Organic Cation Permeation through the Channel Formed by Polycystin-2*

Polycystin-2 (PC2), a member of the transient receptor potential family of ion channels (TRPP2), forms a calcium-permeable cation channel. Mutations in PC2 lead to polycystic kidney disease. From the primary sequence and by analogy with other channels in this family, PC2 is modeled to have six transmembrane domains. However, most of the structural features of PC2, such as how large the channel is and how many subunits make up the pore of the channel, are unknown. In this study, we estimated the pore size of PC2 from the permeation properties of the channel. Organic cations of increasing size were used as current carriers through the PC2 channel after PC2 was incorporated into lipid bilayers. We found that dimethylamine, triethylamine, tetraethylammonium, tetrabutylammonium, tetrapropylammonium, and tetrapentylammonium were permeable through the PC2 channel. The slope conductance of the PC2 channel decreased as the ionic diameter of the organic cation increased. For each organic cation tested, the currents were inhibited by gadolinium and anti-PC2 antibody. Using the dimensions of the largest permeant cation, the minimum pore diameter of the PC2 channel was estimated to be at least 11 Å. The large pore size suggests that the primary state of this channel found in vivo is closed to avoid rundown of cation gradients across the plasma membrane and excessive calcium leak from endoplasmic reticulum stores.

Polycystin-2 (PC2), a member of the transient receptor potential family of ion channels (TRPP2), forms a calcium-permeable cation channel. Mutations in PC2 lead to polycystic kidney disease. From the primary sequence and by analogy with other channels in this family, PC2 is modeled to have six transmembrane domains. However, most of the structural features of PC2, such as how large the channel is and how many subunits make up the pore of the channel, are unknown. In this study, we estimated the pore size of PC2 from the permeation properties of the channel. Organic cations of increasing size were used as current carriers through the PC2 channel after PC2 was incorporated into lipid bilayers. We found that dimethylamine, triethylamine, tetraethylammonium, tetrabutylammonium, tetrapropylammonium, and tetrapentylammonium were permeable through the PC2 channel. The slope conductance of the PC2 channel decreased as the ionic diameter of the organic cation increased. For each organic cation tested, the currents were inhibited by gadolinium and anti-PC2 antibody. Using the dimensions of the largest permeant cation, the minimum pore diameter of the PC2 channel was estimated to be at least 11 Å. The large pore size suggests that the primary state of this channel found in vivo is closed to avoid rundown of cation gradients across the plasma membrane and excessive calcium leak from endoplasmic reticulum stores.
PC1 is a 4302-amino acid integral membrane protein with an extracellular N terminus of ϳ3000 residues, which contains a novel combination of known motifs originally found in other proteins that are involved in cell-cell and cell-matrix interactions (4). PC2 is a 968-amino acid integral membrane protein that functions as a calcium-permeable cation channel (5,6). PC2 (also called TRPP2) belongs to the family of transient receptor potential (TRP) channels (7). PC1 and PC2 co-assemble through their coiled-coil domains (8,9). Because many of the pathogenic mutations are in the interacting domains, the co-assembly of PC1 and PC2 is thought to be essential for the function of these proteins in kidney.
PC2 is a nonselective cation channel with multiple subconductance states and a high permeability to calcium (5). PC2 is also permeable to monovalent cations such as Na ϩ , Cs ϩ , and K ϩ and divalent cations such as Ba 2ϩ and Mg 2ϩ , but it is more permeable to divalent than to monovalent cations (5,6). This channel is activated at low concentrations of calcium and inhibited at high concentrations of calcium (6). Other inhibitors include lanthanum (La 3ϩ ), gadolinium (Gd 3ϩ ), amiloride, and a reduction in pH (5).
PC2 comprises six transmembrane-spanning domains where both the C and N termini are intracellular. The fifth and sixth transmembrane domains of PC2 are similar to those of the voltage-gated potassium channels (K v ) and are hypothesized to form the pore region (10). It is still unknown how many subunits make up the pore of the PC2 channel. The ion selectivity differs between K v , which is highly potassium-selective, and PC2, which does not select among the cations. To investigate this difference, we examined the permeation properties of PC2. In previous reports, organic cations with known ionic diameters were employed to determine the pore diameters of ion channels such as the ryanodine receptor (RyR), potassium channel KcsA, and the vanilloid receptor (TRPV6) (11)(12)(13)(14). These previous reports revealed that the RyR has one pore of a minimum diameter of ϳ7 Å (13). TRPV6, which belongs to the family of TRP channels, has a minimum pore diameter estimated to be ϳ5.4 Å (14). RyR and TPRV6 are highly permeable to calcium and exhibit greater permeability for divalent cations compared with monovalent cations, which are biophysical characteristics similar to those of PC2 channels (14). Unfortunately, the pore size of PC2 cannot be inferred from the some of the aforementioned biophysical properties.
To estimate the pore size of the PC2 channel, we used a series of organic cations with varying ionic diameters. Organic amines such as dimethylamine (DMA ϩ ) and triethylamine (TriEA ϩ ) and large tetraalkylammonium (TAA ϩ ) derivatives such as tetraethylammonium (TEA ϩ ), tetrabutylammonium (TBA ϩ ), tetrapropylammonium (TPA ϩ ), and tetrapentylammonium (TPeA ϩ ) were used as current carriers through the PC2 channel after incorporation of the channel into planar lipid bilayer membranes. The currents carried by these organic cat-* This work was supported in part by National Institutes of Health grants. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  ions through PC2 were inhibited by Gd 3ϩ , an antagonist of PC2, and also by anti-PC2 antibody. These data show that the PC2 channel pore is able to accommodate large organic cations such as TBA ϩ and TPeA ϩ . From our results, we conclude that PC2 has a minimum pore diameter greater than 11 Å, a diameter that is nearly twice that of the RyR or TRPV6.

EXPERIMENTAL PROCEDURES
Materials-All chemicals were purchased from Sigma-Aldrich. Chloride salts of the organic cations were used.
Planar Lipid Bilayer Measurements-Membrane vesicles enriched in ER from LLC-PK 1 porcine kidney cell lines stably expressing PC2 (6,15) were isolated by differential centrifugation in the presence of protease inhibitors (16). Vesicles enriched in ER membranes were fused to lipid bilayers containing phosphatidylethanolamine and phosphatidylserine (3:1 w/w; Avanti Polar Lipids, Alabaster, AL) dissolved in decane (40 mg lipid ml Ϫ1 ) (17). In experiments using Cs ϩ as the primary current carrier, ER membranes containing PC2 were added to the cis side, which contained 500 mM CsCl, 7 M CaCl 2 , and 10 mM HEPES, 0.2 mM EGTA, pH 7.35, and the trans side, which contained 250 or 50 mM CsCl, 10 mM HEPES, pH 7.35. After fusion of vesicles, the cis side was rapidly perfused with 15 ml of the trans buffer with no added calcium. In all experiments single-channel currents were measured at various holding potentials with respect to the trans (ground) side.
In experiments using organic amines and large TAA ϩ derivatives as the current carrier, ER vesicles were fused to the bilayer in the presence of Cs ϩ as described above, but the TAA ϩ was in the trans compartment. For current-voltage experiments performed under asymmetrical conditions, organic amines or TAA ϩ (250 mM amine or TAA ϩ , 10 mM HEPES, pH 7.35) were present on the trans side, and the cis buffer comprised 250 mM CsCl, 10 mM HEPES, pH 7.35. In experiments using gadolinium (Gd 3ϩ ) to modify the cesium currents, Gd 3ϩ (0 -500 M) was added to the cis side for the concentration dependence. Subsequent experiments were performed at 400 M Gd 3ϩ . In experiments using anti-PC2 antibody, a 1:350 dilution of anti-PC2 antibody was added to the cis side. This antibody was generated using a 275-amino acid peptide from the C-terminal portion of PC2 (15).
Analysis of Current Amplitudes-Data were amplified (BC-525C, Warner Instruments, Hamden, CT), filtered at 1-3 kHz with a low-pass eight-pole Bessel filter, digitized at 10 kHz, (Digidata 1322, Axon Instruments, Foster City, CA), and stored on a personal computer. Subsequent data analysis was performed using pCLAMP (Axon Instruments). Current amplitudes for the current-voltage relationships as shown in Fig. 2A were generated using all-points histograms for each current carrier. All-points histograms were generated from at least 2 s of continuously recorded data and fit with a Gaussian curve. Open channel current amplitudes were collected for each cation tested. The junction potentials were factored into all of the current amplitude analyses performed.
Current amplitudes were calculated as the difference between the Gaussian means of the open and closed states. Errors shown in the figures were derived using the standard deviations from the Gaussian fit described above. Permeability ratios relative to Cs ϩ (P x /P Cs ) were calculated from the asymmetrical reversal potentials (18). The slope conductance was determined by fitting a linear regression line through holding potentials ranging from Ϫ10 to Ϫ60 mV. A graph was plotted of slope conductance versus the minimum ionic diameter of the cations, and points were fit to an exponential equation. All representative current traces shown in the figures were filtered at 500 Hz and are at least 500 ms long.
Western Blot-Membrane vesicles derived from LLC-PK 1 cells were solubilized with 1% CHAPS for 1 h under constant agitation at 4°C. Samples were collected and centrifuged at 10,000 rpm for 15 min at 4°C. Protein concentration was determined using the Bradford assay (Pierce). 50 g of supernatant sample was added to 2ϫ Laemmli buffer containing 5% ␤-mercaptoethanol, and the proteins were resolved on 4 -20% SDS polyacrylamide gels at 200 mV for 4 -5 h. Proteins were transferred to a polyvinylidene difluoride membrane at 30 mV and 250 mA for 18 -20 h at 4°C. The polyvinylidene difluoride membrane was blocked with 1ϫ phosphate-buffered saline (containing 0.5% Tween 20 and 5% nonfat milk powder) and incubated with primary antibodies anti-RyR (clone MA3-925; Affinity BioReagents) at 1:1,1000 for 2 h or anti-calnexin antibody (Stressgen, Victoria, British Columbia, Canada) at 1:2,000 for 1 h. Secondary antibodies used were either anti-mouse or anti-rabbit horseradish peroxidase-conjugated antibodies at 1:40,000 for 1 h. Samples were visualized using enhanced chemiluminescence detection reagents (Amersham Biosciences). Representative results of at least three independent experiments are shown.

Conductance through PC2 Channels by Organic Cations-
The organic amines DMA ϩ and TriEA ϩ and the large tetraalkylammonium derivatives TEA ϩ , TBA ϩ , TPA ϩ , and TPeA ϩ were used to probe the pore of the PC2 channel. The ionic diameters of these organic cations are shown in Table I. Cs ϩ , which has the smallest diameter (3.4 Å), was used as the cation for comparisons. There is some variation in the diameters of several of the TAA ϩ derivatives because the alkyl chains can be flexible with free rotation occurring around their carbon-carbon bonds. Fig. 1 shows a recording of single-channel currents for some of the organic cation used at a fixed holding potential under asymmetric conditions (Cs ϩ on the cis side, test cation on the trans side). The relative difference in current amplitude among the test cations (DMA ϩ , TriEA ϩ , TEA ϩ , TBA ϩ , TPA ϩ , and TPeA ϩ ) was measured under asymmetrical conditions at holding potentials ranging from 0 to Ϫ60 mV ( Fig. 2A). The current amplitude decreased as the ionic diameter of the current carrier increased ( Fig. 2A). This observation is similar to that seen when organic amines of varying sizes conduct through the intracellular calcium channel, the RyR (11,13,19).
The trend in current amplitudes is also reflected in the reversal potentials and permeability ratios obtained for each cation (Table II). The reversal potential for the cations decreases as the ionic diameter of the cation changes (Table II). Although TPA ϩ (9.8 Å) has a slope conductance similar to TBA ϩ (9.5-11.6 Å), it appears to be more permeable through PC2 than TBA ϩ ; however, the reversal potential of TBA ϩ is smaller. This may be due to the flexibility of TBA ϩ , where the diameter has been calculated to vary between 9.5 and 11.6 Å. Interestingly, the similarity of the currents by the least permeable cations, TBA ϩ and TPeA ϩ (Table II), implies that the diameter of TBA ϩ being "sensed" in our experiments may actually be closer to 11.6 Å. From the permeability ratios, it is  clear that all of the cations tested permeate through PC2. The slope conductance measured from single-channel current amplitudes at negative holding potentials under asymmetrical conditions also decreased as the size of the cation increased (Fig. 2B). The largest slope conductance of 275 Ϯ 15 picosiemens was found for Cs ϩ , the most permeant cation. The slope conductance (in picosiemens for each cation) for DMA ϩ and TriEA ϩ are 220 Ϯ 8 and 206 Ϯ 20; TEA ϩ and TBA ϩ are 160 Ϯ 10 and 125 Ϯ 15; TPA ϩ and TPeA ϩ are 124 Ϯ 5 and 100 Ϯ 10 ( Table II). Inhibition of Cation Currents through PC2 by Gadolinium-Previous studies have shown that the PC2 channel is inhibited by Gd 3ϩ (5). In this series of experiments, we used Gd 3ϩ solely to determine whether the cation currents we observed are through PC2. We first examined the effect of Gd 3ϩ on the Cs ϩ current. The concentration dependence of channel block by Gd 3ϩ was determined to identify the concentration needed for complete inhibition of channel activity. This concentration then was used to inhibit the currents carried by the test cations. Gd 3ϩ inhibited PC2 channel activity by decreasing the channel open probability with an IC 50 of 206 M and full inhibition at 400 M (Fig. 3). The concentration of Gd 3ϩ needed to inhibit the PC2 channel is similar to that used to inhibit TRP channels, further strengthening the structural similarity between PC2 and TRP channels (20). Concentrations of Gd 3ϩ that partially blocked PC2 channel activity showed no effect on the singlechannel conductance. Gd 3ϩ failed to inhibit Cs ϩ currents when it was added to the trans side (Fig. 3B).
The effect of Gd 3ϩ was tested to determine whether the currents carried by each of the test compounds were passing through the PC2 channel. Addition of 400 M Gd 3ϩ inhibited currents carried by the test cations; representative recordings using DMA ϩ , TEA ϩ , TBA ϩ , and TPeA ϩ are shown (Fig. 4). The effect of Gd 3ϩ on all of the cations tested, including Cs ϩ , was reversible upon perfusion of the recording chamber with buffer lacking the inhibitor. These results show that the currents carried by the test compounds are through the PC2 channel.
We also utilized anti-PC2 antibody to inhibit the channel because Gd 3ϩ can inhibit other ion channels. Channel activity was observed at Ϫ40 mV, and then the antibody was added to the cis side. When the current was carried by Cs ϩ , DMA ϩ , TEA ϩ , or TBA ϩ , channel activity was inhibited after addition of the antibody (Fig. 5). Addition of preimmune serum had no effect on PC2 channel activity (data not shown). These results are similar to those obtained with Gd 3ϩ , supporting our assumption that the activity of PC2 was responsible for the measured currents.
The PC2 channels used for these measurements were from ER membranes that could contain additional calcium-permeable channels, specifically the RyR or the inositol 1,4,5trisphosphate receptor (IP 3 R). An immunoblot of samples from vesicles derived from LLC-PK 1 cells using an anti-RyR antibody that recognizes all three RyR subtypes showed that the vesicles do not express the RyR (565 kDa) (Fig. 6). The same immunoblot was reprobed for the ER resident protein calnexin (90 kDa) to show that similar amounts of samples were loaded (Fig. 6). The IP 3 R is present in LLC-PK 1 cells, but it was not activated because of the absence of its agonist, IP 3 , during the experiments. From these results, we conclude that the currents observed were not carried by the RyR or IP 3 R. In addition to possibly harboring these other intracellular channels, LLC-PK 1 may also contain endogenous PC1 (15,21). PC1 is expressed in the plasma membrane, and plasma membrane proteins such as the epidermal growth factor receptor and the  sodium hydrogen exchanger-3 were previously shown to be removed from the ER fraction of the kidney tissue lysate (6). Therefore, PC1 should not influence the activity of PC2 channels derived from the ER microsome preparations.

DISCUSSION
In this study, we examined the pore size of the PC2 channel. The conductance of a series of organic cations of increasing diameter through PC2 was used to address this question. Six organic cations (dimethylamine, triethylamine, tetraethylammonium, tetrabutylammonium, tetrapropylammonium, and tetrapentylammonium) were selected because they spanned a range of molecular sizes and consequently a range of conduc-tances through the PC2 channel. We found the following: 1) the PC2 channels are permeable to large organic cations; 2) the slope conductance of the PC2 channel decreased as the ionic diameter of the cations increased; and 3) PC2 channel currents carried by the organic cations were inhibited by Gd 3ϩ , a known inhibitor of PC2 (5). The observation that very large cations such as TPeA ϩ were able to conduct through PC2 shows that the conduction pathway of PC2 is able to accommodate ions with a diameter up to at least 11 Å. TPeA ϩ was the largest cation used in our experiments because tetraalkylammonium derivatives larger than TPeA ϩ were insoluble in water making it impossible to conduct comparable electrophysiological measurements. The findings in the present study highlight the inherent differences between the biophysical properties of PC2 and other calcium channels such as the RyR and TRP channels. With the RyR, three main differences can be highlighted. The first difference is the permeation properties of large tetraalkylammonium derivatives. In the RyR, organic amines such as trimethylamine (TriMA ϩ ) were used to show that it had a minimum pore diameter of ϳ7 Å (22). However, large tetraalkylammonium derivatives such as TEA ϩ , TPA ϩ , and TPeA ϩ are impermeant through RyR and inhibit conduction of K ϩ through RyR in a voltage-dependent manner (23). The blocking effect of these derivatives was only observed when added to the cis side but not the trans side of the bilayer (23). Unlike RyR, these large cations conduct through the PC2 channel (see Figs.  1 and 2). It is interesting to note that Cs ϩ has a higher slope conductance for the RyR than the PC2 channel suggesting that Cs ϩ is more permeable through the RyR than through the PC2 channel (Fig. 7). However, the curve relating the slope conductance and the diameter of the test cations drops off more quickly for the RyR than for the PC2 channel. In addition, TriEA ϩ (7.2 Å) is impermeant through RyR (22), whereas this and several even larger cations are still able to permeate through the PC2 channel as shown in our experiments. This comparison shows that the pore size and permeability characteristics of the PC2 channel are drastically different from those of the RyR.
The second difference between the two ion channels is the sensitivity to inhibition by Gd 3ϩ . The concentration needed to achieve half-block of the PC2 channel is 206 M, and full inhibition occurred at 400 M (Fig. 3). In contrast, full inhibition of the RyR occurred at 10 M (24). The concentration of Gd 3ϩ needed to block PC2 is more in line with that needed to inhibit other members of the TRP family (20).
The third difference between the two ion channels is modification of cationic current by methanethiosulfonate ethylammonium (MTSEA ϩ ). MTSEA ϩ is known to covalently modify cysteine residues within the conduction pathway of various ion channels such as the acetylcholine receptor, potassium channels, and the ␥-aminobutyric acid type A receptor (25)(26)(27). Previous work using RyR showed that MTSEA ϩ could modify and irreversibly inhibit the current amplitudes of monovalent and divalent cations such as MA ϩ , DMA ϩ , EA ϩ , TriMA ϩ , and Ba 2ϩ (11,28). In contrast, we found that MTSEA ϩ has no effect on PC2 at concentrations shown to inhibit RyR (data not shown). The basis for this discrepancy is not clear. It is possible that the number of cysteines available for modification in PC2 is reduced compared with those in RyR. Indeed this appears to be the case because the RyR is postulated to have 80 -100 cysteines/subunit, where ϳ25% of them are accessible for covalent modification by MTSEA ϩ , and the remaining are unavailable because they do not reside on the surface of the protein, or they form intraprotein disulfide bonds (29). On the other hand, the sequence of PC2 shows that this channel has only 12 cysteine residues. Although the surface availability of these cysteines is unknown, none of them is located in the putative pore-forming regions of the fifth and sixth transmembrane domains (30).
Although PC2 belongs to the family of TRP channels (7), its permeation/pore properties are distinct from TRP channels. The pore of TRP channels has not been well studied; however, a recent report (14) focusing on TRPV6 (vanilloid receptor) revealed, using organic cations such as MA ϩ , DMA ϩ , TriMA ϩ , tetramethylammonium, and N-methyl-D-glucamine, that its pore diameter is ϳ5.4 Å. TRPV6 has very low or zero permeability to TriMA ϩ (5.2 Å) and tetramethylammonium (5.8 Å), respectively (14). Another difference is in the effect of the divalent cation, Mg 2ϩ , on both channels. Mg 2ϩ (2 mM) can block and unblock TRPV6 in a voltage-dependent manner (31), whereas Mg 2ϩ (55 mM) permeates through PC2 and exhibits no inhibitory effect on PC2 (6). Although differences exist in the pore diameter of these related channels, they produce similar responses to MTSEA ϩ . Wild-type TRPV6 is resistant to modification by MTSEA ϩ (14), as is PC2. However, currents through TRPV6 could be blocked by MTSEA ϩ after introducing cysteines in the putative pore-lining domains (14). These observations further support the comparisons between the fifth and sixth transmembrane domains as pore-lining domains of TRP and PC2 channels.
Ion channels similar to PC2 that have large pore diameters include MthK (a calcium-activated potassium channel) (12,32) and the colicin E1 receptor (33). MthK has previously been used as a model protein to understand the structure-function relationships of K ϩ channels and is structurally similar to the P-loop of KcsA, the region of the protein thought to be critical for formation of the MthK pore (12,32,34). Unlike KcsA, which has a pore diameter of 4 Å, MthK has a pore diameter of ϳ12 Å at its narrowest point (32). Coincidentally, the MthK pore has a diameter that can accommodate the cations TBA ϩ and TPeA ϩ , both of which permeate through the PC2 channel. The colicin E1 receptor is an antibacterial toxin that forms voltagedependent channels when inserted into lipid bilayer membranes (33). Interestingly, the colicin E1 receptor has a pore diameter that supports currents carried by nicotine adenine  Table II. The curve for RyR was obtained from previously published values for the slope conductances and ionic diameters of Cs ϩ , methylamine, DMA ϩ , ethylamine, TriMA ϩ , and triethylamine (11,13). dinucleotide (NAD ϩ ), a cation with a diameter of 12-16 Å (33). Thus, the PC2 pore size of at least 11 Å is larger than most calcium and potassium channels but reminiscent of MthK and the colicin E1 receptor.
When compounds of known sizes can pass through an ion channel, it is possible to calculate the number of transmembrane helices needed to form the pore. For the RyR, TRP, and K ϩ channels it has been proposed that two transmembrane domains (S5 and S6) are contributed from each subunit (35)(36)(37). With the tight packing of the transmembrane domains shown for the K ϩ channels (38) and similar tight packing expected for the RyR, at least eight transmembrane helices are needed to form such a large pore (37). Assuming that a minimum of eight transmembrane domains are needed and presuming that the same two transmembrane domains (S5 and S6) are contributed to form the pore (as in RyR/IP 3 R/TRP), then our present experiments suggest a tetrameric assembly of the PC2 protein. Intriguingly, using biochemical techniques, PC2 has been shown to homodimerize in solution (10); however, it is not known whether the dimers associate in its native environment.
The results presented here also strongly suggest that the PC2 channel must be in the closed conformation most of the time to prevent leakage of cellular components of the cell and a dissipation of ion gradients. Dissipation of ionic gradients as a result of mislocalization of membrane proteins and the overexpression of ion channels such as aquaporins have been proposed as factors leading to fluid accumulation, a characteristic of kidney cysts from patients with polycystic kidney disease (39,40). The mechanism of fluid accumulation in the cysts is still yet to be elucidated; however, various hypotheses such as increases in adenosine triphosphate (ATP) and/or cyclic adenosine monophosphate (cAMP) levels could lead to fluid secretion (41,42). The data presented here suggest that mislocalization or excessive openings of PC2 channels can now be added to the list of factors leading to fluid accumulation. When the PC2 channel is not in its closed conformation, the contents of the cell could be compromised, thus contributing, at least in part, to the fluid accumulation observed in the kidney cysts of patients with polycystic kidney disease.