Intracellular Trafficking and Membrane Targeting Mechanisms of the Human Reduced Folate Carrier in Mammalian Epithelial Cells* 210

The major pathway for cellular uptake of the water-soluble vitamin folic acid in mammalian cells is via a plasma membrane protein known as the reduced folate carrier (RFC). The molecular determinants that dictate plasma membrane expression of RFC as well as the cellular mechanisms that deliver RFC to the cell surface remain poorly defined. Therefore, we designed a series of fusion proteins of the human RFC (hRFC) with green fluorescent protein to image the targeting and trafficking dynamics of hRFC in living epithelial cells. We show that, in contrast to many other nutrient transporters, the molecular determinants that dictate hRFC plasma membrane expression reside within the hydrophobic backbone of the polypeptide and not within the cytoplasmic NH2- or COOH-terminal domains of the protein. Further, the integrity of the hRFC backbone is critical for export of the polypeptide from the endoplasmic reticulum to the cell surface. This trafficking is critically dependent on intact microtubules because microtubule disruption inhibits motility of hRFC-containing vesicles as well as final expression of hRFC in the plasma membrane. For the first time, these data define the mechanisms that control the intracellular trafficking and cell surface localization of hRFC within mammalian epithelia.

The major pathway for cellular uptake of the watersoluble vitamin folic acid in mammalian cells is via a plasma membrane protein known as the reduced folate carrier (RFC). The molecular determinants that dictate plasma membrane expression of RFC as well as the cellular mechanisms that deliver RFC to the cell surface remain poorly defined. Therefore, we designed a series of fusion proteins of the human RFC (hRFC) with green fluorescent protein to image the targeting and trafficking dynamics of hRFC in living epithelial cells. We show that, in contrast to many other nutrient transporters, the molecular determinants that dictate hRFC plasma membrane expression reside within the hydrophobic backbone of the polypeptide and not within the cytoplasmic NH 2 -or COOH-terminal domains of the protein. Further, the integrity of the hRFC backbone is critical for export of the polypeptide from the endoplasmic reticulum to the cell surface. This trafficking is critically dependent on intact microtubules because microtubule disruption inhibits motility of hRFC-containing vesicles as well as final expression of hRFC in the plasma membrane. For the first time, these data define the mechanisms that control the intracellular trafficking and cell surface localization of hRFC within mammalian epithelia.
Folate is necessary for the synthesis of precursors of nucleic acids, initiation of protein synthesis in mitochondria, and metabolism of certain amino acids (1,2). Deficiency of this essential micronutrient leads to a variety of abnormalities including derangement of one-carbon metabolism and inhibition of growth. Humans and other mammals do not possess the ability to synthesize folate and must obtain this vitamin from exogenous sources by intestinal absorption. Absorption occurs from both dietary sources in the small intestine and from bacterially synthesized folate in the large intestine (3,4). Thereafter, folate is distributed into different body compartments where it is taken up by individual cell types for use in different metabolic reactions. Many previous studies have shown that the main folate uptake pathway in cells occurs via a specialized carrier-mediated mechanism, the reduced folate carrier (RFC 1 ; in humans, known as hRFC) (4 -6). The RFC is also involved in intestinal absorption of folate from the small and large intestine (5). The functional properties of this transport pathway have been well characterized (4 -6) and its molecular identity established after cloning of its cDNA (7)(8)(9) and characterization of its flanking regulatory regions (4).
In contrast, comparatively little is known about the mechanisms that control the intracellular trafficking of the RFC protein and its appropriate expression in the plasma membrane, topics of particular importance for transport of folate by mammalian epithelial cells. Genetic defects in such folate transport processes have been reported previously; however, little is known about the exact sites of these defects (10). To address these questions, we used high resolution confocal imaging techniques to monitor the distribution and transport of hRFC fusion proteins tagged with the enhanced green fluorescent protein (EGFP) in living mammalian epithelial cell lines. By comparing the expression patterns of nine truncated hRFC mutants, we show that targeting of hRFC to the plasma membrane depends on the integrity of the membrane-spanning "backbone" of the protein, rather than on targeting sequences in the NH 2 -or COOH-terminal cytoplasmic regions. Furthermore, by using video-rate confocal microscopy, we resolved the intracellular trafficking dynamics of hRFC-containing vesicles and demonstrate that delivery of hRFC-EGFP to the cell surface is critically dependent on intact microtubules but not microfilaments. Taken together, these results highlight how the hRFC polypeptide sequence is targeted and delivered to the cell surface within a physiologically relevant context, and they define the regions of the hRFC protein in which mutations would impair folate transport by disrupting trafficking and targeting mechanisms.

Construction of hRFC-EGFP and Truncated Constructs-cDNA en-
coding the open reading frame of the full-length hRFC and various truncated constructs (Fig. 1) were generated by PCR using the primer sequences and combinations shown in Table I. PCRs were carried out as described previously (11), and the resulting products were gel isolated and ligated together with the EGFP coding sequence digested from the EGFP-N3 vector. This method generated in-frame fusion proteins under the control of the human cytomegalovirus promoter. The nucleotide sequence of each resulting construct was confirmed by sequencing (SeqWright, Houston, TX and Laragen, Inc., CA).
Cell Culture and Transient Transfection-Human embryonic kidney (HEK-293) cells, human duodenal derived intestinal epithelial (HuTu-80) cells, and Madin-Darby canine kidney cells were maintained in minimum Eagle's medium containing 10% fetal bovine serum. Human colon adenocarcinoma (Caco-2) cells were maintained in Dulbecco's modified Eagle's medium (20% fetal bovine serum) and human colonic (NCM 460) cells in Ham's F-12 medium (20% fetal bovine serum). All media contained glutamine, sodium bicarbonate, penicillin, and streptomycin. For transient transfection, cells were grown on poly-D-lysinecoated cover slips cemented in holes at the bottom of sterile Petri dishes (MatTek, MA) and were transfected at ϳ90% confluence with plasmid DNA complexes using LipofectAMINE 2000 (Invitrogen). For all transfection protocols, conditions were optimized to minimize the amount of DNA used (typically ϳ0.6 g of cDNA of the indicated construct and less than recommended by the manufacturers) to minimize any nonspecific effects of fusion protein overexpression. Monolayers were subsequently imaged using confocal microscopy at indicated times after transfection (typically ϳ48 h).
Confocal Imaging of hRFC Constructs-Cell monolayers were monitored for hRFC-EGFP expression using a custom built laser scanning confocal microscope (12) or a video-rate scanning confocal microscope (13). Both instruments were based on Olympus IX70 inverted microscopes fitted with 40ϫ, NA 1.35 oil immersion objectives. EGFP, EYFP, DsRed, and FM4-64 were all excited using the 488 nm line from an argon ion laser, and emitted fluorescence was monitored using a 530 Ϯ 20 nm band pass filter (EGFP, EYFP) or 650LP filter (DsRed, FM4-64). For slow frame scanning, confocal images were obtained by scanning either laterally (top view; x-y scans) or axially (side view; x-z scans) across the specimen.
For video-rate imaging experiments, images were recorded at 30 Hz (one frame every 33 ms) from confocal sections with the microscope focused about 2 m above the coverglass within cells maintained at 22 or 37°C. Data were digitized using the stream acquisition function of the Metamorph processing package (Universal Imaging, Downingtown, PA). The resulting image stacks (x-y time) were then smoothed using a two-frame average to yield a final image stack at 66-ms frame intervals. The motions of individual vesicles were tracked using the point-to-point tracking function in Metamorph, and the resulting paths were exported to the Origin graphing package for analysis. QuickTime videos of image sequences are appended as supplementary material.
Flow Cytometry-HEK-293 cells were used for these experiments because their high (ϳ40%) efficiency of transfection allowed counting of large cell populations (15,000 cells for each construct, three experiments). Monolayers of HEK-293 cells were transfected in situ within individual T75 tissue culture flasks using ϳ4 g of cDNA encoding individual constructs. Cells were trypsinized and resuspended in Ca 2ϩfree phosphate-buffered saline, and populations of Ͼ15,000 transfected cells were measured for intensity of fluorescence emission (488 nm excitation, 530 Ϯ 15 nm emission) using a FACStation fluorescence analysis system (Becton Dickinson Immunocytometry Systems).

RESULTS
Cellular Distribution of hRFC-EGFP-To visualize the targeting of hRFC in mammalian cells, we transiently transfected a variety of epithelial cell lines with cDNA encoding the fulllength fusion construct (hRFC-EGFP) and analyzed the resulting fluorescence distribution by confocal microscopy. In HuTu-80 cells (human duodenal epithelium), hRFC-EGFP expression was evident at the cell surface and in cellular processes extending from the cell membrane as well in a population of discrete, intracellular vesicular structures ( Fig. 2A, see also later). This distribution of fluorescence contrasted markedly with HuTu-80 cells expressing EGFP alone, in which the entire cytosolic volume was fluorescent (Fig. 2B). These contrasting distributions were also evident in axial (x-z) sections of cells expressing hRFC-EGFP (Fig. 2C) or EGFP alone (Fig. 2D), which suggested that hRFC-EGFP was targeted to all plasma membrane domains of the cell. The localization of hRFC-EGFP to the plasma membrane was not unique to HuTu-80 cells because a similar targeting was evident from lateral (x-y) sections images in a variety of other mammalian epithelial cell lines including HEK-293, Caco-2, Madin-Darby canine kidney, and NCM-460 cells (Fig. 2E).
Identification of Polypeptide Domains Important for hRFC Expression and Targeting-To determine the domains within the hRFC protein which are important for its correct trafficking and targeting we compared the cellular distribution seen with the full-length hRFC-EGFP with that observed with the series of hRFC mutants diagrammed in Fig. 1.  Fig. 3D). Expression of hRFC[28 -591]-EGFP was evident at the cell surface as well as throughout the intracellular compartments, whereas hRFC[302-452]-EGFP was restricted to intracellular membranes soon after transfection and thereafter appeared in the cytoplasm. This latter profile is suggestive of a poorly tolerated protein structure that is actively targeted for degradation (14).
To quantify the relative amounts of hRFC protein expressed in the plasma membrane versus that in other cellular compartments, we imaged transfected cells that were costained by   Table 1 shows the sequence and combination of primers used for the preparation of the different constructs by PCR. Restriction sites for BamHI (boldface text) and XhoI (underlined) were added to specific hRFC primers to assist cloning into the pEGFP-N3 vector.  In addition to the differing cellular localization of hRFC-EGFP proteins expressed by the different constructs, it was also apparent that the total amount of protein expressed varied greatly among constructs. For example, although hRFC-EGFP and hRFC[1-452]-EGFP were both expressed selectively at the cell surface, cells expressing the latter construct were much dimmer (at equivalent laser power) than hRFC-EGFP-transfected cells. Given identical transfection conditions (0.6 g of DNA, imaged after 48 h) and promoter expression, this variability presumably reflects differences in cellular processing for each construct (e.g. relative rates of protein synthesis and degradation). To quantify these differences, we used flow cytometry to measure the distributions of total fluorescence of cells transfected with constructs that showed predominant targeting to the plasma membrane (hRFC-EGFP, hRFC [ . HEK-293 cells were used for these experiments because their high transfection efficiency permitted measurements from large (Ͼ15,000) populations of cells. We note however, that the subcellular distribution of each hRFC construct was identical in both HEK-293 and HuTu-80 cell lines (Fig. 2C and data not shown). Fig. 4 shows that although the level of protein expression was not reduced appreciably by either partial truncation of the COOH-terminal cytoplasmic tail (mean fluorescence of the hRFC[1-530]-EGFP population 102 Ϯ 3% of that of hRFC-EGFP, three independent runs) or by partial truncation of both the COOH-terminal and NH 2 Figs. 3 and 4 show that the hRFC structure dictates the subcellular targeting as well as the efficiency of protein expression within mammalian cells.
Trafficking of hRFC-EGFP Involves Microtubule-based Transport-In contrast to other membrane transporters (16 -18), little is known about the cytoskeletal mechanisms that direct the intracellular trafficking of hRFC (11). To resolve this issue, we examined the possible roles of microfilaments and microtubules in transporting hRFC-EGFP to the cell surface by using pharmacological methods to disrupt the cytoskeleton selectively.
To identify an appropriate time window for pharmacological intervention, we began by characterizing the kinetics of hRFC-EGFP expression after transfection. Fluorescence was initially apparent after 4 h as a weak juxtanuclear signal. Subsequently, the fluorescence increased, spread throughout the cell, and became increasingly localized to the plasma membrane (Fig. 5A). The relative extent of hRFC-EGFP expression at the cell surface was estimated by costaining with FM4-64, as before. At early times after transfection (ϳ 4 -5 h) there was little colocalization between hRFC-EGFP distribution and FM4-64 staining (12.1 Ϯ 2.5%, Fig. 5, B left and C), and this value was similar to that with EGFP alone (16.7 Ϯ 6%; Fig. 5C). However, by 36 h colocalization between hRFC-EGFP and FM4-64 had increased to a maximal value (63.5 Ϯ 6%, Fig. 5, B right and C) as a result of trafficking of hRFC-EGFP to the cell surface, whereas colocalization in cells transfected with EGFP alone (9.1 Ϯ 6%) or hRFC[1-304]-EGFP (13.8 Ϯ 9%, Ն15 cells) failed to show any further increase (Fig. 5C).
As shown in Fig. 5C, hRFC-EGFP fluorescence at the cell surface began to increase rapidly about 12 h after transfection, and we thus selected this as a suitable time point at which to apply cytoskeletal disrupting drugs, either cytochalasin D, a microfilament disrupting agent, or nocodazole, a microtubuledisrupting agent. Both drugs were added to the incubation medium at final concentrations (400 nM) that are sufficient to disrupt cytoskeletal architecture but minimize toxicity during the prolonged (8 h) application (11,20). Colocalization of FM4-64 and hRFC-EGFP fluorescence was then assessed 8 h later (i.e. 20 h after transfection). Measurements are shown in Fig. 5D, expressed as the change in colocalization relative to that measured at 12 h to reflect trafficking of hRFC-EGFP to the plasma membrane during the time the drugs were present. Although colocalization of fluorescence between hRFC-EGFP and FM4-64 increased to a similar extent in untreated cells (2.2 Ϯ 0.4-fold, n ϭ 12 cells) and cells treated with cytochalasin D (2.0 Ϯ 0.3-fold, n ϭ 21 cells), a slight decrease was seen after incubation with nocodazole (0.91 Ϯ 0.42-fold, n ϭ 39 cells). Thus, disruption of the microtubular cytoskeleton almost completely inhibited hRFC-EGFP trafficking to the plasma membrane, whereas disruption of the actin cytoskeleton was without effect.
Transport Dynamics of hRFC-containing Vesicles-Confocal images of hRFC-EGFP distribution in HuTu-80 cells revealed numerous intracellular vesicular-like structures ( Figs. 2A and  6A). To measure the dynamics of these structures and further investigate the role of cytoskeletal elements in hRFC transport, we employed video-rate confocal imaging to capture image sequences with sufficient temporal resolution to track move-  Fig. 6 as well as in the supplemental video material (Videos 1 and 2). Fig. 6A shows a single frame from a video collected at 37°C (left) and illustrates the method used to track the motion of hRFC-EGFP-containing structures (right). The coordinates of discrete fluorescent structures were tracked at 66-ms intervals (two-frame averages) to generate color-coded tracks representing movements of several vesicles throughout the imaging period (Fig. 6A, right). Fig. 6B shows similar processing applied to a cell maintained at 22°C from which several tracks have been expanded to illustrate generalized aspects of hRFC-EGFP dynamics.
Four general features were evident from cells incubated at both temperatures. First, hRFC-EGFP was localized within structures of varied size and shape (diameters ϳ0.3-1.6 m), including spherical as well as tubular-like vesicles (Videos 1 and 2). Second, there was a wide variability in the dynamics of hRFC-containing vesicles. Some showed little motion (e.g. vesicle a, Fig. 6B), whereas others remained highly dynamic over the entire imaging period or displayed interspersed periods of relatively static and dynamic behavior (e. g. vesicle b, Fig. 6B). Third, the motion of dynamic vesicles was strikingly multidirectional rather than showing unidirectional progression toward the plasma membrane. Vesicles frequently retraced their paths, moved toward and away from the cell surface (e.g. vesicle d, Fig. 6B) or tracked circumferentially beneath the plasma membrane (e.g. vesicle c, Fig. 6B). Finally, vesicles located within the same portion of the cell often showed significant overlap in their tracks (examples in Fig. 6, A-C).
The high temporal resolution of these video-rate confocal images revealed two discrete components to the observed vesicular dynamics, namely, periods of rapid, approximately linear motion (e.g. regions marked by black lines for vesicle b, Fig.  6B), interspersed with lengthy periods of relative immobility during which vesicles displayed Brownian-like movements constrained within a small area (Fig. 6B). These alternating patterns of motion were apparent at both temperatures, but at 22°C the dwell time of vesicles in the stationary state became longer, and their velocities during linear motions were slower (Fig. 6C). Further reduction in temperature to 4°C almost completely inhibited motility (data not shown).
To quantitate the effect of temperature better, we measured the velocities of vesicles during periods of rapid, directed linear transport. Measurements at 37°C from 79 vesicles (15 cells) showed linear movement over a mean distance (run length) of 3.16 Ϯ 0.16 m at a velocity of 1.56 Ϯ 0.22 m s Ϫ1 compared with an average velocity of 0.66 Ϯ 0.07 m s Ϫ1 over a linear distance of 2.99 Ϯ 0.42 m at 22°C (Fig. 6D). Further, the dwell time at rest was 70.4 Ϯ 3.6% at 22°C, decreasing to 49.8 Ϯ 5.1% at 37°C (125 vesicles, Ն9 cells, Fig. 6D).
Having determined these parameters under control conditions, we proceeded to examine the effect of cytoskeletal disruption on vesicle motion. Video data were recorded before and after addition of the microtubule disrupting drugs nocodazole (10 M) or colchicine (10 M), with ␥-lumicolchicine (100 M) as a negative control, and the actin-disrupting agent cytochalasin D (10 M). Videos were taken at 5-min intervals after drug addition, and vesicular dynamics during linear motions were analyzed as above. Observation of cells for up to 30 min (30-s imaging periods every 5 min) showed that exposure to the laser scan had little deleterious effect on vesicular dynamics (average velocities at the start and end of the 30 min recording were 1.82 Ϯ 0.4 m s Ϫ1 versus 1.65 Ϯ 0.4 m s Ϫ1 , respectively; Ն8 cells, Fig. 7A). In contrast, incubation of cells with 10 M nocodazole inhibited the linear motion of vesicles within 5 min (n ϭ 15 cells, Fig. 7, A and B). Colchicine had a similar effect, albeit over a slower time course. After a 5-min incubation in 10 M colchicine, the linear motion of vesicles had decreased to 0.66 Ϯ 0.07 m/s (from 11 cells), and by 20 min motion was inhibited (Fig. 7A). ␥-Lumicolchicine, even at concentrations as high as 100 M for 30 min, failed to inhibit vesicular motility (linear velocity of 1.69 Ϯ 0.23 m/s, Fig. 7A). Similarly, the actin-destabilizing drug cytochalasin D was without effect on the velocity (1.47 Ϯ 0.22 m/s; 15 cells) of linear vesicular movement (Fig. 7, A and C). Representative tracks from cells treated with nocodazole and cytochalasin D are shown, respectively, in Fig. 7, B and C, and the videos from which they are derived are available as supplemental material (Videos 3 and 4). DISCUSSION In this study, we employed confocal imaging methods to answer three questions regarding the cell biology of hRFC. Where is the full-length hRFC-EGFP fusion protein localized in mammalian epithelial cells? How does the sequence of the hRFC protein control its cellular targeting? What cellular mechanism(s) underlie the trafficking of hRFC to the cell membrane? Our results in each of these areas are discussed separately below.
Cellular Localization of hRFC-EGFP-Ligation of the green fluorescent protein to the COOH terminus of the full-length hRFC sequence generates a functional fusion protein (11,21) that we had shown previously localizes to the plasma membrane of Xenopus oocytes (11). Here we extended those studies to mammalian cells and show that hRFC-EGFP similarly targets to the plasma membrane of a human duodenally derived cell line (HuTu-80; Fig. 2, A and C) as well as several other epithelial cell lines (Fig. 2E). However, in addition to hRFC-EGFP expressed at the plasma membrane, intracellular structures also showed appreciable fluorescence (Figs. 2, 3, 6, and 7). The proportion of fluorescent protein in intracellular structures was greatest soon after transfection and likely represents biosynthetic and trafficking pools of hRFC-EGFP. Nevertheless, about 30% of cellular hRFC-EGFP was retained intracellularly (Fig. 6C) even Ͼ30 h post-transfection. The identity of these intracellular structures and the possibility that they are in exchange with the plasma membrane compartment are being investigated currently.
Confocal images of the axial distribution of hRFC-EGFP in HuTu-80 monolayers (e.g. Fig. 2C) indicate that hRFC-EGFP targets to the entire cell surface rather than showing a strongly asymmetric distribution across the cell. These data suggest that folate transport can occur over the entire cell surface, consistent with previous results from in vitro assays (22,23). However, it is important to state that the targeting of membrane proteins to specific cell surface domains is likely dependent on the culture conditions as well as cell type (24 -26), and our results should be considered only in that context and may not, therefore, reflect the polarity of hRFC distribution in vivo.
Molecular Determinants of hRFC Targeting in Mammalian Cells-We show that the polypeptide domains essential for targeting hRFC to the plasma membrane reside within the polypeptide backbone of the protein rather than in the NH 2terminal (amino acids 1-27) or COOH-terminal (amino acids 452-591) cytoplasmic portions of the protein. Partial or full truncation of the cytoplasmic tail of hRFC (hRFC -EGFP and hRFC -EGFP, respectively) did not affect the plasma membrane targeting of these constructs (Fig. 3). Similarly, ablation of the NH 2 -terminal sequence (hRFC [28 -591]-EGFP) did not prevent cell surface localization. Further, the doubly truncated construct (hRFC [19 -466]-EGFP), with ablation of both the NH 2 -terminal and COOH-terminal tails, still localized to the plasma membrane (Fig. 3).
Although the NH 2 -and COOH-terminal cytoplasmic regions are nonessential for conferring cell surface localization, these regions play accessory roles in modulating the efficiency of expression. For example, although the NH 2 -terminal-deleted construct (hRFC [28 -591]-EGFP) reached the cell surface, a considerable portion of this protein was retained intracellularly (Fig. 3). Further, measurements of the total fluorescence intensity showed differences that likely relate to their expression level (Fig. 4). Fluorescence intensity decreased on full (hRFC[1-452]-EGFP), but not partial (hRFC[1-530]-EGFP), truncation of the COOH-terminal cytoplasmic tail (Fig. 4). This consideration could be of clinical significance in that truncation mutations (14) may affect not only the function and trafficking of hRFC to the cell surface, but also the rates of synthesis and degradation of the mutated proteins. Finally, we note that in the Xenopus oocyte expression system hRFC-EGFP, hRFC[1-530]-EGFP and hRFC -EGFP targeted to the plasma membrane in a similar manner to that in HuTu-80 cells, with hRFC[1-452]-EGFP again being expressed at a similarly lower efficiency (11). Thus, the targeting mechanisms of hRFC appear well conserved among very different cell types and organisms.
A third important point is that the integrity of the backbone of hRFC-EGFP is important for export of the polypeptide from the endoplasmic reticulum to the cell surface. Proteins in which the COOH-terminal end of the polypeptide backbone was truncated (hRFC[1-301]-EGFP and hRFC[1-144]-EGFP) remained trapped within intracellular membranes (Fig. 3). Similarly, truncation of the NH 2 -terminal end of the polypeptide backbone (hRFC[302-452]-EGFP), resulted in low expression and intracellular localization. Because the constructs hRFC[1-301]-EGFP and hRFC[302-452]-EGFP, taken together, comprise the entire sequence of hRFC-EGFP, this suggests that no single contiguous amino acid motif dictates export to the cell surface. Rather, a more complex interplay between two or more determinants distributed widely along the polypeptide backbone appears to be involved.
Sequences critical for plasma membrane targeting have been identified in many different cell surface proteins. In contrast to our findings with hRFC, such motifs frequently reside within the cytoplasmic NH 2 -or COOH-terminal tails of these proteins (27)(28)(29)(30)(31)(32). For example, truncation of the COOH-terminal tails of the rat liver-rat Na ϩ /bile acid cotransporter (29), the type iib NaP i cotransporter (31), or the NH 2 -terminal domain of the SGLT1 glucose transporter prevents cell surface expression of the respective proteins (28). However, these cytoplasmic regions seem to play a less essential role in directing the cell surface targeting of hRFC. This observation is not without precedent within the major facilitator superfamily of transporters (33) because similar results have been obtained with other proteins that share the 12 membrane-spanning topology of hRFC (21). For example, the COOH-terminal tail is not essential for cell surface targeting, stability, or function of the Escherichia coli H ϩ /lactose permease (34), and total truncation of the COOH-terminal tail of the glucose transporter GLUT1 impaired function, but not cell surface localization (35). However, it is probably unwise to generalize about conserved tar- geting mechanisms among different classes of proteins, especially because of the low sequence similarity between the RFC family and other transporters within this group. Our data simply underscore the plethora of targeting signals that can act to route polypeptides to their final cellular destinations.
Intracellular Trafficking of hRFC-EGFP-Our previous studies (11) involving expression of hRFC in Xenopus oocytes indicated that delivery of hRFC-EGFP to the plasma membrane involves microtubule based transport. Here, we confirm that finding in mammalian epithelial cells and describe the subcellular dynamics of this vesicular transport process.
First, the microtubule-disrupting agent nocodazole almost completely abolished cell surface expression of hRFC-EGFP, whereas disruption of actin filaments by cytochalasin D was without effect. Second, measurements of the dynamics of hRFC-containing vesicles yielded velocities (ϳ1.2-1.5 m s Ϫ1 ) and linear run lengths (average of ϳ3 m, ranging up to ϳ8 m) consistent with values reported for microtubule-based vesicular motion in a variety of other cell types (36,37) and with the speeds of the microtubule-based motors kinesin and dynein measured in vitro (38). For example, microtubule-based transport of GLUT4 vesicles in adipocytes (albeit resolved at ϳ10fold slower frame rates) occurs with linear velocities of ϳ0.8 m s Ϫ1 over distances of 5-9 m (39). Similarly, microtubuledependent velocities of exocytotic vesicles in Chinese hamster ovary cells were measured to be ϳ1 m s Ϫ1 with run lengths of ϳ3 m (36). Finally, nocodazole rapidly inhibited the rapid, linear movements of hRFC-EGFP-containing vesicles (Fig. 7). This effect was mimicked by colchicine but not ␥-lumicolchicine, a non-tubulin-binding analog (19). Actin-destabilizing drugs were without effect (Fig. 7).
A strikingly feature was that movement of hRFC-EGFPcontaining vesicles was multidirectional. Individual vesicles displayed an apparent haphazard pattern of motion, frequently retracing their steps, despite the net progression of protein to the cell surface. This behavior likely results because association with microtubules is only transient and is interrupted by breaks, with subsequent directional changes on reassociation with individual microtubules (Fig. 6B). Such multidirectional transport, and especially, rapid reversal of individual vesicles on single tracks, supports roles for both plus end-and minus end-directed motors attached to individual hRFC-EGFP-containing vesicles (36). Further, our observation of frequent circumferential movements close to the cell surface suggests that the final stage of insertion, and possibly retrieval, of hRFC-EGFP from the cell membrane may be a precisely regulated process.
In summary, our results demonstrate that the molecular determinants responsible for hRFC targeting to the cell surface of mammalian epithelia are integral to the hydrophobic backbone of the polypeptide, rather than lying within the NH 2 -and COOH-terminal cytoplasmic regions. Further, the integrity of the backbone of hRFC is essential for export of the protein from the endoplasmic reticulum and its trafficking via microtubules to the cell surface.