Identification of a lysosome membrane protein which could mediate ATP-dependent stable association of lysosomes to microtubules.

We have previously reported that purified thyroid lysosomes bind to reconstituted microtubules to form stable complexes (Mithieux, G., Audebet, C., and Rousset, B. (1988) Biochim. Biophys. Acta 969, 121-130), a process which is inhibited by ATP (Mithieux, G., and Rousset, B. (1988) Biochim. Biophys. Acta 971, 29-37). Among detergent-solubilized lysosomal membrane protein, we identified a 50-kDa molecular component which binds to preassembled microtubules. The binding of this polypeptide to microtubules was decreased in the presence of ATP. We purified this 50-kDa protein by affinity chromatography on immobilized ATP. The 50-kDa protein bound to the ATP column was eluted by 1 mM ATP. The purified protein, labeled with 125I, exhibited the ability of interacting with microtubules. The binding process was inhibited by increasing concentrations of ATP, the half-maximal inhibitory effect being obtained at an ATP concentration of 0.35 mM. The interaction of the 50-kDa protein with microtubules is a saturable phenomenon since the binding of the 125I-labeled 50-kDa protein was inhibited by unlabeled solubilized lysosomal membrane protein containing the 50-kDa polypeptide but not by the same protein fraction from which the 50-kDa polypeptide had been removed by the ATP affinity chromatography procedure. The 50-kDa protein has the property to bind to pure tubulin coupled to an insoluble matrix. The 50-kDa protein was eluted from the tubulin affinity column by ATP. These findings support the conclusion that a protein inserted into the lysosomal membrane is able to bind directly to microtubules in a process which can be regulated by ATP. We propose that this protein could account for the association of lysosomes to microtubules demonstrated both in vitro and in intact cells.


Gilles Mithieux and Bernard Rousset
From the Znstitut National de la Sante et de la Recherche Medicale,Unite 197,Faculte de Medecine Alexis Carrel,69372 Lyon Ceder 08,France We have previously reported that purified thyroid lysosomes bind to reconstituted microtubules to form stable complexes (Mithieux, G., Audebet, C., and Rousset, B. (1988) Biochirn. Biophys. Acta 969, 121-130), a process which is inhibited by ATP  Biochim. Biophys. Acta 971, 29-37). Among detergent-solubilized lysosomal membrane protein, we identified a 50-kDa molecular component which binds to preassembled microtubules. The binding of this polypeptide to microtubules was decreased in the presence of ATP. We purified this BO-kDa protein by affinity chromatography on immobilized ATP. The BO-kDa protein bound to the ATP column was eluted by 1 mM ATP. The purified protein, labeled with 12'1, exhibited the ability of interacting with microtubules. The binding process was inhibited by increasing concentrations of ATP, the half-maximal inhibitory effect being obtained at an ATP concentration of 0.35 mM. The interaction of the 50-kDa protein with microtubules is a saturable phenomenon since the binding of the l2'I-1abeled BO-kDa protein was inhibited by unlabeled solubilized lysosomal membrane protein containing the 50-kDa polypeptide but not by the same protein fraction from which the BO-kDa polypeptide had been removed by the ATP affinity chromatography procedure. The BO-kDa protein has the property to bind to pure tubulin coupled to an insoluble matrix. The 50-kDa protein was eluted from the tubulin affinity column by ATP. These findings support the conclusion that a protein inserted into the lysosomal membrane is able to bind directly to microtubules in a process which can be regulated by ATP. We propose that this protein could account for the association of lysosomes to microtubules demonstrated both in uitro and in intact cells.
Interphase microtubules constitute the tracks supporting the intracellular movements of organelles (1)(2)(3)(4)(5)(6)(7)(8)(9)(10). Microtubules are also involved in the positioning of vesicles inside the cell. Indeed, the intracellular distribution of subcellular vesicles is dependent on an intact microtubule framework (11-18) suggesting that, when not moving, organelles can bind to microtubules in a stable manner. It is now admitted that the microtubule-based translocations of organelles involve cytosolic protein motor systems hydrolyzing ATP, such as kinesin (19)(20)(21)(22). The molecular basis of stable association phenomena is still poorly understood. It is not known whether such associations require cytosolic component(s) or molecular * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. species which are preferentially attached to microtubules or to vesicles.
With intent to investigate the molecular mechanism(s) of the interaction of thyroid lysosomes (involved in the proteolytic cleavage of thyroglobulin, the thyroid prohormone) with microtubules, we have previously designed an in uitro model system in which purified lysosomes associate to reconstituted microtubules. We recently reported the characteristics of the interaction of microtubules with lysosomes in a complete acellular system (23). The formation of microtubule-lysosome complexes was visualized by negative staining electron microscopy and quantitatively analyzed using a radiometric assay based on the use of '251-labeled microtubules and a sedimentation procedure allowing the separation of free microtubules from lysosome-bound microtubules. Using the latter approach, we have been able to assess the possible regulatory actions of nucleotides. It was observed that ATP causes a concentration-dependent inhibition of the microtubule-lysosome interaction, this effect being unrelated to ATP hydrolysis (24).
Highly purified thyroid lysosomes have the capacity to bind to in uitro reconstituted microtubules without addition of cytosolic elements, and the microtubule-lysosome interaction was shown to be independent upon the presence of microtubule-associated proteins (23). These observations led us to postulate that an integral membrane protein of the thyroid lysosomes could be responsible for the association. To try to verify this hypothesis, we have used a double strategy based on the expected properties of such a protein. Attempts have been made to identify a protein extractable from lysosomal membrane preparations using detergents (a) capable of interacting with microtubules and (b) exhibiting an ATP binding activity. Here, we present evidence for the existence of a 50-kDa protein which fulfills these requirements.

EXPERIMENTAL PROCEDURES
Purification and Solubilization of Lysosomal Membrane Protein-Lysosomes were purified from pig thyroid according to the procedure previously described (25) and modified by the use of a hyperosmotic medium from the beginning of the purification (23). After osmotic pressure-dependent lysis, the lysosomal membranes were sequentially washed in 10 mM Tris-HC1, pH 7.4 (buffer T), then in the same buffer containing 0.6 M NaCI, and again in buffer T. Washings were made after centrifugation of the membranes at 100,000 X g for 30 min at for 30 min, and then insoluble material was removed by centrifugation at 100,000 X g for 30 min at 25 "C. Soluble lysosomal membrane protein (LMP) represented 35-45% of the starting membrane protein.
Purification of Microtubule Proteins-Microtubule protein was purified from rat brain by two cycles of temperature-dependent assembly-disassembly (26). Polymerizing steps were performed in buffer A (100 mM Mes, 1 mM EGTA, 0.5 mM MgCL, pH 6.4) supplemented by 1 mM GTP and 4 M glycerol. Tubulin was further purified from twicecycled microtubule protein by phosphocellulose chromatography (27).
Coupling of Tubulin to Affi-Gel IO-Affi-Gel 10 (Bio-Rad) was sequentially washed with isopropyl alcohol and distilled water (5 ml/ ml of gel). One mg of phosphocellulose-purified tubulin in buffer A was added to 1 ml of gel and placed at 4 "C for 4 h under gentle agitation. The mixture was then supplemented with 1 M ethanolamine, pH 8.0 (0.1 ml/ml of gel) for 1 h to block remaining reactive sites. The gel suspension was then placed in a column and washed with 10 bed volumes of buffer A.
ATP Affinity Chromatography-ATP affinity chromatography was performed using a column containing 1 ml of swollen ATP-agarose gel having an 8-carbon spacer (A 9264,Sigma). The LMP fraction (about 1 mg of protein) was diluted 5 times to reach a CHAPS concentration of 0.1% and applied on the gel at a flow rate of 1 ml/ min at room temperature. The column was sequentially washed with (a) 20 ml of buffer T; (b) 2 ml of buffer T containing 100 mM NaCI; and (c) 10 ml of buffer T. The column was then eluted with 2 ml of 1 mM ATP in buffer T. After extensive washing with buffer T containing 1.5 M NaC1, the ATP column was kept at 4 "C in 0.05% azide in buffer T for reuse.
The fractions eluted from the column were desalted by gel filtration on Sephadex G-25 (PD-10 columns from Pharmacia, Uppsala, Sweden) and concentrated up to a volume of 100 pl using the Speed Vac system from Savant. Bovine serum albumin was used as a carrier protein to coat the tubes and saturate the PD-10 column.
Labeling Procedures-Lysosome membranes were labeled with ' "I using the Bolton-Hunter reagent (Amersham, Les Ulis, France). Pelleted membranes (1 mg of protein) were suspended in 50 mM phosphate buffer, pH 7.0, and incubated with 0.5 mCi of lZ5I-labeled Bolton-Hunter reagent. After 30 min at 0 "C, the reaction was stopped by the addition of 10 mM glycine. '251-Labeled membranes were washed once in buffer T, and '251-labeled membrane proteins (lz5I-LMP) were extracted by CHAPS treatment as described above. The specific radioactivity of lZ5I-LMP was 0.15 pCi/pg of protein.
The ATP affinity-purified fraction, desalted and concentrated, was labeled with lz5I using the IODO-GEN reagent (Pierce, Cambridge, United Kingdom). Two pg of the ATP-purified protein in 50 p1 of buffer T were mixed with 0.25 mCi of ['251]iodide in a conical tube coated with IODO-GEN. After 2 min at 0 "C, the mixture was diluted to 500 pl with buffer T and transferred on a PD-10 Sephadex column presaturated with BSA to separate residual ['251]iodide from the labeled protein. '"I-Labeled ATP affinity-purified protein (specific activity, 3 pCi/pg of protein) was stored in liquid nitrogen in the presence of 0.5 mg/ml BSA in buffer T.
Microtubule Binding Experiments-Microtubules (2.5 mg of microtubule protein/ml) were assembled for 30 min at 37 "C and sheared by four to five passages through the needle (25 G) of a syringe and kept for 5 min at 25 "C. One-hundred p1 of the microtubule suspension were mixed with the '251-labeled protein (either LMP or ATP affinitypurified protein) in a total volume of 400 p1 of buffer A containing 10 mg/ml BSA. When present, ATP was added just before microtubules. After 30 min at 25 "C, the incubation mixtures were layered on 1 M sucrose in buffer A and centrifuged at 100,000 X g for 1 h at 25 "C. Pellets and supernatants were counted for radioactivity in a Packard scintillation y counter and analyzed for protein by SDS-PAGE and autoradiography.
Other Methods-Protein was assayed according to the method of Lowry using BSA as a standard. PAGE was performed on 9% polyacrylamide gel in the presence of SDS according to Laemmli (28) including slight modifications (29).

RESULTS
Attempts to Evidence Solubilized Lysosomal Membrane Protein with Microtubule Binding Activity-Microtubules were incubated with lZ5I-LMP, in the absence or in the presence of ATP, with the aim to detect LMP component(s) which could bind to microtubules in a process dependent on ATP. SDS-PAGE and autoradiography analyses of the labeled polypep-tide composition of microtubule pellets revealed that only few polypeptides of LMP fraction associated to preformed microtubules. They represented about 8% of the total lZ5I-LMP radioactivity. In the absence of microtubules, no significant radioactivity could be observed as sedimentable material. In the presence of 1 mM ATP, the amount of one '251-labeled polypeptide bound to microtubules was decreased by about 40% (range 30-61), as estimated by densitometric analysis (Fig. 1). This effect of ATP was selective since ATP did not significantly decrease the radioactivity associated with the microtubule pellet. Since ATP in the same concentration range was shown to prevent the formation of microtubulelysosome complexes (24), this result strongly suggested that this polypeptide, the molecular mass of which was about 50 kDa, could account for the binding of intact lysosomes to microtubules.
Purification of the 50-kDa Protein by ATP Affinity Chromatography-Data presented above indicate that the 50-kDa polypeptide likely exhibits binding site(s) for ATP. We have thus tried to purify this protein by specific retention-elution on an ATP-agarose matrix. LMP were applied on the ATP affinity column. Several LMP were retained on the ATPagarose matrix. Most of them were found in the 0.1 M NaCl wash, an elution step carried out to eliminate proteins bound to the affinity matrix via electrostatic interactions. The elution of the column with 1 mM ATP yielded a protein fraction which contained a major polypeptide (Fig. 2 ) . This polypeptide co-migrated on SDS-PAGE with the 50-kDa component identified in Fig. 1. It is noteworthy that neither AMP nor ADP at the same concentration gave rise to the elution of this polypeptide. The 50-kDa polypeptide represented about 80% of the protein of the ATP-eluted fraction (BSA excluded) as judged by Coomassie Blue staining. From protein determinations and densitometric analysis, we estimated that about 1 pg of the 50-kDa polypeptide was obtained from 1 mg of total membrane protein.
Interaction of the 50-kDaATP Affinity-purified Polypeptide with Microtubules-The protein fraction, purified on the ATP affinity column, which mainly contained the 50-kDa polypeptide, was labeled with ['251]IODO-GEN and studied for its capacity to bind with microtubules. In the absence of ATP, 25 k 1.5% (n = 4) of the total radioactivity of the lZ5I-labeled ATP affinity-purified protein fraction were found in the microtubule pellet. No radioactivity was sedimentable when microtubules were not present. Upon addition of increasing concentrations of ATP (0.1-5 mM), the radioactivity bound to microtubules progressively decreased to a value of about 8%. The concentration of ATP which induced a 50% decrease of the binding was about 0.35 mM whereas 8045% of the maximal inhibition of the binding was obtained at an ATP concentration of 1 mM. PAGE and autoradiography analyses revealed that the radioactivity bound to the microtubule pellets corresponded to lZ5I-labeled 50-kDa polypeptide (Fig. 3C). The '251-labeled ATP affinity-purified protein fraction also contained a small amount of BSA used as carrier protein during the purification steps (see the legend of Fig. 2 ) and some other components in trace amounts (Fig. 3B). These contaminants did not bind to microtubules and remained in the supernatants, irrespective of the ATP concentration (Fig.  30). Increasing the concentration of ATP caused the progressive decrease of the amount of the '251-labeled 50-kDa polypeptide bound to microtubules (Fig. 3C) and a concomitant appearance of this molecular species in the microtubule assembly supernatant (Fig. 3 0 ) .
The amount of lZ5I-labeled 50-kDa polypeptide bound to microtubules as a function of ATP concentration was deter- From densitometric analysis, it was estimated that the 50-kDa polypeptide present in the ATP eluate (indicated by the arrow) represents 2-3 pg of protein. The other polypeptide seen in lane d is BSA used as carrier protein to coat the tubes and saturate the PD-10 column during the preparation of the ATP-eluted fraction; a significant loss of the 50-kDa protein was observed in the absence of carrier protein.
mined by counting the radioactivity of the 50-kDa labeled spots on polyacrylamide gels. The 50-kDa polypeptide contained 38% of the 12' 1 radioactivity incorporated into protein of the ATP-eluted fraction by the IODO-GEN labeling procedure. In the absence of ATP, 65-70% of the 12'I-labeled 50-kDa polypeptide were able to bind to microtubules (a value in complete agreement with the percentage of the radioactivity of the I2'I-labeled ATP affinity-purified protein fraction which was found associated with microtubules, i.e. 25%). Increasing ATP concentration (0-5 mM) progressively decreased the binding of the 12'I-labeled 50-kDa protein to microtubules from 65-70% to 15-20%. A representative experiment is shown in p a e l A of Fig. 3.
The binding of the 12'I-labeled 50-kDa polypeptide to microtubules was a very rapid process; the binding was maximum after 5 min of incubation before centrifugation and remained constant for at least 45 min. The addition of ATP after 15 min of incubation of microtubules with the 12'I-labeled ATP affinity-purified protein fraction induced the dissociation of the 12'I-labeled BO-kDa polypeptide from microtubules; binding values resulting from radioactivity measurements were the same when ATP was present since the beginning of the incubation or added after 15 min.
The binding of the '251-labeled BO-kDa polypeptide to microtubules was significantly decreased in the presence of purified unlabeled 50-kDa protein at a concentration of 1 pg/ ml; it was also inhibited by the addition of unfractionated LMP at a concentration expected to contain about 1 pg/ml 50-kDa polypeptide, but not by the LMP fraction lacking the 50-kDa polypeptide (removed by filtration on the ATP affinity column). Results of a representative experiment are reported in Table I. These data indicate that the interaction of the 50-kDa protein with microtubules is a saturable phenomenon.
The 50-kDa Protein Binds to Pure Tubulin in an ATPdependent Manner-In order to test whether the 50-kDa polypeptide could bind to microtubules through a direct interaction with tubulin dimers, we have studied the interaction of the 50-kDa polypeptide with pure tubulin immobilized on an agarose matrix. The ATP affinity-purified protein fraction was mixed with BSA as a carrier and applied to the tubulin affinity column. The results reported in Fig. 4 show that BSA was not retained by the tubulin column. On the contrary, the 50-kDa polypeptide interacted with the affinity matrix; it was

FIG. 3. ATP-dependent interaction of the SO-kDa protein with microtubules.
The 1261-labeled ATP affinity-purified protein fraction (0.2 pCi, 70 ng of protein) was incubated in buffer A containing BSA (10 mg/ml) for 30 min at 25 "C with preassembled microtubules (250 pg of microtubule protein) in the absence and in the presence of increasing concentrations of ATP. Microtubules were then sedimented, and the pellets were counted for radioactivity. Half of the microtubule pellets and a fortieth of the supernatants (to avoid the overloading of the gel with BSA) were analyzed by SDS-PAGE and autoradiography. To allow a quantitative comparison of autoradiograms, the time of exposure of gels corresponding to supernatants was 20 times higher than that of the gels of microtubule pellets. The '251-labeled 50-kDa protein was quantified in pellets and supernatants by counting the radioactivity of the 50-kDa polypeptide spots on polyacrylamide gels. Panel A, percent of 50-kDa protein found in the microtubule pellets versus ATP concentration. Panel B, autoradiogram showing the composition of the 1251-labeled ATP affinitypurified fraction. Panel C, autoradiographic analysis of microtubule pellets resulting from incubations in the presence of various concentrations of A T P 0, 0.1, 0.5, 1, 2, and 5 mM ATP from the lejt lane to the right lone, respectively. Panel D, autoradiographic analysis of the corresponding supernatants. not eluted by extensive washing with buffer T. The 50-kDa polypeptide was eluted from the tubulin affinity column with 1 mM ATP.

DISCUSSION
In our preceding papers, we reported the characteristics of the interaction phenomenon taking place in vitro between highly purified lysosomes and reconstituted microtubules (23,24). In this work, we report the identification of a lysosomal membrane protein (a) which exhibits an ATP-binding site and a tubulin-binding site, and ( b ) which is able to bind to preformed microtubules in an ATP-dependent manner. This protein has the property of an integral membrane protein since: (i) it remains associated to the lysosomal membrane through multiple steps of sedimentation along the lysosome purification procedure; (ii) it is not removed from the mem-  Fig. 3. Microtubules (200 pg of microtubule protein) were incubated with 1251-labeled ATP affinity-purified fraction (0.12 pCi, 40 ng) with or without unlabeled protein fractions as competitor in 400 pl of buffer A containing 10 mg/ml BSA. LMP* corresponds to the fraction of LMP which is not retained on the ATP column plus the NaC1-eluted fraction. The concentration of the 50-kDa protein in LMP (number in parentheses) was estimated from the yield of purification (see the second paragraph under "Results").  4. Interaction of the 60-kDa polypeptide with pure tubulin coupled to Affi-Gel 10. The ATP affinity-purified protein fraction obtained from 2 mg of membrane protein (corresponding to about 2 pg of the 50-kDa polypeptide) was mixed with 60 pg of BSA in a total volume of 500 pl of buffer T and applied on the tubulin affinity column (1 ml of gel). The flow-through was collected within 1 ml. After washing with 20 ml of buffer T, the column was eluted with 1 ml of 1 mM ATP in buffer T. The flow-through fraction and the ATP-eluted fraction were analyzed by SDS-PAGE. Lane a, 60 pg of BSA and 2 pg of the 50-kDa protein were loaded as a control; lane by ATP. b, fraction which flowed through the column; lane c, fraction eluted brane by high salt treatment (0.6 M NaCl); and (iii) it is solubilized by means of a detergent. The 50-kDa protein, inserted into the lysosomal membrane, may account by itself for the in vitro interaction of lysosomes with microtubules. It is noteworthy that the ATP-dependent inhibition of the interaction of the 50-kDa polypeptide with microtubules and the ATP-dependent inhibition of the formation of lysosome microtubule complexes appeared as closely related phenomena. Indeed, the two ATP effects were observed in the same ATP-dependent Binding of a Membrane Protein to Microtubules concentration range (between 0 and 2.5 mM) (Ref. 24 and Fig.  3).
The interaction of the 50-kDa polypeptide with microtubules is a saturable process. A concentration of 1 pg/ml unlabeled 50-kDa protein decreased the binding of lZ5I-labeled 50-kDa protein by about 60%. The apparent dissociation constant of the 50-kDa protein-microtubule complex should thus be close to or slightly lower than 2. lo-' M.
We propose that this 50-kDa protein could be involved in the establishment of stable associations of lysosomes to microtubules which could determine the intracellular distribution of lysosomes. The positioning of lysosomes inside the cell has indeed been reported to be highly dependent upon the microtubular organization (17, 30-34). Lysosomes are generally clustered close to the centrosome in a region of the cell which is characterized by a high density of microtubules.
The present results support and further substantiate our previous tentative model (24) in which we proposed that a given vesicle could be subjected to two different types of interaction with microtubules: stable association or motile association. The shift from one type of association to the other could be regulated. In this respect, it is interesting to notice that ATP (without hydrolysis) which causes the dissociation of lysosome-microtubule stable complexes (24) promotes the formation of "transient" vesicle-microtubule complexes and sustains vesicle translocation through its hydrolysis (2,6,19,21). A given regulatory factor could, in this way, control two complementary mechanisms ensuring the distribution and the dynamics of vesicles. Different aspects of our model are in keeping with previous findings from other laboratories. The requirement for a specific mechanism for organelle segregation around the centrosome has been put forward by Matteoni and Kreis (31). The activity of ATP as a dissociating agent for microtubule-secretory granule complexes was previously reported by Sherline et al. (35) and Suprenant and Dentler (36).
Motile microtubule vesicle associations and stable microtubule vesicle associations would take place through different molecular components. The former process involves translocator units such as kinesin (19-22) and microtubule-associated protein 1C (37, 38) corresponding to soluble or cytosolic protein components when not operative; the latter process could be based on anchoring factors, present on the vesicle membranes, such as the 50-kDa protein described here. The functions of these factors would be to mediate the adhesion and then the immobilization of the vesicles onto the microtubules. We recently proposed to name such proteins immotilins (17). There is now a need for similar studies on other intracellular vesicles to determine whether the 50-kDa protein or another protein with a similar function exists on the membrane of other organelles, which could be subjected to immobilization or movement depending on regulatory signals.