Mapping multivalency in the CLIP-170–EB1 microtubule plus-end complex

Cytoplasmic linker protein 170 (CLIP-170) is a microtubule plus-end factor that links vesicles to microtubules and recruits the dynein–dynactin complex to microtubule plus ends. CLIP-170 plus-end localization is end binding 1 (EB1)–dependent. CLIP-170 contains two N-terminal cytoskeleton-associated protein glycine-rich (CAP-Gly) domains flanked by serine-rich regions. The CAP-Gly domains are known EB1-binding domains, and the serine-rich regions have also been implicated in CLIP-170's microtubule plus-end localization mechanism. However, the determinants in these serine-rich regions have not been identified. Here we elucidated multiple EB1-binding modules in the CLIP-170 N-terminal region. Using isothermal titration calorimetry and size-exclusion chromatography, we mapped and biophysically characterized these EB1-binding modules, including the two CAP-Gly domains, a bridging SXIP motif, and a unique array of divergent SXIP-like motifs located N-terminally to the first CAP-Gly domain. We found that, unlike the EB1-binding mode of the CAP-Gly domain in the dynactin-associated protein p150Glued, which dually engages the EB1 C-terminal EEY motif as well as the EB homology domain and sterically occludes SXIP motif binding, the CLIP-170 CAP-Gly domains engage only the EEY motif, enabling the flanking SXIP and SXIP-like motifs to bind the EB homology domain. These multivalent EB1-binding modules provided avidity to the CLIP-170–EB1 interaction, likely clarifying why CLIP-170 preferentially binds EB1 rather than the α-tubulin C-terminal EEY motif. Our finding that CLIP-170 has multiple non-CAP-Gly EB1-binding modules may explain why autoinhibition of CLIP-170 GAP-Gly domains does not fully abrogate its microtubule plus-end localization. This work expands our understanding of EB1-binding motifs and their multivalent networks.

(1). Although microtubules are inherently dynamic, the dynamics are spatially and temporally regulated by a host of microtubule-associated proteins (MAPs). 2 A subset of MAPs preferentially localize to polymerizing microtubule plus ends and are termed plus-end tracking proteins (ϩTIPs) (2,3). ϩTIPs include a diverse set of proteins that have assorted structural domains and functional properties. Many ϩTIPs regulate microtubule dynamics or serve as adaptors that link microtubule plus ends to other factors, including cell polarity factors and kinetochores (4,5).
The microtubule EB family members, including human EB1, EB2, and EB3, are central ϩTIPs, as they directly bind the microtubule plus end and recruit a host of other ϩTIPs, establishing a ϩTIP network (3,6,7). The EB proteins have a common domain architecture that consists of an N-terminal calponin homology (CH) domain, a central linker, and a C-terminal dimerization domain (EBHD) that confers homodimerization as well as heterodimerization for the EB1-EB3 pair (Fig. 1A) (4, 8 -11). The dimerization domain is followed by a conserved C-terminal motif consisting of the residues EEY that is similar to the C-terminal segment of ␣-tubulin (12). The CH domain binds the microtubule lattice between four tubulin heterodimers, where it specifically engages tubulin in a transition state/post-GTP-hydrolysis state just distal of the microtubule's polymerizing GTP cap (7). Both the EB1 dimerization domain and the EEY motif serve to bind and recruit factors to the microtubule plus end (12)(13)(14)(15).
The EBHD forms a coiled-coil structure that transitions to a four-helix bundle at its C-terminal region (Fig. 1B) (4,11). The dimerization domain forms two homologous sites at the coiled coil-four-helix bundle interface that are used to bind SXIP or LXXPTPh motif-containing proteins (Fig. 1B) (13,16,17). Over 50 proteins have been identified that contain SXIP motifs and show EB1-dependent microtubule plus end-tracking activity (3). The EB1 C-terminal EEY motif is capable of binding CAP-Gly domains such as those found in p150 Glued of the dynactin complex as well as CLIP-115 and CLIP-170 ( Fig. 1C) (14,15,18). The p150 Glued CAP-Gly domain dually engages the EB1 EEY motif as well as the EBHD, where its binding site overlaps with the binding site for SXIP and LXXPTPh motifs, suggesting mutually exclusive binding per site on the EB1 dimer (13)(14)(15)(16)(17). Whether CLIP-170 CAP-Gly domains engage the EBHD in a similar manner remains to be determined.
CLIP-170 was the first identified ϩTIP protein and is involved in linking endocytic vesicles to microtubules and recruiting the dynein-dynactin complex to microtubule plus ends by binding the dynactin component p150 Glued (14, 15, 19 -23). The domain architecture of CLIP-170 includes three prime parts: an N-terminal region of ϳ350 residues that encompasses two CAP-Gly domains flanked and separated by serine-rich regions, a central coiled-coiled region that mediates homodimerization and is involved in binding the microtubule pause factor CLIP-associated protein (CLASP), and a C-terminal segment that contains two zinc knuckles and an ETF motif that can bind and autoinhibit the CAP-Gly domains (19, 21, 22, 24 -27). When not autoinhibited, the CLIP-170 CAP-Gly domains can each bind C-terminal EEY motifs, as found in both the ␣-tubulin acidic tail and EB1 (26 -28). In cells and in vitro, the CLIP-170 N-terminal CAP-Gly domain-containing region (residues 1-350) is sufficient to track polymerizing microtubule plus ends in an EB1-dependent manner (20,28,29). In vitro, in the absence of EB1, CLIP-170 mostly binds along the lengths of microtubules (28 -30). Why CLIP-170 preferentially engages EB1 versus the tubulin C-terminal tail when EB1 is present remains unclear. One hypothesis is that CLIP-170 has a higher affinity/greater avidity for EB1 than for tubulin.
The CLIP-170 central coiled coil has a predicted hinge region that enables the C-terminal zinc knuckles and ETF motif to engage and autoinhibit the CAP-Gly domains from binding EB1 (21,27). CLIP-170 autoinhibition is regulated by phosphorylation of Ser-311 (31, 32). Phosphorylated CLIP-170 preferen- A, domain architecture of EB1 and CLIP-170. EB1 contains an N-terminal CH domain that binds the microtubule lattice and a C-terminal dimerization domain termed the EBHD. EB1 also contains a C-terminal EEY motif that binds CAP-Gly domains. CLIP-170 contains two N-terminal CAP-Gly domains (denoted A and B) and a central coiled coil involved in dimerization. Two C-terminal zinc knuckle domains are involved in binding and inhibiting CAP-Gly binding to EEY motifs and a C-terminal ETF motif that also binds CAP-Gly domains in a mode similar to EEY motifs. B, structure of EBHD-MACF2 SXIP complex (PDB code 3GJO) (13). The EBHD is shown in cartoon format in green and yellow, and the MACF2 SXIP peptide is shown in stick format in orange. C, composite structural model of the EB1-p150 Glued CAP-Gly domain complex (PDB codes 2HKQ and 2HL3 superimposed, dually aligned over their respective CAP-Gly domains) (15). The EBHD is presented as in B, with the EEY motif depicted in stick format. The two p150 Glued CAP-Gly domains (shown in cyan with a transparent surface depicted on the molecule on the left) engage the EB1 EEY motif as well as the coiled-coil portion of the EBHD. The CAP-Gly residues that engage the EBHD are colored red. The p150 Glued CAP-Gly domain engages the same region on the EBHD as the SXIP peptide (shown in B), suggesting that p150 Glued CAP-Gly domain and SXIP motif binding to the EBHD are mutually exclusive. D, sequence comparison of the EBHD binding region of the p150 Glued CAP-Gly domain versus CLIP-170 CAP-Gly A and B. p150 Glued EBHD-binding residues are shown in red. The position-equivalent residues in CLIP-170 CAP-Gly domains that are not conserved are highlighted in yellow.

Diverse CLIP-170 -EB1 interaction modes
tially adopts a folded/autoinhibited conformation. Interestingly, when phosphorylated at Ser-311, CLIP-170 still accumulates at the microtubule plus end (albeit with diminished comet length relative to nonphosphorylated CLIP-170), suggesting that it is still bound to EB1 (31,32). How CLIP-170 accumulates at microtubule plus ends when its CAP-Gly domains are autoinhibited is unclear but suggests that CLIP-170 may contain additional EB1-binding determinants.
In this study, we performed structure/function analysis of the N-terminal CLIP-170 region that displays EB1-dependent microtubule plus-end tracking activity. We determined the affinity of each CAP-Gly domain for EB1 and identified two distinct regions outside of the CAP-Gly domains that bind EB1. One binding element is an SXIP motif located between the two CAP-Gly domains that has a weak binding affinity for the EBHD. Another EB1 binding site includes multiple SXIP-like motifs, is located N-terminally to the first CAP-Gly domain, and includes ϳ20 conserved residues that bind the EBHD with high affinity. These interactions, when analyzed in the presence of the CAP-Gly domains, strongly enhanced EB1 binding in vitro. Thus, CLIP-170 uses a diverse, multivalent set of elements to engage EB1 and form a stable complex. The multivalent nature of the CLIP-170 -EB1 interaction likely clarifies the ability of CLIP-170 to bind EB1 and localize to microtubule plus ends even when its CAP-Gly domains are autoinhibited. The high avidity of the complex is anticipated to preferentially select EB1 binding over ␣-tubulin binding.

The CLIP-170 CAP-Gly domains engage the EB1 C-terminal tail and do not appreciably affect SXIP motif binding to the EBHD
Previous work suggests that CLIP-170 may be able to interact with EB1 using determinants outside of its CAP-Gly domains. This is based on two sets of data. First, CLIP-170 phosphorylated at residue Ser-311 (a readout for the autoinhibited state in which the CAP-Gly domains are competitively bound) still tracks microtubule plus ends (31,32). Second, although the p150 Glued CAP-Gly domain has a higher EB1 binding affinity than a CLIP-170 construct containing both of its two CAP-Gly domains and the bridging linker, a larger CLIP-170 construct interacts with EB1 more robustly than does p150 Glued (26,33). This caused us to ask the following question: are there determinants outside of the CLIP-170 CAP-Gly domains that engage EB1? Potential non-CAP-Gly EB1 binding modes include SXIP or LXXPTPh motif binding to the EBHD (13,16,17). However, the structure of the p150 Glued CAP-Gly domain bound to EB1 indicated that this CAP-Gly domain engaged both the EB1 EEY motif as well as the EBHD, which sterically occluded SXIP and LXXPTPh motif binding (Fig. 1, B and C) (14,15). We questioned whether CLIP-170 CAP-Gly domains could also engage the EBHD or whether they only engaged the EEY motif, leaving the EBHD free to bind SXIP or LXXPTPh motifs that might be found flanking the CLIP-170 CAP-Gly domains. Comparative analysis of the nine p150 Glued CAP-Gly residues that contact the EBHD (Fig. 1C, residues colored in red) revealed that only four are conserved in CLIP-170 CAP-Gly A and B, with two of these being conserved glycine residues involved in the domain's fold (Fig. 1D). This suggested that, unlike p150 Glued , CLIP-170 CAP-Gly domains might bind EB1 in a mode that left the EBHD free to bind SXIP or LXXPTPh motifs.
We next used isothermal titration calorimetry (ITC) to determine whether CLIP-170 CAP-Gly domains, like those of p150 Glued , bind EB1 through dual engagement of the C-terminal EEY motif and the EBHD. We generated two EB1 constructs. The first included both the EBHD domain and EEY C-terminal tail (EB1 268 , residues 191-268), and the second contained only the EBHD (EB1 260 , residues 191-260), and we measured the binding affinity between these constructs and CLIP-170 CAP-Gly A (residues 57-130) and B (residues 211-279) (Fig. 2, A-D). CAP-Gly A and B each showed an exothermic binding isotherm with EB1 268 . The binding curves were best fit to a one-site model (two CAP-Gly domains binding independently to one EB1 dimer; potential binding models were evaluated based on 2 analyses) yielding K D values equal to 1.5 M and 4.0 M for CAP-Gly A and B, respectively (Fig. 2, A and C). In contrast, neither CAP-Gly domain revealed detectable binding to EB1 260 (Fig. 2, B and D), suggesting that CLIP-170 CAP-Gly domains do not robustly engage the EBHD.
We next tested whether CLIP-170 CAP-Gly domain binding to the EBHD affected the ability of an SXIP motif to bind the EBHD (Fig. 3, A-F). We synthesized two peptides corresponding to the SXIP motifs found in SLAIN2 (QASAIPSPGK-FRSPAY) and CLASP2 (KRSKIPRSQGSSRETY) (13,34). We measured the binding affinity of these peptides for EB1 268 alone and when EB1 268 was prebound to CLIP-170 CAP-Gly A or CAP-Gly B. The SLAIN2 peptide and the CLASP2 peptide bound free EB1 268 with a K D equal to 17.9 M and 5.3 M, respectively (both fit to a one-site model; two peptides binding independently to one EB1 dimer) (Fig. 3, A and D). Prebinding CAP-Gly A to EB1 268 did not appreciably affect the binding affinity of the SLAIN2 peptide (K D ϭ 18.9 M); however, it did have a moderate effect on the CLASP2 peptide's affinity, decreasing the apparent affinity to a K D equal to 14.4 M (Fig. 3, B and E). Similar results were obtained when EB1 268 was prebound to CAP-Gly B; the SLAIN2 peptide bound with a K D of 22.5 M and the CLASP2 peptide bound with a K D of 14.7 M (Fig. 3, C and F). This indicates that CLIP-170 CAP-Gly domain binding to EB1 does not occlude the SXIP motif binding site. It is also of note that we did not observe CAP-Gly-dependent cooperative effects on trans SXIP motif binding to the EBHD; i.e. CAP-Gly domain binding to the EB1 EEY motif did not potentiate the binding of an SXIP motif-containing peptide to the EBHD (Fig. 3, A-F).

An SXIP motif between the CLIP-170 CAP-Gly domains has weak EB1-binding activity
Given that the CLIP-170 CAP-Gly domains do not sterically occlude SXIP motif binding, we inquired how a previously identified SXIP motif located between CAP Gly A and B (residues 165-168) affected CLIP-170 binding to EB1 (33,35). The motif in human CLIP-170, SNIP, is conserved across many animals from frog to human, with conservation extending to flanking regions (consensus: PS(N/G)IPXKXS) (Fig. 4A). We first tested the binding of a CLIP-170 construct (CLIP-170 A-SNIP) that embodied both the CAP-Gly domain A and the C-terminally

Diverse CLIP-170 -EB1 interaction modes
flanking SNIP motif (residues 57-210) to EB1 260 . Although CAP-Gly A did not bind EB1 260 (Fig. 2C), CLIP-170 A-SNIP bound EB1 260 exothermically and fit best to a one-site model with a K D equal to 12.7 M (Fig. 4B). When we examined CLIP-170 A-SNIP binding to EB1 268 , the exothermic binding curve was best fit to a two-site model (two CLIP-170 A-SNIP constructs binding to one EB1 dimer, with each EB1 dimer containing two CAP-Gly domain binding sites (the EEY tail) and two SXIP motif binding sites (on the EBHD)), yielding K D1 equal to 19.7 M and K D2 equal to 1.5 M (Fig. 4, C and D). When we mutated the SNIP motif to SNNN (mutations shown previously to ablate EB1 binding (13)), the CLIP-170 A-SNNN construct still bound EB1 268 as expected but fit best to a one-site model with a K D equal to 1.2 M (Fig. 4E). This binding is on par with the value expected for just CAP-Gly A binding to the EB1 268 C-terminal EEY motif. ITC analysis of CLIP-170 A-SNIP binding to EB1 260 yielded a K D equal to 15.6 M, whereas CLIP-170 A-SNNN did not exhibit binding to EB1 260 (Fig. 4E). Similar

Diverse CLIP-170 -EB1 interaction modes
EB1 268 and EB1 260 binding patterns were observed for a CLIP-170 construct that embodied the SNIP motif and the flanking CAP-Gly B domain (CLIP-170 SNIP-B, residues 156 -279) and its mutant SNNN version (CLIP-170 SNNN-B) (Fig. 4F). We next synthesized a peptide corresponding to the human CLIP-170 SNIP-containing sequence (residues 162-176; TPSNIPQKPSQPAAKY) and analyzed the binding of this peptide to EB1 260 . The binding isotherm was best fit to a one-site binding model (two peptides binding independently to one EB1 dimer) with a K D equal to 19.5 M (Fig. 4G). This binding affinity is similar to that determined for the SLAIN2 SAIP motif peptide (17.9 M, Fig. 3A) but slightly weaker than the CLASP2 SKIP motif peptide (5.3 M, Fig. 3D).
These findings indicate that the EB1 dimer can engage CLIP-170 using multiple binding sites, two C-terminal EEY motifs, used to bind two CAP-Gly domains (either CAP-Gly A or B), as well the EBHD's two SXIP motif binding sites, used to bind the CLIP-170 SNIP motif (for which there would be two in the functional CLIP-170 homodimer), and that these binding sites function independently.

The CLIP-170 CAP-Gly domains and SNIP motif are not the only EB1-binding modules in CLIP-170
We next analyzed whether the CLIP-170 SNIP motif was sufficient to bind and shift the EB1 elution profile using size-

Diverse CLIP-170 -EB1 interaction modes
exclusion chromatography (SEC). To assay the interaction, we constructed and purified a CLIP-170 fragment containing both CAP-Gly A and B and the intervening SNIP motif-containing linker (CAP-Gly A-B, residues 57-279). Co-injection of an equimolar CAP-Gly A-B:EB1 268 mixture yielded a peak that eluted at 28.5 min that was distinct and eluted earlier than either component run individually (Fig. 5A). This peak shift suggested that CAP-Gly A-B and EB1 268 formed a robust complex over gel filtration. We next analyzed CAP-Gly A-B binding to EB1 260 but did not detect a shift in the elution profile (Fig.  5B). Because the EB1 260 construct cannot engage the CAP-Gly domains, this result indicates that the CLIP-170 SNIP motif alone is not sufficient to form a stable complex with the EBHD over gel filtration. This may suggest that the SNIP motif may not be accessible to EB1 when flanked by both CAP-Gly domains. This finding is aligned with previous work demonstrating that the SNIP motif did not contribute significantly to EB1 binding in the context of a CLIP-170 construct containing both CAP-Gly domains (33). Surprisingly, when we analyzed the ability of a larger CLIP-170 construct embodying residues 1-350 to bind EB1 260 , we found that it robustly interacted with the EBHD, and the proteins co-eluted at 25 min, earlier than either component analyzed individually (Fig. 5C). In agreement with our finding that the CLIP-170 SNIP motif was not sufficient for the interaction with EB1 260 , when we tested a CLIP-170 1-350 construct in which the central SNIP motif was mutated to SNNN, we still detected a robust interaction with EB1 260 over gel filtration (Fig. 5C). This suggested that a region either N-terminal to CAP-Gly A (within residues 1-56) or C-terminal to CAP-Gly B (within residues 280 -350) was binding to the EB1 260 EBHD.
We next analyzed the binding affinity between EB1 260 and CLIP-170 constructs containing both CAP-Gly domains using ITC. The CLIP-170 CAP-Gly A-B construct (residues 57-279)

Diverse CLIP-170 -EB1 interaction modes
bound to EB1 260 with a K D equal to 30.4 M, presumably mediated exclusively by the central SNIP motif (Fig. 5D). This binding affinity is slightly weaker than that detected for CAP-Gly A-SNIP (12.7 M, Fig. 4B) or SNIP-CAP-Gly B (15.6 M, Fig.  4F) and may reflect binding hindrance when both CAP-Gly domains are present and flank the SNIP motif. In contrast, the CLIP-170 1-350 construct bound to EB1 260 robustly with a K D value equal to 0.5 M (Fig. 5E). In agreement with our size exclusion chromatography data, this suggested that a region outside of the CLIP-170 A-B (57-279) construct but present in the CLIP-170 1-350 construct was binding EB1 with high affinity. We purified a CLIP-170 construct embodying CAP-Gly B and residues C-terminal to it (residues 205-350) but failed to detect binding to EB1 260 (Fig. 5F). This suggested that determinants N-terminal to CLIP-170 CAP-Gly A were capable of binding the EB1 260 EBHD.

The CLIP-170 N-terminal region contains a unique EB1-binding motif
To map specific EB1 binding determinants in the CLIP-170 N-terminal region, we generated a number of CLIP-170 N-terminal deletion constructs that spanned through the end of CAP-Gly A and tested the ability of these constructs to bind EB1 260 (Fig. 6, A-E). As EB1 260 cannot bind CAP-Gly A, we reasoned that any EBHD binding activity would be due to the region N-terminal to CAP-Gly A. Inclusion of CAP-Gly domain A in the constructs facilitated ease of detection during purification (construct size and determination of protein concentration because of residues in the CAP-Gly domain that absorb at 280 nm). A CLIP-170 construct spanning residues 1-130 bound EB1 260 with a K D equal to 1.3 M (Fig. 6A). Deleting the first 12 residues of CLIP-170 (construct, residues 13-130) did not affect EB1 260 binding (K D ϭ 1.4 M, Fig. 6B); however, deleting the first 18 residues ablated detectable binding (Fig. 6E). This suggested that the CLIP-170 13-18 region contained key EBHD binding residues.
To gain insight into whether residues in the CLIP-170 N-terminal region were conserved, we used Clustal 2.1 to align CLIP-170 homologs from a diverse set of animals, including mammals, bird, reptiles, fish, and insects (tarantula and bee) (Fig.  6D) (36). Across these species, the first 20 amino acids of CLIP-170 are well conserved. The alignment reveals a trimeric array of SXIP-like motifs: SMLKP, SGLKAP, and TKILKP. The first motif takes the form (S/T)X(I/L)XP, whereas the second and third motifs take the form (S/T)X(I/L)XP, where X denotes any amino acid (excluding those with acidic side chains) with at least one of the Xs in the motif being a basic amino acid (lysine or arginine). To examine the potential role of these motifs in EBHD binding, we introduced mutations into these SXIP-like motifs and tested how these mutations affected binding to EB1 260 . First, we introduced two point mutations in the CLIP-170 13-130 construct (K17N and P18L) and did not detect binding to EB1 260 (Fig. 6E). Similarly, introducing a triple point mutation in the CLIP-170 13-130 construct (T13N, K14N, and  I15N) resulted in no detectable binding to EB1 260 . Interestingly, when we engineered the T13N, K14N, and I15N point mutations in a CLIP-170 6 -130 construct, we found EB1 260 binding activity (K D ϭ 5.6 M), indicative that additional determinants in the 6 -12 region were contributing to EBHD binding. The introduction of three point mutations into the 6 -12 region (S7N, L9N, and K10N) in concert with the T13N, K14N, and I15N mutations ablated the ability of the CLIP-170 6 -130 construct to bind EB1 260 . However, introducing these six point mutations into the larger CLIP-170 1-130 construct yielded EB1 260 binding with a K D value equal to 10.8 M. Collectively, these data indicate that determinants spanning CLIP-170 residues 1-18 are involved in binding the EBHD. To test whether this region was sufficient for EBHD binding, we synthesized a peptide corresponding to CLIP-170 residues 1-24. This CLIP-170 N-terminal peptide bound to EB1 260 with a K D equal to 1.4 M, on par with the binding activity of the CLIP-170 1-130 construct (Fig. 6, A, C, and E).
We next confirmed the interaction between the CLIP-170 N-terminal segment and the EB1 EBHD using a SEC multiangle static light scattering (SEC-MALS) assay. In this assay, components and potential complexes are separated on a SEC column based on molecular size and radius of gyration. The eluate is then run in-line through a MALS system that calculates an average experimental mass based on the component's or the complex's multi-angle light scattering profile. The calculated experimental mass provides information regarding oligomeric states, potential binding stoichiometries, and complex dissociation during the SEC run and aids the characterization of elution profiles in which the macromolecule's shape is elongated. CLIP-170 1-130 was able to robustly bind and co-elute with EB1 268 after incubation and co-injection of a 1:1 protein ratio (Fig. 7A). The elution profile of this peak was nearly Gaussian in form and was shifted relative to the peaks of the individual components. The experimentally determined mass of this peak was 41.3 kDa, a value that is between that of a 2:2 (47.5 kDa) and a 2:1 (33.2 kDa) EB1 268 -CLIP-170 1-130 complex, indicative that some dissociation may have occurred during the run. Presumably, formation of this complex relied on interactions between CAP-Gly A and the EB1 EEY motif as well as SXIP-like interactions with the EBHD. We next analyzed whether CLIP-170 1-130 could bind and shift EB1 260 over gel filtration. The use of EB1 260 would prevent complex formation via CAP-Gly A binding to the EEY motif. In agreement with our ITC results, CLIP-170 1-130 was able to bind and shift EB1 260 over gel filtration, indicative that the N-terminal SXIP-like array was sufficient for EBHD binding (Fig. 7B). The elution profile of the peak was not Gaussian but had a trailing tail, indicative of complex dissociation. The experimentally determined average mass across the peak was 27 kDa, which is less than the calculated mass of a 2:1 EB1 260 -CLIP-170 1-130 complex (31.2 kDa). Although the main peak gave an experimental mass of ϳ31 kDa, the trailing tail yielded a lower mass, bringing the overall average down. Thus, although the N-terminal SXIPlike motifs of CLIP-170 are sufficient for EB1 binding, binding is enhanced through avidity, mediated by CAP-Gly domain binding to the EB1 EEY motif. We next used SEC to test whether deleting or mutating some of the CLIP-170 N-terminal SXIPlike motifs affected EB1 binding. Using excess of the CLIP-170 construct, we found that both CLIP-170 13-130 (lacking the first two SXIP-like motifs) and CLIP-170 6 -130 (T13N, K14N,  I15N) (lacking the first SXIP-like motif and containing muta-

Diverse CLIP-170 -EB1 interaction modes
tions in the third motif) were able bind and shift EB1 to an earlier elution time. In accord with our ITC binding data (Fig.  6E), we found that CLIP-170 13-130 was best able to bind and shift EB1 compared with CLIP-170 6 -130 (T13N, K14N, I15N) (Fig. 7C). These results confirm our finding that CLIP-170 contains an array of EB1-binding SXIP-like motifs located N-terminally to CAP-Gly A.

Discussion
We delineated multiple, distinct EB1-binding determinants in the CLIP-170 N-terminal region. These include the two known CAP-Gly domains (A and B) that engage C-terminal EEY motifs as found in EB1 and EBHD-binding determinants that include an SXIP motif in the linker bridging CAP-Gly domains A and B and a unique array of SXIP-like motifs located N-terminally to CAP-Gly A. These N-terminal motifs bear some resemblance to SXIP motifs and LXXPTPh motifs that have been structurally characterized to date (Fig. 8, A and B).
Here we describe SXIP and LXXPTPh EBHD binding modes and compare these binding determinants with the CLIP-170 SXIP-like array to gain insight into how this region of CLIP-170 might bind the EBHD. SXIP and LXXPTPh motifs both engage the EBHD at a hydrophobic site at the domain's coiled coil and four-helix bundle junction (Fig. 8A) (13,16,17). Although each motif takes a slightly different trajectory on the domain, three residues structurally overlap at this site and involve the serinelysine-isoleucine residues (SXI) of the SXIP motif, as observed in the EB1-MACF2 complex, and the threonine-arginine-leucine residues of the LXXPTPh motif involving the first leucine of the motif and the two residues proceeding the leucine (-2 Ϫ 1L), as observed in the Bim1-Kar9 complex. In the CLIP-170 SXIP-like array, we conjecture that the three corresponding structurally homologous residues in each motif are methionine-serine-methionine (MSM, first motif), serine-glycineleucine (SGL, second motif), and threonine-lysine-isoleucine (TKI, third motif). We anticipate that these three residues bind the EBHD in a mode similar to the SXI of the SXIP motif and the -2 Ϫ 1L of the LXXPTPh motif. Across all motifs, there is a conserved proline at a position equivalent to the first proline in the LXXPTPh motif. Structural studies of Dis1 bound to the Mal3 EBHD reveals that Dis1 uses an LXXPTPh-like motif where the second proline is replaced with a glutamine, demonstrating residue variability at this site, but followed by a hydrophobic residue (phenylalanine). In line with Dis1, none of the CLIP-170 motifs have a second proline; however, the first and second motifs in the array do have a final hydrophobic residue (leucine and isoleucine, respectively). The motifs in the CLIP-170 array overlap, suggesting that EBHD binding along the array may occur in different frames, as binding to one site likely precludes the binding of either another EBHD dimer or the second site on the same EBHD to a second motif in the array. Across vertebrates, the SXIP-like motif array is well conserved (Fig. 6D). When compared with more distant animals or with other CLIP family proteins such as CLIP-115, the residues of the second SXIP-like motif are best conserved (Fig. 8C). We note that fly species (e.g. Drosophila melanogaster and Drosophila busckii) that lack the first and third SXIP-like motifs have  . When co-incubated and injected, the proteins co-eluted at 28.5 min with a tail to the peak, indicative of dissociation. The peak yielded an experimentally determined mass of 27 kDa (less than the calculated mass (31.2 kDa) of a 2:1 EB1 260 -CLIP-170 1-130 complex), indicating that the interaction with EB1 is not as robust when the CAP-Gly domain A interaction with the EB1 EEY motif is removed. C, gel filtration analysis of CLIP-170 N-terminal fragment constructs binding to EB1 260 . Excess amounts of CLIP-170 13-130 (which only contains one of the three SXIP-like motifs) was co-incubated with EB1 260 and injected onto a size-exclusion column. All EB1 260 was shifted to a higher mass species, indicating that the third SXIP-like motif is sufficient for detectable EB1 binding in this assay. A CLIP-170 6 -130 construct containing the second SXIP-like motif with mutations in the third SXIP-like motif (T13N, K14N, I15N) also produced a stable shift of EB1 260 to a higher mass species, indicating that the second motif can also bind the EBHD, although not as robustly as the third repeat (compare relative shifts).

Diverse CLIP-170 -EB1 interaction modes
other SXIP motifs that flank this region, potentially suggesting convergent use of diverse EB1-binding motifs in the sequence located N-terminally to CAP-Gly A.
Delineation of multiple distinct EB1-binding determinants in CLIP-170 suggests that CLIP-170 can bind EB1 in a variety of modes that range from the use of a single determinant (no avidity) up to four determinants (high avidity), at which point it saturates the known, available EB1 dimer's binding sites. Alternatively (or additionally), CLIP-170's multiple EB1-binding determinants can be used to engage more than one EB1 dimer. Our data indicate that CLIP-170 CAP-Gly domains only engage the EB1 EEY motif and do not dramatically occlude the EBHD, leaving the domain accessible to SXIP, SXIP-like, or LXXPTPh motif binding. This interaction mode is distinct from p150 Glued CAP-Gly domain binding, which dually engages the EEY motif and the EBHD and competes for SXIP binding (14,15). Our work aligns with a recent analysis of the yeast Bik1-Bim1 (CLIP-170 and EB1 homologs, respectively) interaction, demonstrating that Bik1 CAP-Gly domain binding does not occlude SXIP motif binding to Bim1 (37). It was thought previously that CLIP-170 CAP-Gly domain binding to EB1 occluded SXIP motif binding and was thus mutually exclusive. This is based on the titration of a SXIP motif that displaced CLIP-170 from growing microtubule plus ends in vitro in the presence of EB1 as well as the reverse: titrated CLIP-170 displaced an SXIP peptide from growing microtubule plus ends (33). Although one interpretation is steric occlusion, another possibility is that the titrated trans SXIP peptide displaced the CLIP-170 SXIP and/or SXIP-like motifs from the EBHD, which lowered the CLIP-170 -EB1 binding affinity, leading to dissociation. Likewise, the CLIP-170 SXIP or SXIP-like motifs, when titrated, could displace a trans SXIP peptide.  (13). The KAR9 LXXPTPh motif is shown in stick format, colored magenta (from PDB code 5N74) (17). Structures were superimposed using the coordinates of the respective EBHD from each structure. Only the EBHD from the KAR9 complex is depicted, shown in surface format, colored light gray. The sequence of the two peptides is shown below each structure. The peptides structurally align over the region boxed in black (residues SKI (SXI) from the MACF2 SXIP motif and residues TRL from the KAR9 LXXPTPh motif, which include the initial leucine in the motif and the two residues N-terminal to it. B, alignment of the MACF2 SXIP motif as well as the LXXPTPh motifs from structures determined to date: KAR9 and Dis1 (PDB codes 5N74 and 5M9E) (16,17) compared with the three SXIP-like motifs in the CLIP-170 N-terminal region. C, comparison of the human CLIP-170, CLIP-115, and fly CLIP-190 N-terminal regions. The second SXIP-like motif of CLIP-170 is conserved in CLIP-115 and in the fly species D. melanogaster and D. busckii (highlighted in purple). Although the fly homologs lack the other SXIP-like motifs found in the CLIP-170 N-terminal region, they do contain other SXIP motifs (highlighted in green), suggesting a potential conserved multivalent EB1 interaction mechanism. D, model depicting the interaction modes between CLIP-170 and EB1. CLIP-170 CAP-Gly domains bind the EB1 C-terminal EEY motif. The CLIP-170 N-terminal region and the SXIP motif located between CAP-Gly A and B bind the EBHD. The ability of the CLIP-170 CAP-Gly domains to bind the EB1 EEY motif is auto-inhibited by the C-terminal zinc knuckles. When the CAP-Gly domains are auto-inhibited, the EBHD binding motifs may serve to confer EB1 binding, albeit weaker than when they can engage EB1 in concert with the CAP-Gly domains.

Diverse CLIP-170 -EB1 interaction modes
As CLIP-170 CAP-Gly domain binding to the EB1 EEY motif leaves the EBHD free, the EBHD might be used to engage CLIP-170 SXIP or SXIP-like motifs or to form a higher-order complex involving CLIP-170, EB1, and a third factor. Examples of higher-order complexes that may implement this interaction mode include CLASP2 and SLAIN2. CLASP2 has an EB1-binding SXIP motif as well a C-terminal domain that binds the CLIP-170 coiled coil (13,24). SLAIN2 has multiple SXIP motifs as well as a C-terminal WRDGCY sequence that can bind the EEY-binding site in either CLIP-170 CAP-Gly A or B (34). For both the CLASP2 and SLAIN2 examples, each protein in the respective three-component complex forms pairwise interactions with the other two proteins that likely promotes complex stability, limiting the exclusion and displacement of one factor from the complex. Thus, multivalency in the EB1-CLIP-170 complex can afford avidity and/or alternative binding modes that accommodate higher-order complex formation. It is likely that a high-avidity, multivalent interaction with the EB1 dimer outcompetes CLIP-170 CAP-Gly binding to the ␣-tubulin C-terminal EEY motif (Fig. 8D). Preferential binding to EB1 would localize CLIP-170 to the microtubule plus end rather than along the length of the microtubule lattice by ␣-tubulin binding. This may explain why, in the absence of EB1, CLIP-170 primarily localizes along the lengths of microtubules in vitro but, upon addition of EB1, preferentially localizes to growing microtubule plus ends (28 -30).
Interestingly, previous work suggested that CLIP-170 in its autoinhibited state still localized to microtubule plus ends (21,32). How CLIP-170, with its CAP-Gly domains bound in cis to the zinc knuckle motifs in an autoinhibited state could still engage EB1 to localize to microtubule plus ends was a mystery. In these experiments, the cellular readout for CLIP-170 was its phosphorylation state, as phosphorylation of Ser-311 promoted autoinhibition. It is possible that not all CAP-Gly domains were bound to zinc knuckles, enabling some CAP-Gly domains in the functional dimer to bind EB1 and therefore localize to microtubule plus ends. Alternatively, autoinhibited CLIP-170 may use the SXIP motif and/or the SXIP-like motifs we delineate here to bind the EBHD and localize to microtubule plus ends independent of its inhibited state (Fig. 8D). Future cellular studies in which distinct CLIP-170 EB1-binding modes are systematically ablated will help address this question. The unique EB1-binding SXIP-like motifs we delineate in the CLIP-170 N-terminal region expand our understanding of EB1-interacting determinants and how diverse networks of MAPs in distinct complexes and stoichiometries can be recruited to microtubule plus ends.

Cloning and purification of EB1 and CLIP-170 constructs
EB1 C-terminal (Ct) fragments and CLIP-170 N-terminal (Nt) fragments were inserted into plasmid pET28a (Millipore Sigma, Burlington, MA) using PCR and engineered NdeI and BamHI restriction sites. EB1 Ct clones included EB1 268 (residues 191-268) and EB1 260 (residues 191-260). An array of CLIP-170 Nt constructs was generated, including constructs with point mutations generated using the QuikChange method (Agilent Technologies, Santa Clara, CA). Plasmids containing EB1 Ct and CLIP-170 Nt fragments were transformed into BL21 DE3 PLysS Escherichia coli and grown in 6 liters of Luria-Bertani medium at 37°C under kanamycin selection. At an optical density of 0.8 (600 nm), protein expression was induced using 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside for overnight induction at 18°C. Cells were harvested and lysed by sonication at 4°C. After centrifugation at 23,000 ϫ g for 45 min, supernatant was loaded onto a nickel-nitrilotriacetic acid (Qiagen, Hilden, Germany) chromatography column and eluted with a linear gradient between 10 mM and 300 mM imidazole (25 mM Tris (pH 8.0), 300 mM NaCl, and 0.1% ␤-mercaptoethanol). EB1 Ct constructs were dialyzed against 25 mM Tris (pH 8.0) and 0.1% ␤-mercaptoethanol with either ␣-thrombin (0.1 mg, Hematologic Technologies, Essex Junction, VT) included in the dialysis tube or added after dialysis to cleave off the N-terminal His 6 tag. EB1 Ct constructs were then filtered over benzamadine-Sepharose (GE Healthcare, Waukesha, WI). The cleaved EB1 Ct protein was loaded onto a Q-Sepharose Fast Flow column (GE Healthcare, Waukesha, WI) and eluted over a linear 0 -1 M NaCl gradient. After dialyzing into 25 mM MES (pH 6.5) and 0.1% ␤-mercaptoethanol, the CLIP-170 Nt fragments were digested by ␣-thrombin (0.1 mg, Hematologic Technologies), filtered over benzamadine-Sepharose (GE Healthcare), loaded onto an SP-Sepharose Fast Flow column (GE Healthcare), and eluted over a linear 0 -1 M NaCl gradient. Proteins were exchanged into 25 mM Tris pH 8.0, 150 mM NaCl, and 0.1% ␤-mercaptoethanol, concentrated in a 3000 molecular weight cutoff Millipore concentrator (Millipore Sigma), and flash-frozen in liquid nitrogen for storage.

Peptide synthesis
Three different SXIP peptides and a CLIP-170 Nt peptide were synthesized at the University of North Carolina Microprotein Sequencing and Peptide Synthesis Facility. The SXIP peptides included SAIP (QASAIPSPGKFRSPAY, human SLAIN2, residues 450 -464), SKIP (KRSKIPRSQGSSRETY, human CLASP2, residues 726 -739, in which the native cysteine at position 736 was replaced with a serine), and SNIP (TPSNIPQKPSQPAAKY, human CLIP-170, residues 162-176). A CLIP-170 Nt peptide was synthesized (Nle-S-Nle-LKPSGLKAPTKILKPGSTALKY, human CLIP-170, residues 1-24) in which norleucine (Nle) was substituted for the two native methionines at positions 1 and 3. A tyrosine residue was added at the end of each peptide to determine the concentration using absorption at 280 nm. In the case of the human CLASP2 SKIP peptide, a penultimate threonine was added before the tyrosine.

ITC
ITC experiments were carried out at 16°C in 25 mM HEPES (pH 6.8), 50 mM NaCl, and 0.1% ␤-mercaptoethanol on a MicroCal AutoITC200 (GE Healthcare). Peptides were exchanged into ITC buffer using G-25 Sephadex quick spin columns (Roche Applied Science, Penzberg, Germany). 2-l volumes of 0.5 to 1 mM protein or peptide were automatically injected into a well containing 360 l of a 30 -50 M EB1 Ct construct. The resulting binding isotherms were fit to a one-site Diverse CLIP-170 -EB1 interaction modes or two-site binding model using the Origin 7.0 software package, and the respective fits were evaluated using 2 analysis in Origin 7.0 (OriginLab, Northampton, MA). Reported K D values are the average of the K D values determined from two or three independent experiments. The standard deviation reported was calculated using the STDEVP function in Microsoft Excel (Microsoft Corp., Redmond, WA).

SEC and SEC-MALS measurements
Before injection onto the SEC column, proteins were incubated at room temperature for 20 min. 100 l of the respective protein solution (each protein component at 30 -100 M) was injected onto a Superdex 200 column (GE Healthcare) in 50 mM NaCl, 25 mM HEPES (pH 6.8), 0.1% ␤-mercaptoethanol, and 0.2 g/liter sodium azide with a flow rate of 0.5 ml/min. The sample was then passed through a Wyatt Optilab rES refractometer (Wyatt Technology Corp., Santa Barbara, CA). When conducted in-line with MALS, the sample was also passed through a DAWN HELEOS II light-scattering instrument (Wyatt Technology Corp., Santa Barbara, CA, USA). The light-scattering and refractive index data were used to calculate the weightaveraged molar mass using the Astra V software program (Wyatt Technology Corp.). SEC and SEC-MALS data shown are representative of independent experiments conducted in duplicate.