Identification of a single aspartate residue critical for both fast and slow calcium-dependent inactivation of the human TRPML1 channel

Transient receptor potential mucolipin subfamily 1 (TRPML1) is a nonselective cation channel mainly located in late endosomes and lysosomes. Mutations of the gene encoding human TRPML1 can cause severe lysosomal diseases. The activity of TRPML1 is regulated by both Ca2+ and H+, which are important for its critical physiological functions in membrane trafficking, exocytosis, autophagy, and intracellular signal transduction. However, the molecular mechanism of its dual regulation by Ca2+ and H+ remains elusive. Here, using a mutant screening method in combination with a whole-cell patch clamp technique, we identified a key TRPML1 residue, Asp-472, responsible for both fast calcium-dependent inactivation (FCDI) and slow calcium-dependent inactivation (SCDI) as well as H+ regulation. We also found that, in acidic pH, H+ can significantly delay FCDI and abolish SCDI and thereby presumably facilitate the ion conductance of the human TRPML1 channel. In summary, we have identified a key residue critical for Ca2+-induced inhibition of TRPML1 channel currents and uncovered pH-dependent regulation of this channel, providing vital information regarding the detailed mechanism of action of human TRPML1.

Lysosomes, derived from membrane-enclosed compartments of late endosomes, play a critical role in a wide range of physiological functions (1). They can degrade macromolecules and participate in intracellular signal transduction as well as in membrane trafficking (2,3). Consequently, lysosome dysfunction causes many diseases (4,5).
The transient receptor potential mucolipin subfamily 1 (TRPML1) channel protein is an important regulator in many lysosome-dependent cellular physiological events (3, 6 -10). It is widely expressed in every tissue in mammals (3). The gene encoding TRPML1 is MCOLN1, whose mutation causes a lysosomal storage disorder called mucolipidosis type IV (6), which is an autosomal recessive genetic disease with the typical features of psychomotor retardation, corneal opacities, retinal degeneration, strabismus, elevated blood gastrin levels, and achlorhydria (11)(12)(13).
Similar to other TRP 2 channels, TRPML1 is a nonselective cation channel (7) with permeability to Na ϩ , K ϩ , Ca 2ϩ , and Fe 2ϩ but not H ϩ . It typically shows inwardly rectifying wholecell or whole-lysosome currents (14,15). TRPML1 is mainly located in the membranes of late endosome and lysosome (LEL) organelle compartments (16,17). These LEL compartments are filled with high concentrations of ions, such as Ca 2ϩ (7,18). Another feature of both late endosomes and lysosomes is the low luminal pH established by vacuolar-type H ϩ -ATPase (v-ATPase), resulting in a pH of ϳ 5-6 in late endosomes and ϳ 4 -5 in lysosomes (2, 7, 19 -21). Because of the special resident environment associated with both high Ca 2ϩ and H ϩ , TRPML1 channel activity is regulated by both Ca 2ϩ and H ϩ (3,6,7,9,22). However, the molecular mechanism of TRPML1 channel regulation by Ca 2ϩ and H ϩ remains elusive. In this study, using site-directed mutagenesis in combination with a whole-cell electrophysiological patch clamp technique, we identified a key aspartic acid residue that determines both FCDI and SCDI of the human TRPML1 channel in environments with different pH values.

Ca 2؉ inhibits TRPML1 currents
TRPML1 is mainly located in LEL membranes (16,17). Therefore, it is a challenge to perform direct LEL patch clamp experiments to record WT TRPML1 currents. We instead conducted whole-cell patch clamp experiments as described previously (14,15). Briefly, an amino acid substitution (V432P) causes the TRPML1 channel to be widely expressed on cell plasma membranes. The acquisition of TRPML1-V432P currents was proven to show the same electrophysiological properties as that of WT TRPML1. For convenience, hereafter we call the TRPML1-V432P mutant TRPML1 Va (14,15).
It has been reported that the whole-cell currents of TRPML1 Va are inhibited by Ca 2ϩ (15 cro ARTICLE recorded TRPML1 Va currents in the presence or absence of Ca 2ϩ under weakly basic conditions (pH 7.4). Our results showed that Ca 2ϩ inhibited the monovalent cation currents of the TRPML1 channel with an IC 50 of 0.7 mM (Fig. 1, A and B, and Fig. S1A). It was proposed that three aspartic acid residues, Asp-111, Asp-114, and Asp-115, located between transmembrane 1 and transmembrane 2, might be responsible for Ca 2ϩinduced inhibition of TRPML1 currents (22). Therefore, we generated the TRPML1 Va (3DQ) mutant, in which Asp-111, Asp-114, and Asp-115 were mutated to glutamines, to record the whole-cell currents before and after addition of 2 mM Ca 2ϩ . Our results showed that Ca 2ϩ -induced inhibition still occurred in cells expressing the TRPML1 Va (3DQ) mutant ( Fig. 1C and Fig. S1B), indicating that other candidate residues are crucial for Ca 2ϩ -induced inhibition of the TRPML1 channel.

Asp-472 is responsible for Ca 2؉ -induced inhibition of the TRPML1 channel
TRP channel members generally share sequence conservation in their transmembrane domains (12,23). Thus, we explored whether residues vital for Ca 2ϩ -induced inhibition were located in the TRPML1 pore region. Sequence alignment of the pore-forming regions from several TRP channel members, including human TRPML1, human TRPP2, human TRPV1, human TRPA1, and rat TRPV1 (for which structural information is available) (24 -30), showed that this region is highly conserved (Fig. 2A). Therefore, we performed whole-cell patch clamp experiments before and after addition of 2 mM Ca 2ϩ in HEK293T cells expressing each human TRPML1 Va mutant (single amino acid mutation from the asparagine (Asn-

Mechanism of Ca 2؉ inhibition of TRPML1
469) close to the selective filter to the phenylalanine (Phe-477) above the selective filter) at pH 7.4. The mutants N469L, G470A, D471Q, D472Q, M473A, F474L, V475A, T476A, and F477L were generated using a site-direct mutagenesis methodology. As shown in Fig. 2, B-J, and Fig. S2, A-I, all mutants showed Ca 2ϩ -induced inhibition properties after addition of Ca 2ϩ , except for the TRPML1 Va (D472Q) mutant, which displayed significantly increased whole-cell currents ( Fig. 2E and Fig. S2D). Then we examined the dose-dependent effects of Ca 2ϩ on TRPML1 Va (D472Q) mutant currents, showing significant current potentiation in a Ca 2ϩ concentration-dependent manner (Fig. 3A). This result is in sharp contrast with Ca 2ϩ inhibition of the TRPML1 Va currents (Fig. 1A). We further utilized NMDG/Ca 2ϩ -containing solutions to test concentrationdependent effects, which also showed an obvious current potentiation phenomenon (Fig. 3B). Consequently, Asp-472 may play a key role in Ca 2ϩ -induced inhibition of TRPML1 Va currents.

The negative charge of Asp-472 is important for Ca 2؉ -induced inhibition
To elucidate the underlying mechanism, we examined the charge effect of Asp-472 on Ca 2ϩ -induced inhibition of the TRPML1 Va channel. We first mutated aspartic acid to glutamic

Mechanism of Ca 2؉ inhibition of TRPML1
acid at residue 472 to maintain the same negative charge properties. Similar to TRPML1 Va , addition of 2 mM Ca 2ϩ dramatically decreased the current amplitude of TRPML1 Va (D472E) ( Fig. 3C and Fig. S3A). At the same time, we monitored extracellular Ca 2ϩ entry using Fura-2 dye. Our results showed that TRPML1 Va (D472E) mediated the movement of extracellular Ca 2ϩ into cells, in line with the Ca 2ϩ conductance of TRPML1 Va (Fig. 3F). We then mutated aspartic acid to lysine at residue 472 to reverse the negative charge with the positive charge. In the presence of 0 and 2 mM Ca 2ϩ , the whole-cell currents were negligible ( Fig. 3D and Fig. S3B), and neither allowed extracellular Ca 2ϩ entry (Fig. 3F), indicating that the D472K mutant lost its cation conductance. Finally, we mutated aspartic acid to alanine, which has a small and hydrophobic side chain. Surprisingly, 2 mM Ca 2ϩ largely increased the current amplitude of TRPML1 Va (D472A) (Fig. 3E and Fig. S3C), and this mutant also induced extracellular Ca 2ϩ entry (Fig. 3F). Notably, despite the opposite effects of adding Ca 2ϩ to the two mutants TRPML1 Va (D472A) and TRPML1 Va (D472E), the nearly identical elevations in their F 340 /F 380 ratio may be attributed to the similar current levels upon 2 mM Ca 2ϩ addition (Fig.  3, C and E, left panels). Thus, the negative charge of Asp-472 plays an essential role in mediating both the monovalent ion conductance and Ca 2ϩ -induced inhibition of the TRPML1 currents.

Asp-472 plays key role in both FCDI and SCDI of Ca 2؉ inhibition
To further verify that Asp-472 is important for Ca 2ϩ -induced inhibition, we employed NMDG ϩ solution with Ca 2ϩ as the only permeable ion. First, the whole-cell currents were recorded as the control experiment in cells expressing WT TRPML1 Va before and after addition of 2 mM Ca 2ϩ . Fairly small Ca 2ϩ currents were observed after addition of Ca 2ϩ ( Fig.  4A and Fig. S4A). Then the currents of the TRPML1 Va (D472Q) and TRPML1 Va (D472A) mutants were acquired in the absence and presence of 2 mM Ca 2ϩ , both showing much larger Ca 2ϩ currents than the control experiment (Fig. 4, B and C, and Fig.  S4, B and C). We also added a Ca 2ϩ chelator (10 mM EGTA or 10 mM BAPTA) to the intracellular pipette solution to record the whole-cell currents of the TRPML1 Va (D472Q) mutant after addition of 2 mM Ca 2ϩ . The resulting currents are similar to those without a Ca 2ϩ chelator (Fig. 4, D and E, and Fig. S4, D and E), indicating that the Ca 2ϩ current potentiation of the TRPML1 Va (D472Q) mutant is not regulated by intracellular Ca 2ϩ .
Ca 2ϩ channels, such as store-operated calcium channels (31, 32) and voltage-gated calcium channels (33), can be modulated by Ca 2ϩ to cause inactivation called calcium-dependent inactivation (CDI). CDI comprises Ca 2ϩ -dependent fast inactivation (FCDI) and Ca 2ϩ -dependent slow inactivation (SCDI), which have distinct spatial and temporal mechanisms; the former at the millisecond level and the latter at the second level (34,35). However, for the TRPML1 channel, FCDI and SCDI processes have not yet been described. We therefore further explored the CDI of the TRPML1 Va channel. To verify that monovalent cations cannot induce the CDI phenomenon in the TRPML1 Va channel, the currents were first recorded with the 0 Ca 2ϩ solu-tion, showing no FCDI-like currents (Fig. S4F). Then obvious FCDI was observed after replacement of the 0 Ca 2ϩ solution with 2 mM Ca 2ϩ solution (Fig. S4F). Ba 2ϩ caused few FCDI currents (Fig. S4F). Therefore, Ca 2ϩ indeed induces the classic CDI of the TRPML1 Va channel.
To further investigate the CDI of the TRPML1 Va channel, NMDG/Ca 2ϩ solution was used. The current trace of TRPML1 Va was first recorded in the NMDG with 2 mM Ca 2ϩ solution, showing a small Ca 2ϩ current with the classic FCDI phenomenon, with two time constants, including the fast value of 1.29 Ϯ 0.14 ms (n ϭ 7) and the slow value of 12.73 Ϯ 1.51 ms (n ϭ 7) (Fig. 4F). TRPML1 Va also showed a decreased current amplitude, by 18.8% Ϯ 3.8% (n ϭ 7), from the start to the end position (Fig. 4F). Interestingly, under the same conditions, the mutant TRPML1 Va (D472Q) not only presented a much larger current amplitude than TRPML1 Va but also dramatically abolished the FCDI, showing an increased current amplitude, by 2.8% Ϯ 0.7% (n ϭ 6), from the start to the end position (Fig. 4F). This was also true for the mutant TRPML1 Va (D472A), which showed an increased current amplitude, by 12.1% Ϯ 2% (n ϭ 8), from the start to end position (Fig. 4G). In the presence of 10 mM BAPTA in the pipette solution, FCDI abrogation by mutant TRPML1 Va (D472Q) was not affected, showing an increased current amplitude, by 4.5% Ϯ 0.4% (n ϭ 4), from the start to end position, suggesting that abrogation was not attributed to the intracellular Ca 2ϩ increase (Fig. 4H). It has been noted that the time constants of TRPML1 va (D472Q), TRPML1 va (D472Q) with 10 mM BAPTA, and TRPML1 va (D472A) were not calculated because their FCDI current traces were best fit with the linear function. The SCDI process of TRPML1 Va was also studied with elapsed time through one repetitive depolarized potential. As illustrated in Fig. 4I, the current amplitude of the TRPML1 Va channel showed a timedependent decay with a small time constant (ϳ9 s) in the presence of 2 mM Ca 2ϩ . However, the time-dependent decay curve of the current amplitude of the TRPML1 Va (D472Q) mutant was significantly right-shifted, showing a much larger time constant than TRPML1 Va (Fig. 4I). Addition of 10 mM EGTA or BAPTA to the pipette solution had no obvious effect on the decay time constant of the SCDI of the TRPML1 Va (D472Q) mutant (Fig. 4I). These results suggest that Asp-472 is a key site to modulate both FCDI and SCDI of the TRPML1 channel.

H ؉ alleviates Ca 2؉ -induced inhibition
Late endosomes and lysosomes are acidic compartments filled with high concentrations of H ϩ (7,20). Therefore, we investigated the effect of pH on the Ca 2ϩ concentration-dependent inhibition of the TRPML1 Va channel. First, the effect of Ca 2ϩ -induced inhibition was alleviated 5-fold with an IC 50 value of 3.33 mM Ca 2ϩ at pH 4.6 compared with that at pH 7.4 (Fig. 1A). Second, in the absence of Ca 2ϩ , the whole-cell currents of TRPML1 Va were similar at pH 7.4 and pH 4.6 ( Fig. 5A and Fig. S5A), indicating that H ϩ did not potentiate the monovalent cation conductance of TRPML1 Va . In contrast, in the presence of 2 mM Ca 2ϩ , the whole-cell currents of TRPML1 Va largely increased as the solution pH changed from 7.4 to 4.6 ( Fig. 5B and Fig. S5B), indicating that H ϩ alleviated the Ca 2ϩinduced inhibition associated with TRPML1 Va currents. To

Mechanism of Ca 2؉ inhibition of TRPML1
further verify whether the negative charge of Asp-472 is involved in the H ϩ regulation of TRPML1 currents, we used the TRPML1 Va mutant (D472Q). In the presence of 0 and 2 mM Ca 2ϩ , the whole-cell currents of the mutant TRPML1 Va (D472Q) similarly decreased as the pH decreased from 7.4 to 4.6 (Fig. 5, C and D, and S5, C and D), indicating that the residue Asp-472 indeed significantly contributes to H ϩ regulation of the TRPML1 channel.
We then investigated the reciprocal regulation of both CDI and H ϩ . As illustrated in Fig. 5E, the FCDI at the physiological pH 4.6 was significantly slowed compared with the pH 7.4 con-dition. At pH 4.6, TRPML1 Va showed only one time constant value for the fast of 2.19 Ϯ 0.39 ms (n ϭ 4), and the current amplitude decreased by 10.7% Ϯ 1.1% (n ϭ 4) (Fig. 5E). We further compared SCDI processes at pH 7.4 and pH 4.6. Surprisingly, SCDI was completely abolished at pH 4.6 (Fig. 5F). In summary, H ϩ can alleviate the Ca 2ϩ -induced inhibition of the TRPML1 channel.

Discussion
TRPML1 is mainly located in late endosomes and lysosomes, whose compartments are filled with high concentrations of

Mechanism of Ca 2؉ inhibition of TRPML1
Ca 2ϩ and H ϩ (7,20). Its special dwelling environment determines the regulation of its functional activity by both Ca 2ϩ and H ϩ (14,15,22). In this study, we identified a key residue, Asp-472, that is important for Ca 2ϩ -induced inhibition of TRPML1 channel currents. We showed that the TRPML1 Va (D472E) mutant had similar ion channel properties as WT TRPML1 Va . However, the TRPML1 Va (D472K) mutant, which included a positively charged amino acid substitution, completely abolished the channel conductance of TRPML1. The neutral amino acid substitution mutant TRPML1 Va (D472A) abrogated Ca 2ϩinduced inhibition of the TRPML1 channel activity but retained the cation conductance of TRPML1.
Calcium-dependent inactivation is an important negative feedback regulation mechanism to prevent excessive Ca 2ϩ entry into the cytoplasm. FCDI occurs rapidly after Ca 2ϩ influx, whereas SCDI occurs upon slow global Ca 2ϩ rise (34,36,37). Generally, there are different regulators to manage FCDI and SCDI. Among these regulating factors, Ca 2ϩ -bound calmodulin is the most widely reported regulator (33, 35, 38 -42). However, the TRPML1 channel showed unconventional behavior in both FCDI and SCDI regulation. The single residue Asp-472 itself can regulate both the FCDI and SCDI of the TRPML1 channel. Ca 2ϩ -dependent calmodulin regulation seemed not to be involved in the regulation of the TRPML1 channel because the presence or absence of the Ca 2ϩ chelator BAPTA caused no significant effects on both FCDI and SCDI. The FCDI induced by Ca 2ϩ can be well-explained by the special localization of Asp-472 near the selectivity filter (24), parallel to the spatial and temporal mechanism of FCDI occurrence (34,36,43). Strikingly, Ca 2ϩ also facilitated the SCDI process through residue Asp-472 because the mutant TRPML1 Va (D472Q) significantly prolonged the decay time constant of SCDI. Importantly, under physiological pH 4.6 conditions, we showed that Asp-472 also played a critical role in the H ϩ alleviation of the Ca 2ϩ -induced inhibition of TRPML1 channel currents. The D472Q mutant fully abrogated the H ϩ attenuation effects on Ca 2ϩ -induced inhibition. Further studies showed that H ϩ was able to not only significantly delay FCDI but also fully abolish SCDI. This unusual regulation of CDI can be further emphasized compared with other TRP channels adopting the Ca 2ϩ -bound calmodulin regulation approach (44 -47). This is probably an important molecular regulation mechanism for TRPML1-dependent cellular physiological events. Because lysosomes and late endosomes are filled with highly concentrated Ca 2ϩ and H ϩ (7,20), the opposite regulation between Ca 2ϩ and H ϩ is required to synergistically manage TRPML1 Ca 2ϩ release. These results together indicated that the residue Asp-472 is indispensable for both TRPML1 ion conductance and its regulation by Ca 2ϩ and H ϩ .
Based on these results, we propose a possible molecular model for the regulation of the TRPML1 channel current by Ca 2ϩ and H ϩ (Fig. 5G). Under weakly basic conditions (pH 7.4), Asp-472 is deprotonated and able to effectively aggregate Ca 2ϩ to form a positively charged microdomain, causing obvious CDI of the TRPML1 channel, thereby inducing Ca 2ϩ inhibition of conductance. Additionally, the positively charged microdomain may also interfere with the access of monovalent cations as well as Ca 2ϩ to the selective filter because of repulsive inter-actions. Consequently, a Ca 2ϩ -induced inhibition phenomenon is observed. However, under acidic conditions (pH 4.6), Asp-472 is protonated and weakens the binding ability of Ca 2ϩ . The FCDI process, especially SCDI, is clearly inhibited. Therefore, access of monovalent cations as well as Ca 2ϩ to the selective filter is similar, with no Ca 2ϩ -induced inhibition phenomenon in the TRPML1 channel.
A previous report showed that Ca 2ϩ and H ϩ can regulate TRPML1 channel activity through three aspartic acid residues (D111Q, D114Q, and D115Q) (22). However, these three residues seem to be dispensable for the regulation of the TRPML1 channel by Ca 2ϩ and H ϩ in our system. The discrepancy may require further clarification in the future. Notably, we used a constitutively activated TRPML1 Va mutant (14) to conduct this research. Although the basic electrophysiological properties of this mutant are identical to those of WT TRPML1 (15), it will be necessary to perform relevant direct late endosome or lysosome patch clamp experiments under physiological conditions in the future.

Plasmids
The human TRPML1 gene (accession number NM_020533) was synthesized by Genewiz and cloned into the pcDNA-EGFP vector in the BamH1 and Xho1 restriction sites using transcription PCR. The mutants, including TRPML1 Va , TRPML1 Va  Table S1.

Whole-cell recordings and data analysis
The recording pipettes were pulled from borosilicate glass using a P-97 glass microelectrode puller (Sutter Instrument) and polished with an MF-830 (Narishige). The pipettes had a resistance of 3-5 megaohms after being filled with the internal recording solution containing 120 mM cesium methanesulfonate, 4 mM NaCl, 2 mM MgCl 2 , 2 mM Na 2 -ATP, 10 mM EGTA, and 20 mM HEPES (pH adjusted to 7.2 with CsOH). This internal solution was used for all non-NMDG solution recordings. After establishment of the whole-cell configuration, the currents were recorded using an Axopatch 700B amplifier (Molecular Devices) and digitized using a Digidata 1550A (Molecular Devices). The voltage protocol included 50-ms voltage steps to Ϫ100 mV from a holding potential of 0 mV, followed by a voltage ramp increasing from Ϫ100 to ϩ 100 mV in 50 ms with a frequency of 0.5 Hz. A 100-ms step potential from 0 mV to

Mechanism of Ca 2؉ inhibition of TRPML1
Ϫ100 mV was applied for investigation of the FCDI process. All currents were sampled at 10 kHz and filtered at 2 kHz by the low-pass filter. Current recordings were acquired through pClamp software (Molecular Devices). The standard extracellular 0 Ca 2ϩ recording solution contained 153 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM glucose, and 20 mM HEPES (pH 7.4 adjusted with NaOH). The standard extracellular Ca 2ϩ -containing recording solutions were made in different concentrations by diluting the 1 M CaCl 2 stock solution in the standard extracellular 0 Ca 2ϩ recording solution. The pH 4. To eliminate the obvious external currents recorded with the above internal pipette solution, we applied the following pipette solution containing 160 mM NMDG, 20 mM HEPES, and 10 mM glucose (pH 7.2 adjusted with HCl) for all NMDG-containing solution recordings. 10 mM glucose was replaced with 10 mM EGTA or BAPTA for the EGTA-or BAPTA-containing pipette solutions (pH 7.2 adjusted with HCl). The solution exchange was performed using a peristaltic pump and was accomplished within several seconds. Data were analyzed using pClamp and Origin software (OriginLab Corp.). All experiments were conducted at room temperature. The method for the normalization of the recorded currents was as follows. The very beginning current density was set as Ϫ1.0, the current densities during the measurement were displayed as the ratio of the very beginning current density to obtain the normalized current densities, and these densities were plotted as a function of the recording time. The time constants of FCDI were obtained through the nonlinear function fit (ExpDec 1 or ExpDec 2) in Origin 9.0 software. Statistical analysis was executed in SPSS Statistics 20 (Statistical Product and Service Solutions, IBM Corp.) by one-way analysis of variance. p Ͻ 0.05 represents statistical significance. Data points represent the mean Ϯ S.E.

Intracellular Ca 2؉ measurements
HEK293T cells expressing the TRPML1 Va , TRPML1 Va (D472E), TRPML1 Va (D472K), and TRPML1 Va (D472A) mutants were plated on glass-bottom dishes coated with poly-L-lysine (Sigma-Aldrich). Twenty-four hours after transfection, the cells were loaded with 5 M Fura-2/AM (Invitrogen) in the standard extracellular 0 Ca 2ϩ recording solution (pH 7.4) at room temperature for 30 min. After loading, the cells were transferred to Fura-2/AM-free solution for 30 min. Fluorescence imaging was undertaken at room temperature using a Leica DMI6000B microscope with the LAS software before and after addition of 2 mM Ca 2ϩ . Consecutive excitation occurred at 340 and 380 nm every 2 s, and the emission was collected at 510 nm. The intracellular Ca 2ϩ concentration change is shown as the 340/380 ratio. Data points are shown as the mean Ϯ S.E.
Author contributions-G. W., X. Y., and Y. S. data curation; G. W. investigation; G. W. and Y. S. writing-original draft; X. Y. and Y. S. supervision; X. Y. and Y. S. project administration; Y. S. conceptualization; Y. S. funding acquisition; Y. S. writing-review and editing.