Biogenesis and Topology of the Transient Receptor Potential Ca²⁺ Channel TRPC1*

The TRPC ion channels are a family of Ca²⁺-permeable cation channels that are activated following receptor-mediated stimulation of phospholipase C (1–3). The TRPC family belongs to the TRP superfamily of non-voltage-gated cation channels that also includes channels involved in pain transduction, epithelial Ca²⁺ transport, osmoregulation, mechanosensitivity, cell growth and differentiation, and other as yet uncharacterized functions (1–3). At least 20 TRP superfamily members including seven TRPC family members (TRPC1–7) have thus far been identified in mammals. Recent studies indicate that members of the TRPC family are candidates for the store-operated Ca²⁺ entry pathway activated in response to depletion of intracellular Ca²⁺ stores, a pathway whose molecular identity has remained elusive. However, the details of their involvement in this process and the downstream mechanisms underlying their activation following store depletion remain controversial (1–3).

Our understanding of the function and regulation of this important class of ion channels as well as the design and interpretation of experiments to probe their structure/function relationships require information concerning their transmembrane topology. There is strong evidence that the N and C termini of the TRPC channels are intracellular (1–4), and hydropathy analyses indicate the presence of eight hydrophobic regions (HRs)¹ that could potentially form α-helical membrane-spanning segments (MSSs). Based on sequence similarities to other apparently structurally related ion channels (5), it is thought that only six of the eight HRs actually form MSSs. Specifically, four of the first five HRs are thought to span the membrane, HRs 6 and 8 are thought to span the membrane and border the ion-conducting pore, and HR 7 is thought to form a part of the pore by dipping into the membrane from the extracellular surface (1–3). Evidence supporting this general topology scheme has been obtained for human TRPC3 by Vannier et al. (4). These authors inferred the intracellular or extracellular location of both native and inserted glycosylation sites in the full-length TRPC3 sequence by transiently expressing appropriate recombinant proteins in COS cells and assaying for their glycosylation. They concluded that HR 1 was left out of the membrane on the cytosolic side and that HRs 2–6 and 8 were MSSs. A site placed near the N-terminal end of HR 7 was found to be glycosylated, consistent with its proposed role as a shallowly membrane-embedded part of the ion-conducting pore. They also verified that the N and C termini of TRPC3 were intracellular by showing that HA tags inserted at these sites were not accessible to extracellular antibodies unless the cells were permeabilized.

However, analysis of the TRPC1 sequence using a number of recently derived algorithms for predicting MSSs as well as a detailed comparison of the TRPC1 and TRPC3 sequences suggested to us that a complementary study of the TRPC1 topology would be worth carrying out. For example, we found that HRs 1 and 7 of TRPC1 (the HR thought to be left out of the membrane in TRPC3 and the HR thought to form a part of the ion-conducting pore, respectively) were quite hydrophobic and almost universally predicted to be MSSs in TRPC1. We also noted that sequence conservation between TRPC1 and TRPC3 was rather poor within some of the HRs and their connecting loops (overall sequence identity over the eight HRs and interconnecting loops of TRPC1 and TRPC3 is ∼28%).

The integration of membrane proteins into the bilayer of the ER has been shown to occur with the aid of a large complex of membrane-bound translocation/insertion machinery termed the “translocon” (6, 7). The functional core of this complex is a transmembrane aqueous channel sufficiently large to accommodate one or more MSS. Ribosomes that are synthesizing membrane proteins associate with the translocon in such a way that successive MSSs are fed into this channel where they are recognized and ultimately transferred laterally into the lipid bilayer in their proper transmembrane orientations (6–9). In the simplest case, each MSS is sequentially recognized, oriented, and integrated into the membrane by the translocon as it is synthesized. But more complex scenarios where the integration of a MSS has been shown to depend on presence of neighboring MSSs or on the structure or charge of its flanking regions have been well documented (Ref. 7 and references therein). It also seems clear that multiple MSSs can occupy the translocon channel simultaneously and that some MSSs may exit the translocon into the bilayer en bloc (10).

In this paper, we examine the biogenesis and topology of TRPC1 in intact HEK-293T cells using a truncation mutant approach (11) that allows us to follow the integration/folding process described above. In contrast to results obtained with TRPC3, we find that HR 1 of TRPC1 spans the membrane consistent with its high hydrophobicity, and that HR 3 is left out of the membrane on the cytosolic side. In addition, our results suggest a mechanism by which a pore-like structure might form from HRs 6–8 during TRPC1 biogenesis. As discussed in more detail later in the paper, amino acids downstream of HR 8, including the highly conserved “TRP box” (Glu-Trp-Lys-Phe-Ala-Arg), appear to play an important role in this process. We suggest that this mechanism may be a common feature of the folding of the TRPC and related ion pores.

MATERIALS AND METHODS

DNA Constructs—Segments of the human TRPC1 sequence were cloned into the mammalian expression vector pEGFP-β whose construction has been described previously (11). This vector drives the expression of a fusion protein consisting of the enhanced green fluorescent protein (EGFP) followed by BglII and HindIII restriction sites for the insertion of additional sequence and finally a C-terminal glycosylation tag. The segments of TRPC1 indicated in the text and/or figure legends were amplified by PCR and cloned directly into pEGFP-β by standard methods. The forward and reverse PCR primers, incorporating 5′- and 3′-BglII and -HindIII sites, respectively, were designed essentially as reported in previous studies from our laboratory (11). In all of the fusion protein constructs, the amino acids Ser-Asp-Leu and Gly-Ser-Phe coded (in part) by BglII and HindIII, respectively, flanked the TRPC1 inserts in pEGFP-β. All of the TRPC1 inserts began at M339. The correctness of all of the PCR products was confirmed by direct sequencing. In our early experiments, we were plagued by PCR errors in constructs extending beyond HR 4 (typically resulting in premature stop codons). This problem was resolved by growing all of the transformed bacteria at 30 °C rather than at 37 °C. We suspect that this problem was related to the production of a toxic protein that inhibited the growth of bacteria harboring correctly coded clones, but we have not explored this further. An HA-tagged human TRPC1 clone (in pcDNA3.1) was used as the template for the PCR reactions (12). For inserts longer than K647, the template was the same vector in which the native TRPC1 Hind III site following HR 8 had been destroyed by a silent mutation. Glycosylation consensus sites (see “Results”) were inserted into the full-length HA-tagged TRPC1 sequence in pcDNA3.1 using the QuikChange kit (Stratagene) according to the manufacturer's instructions.

Growth and Transfection of HEK-293T and HEK-293 Cells—HEK-293T cells (from American Type Culture Collection) were cultured in Dulbecco's modified essential medium supplemented with 2 mm glutamine, 100 μg/ml each of penicillin and streptomycin (all from Biofluids), and 10% heat-inactivated fetal bovine serum (Invitrogen). Cells were grown in 10-cm plastic dishes in a humidified incubator at 37 °C and 5% CO₂ and subcultured every 2–3 days. Subconfluent (∼80%) HEK-293T monolayers were transiently transfected overnight (19–24 h) with the expression vectors indicated using Polyfect (Qiagen) according to the manufacturer's instructions.

HEK-293 cells (from Microbix Biosystems Inc.) were cultured and transfected as above with the exception that Earle's minimal essential medium was used in place of Dulbecco's modified essential medium. To obtain stably transfected HEK-293 cells, G418 (0.9 mg/ml) was added to the medium 2 days after transfection and cells were subcultured as necessary. Confluent cultures of G418-resistant cells were harvested ∼3 weeks later.

Preparation of Particulate and Membrane Fractions—Particulate and membrane fractions from HEK-293 and HEK-293T cells were obtained essentially as described previously (11). Cells were washed in phosphate-buffered saline and then homogenized in ice-cold TEEA buffer consisting of 20 mm Tris-HCl, pH 8.0, 1 mm EDTA, 3 mm EGTA, 300 μm AEBSF (4-(2-aminoethyl)benzenesulfonyl fluoride, ICN), 10 μm leupeptin, 10 μm pepstatin A, and 2.5 μg/ml aprotinin (all from Roche Applied Science). This material was centrifuged at 1,000 × g for 10 min, and the supernatant was saved. The pellet was resuspended in TEEA buffer, rehomogenized, and centrifuged as before. The combined supernates from these two homogenization steps were centrifuged at 100,000 × g for 30 min, and the resulting “particulate fraction” was resuspended in TEEA buffer, snap-frozen, and stored above liquid nitrogen (protein concentration was typically 5–10 mg/ml measured using the Bio-Rad protein assay kit with bovine IgG as the standard).

The “membrane fraction” was prepared from the above particulate fraction by an alkaline floatation step (11, 13, 14) as follows. An aliquot of the particulate fraction containing 50–100 μg of protein was diluted to 25 μl with TEEA, and 25 μl of 200 mm Na₂CO₃ (pH 12.0) was added. This mixture was incubated on ice for 30 min and then mixed with 90 μl of 2.5 m sucrose in 100 mm Na₂CO₃. 50 μl of 1.25 m sucrose and 50 μl of 0.25 m sucrose, both containing 0.2 mm EDTA and 10 mm Tris-HCl (pH 8.0), were overlaid next on the alkaline mixture, and the tube was centrifuged at 100,000 × g for 60 min in a Beckman TL100 ultracentrifuge equipped with a TLA100.3 rotor. The 0.25 and 1.25 m sucrose layers and the interface between the 1.25 m sucrose layer and the alkaline mixture were recovered as the membrane fraction.

Deglycosylation of the Membrane Fraction—Aliquots of the above membrane fractions were treated with peptide N-glycosidase F (PNGase F, New England Biolabs) as follows. A 10-μl aliquot of the membrane fraction was diluted to 20 μl in 50 mm sodium phosphate (pH 7.5), 0.5% SDS, and 1% β-mercaptoethanol (final concentrations) and incubated at room temperature for 10 min. 2.22 μl of 10% Nonidet P-40 and 1 μl (1,000 units) of PNGase F next were added, and this mixture was incubated at 37 °C for 2 h. In control samples, PNGase F was substituted by its storage buffer (20 mm Tris-HCl, pH 7.5, 50 mm NaCl, 5 mm Na₂EDTA, 50% glycerol).

Western Blotting and Analysis—SDS-PAGE using 4–20% Tris-glycine Ready gels (Bio-Rad) and Western blotting using a rabbit antigreen fluorescent protein polyclonal antibody (Molecular Probes) were carried out as described previously (11). SDS-PAGE using 10% Tris-glycine Ready Gels (Bio-Rad) and Western blotting using a peroxidase-conjugated anti-HA antibody (Roche Applied Science) were carried out similarly with the exception that incubation with the primary antibody was for 90 min and no secondary antibody was used. Quantitation of Western blots was done using a Molecular Dynamics computing densitometer. Quantitative results shown are means ± S.E. for three or more independent experiments.

RESULTS

Predictions of Membrane-spanning Segments—The MSSs of human TRPC1 predicted by a number of recent theoretical methods are summarized in the upper portion of Fig. 1. In each case, the positions of predicted MSSs in the amino acid sequence are indicated by horizontal lines. The results illustrated are from the DAS (15), SOSUI (16), PRED-TMR (17), Toppred2 (18), the PHDhtm (19), TMHMM (20), and HMMTOP (21) methods. All of these methods have been shown to be considerably more accurate in the prediction of MSSs than a simple hydropathy analysis (22), and all are available for use over the Internet (see Refs. 15–21). In the lower part of Fig. 1, we show a hydrophobicity plot obtained by the classical method of Kyte and Doolittle (23) using a 19-amino acid window. The eight HRs are labeled. Note that all of the theoretical methods predict that HR 1 is a MSS, and most predict that HR 7 is membrane spanning but that HR 3 is not. There is also some disagreement among the methods concerning the integration and location of HRs 4 and 5. The amino acid sequence of the central hydrophobic domain of TRPC1 with the approximate positions of HRs 1–8 indicated is shown in Fig. 2.

The Experimental System—To explore the topology of TRPC1, we have used a truncation mutant approach (11) where portions of the TRPC1 sequence beginning at Met-339 and containing one or more possible MSSs are fused (see “Materials and Methods”) between EGFP and the extracellular tail (177 amino acids) of the β-subunit of the rabbit gastric H,K-ATPase, a glycosylation tag. This latter sequence contains five consensus sites for N-linked glycosylation (24). When translocated into the interior of the ER, it acquires ∼14 kDa of apparent molecular weight due to glycosylation (25), an increase that is easily detected by SDS-PAGE electrophoresis. The use of this glycosylation tag in membrane topology determinations is now well established (11, 24–27). EGFP at the N terminus of our constructs acts as both a cytosolic anchor and a convenient marker for the detection of fusion proteins on Western blots.

The Membrane Topology of HRs 1–3—In Fig. 3A, we show the results of a series of experiments where TRPC1 fragments including HR 1 (K373), HRs 1 and 2 (M416), and HRs 1, 2, and 3 (W443 and Q457) were expressed as EGFP/β-subunit fusion proteins in HEK-293T cells. In each of the panels in Fig. 3A, we show the results of a typical experiment where the membrane fraction from HEK-293T cells, transiently transfected with the truncation mutant indicated, was treated with (+) or without (–) PNGase F (see “Materials and Methods”). These membrane fractions were run on SDS-PAGE and probed by Western blotting to determine the extent of glycosylation of the β-subunit and thus its location inside or outside the ER lumen. The percentage of glycosylated recombinant protein is given below each panel. Thus, for example, the mutant truncated after HR 1 (K373) is highly glycosylated (∼90%), indicating that HR 1 crosses the ER membrane in a N_cyt/C_lum orientation consistent with its high hydrophobicity and with the theoretical predictions discussed above (cf. Fig. 1).

As detailed under “Materials and Methods,” the membrane preparations analyzed in Fig. 3A were obtained from particulate fractions of HEK-293T cells using an alkaline floatation procedure (11, 13, 14). In this procedure, the particulate fractions were first incubated in alkaline 100 mm Na₂CO₃. Under these conditions, any membrane vesicles present are converted to sheets and most protein-protein interactions are disrupted; however, protein-lipid (hydrophobic) interactions are not disrupted and the membrane bilayer remains intact (28). Thus, this incubation is expected to strip away most peripheral membrane proteins but leave integrated proteins in the bilayer. Following this incubation in alkaline medium, the membrane fraction was isolated by floatation on a sucrose gradient (see “Materials and Methods”) (11, 13, 14). In addition to separating the membranes from the peripheral proteins removed by alkaline treatment, this step also leaves any aggregated recombinant proteins arising from overexpression in the lower phase. An analysis of the particulate and membrane fractions from cells expressing K373 showed that 73 ± 3% of this recombinant protein was glycosylated in the particulate fraction and that 24 ± 8% of the total K373 found in the particulate fraction was recovered in the membrane fraction (data not shown). From these results, one can calculate that ∼70% of the glycosylated K373 found in the particulate fraction does not copurify with the membrane fraction. Because glycosylation can only occur in the lumen of the ER, these non-membrane-associated recombinant proteins must have been formerly inserted into the membrane and then later removed, presumably for degradation. Also, because membrane proteins extracted from the ER and targeted for degradation are first deglycosylated (29), the percentage of glycosylated K373 in the particulate fraction (73 ± 3%) is expected to be a lower limit on the percentage of these recombinant proteins targeted to the membrane.

Returning to Fig. 3A, we see that the mutant truncated after HR 2 (M416) is ∼60% glycosylated, indicating that HR 2 crosses the membrane in a N_lum/C_cyt orientation in only 40% of these recombinant proteins. Extending the length of the TRPC1 insert to W430 to include the N-terminal end of HR 3 (Fig. 2) has no significant effect on this result (data not shown); however, when the TRPC1 sequence was extended to include all of HR 3 (W443), almost no glycosylation was observed, indicating that the C terminus of this truncation mutant is mainly cytosolic. Further extension of the TRPC1 sequence to Gln-457 to include the sequence between HR 3 and HR 4 yielded a similar result (Fig. 3A). Taken together, these results indicate that HR 1 is a MSS with a N_cyt/C_lum orientation and that a second MSS with a N_lum/C_cyt orientation, presumably HR 2 or HR 3, exists between HR 1 and Trp-443.

To understand the possible interactions of HRs 2 and 3 with the membrane, we consider the biogenesis of TRPC1 as HRs 1–3 emerge from the ribosome. Fig. 3B shows a schematic representation of HR 1 within the translocon channel during the synthesis of the recombinant protein K373. The black dot at the C-terminal end of HR 1 indicates the end of the TRPC1 sequence (at K373), and the N represents glycosylation sites on the glycosylation tag as it passes through the translocon toward the luminal space. In this depiction, the glycosylation tag is still being synthesized by the ribosome and thus is still attached to it. Note that the translocon, translocon channel, and HR 1 (assumed to be α-helical) have been drawn approximately to scale (diameters ∼10, ∼5, and ∼1 nm, respectively), whereas the ribosome is drawn at an approximate one-tenth scale (diameter ∼25 nm). In Fig. 3C, we show a schematic representation of HRs 1 and 2 of TRPC1 just after the synthesis of HR 2. Because HR 1 has a N_cyt/C_lum orientation and HR 2 is tethered to the ribosome, HR 2 must have a N_lum/C_cyt orientation at this point. Our results do not address the question of whether HR 1 leaves the translocon to enter the lipid bilayer before or after the synthesis of HR 2, so both HRs 1 and 2 have been arbitrarily drawn within the translocon pore in Fig. 3C. In subsequent diagrams of this type, a depiction of the translocon has been omitted for simplicity.

Fig. 3, D and E, show two possible topology schemes that could account for the data presented in Fig. 3A. In Fig. 3D, HR 3 is excluded from the membrane and serves to anchor the TRPC1 sequence on the cytosolic side of the bilayer, possibly by interacting with the membrane surface. In this model, even though HR 2 does not efficiently integrate into the membrane on its own (cf. M416 in Fig. 3A), it is constrained to cross the membrane because HR 3 is retained in the cytosol (i.e. TRPC1 biogenesis proceeds from panel C to panel D). An alternate topology for HRs 1–3 of TRPC1 is illustrated in Fig. 3E. Here HR 3 is integrated into the membrane instead of HR 2. In this model, HR 2 is displaced from the translocon and replaced by HR 3 during TRPC1 biogenesis (i.e. biogenesis proceeds from panel C to panel E). In both cases, the C-terminal end of HR 3 is cytosolic as indicated by the results shown in Fig. 3A (W443 and Q457). To obtain data that distinguish between these two topology schemes, we carried out the experiments illustrated in Fig. 4A. The two constructs studied, W443(R) and W443(RD), are identical to W443 (Fig. 3A) with the exception that W443(R) contains the mutation I431R and W443(RD) contains the double mutation I431R/I432D. These mutations of hydrophobic to charged amino acids dramatically decrease the hydrophobicity of HR 3 (cf. Fig. 4B) and would be expected to seriously compromise its ability to act as a MSS. Accordingly, we would expect that the topology scheme shown in Fig. 3E would be unlikely if not impossible to achieve for these mutants. In fact, both of these mutants are similarly and only weakly glycosylated (Fig. 4A), indicating that the topology of this region of TRPC1 is not dependent on the hydrophobicity of HR 3. Thus, these results are consistent with the hypothesis that Fig. 3D is the correct topology scheme for HRs 1–3 of TRPC1 and that HR 2 crosses the membrane in an N_lum/C_cyt orientation in these constructs.

The above observations suggest that HR 2 of TRPC1 is rather loosely constrained within the membrane at this stage of TRPC1 biogenesis and at least partially dependent on the synthesis of HR 3 to hold it in place. More experimentation will be required to establish whether additional interactions (e.g. with other HRs) further stabilize HR 2 at later stages of TRPC1 biogenesis and/or whether the relative mobility of HR 2 may play a role in TRPC1 function.

Some insight into the behavior of HR 3 can be gained by inspection of the helical wheel plot of its sequence shown in Fig. 4C. This (putative) helix contains a very hydrophobic face (the darkly shaded positions in Fig. 4C) while the rest of the surface is dominated by polar and charged amino acids. The last two turns of the helix on this latter surface are particularly hydrophilic (positions 12, 13, 14, 16, and 17 in Fig. 4C; amino acids Trp-436, Ser-437, Asp-438, Lys-440, and Arg-441, respectively). This highly amphipathic structure seems ideally suited to residing at a lipid/aqueous interface.

The Membrane Topology of HRs 4 and 5—In Fig. 5A, we show the results of experiments where the TRPC1 insert in pEGFP-β was truncated after HR 4 (D488) or HR 5 (M530). A modest but significant level of glycosylation was observed for the mutant truncated after HR 4, and a smaller glycosylation signal was found for the mutant truncated after HR 5. Although these data suggest a membrane association for HRs 4 and 5, they do not answer the question of whether they are MSSs as found for the homologous regions of TRPC3 (4) and as predicted by most of the theoretical methods examined in Fig. 1. After inspecting the sequence of this region of TRPC1, we did not feel that results from additional truncation mutants would be likely to help in resolving this issue. Accordingly, to address this question from another perspective, we inserted a glycosylation consensus site (Asn-Gly-Thr) into full-length TRPC1 after W489 in the hydrophilic loop between HRs 4 and 5 and assayed for its glycosylation in stably transfected HEK-293 cells. As illustrated in Fig. 5B, this modified protein (Gly 4-5) was, in fact, strongly glycosylated, whereas no glycosylation of wild-type TRPC1 (wt TRPC1) was observed. This result indicates that the loop between HRs 4 and 5 is exposed to the ER lumen in full-length TRPC1. We have already demonstrated the cytosolic location of the N-terminal flanking sequence of HR 4 (Fig. 3), and in experiments presented below, we provide strong evidence that HR 6 is a MSS with an N_cyt/C_lum orientation. Taken together with the results of Fig. 5, these data indicate that that HRs 4 and 5 are MSSs with N_cyt/C_lum and N_lum/C_cyt orientations, respectively, in full-length TRPC1.

The Membrane Topology of HRs 6–8 and Formation of the TRPC1 Pore—The data shown in Fig. 6A illustrate the results of a series of experiments where the TRPC1 insert in pEGFP-β was truncated after HR 6 (K561), HR 7 (R610), at two places within HR 8 (V636 and V642), and at several places in hydrophilic sequence following HR 8 (K647, D657, and R664). The mutant truncated after HR 6 (K561) is highly glycosylated (>90%), indicating that HR 6 is inserted into the ER membrane in a N_cyt/C_lum orientation. The mutant truncated after HR 7 (R610) shows little glycosylation, indicating that HR 7 crosses the membrane in a N_lum/C_cyt orientation in this construct. This observation was somewhat surprising to us because, as discussed in the Introduction, HR 7 is thought to form a part of the TRPC1 pore; however, this result is consistent with the high hydrophobicity of HR 7 (cf. Fig. 1). Extending the insert to include the loop between HRs 7 and 8 as well as ∼14 residues of HR 8 (V636) did not significantly change this glycosylation pattern. But ending the insert at Val-642 near the C-terminal end of HR 8 resulted in a highly glycosylated recombinant protein, indicating that HR 8 was integrated into the membrane in an N_cyt/C_lum orientation in this construct. However, as shown by the remaining results shown in Fig. 6A, the (hydrophilic) sequence downstream of HR 8 had a dramatic effect on the apparent topology of this region of TRPC1. As the insert in pEGFP-β was extended to K647, D657, and R664, the level of glycosylation of the expressed protein was successively reduced, such that the C-terminal end of the fusion protein R664 had adopted a cytosolic orientation.

To further examine the configuration of HR 6–8 in intact TRPC1, we inserted glycosylation consensus sites (Asn-Gly-Thr) into full-length TRPC1 in the hydrophilic loop between HRs 6 and 7 (Gly 6-7) or in the hydrophilic loop between HRs 7 and 8 (Gly 7-8) and then assayed for their glycosylation in HEK-293 cells. As illustrated in Fig. 6B, Gly 6-7 was strongly glycosylated, indicating that the loop between HRs 6 and 7 is exposed to the lumen of the ER as predicted by our results from truncation mutants (Fig. 6A). Significant glycosylation is also seen for Gly 7-8, indicating that the loop between HRs 7 and 8 is also exposed to the ER lumen. The incomplete glycosylation of Gly 7-8 is probably the result of the short length of the loop between HRs 7 and 8. It is well known that glycosylation sites located close to the ER luminal membrane are inefficiently glycosylated, apparently because they are poorly accessible to oligosaccharyl transferase (30–32).

The above results can be accounted for as follows. Fig. 6C shows a schematic representation of this region of TRPC1 as HR 8 begins to emerge from the ribosome. At this point, HRs 6 and 7 have adopted N_cyt/C_lum and N_lum/C_cyt orientations, respectively, as indicated by the data in Fig. 6A. If the TRPC1 sequence is truncated at V642 near the C-terminal end of HR 8, our results indicate that HR 8 will adopt an N_cyt/C_lum orientation in the resulting fusion protein so that HRs 6, 7, and 8 all act as MSSs. However, when HR 8 is followed by its C-terminal flanking sequence, this downstream sequence is retained in the cytoplasm (this presumably occurs because this sequence is excluded from the translocon but our results do not address this point). This apparently constrains HR 8 to cross the membrane in an N_lum/C_cyt orientation. As a consequence, the intrinsic topogenic properties of HR 7 are overridden so that both its N-terminal and C-terminal ends are exposed to the ER lumen (Fig. 6, B and D).

DISCUSSION

The final transmembrane topology of TRPC1 indicated by our experiments is illustrated in Fig. 7. As discussed in the Introduction, in contrast to our results for TRPC1, Vannier et al. (4) conclude that HR 1 of TRPC3 was left out of the membrane on the cytosolic side, whereas HRs 2 and 3 were MSSs. This conclusion was based largely on their observation that a glycosylation site located between HRs 2 and 3 was glycosylated in full-length TRPC3 (a site inserted between HRs 1 and 2 of TRPC3 was not glycosylated, but the loop between HRs 1 and 2 may be too short to permit efficient glycosylation). We have no reason to question these earlier results for TRPC3 but we find our data for TRPC1 to be equally convincing. These differences may indicate a structural heterogeneity within the TRPC family. The extent to which homologous membrane proteins adopt the same transmembrane topologies is still unknown. Although similar topologies may be the norm, Tatishchev et al. (33) have recently provided evidence for a 10 MSS topology of the human pancreatic sodium bicarbonate transporter, pNBC1, whereas a 14 MSS topology for the homologous human chloride bicarbonate exchanger, AE1, has been proposed (34). Interestingly, over their central hydrophobic domains human pNBC1 and AE1 are 39% identical, whereas human TRPC1 and TRPC3 are only 28% identical (as determined from ClustalW alignments). Differing topologies have likewise been suggested for members of the SGLT and SLC7 transporter gene families (35, 36). It is also interesting to note that our topology scheme for HRs 1–3 of TRPC1 (Fig. 3B) is the same one originally proposed by Phillips et al. (5) for this region of the Drosophila TRP family member TRPL.

Our data also indicate that although HR 4 does not integrate efficiently into the membrane on its own, HRs 4 and 5 form the third and fourth MSSs, respectively, of full-length TRPC1. Our results do not address the questions of how or when these HRs enter the membrane. However, it is reasonable to speculate that this is the result of a synergistic interaction between HRs 4 and 5 themselves. For example, it is possible that these two rather weakly hydrophobic segments (Fig. 1) associate with one another during TRPC1 biogenesis in such a way that they mask some of one another's hydrophilic residues. The resulting more hydrophobic structure may then be readily integrated into the membrane via the translocon complex. In fact, exactly this mechanism has been suggested by Sato et al. (37) for the integration of the corresponding third and fourth MSSs of the Shaker-type K⁺ channel KAT1, a channel that is thought to be structurally related to TRPC1 (see below). There have also been a number of other examples in the literature where the integration of one HR of a membrane protein has been shown to be dependent on the presence of a neighboring (typically downstream) MSS (38–41).

Our (schematic) model for the pore region of TRPC1 (Fig. 7, HRs 6–8) is in agreement with the studies of TRPC3 carried out by Vannier et al. (4) and also corresponds with predictions made some time ago for the structure of this region based on similarities between the TRP channels and other ion channels (1–3, 5). Recent sophisticated sequence analyses have confirmed that the TRP channels are members of the Pfam family (42) PF00520 that includes numerous sodium, potassium, and calcium ion channels, including KAT1 (see above paragraph). These channels are all thought to have a similar structure of six MSSs, the last two of which flank a pore-like ion selectivity filter (in some family members, this six MSS domain is repeated). Interestingly, Sato et al. (37) found that the hydrophobic pore-forming region of KAT1 displayed no topogenic function. In contrast, we find that HR 7 of TRPC1 is apparently capable of integrating into the membrane in a N_lum/C_cyt orientation (Fig. 6). But our results indicate that this ability of HR 7 to form a MSS is overcome by the strong tendency of HR 8 of TRPC1 to adopt an N_lum/C_cyt orientation (Fig. 6). This effect is largely due to the influence of ∼19 residues of its downstream flanking sequence (H646-R664). This sequence includes a number of charged amino acids (cf. Fig. 2) along with the highly conserved TRP box (Glu-Trp-Lys-Phe-Ala-Arg; amino acids 659–664). Past studies have demonstrated that clusters of positively charged amino acids can act as strong topogenic determinants because they tend to be retained in the cytoplasm (7, 43). However, this effect is much more significant in bacteria than in eukaryotes (43) and in any case can only partially account for the orientation of HR 8 because much of the effect of this flanking sequence is seen in the recombinant protein D657, which contains a single downstream positive charge. Nevertheless, it is clear that the sequence that follows HR 8 acts to anchor its C-terminal end on the cytosolic side of the membrane. This result together with our observations that glycosylation sites inserted on either side of HR 7 are exposed to the ER lumen (Fig. 6B) provide strong supporting evidence for existence of the previously envisioned pore-like structure depicted schematically in Fig. 7.

Because of the high hydrophobicity of HR 7 and its ability to act as a MSS, the TRPC1 pore appears to be an example of the phenomenon of topological “frustration.” This effect was first described by Gafvelin and von Heijne (44) in model membrane proteins. It occurs when two HRs within a molecule prefer to insert into the membrane with incompatible transmembrane orientations resulting in one being constrained to assume a non-membrane-spanning state. We speculate that this might be a common mechanism for the formation of the pore-like structures thought to be characteristic of many ion channels, particularly those that have a rather hydrophobic pore region such as TRPC1.