ABSTRACT Growing microtubule ends organize end-tracking proteins into comets of mixed composition. Here using a reconstituted fission yeast system consisting of end-binding protein Mal3,

kinesin Tea2 and cargo Tip1, we found that these proteins can be driven into liquid-phase droplets both in solution and at microtubule ends under crowding conditions. In the absence of

crowding agents, cryo-electron tomography revealed that motor-dependent comets consist of disordered networks where multivalent interactions may facilitate non-stoichiometric accumulation of

cargo Tip1. We found that two disordered protein regions in Mal3 are required for the formation of droplets and motor-dependent accumulation of Tip1, while autonomous Mal3 comet formation

CONTENT BEING VIEWED BY OTHERS EVIDENCE FOR A HURP/EB FREE MIXED-NUCLEOTIDE ZONE IN KINETOCHORE-MICROTUBULES Article Open access 10 August 2022 PHASE SEPARATION OF EB1 GUIDES MICROTUBULE

PLUS-END DYNAMICS Article 19 December 2022 MITOCHONDRIA-ADAPTOR TRAK1 PROMOTES KINESIN-1 DRIVEN TRANSPORT IN CROWDED ENVIRONMENTS Article Open access 19 June 2020 MAIN Growing microtubule

plus-ends recruit an evolutionary conserved network of proteins interacting with end-binding (EB) proteins1. This network exists as a multivalent protein assembly that recognizes features of

growing microtubule ends, such as GTP hydrolysis intermediates2, bent tubulin protofilaments3 and tubulin interfaces that are unavailable on closed microtubules4. In fungi, the microtubule

plus-end tracking (MPET) system is crucial to establish cell polarity by asymmetrically transporting polarity markers to the cellular cortex5,6. Once associated with the cellular cortex,

many of these markers behave like clusters7, which raises the question whether clusters may already be formed at growing microtubule ends before being deposited at the cortex8. A minimal

protein network for MPET was first reconstituted in vitro using purified proteins from _Schizosaccharomyces pombe_9. The three proteins that are necessary and sufficient for successful in

vitro plus-end tracking are Mal3 (EB homologue), Tea2 (kinesin-7 homologue) and Tip1 (CLIP-170 homologue). Accumulation of Tip1 and Tea2 at the microtubule end is Mal3-dependent both in

vitro and in vivo5,10. Mal3 is needed for ATPase activity and processive transport of Tea2 (ref. 11). However, affinity of Mal3 for microtubules is independent of Tea2 and Tip1. Tip1 has

been shown to interact with the EB homology domain of Mal3 through its CAP-Gly domain10, as also shown for Tip1 homologue CLIP-170 and other plus-end tracking proteins (+TIPs) interacting

with EB proteins1,12,13,14. Tea2 interacts with Mal3 through its N-terminal extension and with Tip1 through its coiled-coil region6,11,15. As many of these interactions happen through

disordered protein regions (Extended Data Fig. 1a), we hypothesize that the Mal3/Tip1/Tea2 network may be formed by multivalent low-affinity interactions that are a hallmark of liquid–liquid

phase separation (LLPS)16,17. LLPS is the phenomenon of reversible de-mixing of miscible components from their homogeneous mixture driven by microscopic interactions between the

molecules18. Eukaryotic cells contain many membrane-bound and membrane-less organelles that form through similar phase separation processes. Examples include Cajal bodies, nuclear speckles,

nucleoli, stress granules and P-bodies19,20,21. Recently, a number of microtubule-associated proteins have been reported to undergo similar de-mixing in vitro with proposed relevance for

microtubule dynamics, nucleation, branching and so on22,23,24,25,26. Note, however, that the importance of these liquid- and gel-like assemblies for cellular function is still

controversial16,27. Also, while it is widely accepted that disordered protein regions often drive interactions leading to phase separation17, it should be noted that some of the phenomena

explained through phase separation of disordered regions could be interpreted as being produced by site-specific interactions as well28. In this Article, we investigate the role of

multivalent interactions in the formation of comets of fission yeast MPET proteins at growing microtubule ends in an in vitro reconstitution experiment. Using a combination of fluorescence

microscopy, electron cryo-tomography (cryo-ET) and protein truncation, we study the formation of both phase-separated droplets and comets under crowding and non-crowding conditions, focusing

on the contribution of intrinsically disordered regions (IDRs) in the Mal3 protein. We conclude that multivalent interactions contribute to a network-like architecture of plus-end comets,

forming disordered structures that are easily driven into phase-separated dense droplets under crowding conditions. We propose that these non-stoichiometric structures allow for the

efficient motor-driven accumulation of Tip1 at microtubule ends. We finally use stochastic modelling of motor-driven cargo transport to explore how multivalent interactions may enhance this

accumulation at microtubule ends. RESULTS MPET PROTEINS FORM A COMPLEX ON MICROTUBULE LATTICE AND ENDS We reconstituted the fission yeast MPET network in vitro using bacterially expressed

proteins Mal3, Tea2 and Tip1, as reported previously9 (Fig. 1a). Using total internal reflection fluorescence (TIRF) microscopy and double labelling of either Mal3-Alexa647 and Tip1:GFP

(Fig. 1b) or Mal3-Alexa488 and Tea2-Alexa647 (Fig. 1c), we observed that all three proteins were transported on the microtubule lattice in the direction of the microtubule plus-end and were

all present in an end-tracking comet, confirming that they form a complex. As all three proteins contain disordered, low-complexity regions (Extended Data Fig. 1a), we hypothesized that

efficient plus-end accumulation of the Mal3/Tea2/Tip1 protein network is facilitated through multivalent or non-stochiometric protein interactions. To test this hypothesis, we investigated

the behaviour of the protein network under crowding conditions, first without and then with microtubules. MAL3, TEA2 AND TIP1 CO-CONDENSE UNDER CROWDING CONDITIONS As Mal3 is an autonomous

end-tracker and also plays a key role in motor activation needed for plus-end tracking of the MPET network9, we first focused on the ability of Mal3 to form condensates. At high

concentrations, Mal3 readily formed condensates in the presence of polyethylene glycol (PEG) 6k (Fig. 1d) that fused together like fluid droplets (Supplementary Video 1). To probe the

robustness of droplet formation, we systematically explored the effects of Mal3 and PEG concentration as well as PEG chain length. At 200 nM Mal3:mCherry, a typical concentration in

microtubule end-tracking assays, and 5% (w/v) of PEG-35k, Mal3 produced robust protein droplets (Extended Data Fig. 1b). In the additional presence of 20 nM Tea2-Alexa647 and 150 nM

Tip1:GFP, typical concentrations for microtubule end-tracking reconstitutions, we observed co-localization of Tea2 and Tip1 with Mal3 condensates (Fig. 1e). Also, Tea2 and Tip1 formed

condensates under similar crowding conditions and concentrations on their own (Extended Data Fig. 5). MAL3, TEA2 AND TIP1 CO-CONDENSE ON MICROTUBULES In the presence of dynamic microtubules

growing from coverslip-anchored seeds and 5% PEG-35k, Mal3:GFP coated the entire microtubule lattice (Fig. 1f). When Tea2 and Tip1:GFP were added to the PEG-containing assay, we observed

both motor traces at the lattice and bright comets at microtubule plus-ends (Fig. 1g). These plus-end-bound comets could transfer from the end of one microtubule to the lattice of another,

spread out, be transported again towards the new plus-end and then merge with the comet of the second microtubule (Fig. 1h and Supplementary Video 2). When only immobilized seeds but no

soluble tubulin were present, we observed Mal3 binding to the GMPCPP seeds, contrary to non-crowding conditions where Mal3 did not interact with the seeds (Extended Data Fig. 1c). In the

presence of all MPET proteins, we observed Tip1:GFP transport towards the plus-end on seeds. Presumably, PEG-assisted Mal3 binding to the seeds was sufficient to induce Tea2 activity and

hence Tip1:GFP transport towards the plus-end. Droplets were observed to form at the plus-ends of the seeds that grew over time owing to continuous Tea2-driven transport along the seeds

(Fig. 1i (top) and Supplementary Video 3). Finally, when seeds were not attached to the coverslips, and motors were non-specifically binding to the glass surface, seeds started gliding and

depositing trails of droplets behind their plus-ends (Fig. 1i (bottom) and Supplementary Video 4), similar to the Plateau–Rayleigh instability29,30. Together, these observations provide

evidence that, in the presence of crowding agent, Mal3, Tip1 and Tea2 together form condensates both in the absence and in the presence of microtubules. The observed condensates are

liquid-like in nature, can coat the microtubule lattice and can be transported by Tea2 motors towards the plus-ends of microtubules. CRYO-ET OF MPET PROTEIN DROPLETS AND COMETS We next asked

whether droplet-like comets also form in the absence of crowding agents. Given the small size of normal comets, we turned to cryo-ET for higher spatial resolution. We first added pre-formed

droplets made by incubating the Mal3/Tip1/Tea2 mixture with 10% PEG-6k to holey carbon grids and vitrified them (Fig. 2a). In these conditions, we observed spherical droplets with fine

internal grain (Fig. 2b,c). To prevent non-specific adsorption in experiments with tubulin, we adapted passivation methods previously established for treatment of glass coverslips31. We

silanized a SiO film on the grids, adsorbed anti-DIG IgG to the silanized surface and then made the film hydrophilic by incubation with Pluronic F-127 (Fig. 2d). This treatment allowed us to

firmly attach DIG-labelled GMPCPP seeds, while rejecting the binding of other proteins from solution. We then added tubulin in the presence or absence of Mal3 or the complete MPET network

and plunge-froze the grids after 5–7 min of microtubule growth. To facilitate analysis of microtubule end structures, we used cryoCARE, a neural network-driven denoising algorithm designed

to increase the signal to noise ratio in individual tomograms32 (for details, see Methods; Extended Data Fig. 2). Microtubule polarity was determined from moiré patterns or protofilament

shapes in microtubule cross-sections33,34 (Extended Data Fig. 3a and Supplementary Table 1). In the absence of end-tracking proteins, we observed microtubules growing with flared

protofilaments at their ends, as described previously35, and no lattice or end decoration (Fig. 2e). Adding Mal3 alone did not produce clearly visible densities at microtubule ends (Fig.

2e), but we observed a clear diffuse coating at the ends of growing microtubules when all MPET components were present (Fig. 2e and Extended Data Fig. 3). To assist the interpretation of the

reconstructed tomograms, we used volume segmentation to highlight tubulin and microtubules (cyan) and non-tubulin densities (yellow) (Fig. 2f). Together with polarity assignment, this

allowed us to visualize massive microtubule end-bound structures at the plus-ends in the presence of Mal3/Tea2/Tip1 (Fig. 2f, Extended Data Fig. 3b and Supplementary Video 5). Structures

binding to minus-ends in the presence of Mal3/Tea2/Tip1 (Supplementary Video 6) to plus-ends in the presence of Mal3 alone (Extended Data Fig. 3c and Supplementary Video 7) or to plus-ends

in the absence of additional proteins (Extended Data Fig. 3d and Supplementary Video 8) appeared much smaller. Interestingly, when we added Mal3/Tea2/Tip1 in the presence of PEG to

microtubules pre-polymerized in the presence of Mal3/Tea2/Tip1 without PEG, we observed a subset of comets that looked similar to the ones we observed in the absence of PEG, and a subset

that were surrounded by diffuse material extending over hundreds of nanometres from the microtubule walls (Fig. 2e and Extended Data Fig. 3e). We further analysed microtubule cross-sections

to obtain quantitative information on the microtubule end-bound structures (Fig. 2g,h). The average thickness of comets extending outwards from the microtubule surface in the presence of

Mal3 alone was 11 ± 7 nm (here and onwards mean ± standard deviation (s.d.)), considerably thinner than 25 ± 12 nm in the presence of Mal3, Tip1 and Tea2 (Fig. 2i). The presence of PEG did

not result in a statistically significant difference in comet length or thickness (Fig. 2h,i) (not taking into account the diffuse material surrounding the comets). The differences in comet

length were not significant (Fig. 2h). Plus-ends carried thicker comets of Mal3/Tip1/Tea2 (29 ± 11 nm) compared with minus-ends in the same sample (19 ± 9 nm) (Fig. 2j, right). The

polarity-dependent thickness of comets is consistent with the plus-end-directed motility of Tea2 bringing its cargo, Tip1, to the plus-ends of microtubules. There is clearly a difference

between the shape and the internal organization of the spherical PEG-driven droplets shown in Fig. 2b,c and the microtubule-bound comets that appear as more loosely structured.

Interestingly, when PEG together with Mal3/Tip1/Tea2 was added to comets pre-formed in the absence of PEG, a separate comet structure remained visible even when surrounded by diffuse

material, hinting that these structures were formed in different ways. Yet, it is possible that multivalent interactions responsible for the formation of droplets under crowding conditions

are also facilitating the formation of the network-like architecture of motor-driven plus-end comets observed in cryo-ET. NON-STOICHIOMETRIC TIP1 ACCUMULATION AT MICROTUBULE ENDS The

network-like architecture described above may facilitate the non-stoichiometric accumulation of Tip1 cargo at microtubule ends. To address this hypothesis, we measured fluorescence

intensities of MPET proteins along the microtubules at two Tea2 concentrations (20 nM and 100 nM; Fig. 3a). The average intensity profiles of Tea2Alexa647, Tip1:GFP and Mal3-Alexa647

demonstrated a similar, specific shape: a shallow intensity increase starting at the microtubule seed, a constant average intensity along the microtubule lattice and a peak at the

microtubule plus-end (Fig. 3b–d). Interestingly, at higher motor concentration, Tip1 intensity increased more than Tea2 intensity itself both on the microtubule lattice and at microtubule

ends (Fig. 3b,c). In contrast, the intensity of Mal3-Alexa647 did not change with Tea2 concentration (Fig. 3d). We summarized the effect of Tea2 concentration on the end accumulation of Mal3

and Tip1 by calculating the ratios of intensities between the two Tea2 concentrations for Mal3 and Tip1 on the microtubule lattice and in the comet (Fig. 3e). An increase in Tea2

concentration had no influence on the amount of labelled Mal3 protein that localized at the microtubule plus-end. On the other hand, Tip1:GFP localization to the plus-end was

disproportionately affected by Tea2 concentration. An increase from 20 nM to 100 nM Tea2 led to a roughly fourfold increase of Tip1:GFP intensity at the plus-end, whereas the Tea2 intensity

itself was increased by only a factor of 2. Apparently, the amount of Tip1 that is present on the microtubule does not follow the density of motor proteins on the microtubule in a

stoichiometric way. In fact, we show that the presence of Tip1 responds in a non-linear way to the concentration of Tea2 over a range of concentrations (Extended Data Fig. 4), in agreement

with previous observations8. Note also that there is large variability in the Tip1 intensity between individual microtubules (Extended Data Fig. 4a), which we interpret as another sign that

the accumulation of the cargo Tip1 is not limited by one-on-one interactions with motor proteins. DISTINCT DOMAINS OF MAL3 DRIVE FORMATION OF COMETS AND LLPS Having established that our

three-component network is capable of both droplet formation and non-stoichiometric protein accumulation at microtubule ends, we set out to elucidate the contributions of disordered protein

regions to both comet formation and LLPS. As Mal3 is central to comet formation of all three proteins, we studied different truncations of Mal3. Full-length Mal3 contains two folded domains:

a calponin-homology (CH) domain and an EB-homology domain (EB HD), and two IDRs: IDR1, which connects the CH domain to EB HD, and the C-terminal IDR2 (Fig. 4a). Note that the C-terminal

IDR2 domain is not present in Mal3’s homologue EB1, which contains a much shorter negatively charged C-terminal tail36. We first focused on dissecting the contributions of these domains to

formation of Mal3 comets on microtubule ends without Tea2 or Tip1, and in the absence of crowding conditions (Fig. 4b,c). In the absence of Tea2 and Tip1, 200 nM of full length Mal3:GFP

coated the entire microtubule lattice without a clear saturation at the plus-end, in contrast to the full MPET network (compare Figs. 4c (top left) and 1b, respectively). Mal3-∆IDR2 showed a

lower binding affinity to the microtubule lattice than full-length Mal3, and formed slightly brighter comets at the microtubule ends (Fig. 4c,d). We did not observe any comet formation or

lattice binding with Mal3 mutants containing only the CH domain (Fig. 4b,c). Other Mal3 mutants were binding mostly to the growing end, rather than the microtubule lattice, and formed comets

with very low intensity (Fig. 4d). We conclude that, in addition to previously described microtubule binding through the CH domain and the role of dimerization through the EB HD37,38, IDR1

also contributes to efficient comet formation by Mal3. In contrast, IDR2 appears not to contribute to Mal3’s affinity to the microtubule end but only to its affinity to the lattice

(potentially via Mal3 self-interactions; see below). Note that in previous work on EB1 truncations, it was observed that removal of the C-terminal tail (where IDR2 is located in Mal3) led to

stronger instead of weaker lattice binding36. This effect was attributed to the removal of a short negatively charged section of the protein, which is expected to destabilize electrostatic

interactions with the positively charged microtubule lattice. While it is difficult to disentangle the effect of charge from the contribution of multivalent interactions, it should be noted

that, unlike Mal3, EB1 does not have a sizeable, disordered region at its C-terminal end. We next wondered which domains of Mal3 were important for the protein’s self-interactions under

crowding conditions. When Mal3 mutants were incubated at a concentration 1 µM with 5% PEG-35k, we observed that domain deletions preventing comet formation on microtubules also prevented

droplet formation (Fig. 5a). In addition, Mal3-∆IDR2, which reduced microtubule lattice- but not microtubule end-binding, also formed smaller condensates than the full-length protein in the

presence of a crowding agent. We conclude that IDR1 and EB HD are necessary both for Mal3 self-interactions and for Mal3 interaction with the microtubule end, while IDR2 is necessary for

Mal3 self-interactions and interaction with the microtubule lattice, but not the microtubule end (Figs. 4b,c and 5a). ROLES OF MAL3 DOMAINS IN DROPLET AND COMET FORMATION To pin-point the

interactions between Mal3, Tea2 and Tip1 under crowding conditions, we next designed a scaffold-client assay (Fig. 5b,c and Extended Data Fig. 5). Scaffold condensates were formed by either

Mal3, Tea2-Alexa647 or Tip1 by incubation with 10% PEG-6k, and 2 nM Mal3:GFP was added as a client. When non-fluorescent Mal3 and Tip1 were used, we additionally added 2 nM full-length

Mal3:mCherry as a tag to visualize the scaffold condensates independent of Mal3:GFP construct localization. Figure 5b shows the outcome of a typical experiment, with non-fluorescent Mal3 as

the scaffold (tagged with Mal3:mCherry), and full-length Mal3:GFP as a client. Deletion of any disordered region from Mal3:GFP prevented its recruitment to the Mal3 scaffold (Fig. 5c),

reinforcing our conclusion that both IDR1 and IDR2 are important for Mal3–Mal3 interactions in crowding conditions. We observed a direct interaction between full-length Tea2 and Mal3 in

crowding conditions in the absence of Tip1 (Extended Data Fig. 5a). However, Mal3 constructs lacking IDR1 or IDR2 were recruited poorly to Tea2-Alexa647 scaffold (Extended Data Fig. 5a).

Deletion of EB HD further disrupted recruitment of Mal3 to the Tea2 scaffold. These data indicate that crowding conditions strengthen Tea2–Mal3 interactions and that these interactions rely

on the disordered regions in Mal3 as well as the EB HD. Finally, we used unlabelled Tip1 as the scaffold (Extended Data Fig. 5b) and Mal3:GFP truncations as the client. We again observed

that Tip1 condensates predominantly recruited full-length Mal3:GFP, and to a much lesser extent Mal3-∆IDR2, but failed to recruit the Mal3 constructs lacking the EB homology domain or IDR1.

We finally set out to correlate the recruitment behaviour observed in the scaffold-client assays with the capacity of truncated Mal3 constructs to couple Tip1/Tea2 transport to plus-end

tracking on dynamic microtubules. Using Tip1:GFP fluorescence as a readout, we observed that Mal3 constructs lacking either IDR1, EB homology domain or both failed to recruit Tip1:GFP to

microtubules altogether (Extended Data Fig. 6). Mal3-ΔIDR2 was still able to support Tip1 localization at microtubule ends, but unlike full-length Mal3, it did not co-localize with Tea2/Tip1

transported along the microtubule lattice (Fig. 5d,e). Furthermore, the intensity of both Mal3-ΔIDR2 and Tip1 in the comets was reduced compared with full-length Mal3 (Fig. 5f,g). Together,

the analysis of Mal3 truncations leads us to conclude that robust three-component comets are formed by a combination of different molecular mechanisms. Mal3 interaction with itself, Mal3

interaction with the microtubule lattice, as well as Mal3 co-localization with motor tracks requires each of Mal3’s IDRs. The formation of Mal3/Tip1 comets requires only Mal3 IDR1 and EB HD,

but the additional presence of IDR2 enhances the motor-dependent accumulation of Tip1 at growing microtubule ends. It thus appears that Mal3 self-interactions are needed to promote

non-stoichiometric Tea2/Tip1 transport on the microtubule lattice. THEORETICAL MODELS FOR MOTOR-DRIVEN END ACCUMULATION To help understand the possible contribution of protein

self-interactions to efficient motor-driven end accumulation, we turned to stochastic simulations, complementing a series of previously published models of single-component traffic jams39.

It should be stressed that these simulations were not designed to exactly reproduce our experimental situation, which is highly complex: plus-end accumulation of the three MPET network

components (Mal3, Tea2 and Tip1) is a result of both motor-driven transport towards the plus-end and autonomous interaction of Mal3 with the comet region near the growing MT ends. Varying

the concentration of each of the components is likely to change the balance of complex formation both in solution and on the microtubules, complicating straightforward predictions about the

resulting effects on both lattice coverage and end accumulation. The phenomenology of motor transport in the absence of cargo is well known40,41, and because the binding/unbinding of cargo

is an equilibrium process, it is not expected that simple 1:1 cargo binding changes any characteristic of these models. We therefore specifically focused on the possible effects of cargo

clustering due to protein self-interactions. A microtubule was represented as a growing one-dimensional lattice42,43, and the motors as particles binding to and unbinding from the lattice

and hopping towards the plus-end39,44,45, where each lattice site can be occupied by only one motor (Fig. 6a). The cargo was represented by a second set of particles that bind to and unbind

from the motor particles. Mal3 was not simulated explicitly, because in our experiments Mal3 localization was not affected by motor concentration (Fig. 3d and Extended Data Fig. 4b).

Instead, to represent the effects of Mal3, we assumed different motor/cargo behaviours at microtubule end and lattice, and studied different scenarios for cargo oligomerization. We first

investigated the effect of motor slowdown in the comet region near the microtubule plus-end. Mimicking the hydrolysis state of GTP using GTPγS microtubules, we observed that motor intensity

increased and motor speed slowed down (Extended Data Fig. 7a–c). Simulations show that motor slowdown near the microtubule end indeed leads to end accumulation (Fig. 6b), an effect that is

due to a traffic jam at the transition from the fast to the slow parts of the lattice (dashed line in Fig. 6b). This type of traffic jam is different from the previously reported formation

of ‘spikes’39, which is due to a reduced motor off-rate at the microtubule end. Inspired by our experimental observations in guest-host and end-tracking assays, by evidence that Tip1 may be

able to oligomerize8,46, and by structural data suggesting interactions between Tip1’s CAP-Gly domain and its C-terminal zinc finger domain13,47, we next considered the effect of lateral

interactions between neighbouring motor-bound cargo molecules (Fig. 6c). When we simulated cargo particles as synchronously moving oligomeric cargo trains48, the effective flux of cargo on

the microtubule increased, but in the absence of any end-specific effects, this did not lead to accumulation of cargo at the microtubule end (Fig. 6c). Accumulation of cargo was, however,

recovered by introducing motor slowdown at the microtubule end as in Fig. 6b. Finally, we explored the effect of increased stability of cargo clusters: motor-cargo neighbouring with at least

one other motor-cargo was given a higher dwell time compared with non-clustered motors (Fig. 6d). Even a three-fold increase in dwell time was not sufficient to cause end accumulation (Fig.

6d). However, addition of the end-dependent slowing down resulted in pronounced cargo accumulation at the microtubule end (Fig. 6d). Note that stabilization of oligomeric clusters also

increases the lattice occupancy away from the microtubule end (Fig. 6d,e). For all scenarios, we also investigated how the accumulation of the cargo depends on the concentration of motors in

the model. Only the scenario in which motors slow down at the microtubule end and cargo clusters are stabilized by lateral interactions resulted in non-linear end accumulation of cargo

(Fig. 6e,f). DISCUSSION In this study, we systematically dissected the role of multivalent interactions within the MPET network reconstituted in vitro using recombinant Mal3, Tip1 and Tea2

from _S. pombe_. We found that in vitro molecular crowding agents, such as PEG, drove these proteins into spherical droplets that displayed liquid-like properties: they fused with each other

over time, wetted the microtubule surface and transferred from one microtubule to another. This behaviour shows similarity to the previously observed transfer of end-tracking protein

clusters from a microtubule end to a solid barrier8, and might be relevant in vivo for the cortical deposition of polarity markers that are crucial for the physiology of fission yeast such

as Tea1, Tea4 and Tea3 in addition to Mal3, Tea2 and Tip1 (refs. 49,50,51,52,53). Under crowding conditions Mal3, Tip1 and Tea2 co-existed in the same condensed phase. Although interactions

between these proteins were reported previously, it remained unclear which domains of Mal3 were involved10,11,15. Here we show that Mal3 IDR1 and IDR2 are responsible for interactions with

Tip1 and Tea2 in the absence of microtubules (Fig. 5c and Extended Data Fig. 5a,b). Deletion of these disordered regions impaired formation of Mal3 droplets in crowding conditions (Fig. 5a),

in accordance with the idea that disordered regions are the main drivers of LLPS16,17. We further found that Mal3 IDR1, in combination with EB HD, is crucial for Mal3’s accumulation at

growing microtubule plus-ends (Fig. 4 and Extended Data Fig. 6). Importantly, Mal3-ΔIDR2, for which we also observed severely impaired droplet formation and interaction with Tip1 and Tea2 in

crowding conditions, was nevertheless able to form comets at growing microtubule ends (Fig. 4b–d) and recruit Tip1 to these comets (Fig. 5e,g). However, in comets formed by Mal3-ΔIDR2, both

Mal3 and Tip1 intensity were reduced, and Mal3 association with the lattice (Fig. 4b,c) and Tea2 transport (Fig. 5e,g) was also reduced. We conclude that robust motor-driven transport and

accumulation of Tip1 at microtubule ends depend on both Mal3 IDR1 and IDR2, leading to the suggestion that Mal3 self-interactions responsible for LLPS are also responsible for protein

interactions in the network-like structures observed at microtubule ends in cryo-ET (Fig. 2e,f). The question that then remains is whether the network-like structures observed in cryo-ET in

the absence of crowding agents show characteristics of liquid-like droplets and/or whether this is expected to be the case for end-tracking complexes in vivo. Clearly, PEG-driven droplets in

the absence of microtubules displayed a characteristic dense internal grain in cryo-ET (Fig. 2b,c) that was not seen in the microtubule end-tracking comets. The comets appeared as loosely

structured densities (Fig. 2e,f) which did not extend further than 55 nm from the microtubule lattice (Fig. 2g,h). This even remained the case when Mal3/Tip1/Tea2 together with PEG were

added to pre-formed comets (Fig. 2e), when we sometimes observed an additional layer around the comet. Given the estimated dimensions of Mal3 (3 × 6 × 10 nm) (refs. 54,55), Tea2 (4 × 4 × 7 

nm) and Tip1 (predicted 40-nm-long coiled coil), it is technically possible that all the molecules inside the comet are directly interacting with microtubule surface. On the other hand, it

is also possible that the loose network represents a liquid-like structure where multiple dynamic, weak interactions between its components facilitate the observed non-stoichiometric

accumulation of plus-end trackers and allow them to behave as a protein cluster8. In vivo, where crowding effects could be different from the conditions of our cryo-ET experiments, these

clusters may again appear as dense droplets as we observed in the presence of PEG. We must also note that preserving droplets at the ends of microtubules during our cryo-ET sample

preparation may be technically challenging, potentially limiting our ability to visualize these structures. In conclusion, our study suggests that microtubule ends may act as platforms where

multivalent interactions condense microtubule-associated proteins into large complexes. Our observations are paralleled by observations in three other biological systems: formation of Kar9

nanodroplets at the ends of specialized microtubules in budding yeast56, formation of droplets by human EB3 and CLIP-170 at microtubule ends57, and condensation of EB1 affecting chromosome

segregation during mitosis58. LLPS at microtubule ends is thus emerging as a general organizing principle that may explain how different end-tracking proteins may (simultaneously) associate

with microtubule ends to perform their wide range of biological functions. METHODS PROTEIN EXPRESSION, PURIFICATION AND LABELLING Full-length _S. pombe_ Mal3 and all of its derivatives (that

is, truncations and superfolder-GFP fusions) were expressed with an N-terminal His8 tag followed by a 3C protease site from a pBADTOPO derived plasmid in _Escherichia coli_ ER2566 cells

(New England Biolabs, _fhuA2 lacZ::T7 gene1 [lon] ompT gal sulA11 R(mcr73::miniTn10–Tet__S__)2 [dcm] R(zgb-210::Tn10–Tet__S__) endA1 Δ(mcrCmrr)114::IS10)_). Mal3(-truncations) were

covalently linked to superfolder-GFP by a flexible ASTGILGAPSGGGATAGAGGAGGPAGLINPGGSTSSRAAEIWPAS ‘happy linker’ sequence. Cells were grown at 37 °C in baffled flasks on LB supplemented with

100 µg ml−1 ampicillin, expression was induced at an OD600 of 0.6, and cells were collected after 3 h (8 min 4,500 rpm, JLA8.1000 rotor). After washing the cells in PBS, they were lysed

using a French Press (Constant Systems) at 20 kpsi, 4 °C, and unbroken cells, debris and aggregates were pelleted in a Ti45 rotor (30 min, 40,000 rpm, 4 °C). The lysate was applied to 2 ml

Talon Superflow resin (Clontech) pre-equilibrated with buffer A (20 mM Tris–HCl pH 7.5, 200 mM NaCl and 5% (w/v) glycerol), and incubated for 1 h while rotating at 4 °C. Subsequently, the

resin was washed with 50 ml of buffer A supplemented with 0.1% Tween20 and 50 ml of buffer A supplemented with additional 500 mM NaCl, and finally Mal3 was eluted in 10 ml of buffer A

supplemented with 1 mM β-mercaptoethanol and homemade 3C protease. Proteins were concentrated using a Vivaspin centrifugal concentrator (10 kDa cut-off) and further purified by size

exclusion chromatography (SEC) on a Superdex 200 Increase 10/300 column pre-equilibrated with buffer B (20 mM Tris–HCl, 100 mM NaCl and 5% (w/v) glycerol). Mal3 was labelled by dialysing ~1 

mg of protein into buffer C (80 mM PIPES pH 6.8, 1 mM MgCl2 1 mM ethylene glycol tetraacetic acid (EGTA) and 100 mM NaCl) and incubating for 1 h at room temperature with 140 µM Alexa Fluor

488 TFP ester or Alexa Fluor 647 TFP ester (Thermo Fisher). After quenching the reaction with excess Tris–HCl, the free label was removed by SEC on a Superdex 200 Increase 10/300 column

pre-equilibrated with buffer B. Full-length _S. pombe_ Tea2 was expressed with an N-terminal Z-tag followed by a TEV protease recognition site, and purified essentially as described9, but

with the following modifications: after washing of the Talon resin with 15 mM imidazole in buffer D (50 mM KPi pH 8.0, 400 mM NaCl, 2 mM MgCl2, 0.2 mM MgATP and 0.05 mM TCEP), Tea2 was

eluted in buffer D supplemented with homemade 3C protease by taking advantage of crossreactivity with the TEV recognition site. Following concentration using a Vivaspin centrifugal

concentrator (10 kDa cut-off), Tea2 was labelled with 138 µM Alexa Fluor 647 NHS ester (Thermo Fisher) by incubating 30 min at room temperature. After quenching the reaction with excess

Tris–HCl, the free label was removed by SEC on a Superdex 200 Increase 10/300 column pre-equilibrated with buffer D. Unlabelled Tea2 was applied to the SEC column directly after

concentrating. FLOW CELL PREPARATION Coverslips and glass slides were cleaned using base Piranha (NH4OH:H2O2 in 3:1 at 75 °C) for 10 min and sonicated in MilliQ water for 5 min. Flow cells

were prepared by sandwiching two strips of parafilms between the glass slide and the coverslip. The strips were placed about 3–5 mm apart approximately from each other. The flow cell was

then placed on top of a hot plate, kept at 120 °C, to let the parafilm melt and seal the glass slid with the coverslip. MICROTUBULE BIOCHEMISTRY GMPCPP-STABILIZED MICROTUBULE SEEDS

Microtubule seeds were prepared by two cycles of polymerization with GMPCPP in MRB80 buffer (80 mM PIPES, 4 mM MgCl2 and 1 mM EGTA, pH 6.8). First 20 µM tubulin (25% HyLite 647, 10%

biotinylated and 65% unlabelled) was polymerized in the presence of 1 mM GMPCPP (NU-405 Jena BioScience) at 37 °C for 30 min. The mix was centrifuged for 5 min at 200,000 _g_ with an

air-driven ultracentrifuge, airfuge (Beckman Coulter), and the pellet was resuspended in MRB80 (80% of the initial volume) and kept on ice for 20 min for depolymerization. For the second

polymerization step, again 1 mM of GMPCPP was added to the mix and the mix was incubated at 37 °C for another 30 min. After 30 min of incubation the mix was ultracentrifuged using an airfuge

at 200,000 _g_ and the pellet was resuspended in 50 µl MRB80 with 10% glycerol. The seeds thus prepared were aliquoted, flash frozen in liquid nitrogen and stored at −80 °C. END-TRACKING

RECONSTITUTION ASSAYS To functionalize the glass surface, the channels in the flow cells were first filled with 0.2 mg ml−1 PLL(20)-g[3.1]-PEG(2)/-PEG(3.4)-biotin(17.5%) (SUSOS AG) then 0.1 

mg ml−1 neutravidin and finally with κ-casein (Sigma). Ten minute incubation at room temperature was maintained before the subsequent steps. The channels were then washed with MRB80 and

incubated with biotinylated seeds for 5 min. After 5 min, the reaction mix was added to the channels. The channels were sealed with VALAP before starting the observations on the microscope

to avoid evaporation. To reconstitute plus-end-tracking assays with full-length Mal3 and Mal3 mutants, the reaction mix contained 200 nM Mal3/Mal3 construct, 20 nM Tea2 and 150 nM Tip1 in

MRB80 buffer containing 14.5 µM tubulin, 0.5 µM rhodamine tubulin, 50 mM KCl, 0.5 mg ml−1 κ-casein, 0.4 mg ml−1 glucose oxidase, 50 mM catalase, 0.1% methylcellulose, 1 mM GTP and 2 mM ATP.

PHASE SEPARATION ASSAYS CONDENSATES ON DYNAMIC MICROTUBULES The assay was performed in two steps. In the first step a dynamic microtubule assay was set up in a flow cell and in the second

step condensates were added. To set up a dynamic microtubule assay, a reaction mix with 14.5 µM tubulin, 0.5 µM rhodamine tubulin, 50 mM KCl, 0.5 mg ml−1 κ-casein, 0.4 mg ml−1 glucose

oxidase, 50 mM catalase, 0.1% methylcellulose, 1 mM GTP and 2 mM ATP in MRB80. The flow cell was then left for incubation at 37 °C for 15 min. After 15 min the tubulin was washed off using

MRB80 (pre-warmed at 37 °C) and condensates were added to the flow cell immediately to the flow cell. The condensates were prepared by incubating 200 nM Mal3, 20 nM Tea2 and 150 nM Tip1 in

MRB80 buffer containing 50 mM KCl, 0.5 mg ml−1 κ-casein, 0.4 mg ml−1 glucose oxidase, 50 mM catalase, 0.1% methylcellulose, 1 mM GTP and 2 mM ATP on ice for 1 h with 5% PEG-35k.

SCAFFOLD-CLIENT EXPERIMENTS Coverslips were cleaned as described above. Glass slides were cleaned in a 250 ml beaker with a custom-made Teflon rack by repeated (2×) sonication and washing

steps as follows 1% Hellmanex (10 min), MilliQ water (5 min), 70% ethanol (10 min), MilliQ (5 min) and stored in the beaker with MilliQ covered by parafilm. Before use, slides were rinsed

with MilliQ and dried with N2. Flow cells were prepared by cutting six channels into a piece of parafilm with a razor blade. The parafilm was sandwiched between the clean glass slide and the

cover glass and heated on a piece of aluminium foil on top of a 120 °C hot plate until the parafilm melted and cover glass was gently pressed with tweezers to assure that channels were

sealed off well. The parafilm overhangs were removed with the blade while the glass was still hot. After cooling to room temperature, the channels were incubated for 10 min with 0.2 mg ml−1

PLL(20)-g[3.1]-PEG(2) (SUSO AG), rinsed and incubated for 10 min with 0.5 mg ml−1 κ-casein (Sigma), all solutions were MTB80 buffer. The scaffold and client condensates were prepared on ice

by first eluting all proteins into MRB80 buffer containing 250 mM KCl, and then further diluting them into the reaction mixture, at a final composition of 1× MRB80, 50 mM KCl, 10% PEG6k and

freshly thawed 2 mM ATP, 1 mM GTP, 2 mM dithiothreitol and 0.5 mM β-mercaptoethanol. Solutions were clarified for 5 min at 200,000 _g_ using an airfuge and kept on ice for 15 more minutes

before being transferred into flow cells. Imaging occurred approximately 30 min after mixing. Mal3 and Tip1 host condensates were prepared with 200 nM FL Mal3 or 215 nM Tip1, 2 nM FL

Mal3:mCherry and 2 nM of each of the constructs (Fig. 3a). Tea2 scaffold condensates were prepared from 200 mM Tea2-Alexa647 and 2 nM of each of the constructs. Experiments for each scaffold

protein were conducted in parallel. Image acquisition was performed using spinning disc confocal microscopy (CSU-W1, Yokogawa; Ilas2, Roper Scientific) with the scanning slide module in the

Ilas2 software. CRYO-ELECTRON TOMOGRAPHY To study PEG-driven droplets, a solution containing 200 nM Mal3, 150 nM Tip1 and 80 nM Tea2 was incubated with 10% of PEG-8k in MRB80; 4 µl of this

solution was mixed with 5 nm gold nanoparticles (OD50, final dilution 1:20) and added to freshly glow-discharged copper grids with R2/2 Quantifoil film. The grid was blotted from the back

side for 4–6 s in a Leica EM GP plunger and immediately plunge-frozen in liquid ethane. To reconstitute comet formation, we used copper mesh grids with holey SiO film (SPI Supplies), coated

with 5 nm gold on one side. The grids were treated with oxygen plasma for 2 min and immediately submerged in Plus-One Repel Silane solution (GE Life Sciences) for 3 min, then washed in

ethanol and dried. A silanized grid was incubated in a drop of anti-DIG IgG (0.2 µM, Roche), washed with MRB80, incubated in a drop of 1% Pluronic F-127 and washed again with MRB80. The

passivated grid was then taken into the chamber of the Leica EM GP2 plunger equilibrated at 95% relative humidity and 26 °C. Inside the chamber, GMPCPP-stabilized, DIG-labelled microtubule

seeds were added to the grid for 1 min followed by a wash with MRB80 supplemented by 0.5 mg ml−1 κ-casein and finally a 4 µl drop of a solution containing 200 nM Mal3, 150 nM Tip1 and 80 nM

Tea2 in MRB80 supplemented with 25 µM tubulin, 0.01% Tween20, 2 mM ATP, 1 mM GTP and 1 mM dithiothreitol. The microtubules were allowed to grow for 7 min, after which 5 nm gold nanoparticles

were added (OD50, final dilution 1:20), the grid was blotted from the back side for 3–4 s and immediately plunge-frozen in liquid ethane. All grids were stored in closed boxes in liquid

nitrogen until further use. Tilt series were recorded on a JEM3200FSC microscope (JEOL) equipped with a K2 Summit direct electron detector (Gatan) and an in-column energy filter operated in

zero-loss imaging mode with a 30 eV slit width. Images were recorded at 300 kV with a nominal magnification of 10,000×, resulting in the pixel size of 3.668 Å at the specimen level.

Automated image acquisition was performed using SerialEM 3.8.5. software60 with a custom-written script, recording bidirectional tilt series ranging from 0° to ±60° with tilt increment of

2°; a total dose of 100 e− Å−2 and the target defocus set to −4 µm. Individual frames were aligned using MotionCor2 (ref. 61), and then split into odd and even frame stacks. Tilt-series

alignment and tomographic reconstructions were performed with the IMOD software package using gold beads as fiducial markers62. Final tomographic volumes were binned two-fold and

subsequently denoised using the cryoCARE procedure32. For this, 3D reconstruction was performed on aligned sets of odd and even frame stacks with identical IMOD parameters. The full even and

odd tomograms obtained in this way were then split into subvolumes for network training, and eventually full volumes were denoised. The images shown in Fig. 5 were obtained from a

voxel-wise average of odd and even denoised tomograms. Automated segmentation of binned and denoised cryo-tomograms was performed using the _tomoseg_ module of EMAN2 v.2.2 (ref. 63) and

visualized using UCSF Chimera64. TIRF MICROSCOPY Imaging was performed using an inverted Nikon Eclipse Ti-E microscope with perfect focus system, an oil immersion objective (Nikon Plan Apo λ

100× NA 1.45), using two EMCCD cameras (Photometrics Evolve 512), which are mounted on a spinning disc unit (CSU-W1, Yokogawa). TIRF illumination was generated with the FRAP/TIRF system

Ilas2 (Roper Scientific). A custom-made objective heater was used for temperature control of the samples. The imaging software used was Metamorph 7.8.8.0 with system specific routines

(Ilas2) for streaming, time lapse and scanning slide acquisition. STOCHASTIC SIMULATIONS Stochastic simulations were performed using Gillespie’s algorithm65 on the TU Delft Applied Science

in-house linux cluster using an implementation in C++. The different implementations of the model were all simulated, tested and prepared independently. Model parameters were chosen as much

as possible in agreement with experimental conditions and corresponded to a low-density regime (LD phase) in terms of the TASEP/LK model on growing microtubules42. System size was 1,000

lattice sites, each corresponding to the size of one tubulin heterodimer (8.4 nm). Simulations were equilibrated for 105 s before 104 motor and cargo distributions were recorded in time

intervals corresponding to the time it takes for one motor to traverse the system (~50 s). Equilibration times were particularly critical for cargo clustering conditions, since the motor

distributions generically deviate from their classical equilibrium owing to the aggregation and fragmentation kinetics, as seen in similar systems48. Data analysis and plotting was performed

using custom programs and scripts written in C++ and Python (Matplotlib). Details regarding all model parameters and corresponding experimental values can be found in Supplementary Table 2.

DATA ANALYSIS PREPARATION OF DENSITY PROFILES Kymographs were extracted from background-subtracted TIRF microscopy data in a semiautomated way using ImageJ (50 pixel rolling ball radius).

Image projections were used to identify dynamic microtubules in movies (function Z Project with option standard deviation) and positional data were stored in the form of linear regions

(thickness 9 pixels) using the ImageJ ROI manager. Saved regions of interest were used to automatically generate kymographs. Subsequently a MATLAB (R2018b) script was used to analyse

kymographs (dual colour where necessary) in a semi-automated way. The script allows to manually mark regions of growing microtubules with comets as polygons (typically triangular), generates

a mask thereof, and extracts the corresponding intensity profiles from the underlying images. The intensity profiles are saved per experimental condition for further processing. In a

separate step, the intensity profiles were sorted into sets by length using a binning of (±0.64 µm). The set of profiles was aligned by finding alignments which minimize the s.d. of the sum

of differences between a randomly chosen first intensity profile and every other profile in a set. Averages of the aligned sets of data are shown in Extended Data Fig. 4c,d. Regions of end

and lattice intensities were defined manually. For Fig. 3e, individual intensities at 100 nM Tea2 (for all microtubule lengths) were divided by the average of all intensities at 20 nM Tea2.

ANALYSIS OF SCAFFOLD-CLIENT EXPERIMENTS Analysis of scaffold-client experiments was performed using a custom script written in MATLAB (R2018b) including the image processing toolbox.

Fluorescence microscopy images of scaffold and client condensates were loaded after rolling ball (50 pixel) background subtraction using ImageJ. Condensates were identified in the mCherry or

Alexa-647 fluorescence channel (‘tag’). The positional information was used to quantitatively evaluate co-localization of Mal3:GFP (client molecules; Fig. 5b and Extended Data Fig. 5a,b)

and the tag (Extended Data Fig. 5c). The procedure consisted of converting the mCherry/Alexa-647 image to a binary image that can be used as an image mask (im2bw function with manually

optimized threshold levels ~0.05). The image mask was then used to detect condensates and evaluate their positions, major and minor axes lengths, and the mean intensity, using the

regionprops function for centroid regions. ANALYSIS OF MAL3/TEA2/TIP1:GFP VELOCITIES Single-molecule traces of Mal3/Tea2/Tip1:GFP complexes were recorded at concentrations of 200 nM Mal3, 1 

nM Tea2 and 150 nM Tip1:GFP under MPET conditions. A total number of _N_ = 148 Tip1:GFP traces remained after automated detection in seven kymographs using KymoButler66, and manual exclusion

of obscure traces (crossings, merging or tracks that reach the microtubule end). We calculated a median of 0.23 µm s−1 and a standard error of the mean of 0.06 µm s−1. The velocity of

Mal3/Tea2/Tip:GFP clusters in the presence of PEG35k (Fig. 1h) was assessed after transfer events between microtubule ends and surrounding microtubules. We manually measured _N_ = 47 traces

with a median velocity of 0.12 µm s−1 and a standard error of the mean of 0.018 µm s−1 (Extended Data Fig. 7d). STATISTICS AND REPRODUCIBILITY Data reported are from at least three

independent repeats for each experiment. Detailed information on reproducibility for individual experiments is available from respective figure legends. _P_ values are reported as a result

of the Mann–Whitney test. No statistical method was used to pre-determine sample size. No data were excluded from the analyses; the experiments were not randomized; the investigators were

not blinded to allocation during experiments and outcome assessment. REPORTING SUMMARY Further information on research design is available in the Nature Portfolio Reporting Summary linked to

this article. DATA AVAILABILITY Tomography data shown in Fig. 2 are available from Electron Microscopy Data Bank (EMDB) using the following accession codes: microtubule plus-end in presence

of Tip1, Tea2 and Mal3 (EMD-14110), microtubule minus-end in presence of Tip1, Tea2 and Mal3 (EMD-14111), microtubule plus-end in presence of Mal3 (EMD-1408), microtubule minus-end in

presence of Mal3 (EMD-14109), microtubule plus-end in absence of additional proteins (EMD-14106), microtubule minus-end in absence of additional proteins (EMD-14107), Tip1, Tea2 and Mal3 in

presence of PEG without microtubules or tubulin (EMD-14112) and Tip1, Tea2 and Mal3 in presence of both PEG and microtubules (EMD-14182). Source data are provided with this paper. All other

data are available upon request. CODE AVAILABILITY Python scripts used for splitting of movie frames, reconstruction of even and odd tomographic volumes, training data generation, model

training and denoising are available at https://github.com/NemoAndrea/cryoCARE-hpc04. The simulation code is available at https://github.com/luiree/TipPhase. CHANGE HISTORY * _ 15 MARCH 2023

A Correction to this paper has been published: https://doi.org/10.1038/s41556-023-01124-w _ REFERENCES * Honnappa, S. et al. An EB1-binding motif acts as a microtubule tip localization

signal. _Cell_ 138, 366–376 (2009). Article  CAS  PubMed  Google Scholar  * Maurer, S. P. et al. EB1 accelerates two conformational transitions important for microtubule maturation and

dynamics. _Curr. Biol._ 24, 372–384 (2014). Article  CAS  PubMed  PubMed Central  Google Scholar  * Guesdon, A. et al. EB1 interacts with outwardly curved and straight regions of the

microtubule lattice. _Nat. Cell Biol._ 18, 1102–1108 (2016). Article  CAS  PubMed  Google Scholar  * Reid, T. A. et al. Structural state recognition facilitates tip tracking of EB1 at

growing microtubule ends. _eLife_ 8, e48117 (2019). Article  CAS  PubMed  PubMed Central  Google Scholar  * Browning, H., Hackney, D. D. & Nurse, P. Targeted movement of cell end factors

in fission yeast. _Nat. Cell Biol._ 5, 812–818 (2003). Article  CAS  PubMed  Google Scholar  * Brunner, D. & Nurse, P. CLIP170-like tip1p spatially organizes microtubular dynamics in

fission yeast. _Cell_ 102, 695–704 (2000). Article  CAS  PubMed  Google Scholar  * Dodgson, J. et al. Spatial segregation of polarity factors into distinct cortical clusters is required for

cell polarity control. _Nat. Commun._ 4, 1834 (2013). Article  PubMed  Google Scholar  * Taberner, N. & Dogterom, M. Motor-mediated clustering at microtubule plus ends facilitates

protein transfer to a bio-mimetic cortex. Preprint at _bioRxiv_ https://doi.org/10.1101/736728 (2019). * Bieling, P. et al. Reconstitution of a microtubule plus-end tracking system in vitro.

_Nature_ 450, 1100–1105 (2007). Article  CAS  PubMed  Google Scholar  * Busch, K. E. & Brunner, D. The microtubule plus end-tracking proteins mal3p and tip1p cooperate for cell-end

targeting of interphase microtubules. _Curr. Biol._ 14, 548–559 (2004). Article  CAS  PubMed  Google Scholar  * Browning, H. & Hackney, D. D. The EB1 homolog Mal3 stimulates the ATPase

of the kinesin Tea2 by recruiting it to the microtubule. _J. Biol. Chem._ 280, 12299–12304 (2005). Article  CAS  PubMed  Google Scholar  * Bieling, P. et al. CLIP-170 tracks growing

microtubule ends by dynamically recognizing composite EB1/tubulin-binding sites. _J. Cell Biol._ 183, 1223–1233 (2008). Article  CAS  PubMed  PubMed Central  Google Scholar  * Hayashi, I.,

Plevin, M. J. & Ikura, M. CLIP170 autoinhibition mimics intermolecular interactions with p150Glued or EB1. _Nat. Struct. Mol. Biol._ 14, 980–981 (2007). Article  CAS  PubMed  Google

Scholar  * Stangier, M. M. et al. Structure–function relationship of the Bik1–Bim1 complex. _Structure_ 26, 607–618.e4 (2018). Article  CAS  PubMed  Google Scholar  * Busch, K. E., Hayles,

J., Nurse, P. & Brunner, D. Tea2p kinesin is involved in spatial microtubule organization by transporting Tip1p on microtubules. _Dev. Cell_ 6, 831–843 (2004). Article  CAS  PubMed 

Google Scholar  * Alberti, S., Gladfetter, A. & Mittag, T. Considerations and challenges in studying liquid-liquid phase separation and biomolecular condensates: cell. _Cell_ 176,

419–434 (2019). Article  CAS  PubMed  PubMed Central  Google Scholar  * Wang, J. et al. A molecular grammar governing the driving forces for phase separation of prion-like RNA binding

proteins. _Cell_ 174, 688–699.e16 (2018). Article  CAS  PubMed  PubMed Central  Google Scholar  * Hyman, A. A., Weber, C. A. & Jülicher, F. Liquid–liquid phase separation in biology.

_Annu. Rev. Cell Dev. Biol._ 30, 39–58 (2014). Article  CAS  PubMed  Google Scholar  * Banani, S. F., Lee, H. O., Hyman, A. A. & Rosen, M. K. Biomolecular condensates: organizers of

cellular biochemistry. _Nat. Rev. Mol. Cell Biol._ 18, 285–298 (2017). Article  CAS  PubMed  PubMed Central  Google Scholar  * Brangwynne, C. P. et al. Germline P granules are liquid

droplets that localize by controlled dissolution/condensation. _Science_ 324, 1729–1732 (2009). Article  CAS  PubMed  Google Scholar  * Woodruff, J. B. et al. Regulated assembly of a

supramolecular centrosome scaffold in vitro. _Science_ 348, 808–812 (2015). Article  CAS  PubMed  PubMed Central  Google Scholar  * Hernández-Vega, A. et al. Local nucleation of microtubule

bundles through tubulin concentration into a condensed tau phase. _Cell Rep._ 20, 2304–2312 (2017). Article  PubMed  PubMed Central  Google Scholar  * King, M. R. & Petry, S. Phase

separation of TPX2 enhances and spatially coordinates microtubule nucleation. _Nat. Commun._ 11, 270 (2020). Article  CAS  PubMed  PubMed Central  Google Scholar  * Setru, S. U. et al. A

hydrodynamic instability drives protein droplet formation on microtubules to nucleate branches. _Nat. Phys._ 17, 493–498 (2021). Article  CAS  PubMed  PubMed Central  Google Scholar  *

Siahaan, V. et al. Kinetically distinct phases of tau on microtubules regulate kinesin motors and severing enzymes. _Nat. Cell Biol._ 21, 1086–1092 (2019). Article  CAS  PubMed  Google

Scholar  * Tan, R. et al. Microtubules gate tau condensation to spatially regulate microtubule functions. _Nat. Cell Biol._ 21, 1078–1085 (2019). Article  CAS  PubMed  PubMed Central  Google

Scholar  * Raff, J. W. Phase separation and the centrosome: a fait accompli? _Trends Cell Biol._ 29, 612–622 (2019). Article  PubMed  Google Scholar  * Musacchio, A. On the role of phase

separation in the biogenesis of membraneless compartments. _EMBO J._ 41, e109952 (2022). Article  CAS  PubMed  PubMed Central  Google Scholar  * Plateau, J. A. F. _Statique expérimentale et

théorique des Liquides soumis aux seules Forces moléculaires_ (Gauthier-Villars, 1873). * Rayleigh, L. On the instability of jets. _Proc. Lond. Math. Soc._ S1–10, 4–13 (1878). Article 

Google Scholar  * Volkov, V. A., Huis In’t Veld, P. J., Dogterom, M. & Musacchio, A. Multivalency of NDC80 in the outer kinetochore is essential to track shortening microtubules and

generate forces. _eLife_ 7, e36764 (2018). Article  PubMed  PubMed Central  Google Scholar  * Buchholz, T.-O., Jordan, M., Pigino, G. & Jug, F. Cryo-CARE: content-aware image restoration

for cryo-transmission electron microscopy data. In _2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019)_ 502–506 (IEEE, 2019). * Chrétien, D., Kenney, J. M., Fuller, S.

D. & Wade, R. H. Determination of microtubule polarity by cryo-electron microscopy. _Structure_ 4, 1031–1040 (1996). Article  PubMed  Google Scholar  * Foster, H. E., Ventura Santos, C.

& Carter, A. P. A cryo-ET survey of microtubules and intracellular compartments in mammalian axons. _J. Cell Biol._ 221, e202103154 (2021). Article  PubMed  PubMed Central  Google

Scholar  * McIntosh, J. R. et al. Microtubules grow by the addition of bent guanosine triphosphate tubulin to the tips of curved protofilaments. _J. Cell Biol._ 217, 2691–2708 (2018).

Article  CAS  PubMed  PubMed Central  Google Scholar  * Buey, R. M. et al. Insights into EB1 structure and the role of its C-terminal domain for discriminating microtubule tips from the

lattice. _Mol. Biol. Cell_ 22, 2912–2923 (2011). Article  CAS  PubMed  PubMed Central  Google Scholar  * Honnappa, S., John, C. M., Kostrewa, D., Winkler, F. K. & Steinmetz, M. O.

Structural insights into the EB1–APC interaction. _EMBO J._ 24, 261–269 (2005). Article  CAS  PubMed  Google Scholar  * Slep, K. C. et al. Structural determinants for EB1-mediated

recruitment of APC and spectraplakins to the microtubule plus end. _J. Cell Biol._ 168, 587–598 (2005). Article  CAS  PubMed  PubMed Central  Google Scholar  * Leduc, C. et al. Molecular

crowding creates traffic jams of kinesin motors on microtubules. _Proc. Natl Acad. Sci. USA_ 109, 6100–6105 (2012). Article  CAS  PubMed  PubMed Central  Google Scholar  * Parmeggiani, A.,

Franosch, T. & Frey, E. Totally asymmetric simple exclusion process with Langmuir kinetics. _Phys. Rev. E_ 70, 046101 (2004). Article  CAS  Google Scholar  * Reese, L., Melbinger, A.

& Frey, E. Crowding of molecular motors determines microtubule depolymerization. _Biophys. J._ 101, 2190–2200 (2011). Article  CAS  PubMed  PubMed Central  Google Scholar  * Melbinger,

A., Reese, L. & Frey, E. Microtubule length regulation by molecular motors. _Phys. Rev. Lett._ 108, 258104 (2012). Article  PubMed  Google Scholar  * Reese, L., Melbinger, A. & Frey,

E. Molecular mechanisms for microtubule length regulation by kinesin-8 and XMAP215 proteins. _Interface Focus_ 4, 20140031 (2014). Article  PubMed  PubMed Central  Google Scholar  *

Lipowsky, R., Klumpp, S. & Nieuwenhuizen, T. M. Random walks of cytoskeletal motors in open and closed compartments. _Phys. Rev. Lett._ 87, 108101 (2001). Article  CAS  PubMed  Google

Scholar  * Parmeggiani, A., Franosch, T. & Frey, E. Phase coexistence in driven one-dimensional transport. _Phys. Rev. Lett._ 90, 086601 (2003). Article  CAS  PubMed  Google Scholar  *

Chen, Y., Wang, P. & Slep, K. C. Mapping multivalency in the CLIP-170-EB1 microtubule plus-end complex. _J. Biol. Chem._ 294, 918–931 (2019). Article  CAS  PubMed  Google Scholar  *

Steinmetz, M. O. & Akhmanova, A. Capturing protein tails by CAP-Gly domains. _Trends Biochem. Sci._ 33, 535–545 (2008). Article  CAS  PubMed  Google Scholar  * Bunzarova, N. Z., Pesheva,

N. C. & Brankov, J. G. One-dimensional discrete aggregation-fragmentation model. _Phys. Rev. E_ 100, 022145 (2019). Article  CAS  PubMed  Google Scholar  * Behrens, R. & Nurse, P.

Roles of fission yeast tea1p in the localization of polarity factors and in organizing the microtubular cytoskeleton. _J. Cell Biol._ 157, 783–793 (2002). Article  CAS  PubMed  PubMed

Central  Google Scholar  * Feierbach, B., Verde, F. & Chang, F. Regulation of a formin complex by the microtubule plus end protein tea1p. _J. Cell Biol._ 165, 697–707 (2004). Article 

CAS  PubMed  PubMed Central  Google Scholar  * Meadows, J. C. et al. Opposing kinesin complexes queue at plus tips to ensure microtubule catastrophe at cell ends. _EMBO Rep._ 19, e46196

(2018). Article  PubMed  PubMed Central  Google Scholar  * Snaith, H. A. & Sawin, K. E. Fission yeast mod5p regulates polarized growth through anchoring of tea1p at cell tips. _Nature_

423, 647–651 (2003). Article  CAS  PubMed  Google Scholar  * Snaith, H. A., Samejima, I. & Sawin, K. E. Multistep and multimode cortical anchoring of tea1p at cell tips in fission yeast.

_EMBO J._ 24, 3690–3699 (2005). Article  CAS  PubMed  PubMed Central  Google Scholar  * Matsuo, Y. et al. An unconventional interaction between Dis1/TOG and Mal3/EB1 promotes the fidelity

of chromosome segregation. _J. Cell Sci._ https://doi.org/10.1242/jcs.197533 (2016). Article  PubMed  PubMed Central  Google Scholar  * von Loeffelholz, O. et al. Nucleotide- and

Mal3-dependent changes in fission yeast microtubules suggest a structural plasticity view of dynamics. _Nat. Commun._ 8, 2110 (2017). Article  Google Scholar  * Meier, S. M. et al.

Multivalency ensures persistence of a +TIP body at specialized microtubule ends. _Nat Cell Biol_. https://doi.org/10.1038/s41556-022-01035-2 (2022). * Miesch, J., Wimbish, R. T., Velluz,

M.-C. & Aumeier, C. Phase separation of +TIP-networks regulates microtubule dynamics. Preprint at _bioRxiv_ https://doi.org/10.1101/2021.09.13.459419 (2022). * Song, X., et al. Phase

separation of EB1 guides microtubule plus-end dynamics. _Nat. Cell. Biol_. https://doi.org/10.1038/s41556-022-01033-4 (2022). * Jones, D. T. & Cozzetto, D. DISOPRED3: precise disordered

region predictions with annotated protein-binding activity. _Bioinformatics_ 31, 857–863 (2015). Article  CAS  PubMed  Google Scholar  * Mastronarde, D. N. Automated electron microscope

tomography using robust prediction of specimen movements. _J. Struct. Biol._ 152, 36–51 (2005). Article  PubMed  Google Scholar  * Zheng, S. Q. et al. MotionCor2: anisotropic correction of

beam-induced motion for improved cryo-electron microscopy. _Nat. Methods_ 14, 331–332 (2017). Article  CAS  PubMed  PubMed Central  Google Scholar  * Kremer, J. R., Mastronarde, D. N. &

McIntosh, J. R. Computer visualization of three-dimensional image data using IMOD. _J. Struct. Biol._ 116, 71–76 (1996). Article  CAS  PubMed  Google Scholar  * Chen, M. et al. Convolutional

neural networks for automated annotation of cellular cryo-electron tomograms. _Nat. Methods_ 14, 983–985 (2017). Article  CAS  PubMed  PubMed Central  Google Scholar  * Pettersen, E. F. et

al. UCSF Chimera—a visualization system for exploratory research and analysis. _J. Comput. Chem._ 25, 1605–1612 (2004). Article  CAS  PubMed  Google Scholar  * Gillespie, D. T. A general

method for numerically simulating the stochastic time evolution of coupled chemical reactions. _J. Comput. Phys._ 22, 403–434 (1976). Article  CAS  Google Scholar  * Jakobs, M. A.,

Dimitracopoulos, A. & Franze, K. KymoButler, a deep learning software for automated kymograph analysis. _eLife_ 8, e42288 (2019). Article  PubMed  PubMed Central  Google Scholar 

Download references ACKNOWLEDGEMENTS We are grateful to all group members of the Dogterom as well as Akhmanova (Utrecht University) labs for many discussions on microtubule end tracking

proteins during ERC Synergy meetings. Initial droplet assays were performed during the Physiology Course 2017 at the Marine Biological Laboratory in Woods Hole. This work was supported by

the following grants awarded to M.D.: FOM programme number 110 from the Netherlands Organisation for Scientific Research (L.R.), European Research Council Synergy grant 609822 (V.A.V.) and

Sinergia grant 160728 from the Swiss National Science Foundation (E.O.v.d.S. and R.M.). AUTHOR INFORMATION Author notes * Vladimir A. Volkov Present address: School of Biological and

Behavioural Sciences, Queen Mary University of London, London, UK * Matthew R. King Present address: Department of Biomedical Engineering, Washington University in St. Louis, St Louis, MO,

USA * These authors contributed equally: Renu Maan, Louis Reese, and Vladimir A. Volkov. AUTHORS AND AFFILIATIONS * Department of Bionanoscience, Kavli Institute of Nanoscience, Delft

University of Technology, Delft, the Netherlands Renu Maan, Louis Reese, Vladimir A. Volkov, Eli O. van der Sluis, Nemo Andrea, Wiel H. Evers, Arjen J. Jakobi & Marileen Dogterom *

Physiology Course 2017, Marine Biological Laboratory, Woods Hole, MA, USA Louis Reese, Vladimir A. Volkov, Matthew R. King & Marileen Dogterom Authors * Renu Maan View author

publications You can also search for this author inPubMed Google Scholar * Louis Reese View author publications You can also search for this author inPubMed Google Scholar * Vladimir A.

Volkov View author publications You can also search for this author inPubMed Google Scholar * Matthew R. King View author publications You can also search for this author inPubMed Google

Scholar * Eli O. van der Sluis View author publications You can also search for this author inPubMed Google Scholar * Nemo Andrea View author publications You can also search for this author

inPubMed Google Scholar * Wiel H. Evers View author publications You can also search for this author inPubMed Google Scholar * Arjen J. Jakobi View author publications You can also search

for this author inPubMed Google Scholar * Marileen Dogterom View author publications You can also search for this author inPubMed Google Scholar CONTRIBUTIONS All authors planned

experiments, analysed data and discussed results. E.O.v.d.S. purified the proteins. R.M. performed TIRF experiments including Tea2 titration experiments. R.M. and L.R. performed droplet

assays, following the initial observation by M.R.K. L.R. performed guest-host assays and theoretical modelling and analysed data of Tea2 titration experiments. V.A.V. performed electron

microscopy and 3D reconstruction with help of A.J.J. W.H.E. coated grids with gold and contributed tilt-series acquisition scripts. N.A. contributed cryo-CARE automation scripts. V.A.V.,

R.M., L.R. and M.D. wrote the paper. CORRESPONDING AUTHOR Correspondence to Marileen Dogterom. ETHICS DECLARATIONS COMPETING INTERESTS The authors declare no competing interests. PEER REVIEW

PEER REVIEW INFORMATION _Nature Cell Biology_ thanks Carsten Janke and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are

available. ADDITIONAL INFORMATION PUBLISHER’S NOTE Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. EXTENDED DATA

EXTENDED DATA FIG. 1 (a) Protein regions of Mal3, Tip1 and Tea2 together with disorder prediction (DISOPRED, blue, and IUPred2A, orange). In particular, the serine-rich part of Tip1 and

tails of Tea2 are predicted differently in DISOPRED3 and IUPred2A. (b) Titration of PEG35k percentage and Mal3 concentration required to achieve droplet formation. (c) Mal3:mCherry

interaction with fluorescently labelled GMPCPP seeds in the absence of PEG (top row) or in the presence of 5% PEG-35k. Experiments were repeated twice, representative images from one repeat

are shown. Scale bars: 5 µm. EXTENDED DATA FIG. 2 Each panel shows a slice from one of the tomograms in Fig. 2 b,e processed in the following way: (left) cryoCARE-denoised (see Materials and

Methods for details), (center) back-projected volume generated using IMOD without further processing, (right) the same volume processed using nonlinear anisotropic diffusion algorithm in

IMOD (k = 0.5, 50 iterations). Unprocessed tomograms are available from EMDB using the accession numbers provided for each condition. Experiments were repeated three times, representative

images from one repeat are shown. EXTENDED DATA FIG. 3 (a) Summed tomogram slices containing microtubules (top) and Fourier-filtered subsets containing microtubule moiré pattern (bottom).

Two examples are shown: a 14-protofilament microtubule with the minus end pointing up, and a 14protofilament microtubule with the plus-end pointing up. a – comet at the minus-end, b –

Mal3/Tip1/Tea2 oligomers bound to microtubule lattice, c – soluble tubulin oligomers, d – gold particles with erased densities. (b) Examples of plus- and minus-ends of microtubules grown in

presence of Mal3, Tip1 and Tea2. Two examples of ends with unclear polarity are shown at the bottom. (c) Examples of plus- and minus-ends of microtubules grown in presence of Mal3. (d)

Examples of plus- and minus-ends of microtubules grown in the absence of additional proteins. Scale bars: 100 nm. Experiments were repeated three times, representative images from one repeat

are shown. EXTENDED DATA FIG. 4 (a) Tip1:GFP intensity on the microtubule lattice as well as on the microtubule tip depends non-linearly on the concentration of Tea2. (b) The intensity of

Mal3-Alexa647 is less effected by the Tea2 concentration. Each data point in panels a and b was obtained by averaging tip and lattice intensities at the corresponding concentration, as shown

in c and d, respectively. The number of observed intensity profiles obtained from one experiment per condition is indicated in the legend of panel c (the same number for Tip1:GFP and

Mal3-647 and for ‘tip’ and ‘lattice’, but different at different Tea2 concentrations as indicated). Data are presented as mean ± S.D. (c,d) Intensity profiles of Tip1:GFP (c) and Mal3-647

(d) in the Tea2 titration experiments. Intensity profiles were extracted from TIRF microscopy images of dynamic microtubules. Source data EXTENDED DATA FIG. 5 (a) Droplets of Tea2-647

(scaffold) were allowed to recruit Mal3:GFP constructs (2 nM, client) in the presence of PEG-6k. The graph shows distribution of Mal3:GFP construct intensity in the scaffold droplets (number

of droplets n = 358, 4178, 5220, 6023, 5317, 6065; from left to right). (b) Droplets of unlabelled Tip1 (scaffold) tagged with FL-Mal3:mCherry (2 nM, tag) were allowed to recruit Mal3:GFP

constructs (2 nM, client) in presence of PEG-6k. The graph shows distribution of Mal3:GFP construct intensity in the scaffold droplets (number of droplets n = 1645, 1400, 1248, 831, 908,

1267; from left to right). (c) Intensity of Mal3:mCherry tag in experiments presented in Fig. 4c (Mal3:GFP constructs recruited to unlabelled Mal3 droplets, number of droplets n = 445, 436,

432, 347, 369, 452), and panels a and b. Data was collected on three different days, one experiment per condition. Source data EXTENDED DATA FIG. 6 Mal3 constructs lacking the disordered

region IDR1 neither show motor transport at the lattice nor Tip:GFP accumulation at the plus end. Scale bars: 5 µm (horizontal) and 60 s (vertical). EXTENDED DATA FIG. 7 (a and b) MPET

reconstituted in the presence of 1 mM GTPγS (a) or 1 mM GTP (b). Scale bars: 5 µm (horizontal) and 60 s (vertical). (c) Speed of motor clusters on the microtubule lattice, estimated from the

kymographys in the presence of GTP and GTPγS. Each data point represents the average speed on one microtubule estimated using the imageJ plugin orientationJ. The number of microtubules

analyzed for each condition was 25, pooled from two independent experiments. (d) A comparison between the velocities of Mal3/Tea2/Tip1:GFP complexes (200 nM Mal3, 1 nM Tea2, 150 nM Tip1; n =

 148 traces from 7 different microtubules in one experiment) and Mal3/Tea2/Tip1:GFP droplets (n = 47 droplet traces from one experiment) in the presence of crowding agents (cf. Figure 1h)

shows an approximate 50% reduction in plus-end directed velocity (see Materials and Methods for details). Source data SUPPLEMENTARY INFORMATION REPORTING SUMMARY PEER REVIEW FILE

SUPPLEMENTARY VIDEO 1 Droplet fusion of full length Mal3. Mal3 forms droplets in the presence of crowding agents. Fusion events of micron-sized droplets were observed at 24 µM Mal3 and 10%

PEG-6k using DIC microscopy. SUPPLEMENTARY VIDEO 2 Droplet transfer between two microtubules. Under crowding conditions, Mal3, Tea2 and Tip1 formed droplets at the microtubule tip. These

droplets spread on the GDP microtubule lattice and fuse at the plus-end. Scale bar, 5 µm. SUPPLEMENTARY VIDEO 3 MPET and fixed seed in the presence of crowding agent. In the presence of 5%

PEG-35k, Mal3 was observed interacting with the seed to support MPET on GMPCPP lattice. Biotinylated seed was made to attach the glass surface using Biotin-neutravidin binding. TIP1:GFP

transported by Tea2 motors can be seen accumulating at the plus-end and form a droplet that grew over time. Scale bar, 5 µm. SUPPLEMENTARY VIDEO 4 MPET and moving seed in the presence of

crowding agent. Deposition of Tip:GFP was observed from the plus-end of the non-attached seed. Scale bar, 5 µm. SUPPLEMENTARY VIDEO 5 3D rendering of denoised and segmented cryo-ET volumes.

Plus-end of a microtubule grown in presence of Mal3, Tip1 and Tea2. SUPPLEMENTARY VIDEO 6 3D rendering of denoised and segmented cryo-ET volumes. Minus-end of a microtubule grown in presence

of Mal3, Tip1 and Tea2. SUPPLEMENTARY VIDEO 7 3D rendering of denoised and segmented cryo-ET volumes. Plus-end of a microtubule grown in presence of Mal3 alone. SUPPLEMENTARY VIDEO 8 3D

rendering of denoised and segmented cryo-ET volumes. Plus-end of a microtubule grown in presence of tubulin, without any additional proteins. SUPPLEMENTARY TABLES 1 AND 2 Supplementary Table

1. Protofilament number and polarity of microtubule ends. Supplementary Table 2. Parameters for the theoretical model. SOURCE DATA SOURCE DATA FIG. 2 Statistical source data for Fig. 2.

SOURCE DATA FIG. 3 Statistical source data for Fig. 3. SOURCE DATA FIG. 4 Statistical source data for Fig. 4. SOURCE DATA FIG. 5 Statistical source data for Fig. 5. SOURCE DATA FIG. 6

Statistical source data for Fig. 6 SOURCE DATA EXTENDED DATA FIG. 4 Statistical source data for Extended Data Fig. 4. SOURCE DATA EXTENDED DATA FIG. 5 Statistical source data for Extended

Data Fig. 5. SOURCE DATA EXTENDED DATA FIG. 7 Statistical source data for Extended Data Fig. 7. RIGHTS AND PERMISSIONS OPEN ACCESS This article is licensed under a Creative Commons

Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original

author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the

article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use

is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit

http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Maan, R., Reese, L., Volkov, V.A. _et al._ Multivalent interactions facilitate

motor-dependent protein accumulation at growing microtubule plus-ends. _Nat Cell Biol_ 25, 68–78 (2023). https://doi.org/10.1038/s41556-022-01037-0 Download citation * Received: 15 October

2021 * Accepted: 25 October 2022 * Published: 19 December 2022 * Issue Date: January 2023 * DOI: https://doi.org/10.1038/s41556-022-01037-0 SHARE THIS ARTICLE Anyone you share the following

link with will be able to read this content: Get shareable link Sorry, a shareable link is not currently available for this article. Copy to clipboard Provided by the Springer Nature

SharedIt content-sharing initiative