![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Vol. 16, Issue 9, 4183-4201, September 2005
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Department of Biochemistry and Molecular Biology, Louisiana State University Health Sciences Center, Shreveport, LA 71130;
Department of Microbiology and Immunology, Louisiana State University Health Sciences Center, Shreveport, LA 71130;
Feist-Weiller Cancer Center, Louisiana State University Health Sciences Center, Shreveport, LA 71130; and
¶ Department of Pediatrics, University of Arkansas for Medical Sciences, Little Rock, AR 72202
Submitted November 9, 2004;
Revised June 8, 2005;
Accepted June 13, 2005
Monitoring Editor: Keith Mostov
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
). The unique sequence, functional importance of the C terminus of GLUT1, and indirect evidence that it served as a regulatory target, prompted a search for proteins capable of interacting with this unique domain. We subsequently identified a protein termed GLUT1CBP that binds to the GLUT1 C terminus (Bunn et al., 1999). Its binding to GLUT1 is absolutely dependent upon the presence of the four C-terminal amino acids comprising the PDZ domain recognition motif in GLUT1. The protein is 333 amino acids in length, possesses a central PDZ domain required for the recognition of GLUT1, and also contains an N-terminal proline-rich region that may serve as a linkage target recognized by other proteins containing SH3 or WW domains (Figure 1). We also demonstrated that GLUT1CBP uses the PDZ domain to bind not only to GLUT1 but also to the actin cross-linking protein 
-actinin and the microtubule linked motor protein KIF-1B (Bunn et al., 1999). The presence of the PDZ domain and the ability of GLUT1CBP to interact with cytoskeletal elements places it among the important class of proteins containing PDZ domains that assemble membrane proteins and signaling cascades into areas of membrane specialization.
|
), GIPC (DeVries et al., 1998), SEMCAP-1 (Wang et al., 1999), NIP-1 (Cai and Reed, 1999), and synectin (Gao et al., 2000), to describe its interactions with the HTLV-1 virus-encoded oncoprotein TAX, the trimeric G protein-directed RGS-GAIP, neuronal semaphorins, neuropilins, and syndecans, respectively. See Table 1 for a more comprehensive list of such interacting proteins.
|
Historically, this protein was identified first as TIP-2, GIPC, and then GLUT1CBP. As GIPC use predominates the current literature and two homologues to GIPC have been identified recently (Kirikoshi and Katoh, 2002a, b; Saitoh et al., 2002), the term GIPC1 is synonymous to GLUT1CBP and is used interchangeably when referring to the GLUT1CBP protein described in our earlier work (Bunn et al., 1999). That study was the first to identify an interaction between the amino acid sequences flanking the PDZ domain in GLUT1CBP and the actin linked motor protein myosin VI, as well as to identify the ability of GLUT1CBP to dimerize (or form higher order oligomers) with other molecules of GLUT1CBP (Bunn et al., 1999). In that article, two of the three models proposed for the function of GLUT1CBP were to 1) cluster or anchor GLUT1 or any of the additional proteins recognized by the PDZ domain to actin filaments via -actinin and 2) link them to microtubule directed transport via the kinesin KIF-1B. These models were depicted (Bunn et al., 1999) for GIPC1 interactions with GLUT1, but potentially include any other protein that interacts with the PDZ domain of GIPC1. The data presented in this article address the third and most intriguing of the proposed models (Bunn et al., 1999) in which GIPC1 links PDZ-bound proteins (soluble, or integrated into vesicle membranes) to F-actin-directed transport via myosin VI (Figure 1).
GLUT1CBP(GIPC1) was the first potential adapter protein identified to interact with the unique myosin VI tail. This interaction is significant because mounting evidence now strongly implicates myosin VI as a motor protein involved in the trafficking of vesicle-bound proteins (for recent reviews, see Buss et al., 2002; Hasson, 2003). The postulated role of this motor in protein trafficking is unusual for myosin motors because myosin Vl moves toward the negative end of the actin filament (Wells et al., 1999). Thus, interactions of myosin VI with GIPC1 would provide cargo proteins bound to the PDZ domain with the capacity to move in a direction along actin fibers that opposes the positive end directed movement characteristic of other myosin motor proteins. Understanding when, where, and with which potentially numerous and important cargo proteins (Table 1) GIPC1 is associated when it is bound to or released from myosin VI could enhance our current understanding of the selectivity of the process of protein sorting within the cell.
The studies in this article address the proposed myosin VI-linked adapter function for GIPC1 by 1) further characterizing the domains required for interactions between GIPC1 and myosin VI; 2) demonstrating that GIPC1 can bind to the C-terminal domain of myosin VI and move as a complex coordinately with myosin VI along F-actin filaments within cellular extensions; and 3) illustrating that faulty trafficking of one of the many potential cargo proteins (GLUT1) occurs when interactions are disrupted between GLUT1 and the PDZ domain of GIPC1, or between GIPC1 and myosin VI.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Reagents
Purified His6-GLUT1CBP(GIPC1) was prepared as described previously (Bunn et al., 1999) after expressing PET30-GLUT1CBP in bacteria. The S-Tag, protease cleavage site, and cloning site encoded amino acids that join the N-terminal His6 tag sequences to the N terminus of GIPC1 are encoded by the PET30 vector and generate a His6-GLUT1CBP that migrates as a larger protein than native GIPC1 in SDS-PAGE. Rabbit anti-GLUT1 was raised against a KLH-peptide conjugate identical to the C-terminal 14 amino acids of GLUT1 and affinity purified on a peptide affinity column. Rabbit anti-GIPC1 was raised against a KLH-peptide conjugate identical to the first 12 amino acids of rat GIPC1 and purified on a peptide affinity column.
Interactions of the Purified Myosin VI C Terminus with Native and Purified GIPC1
For pull-down assays of endogenous native MDCK GIPC1, cell extracts were prepared by scraping confluent 10-cm dishes of MDCK cells into 0.5 ml/plate of ice-cold phosphate-buffered saline (PBS) containing 5% NP-40, 0.1 mM phenylmethylsulfonyl fluoride, and 40 µl/ml stock protease inhibitor cocktail (Roche Diagnostics, Indianapolis, IN). Lysates were treated eight times with 10 strokes each in a glass-teflon homogenizer with a total incubation time of 1 h on ice and then were centrifuged at 100,000 x g for 1 h at 4°C. One milliliter of clear supernatant was added to 40 µl of a 50% suspension of glutathione beads containing bound glutathione S-transfersae (GST) or GST-myosin VI(955-1254) and incubated for 3 h at 4°C. The beads were washed three times quickly and 1 x 10 min with 1 ml of PBS/5% NP-40, three times quickly and 1 x 10 min with 1 ml of PBS/0.1% NP-40, and once with Laemmli buffer lacking SDS and urea.
For pull-down assays of purified His6-GLUT1CBP(GIPC1), glutathione beads containing bound GST and GST-myosin VI(955-1254) were incubated for 1 h at 4°C with or without 20 µg of purified His6-GLUT1CBP(GIPC1) in buffer containing 10 mM HEPES, pH 7.5, 1 M NaCl, 1 mM EDTA, and 1 mM 
-mercaptoethanol. The beads were washed two times quickly and 2 x 10 min with 1 ml of PBS/0.1% NP-40, and then once with Laemmli buffer lacking SDS and urea.
Bound proteins were eluted from the washed beads in 70 µl of gel-loading buffer (Laemmli) containing 4% SDS/6 M urea by heating 8 min at 80°C. Twenty-five microliters of the supernatant was resolved by SDS-PAGE in 7.5% gels, and the separated proteins transferred to a polyvinylidene fluoride membrane in Tris-glycine buffer containing 20% (vol/vol) methanol. Membranes were blocked overnight in buffer containing 5% dry milk and 0.1% Tween 20 and then incubated with a 1:1000 dilution of either purified rabbit anti-N-terminal-GIPC1 or rabbit anti-C-terminal GLUT1 as indicated. Membranes were washed, and bound anti-GIPC1 or anti-GLUT1 was detected with horseradish peroxidase-conjugated monoclonal anti-rabbit IgG (Jackson ImmunoResearch Laboratories) and enhanced chemiluminescence. Images were captured using a Molecular Dynamics PhosphorImager and Image-Quant software.
For pull-down assays of labeled native GIPC1(1-333) and mutant GIPC1(1-300), [35S]methionine-labeled GIPC1(1-333) and GIPC1(1-300) proteins were produced by in vitro transcription and translation from the DNA templates pcDNA3.1(+)-GLUT1CBP(1-333) and pcDNA3.1(+)-GLUT1CBP(1-300) by using the TNT-coupled reticulocyte system (Promega, Madison, WI). Labeled proteins were incubated with beads containing bound GST, GST-myosin VI(955-1254), GST-myosin VI(1065-1160), or GST-myosin VI(1065-1254) as indicated for 1 h at room temperature in PBS containing 0.1% NP-40 (PBS-NP40) and then washed extensively in PBS-NP40. Bound labeled proteins were separated by SDS-PAGE and detected using a Molecular Dynamics PhosphorImager after exposure of the dried gel to a PhosphorImager screen.
Construction and Expression of Cyan Fluorescent Protein (CFP), Yellow Fluorescent Protein (YFP), and Green Fluorescent Protein (GFP) Fusion Proteins
YFP-GIPC1(1-333), YFP-GIPC1(1-249), CFP-myosin VI(955-1254), GFP-GLUT1, GFP-GLUT1
4, GFP-GLUT125, CFP-myosin VI(1-1254), and GFP-GLUT1synd4ctrm were constructed by inserting the cDNA fragments generated from restriction digests or PCR amplification of cDNA's encoding the appropriate amino acid sequences into pECFP, pEYFP, or pEGFP vectors (BD Biosciences Clonetech, Palo Alto, CA). The sequences derived from PCR-generated fragments were verified by sequencing. GIPC1 sequences were derived from pBSKII-GLUT1CBP (Bunn et al., 1999). GLUT1 sequences were from isolated mouse GLUT1 cDNA (Reed et al., 1990). Myosin VI(955-1254) sequences were derived from previously isolated rat myosin VI(955-1254) (Bunn et al., 1999) or from full-length pig myosin VI(1-1254), a generous gift of Dr. Tama Hasson (University of California, San Diego, CA). cDNAs were introduced into the appropriate cell lines using Lipofectamine 2000 according to manufacturer's protocols and analyzed after allowing 18 h or 2-4 d for protein expression as indicated.
Fluorescence Microscopy
For Figure 7A alone, a Bio-Rad MRC1000 confocal scope and Nikon plan apo 60x (numerical aperture [N.A.] 1.4) lens was used. Cells expressing GFP-GIPC1 were fixed for 1 h in 2% paraformaldehyde/phosphate-buffered saline and then permeabilized by incubation with 0.2% Triton X-100/phosphate-buffered saline for 0.5 h before incubation with rhodamine-phalloidin and mounting in Vectashield mounting media (Vector Laboratories, Burlingame, CA). A 488-nm excitation laser and 522/34 emission filter were used to collect the GFP-GIPC1 signal, and a 568-nm excitation laser and 605/32 emission filter were used to collect the rhodamine-phalloidin signal. All other microscopy was performed using a Bio-Rad Radiance 2000/AGR-3 (Q) confocal microscope equipped with a three-channel, three-detector system mounted on a Nikon TE 300 inverted microscope and three available lasers: 1) a 25-mW argon ion laser emitting at 457, 476, 488, and 514 nm; 2) a 1-mw green HeNe laser emitting at 543 nm; and 3) a red diode laser emitting at 638 nm. Images were acquired using plan apo 100x oil (N.A. 1.40) lens and Lasersharp I software with 32-bit acquisition, display, and three-dimensional volume rendering. Images of cells expressing native or mutant GFP-GIPC1 or GFP-GLUT1 were obtained by excitation at 488 nm with the argon laser and collecting emitted light through a HQ500LP emission filter. Live cell normal and time-lapse images were obtained using cells grown on temperature regulated T3-plates (Bioptechs, Butler, PA) that were washed and incubated in HEPES-glucose-saline solution (140 mM NaCl, 5.4 mM KCl, 1.3 mM CaCl2, 1.0 mM MgCl2, 25 mM HEPES, and 33 mM glucose, pH 7.4). The temperature was maintained at 25°C using a T3 temperature regulator (Bioptics). Simultaneous imaging of normal and mutant CFP-myosin VI and YFP-GIPC1 fusion proteins were collected using the strobe mode to block cross-channel signal bleed by sequentially exciting with either the 457- or 514-nm single laser excitation line and collecting with emission filters HQ485/30 for CFP emissions and HQ545/40 for YFP emissions. Three-dimensional reconstructions of z-sections or generation of videos from time-lapse images were obtained using MetaMorph imaging software (Universal Imaging, Downigntown, PA). GFP signals with rhodamine isothiocyanate (RITC)-dextran, LysoTracker red, or Cy3-EEA1 were collected by exciting with the 488 laser line and the HQ515/30 emission filter or with the 543 laser line and HQ590/70 emission filter, respectively, in strobe mode. GFP signals with LysoTracker Red and Cy5-EEA1 were collected with the 488 laser line and the HQ515/30 emission filter, the 543 laser line and the HQ590/70 emission filter, and the 637 laser line and the HQ660LP filter setting, respectively, in strobe mode.
|
4 in Figure 10 was obtained by collecting z-sections in 0.35-µm steps throughout the entire height of each cell. The total cellular fluorescence was obtained by integration of the fluorescent intensity of each section within a software defined boundary paralleling the outer contour of the cell using MetaMorph software. Likewise the total plasma membrane-associated fluorescence was determined by integrating and summing the fluorescence intensity within a 0.4-µm-thick boundary region paralleling the contour of the outer and inner surfaces of the plasma membrane for each z-section. The contribution of the basal plasma membrane fluorescence was obtained by integrating the z-section image centered within the plane of the membrane lying parallel to the coverslip. The calculated vertical resolution is 0.70 µm and therefore collects fluorescence within a 0.35-µm boundary either side of the basal membrane. Both total and plasma membrane fluorescence were corrected for endogenous fluorescence by integrating identical volumes in adjacent fields containing cells not expressing CFP-GLUT1 or CFP-GLUT14. The laser settings were held constant throughout the integration of all cells in Figure 10 and set initially to maintain all pixel intensities within the linear response range of the system. Statistical differences in the means were evaluated by a Student's t test.
|
Mapping of the Myosin VI and GIPC1 Interaction Domains Using the Yeast Two-Hybrid System
Plasmid vectors pGBT9 (BD Biosciences Clontech) or pGBDC3 were engineered to contain GLUT1CBP constructs fused to the Gal4 DNA binding domain (DBD) and are marked with TRP1 for nutritional selection in yeast. The plasmid vector pGADT7 containing rat myosin VI constructs fused to the Gal4 activation domain (ACT) were marked with LEU2 for nutritional selection in yeast. The procedures for selection and elimination of false positives are described in detail previously (BD Biosciences Clontech two-hybrid manuals). For Figure 3, plasmids encoding Gal4 DBD and ACT fusion proteins were cotransfected into yeast strains HF7c and the transformants were plated onto selection media lacking tryptophan and leucine. For each interaction tested, three or four viable colonies containing both plasmids were patched onto an identical plate and after 2-3 d were replicate plated onto selection plates lacking tryptophan and leucine or plates containing 30 mM 3-aminotriazole that lacked tryptophan, leucine, and histidine. Significant growth relative to the control plates indicated a positive interaction. For Figure 4, plasmids were introduced into yeast strain PJ69-4A (James et al., 1996) and selected in a similar manner on plates lacking tryptophan, leucine, and adenine. Appropriate controls were included for each fusion protein with the appropriate corresponding empty vector and a positive control to eliminate any self-activating sequences that would yield false positives.
|
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
) to define myosin VI C-terminal protein fragments and use the numbering for the full-length rat GIPC1 (amino acids 1-333) (Bunn et al., 1999).
A Characterization of the Interactions between Purified GIPC1 and Myosin VI Proteins and Their Endogenous Cellular Counterparts
A His6 fusion protein containing full-length GIPC1 when expressed in bacteria and purified by nickel affinity chromatography retains its ability to bind to the C terminus of GLUT1 expressed as a GST-fusion protein (Bunn et al., 1999). In addition, purified His6-GIPC1 covalently coupled to agarose beads can bind to and pull-down native GLUT1 present in detergent extracts of MDCK cells (Bunn et al., 1999). These previous studies establish the ability of GLUT1 to interact with purified GIPC1. Before defining the interacting domains between myosin VI and GIPC1, it was necessary to establish that both native and purified forms of GIPC1 and myosin VI were capable of forming a complex and that other unidentified proteins were not required for their interaction.
Purified His6-GIPC1 coupled to agarose beads is fully capable of binding tightly to native full-length myosin VI present in detergent extracts of MDCK cells (Bunn et al., 1999). Conversely, to demonstrate the ability of endogenous native GIPC1 to interact with the purified C terminus of myosin VI, MDCK cell extracts were incubated with GST or the purified GST-myosin VI(955-1254) fusion protein bound to agarose beads, and the associated proteins were separated by SDS-gel electrophoresis. Western blots of the gels probed with rabbit anti-GIPC1 antibody demonstrated that only GST-myosin VI beads interact with endogenous GIPC1 (Figure 2A, lane 4) present in MDCK cell extracts. This confirms the ability of the C-terminal amino acids 955-1254 of myosin VI to bind to GIPC1 protein synthesized and folded in its native environment. The purified GST-myosin VI(955-1254) fusion protein bound to agarose beads also interacts with purified His6-GIPC1(Figure 2B, lane 8). GST bound to agarose beads does not interact with either native GIPC1 (Figure 2A, lane 3) or purified His6-GIPC1 (Figure 2B, lane 7). These data establish the ability of the myosin VI C terminus to interact directly with His6-GIPC1, and they eliminate a concern that the interactions studied in the yeast two-hybrid or other cell systems might be mediated solely by an unknown protein or proteins capable of binding simultaneously to both GIPC1 and myosin VI.
|
GIPC1-dependent Formation of Complexes Containing Myosin VI(955-1254) and GLUT1
Agarose beads containing bound GST-myosin VI(955-1254) that pull-down GIPC1 (Figure 2A, lane 4) also pull-down cellular GLUT1 bound to endogenous GIPC1 (Figure 2C, lane 12) from MDCK cell extracts. Addition of purified His6-GIPC1 to the extract dramatically increases the amount of GLUT1 associating with GST-myosin VI(955-1254) (Figure 2C, lane 14), emphasizing the role of GIPC1 to mediate the interaction of GLUT1 with myosin VI. Agarose beads containing bound GST fail to interact with cellular GLUT1 in the absence (Figure 2C, lane 11) or presence (Figure 2C, lane 13) of added purified His6-GIPC1. These data establish the ability of GLUT1, GIPC1, and the myosin VI tail to form a trimeric complex.
Mapping of the Myosin VI and GIPC1 Interaction Domains
With the knowledge that mediating proteins were not required for binding, the mapping of the regions required for direct binding of GIPC1 to myosin VI was accomplished using a yeast two-hybrid system.
Our initial studies documented that removal of both N-terminal and C-terminal sequences flanking the PDZ domain blocked dimerization (or oligomerization) of GIPC1 with another molecule of GIPC1 in addition to destroying the ability to interact with myosin VI (Bunn et al., 1999). Previously, it was demonstrated (Gao et al., 2000) that removal of N-terminal and not C-terminal flanking sequences of synectin (GIPC1) destroyed the ability of GIPC1 to form dimers or higher oligomers. However, the data did not determine whether the dimerization and myosin VI interaction domains resided together in the N-or C-terminal regions of GIPC1 or whether they were located in separate regions.
To resolve this issue, the experiment presented in Figure 3 was performed. A diagram of each of the Gal4 binding domain (GBD)-GIPC1 fusion proteins tested for their interaction with the activation domain fusion proteins to myosin VI(955-1254) and GIPC1(1-333) are presented in the left column. Strong interactions (+) are indicated by rapid growth on -Trp, -Leu, -His plates containing 30 mM aminotriazole. The results demonstrate that constructs lacking the N-terminal flanking sequences (row 2) lose the capacity to interact with GIPC1(1-333) but maintain the ability to interact with myosin VI(955-1254). The converse is true for GIPC1 constructs lacking the C-terminal flanking sequence (row 3). GIPC1 constructs missing both N-terminal and C-terminal flanking sequences lose both the ability to dimerize and interact with myosin VI (row 4). The microtubule based motor protein KIF-1B and -actinin were included as positive controls, and consistent with their known ability to bind to the PDZ domain of GIPC1 (Bunn et al., 1999), interacted with all forms of GIPC1 tested. Thus, the stretches of amino acid sequences required for either dimerization or binding to myosin VI are separate and reside in the N-terminal and C-terminal regions of GIPC1, respectively.
With the knowledge that the myosin VI binding domain resides in the C-terminal half of the GIPC1 protein and is separate from the dimerization domain, a more thorough characterization of the GIPC1 and myosin VI interaction domains was undertaken. A diagram of each of the Gal4 DBD-GIPC1 fusion proteins and ACT-myosin VI fusion proteins tested is presented in Figure 4. Strong interactions tested in a more stringent system (yeast strain PJ69-4A) are indicated by rapid growth on -Trp, -Leu, -Ad plates and are designated by a "+," whereas no interactions (no growth) are indicated by a "-". The rat myosin VI fragment we isolated during the initial two-hybrid screen for proteins interacting with GLUT1CBP(GIPC1) extends from amino acid 955 to the C-terminal amino acid 1254 (using numbering for pig myosin VI (Hasson and Mooseker, 1994). This includes a portion of the conserved coiled-coil domain and the entire unique C-terminal domain of myosin VI (Figure 1). This myosin VI isoform is identical in this region to pig myosin VI and lacks both the short and long inserts that have been reported to be present in less common isoforms of myosin VI (Buss et al., 2001).
Using this approach, one binding domain for GIPC1 was identified in the myosin VI tail. The results of experiments defining the minimal myosin VI domain capable of interacting with full-length GIPC1 are illustrated by row 1 in Figure 4. A region as small as myosin VI(1055-1169) interacts well with full-length GIPC1(1-333), and a smaller myosin VI(1065-1160) segment also binds, but less effectively as evidenced by marginal growth (±) of yeast expressing these constructs. Thus, these data define a minimal interaction domain within the myosin VI tail that is marked by the light-gray horizontal rectangle (Figure 4) and the solid white box (Figure 1). In agreement with this assignment, GIPC1(1-333) fails to interact with a myosin VI fragment containing a portion of the coiled-coil region (955-1066) or with more C-terminal regions such as myosin VI(1127-1254), (1162-1254), and (1182-1254) shown in Figure 4, row 1.
Rows 2-6 in Figure 4 depict the N-terminal truncations of GIPC1 used to determine the N-terminal boundary of the GIPC1 domain (D#1) that recognizes myosin VI. Fusion protein fragments as small as GIPC1(261-333) retain binding activity to all myosin VI constructs containing amino acids 1065-1160 and indicate amino acids 1-260 of GIPC1 are not required for binding of the C-terminal portion of GIPC1 to myosin VI. DBD fusion proteins containing further N-terminal truncations equal to or smaller in length than GIPC1(279-333) inexplicably become self-activating (SA; Figure 4, row 6) and could not be evaluated in this system.
Experiments depicted by rows 7-10, column 1, in Figure 4 represent the experiments to define the C-terminal boundary of the myosin VI binding domain D#1 in GIPC1. No interactions were detected when the binding to the myosin VI(955-1254) fragment containing the coiled-coil domain was tested using a truncated GIPC1(1-300) or any other constructs missing more of the GIPC1 C terminus. Thus, when evaluated in the yeast system, the region defined by amino acids 261-333 of GIPC1 contains amino acid sequences critical for interactions of GIPC1 with the myosin VI(955-1254) tail. These data demonstrated that the first of two myosin VI binding domains in GIPC1 resides within the region marked by the light gray vertical rectangle and designated D#1 in Figure 4.
An unexpected finding in these studies was the existence of a potential second myosin VI binding domain (D#2, Figure 4) located within the C-terminal half of the PDZ domain in GIPC1. This domain is revealed (Figure 4, row 7) only when the binding of GIPC1(1-300) is tested with myosin VI fusion proteins that lack the coiled-coil domain, but retain amino acids 1065-1160; for example, the myosin VI(1065-1204) fragment (Figure 4, column 5, row 7). Apparently, the conformation of the myosin VI(955-1254) fragment containing a portion of the coiled-coil domain blocks domain D#2 interaction with myosin VI.
To locate the C-terminal boundary of the second binding domain D#2, additional C-terminal amino acids were removed from GIPC1(1-300) to form the GIPC1(1-277), (1-249), and (1-231) fusion proteins. Binding of each of these proteins was tested with myosin VI constructs lacking the coiled-coil domain. All of the GIPC1 truncations (Figure 4, rows 8-10) produced a marginal interaction (marginal growth) with both the myosin VI(1065-1160) and (1055-1169) fusion proteins, but they maintained a strong interaction with each of the myosin VI(1065-1204) and (1065-1254) fusion proteins.
Experiments to define the N-terminal boundary of domain D#2 are presented in rows 11-13. Results using successive N-terminal truncation mutants of GIPC1(1-300) demonstrate that both GIPC1(173-300) and GIPC1(189-300) are unable to bind to the shorter myosin VI(1065-1160) and myosin VI(1065-1169) fragments, but they retain their ability to bind to myosin VI(1065-1254) and myosin VI(1065-1204) constructs that contain additional C-terminal amino acid residues within the myosin VI tail (rows 11 and 12). Further truncation to form the GIPC1(214-300) fragment (row 13) abolishes binding to any of the shorter or longer myosin VI fragments tested.
The data from rows 7-13 indicate that the smallest region comprising this second binding domain should reside within amino acids 189-231 of GIPC1. This short fragment, however, must not contain sufficient structural information as it fails to bind to any of the myosin VI fragments tested (row 16). The shortest interacting sequence capable of binding to myosin VI was GIPC1(173-231) (Figure 4, row 15) and defines the potential second interaction domain in GIPC1 marked by the light gray vertical rectangle and labeled D#2. The interaction pattern exhibited by domain D#2 in GIPC1 suggests that it binds to a myosin VI region that either overlaps with the myosin VI region recognized by GIPC1 domain D#1, or is located slightly more toward the C terminus of myosin VI. This is illustrated in Figure 4, row 15, by the ability of fusion proteins containing only the D#2 domain to interact with myosin VI(1065-1204) but not myosin VI(1065-1160) or myosin VI(1055-1169).
|
Analysis of Intracellular Interactions between Full-Length GIPC1(1-333) or GIPC1(1-249), and Myosin VI Using a Mammalian Cell System
To analyze the binding interactions in a normal environment in which these proteins exist, CHO cells were used to process and fold native and mutant proteins in the most natural environment (an intact mammalian cell) to test for myosin VI and GIPC1 domain interactions. The cells are convenient to use for binding assays because the exogenous YFP-tagged native GIPC1 and GIPC1 mutants form punctuate intracellular distributions when expressed alone in CHO cells. Under the conditions of this assay, exogenous native and mutant myosin VI proteins conveniently exhibit a diffuse distribution unless coexpressed with an interacting form of GIPC1, in which case they will redistribute and colocalize with the punctuate GIPC1 proteins. To illustrate the behavior of the assay with full-length GIPC1 and myosin VI proteins, a CFP fusion protein to the N terminus of full-length pig myosin VI and a YFP fusion protein to the N terminus of full-length GIPC1(1-333) were constructed and expressed in CHO cells. When cells expressing only CFP-myosin VI(1-1254) were analyzed (Figure 6A, solid arrows), CFP-myosin VI was excluded from the nucleus, but it was otherwise distributed diffusely throughout the cell. Although YFP-GIPC1 is a soluble protein, the native and mutant fusion proteins nevertheless exhibit a punctate distribution (A-E, middle), presumably by binding to the termini of integral membrane proteins associated with membranous and vesicular structures. When YFP-GIPC1 and CFP-myosin VI were expressed in the same cell (A, dotted arrows), the normal diffuse distribution of CFP-myosin VI shifted to the plasma membrane and to internal vesicular-like structures that colocalize with YFP-GIPC1 (Figure 6A merged). The redistribution and colocalization of CFP-myosin VI with YFP-GIPC1 is more apparent in a separate cell viewed at a higher magnification (Figure 6B).
|
In experiments that are directly analogous to the yeast two-hybrid tests, full-length YFP-GIPC1(1-333) was coexpressed with the truncated CFP-myosin VI(955-1254) containing the C-terminal domain of myosin VI and a portion of the coiled-coil domain. As expected, the redistribution of the myosin VI fusion protein to a punctate pattern that colocalizes with YFP-GIPC1(1-333) was observed (Figure 6D, merged). However, as noted in the yeast studies, when truncated YFP-GIPC1(1-249) containing the D#2 domain and missing the D#1 domain was coexpressed with truncated CFP-myosin VI(955-1254), the redistribution and colocalization of CFP-myosin VI(955-1254) with YFP-GIPC1(1-249) did not occur (Figure 6E). These experiments establish the importance of D#1 for proper interactions of GIPC1 with myosin VI in the intact mammalian cell. When it is missing, GIPC1 does not bind to myosin VI(955-1254) and more importantly does not interact with full-length myosin VI. Domain D#2 alone simply is not sufficient.
Demonstration of Actin-dependent and Microtubule-independent Movement of GIPC1
If GIPC1 functions as an adapter molecule that links cargo proteins to myosin VI movement as proposed, then it should be possible to demonstrate motion of GIPC1 that is actin dependent and characteristic of motion catalyzed by a myosin VI motor protein.
While analyzing the characteristic distributions of GFP-GIPC1(1-333) in a number of cell lines, it was noted that GFP-GIPC1 is often concentrated in small structures or aggregates within tubular extensions from the cell that seemed similar to retraction fibers and in many cases, filopodia. Small GFP-GIPC1-containing structures were evident in tubular extensions protruding from both the leading and trailing edges of a migrating cell. Because retraction fibers and filopodia often are difficult to discern from shape alone and can interconvert (Litman et al., 2000; Svitkina et al., 2003), these processes are referred to collectively as cellular extensions (Svitkina et al., 2003) with the realization that we are analyzing potentially both types of structures.
Green fluorescent GIPC1 particles can be observed throughout the length of such cellular extensions (Figure 7A, arrows), when CHO cells expressing GFP-GIPC1(1-333) are fixed and the F-actin is counterstained with rhodamine phalloidin. Phalloidin staining (red) of the F-actin filaments is evident throughout the entire length of these cellular extensions. Furthermore, when analyzing time-lapse images of live cell preparations, GFP-GIPC1(1-333) particles were observed to stream continuously toward the cell body. This movement toward the cell body was observed in cellular membrane extensions from both the advancing (Ad) and trailing (Tr) edge of motile cells (Figure 7B). This is more readily observed in time-lapse images of live cells presented in the supplemental movie (Fig 7B.mov) from which the representative single frame image shown in Figure 7B was taken. In this movie, membrane protrusions are being extended on the advancing edge (Ad) and are being retracted from the trailing edge (Tr) of the cell, whereas the direction of GFP-GIPC1 particle movement from either the trailing or leading edge is always toward the cell body and therefore, importantly, toward the negative end of the F-actin filament. This is consistent with its movement being catalyzed by a myosin VI motor that has been shown to move toward the negative end of the actin filament (Wells et al., 1999).
Because GIPC1 is known to associate with the microtubule-associated kinesin motor protein KIF-1B (Bunn et al., 1999) (Figure 3), and we cannot rule out the possibility that microtubule filaments also might be present within many of the cellular extensions being analyzed, it was necessary to test whether the observed GIPC1 movement was dependent upon the integrity of actin or microtubule structures within these cellular extensions. Time-lapse images were collected to establish the rate of GFP-GIPC1 movement along the extensions of living CHO cells. Then, colchicine or nocodazole was added at concentrations known to disrupt microtubules, or cytochalasin D or latrunculin B was added at concentrations known to disrupt F-actin filaments. The acquisition of images was continued after the addition of each agent to assess its effect on GFP-GIPC1 particle movement. The right panels in Figure 8 show single-frame images of a representative cell used for each experiment. The associated videos are provided as Fig 8a, b, c, and d.mov in the supplement. Individual particles were tracked using MetaMorph software to determine the distance each traveled along the cellular extensions during the interval between successive image acquisitions. For each particle tracked, the data for the distance traveled versus time were plotted (Figure 8, left). The slope of this plot provides an estimate of the average speed of each particle before and after the addition of each disruptive agent.
|
The addition of colchicine or nocodazole to disrupt microtubules does not stop or retard the movement of the GFP-GIPC1 particles (Figure 8, A and B). Colchicine in fact seemed to accelerate their movement by at least twofold. In contrast, cytochalasin D, after a small delay, and latrunculin, immediately, stopped particle movement after disrupting F-actin structure (Figure 8, C and D). This behavior is consistent with an actin-dependent and myosin VI-catalyzed movement of GIPC1 as proposed. The videos provided for each of the experimental drug treatments more emphatically demonstrate the direction of movement and the dependence of GFP-GIPC1 movement on intact F-actin (Supplemental Figures 8A-D.mov). The rate of movement of GFP-GIPC1 particles before drug treatment was estimated from measurements taken for 24 particles from three different cells. The mean velocity ± SE was 0.94 ± 0.024 µm/min. The maximum velocity over any given 30-s interval was 5.7 µm/min.
Although the direction of movement of the GFP-GIPC1 complexes is consistent with myosin VI-catalyzed motion, it is also consistent with the movement that could arise from actin tread-milling. To further substantiate that the observed motion arises from association with a functional myosin VI molecule, cells expressing both YFP-GIPC and a CFP-myosin VI(955-1254) mutant lacking the motor domain were also analyzed. As presented in the left panel of Figure 8E, after 10 min the majority of the particles have shown little to no movement toward the cell body. One particle has moved at most 1 µm closer after 10 min, compared with the 9-10 µm traveled in 10 min by those particles tracked before drug treatment in Figure 8B. The average velocity and SE obtained from measurements of a total of 53 YFP-GIPC1 particles from 12 different cells was -0.009 ± 0.001 µm/min (-signifies movement away from the cell body) in the presence of CFP-myosin VI(955-1254), whereas in its absence the average rate is nearly 1 µm/min. Thus, for all practical purposes motion toward the cell is stopped by disrupting potential interactions of GIPC1 with native myosin VI. This is also illustrated in the supplemental video Fig 8E.mov illustrating particle movement in the representative cell presented in the right panel of Figure 8E.
Occasionally, a particle can be found that is moving more rapidly away from the cell, for example the track with the negative slope in Figure 8E (velocity -0.25 µm/min). This suggests that other forces might function to move GIPC1 toward the tips of the cell extensions when myosin VI interactions are disrupted. This possibility is also supported by the observation that some particles reverse their direction and begin moving outward when F-actin structure is disrupted by cytochalasin D or latrunculin (see the left panels of Figure 8, C and D, and movies Fig 8C.mov and Fig 8D.mov).
Assays for Coordinated Movement of GIPC1-Myosin VI Complexes
This experiment was designed to demonstrate that GIPC1 movement within cellular extensions is not restricted to CHO cells and to test the model for myosin VI-assisted GIPC1 movement by determining whether GIPC1-myosin VI complexes could be observed to move together along an actin network within cellular extensions. This was accomplished by cotransfecting 293 cells with CFP-myosin VI(1-1254) and YFP-GIPC1(1-333). Then, 2 d after transfection, time-lapse images were collected with the confocal microscope using the same settings as before (Figure 6) to separate CFP and YFP emission signals. Particles of CFP-myosin VI in the CFP channel (Figure 7C, top left) and YFP-GIPC1 in the YFP channel (top right) were located in identical positions along the cellular extensions (marked by arrows). The merged image (Figure 7C, bottom left) of the CFP-myosin VI (green) and YFP-GIPC1 (red) signals supports their colocalization (yellow) and coordinated movement as a complex toward the cell body. An analysis of their rate of movement using MetaMorph software indicates that each of the three GIPC1-myosin VI complexes was moving at a different rate (Figure 7D), whereas the CFP-myosin VI and YFP-GIPC1 components of each complex moved together and did not separate. The movie (Figure 7C.mov) provided in the supplement more readily illustrates this point. Measurement of the movement of 22 separate particles from five different cells gave a mean velocity ± SE of 0.76 ± 0.037 µm/min for the CFP-myosin VI/GIPC1 particle complexes. The maximum rate observed within the 20-s sampling interval was 5.0 µm/min.
In a recent report (Rustom et al., 2004), structures arising from extending filopodia that establish an actin containing bridge between two well separated cells were reported to serve as conduits for unidirectional transport of small membrane organelles between the two cells. These structures were named nanotubes. A bridge between 293 cells identical in appearance to the nanotubes described in that report was noted in our study of myosin VI/GIPC1 complex movement (Figure 7E). Several CFP-myosin VI/YFP-GIPC1 complexes were noted within this bridging structure. The complex marked by the arrow was moving in the direction indicated at a rate of 1.2 µm/min.
Disruption of GLUT1 Interactions with GIPC1 Induces Faulty Trafficking of the Cargo Protein GLUT1
The studies described above support the proposed adapter function for GIPC1 to link PDZ bound proteins to actin-dependent movement catalyzed by myosin VI. As further evidence of the potential trafficking function for GIPC1, we have analyzed the processing of one of its potential cargo proteins, GLUT1, after disrupting interactions between the two proteins.
MDCK cells were used in these experiments because GLUT1 is efficiently targeted to the basolateral membrane (Pascoe et al., 1996), and in addition to GLUT1, they normally express both GIPC1 and myosin VI. If the interaction of GLUT1 with the GIPC1-myosin VI complex is important for the delivery of GLUT1 to this region of the plasma membrane, then disrupting the interactions should alter proper GLUT1 movement within the MDCK cell.
To test this hypothesis, a GFP-fusion protein to GLUT1 was expressed in polarized MDCK cells to confirm that the N-terminal fluorescent fusion protein would not interfere with normal GLUT1 targeting to the basolateral membrane. After 2 d, reconstructed X-Z sections from confocal scans of MDCK cells expressing GFP-GLUT1 (Figure 9C) demonstrate the predicted basolateral distribution for GFP-GLUT1. This is also evident in the three-dimensional reconstruction of the GFP-GLUT1 protein distribution provided as a supplemental video (Fig 9A.mov) that is represented by the single-frame image (Figure 9A). The majority of the protein is confined to the basolateral membrane and very little transporter is concentrated in either the apical membrane or in any large vesicle populations within the cell (Figure 9C).
|
4 mutant that no longer binds to the PDZ domain of GIPC1 (Bunn et al., 1999). When cells expressing this mutant form of GLUT1 are analyzed after the same interval of expression used for GFP-GLUT1 (2 d), a significantly higher percentage of the total cellular transporter mass seems to accumulate in an abundant vesicle population within the cell (Figure 9, B and D). This is evident when comparing both the three-dimensional projections as well as the X-Z section images of cells expressing "native" GFP-GLUT1 (Figure 9A, 9A.mov and C) or mutant GFP-GLUT14 (Figure 9B, 9B.mov and D).
To determine whether the abnormal concentration of GLUT14 protein in internal vesicle pools observed 2 d after transfection was a transient phenomenon, the protein distribution of native GFP-GLUT1 and mutant GFP-GLUT14 proteins were analyzed at 2, 3, and 4 d after transfection. The level of GFP-GLUT1 present in internal structures (Figure 9G) remains very low relative to the amount present in the plasma membrane. In addition, GFP-GLUT1 remains effectively excluded from the apical domain throughout the 4-d period of expression. In contrast, the level of GFP-GLUT14 present in internal membranous structures remains high relative to that present in the plasma membrane throughout the same interval. The majority of the GFP-GLUT14 in the plasma membrane, however, was still restricted to the basolateral domain of the cell (Figure 9G, day 4). Small amounts of GFP-GLUT14 protein are evident near the apical region of the cell throughout the interval of expression, but it is unclear whether this small signal resides in the apical membrane or may actually represent a small portion of the enlarged vesicle pool of mutant transporter residing below the apical membrane surface.
These data strongly suggest that relative to cells expressing native GLUT1, cells expressing the mutant GLUT14 exhibit some defect in processing of GLUT14 that persists throughout the entire 4-d period of expression. It is also clear that the C-terminal four amino acids are not required for newly synthesized GLUT14 to reach the plasma membrane. Furthermore, the presence of the C-terminal four amino acids of GLUT1 does not seem necessary to target GLUT1 to the basolateral membrane as their loss does not result in faulty targeting of the transporter to the apical membrane.
An alternative interpretation for the inefficient trafficking
