar= 5 µm. spreads as it moves along the axon. polymers. For example, it is important to note (Figure adapted from REF. 14.) (See movie online).

154 | NOVEMBER 2000 | VOLUME 1 www.nature.com/reviews/molcellbio PERSPECTIVES

Why so slow? Proximal Distal The rate of slow axonal transport in radioiso- topic pulse labelling experiments is generally quoted as the rate of movement of the wave peak, but the spreading of the transport wave indicates that the radiolabelled proteins actu- – + ally move at a broad range of rates16,19(FIG. 1). The motile behaviour of GFP-tagged neurofil- aments described above suggests a model for slow axonal transport that can account for the – + slow rate and the spreading of the transport wave. Consider a pulse of radioactive neurofil- ament proteins that assemble into filaments in the neuronal cell body20. Let us assume that each neurofilament moves rapidly along the Microtubule Retrograde motor (minus-end-directed) axon but that the overall rate of movement is Neurofilament Anterograde motor (plus-end-directed) slow because the filaments spend a large pro- portion of their time pausing. By chance, or perhaps due to intrinsic differences, some fila- Figure 3 | A model for the movement of neurofilaments in axons. In this model, neurofilaments are considered to move bidirectionally along microtubules through the action of a plus-end-directed motor ments move more frequently than others, and such as a kinesin-related protein, and a minus-end-directed motor such as dynein. Note that axonal this causes the population to spread out as it microtubules are all orientated with their plus-ends distal, towards the axon tip. Only a small fraction of the moves along the axon. The frequency with axonal neurofilaments move at any one point in time. which filaments move, or the amount of time that they spend pausing, could be determined simply by proximity to the transport machin- movement36. If the motility of microtubules is cytosolic proteins move, but also the motors ery or substrate, or by local variations in the as rapid and infrequent as for neurofilaments, that move them, and the substrates that they resistance to movement, or by some regulato- then it is possible that their movement might interact with. Potential substrates for motor- ry process. Neurofilaments that move most have gone undetected using the speckling driven movements of cytoskeletal polymers often will end up at the leading edge of the technique. in axons include the plasma membrane, the transport wave19, whereas neurofilaments that Slow axonal transport represents the endoplasmic reticulum and other cytoskeletal move least often will end up at the trailing movement of a myriad of other cytosolic pro- filaments21. In principle, neurofilaments could edge16. According to this hypothesis, the trans- teins in addition to cytoskeletal proteins move by direct interaction with molecular port wave represents the distribution of many (TABLE 1). One attractive hypothesis is that motors, or they could ride ‘piggyback’ by thousands of neurofilaments whose individ- these cytosolic proteins are transported by attachment to other moving structures. ual movements and pauses are summed over forming physical associations with moving Evidence for a direct interaction has come the days, weeks or months that they spend cytoskeletal polymers1. The relatively simple from a recent report that neurofilaments travelling down the axon (FIG. 1). protein composition of slow component ‘a’ purified from bovine spinal cord can move indicates that neurofilaments and micro- rapidly along microtubules in an ATP- Polymers as carrier structures tubules could be the sole carrier structures for dependent manner in vitro37, at peak rates of The mechanism of slow axonal transport has this rate component; all of the proteins that up to 1 µm s–1. A similar motile mechanism been debated for almost 15 years, and most of move in slow component ‘a’ are either integral has also been described for vimentin fila- the controversy has focused on the structural parts of these cytoskeletal polymers or are ments and their precursors along micro- form in which the cytoskeletal subunit pro- known to associate with these polymers in tubules in non-neuronal cells38. teins move21,22. Some studies have concluded vivo. In contrast, the protein composition of Slow axonal transport has generally been that cytoskeletal proteins move as assembled slow component ‘b’ is extremely complex and assumed to be exclusively anterograde, mov- polymers23–28, and some have concluded that includes more than 200 proteins, many of ing towards the axon tip, but in the observa- they move as unassembled subunits28–32,but which are traditionally described as ‘soluble’1. tions on GFP-tagged neurofilaments none has been conclusive. The observations on The presence of actin indicates that microfila- described above, about 20–30% of the GFP-tagged neurofilaments described ments could function as carrier structures for observed filaments actually moved in a retro- above14,15 have shown unequivocally that neu- this rate component. However, given the large grade direction, towards the cell body14,15. rofilament polymers do move in axons. number of diverse proteins in slow compo- Similarly, in the in vitro study described Whether actin and tubulin also move in the nent ‘b’,it is likely that the carrier structures for above, neurofilaments were observed to move form of assembled polymers in axons remains this rate component are complex and may towards the minus as well as the plus ends of to be determined, although there is clearly comprise several macromolecular complexes microtubules37. One possible explanation is precedent for such movements in non-neu- that move by direct or indirect association that the retrogradely moving neurofilaments ronal cells (for example, REFS 33,34). with the moving microfilaments. represent a distinct population, as proposed Microtubule polymers have been observed to by Griffin and colleagues39 on the basis of move in growth cones and developing axonal Motors and substrates their studies on the redistribution of branches of cultured neurons35, whereas To understand the mechanism of slow axonal cytoskeletal proteins in transected peripheral experiments using fluorescence-speckle transport, we must identify not only the nerves. Alternatively, the retrograde move- microscopy have not detected any structural forms in which cytoskeletal and ments could represent transient reversals of

NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 1 | NOVEMBER 2000 | 155 PERSPECTIVES

filaments that all eventually move in a net ity to observe the slow axonal transport of (1997). 22. Hirokawa, N., Terada, S., Funakoshi, T. & Takeda, S. anterograde direction. If there is a distinct neurofilament polymers in living axons now Slow axonal transport: the subunit transport model. population of retrogradely moving neurofila- permits, for the first time, direct analysis of Trends Cell Biol. 7, 384–388 (1997). 23. Terasaki, M., Schmidek, A., Galbraith, J. A., Gallant, P. E. ments in axons, previous studies indicate that the molecular mechanism of this remarkable, & Reese, T. S. Transport of cytoskeletal elements in the it represents no more than 5% of the total and once intractable, motile phenomenon. squid giant axon. Proc. Natl Acad. Sci. USA 92, 40 11500–11503 (1995). transported neurofilament protein . Anthony Brown is at the Neuroscience Program, 24. Ahmad, F. J. & Baas, P. W. Microtubules released from Bidirectional movement of neurofilaments Department of Biological Sciences, Ohio the neuronal centrosome are transported into the axon. along microtubules could be achieved by a University, Athens, Ohio 45701, USA. e-mail: J. Cell Sci. 108, 2761–2769 (1995). 25. Yu, W., Schwei, M. J. & Baas, P. W. Microtubule plus-end-directed motor such as kinesin or a [email protected] transport and assembly during axon growth. J. Cell Biol. kinesin-related protein, and a minus-end- 133, 151–157 (1996). Links 26. Slaughter, T., Wang, J. & Black, M. M. Microtubule directed motor such as cytoplasmic dynein transport from the cell body into the axons of growing FURTHER INFORMATION Movies of moving (FIG. 3). Dynein, dynactin and several putative neurons. J. Neurosci. 17, 5807–5819 (1997). neurofilaments | The Brown lab page 27. Ahmad, F. J., Echeverri, C. J., Vallee, R. B. & Baas, P. W. kinesin-related proteins have been identified Cytoplasmic dynein and dynactin are required for the in neurofilament preparations by immun- 1. Lasek, R. J., Garner, J. A. & Brady, S. T. Axonal transport transport of microtubules into the axon. J. Cell Biol. 140, of the cytoplasmic matrix. J. Cell Biol. 99, S212–S221 391–401 (1998). oblotting, and the retrograde movement of (1984). 28. Galbraith, J. A., Reese, T. S., Schlief, M. L. & Gallant, neurofilaments on microtubules in vitro can 2. Hirokawa, N. Kinesin and dynein superfamily proteins and P. E. Slow transport of unpolymerized tubulin and the mechanism of transport. Science 279, polymerized neurofilament in the squid giant axon. Proc. be partly inhibited by pharmacological 519–526 (1998). Natl Acad. Sci. USA 96, 11589–11594 (1999). inhibitors of dynein and by monoclonal anti- 3. Lim, S.-S., Sammak, P. J. & Borisy, G. G. Progressive 29. Terada, S., Nakata, T., Peterson, A. C. & Hirokawa, N. bodies specific for dynein intermediate and spatially differentiated stability of microtubules in Visualization of slow axonal transport in vivo. Science developing neuronal cells. J. Cell Biol. 109, 253–263 273, 784–788 (1996). 37 chains . A potential role for dynein as a slow (1989). 30. Funakoshi, T., Takeda, S. & Hirokawa, N. Active transport axonal transport motor is also indicated by 4. Lim, S.-S., Edson, K. J., Letourneau, P. C. & Borisy, G. G. of photoactivated tubulin molecules in growing axons A test of microtubule translocation during neurite revealed by new electron microscopic analyses. J. Cell the fact that a substantial proportion of the elongation. J. Cell Biol. 111, 123–130 (1990). Biol. 133, 1347–1354 (1996). dynein and dynactin in axons moves in slow 5. Okabe, S. & Hirokawa, N. Turnover of fluorescently 31. Miller, K. W. & Joshi, H. C. Tubulin transport in neurons. 41 labelled tubulin and actin in the axon. Nature 343, J. Cell Biol. 133, 1355–1366 (1996). component ‘b’ . Less is known about the 479–482 (1990). 32. Yabe, J. T., Pimenta, A. & Shea, T. B. Kinesin-mediated potential roles of kinesin and kinesin-related 6. Okabe, S., Miyasaka, H. & Hirokawa, N. Dynamics of the transport of neurofilament protein oligomers in growing 42 neuronal intermediate filaments. J. Cell Biol. 121, axons. J. Cell Sci. 112, 3799–3814 (1999). proteins in slow axonal transport.Yabe et al. 375–386 (1993). 33. Cao, L.-G. & Wang, Y.-L. Mechanism of the formation of have reported that conventional kinesin, 7. Takeda, S., Okabe, S., Funakoshi, T. & Hirokawa, N. contractile ring in dividing cultured animal cells. I. which is a known vesicle transport motor, Differential dynamics of neurofilament-H protein and Recruitment of preexisting actin filaments into the neurofilament-L protein in neurons. J. Cell Biol. 127, cleavage furrow. J. Cell Biol. 110, 1089–1095 (1990). associates with axonally transported neurofil- 173–185 (1994). 34. Keating, T. J., Peloquin, J. G., Rodionov, V. I., aments, whereas Elluru et al.43 detected little or 8. Sabry, J., O’Connor, T. P. & Kirschner, M. W. Axonal Momcilovic, D. & Borisy, G. G. Microtubule release from transport of tubulin in Ti1 pioneer neurons in situ. the centrosome. Proc. Natl Acad. Sci. USA 94, no conventional kinesin in either of the slow 14, 1247–1256 (1995). 5078–5083 (1997). components. Further investigation of the 9. Takeda, S., Funakoshi, T. & Hirokawa, N. Tubulin 35. Dent, E. W., Callaway, J. L., Szebenyi, G., Baas, P. W. & dynamics in neuronal axons of living zebrafish embryos. Kalil, K. Reorganization and movement of microtubules axonal transport of kinesin and kinesin-relat- Neuron 14, 1257–1264 (1995). in axonal growth cones and developing interstitial ed proteins is clearly required. 10. Reinsch, S. S., Mitchison, T. J. & Kirschner, M. W. branches. J. Neurosci. 19, 8894–8908 (1999). Microtubule polymer assembly and transport during 36. Chang, S., Svitkina, T. M., Borisy, G. G. & Popov, S. V. axonal elongation. J. Cell Biol. 115, 365–379 (1991). Speckle microscopic evaluation of microtubule transport Some questions for the future 11. Okabe, S. & Hirokawa, N. Differential behavior of in growing nerve processes. Nature Cell Biol. 1, 399–403 The motile behaviour of neurofilaments in photoactivated microtubules in growing axons of mouse (1999). and frog neurons. J. Cell Biol. 117, 105–120 (1992). 37. Shah, J. V., Flanagan, L. A., Janmey, P. A. & Leterrier, axons lends support to a general model for 12. Okabe, S. & Hirokawa, N. Do photobleached fluorescent J.-F. Bidirectional translocation of neurofilaments along slow axonal transport characterized by the microtubules move? Re-evaluation of fluorescence laser microtubules mediated in part by dynein/dynactin. Mol. photobleaching both in vitro and in growing Xenopus Biol. Cell (In the press). rapid, infrequent and highly asynchronous axons. J. Cell Biol. 120, 1177–1186 (1993). 38. Prahlad, V., Yoon, M., Moir, R. D., Vale, R. D. & Goldman, movement of cytoskeletal polymers and their 13. Chang, S. H., Rodionov, V. I., Borisy, G. G. & Popov, S. V. R. D. Rapid movements of vimentin on microtubule Transport and turnover of microtubules in frog neurons tracks: Kinesin-dependent assembly of intermediate associated proteins. Proof of this model will depend on the pattern of axonal growth. J. Neurosci. 18, filament networks. J. Cell Biol. 143, 159–170 (1998). require identification of both the structural 821–829 (1998). 39. Glass, J. D. & Griffin, J. W. Retrograde transport of 14. Wang, L., Ho, C.-L., Sun, D., Liem, R. K. H. & Brown, A. radiolabeled cytoskeletal proteins in transected nerves. forms in which other cytoskeletal and cytoso- Rapid movement of axonal neurofilaments interrupted by J. Neurosci. 14, 3915–3921 (1994). lic proteins move in slow axonal transport, prolonged pauses. Nature Cell Biol. 2, 137–141 (2000). 40. Koehnle, T. J. & Brown, A. Slow axonal transport of and of the kinetics of their movement. For 15. Roy, S. et al. Neurofilaments are transported rapidly but neurofilament protein in cultured neurons. J. Cell Biol. intermittently in axons: implications for slow axonal 144, 447–458 (1999). example, do actin and tubulin also move transport. J. Neurosci. 20, 6849–6861 (2000). 41. Susalka, S. J., Hancock, W. O. & Pfister, K. K. Distinct along axons as assembled polymers, and do 16. Lasek, R. J., Paggi, P. & Katz, M. J. Slow axonal cytoplasmic dynein complexes are transported by transport mechanisms move neurofilaments relentlessly different mechanisms in axons. Biochim. Biophys. Acta cytoskeletal polymers serve as the carrier in mouse optic axons. J. Cell Biol. 117, 607–616 (1992). 1496, 76–88 (2000). structures for cytosolic proteins? The answers 17. Nixon, R. A. & Logvinenko, K. B. Multiple fates of newly 42. Yabe, J. T., Jung, C. W., Chan, W. K. H. & Shea, T. B. synthesized neurofilament proteins: Evidence for a Phospho-dependent association of neurofilament to these questions may shed light on funda- stationary neurofilament network distributed non- proteins with kinesin in situ. Cell Motil. 45, mental organizational principles of the cyto- uniformly along axons of retinal ganglion cells. J. Cell Biol. 249–262 (2000). 102, 647–659 (1986). 43. Elluru, R. G., Bloom, G. S. & Brady, S. T. Fast axonal plasm that are applicable to all eukaryotic 18. Nixon, R. A. Dynamic behavior and organization of transport of kinesin in the rat visual system: Functionality cells. Many questions also remain regarding cytoskeletal proteins in neurons: reconciling old and new of kinesin heavy chain isoforms. Mol. Biol. Cell 6, 21–40 findings. Bioessays 20, 798–807 (1998). (1995). the axonal transport of neurofilaments. For 19. Lasek, R. J., Paggi, P. & Katz, M. J. The maximum rate of 44. Dahlstrom, A. B., Czernik, A. J. & Li, J. Y. Organelles in example, do these cytoskeletal polymers asso- neurofilament transport in axons: a view of molecular fast axonal transport — what molecules do they carry in ciate directly with motor proteins or do they transport mechanisms continuously engaged. Brain Res. anterograde vs retrograde directions, as observed in 616, 58–64 (1993). mammalian systems. Mol. Neurobiol. 6, 157–177 (1992). ride piggyback on other moving structures? 20. Black, M. M., Keyser, P. & Sobel, E. Interval between the And what is the significance of the retro- synthesis and assembly of cytoskeletal proteins in cultured neurons. J. Neurosci. 6, 1004–1012 (1986). Acknowledgements gradely moving neurofilaments in axons? 21. Baas, P. W. & Brown, A. Slow axonal transport: the The author thanks Ray Lasek and Peter Baas for stimulating There is still much to be learned, but our abil- polymer transport model. Trends Cell Biol. 7, 380–384 discussions.

156 | NOVEMBER 2000 | VOLUME 1 www.nature.com/reviews/molcellbio