Systematic identification of microtubule lumen proteins reveals a taxane-sensitive luminal resident JPT2 regulating MEC17 accessibility

Significance

Microtubules are traditionally regarded as polymers regulated through their outer surface and dynamic ends, while the internal lumen has remained largely unexplored. In this study, we develop a spatially resolved proximity-labeling strategy that enables the identification of regulatory factors specifically residing within the singlet microtubule lumen. We show that the conserved protein Jupiter microtubule-associated homolog 2 (JPT2) enters the lumen via its C-terminal domain and uncover that this compartment exhibits selective accessibility and potential regulatory capacity. These findings demonstrate that the microtubule lumen is not an inert void but a structurally constrained and protein-enriched regulatory space. Our work provides a conceptual and experimental foundation for dissecting the internal compartmentalization of the cytoskeleton.

Abstract

Microtubules have been extensively studied for their interactions at the outer surface and ends, yet the regulatory mechanisms within the microtubule lumen, particularly in singlet microtubules, remain poorly understood. Here, we developed a proximity-labeling strategy to distinguish intraluminal from external proteins, enabling the systematic identification of key regulators residing within the singlet microtubule lumen. Among these, we focused on Jupiter microtubule-associated homolog 2 (JPT2), a conserved microtubule-binding protein. JPT2 binds directly to microtubules and localizes to the lumen in a C-terminal-dependent manner. Within this compartment, it modulates the distribution of the α-tubulin acetyltransferase MEC17 and contributes to luminal homeostasis. Notably, the luminal accessibility of JPT2 is highly sensitive to microtubule structural states and is markedly reduced by Paclitaxel treatment, suggesting that drug-induced conformational changes or steric hindrance can restrict luminal entry. Together, our findings identify JPT2 as a key organizer of the singlet microtubule lumen and provide a framework for the spatial and functional analysis of microtubule inner proteins. Beyond exogenous perturbations, our observations suggest that luminal protein distribution may also be modulated by steric competition or selective entry among intraluminal components, indicating a more intricate regulatory landscape within the microtubule interior.

Microtubules are a crucial component of the cytoskeleton, essential for maintaining cell structure, facilitating intracellular transport, and playing key roles in developmental diseases and cancer (1–4). While the interactions of proteins to the outer surfaces and ends of microtubules have been extensively studied, the regulatory mechanisms within the microtubule lumen remain largely unexplored, particularly in singlet microtubules, which are vital for essential cellular processes such as cell division, intracellular transport, and signal transduction (5–7). Understanding the luminal environment is essential for elucidating how microtubule behavior shapes cellular responsiveness and adaptation.

The hollow lumen of microtubules, which can be up to 15 to 17 nm wide, contains the sites of posttranslational modifications, MIPs (8–10), and small molecules (11–14). These components collectively form the microtubule luminal environment, playing a key role in microtubule dynamics and cellular pharmacology. Studying how MIPs regulate this environment will provide insights into microtubule biology and its molecular regulation.

Since the 1960s, electron microscopy has revealed dense granules and protein particles within microtubules across a variety of organisms (9, 15). More recently, cryoelectron tomography has uncovered periodic protein arrangements within the lumen of doublet microtubules, such as those found in the respiratory cilia of bovines and the flagella of human and mouse sperm (16, 17). However, despite their broader distribution and involvement in diverse cellular processes, research into the luminal regulations of the singlet microtubules has been lagged behind. Investigating the regulatory mechanisms of singlet microtubules is critical for a comprehensive understanding of microtubule biology.

Nonetheless, several key findings have emerged: Tau and MAP6 proteins have been found localized to the inner walls of microtubules, where they stabilize microtubules and induce bending, respectively (18, 19). CSPP1, related to spindles and cilia, has also been shown to localize in the lumen of microtubules, contributing to their stabilization (20). Additionally, acetylation is the only known posttranslational modification that occurs within the microtubule lumen, and studies have shown that the acetyltransferase MEC17/ATAT1 enters the microtubule lumen through its ends to promote microtubule acetylation (21–23).

To address the limited understanding of MIPs in singlet microtubules, we developed a screening strategy based on the proximity-dependent biotinylation (Bio-ID) coupled to mass spectrometry. This approach allowed us to systematically identify candidate MIPs, including Jupiter microtubule-associated homolog 2 (JPT2). Our experiments revealed that JPT2 directly binds to microtubules and localizes within the lumen, where it modulates the luminal environment. Notably, JPT2 alters the localization of the acetylase MEC17 and the microtubule-binding compound Paclitaxel. Structural analysis using AlphaFold3 suggests that JPT2 may interact with β-Tubulin at the Paclitaxel-binding site, supporting its role as a key regulatory protein within the microtubule lumen. In summary, our study introduces a strategy for MIP discovery and highlights their crucial roles in regulating singlet microtubules.

Results

Establishment of the MIPs Identifying System.

To comprehensively identify MIPs, we developed a protein capture system using proximity labeling combined with mass spectrometry analysis (Fig. 1A). The N terminus and C-terminus of β-Tubulin are known to localize inside the microtubule lumen and outside, respectively (11, 24). For further study, we selected TUBB4B because it is highly expressed in the cervical cancer cell line HeLa cells and exists as the single isotype (25). We integrated the HA-tagged BirA* biotin ligase gene at both the start and stop codons of TUBB4B using the CRISPR-Cas9 system, successfully tagging both the N terminus and C-terminus of the TUBB4B protein (26–28) (SI Appendix, Fig. S1 A–F).

Fig. 1.

Establishment of the singlet MIPs screening system. (A) Schematic diagram of the MIP screen system. HA-BirA*-TUBB4B knock-in cells catalyzed biotinylation of the MIPs, then enrichment of biotinylated proteins using streptavidin-coated beads for isolation. (B) Overexpression of exogenous cytoplasm MIPs (CSPP1, MAP6, MEC17) and the microtubule surface protein MATCAP1 in HA-BirA*-TUBB4B knock-in cells, followed by immunoprecipitation using Flag tag to purify these proteins. Western blot was then used to detect the biotin modification level and Flag expressed level on the same PVDF membrane. (C) Bioinformatic analysis of HA-BirA*-TUBB4B cells and TUBB4B-BirA*-HA cells proteomic data. The volcano plot’s vertical axis represents the proteins’ molecular weight, while the horizontal axis shows the Log2 ratio of HA-BirA*-TUBB4B (N termini) to TUBB4B-BirA*-HA (C termini). Compared to the TUBB4B-BirA*-HA (C termini) cells, the red portion of the plot indicates proteins that are upregulated in the HA-BirA*-TUBB4B (N termini) cells, and the blue portion indicates proteins that are downregulated in the HA-BirA*-TUBB4B (N termini) cells. The data were corrected using the assembly coefficient of N termini and C termini, and the calculation of the assembly coefficient is shown in SI Appendix, Fig. S1 G and H.

To validate the functionality of this system, immunofluorescence staining confirmed that the engineered TUBB4B fusion proteins were correctly expressed in HeLa cells and formed microtubules (SI Appendix, Fig. S1 B, C, E, and F). Additionally, Western blot analysis using biotin detection demonstrated that the fused BirA* biotin ligase retained its enzymatic activity (SI Appendix, Fig. S1 C and F). To quantify the effect of N-terminal and C-terminal BirA* knock-in on the incorporation efficiency of tubulin into microtubules, we performed microtubule sedimentation assays using TUBB4B-BirA*-HA-N and TUBB4B-BirA*-HA-C cell lines. Western blot analysis revealed that TUBB4B-BirA*-HA-N had an incorporation efficiency of 23%, whereas TUBB4B-BirA*-HA-C showed a slightly lower efficiency of 16%. Both variants were successfully incorporated and properly assembled into microtubules (SI Appendix, Fig. S1 G and H). Furthermore, we exogenously expressed HA-MATCAP1, which is known to localize to the outer surface of microtubules (29), alongside Flag-mEGFP-CSPP1, Flag-mEGFP-MAP6, and MEC17-mEGFP, which are located within the lumen of microtubules (19, 20, 22) (SI Appendix, Fig. S1I). Proximity labeling experiments followed by immunoprecipitation using anti-Flag antibodies were then conducted (Fig. 1B). The results revealed that CSPP1, MAP6, and MEC17 were effectively labeled with biotin, except for MATCAP1, a known outer-surface microtubule protein (29), which showed no detectable biotinylation. These findings demonstrate that our MIPs identification system is robust and effective.

To ensure accurate interpretation of BioID enrichment results, we normalized the protein enrichment data based on the incorporation efficiency of BirA*-Tubulin fusion proteins, as determined by sedimentation assays (SI Appendix, Fig. S1 G and H). Using this system, we identified several candidate MIP proteins (Fig. 1C). Comparison of proteins labeled with TUBB4B-BirA* at the N- vs. C-terminus revealed that MCAF1, CLASP1, MAP4, MAP7, and various KIF proteins—known to interact with the outer surface of microtubules (30–35)—were more abundant in proteins labeled from the C-terminus. Conversely, proteins known to reside in the microtubule lumen, such as CSPP1 (20), were more enriched in proteins labeled with BirA* at the N terminus. These results provided us with candidate proteins for potential MIPs. Among these, JPT2 showed robust N-terminal labeling enrichment and was selected for in-depth analysis.

JPT2 Is a MIP.

We prioritized proteins with lower molecular weights, as most reported MIPs in singlet or doublet microtubules have molecular weights below 100 kD (SI Appendix, Fig. S1J) (16–20, 23, 36–38). Among the identified proteins, JPT2 (also known as HN1L) attracted our attention (Figs. 1C and 2A). JPT2 is a homolog of the Drosophila Jupiter protein (39–41). While JPT2 has been strongly linked to multiple cancers in human cells and to resistance against Paclitaxel treatment, its molecular mechanism remains poorly understood (42–48).

Fig. 2.

JPT2 is a singlet MIP. (A) Schematic illustration of the full-length JPT2 and the JPT2 mutant, which lacks microtubule binding capability. (B) Overexpressed MEC17 (MIP), JPT2, and MATCAP1 (outer MAP) in HA-BirA*-TUBB4B knock-in cells were purified using immunoprecipitation with the Flag tag. Western blot was then used to detect the biotin modification and Flag expression levels on the same PVDF membrane. (C) Immunofluorescence staining of α‐Tubulin (red) in CopGFP, CopGFP-JPT2, and CopGFP-JPT2-Mutant expressing HeLa cells. (Scale bars: 10 μm.) (D) Quantification of the localization of CopGFP-JPT2 and CopGFP-JPT2-mutant with α-Tubulin. Each symbol represents an individual cell. For CopGFP-JPT2, n = 67 cells; for CopGFP-JPT2-mutant, n = 63 cells. (E) 3D single-molecule localization microscopy (ROSE-Z) image of CopGFP-HA-JPT2 and CopGFP-HA-MAP7 (green) and α-Tubulin (red) in Caco2 cells, (Scale bars: 500 nm for XY plane; 30 nm for XZ plane.) (F) Diagram illustrating the statistical methods for analyzing JPT2, MAP7, and α-Tubulin ROSE-Z signals. For more details, see the Methods section for a comprehensive analysis pipeline. (G) ROSE-Z superresolution cross-sectional images were used to measure the radial distances from the fluorescence centroids of CopGFP-HA-JPT2 (red) and CopGFP-HA-MAP7 (blue) to the center of individual microtubule rings. For each ring, the distance was normalized by its own fitted microtubule radius, and values are presented as centroid-to-center distance divided by the corresponding microtubule radius (D/R). Each dot represents a single centroid measurement. For CopGFP-HA-JPT2, n = 101 from three cells; for CopGFP-HA-MAP7, n = 114 from three cells. A dashed line at D/R = 1 indicates the normalized position of the microtubule wall. Fluorescence centroids with D/R < 1 are considered luminal, while those with D/R > 1 are outside the lumen. (H) Quantification of ratios of JPT2 and MAP7 radius to α-Tubulin radius. (I) TIRF microscopy images of dynamic microtubules in the presence of 30 nM CopGFP or 10 nM CopGFP-JPT2. CopGFP-JPT2 fluorescence (green) shows a gradient distribution along microtubules (red), with decreasing intensity toward the center. GMPCPP-stabilized seeds are shown in blue. (Scale bars: 10 µm.) Kymograph of dynamic microtubules polymerized in the presence of 10 nM CopGFP-JPT2 is shown in the Right panel. (Scale bars: vertical, 2 μm; horizontal, 1 min.) (J) Quantification of CopGFP-JPT2 fluorescence along microtubules reveals an exponential decay from the plus end, with a fitted decay length (τ) of 0.96 ± 0.49 µm (n = 13 microtubules from at least two independent experiments). (K) Kymographs of dynamic microtubules polymerized in the presence of 200 nM CopGFP or 200 nM CopGFP-JPT2. (Scale bars: vertical, 2 μm; horizontal, 1 min.) (L–O) Quantification of microtubule dynamic parameters. At 200 nM, CopGFP-JPT2 significantly increases growth rate (n = 79 events for CopGFP group; n = 94 events for CopGFP-JPT2 group), shrinkage rate (n = 63 events for CopGFP group; n = 70 events for CopGFP-JPT2 group), catastrophe frequency (n = 30 microtubules for CopGFP and CopGFP-JPT2 group), and rescue frequency (n = 30 microtubules for CopGFP and CopGFP-JPT2 group). Thirty microtubules from at least two independent experiments were analyzed. Values = means ± SEM. Statistical analyses: Student’s t test. N.S P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To confirm that JPT2 is a potential microtubule inner protein, we transfected the HA-BirA*-TUBB4B (N termini) knocked-in HeLa cell line with MEC17-Flag, Flag-MATCAP1, and Flag-mEGFP-JPT2, respectively, and conducted proximity labeling experiments (Fig. 2B). Notably, the luminal BirA* specifically biotinylated JPT2 and MEC17, but not MATCAP1, indicating that JPT2 is a MIP candidate.

To investigate the localization of JPT2 within cells, we faced the challenge of the lack of suitable specific antibodies for immunofluorescence staining of endogenous JPT2. Therefore, we overexpressed CopGFP-JPT2 in cells to observe the localization of JPT2 on microtubules (Fig. 2 A and C). Immunofluorescence staining with anti-α-Tubulin antibodies revealed colocalized CopGFP and α-Tubulin signals, suggesting that JPT2 localizes to microtubules (Fig. 2 C and D).

Furthermore, to confirm JPT2’s localization within the microtubule lumen and to intensify the signal, we transfected CopGFP-HA-JPT2 into Caco2 cells and performed immunostaining. Superresolution microscopy (ROSE-Z), which allows visualization of the microtubule lumen due to its high axial resolution, revealed that the HA signal was localized within the microtubule lumen, enclosed by α-Tubulin, further supporting the luminal localization of JPT2 (Fig. 2E and SI Appendix, Fig. S2A) (49). As a control, we transfected CopGFP-HA-MAP7 into Caco2 cells and observed its localization on microtubules. MAP7 is a microtubule-binding protein that localizes to the outer surface of microtubules (34). Indeed, the ROSE-Z imaging confirmed that its signal is primarily localized to the outer regions of the microtubules (Fig. 2E and SI Appendix, Fig. S2B). ROSE-Z imaging further revealed that the spatial distribution of JPT2 signals was significantly closer to the microtubule inner surface compared to that of MAP7 (Fig. 2 F–H). Quantitative analysis confirmed that the relative localization radius of JPT2 was smaller than that of tubulin signal and that of MAP7, supporting its luminal localization.

PPGGK*S Motif Is Important for JPT2 Localization at the Microtubule.

To further investigate the key sequences responsible for JPT2’s localization to microtubules, a series of JPT2 mutants were generated for screening (SI Appendix, Fig. S3A). We expressed full-length JPT2 and its mutants, followed by immunostaining using anti-α-Tubulin antibodies. The results indicate that both amino acids 61th to 80th, 81th to 100th, and 178th to 190th can localize to microtubules, suggesting that JPT2 has multiple regions associated with its microtubule localization (Fig. 2C and SI Appendix, Fig. S3 B and C). Interestingly, JPT2 possesses four incompletely repeated conserved sequences, which include the PPGG amino acid residues. Notably, the last three of these sequences are located in regions critical for microtubule localization (Fig. 2A). This spatial overlap supports their potential functional role.

We found that the characteristic sequence within the PPGG sequence of JPT2 is similar to the PGGG motif, a sequence found in the Tau protein that plays a key role in its localization to the microtubule lumen (18, 50). Further experiments found that mutants lacking the last three incompletely repeated conserved PPGG sequence groups completely lost the ability to localize to microtubules (Fig. 2 C and D). Unlike the first motif, the last three contain an additional lysine (K) residue, forming a distinct PPGGK motif potentially important for microtubule interaction (Fig. 2A). These results suggest that the PPGGK*S motif contributes critically to JPT2’s microtubule association.

JPT2 Directly Binds to Microtubules.

To further investigate the interaction between JPT2 and microtubules, we purified CopGFP-JPT2 using the Strep-tagII expression system in HEK293T cells (SI Appendix, Fig. S4A). Cosedimentation assays confirmed JPT2’s ability to interact directly with microtubules (SI Appendix, Fig. S4B).

We performed in vitro microtubule polymerization assays and visualized dynamic behaviors using total internal reflection fluorescence (TIRF) microscopy (Fig. 2I and Movie S1). The CopGFP-JPT2 signal (10 nM) exhibited a gradient distribution along microtubules, with enrichment at both ends and decreasing intensity toward the lattice (Fig. 2 I and J and Movie S1). Quantitative analysis revealed that the signal decayed exponentially from the microtubule plus end, with a characteristic decay length of 0.96 ± 0.49 µm (Fig. 2J). This distribution is consistent with JPT2 functioning as a MIP. If JPT2 bound to the outer surface, we would expect either 1) a uniform distribution due to constant free JPT2 concentration, or 2) a plus-end-enriched pattern with a decay length corresponding to the ~200 nm tapered tip region, similar to EB proteins. Neither pattern matches the observed ~1 µm decay. To further investigate the binding of JPT2 to stable microtubules, we introduced CopGFP-JPT2 (200 nM) to prepolymerized GMPCPP-stabilized microtubules. The CopGFP-JPT2 signal appeared at the microtubule ends and extended inward along the lattice, showing an end-proximal accumulation pattern (SI Appendix, Fig. S4C). Quantitative analysis revealed that the fluorescence intensity decayed exponentially along the microtubule lattice, yielding characteristic decay lengths (τ) of 0.34 ± 0.23 µm and 0.31 ± 0.17 µm, respectively (SI Appendix, Fig. S4 D and E). Furthermore, a paired-sample analysis of decay constants measured at both microtubule ends showed no significant difference (P = 0.68, paired t test), indicating that the JPT2 signal decays symmetrically at the two ends (SI Appendix, Fig. S4F). These findings suggest that JPT2 selectively enters the microtubule lumen through the open ends.

We then used a higher concentration to measure the effects of JPT2 on microtubule dynamics. We found that at this concentration, JPT2 had a significant decoration on microtubule lattice, and it showed a significant, yet modest, increase in microtubule growth rate, shrinkage rate, catastrophe frequency, and rescue frequency (Fig. 2 K–O and Movie S2). These results indicate that JPT2 directly affects microtubule dynamics, although its overall impact remains relatively modest.

As a comparison, JPT1 exhibited a similar gradient distribution with an exponential decay length of 0.58 ± 0.19 µm (SI Appendix, Fig. S4 G and H and Movie S1). At 200 nM, JPT1 had a similar effect on the growth rate, shrinkage rate, catastrophe frequency, and rescue frequency (SI Appendix, Fig. S4 I–M and Movie S2), suggesting that JPT1 had a similar binding mechanism onto microtubules and effects on microtubule dynamics.

Our findings were further validated in vivo. We transfected CopGFP-IRES-EB3-mScarlet and CopGFP-JPT2-IRES-EB3-mScarlet into HeLa cells to study the microtubule plus end dynamics (SI Appendix, Fig. S4N). Live-cell imaging demonstrated an increase in the growth rate of EB3-labeled growing microtubule ends in cells expressing JPT2, along with a shortened life cycle (SI Appendix, Fig. S4O). Consistent with our in vitro findings (Fig. 2 K–O), these results indicate that JPT2 directly interacts with microtubules and exerts a modest yet reproducible influence on their dynamics.

JPT2 Prevents MEC17 Microtubule Localization.

The 15 to 17 nm lumen of microtubules constitutes a microenvironment that includes various luminal proteins, amino acids of tubulin exposed to the lumen, and small molecules bound to the microtubule inner surface (14, 19, 20). Since K40 acetylation of α-Tubulin occurs on the inner surface of microtubules (51, 52), and JPT1 has been shown to affect this modification (53). We explored whether these microtubule luminal environmental factors might be associated with each other.

To test this hypothesis, we generated a CRISPR/Cas9-engineered HeLa cell line in which a CopGFP tag was knocked in at the endogenous JPT2 locus, allowing us to monitor native JPT2 localization. Immunofluorescence staining revealed that CopGFP-JPT2 exhibited a discontinuous distribution along microtubules even at endogenous expression levels (Fig. 3 A and B). Importantly, its colocalization with acetylated microtubules was significantly lower than with total microtubules (Fig. 3 A–D), suggesting a potential spatial exclusion or inverse relationship between JPT2 and microtubule acetylation.

Fig. 3.

JPT2 inhibits microtubule acetylation by restricting MEC17 binding to microtubules. (A) Immunofluorescence staining of α‐Tubulin (red) in cop-GFP-JPT2 knock-in Hela cells, (Scale bars: 10 μm.) (B) Western blot was used to detect JPT2 expression levels in wild-type HeLa cells and CopGFP-JPT2 knock-in HeLa cells. GAPDH as a reference protein. (C) Immunostaining for Ace-Tubulin (red) in CopGFP-JPT2 knock-in HeLa cells, (Scale bars: 10 μm.) (D) Quantification of the colocalization of CopGFP-JPT2 and Ace-Tubulin or α-Tubulin. Each symbol represents an individual cell. For the α-Tubulin group, n = 101 cells; for the Ace-Tubulin group, n = 88 cells. (E) Western blot was used to detect Ace-Tubulin expression level in CopGFP-JPT2 and CopGFP overexpressing HeLa cells, α-Tubulin as a reference protein. (F) Quantification of relative protein level of Ace-Tubulin signal from (E), n = 3. (G) Immunostaining for Ace-Tubulin (red) in CopGFP-JPT2 and CopGFP overexpressing HeLa cells, (Scale bars: 10 μm.) (H) Quantification of the ratio of Ace-Tubulin/α-Tubulin intensity in CopGFP-JPT2 and CopGFP overexpressing HeLa cells. The mean intensity of cells overexpressing CopGFP-JPT2 was normalized to the mean intensity in control cells. Total number of cells analyzed: control cells, n = 77; cells overexpressing CopGFP-JPT2, n = 55. Quantification of the fluorescence intensity of α-Tubulin in CopGFP-JPT2 and CopGFP overexpressing HeLa cells. Total number of cells analyzed: control cells, n = 42; cells overexpressing CopGFP-JPT2, n = 56. Bars represent pooled data from two independent experiments. (I) TD-tomato (red) and CopGFP (Green) signals were obtained in MEC17-TD-tomato-Flag-IRES-Flag-CopGFP and MEC17-TD-tomato-Flag-IRES-Flag-CopGFP-JPT2 overexpressing in MEC17-JPT2 double KO HeLa cells respectively, (Scale bars: 10 μm.) (J) Quantification of the mean MEC17 intensity from images as in (I). The total number of cells analyzed: MEC17-TD-tomato-IRES-CopGFP-OE cells, n = 349; cells overexpressing MEC17-TD-tomato-IRES-CopGFP-JPT2 cells, n = 237. Bars represent pooled data from two independent experiments. Values = means ± SEM. Statistical analyses: Student’s t test. N.S P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To determine whether microtubule acetylation status affects JPT2 localization, we treated cells with the deacetylase inhibitor TSA to globally elevate tubulin acetylation. This manipulation had no discernible impact on the microtubule distribution of JPT2 (SI Appendix, Fig. S5 A and B), indicating microtubule acetylation does not affect the localization of JPT2 on microtubules.

Next, we explored whether JPT2 could influence the level of microtubule acetylation. We overexpressed or knocked out JPT2 in HeLa cells and examined the changes in acetylation. The results showed that overexpression of JPT2 significantly reduced the acetylation level of microtubules, while the total microtubule mass remained unchanged in cells, and knockout of JPT2 did not lead to significant changes (Fig. 3 E–H and SI Appendix, Fig. S5 C–G). Given that JPT2 is already localized to microtubules at endogenous levels, this suggests that JPT2 is not essential for maintaining basal acetylation levels but may suppress microtubule acetylation in a dose-dependent manner.

Given that microtubule acetylation and deacetylation are dynamically regulated by specific enzymes (23, 54, 55), we further examined whether JPT2 affects acetylation by interacting with these enzymes. Coimmunoprecipitation experiments showed that HA-BirA*-JPT2 does not interact with HDAC6 or SIRT2, the two main deacetylases of microtubules (SI Appendix, Fig. S6A), thereby excluding the possibility that JPT2 directly regulates acetylation through these deacetylases.

We then focused on whether the function of the acetyltransferase MEC17 is affected by JPT2. To eliminate background acetylation, we constructed a HeLa cell line with double knockout of JPT2 and MEC17 and transfected either MEC17-tdTomato-Flag-IRES-Flag-CopGFP or MEC17-tdTomato-Flag-IRES-Flag-CopGFP-JPT2 into these cells. Immunofluorescence results showed that MEC17 could normally localize to microtubules when expressed with CopGFP alone (Fig. 3I), but in cells coexpressing JPT2, MEC17 clearly lost its ability to localize to microtubules (Fig. 3J and SI Appendix, Fig. S6 B and C). These findings indicate that JPT2 may modulate microtubule acetylation via a nonenzymatic mechanism, potentially by interfering with the microtubule localization of MEC17. This possibility is consistent with JPT2’s distinct distribution along microtubules, as shown above (Fig. 3 A and B).

To examine whether this phenomenon is conserved within the Jupiter family, we conducted sequence conservation analysis and functional validation. Sequence alignment showed that the N- and C-terminal regions of Drosophila Jupiter and mammalian JPT1/JPT2 are highly conserved (SI Appendix, Fig. S7A). When CopGFP-tagged Drosophila Jupiter was transfected into HeLa cells, we observed strong nuclear signals and clear colocalization with microtubules (SI Appendix, Fig. S7B). Furthermore, when a 17-amino acid peptide containing a partial repeat of the conserved PPGGK*S motif at Jupiter’s C-terminus was overexpressed (SI Appendix, Fig. S7C), we unexpectedly observed a marked change in the pattern of microtubule acetylation, with the continuous signal becoming dotted (SI Appendix, Fig. S7 C–E). Similarly, overexpression of CopGFP-JPT1 in HeLa cells also led to a decrease in the acetylation level of intracellular microtubules (SI Appendix, Fig. S7 F and G). These findings suggest that the ability to regulate the microtubule lumen and its acetylation status may be a conserved function across the Jupiter family of proteins.

Together, our findings support the initial hypothesis that JPT2 modulates microtubule acetylation by shaping the luminal environment through a nonenzymatic mechanism. This regulatory model provides insights into how nonenzymatic luminal proteins can influence microtubule modifications.

JPT2 and Paclitaxel Interfere within the Microtubule Lumen.

Small molecules bound to the inner surface of microtubules are also important elements of the microtubule environment (14). Based on the known association of the PGGG motif with the β-Tubulin Paclitaxel binding site (18), we used AlphaFold3 to predict how JPT2 interacts with microtubules (Fig. 4 A and B). Our simulations focused on the 13 amino acids at the C-terminus of JPT2 (Fig. 4A), as our experiments demonstrated the presence of the PPGGK*S motif and significant localization with microtubules (SI Appendix, Fig. S3). The results showed that the serines at positions 186 and 188 of the 13 amino acids (LNPPGGKSSISFY) at the C-terminus of JPT2 could bind to aspartic acid at position 224 (Asp224) and arginine (Arg) at position 276 of the core helix H7 and M-loop of β-Tubulin (56), respectively (Fig. 4B). When we mutated these two key serines to alanine, the colocalization phenomenon between the C-terminus of JPT2 and the microtubules disappeared (Fig. 4 C–E). This result is consistent with the AlphaFold3 prediction.

Fig. 4.

Paclitaxel disassociates JPT2 from microtubules. (A) Schemes of the domain organization of the full-length JPT2, JPT2-178-190 a.a. which contains the minimal microtubule-binding motif of JPT2. (B) The AlphaFold3 prediction suggests that the binding site of JPT2-178-190 a.a. with β-Tubulin is similar to the binding site of Paclitaxel. (C) Immunostaining for α-Tubulin (red) in CopGFP-JPT2-178-190 a.a. and CopGFP-JPT2-178-190 a.a.-SS > AA expressing HeLa cells were shown in the Right panel, (Scale bars: 10 μm.) (D) Quantification of the CopGFP-JPT2-178-190 a.a. and CopGFP-JPT2-178-190 a.a.-SS > AA intensity in HeLa cells. For the CopGFP- JPT2-178-190 a.a. group, n = 83 cells; for the CopGFP-JPT2-178-190 a.a.-SS > AA group, n = 79 cells; Bars represent pooled data from two independent experiments. (E) Quantification of the CopGFP-JPT2-178-190 a.a. and CopGFP-JPT2-178-190 a.a.-SS > AA colocalization with α-Tubulin in HeLa cells. For the CopGFP- JPT2-178-190 a.a. group, n = 35 cells; for the CopGFP-JPT2-178-190 a.a.-SS > AA group, n = 42 cells; Bars represent pooled data from two independent experiments. (F) Flag-CopGFP-JPT2-178-190 a.a. overexpressed in the HA-BirA*-TUBB4B knock-in cells, catalyzed biotinylation, and treated with increasing concentrations Paclitaxel, followed by IP purify Flag-CopGFP- JPT2-178-190 a.a. protein, Western blot detected proteins Biotin modification level and Flag expressed level in the same PVDF membrane. (G) Quantification of the relative protein level of Biotin signal from (F), n = 3. (H) Flag-CopGFP-JPT2 overexpressed in the HA-BirA*-TUBB4B knock-in cells, catalyzed biotinylation, and treated with increasing concentrations Paclitaxel, followed by IP purify Flag-CopGFP-JPT2. Western blot detected proteins Biotin modification level and Flag expressed level in the same PVDF membrane. (I) Quantification of the relative protein level of Biotin signal from (H), n = 3. (J) Immunostaining for α-Tubulin (red) in DMSO and Paclitaxel-treated CopGFP-JPT2-178-190 a.a stably expressing HeLa cells, (Scale bars: 10 μm.) (K) Quantification of the JPT2-178-190 a.a intensity in DMSO and Paclitaxel-treated HeLa cells. The mean intensity of cells treated with Paclitaxel was normalized to that in cells treated with DMSO. Total number of cells analyzed: control cells, n = 111; Paclitaxel group, n = 118. Bars represent pooled data from two independent experiments. Values = means ± SEM. Statistical analyses: Student’s t test. N.S P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

This allowed us to examine how JPT2 and Paclitaxel influence the binding of Asp224 and Arg227, which are critical amino acids for Paclitaxel binding (56). We transfected HA-BirA*-TUBB4B knock-in HeLa cells with Flag-CopGFP-JPT2 (amino acids 178 to 190) and Flag-CopGFP-JPT2, then performed proximity labeling assays using various concentrations of Paclitaxel (Fig. 4 F–I). We observed a decline in the biotinylation of both Flag-CopGFP-JPT2 (178 to 190) and full-length Flag-CopGFP-JPT2 with increasing Paclitaxel levels, indicating that Paclitaxel impacts the biotinylation of JPT2. Moreover, we treated HeLa cells overexpressing CopGFP-JPT2 (178 to 190) with 100 nM Paclitaxel for 12 h and conducted immunostaining, which showed reduced microtubule localization of JPT2 after Paclitaxel treatment (Fig. 4 J and K). These findings suggest that Paclitaxel may influence the distribution of JPT2 within the microtubule lumen in cells.

To investigate whether JPT2 displacement was driven by conformational remodeling or increased steric hindrance within the microtubule lumen, we exposed CopGFP-JPT2-expressing HeLa cells to a panel of microtubule-binding ligands targeting distinct sites on the tubulin lattice (SI Appendix, Fig. S8 A and B). Ligands engaging the taxane-binding pocket—such as Paclitaxel, Epothilone D, and TPI287—disrupted JPT2 localization, while ligands binding alternative regions (e.g., Noscapine and TTI237) did not (SI Appendix, Fig. S8 C and D). Proximity labeling assays at different concentrations yielded consistent results (Fig. 4 G and H and SI Appendix, Fig. S8 E–H). To visualize real-time changes in JPT2 localization, we performed TIRF microscopy upon compound exposure (SI Appendix, Fig. S9 and Movie S3). These results demonstrate that ligands binding the taxane pocket specifically impair JPT2–microtubule association, likely through lattice remodeling and steric exclusion, rather than general changes in microtubule stability.

Last, live-cell imaging of HeLa cells overexpressing Flag-mEGFP-JPT2 or Flag-mEGFP and treated with low concentrations of deep red fluorophore-conjugated Docetaxel revealed a reduction in Docetaxel association with microtubules in regions enriched with JPT2 (SI Appendix, Fig. S8 I and J). These results suggest a mutual influence between JPT2 and ligands targeting the taxane-binding site, highlighting the dynamic interactions occurring within the microtubule lumen.

Intraluminal Protein–Compound Competition Defines a Dynamic Regulatory Axis within Microtubules.

Having observed that JPT2’s luminal localization is sensitive to Paclitaxel, we next investigated whether other microtubule inner proteins (MIPs) display similar responses. Using a biotin-based proximity labeling assay under different concentrations of Paclitaxel (Fig. 5A), we performed proteomic screening and identified 29 proteins whose microtubule association was altered, including JPT2, JPT1, and CSPP1 (Fig. 5B), the latter of which has previously been localized to the microtubule lumen (20).

Fig. 5.

JPT1 and CSPP1 were new Paclitaxel-sensitive MIPs (A) Schematic diagram showed HA-BirA*-TUBB4B knock-in cells undergoing biotinylation of MIPs and treatment with increasing concentrations of Paclitaxel, followed by streptavidin bead enrichment to capture the proteins. (B) Bioinformatic analysis of proteomic data found two new Paclitaxel-sensitive MIPs. The heat map shows the changes in MIPs in different concentrations of Paclitaxel treatment groups. (C and D) Falg-CopGFP-JPT1 (E and F) mEGFP-CSPP1 (G and H) Falg-mEGFP-MAP6 overexpressed in the HA-BirA*-TUBB4B knock-in cells, catalyzed biotinylation in the MIP and treated with increasing concentrations Paclitaxel, followed by IP purify MIP, Western blot detected proteins Biotin modification level and Flag/GFP expressed level in the same PVDF membrane. The relative protein levels of Biotin signals were quantified, n = 3. (I) Immunostaining for mEGFP-CSPP1, CopGFP-JPT1, mEGFP-MAP6 (Green) and α-Tubulin (red) in DMSO and Paclitaxel-treated HeLa cells, (Scale bars: 10 μm.) (J) Quantification of the CSPP1, JPT1, and MAP6 intensity in DMSO and Paclitaxel-treated HeLa cells are shown in the Right panel. For the CSPP1 DMSO group, n = 70 cells; for the CSPP1 Paclitaxel group, n = 100 cells. For the JPT1 DMSO group, n = 64 cells; for the JPT1 Paclitaxel group, n = 76 cells. For the MAP6 DMSO group, n = 67 cells; for the MAP6 Paclitaxel group, n = 91 cells. Values = means ± SEM. Statistical analyses: Student’s t test. N.S P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To validate these results, we compared the proximity labeling signal of several representative proteins. We found that the biotinylation levels of JPT1, and CSPP1 significantly decreased upon Paclitaxel treatment (Fig. 5 C–F), whereas MAP6, a known microtubule-lumen-associated protein (19), showed no change in biotinylation (Fig. 5 G and H). This suggests that different MIPs vary in their response to Paclitaxel.

To determine whether this effect was specific to taxane-class compounds, we compared the effects of taxane (Paclitaxel, Epothilone D, TPI287) and nontaxane (Noscapine, TTI237) stabilizers. Taxanes significantly reduced the biotinylation of JPT1 and CSPP1, while nontaxane stabilizers showed no detectable effects (SI Appendix, Fig. S10 A–H). These results were corroborated by immunofluorescence staining (Fig. 5 I and J), confirming a consistent pattern of taxane sensitivity.

Furthermore, time-lapse imaging demonstrated that overexpression of JPT1 and CSPP1 reduced the binding of fluorophore-conjugated Docetaxel to microtubules (SI Appendix, Fig. S10I), similar to our previous observations with JPT2 (Fig. 4 K and L). These results highlight a previously underappreciated mechanism of spatial competition within the microtubule lumen, where endogenous proteins and small-molecule binders can mutually influence their intraluminal accessibility.

Discussion

In this study, we established a proximity-labeling strategy based on CRISPR-engineered BirA*-Tubulin to identify key MIPs systematically. Among the identified candidates, JPT2 emerged as a key regulator of the microtubule luminal environment. Our findings demonstrate that JPT2 localizes within the microtubule lumen and spatially interferes with the luminal positioning of the acetyltransferase MEC17, as well as the microtubule binding of the tubulin-interacting compound Paclitaxel. These observations underscore the importance of luminal protein–protein and protein–compound interactions in organizing the internal architecture of microtubules and establish a robust framework for functional dissection of MIPs in diverse cellular contexts. Importantly, they also reveal that luminal proteins may modulate microtubules through multiple interconnected layers, including spatial accessibility, enzymatic modification, and dynamic behaviors.

Systematic Identification of MIPs in Singlet Microtubules.

In this study, we developed a systematic proximity-labeling strategy to identify proteins localized within the lumen of singlet microtubules—an intracellular space that remains largely underexplored. By integrating spatially targeted biotinylation, quantitative proteomics, and functional validation, our platform enables high-specificity detection of MIPs at the cellular level (Fig. 1A).

Taking advantage of the structural polarity of β-tubulin, we generated CRISPR knock-in cell lines expressing BirA*-tagged TUBB4B at either the N terminus (facing the lumen) or C-terminus (facing the cytoplasm). TUBB4B was chosen as the tagging target due to its unique expression among β-tubulin isoforms in HeLa cells, minimizing background. Comparative analysis of biotinylated proteomes under the two configurations, normalized by microtubule incorporation efficiency, allowed us to identify luminally enriched MIP candidates (Fig. 1C and SI Appendix, Fig. S1 G and H). Subsequent validation via structural prediction, high-resolution imaging, and colocalization with microtubules confirmed their luminal localization and functional roles in regulating microtubule dynamics and acetylation. This approach successfully identified known MIPs such as CSPP1, as well as previously uncharacterized candidates including JPT2 and JPT1 (Fig. 1C) (20).

Our system demonstrates high spatial precision, low background interference, and strong reproducibility, and holds promise for future in vivo applications. It could be extended to animal models to explore the dynamic behavior and physiological relevance of MIPs under various biological conditions. Nonetheless, the method has certain limitations. The confined geometry of the microtubule lumen may impose steric hindrance, and the size of BirA* itself could restrict access to transient or low-abundance proteins, potentially leading to underrepresentation. Furthermore, the system’s applicability across diverse cell types and its ability to capture dynamic protein interactions remain to be fully validated.

Future efforts should focus on implementing smaller proximity labeling enzymes (such as miniTurbo or APEX2), optimizing insertion sites, and combining orthogonal strategies to enhance labeling sensitivity and coverage (57–59). Expanding the platform to more cell types and in vivo contexts will provide new opportunities to investigate the composition, regulation, and functional significance of the microtubule lumen proteome.

Microtubule Luminal Localization of JPT2.

JPT2 was identified as one of the MIPs through our system. JPT2 is a homolog of Drosophila’s microtubule-associated protein Jupiter and is highly associated with cancer progression and prognosis, though its interaction with microtubules is less studied (39, 60). We confirmed that JPT2 is a direct microtubule-binding protein, with superresolution microscopy verifying its localization within the microtubule lumen.

To further characterize its spatial positioning, we performed ROSE-Z analysis, which revealed a significant inward radial shift of JPT2 fluorescence relative to the microtubule wall, consistent with luminal localization (Fig. 2 E–H). Complementary TIRF imaging showed that JPT2 fluorescence decays exponentially from the microtubule plus end along the lattice, with a decay constant (τ) of 0.96 ± 0.49 μm (Fig. 2 I and J and Movie S1). This decay length exceeds the typical size of the GTP-cap (~100 to 200 nm), indicating that the signal is not restricted to the open tip region (61). The observed decay pattern supports a model in which JPT2 enters through the open plus end and distributes longitudinally within the confined microtubule lumen, rather than binding uniformly along the outer surface. Together with the strong N-terminal biotinylation preference and ROSE-Z-revealed inward radial distribution, this spatial analysis further substantiates the intraluminal localization of JPT2 (Fig. 2 B and E–H).

We next explored the sequence and structural basis for JPT2’s interaction with microtubules. JPT2 contains four imperfect repeat motifs, primarily composed of PPGG. Except for the N-terminal repeat, the three repeats starting from JPT2’s C-terminal 13 amino acids can localize to microtubules and exhibit the conserved PPGGK*S sequence (Fig. 4A). We observed that the PPGGK*S sequence bears a high similarity to the PGG motif, which plays a crucial role in the Tau protein, especially in the Paclitaxel binding pocket of β-Tubulin. PGGG motif includes proline (P) and glycine (G), where proline induces specific folding and glycine provides flexibility. This combination forms a hairpin structure that fits well into the binding pocket, stabilizing microtubules and promoting their assembly (18). Using AlphaFold3, we predicted the interaction of JPT2’s C-terminal 13 amino acids containing PPGGK*S with α/β-Tubulin dimer, suggesting it can bind the Paclitaxel binding pocket of β-Tubulin similarly to the PGGG motif (Fig. 4B). This finding supports our observation that JPT2 and Paclitaxel mutually influence microtubule localization (Fig. 4 F–I). We hypothesize that the additional proline in PPGGK*S enhances structural rigidity, while glycine retains the necessary flexibility. This combination of rigidity and flexibility is crucial for fitting the complex three-dimensional structure of the binding pocket. The structural features of PPGGK*S motifs may inspire future studies on microtubule-targeting tools.

JPT2 Regulates the Microtubule Luminal Environment through Multilayered Mechanisms.

Having established the intraluminal localization of JPT2, we next investigated how its positioning affects microtubule structure and function. Microtubule acetylation is one of the few known posttranslational modifications that occur within the microtubule lumen (52, 62). We found that JPT2 disrupts the microtubule localization of the acetyltransferase MEC17, leading to reduced acetylation upon overexpression (Fig. 3 E–J). Interestingly, residual acetylation persists even in the absence of MEC17 on microtubules, suggesting that MEC17 may act in a transient, nonresident fashion or function upstream of additional, unidentified secondary acetyltransferases (23, 63, 64). These possibilities remain to be explored.

Beyond modulating enzymatic localization, JPT2 also affects intraluminal accessibility of small molecules (14, 19, 20). For example, Paclitaxel binds to the luminal face of β-tubulin to stabilize microtubules (56, 65). We observed that JPT2 expression reduces Paclitaxel accessibility, while Paclitaxel in turn interferes with JPT2 binding (Fig. 4 F–I). This mutual competition is not unique to JPT2, our biotinylation screen identified other luminal proteins whose microtubule association was similarly perturbed by compound treatment, suggesting a competitive environment within the lumen shaped by dynamic protein–compound interactions (Fig. 5 and SI Appendix, Fig. S10).

Furthermore, in vitro reconstitution assays demonstrated that both JPT2 and JPT1 promote microtubule growth and shrinkage rates, with minimal effects on catastrophe or rescue frequency (Fig. 2 K–O, SI Appendix, Fig. S4 I–M, and Movie S2). These findings suggest that the JPT family contributes not only to spatial occupancy within the lumen but also to the fine-tuning of microtubule dynamic instability.

Collectively, our results reveal that JPT2 coordinates multiple regulatory layers within the microtubule lumen—including enzymatic access, small-molecule binding, and dynamic behavior—underscoring how luminal proteins can act as integrative regulators. This supports a broader model in which the microtubule lumen serves as a multifunctional regulatory space where biochemical and structural signals converge to control microtubule function.

Physiological Relevance and Broader Implications.

In addition to its structural role, JPT2 has been previously identified as an NAADP-binding protein involved in calcium signaling from the endoplasmic reticulum and endolysosomal compartments (60, 66, 67). This raises the intriguing possibility that the localization and function of luminal proteins may themselves be regulated by intracellular signaling cues, pointing to a potential interface between microtubule architecture and broader cellular signaling networks.

Together, these findings not only reveal JPT2 as a functionally relevant MIP but also establish a generalizable framework for investigating how luminal proteins modulate microtubule behavior through spatial and molecular coordination. As new luminal regulators emerge, understanding their distribution, accessibility, and interactions will be critical to decoding the functional logic of the microtubule interior.

Materials and Methods

All plasmids, antibodies, and reagents used in this study are detailed in SI Appendix, Materials and Methods, which also describes the experimental protocols for cell culture, generating knock-in and knock-out cell lines, BioID and LC–MS/MS, Bio-IP and western blot, immunofluorescence staining and data statistics, protein purification, tubulin purification and in vitro microtubule dynamic assay, ROSE-Z imaging, EB3 tracking, and data analysis.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

The distribution of JPT1 and JPT2 along microtubules shows an exponentialdecay in fluorescence intensity from the plus end. Live TIRF imaging shows microtubules polymerized in the presence of 30 nM CopGFP, 30 nM CopGFP-JPT1 and 10 nM CopGFP-JPT2. The distribution of JPT1 and JPT2 proteins along microtubules reveals an exponential decay in fluorescence intensity from the plus end. CopGFPprotein is used as a negative control. GMPCPP-stabilized microtubule seeds (blue), GTPpolymerized microtubules (red), CopGFP, CopGFP-JPT1 and CopGFP-JPT2 (green). Imaging interval: 3 sec/frame. Scale bars: 5 μm.

Movie S2.

JPT1 and JPT2 proteins regulate microtubule dynamics. Live TIRF imaging shows microtubules polymerized in the presence of 200 nM CopGFP, 200 nMCopGFP-JPT1 and 200 nM CopGFP-JPT2. Both JPT1 and JPT2 proteins increase microtubule growth and shrinkage rates, with minimal effects on catastrophe or rescue frequency. CopGFP protein is used as a negative control. GMPCPP-stabilized microtubule seeds (blue), GTPpolymerized microtubules (red), CopGFP, CopGFP-JPT1 and CopGFP-JPT2 (green). Imaging interval: 3 sec/frame. Scale bars: 5 μm.

Movie S3.

Taxanes dissociate JPT2 from microtubules in vitro. Live TIRF imaging shows that taxane-class compounds (Paclitaxel, Epothilone D, TPI287) rapidly displace CopGFP-JPT2 from polymerized microtubules, whereas non-taxane molecules (TTI237, Noscapine) have no detectable effect. DMSO is used as a negative control. GMPCPP-stabilized microtubule seeds (blue), GTP-polymerized microtubules (red), CopGFP-JPT2 (green). Imaging interval: 5 sec/frame. Scale bars: 5 μm.

Data, Materials, and Software Availability

Study data are included in the article and/or supporting information. The mass-spectrometry datasets generated in this study are available from the Science Data Bank as entry: https://www.scidb.cn/s/aEZjey (68).