Genetically encoded live cell sensor for tyrosinated microtubules

Microtubule cytoskeleton exists in various biochemical forms in different cells due to tubulin post-translational modification (PTMs). These PTMs are known to affect microtubule stability, dynamics and interaction with MAPs and motors in a specific manner, widely known as tubulin code hypothesis. At present there exist no tool that can specifically mark tubulin PTMs in live cells, thus severely limiting our understanding of tubulin PTMs. Using yeast display library, we identified a binder against terminal tyrosine of alpha tubulin, a unique PTM site. Extensive characterization validates the robustness and non-perturbing nature of our binder as tyrosination sensor, a live cell tubulin nanobody specific towards tyrosinated or unmodified microtubules. Using which, in real time we followed nocodazole, colchicine and vincristine induced depolymerization events of unmodified microtubules, and found each distinctly perturb microtubule polymer. Together, our work describes the tyrosination sensor and potential applications to study microtubule and PTM processes in living cells.


Introduction
The cytoskeleton tubular polymer, microtubules performs diverse cellular functions, including but not limited to intracellular cargo transport, chromosome segregation and motility. These cellular processes involving microtubules are mediated by interactions with a cohort of molecular motors and microtubule associated proteins (MAPs). A key regulatory process that governs microtubule interaction with its cognate proteins is the diversity of tubulin genes and post-translational modifications (PTMs) (Janke, 2014). Most of the PTMs are reversible and defects in these PTM enzymes leads to abnormal levels of microtubule modifications, manifested in different disease pathologies (Magiera et al., 2018a) causing neurodegeneration (Magiera et al., 2018b) and cardiomyopathies (Chen et al., 2018;Robison et al., 2016).
Among the PTMs, tyrosination and detyrosination cycle of alpha tubulin carboxy-terminal site was the first PTM reported from rat brain extracts (Barra et al., 1973) and later in metazoans, ciliates and flagellates (Janke, 2014). The genetically encoded terminal tyrosine residue can be enzymatically removed by Vasohibin-SVBP complex, a detyrosinase identified recently (Aillaud et al., 2017;Nieuwenhuis et al., 2017). The tubulin tyrosine ligase (TTL), was the first tubulin PTM enzyme discovered, which reverses the detyrosination modification by adding tyrosine back to the terminal site of alpha tubulin (Barra et al., 1973). Over the years several additional tubulin modifications and respective enzymes have been identified across species; acetylation (L' Hernault and Rosenbaum, 1985), glutamylation (Eddé et al., 1990) and glycylation (Redeker et al., 1994). These tubulin modifications, except acetylation, occur at the carboxy-termini tails (CTT) of either alpha and/or beta tubulin gene products. The PTMs can also be combinatorial, overlapping with the diverse tubulin gene products creating diverse biochemical forms of microtubules across cell types (Janke, 2014), which makes tubulin PTM studies a challenging prospect to probe. Recent advances in protein engineering and expression have allowed creating homogenous microtubules with a particular PTM (Sirajuddin et al., 2014;Ti et al., 2018;Valenstein and Roll-Mecak, 2016;Vemu et al., 2014). These studies have highlighted that each PTM uniquely modulates different molecular motors (Barisic et al., 2015;McKenney et al., 2016;Sirajuddin et al., 2014), MAPs (Bonnet et al., 2001) and severing enzymes (Valenstein and Roll-Mecak, 2016), providing first insights into the regulatory roles of tubulin diversity.
A shortcoming of in vitro reconstitution experiments using homogenous modified microtubules is that it may not reflect the in vivo scenario, since microtubules inside cells can possess multiple PTMs at the same time. For example, the stable microtubules have been frequently associated with detyrosinated and acetylated microtubules (Bulinski et al., 1988). Similarly, the glutamylation and glycylation can occur at multiple sites of same tubulin CTTs and have been reported to co-exist in axonemal microtubules (Wloga et al., 2017). A typical cellular or in vivo study involving tubulin PTM involves genetic (Barisic et al., 2015;Magiera et al., 2018b), ectopic expression (van Dijk et al., 2007;Souphron et al., 2019) and chemical perturbations (Aillaud et al., 2017;Nieuwenhuis et al., 2017) of PTM enzymes, followed by antibody staining that are specific towards the respective PTM epitope (van Dijk et al., 2007;Gadadhar et al., 2017;Janke, 2014). Although the antibodies have illuminated the tubulin modifications, it severely limits our understanding of the spatial-temporal component of tubulin PTMs. Therefore, a cellular sensor which can detect and track tubulin modifications in real time will aid studying tubulin PTM dynamics and their function in vivo.
In general, the most common methods to label microtubules in living cells either involve fluorescent tagged alpha tubulin (Gierke et al., 2010;Kamath et al., 2010;Rusan et al., 2001), MAPs (Bulinski et al., 2001) or SiR-tubulin, a taxol derivative (Lukinavičius et al., 2014). The fluorescent protein tagged alpha tubulin can impede incorporation into microtubule polymer, this could be due to improper folding or incompatible tubulin isotypes. The fluorescent markers using MAPs and SiR-tubulin are known to affect microtubule assembly and/or dynamics. Nanobody and single chain antibody approaches have been successfully employed to study actin cytoskeleton and other PTMs (Helma et al., 2015), but without much success against microtubules (Traenkle and Rothbauer, 2017). So far two studies have attempted in this direction; nanobodies against microtubules, which was used to reconstruct super-resolution structures of microtubules (Mikhaylova et al., 2015). Another study has reported the identification of single chain antibody (anti-tubulin scFv) against tyrosinated form of microtubules (Cassimeris et al., 2013). However, the nanobody could not be employed in live cells (Mikhaylova et al., 2015) and no further study of anti-tubulin scFv application has been reported till date. Altogether there is a severe dearth of tools that can mark generic microtubules and/or tubulin PTMs in live cells.
To overcome this, we screened yeast display library (Gera et al., 2012) against CTT of alpha tubulin and identified a binder. Our study with the binder shows specificity against the tyrosinated state of microtubules, which does not interfere with the cellular or microtubule-based functions. The tyrosination sensor reported here, therefore becomes the first thoroughly characterized tubulin nanobody, that can be employed to follow unmodified (tyrosinated) microtubules in living cells.

Strategy for screening binders against tyrosinated microtubule
Several studies have successfully employed carboxy-termini peptides of tubulins as epitopes to identify antibodies specific for a particular tubulin PTM (Bré et al., 1996;Gadadhar et al., 2017;Gundersen et al., 1984). Keeping this in mind, we synthesized the carboxy-terminus of TUBA1A (amino acids 440-451) with a biotin at the amino-terminus . To obtain a binder protein specific for biotin-TUBA1A 440-451 (termed as Hs_TUBA1A), we employed a combinatorial yeast display library of SSO7d mutants screen as described previously (Gera et al., 2012). In order to select binders that are specific for the tyrosinated form, we also performed a negative selection of SSO7D library against biotin-
To understand the interaction of TUBA1A 440-451 peptide (epitope) with A1aY1 binder better, we performed NMR experiments to gain three-dimensional structural information (Methods) ( Table 1). The highest ranked ensemble structure of binder with epitope show the AlaY1 binding site overlap with the diversified regions of SSO7D protein ( Figure 1C and Supplement Figure 2A, B, C, D & E). The key interacting residues from TUBA1A 440-451 peptide includes; Y451, E449, E447 and E445, the most common glutamylation site (Eddé et al., 1990) (Figure 1C and D). The terminal tyrosine (Y451, n th residue) is latched with the aid of L32 and Y29 of A1aY1 binder, additionally the side chain hydroxy group of tyrosine makes a hydrogen bond with main chain amino group of G30 residue ( Figure 1C and Supplement Figure 2). The remaining glutamic acid residues of TUBA1A 440-451 peptide alternatively make electrostatic contacts; E449 (n-2 nd ), E447 (n-4 th ) and E445 (n-6 th ) with K12, H34 and K16 of the A1aY1 binder respectively ( Figure 1C).
Guided by the structure, we then compared ubiquitous sequences of alpha tubulin carboxytermini from human, drosophila, worm, plant and fission yeast ( Figure 1D). The terminal tyrosine (n th residue) and the alternating glutamic acids (n-2 nd , -4 th , -6 th residues) are invariant across different species ( Figure 1D). To further probe the CTTs interaction with binder, we tested Dm_Tub84B and Ce_TB1A of D. melanogaster and C. elegans alpha tubulin CTT peptides respectively, against our A1aY1 binder (Methods). Titration assays show that the Dm_Tub84B and Ce_TB1A CTTs interact with 1.0μM and 4.0μM affinities similar to the Hs_TUBA1A 440-451 CTT peptide ( Figure 1B). In summary, the A1aY1 binder recognizes alpha tubulin CTTs from different organisms and the terminal tyrosine is an important element for A1aY1 binder interaction.

A1aY1 binder labels specifically tyrosinated microtubules in cells
So far, our biochemical and structural experiments with A1aY1 binder have been limited to CTT peptide. In order to check whether A1aY1 binder can recognize microtubules in cells we generated a series of fluorescent protein fusion constructs and transiently expressed them in U2OS cells (Methods). Among them only TagBFP and TagRFP-T fused A1aY1 showed good colocalization with microtubules (Supplement Figure 3). Compared to the Entacmaea quadricolor fluorescent protein variants (TagBFP and TagRFP-T) the other fusion proteins showed poor signal to noise ratio and did not feature prominent microtubule signal (Supplement Figure 3). Both the TagBFP and TagRFP-T A1aY1 sensor illuminated similar microtubule features with amino-and carboxy-terminal fusions ( Figure 2 and Supplement Figure 3). Therefore, we focused on TagBFP and TagRFP-T fused A1aY1 (hereafter, A1aY1 blue and red sensor respectively) and stably expressed them in U2OS cells ( Figure 2) (Methods). In our stable cell lines, we also observed that upon high expression the microtubules begin to bundle in few cells (Supplement Figure 4). Thus, a medium to lowlevel expression of the blue and red A1aY1 sensor offers its application as a live cell biosensor for labelling microtubules (Figure 2 & Supplement Figure 4). Tyrosinated microtubules are abundant at the interphase stage of fibroblasts and epithelial cells, therefore we inferred that our A1aY1 binder is marking bulk of the microtubules in our experiments. Our biochemical results show that A1aY1 binder is specific for tyrosinated peptide, we then checked the specificity of A1aY1 sensor towards detyrosinated and glutamylated microtubules. The U2OS stable cell lines of red A1aY1 sensor were cotransfected with detyrosinase (Aillaud et al., 2017;Nieuwenhuis et al., 2017)  Co-transfection of detyrosinase in the stable lines with A1aY1 red sensor leads to complete loss of microtubule signal by A1aY1 sensor ( Figure 3A & B). In the case of TTLL5 and its catalytically inactive version, we see a decrease and retention of microtubule signal by A1aY1 sensor respectively ( Figure 3A & B). Conversely, TTLL7, an enzyme which is more specific towards beta tubulin CTT (van Dijk et al., 2007), shows no loss of colocalization by A1aY1 sensor towards microtubules ( Figure 3A & B). This indicates that polyglutamylation modification at the alpha-tubulin CTT (E445 residue) will sterically interfere with A1aY1 sensor ( Figure 3C, D & E). Thereby reducing the microtubule binding ability of A1aY1 sensor, which is similar to our observation with biochemistry using mono-glutamylated peptide ( Figure 1B).
In summary, our biochemistry and specificity experiments unequivocally suggest that A1aY1 sensor specifically recognizes tyrosinated microtubules. Therefore, hereafter we refer our A1aY1 binder as tyrosination sensor.

Tyrosination sensor does not alter the cellular function of microtubules
Microtubules are majorly involved in mitotic spindle organization and chromosome segregation; in addition, a hallmark property of microtubules is their ability to undergo dynamic instability. Therefore, the next step in validation of tyrosination sensor as a live cell marker is to check whether they interfere with microtubule function. We first checked the viability and proliferative ability of stable U2OS cell lines expressing blue and red tyrosination sensor (Methods). Trypan blue assays, propidium iodide and DAPI based FACS for both blue and red sensor stable lines show that ~95% of the cells are viable and in their proliferative state (Supplement Figure 6A & B). We also checked the ability of stable cell lines to undergo mitosis and chromosome segregation, here we could observe all the mitotic stages and image them using our tyrosination sensor (Supplement Figure 6A & B). This confirms that by stably integrating the sensor genes and constitutive expression of either blue or red tyrosine sensors does not interfere with cell viability and division.
Next, we measured microtubule dynamics in cells expressing tyrosination sensor using fluorescence from the TagRFP-T fused to A1aY1 binder (Methods). Time lapse images show that the microtubule polymerization events can be followed by TagRFP-T-A1aY1 fluorescence signal. Microtubules labelled with TagRFP-T-A1aY1 undergo typical dynamic instability states of growth, catastrophe and rescue ( Figure 4A and Supplement Movie 1), suggesting that tyrosination sensor does not interfere with microtubule dynamics. In order to quantify the growth rates, we then transfected EB3-GFP and imaged microtubules along with EB3 comets (Methods) ( Figure 4B, Supplement Figure 7A, B & C and Supplement Movie 2). We observed 0.35±0.09µm/s (n=450), 0.36±0.1µm/s (n=468) and 0.19±0.06µm/s (n=414) microtubule growth rates in the presence and absence of tyrosine sensor and with 0.5µM SiR-tubulin respectively ( Figure 4C). Our live cell imaging with and without EB3-GFP comets also reveals that the tyrosination sensor binds to microtubules promptly during polymerization and disappears during catastrophe events ( Figure 4A, B and Supplement Movie 1 & 2). Simultaneously, we measured EB3-GFP comets in the presence of SiRtubulin, which shows the significant decrease in growth rates compared to the untreated and cells with tyrosination sensor (Supplement Figure 7). These live cell imaging experiments reveal that the tyrosination sensor can be used to follow microtubule polymerization and depolymerization events, without affecting their growth rates and dynamics (Supplement Another important microtubule function is their role as tracks for motor proteins, which facilitates intracellular cargo transport. In order to check if tyrosination sensor interferes with motor motility, we performed in vitro motility experiments using kinesin-1 (K560-SNAP) or Combinedly these experiments suggest that tyrosination sensor does not affect microtubule properties and their related cellular function, thus validating the suitability of our sensor for live cell experiments.

Live cell imaging and mechanism of drugs that target microtubules
Currently live cell imaging of microtubules is carried out by fluorescent tagged alphatubulin, MAPs and SiR tubulin (fluorescent version of taxol) that stabilize microtubules (Supplement Figure 7). We envision that our tyrosination sensor, which does not affect dynamics ( Figure 4) and specifically labels unmodified i.e., tyrosinated microtubules can be used to study microtubules in live cells. As a proof of concept, we employed drugs such as nocodazole, colchicine and vincristine that are known to target microtubules and are commonly used in cell biology studies to perturb microtubules. Although there are several reports about mode of drugs action, so far there are no reports of live cell imaging and description of depolymerization events in real time. To address this, we took advantage of our U20S cells stably expressing AlaY1-TagRFP-T, and were individually treated with either 10µM nocodazole, 500µM colchicine or 1µM vincristine (Methods). Upon nocodazole addition, we observed that the microtubules begin to shrink from the ends ( Figure 5A In the case of colchicine, we also observe that majority of the microtubules undergo end-on depolymerization events, with frequent severing like activity ( Figure 5C, E & F, Supplement Movie 5). Similarly, we imaged vincristine mediated depolymerization events, vincristine when applied to cells the microtubules become brittle, similar to the filament severing activity ( Figure 5D, E & F, Supplement Movie 6). Quantification of the depolymerization events by each drug shows that nocodazole, colchicine and vincristine have distinct mechanism of depolymerization as attributed by structural and biochemical studies (Gigant et al., 2005;Lee et al., 1980;Ravelli et al., 2004) (Figure 5E & F). While the mode of action for these drugs has been suggested earlier (Jordan and Kamath, 2007), here we are able to capture and follow the depolymerization events in real time. Thus, the A1aY1 sensor presents a great opportunity as a tool to study new microtubule targeting drugs and understand their mechanism in real time.

Live cell super-resolution microscopy with tyrosination sensor:
Cytoskeleton filaments and in general microtubules have been favorite test subjects for developing new methodology towards super-resolution imaging (Demmerle et al., 2015).
Using the existing fluorescent tags, we performed 3color 3D-SIM on cells stably expressing tyrosination red sensor i.e., TagRFP-T-A1aY1 ( Figure 6A) in interphase. Z-stacks were acquired for a total width of 2µm and all planes images were reconstructed, 3D volume rendered using alpha blending (Methods). Correlative analysis of SiR-tubulin versus TagRFP-T-A1aY1 signal in interphase cells shows a good agreement of colocalization (Pearson's coefficient -0.75; Spearman's rank correlation -0.88; Mander's coefficient -0.93) ( Figure   6B), indicating the abundance of tyrosinated microtubules in during interphase cycle. Line scan comparison between 3D-SIM versus widefield filament width shows a gain of ~300 nm in resolution ( Figure 6C). Quantification of microtubules width for ~50 filaments showed a mean FWHM value of 107±0.35 nm, in the region of half the diffraction limit as compared to the conventional resolution limit of ~360 nm in a widefield setting ( Figure 6D). We have also obtained 3D-SIM images for the mitotic stages in U2OS cells to confer the suitability of the probe for live cell super-resolution imaging over long periods of time (Supplement Movie 7).

Discussion
Fluorogenic nanobodies against intracellular structural components including actin cytoskeleton have been successfully employed in unraveling new biology. A notable exception has been the microtubule cytoskeleton (Traenkle and Rothbauer, 2017), in particular the tubulin PTMs where the epitopes are well-characterized and have yielded specific antibodies (Janke, 2014). In this study, we have employed alpha tubulin CTT peptide as epitope and discovered a binder from SSO7D library. The binder was then developed and validated as an intracellular nanobody against microtubules, specifically the tyrosinated form of alpha tubulin called tyrosination sensor. This sensor represents the first robust tubulin nanobody reported in the field, which does not affect microtubule and cellular functions. The imaging experiments with tyrosination sensor shows that single microtubule events can be followed in real time (Supplement Movie 1-6). Our EB3 comet assay also suggests that the tyrosination sensor does not interfere with binding of +TIP proteins, which contains CAP-Gly domains. We further extended the imaging capability of our tyrosination While the terminal tyrosine residue is an indispensable element in recognition by the sensor, our experiments indicates that the glutamic acid residues of alpha tubulin CTT also play important roles in tyrosination sensor binding. Tubulin CTTs contain a series of glutamic acid residues and provided that a terminal tyrosine is available, we predict that our tyrosination sensor will be able to bind to the tubulin/microtubule. This is strongly supported by our biochemical experiments with fly and worm alpha tubulin CTTs, which binds to the tyrosination sensor with similar affinities as that of human. A key element in this interaction is the third alternating glutamic acid residue (n-6 th residue) of alpha tubulin CTT, the most common site for glutamylation modification (Janke, 2014). Specificity experiments show that glutamylation modification at this site abolishes the interaction, suggesting that crosslinked glutamic acid residue will sterically hinder tyrosination sensor binding.  , 2014). Additionally, the CAP-Gly domains bind to the carboxy-termini EEY motif of end-binding (EB) proteins, which is consensus to the alpha tubulin CTT terminus (Honnappa et al., 2006). Structural studies show that the sextette acidic motif (EEGEEY/F) of alpha tubulin CTT or the EEY motif of EBs are important for binding with CAP-Gly domain (Honnappa et al., 2006;Mishima et al., 2007). Vasohibin bound to alpha tubulin CTT complex structure also shows that the last five residues of alpha tubulin CTT (EGEEY) binds to the active site (Liao et al., 2019). In both cases the tyrosine recognition by CAP-Gly and VASH proteins reveals the importance of free mainchain carboxyl group of the terminal tyrosine residue. In contrast, our A1aY1:alpha tubulin CTT complex structure, reveals a unique mode of tyrosine sensing, which involves the interaction of phenyl ring and the hydroxyl group of tyrosine side chain ( Figure 1C & 3D). Additionally, none of the naturally occurring proteins that bind to alpha tubulin CTT extend their recognition towards the glutamylation site (i.e., n-6 th residue). Therefore, we conclude that our tyrosination sensor senses the glutamylation state of alpha tubulin, in addition to the tyrosination state of microtubules.
Microtubules are well-known target for anti-cancer therapeutics, several reports describe drugs that stabilize and destabilize microtubules (Jordan and Kamath, 2007). Most of the drugs that are either used in therapeutics or for research purposes, have detailed account of their activity and binding site, from which the mechanism of action has been proposed.
However, none of the work reported so far in this regard show drug induced microtubule depolymerization events. Here we applied our tyrosination sensor to image nocodazole, colchicine and vincristine induced microtubule depolymerization events in real time ( Figure   5). Nocodazole is known to bind free tubulin dimers thus preventing their addition to microtubule polymer (Lee et al., 1980). Our results show that upon nocodazole treatment, the microtubules are in a constant state of catastrophe without any rescue events, in line with the proposed nocodazole mode of action. Colchicine was originally used to identify the tubulin (Borisy and Taylor, 1967), biochemical and structural investigations point that colchicine can bind to both soluble tubulin and as well as microtubule lattice (Jordan and Kamath, 2007;Ravelli et al., 2004). Cells when treated with colchicine show a combination of severed polymers and depolymerization events from both of the severed ends, underscoring its dual binding mode and twofold action. Vincristine is a potent anti-cancer drug, which binds only to microtubule polymer and destabilizes the lattice (Dhamodharan et al., 1995;Gigant et al., 2005;Jordan and Kamath, 2007;Jordan et al., 1992). In line with this structural finding, we observed more polymer severing and rapid depolymerization of double-ended microtubules.
The three drugs studied here show distinct modes of microtubule depolymerization, uniquely they also have commonalities amongst them. For example, nocodazole and colchicine when binds to the free tubulin traps them in curved state, which is incompatible for incorporation at the growing plus ends (Brouhard and Rice, 2014). Similarly, when colchicine or vincristine when binds to the microtubule, it induces lattice defects by kinking the longitudinal interaction between tubulins (Gigant et al., 2005;Ravelli et al., 2004). These lattice defects are then amplified leading to breaking of polymer, akin to severing like activity. In summary, the tyrosination sensor described here can be applied to study mechanisms pertaining to microtubule destabilizing drugs. In addition, since our tyrosination sensor marks unmodified microtubules without affecting dynamics, we predict that our sensor will be a valuable tool in screening new drugs that target microtubules. The specificity experiments described in Figure 3 outsets application towards studying PTM enzymes. The retention of sensor binding to microtubules upon the expression of catalytically dead TTLL5 suggests, that a similar approach will be advantageous in screening regulatory factors and drugs that affect PTM enzymes, such as vasohibin and TTLLs.
Together, the tyrosination sensor described here presents an opportunity to study microtubules and tubulin PTMs in live cells using fluorescence and super-resolution microscopy methods. Further providing a prospect to expand this methodology to generate sensors against other microtubule PTMs such as detyrosination, acetylation, glutamylation and glycylation.

Materials and Methods:
Detailed materials methods can be found in Supplement Information.    Dissociation constant (kd) of A1aY1 binder against biotinylated alpha tubulin CTT peptides of Human (Hs_TUBA1A; 1.6 ± 0.3µM, Hs_TUBA1A-DY; >60µM*, Hs_TUBA1A-mG; 13.6 ± 4.5µM), Drosophila melanogaster (Dm_Tub84B; 4.1 ± 0.9µM) and C. elegans (Ce_TBA1; 1.0 ± 0.2µM) measured using surface plasmon resonance state binding assay. Experiments were performed in triplicates with at least two different batches of the protein A1aY1 (* represents that the binding affinities can't be uniquely determined with the current fit). The inset shows titration response up to 4µM A1aY1 binder concentration. C. The NMR structure of A1aY1 binder (magenta surface representation) bound to Hs_TUBA1A peptide (blue cartoon with key residues in stick representation). The key interacting residues from A1aY1 binder and alpha tubulin CTT are labeled in white and blue respectively. D. Sequence alignment of alpha tubulin CTTs from Human (Hs), Drosophila melanogaster (Dm), Caenorhabditis elegans (Ce), Arabidopsis thaliana (At), Schizosaccharomyces pombe (Sp) with gene names as indicated. The red asterisk indicates residues involved in A1aY1 binder interaction, the terminal tyrosine and alternating glutamate residues indicated as n series.    A typical diffused cytoplasmic signal will have ~ 50% or less overlap, whereas a complete co-localization will show near 100% overlap. n=25 cells, representing each dot for respective column. C. The NMR structure of A1aY1 (grey surface representation) and alpha tubulin CTT peptide (blue cartoon representation), the carboxy-terminal, tyrosine and glutamic acid residues are indicated. D. and E. Closer view of the terminal tyrosine (Y451) and polyglutamylation site glutamic acid (E445) with key interacting residues from A1aY1 binder as indicated.

Hs_TUBA1A-mG
All the above peptide sequences were ordered from Thermo fisher scientific.

Yeast Surface Display Library Screening
A combinatorial SSO7d yeast display library was obtained as a kind gift from Dr Balaji

M. Rao's lab at Department of Chemical and Biomolecular Engineering, North
Carolina State University, Raleigh, NC 27695, USA and the detailed protocol for screening binders was adopted as described in [(Gera et al., 2011)]. To screen a binder for CTT of human alpha tubulin specific for terminal tyrosine residue, SSO7d library was applied for a stringent negative selection with biotinylated detyrosinated (biotin-Hs_TUBA1A-DY) and mono-glutamylated (biotin-Hs_TUBA1A-mG) peptides to reduce the diversity and non-specific binders from the library, followed by a positive selection with tyrosinated peptide (biotin-Hs_TUBA1A) yielded a population of binders with varying degree of binding affinities.
The library (diversity around 10ˆ8 cells) was propagated in 10-fold excess of its  The washed beads were transferred to a fresh 5ml SDCAA media for growth at 30°C, 250rpm for 48hr. Beads were removed from the grown culture after and cells were grown in large culture volume for stocks preparation (10ˆ9 cells) and use for FACS.

Protein purification
The gene sequences of A1aY1 and A1aY2 binders were cloned in pET28a(+) vector between NdeI and NotI restriction sites with amino-terminus 6xHis-tag followed by a thrombin site and the gene sequence. The protein expression was achieved in Rosetta (DE3) competent cells by inducing the culture at OD600 of 0.5 with 1mM IPTG (Sigma cat. no. I6758) at 25°C for 8-10hr in terrific broth (Yeast extract 24g/L, Tryptone 20g/L, 17mM KH2PO4, 72mM K2HPO4, 4ml/L Glycerol) and 50µg/ml Kanamycin antibiotic. Only A1aY1 could be purified successfully as A1aY2 protein was toxic to the bacterial expression host (DE3 rosetta) and hence was purified with a carboxyterminus GFP tag (mentioned below). Overnight-induced culture of A1aY1 was harvested in 50mM Tris-HCl (pH 7.5), 100mM NaCl, 1mM PMSF with 1 tablet of EDTA free protease inhibitor cocktail (Roche, cat. no. 11836170001) for 1L of the culture. The cells were lysed with the Avestin Emulsiflex C3 homogenizer (ATA Scientific instruments) and protein was purified using 5ml Ni-NTA affinity column after 10-column volume (CV) wash with 50mM Tris-HCl (pH 7.5), 500mM NaCl and 25mM Imidazole (pH 7.5) and elution in 5CV of elution buffer (50mM Tris-HCl, pH 7.5, 100mM NaCl and 350mM Imidazole, pH 7.5). The eluted protein was concentrated up to 5ml using 3KDa millipore amicon filter (Merck UFC900324). Further, size exclusion chromatography was carried out in 50mM Tris-HCl (pH 7.5), 100mM NaCl buffer in Superdex-75, 16/600 column (cat. no. 28989333 GE). The resulting fraction of the pure A1aY1 protein was concentrated (mol. wt. 9.3KDa) and frozen in small aliquots for long term storage in -80°C.
For NMR experiments, 13 C-15 N isotope-labelled A1aY1 protein was purified from the the DE3 rosetta bacterial cells grown in M9 media (Na2HPO4 6g/L, KH2PO4 3g/L, NaCl 0.5g/L, 15NH4Cl 1g/L, 13C labelled glucose 2g/L, divalent cations, Vit-B12, Thiamine and trace elements and 50µg/ml Kanamycin antibiotic) using Ni-NTA affinity chromatography in Tris buffer as mentioned above. Further purification using size exclusion chromatography was performed in 50mM sodium phosphate buffer (pH at room temperature for 4hr. The cleaved protein was purified again with size exclusion chromatography (using S75 16/600 column) in 50mM sodium phosphate buffer (pH 6.5) with 200mM NaCl. The resulting protein corresponds to 7.8KDa which was concentrated and frozen in -80°C in small aliquots for future use in NMR experiments.
A GFP tag was attached to the above A1aY1 construct at carboxy-terminus in pET28a(+) vector, such that it contains an amino-terminus 6x-His, a thrombin site, A1aY1 followed by a GFP sequence separated by a glycine-serine linker. Similarly, binder A1aY2 was cloned with a carboxy-terminus GFP tag to obtain its soluble expression. Both the proteins were purified in 50mM potassium phosphate buffer (pH 6.0) with 100mM potassium chloride and 5mM beta-mercaptoethanol (BME) using Ni-NTA chromatography. Further purification of the protein was carried out using S-200, 16/600 column for size exclusion chromatography. For all the in-vitro experiments with the polymerized HeLa microtubules, these proteins were diluted in 1xBRB80 (80mM PIPES, 1mM MgCl2, 1mM EGTA, pH 6.8) buffer. For the SPR steady-state binding assay these proteins were diluted in 1xHBS-P+ buffer (10mM HEPES, 150mM NaCl and 0.05% v/v surfactant P20, GE healthcare life science cat. no. BR100671).

K560-SNAP and KIF1A-SNAP Purification:
Truncated Rat kif1a (1-393 amino acids) followed by a GCN4 leucine zipper was cloned into a pET-17b vector with a Snap-tag followed by a 10X Histidine-tag at the carboxy-terminus. K560-Snap and Kif1a-LZ-Snap were expressed using rosetta (DE3) bacterial expression system. 10mM Glycine-HCl pH 2.0 and 2.5 respectively followed by 60sec stabilization period between two titrations. The relative responses of binding (RUmax) with each peptide were determined and normalised (maximum response as 100) by subtracting the blank flow channel 1 (as FC2-1, FC3-1 and FC4-1), and was plotted with the corresponding concentrations to fit a curve (one site total fitting on GraphPad Prism6) and determine the value of Kd.

NMR Spectroscopy
Protein samples were prepared as described above. All NMR spectra were acquired at 25ºC on an 800MHz/600MHz Bruker Avance III spectrometers equipped with a 5mm TCI CryoProbe. The sample was loaded in 5mm Shigemi tube. 1 H-15 N heteronuclear single quantum coherence (HSQC) experiments were performed with 2048 × 256 complex data points. All NMR spectra were processed by using NMRPipe [1] and analyzed by using Sparky [2] .
Assignment of the backbone resonances ( 1 H, 13 C, and 15 N) of the proteins were carried out by using Bruker's BEST pulse program (b_HNCO, b_HNCACO, b_HNCACB, and b_CBCACONH) experiment. 1 H and 13 C resonance assignments of side-chain atoms in A1aY1 were obtained by collecting H(CC)CONH (H)CC(CO)NH and HCCH-TOCSY respectively. The resonance list for the above experiments generated by using sparky and submitted to I-PINENMR [3] server for automatic assignment. The I-PINE results are manually cross-checked and corrected. The model structure of A1aY1 was calculated by using CS-ROSETTA [4] . The interface between A1aY1 and nonbiotinylated Hs_TUBA1A peptide was identified using 13 C filtered NOESY experiments on a 1:1 sample of 13 C, 15 N enriched protein, and unlabeled (natural abundance) Peptide.

Cell fixation, immunofluorescence and imaging
HeLa cells plated on glass coverslips were transduced with lentiviruses encoding different constructs of mVash-mSVBP for 24 hours and were fixed according to previously described protocols 5 . Briefly, the cellular proteins were cross-linked using the homo-bifunctional cross-linker, Dithio-bis(succinimidyl propionate; DSP; #22585 Thermo Fisher Scientific) diluted in microtubule-stabilizing buffer (MTSB), followed by fixation with 4% paraformaldehyde for 15 mins. The cells were then permeabilized with 0.5% Triton X-100 in MTSB for 5 mins and were blocked in 5% bovine serum albumin (BSA) prepared in PBS containing 0.1% Triton X-100 (PBST).