Properties of the TRPML3 Channel Pore and Its Stable Expansion by the Varitint-Waddler-causing Mutation*

TRPML3 is a H+-regulated Ca2+ channel that shuttles between intracellular compartments and the plasma membrane. The A419P mutation causes the varitint-waddler phenotype as a result of gain-of-function (GOF). The mechanism by which A419P leads to GOF is not known. Here, we show that the TRPML3 pore is dynamic when conducting Ca2+ to change its conductance and permeability, which appears to be mediated by trapping Ca2+ within the pore. The pore properties can be restored by strong depolarization or by conducting Na+ through the pore. The A419P mutation results in expanded channel pore with altered permeability that limits modulation of the pore by Ca2+. This effect is specific for the A419P mutation and is not reproduced by other GOF mutations, including A419G, H283A, and proline mutations in the fifth transmembrane domain. These findings describe a novel mode of a transient receptor potential channel behavior and suggest that pore expansion by the A419P mutation may contribute to the varitint-waddler phenotype.

TRPML3 is a H ؉ -regulated Ca 2؉ channel that shuttles between intracellular compartments and the plasma membrane. The A419P mutation causes the varitint-waddler phenotype as a result of gain-of-function (GOF). The mechanism by which A419P leads to GOF is not known. Here, we show that the TRPML3 pore is dynamic when conducting Ca 2؉ to change its conductance and permeability, which appears to be mediated by trapping Ca 2؉ within the pore. The pore properties can be restored by strong depolarization or by conducting Na ؉ through the pore. The A419P mutation results in expanded channel pore with altered permeability that limits modulation of the pore by Ca 2؉ . This effect is specific for the A419P mutation and is not reproduced by other GOF mutations, including A419G, H283A, and proline mutations in the fifth transmembrane domain. These findings describe a novel mode of a transient receptor potential channel behavior and suggest that pore expansion by the A419P mutation may contribute to the varitint-waddler phenotype.
TRPML3 belongs to the TRPML subfamily of the transient receptor potential (TRP) 3 channels superfamily (1,2). The subfamily consists of TRPML1, mutations in which cause the lysosomal storage disease mucolipidosis type IV (3), TRPML2 and TRPML3. The A419P mutation in the putative fifth transmembrane domain (TMD) of TRPML3 causes the varitint-waddler phenotype that is characterized by pigmentation defect, hearing loss, circling behavior, and embryonic lethality (4,5).
TRPML3 functions as an inwardly rectifying cation channel (6), with high Ca 2ϩ /K ϩ selectivity (7)(8)(9)(10) and shuttles between the plasma membrane and intracellular organelles to regulate membrane trafficking (11,12) and autophagy (11). Recently, we reported a unique form of regulation of TRPML3 by extracytosolic (luminal) Na ϩ and H ϩ (H ϩ e-cyto ), which is mediated by a string of three histidines (His 252 , His 273 , and His 283 ) in the large extracytosolic loop between TMD1 and TMD2 (7). Binding of H ϩ to His 283 inhibits the channel, whereas His 252 and His 273 hinder H ϩ access to His 283 . Interestingly, preventing H ϩ binding to His 283 resulted in a gain-of-function (GOF) that increases Ca 2ϩ influx and causes marked cell toxicity (7). Regulation of TRPML3 function by binding of H ϩ to His 283 suggested the possibility that the loop communicates with the pore domain and the communication is regulated by H ϩ binding to His 283 . This possibility is further suggested by elimination of the regulation by H ϩ e-cyto by the A419P mutation in the fifth TMD of TRPML3 (7).
We (6) and others (8 -10) have shown that the A419P mutation is a GOF mutation that locks TRPML3 in an open conformation. The mechanism by which the A419P mutation leads to GOF is not known. Although elimination of the inhibition of TRPML3 activity by H ϩ e-cyto plays a role in the GOF (7), the A419P mutation may also affect the pore selectivity and conductance. Because proline introduces a kink or a break in ␣-helical transmembrane spans, and the A419G mutation that may disrupt ␣-helical structures results in a similar GOF in TRPML3 as the A419P (6,9), it was postulated that the GOF by TRPML3(A419P) is caused by destabilization of the fifth TMD helix of TRPML3 (9,13). However, it is still unclear how the proline-induced destabilization of the fifth TMD affects the behavior of the TRPML3 pore to lead to the GOF.
Very little is known of the properties of the TRPML3 pore. As indicated above, the pore of the wild-type (WT) channel shows high Ca 2ϩ /K ϩ and Ca 2ϩ /Cs ϩ selectivity with similar Na ϩ and Ca 2ϩ conductance at isotonic ion concentrations (7). With limited exceptions, ionic selectivity and conductance are considered constant and defining features of ion channels. The best known examples of ion channels with variable permeability are the ionotropic P2X receptors, which undergo time-dependent slow pore expansion subsequent to activation with their ligands (14 -16). Other recent examples are the TRPV1 and TRPA1 channels. TRPV1 undergoes time-dependent pore expansion following agonist stimulation (17), with the agonist type controlling the rate of pore expansion (17) and Ca 2ϩ /Na ϩ conductance ratio (18). Ligand stimulation dilates the TRPA1 pore (19,20) and increases its Ca 2ϩ /Na ϩ selectivity ratio (19).
To understand better the molecular mechanism of the varitint-waddler phenotype, we characterized the pore properties of WT TRPML3 and several TRPML3 mutants, including TRPML3(A419P). We report that the TRPML3 pore is dynamic and undergoes spontaneous, Ca 2ϩ -dependent modulation when conducting Ca 2ϩ . The A419P mutation alters this behavior by stably expanding the pore. Interestingly, several GOF mutations, and in particular H283A in the extracytosolic loop and the A419G in the fifth TMD, do not affect pore behavior. These findings describe a unique behavior of the TRPML3 pore and provide a molecular mechanism for the GOF in the varitint-waddler phenotype.

MATERIALS AND METHODS
Plasmid Construction, Mutagenesis, Reagents, and Cells-Full-length human TRPML3 was amplified from human placenta and cloned into the pEGFPC1 and p3XFLAG-CMV-7.1 vectors, as described before (7). The full-length sequence corresponding to the human TRPML1 coding region was amplified by PCR using IMAGE clone BF 529860 as template, and the amplified product was subcloned into pEGFPC1. The QuikChange mutagenesis kit was used to introduce mutations by following the manufacturer's instructions. All mutations were confirmed by sequencing the entire open reading frame to verify the presence of the desired mutation and the absence of extraneous mutations. All reagents were purchased from Sigma. HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. One day after plating, cells were transfected with Lipofectamine 2000 and used for current recording 24 h later.
Current Recordings-The TRPML3 current was recorded at room temperature using the whole-cell configuration of the patch clamp technique. The patch electrodes had a resistance of 3-5 M⍀. Series resistance values were generally below 15 M⍀ and were not compensated for. Data were collected and ana-lyzed using an Axopatch 200 amplifier and pClamp 9.2 software. Currents were measured by application of Ϫ100 to ϩ100 mV, 100-ms voltage RAMPs every 1 or 5 s from a holding potential of 0 or ϩ80 mV. Unless otherwise indicated, the pipette solution contained 130 mM cesium aspartate, 20 mM CsCl, 1 mM MgCl 2 , 4 mM Na 2 -ATP, 10 mM HEPES, and 10 mM BAPTA. In some experiments, as indicated, 2 mM CaCl 2 and 2 mM EGTA were added to the standard pipette solution instead of 10 mM BAPTA to clamp Ca 2ϩ at 10 M. Effectively, this solution has no mobile Ca 2ϩ buffer because most of the EGTA is bound with the added Ca 2ϩ . The standard bath solution contained 140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 ,10 mM glucose, and 10 mM HEPES, adjusted to pH 7.4 with NaOH. The Na ϩ -free solution contained 150 mM NMDG-Cl, 5 mM EGTA, and 10 mM HEPES, adjusted to pH 7.4. Ca 2ϩ -containing solutions were prepared by isosmotically replacing NMDG ϩ with Ca 2ϩ in the Na ϩ -free solution. Monovalent ion solutions contained 150 mM monovalent ions instead of NMDG ϩ in the Na ϩ -free solution, adjusted to pH 7.4 with NMDG-OH.
Orai1 current was recorded with the same pipette solution used to record the TRPML3 current. The channel was activated by passive store depletion with 10 mM BAPTA in the pipette solution, and the current was initiated by perfusing the cells with a solution containing 105 mM CaCl 2 and 10 mM HEPES, pH 7.4. Current was measured by Ϫ100 to ϩ100 mV, 100-ms RAMPs every 1 s from a holding potential of 0 mV.
Data Analysis-The permeability ratios of monovalent cations relative to Na ϩ were calculated from the shift in reversal potential (⌬E rev ) on changing the extracellular solutions using the question P X /P Na ϭ exp(⌬E rev xF/RT), where R, T, and F have their usual meaning. The shift in E rev was determined relative to Na ϩ in the same cells to nullify any effect of the leak current on E rev . All potentials were corrected for liquid junction potentials using the program JPCalcW (Axon Instruments). 2 -distribution for two exponential fits of change in E rev time course was about 10 times better than single exponential fits, and thus the changes in E rev are mostly fitted with two exponentials. Pore diameter was estimated by fitting the currents to the modified excluded volume equation: P X /P Na ϭ (dP Ϫ dX/dP Ϫ dNa) 2 , where dP, dX, and dNa are diameters of the pore, cation X ϩ , and Na ϩ , respectively. Data are shown as mean Ϯ S.E. of the indicated number of observations. FIGURE 1. Dynamic behavior of the TRPML3 pore. a, whole cell current was measured in HEK cells transfected with WT TRPML3. The major cation in the pipette solution was 150 mM Cs ϩ , and Ca 2ϩ was buffered close to 0 with 10 mM BAPTA. The major cation in the bath solution was 150 mM Na ϩ (black bars), 150 mM NMDG ϩ (light gray bars), or 100 mM Ca 2ϩ (black bars). TRPML3 was activated by exposing the cells to Na ϩ -free bath solution. After measurement of the maximal Na ϩ current, the cells were exposed to 100 mM Ca 2ϩ . Shown is the current measured at Ϫ100 (lower circles) and ϩ100 mV (upper circles). Dashed lines here and in all other figures indicate the 0 current levels. b and c, I/V relationships of the TRPML3 current recorded at the times shown in the large black circles in a. d, changes in E rev during the first 1 min of WT TRPML3 in 150 Na ϩ (squares) and 100 Ca 2ϩ (circles). The results are the mean Ϯ S.E. of seven experiments. e, current clamp protocol was used to record the membrane potentials.

RESULTS AND DISCUSSION
Behavior of the TRPML3 Pore-In a previous study, we reported that WT TRPML3 is a selective Ca 2ϩ channel (6). The current was recorded by Ϫ100 to ϩ100 mV voltage RAMPs from a holding potential of 0 mV every 5 s. As was shown previously (6, 7), short incubation of cells expressing WT TRPML3 in Na ϩ -free solution and readdition of Na ϩ resulted in a large strongly inward rectifying current that slowly inactivates in the continuous presence of Na ϩ o (Fig. 1a). Incubation in isosmotic Ca 2ϩ following the exposure to Na ϩ -free medium evoked a large inward current that undergoes partial fast inactivation and stabilization at a plateau of about 25% of the initial current (Fig. 1a). The current/voltage (I/V) relationship of the Na ϩ current ( Fig. 1b) showed only a slight shift in E rev , even 5 min after channel opening. By contrast, the Ca 2ϩ current displays a large shift in E rev during the first two to three sweeps (Fig. 1c). The analysis in Fig. 1d and supplemental Table 1 show that the shift in E rev of the Ca 2ϩ current could be fitted well with two exponentials. E rev was also recorded using current clamp (Fig. 1e). The hyperpolarization observed on exposure to Na ϩ -free medium reflects the rate of channel activation in the absence of external permeable cations. Incubation in isotonic Na ϩ resulted in a small stable depolarization. On the other hand, incubation in isosmotic Ca 2ϩ resulted in a large but transient depolarization, similar to the shift in E rev in Fig. 1d. However, notably, the decay in E rev of the Ca 2ϩ current in Fig. 1e was fitted best to a single exponential.
Although cytoplasmic Ca 2ϩ accumulation likely accounts for part of the shift in E rev , most of the shift may be caused by a change in channel pore properties. An attempt to increase the Ca 2ϩ buffer by increasing pipette BAPTA to 30 and 50 mM, to reduce the effect of accumulated Ca 2ϩ on E rev , was not feasible because this resulted in inactivation (50 mM) or altered properties (30 mM) of the channel. Therefore, to determine whether a change in channel Ca 2ϩ selectivity contributes to the shift, we determined whether the E rev shift is influenced by current density and the cytoplasmic mobile Ca 2ϩ buffer. In Fig. 2 we compared the shift in E rev in cells dialyzed with pipette solution containing 10 M Ca 2ϩ buffered with 2 mM EGTA (no free buffer) and pipette solution containing 10 mM BAPTA and at two current densities. External Ca 2ϩ was reduced to 10 mM to reduce Ca 2ϩ influx, and the current was sampled every second to resolve the time constant of the E rev shifts better. The current at 10 mM BAPTA was fit best with two exponentials of 4.3 Ϯ 0.5 and 70.2 Ϯ 36.3 ( 2 for two-exponential 0.4 Ϯ 0.1 and for single exponential 3.0 Ϯ 1.6). However, the Ca 2ϩ current at 10 M Ca 2ϩ can be fit best with a single exponential time course, and thus Fig. 2 shows the time constants of the single exponential fits. Fig. 2 show that the time constant of the shift in E rev is similar in the presence and absence of Ca 2ϩ buffer and at a current density range of 80 -550 pA/pF. If most of the shift in E rev was because of Ca 2ϩ accumulation, the time constant would have been strongly affected by current density and the Ca 2ϩ buffer.
Additional evidence that part of the shift in the Ca 2ϩ current E rev is because of a change in channel pore properties is shown in supplemental Fig. S1: (i) TRPML3(A419P) retains part of   Fig. 1a, except that the holding potential (HP) was ϩ80 mV. b, E rev changes during the first 1 min of current recording for WT TRPML3 in 100 Ca 2ϩ solutions at holding potentials of ϩ80 mV (filled circles) and 0 mV (open circles). The latter is reproduced from Fig. 1d to facilitate comparison. The results at ϩ80 mV are the mean Ϯ S.E. of three experiments. c, Na ϩ and Ca 2ϩ currents of WT TRPML3 were measured as in Fig. 1, except that at the periods marked by gray bars the cells were incubated in 10 mM Ca 2ϩ -containing Na ϩ -free bath solution. All Na ϩ -free solutions contained 5 mM EGTA. Note that after conducting Ca 2ϩ , both Na ϩ and Ca 2ϩ currents are strongly inhibited. d, mean Ϯ S.E. (error bars) of the E rev recorded at the times shown in the large filled circles in c in 10 mM Ca 2ϩ during the first and second exposure to Ca 2ϩ and the recovery after relief of the inhibition by conduction of Na ϩ by the channel (n ϭ 4). The first columns are the initial E rev , and the striped columns are the E rev after 30-s exposure to 10 mM Ca 2ϩ .
the E rev shift (see below), whereas TRPML1(V432P), which shows similar inward rectifying Ca 2ϩ current density as TRPML3(A419P), shows no E rev shift (supplemental Fig.  S1, d-f); (ii) TRPML3 mutants that reduce the Ca 2ϩ conductance did not prolonged the time constant of the shift (supplemental Fig. S1g); (iii) Ca 2ϩ current by the inward rectifying Orai1 had no obvious effect of the E rev for Ca 2ϩ . Although none of these criteria alone is sufficient, together they do indicate that the E rev shift is at least in part due to a change in the TRPML3 pore properties for Ca 2ϩ .
To examine the effect of the membrane potential on the shift in E rev , we recorded the Ca 2ϩ current while holding the membrane potential at ϩ80 mV. Fig. 3a shows that the Ca 2ϩ current develops normally at ϩ80-mV holding potential, but the current does not undergo the fast inactivation observed at 0-mV holding potential (Fig. 1a). Fig. 3b shows that the shift in E rev is prevented when the holding potential is ϩ80 mV.
The effects of holding the membrane potential at ϩ80 mV raise the possibility that the shift in E rev and the fast inactivation of the Ca 2ϩ current are mediated by Ca 2ϩ binding to a site that regulates the pore. The positive membrane potential may prevent interaction of Ca 2ϩ with or dislodge Ca 2ϩ from this site. We tested this prediction first by repeated application of a solution containing 10 mM Ca 2ϩ . Fig. 3c shows that after the initial fast inactivation, the channel remains partially inactivated on repeated application of 10 mM Ca 2ϩ . Notably, Fig. 3c shows that E rev remained shifted, even when the cells were incubated in Ca 2ϩ -free solution containing 5 mM EGTA. Because of gradual channel inactivation we were able to alternately expose the channel to 10 mM Ca 2ϩ and 5 mM EGTA up to four times and as long as 10 min. Yet, E rev never recovered or showed significant farther shift. Fig. 3d shows that almost complete recovery of E rev was obtained when TRPML3 was made to conduct Na ϩ between Ca 2ϩ current measurements, although the current magnitude did not recover. This suggests that Na ϩ flow through the channel is necessary to displace Ca 2ϩ from an inhibitory site in the pore.
It is possible that the shift in E rev is mediated by Ca 2ϩ binding to a deep pore site because removing intracellular Ca 2ϩ with 10 BAPTA and extracellular Ca 2ϩ by including 5 mM EGTA in the Na ϩ -free bath solution did not restore E rev (Fig. 3d). This may be mediated by pore-lining charged residues that affect in the E rev shift. A search for such potential residues is shown in supplemental Fig. S2a, which depicts the predicted lining of the negatively charged Glu 449 , Asp 458 , and Asp 459 in the TRPML3 pore region. Glu 449 is predicted to be in the pore helix, whereas Asp 458 and Asp 459 are predicted to line the selectivity filter. Mutation of Asp 458 to lysine, alanine, or asparagine resulted in an inactive channel. The E449K mutation completely inhibited the channel. However, supplemental Fig.  S2, b-d shows that the E449A mutation converted the channel to constitutively active, nearly eliminated inactivation of the Ca 2ϩ current, and altered the shift in E rev . The D459A mutation resulted in a marked decrease in Ca 2ϩ permeability and selectivity and eliminated the shift in E rev (supplemental Fig. S2, e-g). These findings suggest that Glu 449 and Asp 459 participate in the E rev shift. . Effect of the varitint-waddler phenotype causing A419P mutation on TRPML3 pore behavior. a and e, whole cell currents of the A419P and H283A mutants were measured as in Fig. 1. b and c, f and  The results so far suggest that Ca 2ϩ has two apparently independent effects on TRPML3. Ca 2ϩ causes partial inhibition of the current and a shift in E rev during Ca 2ϩ conduction to reduce the pore selectivity for Ca 2ϩ . The independence of the two effects is concluded from the restoration of E rev for Ca 2ϩ but not of the magnitude of either Ca 2ϩ or Na ϩ currents. On the other hand, the Na ϩ current is fully restored on repeated alternate exposure of TRPML3 to Na ϩ -free and Na ϩ -containing media (6,7). The shift in TRPML3 E rev for Ca 2ϩ is different from the shift in E rev reported for TRPV1 (17,18), TRPA1 (19,20), and the P2X receptors (14,16) in that it is not caused by agonist-mediated events, but is mediated by the conducted ion itself.
Effect of the Varitint-Waddler Phenotype-causing Mutation A419P on the Properties of the TRPML3 Pore-The A419P mutation locks TRPML3 in an open state and reduces its Ca 2ϩ permeability (7), suggesting that it affects pore behavior. Hence, we characterized in detail the pore behavior of TRPML3(A419P). Fig. 4a confirms our previous finding (7) of constitutive activity of TRPML3(A419P) and low Ca 2ϩ /Na ϩ current ratio even at 100 mM Ca 2ϩ . More important, Fig. 4, b and c, shows significantly reduced shift in E rev when TRPML3(A419P) conducts Na ϩ or Ca 2ϩ . Fig. 4d and supplemental Table 1 show that the main effect of the A419P mutation is an increase in 2 to 21.48 Ϯ 6.74 (p ϭ 0.03). An additional effect of the A419P mutation is reduced selectivity for Ca 2ϩ . The initial E rev for Ca 2ϩ of TRPML3 is 69.1 Ϯ 4.6 mV whereas that for TRPML3(A419P) is 49.7 Ϯ 4.5 mV (Fig. 4d and supplemental Table 2).
The findings with TRPML3(A419P) raised the question of whether the behavior of the TRPML3(A419P) pore is related to the GOF of this mutant. We addressed this question first by examining the pore properties of TRPML3(H283A) that have the same functional and cellular phenotype as TRPML3(A419P) (7). Fig. 4, e-h, shows that although TRPML3(H283A) shows reduced Ca 2ϩ conductance (0.49 Ϯ 0.08 Ca 2ϩ /Na ϩ current ratio), it retains the high Ca 2ϩ selectivity, and the pattern of E rev shift of the Ca 2ϩ current is the same as that of TRPML3. Notably, although the Na ϩ current by TRPML3(H283A) is spontaneously active and does not show time-dependent inactivation, TRPML3(H283A) retains the Ca 2ϩ -dependent inactivation of the Ca 2ϩ current. Hence, the altered properties of the TRPML3(A419P) pore are not simply related to the GOF.
Another proposed mechanism by which the A419P mutation can alter the TRPML3 pore is by disruption of the ␣-helical structure of the fifth TMD (7,9,13). This suggestion is based on the finding that mutation of Ala 419 to glycine, another ␣-helix-disrupting residue, resulted in a GOF in TRPML3 (7,9). We therefore analyzed the pore properties of TRPML3(A419G). Fig. 5a confirms the GOF of TRPML3(A419G) and shows that the A419G mutation reduced the Ca 2ϩ conductance of the channel (0.27 Ϯ 0.05 Ca 2ϩ /Na ϩ current ratio). Although the A419G mutation significantly changed 1 to 1.06 Ϯ 0.12 and 2 to 21.77 Ϯ 2.57 (supplemental Table 1), TRPML3(A419G) retains high Ca 2ϩ selectivity and a large E rev shift of the Ca 2ϩ current (Fig. 5, a-d). Hence, the GOF and pore properties of Because proline introduces kinks in ␣-helices and mutations of residues in the fifth TMD other than Ala 419 to prolines result in GOF in TRPML3 (9), we tested the effect of mutating to proline residues N-and C-terminal to Ala 419 . As reported previously (9), the R414P, C417P, C418P, and G425P mutations caused GOF in TRPML3. Fig. 5, e-h, and supplemental Tables 1 and 2 show that although all mutants are spontaneously active, the extents of their effect on pore properties are different. R414P and G425P had minimal effects, C418P had modest effect, and only C417P behaved similarly to the A419P mutant.
Together, the analysis in Figs. 4 and 5 suggests that the kink caused by A419P specifically alters properties of the TRPML3 pore. Below,  . Measurement of pore diameter. a-c, HEK cells expressing the indicated TRPML3 constructs were exposed to extracellular solutions containing either 150 mM Na ϩ or the indicated organic monovalent cations. d, permeability of the organic cations relative to Na ϩ is plotted against the size of each cation (n ϭ 4 -5).
we characterize in greater details the effect of the A419P mutation on the TRPML3 pore.
A419P Mutation Affects TRPML3 Pore Field Ionic Strength-The results shown in Figs. 4 and 5 indicate that the A419P mutation directly affects the TRPML3 pore. To classify the pore properties of WT and mutants TRPML3, we determined the relative monovalent cation permeability of the channels. To measure the maximal current for each monovalent ion, the bath solution was alternately switched between Na ϩ -free solution and solutions containing a single permeant cation species to calculate the permeability and selectivity relative to Na ϩ . Fig. 6 shows that for TRPML3, the permeability ratios for Li ϩ , K ϩ , Rb ϩ , and Cs ϩ relative to Na ϩ were 1.10 Ϯ 0.04, 0.80 Ϯ 0.02, 0.62 Ϯ 0.02, and 0.41 Ϯ 0.03, respectively. This sequence (Li ϩ ϾNa ϩ ϾK ϩ ϾRb ϩ ϾCs ϩ ) corresponds to Eisenman sequence XI, indicating a strong field strength site. The GOF mutation H283A has the same permeability sequence and thus does not change the pore architecture. By contrast, TRPML3(A419P) poorly discriminates between monovalent cations and resulted in the selectivity sequence of Na ϩ ϾLi ϩ ϾK ϩ ϾRb ϩ ϾCs ϩ , which corresponds to Eisenman sequence X. These results are consistent with the conclusion that the A419P mutation causes a conformational change in the TRPML3 pore to change its architecture.
A419P Mutation Increases the Diameter of the TRPML3 Pore-The change in pore selectivity of TRPML3(A419P) to a weaker field strength can reflect an increased pore size. To test this possibility, we estimated the pore diameter of WT and mutants TRPML3 by measuring their relative permeability to organic monovalent cations of increasing size (19,21,22). We tested the permeability of ions with the following sizes (23): ammonium (A ϩ ), 3.2 Å; methylammonium (MA ϩ ), 3.78 Å; dimethylammonium (DMA ϩ ), 4.6 Å; and tetramethylammonium (TMA ϩ ), 5.6 Å. Fig. 7, a-c, show that TRPML3 and TRPML3(H283A) conducted a small current with MA ϩ but failed to conduct DMA ϩ and TMA ϩ . By contrast, DMA ϩ current was measurable with the A419P mutant. The pore diameter of the channels was estimated using the excluded volume equation (17,19,21,24). Estimates of pore diameter for TRPML3, TRPML3(H283A), and TRPML3(A419P) are 4.73, 4.82, and 5.94 Å, respectively (Fig.  7d). These results suggest that the GOF of TRPML3(A419P) results from about 30% stable expansion of the channel pore.

CONCLUSIONS
The present findings reveal a unique form of pore behavior of the TRPML3 Ca 2ϩ channel. Measurement of Ca 2ϩ current shows that the current undergoes a rapid shift in E rev to result in altered pore properties. This process is regulated by Ca 2ϩ that interacts with a pore site that does not have readily access to cytosolic or extracytosolic face of the channel. The pore lining-charged residues Glu 449 and Asp 459 appear to participate in the potential binding site for the regulatory Ca 2ϩ . Modeling suggests that in the tetramer these residues form two adjacent rings of negative charges next to the cytoplasmic channel vestibule. Ca 2ϩ binding to this site reduces the selectivity and permeability of the channel for Ca 2ϩ . This may serve as a protective mechanism to limit Ca 2ϩ flux by the channel. Maintained activity of TRPML3 is highly toxic, as evident from the varitint-waddler phenotype (4) and from cell death that occurs by expression of TRPML3(A419P) (6, 8 -10, 25) and the TRPML3(H283A) (7) GOF mutations in model systems. The toxicity is caused by excessive Ca 2ϩ influx (6 -10, 25). Hence, the reduction in selectivity and permeability of TRPML3 by Ca 2ϩ allows limited Ca 2ϩ flow through the active channel by rapidly reducing the flow and by increasing the conduction of monovalent ions relative to Ca 2ϩ .
Change in TRP channel selectivity as described here for TRPML3 may be a more general mechanism for regulation of Ca 2ϩ flow by the TRP channels. Many TRP channels are regulated by Ca 2ϩ (1,2), and negative charges are present in the predicted pores of many TRP channels (1). It is thus possible that these negatively charged residues have a function similar to Glu 449 and Asp 459 found here for TRPML3. This possibility is supported by recent findings showing that Asp 646 in TRPV1 (17) and Asp 918 in TRPA1 (19) participate in the alteration of E rev observed after activation of the channels by ligands.