A C-terminally truncated mouse Best3 splice variant targets and alters the ion balance in lysosome-endosome hybrids and the endoplasmic reticulum

The Bestrophin family has been characterized as Cl− channels in mammals and Na+ channels in bacteria, but their exact physiological roles remian unknown. In this study, a natural C-terminally truncated variant of mouse Bestrophin 3 (Best3V2) expression in myoblasts and muscles is demonstrated. Unlike full-length Best3, Best3V2 targets the two important intracellular Ca stores: the lysosome and the ER. Heterologous overexpression leads to lysosome swelling and renders it less acidic. Best3V2 overexpression also results in compromised Ca2+ release from the ER. Knocking down endogenous Best3 expression in myoblasts makes these cells more excitable in response to Ca2+ mobilizing reagents, such as caffeine. We propose that Best3V2 in myoblasts may work as a tuner to control Ca2+ release from intracellular Ca2+ stores.

The Bestrophin family was initially regarded as a candidate Ca 2+ activated Cl − channel (CACC) [1][2][3][4] . In 2008, people showed that the Anoctamin was the typical CACC 5,6 . Thus, the in vitro roles of Bestrophins became more confusing at that time. Bestrophin was also revealed to be a putative candidate volume-regulated ion channel 7,8 , which conducts HCO3 − or gamma amino butyric acid (GABA) [9][10][11] . Recently, the crystal structure of Bestrophin was generated and revealed a clear channel core for Cl − in human Best1 12 . The crystal structure of human Best1 also displayed a Ca 2+ clasp or Ca 2+ bowl in aa 300-304 12 . However, the successful generation of crystals of the bacterial Bestrophin homolog revealed that it is a Na + channel 13 . The Bestrophin family may have diverse or versatile roles.
The Bestrophin family has 4 members in humans and 3 or 4 Bestrophin paralogs in other animals 14 . BEST1 or VMD2 (vitelliform macular dystrophy 2) has been linked to an autosomal dominant form of juvenile blindness known as Best vitelliform macular dystrophy 15,16 , which was regarded as a type of lysosome storage disease. Unfortunately, people still do not understand the direct correlation between Best1 and the lysosomes 14 . However, a role for Best1 in maintaining intracellular Ca 2+ balance has been proposed in RPE cells 17 . Best2 was shown to be involved in generating aqueous flow and controlling the intraocular pressure in eyes 18,19 . Best3 is very abundantly expressed in muscle tissues, along with diverse splice variants 20 , but their exact roles in muscles remain largely undefined.
In this report, a natural splice variant (Best3V2) of mouse Best3 without the long C-terminus (an alternative splice variant after amino acid L360 that introduces a premature stop codon at amino acid 364 due to a frame shift) was cloned from mouse myoblasts. This splice variant was capable of inducing cell vacuolization in cultured cell lines. Therefore, these vacuoles can help track their recruitment to certain intracellular organelles and facilitate the monitoring of the factors contributing to vacuolization. More importantly, this splice variant provides new insights into its role in myoblasts.

Figure 1. The expression of Best3 in myoblasts, muscle tissues and fibroblasts. (A)
The expression of mouse Best3 in NIH3T3, myoblasts differentiated from satellite cells, myoblasts differentiated from C2C12 (4 days after confluency), skeletal muscle (SK) and heart was assessed by RT-PCR. Two sets of primers for mouse Best3 sequence (NM_001007583) were used. Alpha actin was used as an internal control. (B) Schematic presentation of full-length mouse Best3 and 2 alternatively spliced variants V1 and V2 (Best3V2). (C) Distinct distribution patterns of full-length Best3 and Best3V2. The ER was stained with PDI, the membrane ruffles were stained with an anti-caveolin 1 antibody, and some focal adhesions were visualized with GFP_FAK. Scale bar, 20 μ m. (D) The expression of endogenous Best3 in differentiating myoblasts derived from satellite cells. The cells were also co-stained with a tropomyosin antibody to determine the levels of endogenous Tropomyosin. (E) The size of the endogenous Best3 proteins and its targeting to the plasma membrane in differentiating C2C12 myoblasts. Plasma membrane targeting was monitored via cell surface protein biotinylation. Whole proteins of skeletal muscle (SK) and C2C12 myoblasts were used as controls. (F) Best3 expression was analyzed in myoblasts differentiating from C2C12 cells on day 0 (D0), day 2 (D2) and day 4 (D4) by RT-PCR. M, DNA marker. The experiments were repeated 3 times.
to the plasma membrane. The Best3 splice variants were expressed at very high levels at the start of C2C12 cell differentiation. However, its role in myoblasts is still unknown.
The Best3V2-induced vacuoles are swelling lysosomes. Vacuoles may not be the physiological consequence of endogenous Best3V2 expression, but the vacuoles provide clues that can be used to track its subcellular targeting. Live cell imaging showed that Vac-V2s were lysosomes, because the vacuoles of the HeLa cells were filled with small LysoTracker Red granules ( Fig. 2A). To characterize the pH within vacuoles, we stained these cells with acridine orange (AO) or the biphasic lysosome dye Lysosensor DND160. AO labeling showed that the lumens of the Vac-V2s were acidic and also exhibited a strong green shift, a sign of a less acidic lumen (Fig. 2B). Staining with the biphasic pH-sensitive dye Lysosensor DND 160 also confirmed that the Vac-V2s were less acidic than the lysosomes (Fig. 2C). We then labeled the vacuoles with endogenous Rab7a or Lamp2 to discriminate the origin of the Vac-V2s. The fluorescence images showed that the membranes of most Vac-V2s in HeLa cells were strongly stained by Rab7a and, to lesser extent, by Lamp2 (Fig. 2D). Noticeably, Best3V2_HA was located in the nuclear envelope and in the vacuole membrane, in contrast to full-length Best3, which was mainly located in the plasma membrane folds shown in the preceding image. A considerable number of Lamp2-positive granules were negative for Best3V2s, suggesting that primary lysosomes were not the origin of the Vac-V2s. Rab7-positive vesicles are usually late endosomes or lysosomes (including endosome-lysosome hybrids and autolysosomes) [21][22][23] . The TEM images showed that the Vac-V2s contained electron dense lysosome granules with undigested membranous structures, further confirming the compromised digestion in the less acidic lumen of the Vac-V2s in HeLa cells (Fig. 2E). Noticeably, the TEM images also showed that the Best3V2-transfected NIH3T3 cells were filled with a large number of multilamellar lysosomes instead of vacuoles, which may explain why the NIH3T3 cells contained fewer Vac-V2s than the HeLa cells (Fig. 2F, Fig. 1S). Therefore, the cells that underwent vacuolization or formed multilamellar lysosomes upon Best3V2 overexpression varied depending on the cell types. However, the common finding for both cell types was compromised lysosome digestion, and lysosomes were the target of BestV2 in both cell types. Vacuoles or multilamellar bodies may not be the physiological consequence of endogenous Best3V2 expression but may result from overexpression. Oil red O staining further confirmed that a number of incompletely digested lipids accumulated within the Vac-V2s in HeLa cells (Fig. 2G). However, the reduced acidification in lysosomes induced by Best3V2 did not affect the normal maturation process of the lysosomal enzyme cathepsin D (CTSD) in HeLa cells (Fig. 2H), suggesting that the enzyme trafficking and maturation through the Golgi apparatus, endosome and ultimately to the primary lysosome was not impaired by Best3V2. In this context, the primary lysosomes were less likely to undergo vacuolization, and the endo-lysosome hybrid may be one candidate target. More interestingly, the phospholipid-binding domain on the cytosolic loop of Best3V2 was able to bind PI 3 P and PI (3,5) P 2 (Fig. 2I), which are two specific phospholipids of the endo-lysosome system 24 . Recombinant aa 114-153 of Best3V2, which are located within the cytosolic loop between TM2 and TM3, were able to bind to phospholipids, such as PI 3 P, PI (3,4) P 2 , PI (3,5) P 2 , PI (3,4,5) P 3 and PA. This result further implies that Best3V2 has the ability to bind endosome-lysosome vesicles. A double mutation at aa 114-153 (R125Q and R126Q) abolished phospholipid binding (data not shown).
In summary, regarding the origin of the Vac-V2s, the evidence indicates that they originate from the lysosomes, particularly the endo-lysosome hybrids.
Failure to reform lysosome-endosome hybrids leads to vacuolization. We next monitored whether these vacuoles were generated from late endosome-lysosome hybrids. The fusion process of endosome and lysosome should be blocked or enhanced to block or increase the formation of endosome-lysosome hybrids, respectively. Rab7 is the key protein that mediates this fusion process 24 . Therefore, dominant negative mutants (Rab7a(T22N), Rab7a(N125I)) and a constitutively active mutant (Rab7a(Q67L)) of Rab7a and shRNA constructs (shRab7a) were applied to either block or enhance Rab7a-mediated fusion between endosomes and lysosomes. Clc3s was also used as a parallel control to show the effect of blocking or enhancing late endosome-lysosome fusion. Unfortunately, our data showed that blocking or enhancing Rab7a activity had contrasting effects on the formation of Vac-V2s and Vac-Clc3s (Fig. 3A). Rab7a inhibition decreased the size of the formed Vac-V2s ( Fig. 3B-D). In contrast, Rab7a inhibition significantly increased the size of the Vac-Clc3s and Rab7a(Q67L)-enhanced Rab7a-mediated fusion abolished the generation of large diameter Vac-Clc3s, but not the small-sized Vac-Clc3s. The results suggest that the large sized Vac-Clc3s are exclusively derived from fusion among endosomes, and upon fusion with lysosomes, the Vac-Clc3s are degraded by the lysosomes. These data clearly indicate that the Vac-V2s are mainly derived from endosome-lysosome hybrids, whereas the Vac-Clc3s are exclusively generated from endosomes. Enhancing the fusion of late endosomes with lysosomes prevented fusion among Vac-Clc3s and thus eliminated the formation of large Vac-Clc3s. This result is different from a previous report showing that the vacuoles generated by Clc3s were lysosomes 25 . Furthermore, TEM confirmed the fusion of a large Vac-Clc3 with a small, emerging Vac-Clc3 (Fig. S2).
Therefore, by comparing the generation of Vac-V2s and Vac-Clc3s by blocking or enhancing endosome and lysosome fusion, we show that the Vac-V2s are derived from endosome-lysosome hybrids and Vac-Clc3s are formed from endosomes.

ER vacuoles induced by mutating the phospholipid binding domain or by adding an ER retention signal to Best3V2 and their properties compared with vacuoles induced from lysosomes.
Then, we determined whether mutation of the phospholipid binding domain, aa 114-153, or the putative channel core of Best3 can abolish the swelling of endosome-lysosome hybrids, because aa 114-153 were shown to bind PI3P and PI(3, 5)P2. Moreover, according to the channel structure of human Best1 or bacterial Bestrophin 12,13 , a segment of transmembrane domain 2 (TM2) formed the putative conducting channel core. Interestingly, mutation of transmembrane domain 2 (TM2) (TM2 (L82N)) and Best3V2 (Δ TM2) derived from a natural alternative splice variant, Best3V1, without TM2 abolished the formation of bright round vacuoles in HeLa cells (Fig. 4A). Mutation of the phospholipid binding domain, aa 114-153 (Best3V2(R125Q, R126Q)), abolished the formation of the bright round vacuoles, and, instead, low-diopter vacuoles formed (morphology was shown in Fig. 4B). More importantly, the low-diopter vacuoles were not blocked by the H + pump blocker BFA. However, the bright vacuoles induced by Best3V2 or other vacuoles derived from acidic endosome-lysosome systems, i.e., Vac-Clc3s and Vac-VPS4(E228Q), can be effectively blocked by BFA, suggesting that the two morphologically distinct vacuoles are derived from different organelles in HeLa cells (Fig. 4C).
We then sought to determine which organelles were induced by Best3V2 (R125Q, R126Q) and then generated the low-diopter vacuoles. We ruled out the contribution of the acidic endo-lysosome system, and the ER, a neutral pH organelle, became our focus. We then added an ER retention signal to Best3V2 and tested whether the retention of Best3V2 in ER also induced the formation of low-diopter vacuoles similar to Best3V2(R125Q, R126Q). Two ER retention signals were introduced. One is located in the cell cytosol and has the _KKXX consensus sequence 26 . The other is within the ER lumen and has the well-known _KDEL consensus motif 27 . According to the topology of Best1, the two termini are facing the cell cytosol 28,12,13 . Therefore, we fused _KKXX (_KKSA) to its Scientific RepoRts | 6:27332 | DOI: 10.1038/srep27332 C-terminus to obtain Best3V2_KKSA. Our data showed that _KKSA was indeed able to eliminate the formation of the bright round vacuoles, but not _KDEL (Fig. 4D,E). For Best3V2_KKSA, instead of bright round vacuoles, a large number of low-diopter vacuoles were formed in the perinuclear region (Fig. 4E), suggesting that the vacuoles induced by retention in the ER were similar to Best3V2 (R125Q, R126Q). In contrast, Clc3s_KDEL with the luminal retention signal nearly eliminated vacuolization and did not form any low-diopter vacuoles, but Clc3s_ KKSA had a similar vacuolization capacity as Clc3s, suggesting that its C-terminus was within ER luminal side, which is opposite to full-length Clc3. According to the topology of Clc3s, both of its termini should be inserted toward the cytosolic side 28 . The result strikingly suggests that Clc3s is inserted in the organelle membrane in a reverse orientation when the N-terminal sequence is naturally deleted, and, alternatively, the C-termini of Clc3s turns to face the ER lumen. Therefore, the natural splice variant Clc3s completely alters the insertion topology in the organelle membrane, which enriches the function of Clc3. This interesting and unique phenomenon for channel genes extends their functions by undergoing alternative splicing to change their insertion topology into membrane. Notably, Clc3s_KDEL retention in the ER did not induce any visible vacuolization, suggesting that not all ion channels display channel activity in ER when they are confined to this organelle. This parallel comparison between the ER retention of Clc3s and Best3V2 further emphasizes that the low-diopter vacuoles formed by Best3V2 (R125Q, R126Q) and Best3V2_KKSA originated from the ER and resulted in an ion imbalance in ER. However, Clc3s retention in the ER is not capable of inducing the formation of any visible vacuoles in the ER. Additionally, our data also suggest that aa 114-153 are critical for Best3V2 targeting and binding to the lysosome.
However, the anion channel blockers NFA, NPPB and DIDS did not substantially inhibit the formation of either the bright or low-diopter vacuoles (Fig. 4F,G). Although only NPPB inhibited both types of vacuoles (p < 0.05), it is not clear whether its inhibition is direct or indirectly mediated by other anion channels. As expected, the formation of the low-diopter vacuoles was not blocked by the V-type H + pump blocker BFA, whereas the formation of the bright vacuoles was completely blocked by BFA. The results further suggest that these low-diopter vacuoles are derived from the ER. Among the ER Ca 2+ channel or Ca 2+ pump blockers, tharpsigargin (TG) very significantly increased the sizes of both vacuoles (p < 0.01), and the IP3R blocker 2-APB decreased the sizes of both types of vacuoles (p < 0.05). The inhibition of the mitochondrion Ca 2+ uniporter by RU360 only slightly decreased the size of both types of vacuoles (statistically not significant). These results imply that the cytosolic free Ca 2+ level exerts an impact on the sizes of both types of vacuoles, because the TG treatment is able to increase the cytosolic free Ca 2+ level, whereas 2-APB decreased the cytosolic free Ca 2+ level by blocking the release of the Ca 2+ stores through IP3R [29][30][31] . This evidence shows the Ca 2+ sensitivity of Best3V2.
These results show that Best3V2 also induces ER swelling when it is confined to the ER, but Clc3s retention in the ER does not induce ER swelling. Amino acids 114-153 of Best3V2 comprise the key domain mediating its ability to bind lysosomes. The traditional anion channel blockers do not effectively block the formation of either the lysosome vacuoles or the ER vacuoles. However, both types of vacuoles display a Ca 2+ -dependent property.
In summary, Best3V2 is capable of exerting impacts on ER while Clc3s is not. Traditional anion channel blockers seem to be inefficiency in blocking both the vacuolization of ER and lysosome induced by Best3V2.
Best3V2 reduces the free intracellular Ca 2+ mobilization, which prevents premature excitation and contraction in differentiating myoblasts. We have shown that the ER can be altered by the retention of Best3V2 in this organelle. We then wondered whether Best3V2 without the ER retention signal impacted the ER ion balance, because BestV2 is synthesized in and transported by the ER. More importantly, Best3V2 was also shown to be specifically located in the ER around the nuclear envelope, as shown in Fig. 2D. Additionally, Best1 has been proposed to be able to alter ion homeostasis in the ER [32][33][34][35] . Therefore, we tested the alterations in ER Ca 2+ release by overexpressing Best3V2 to determine whether the ER ion balance is altered. NIH3T3 cells were used for this experiment because these cells endogenously express the Best3 splice variants and thus have a more similar endogenous cell environment than HeLa cells. As expected, in the absence of the ER retention signal, we still detected a striking decrease in Best3V2-mediated ER Ca 2+ release in bradykinin-stimulated NIH3T3 cells ( Fig. 5A-C).
We then determined the role of endogenous Best3V2 in ER Ca 2+ release. The expression of endogenous Best3V2 was knocked down in C2C12 myoblasts. To knockdown endogenous Best3, we constructed 2 shRNAs (shBest3-I and shBest3-II) against mouse Best3. The knockdown efficiency of shBest3-I was approximately 80% and that of shBest3-II was approximately 50%, respectively, which was validated by real-time PCR (data not shown). Unfortunately, it is difficult to get the myoblasts (infected with shBest3-I or shBest3-II) to attach to the laminin-coated coverglasses on day 5 postconfluency when they are treated with even low concentrations of caffeine (0.1 mM) to trigger ER Ca 2+ release (Fig. 5D). In contrast, the Scr control was better at resisting detachment. We then tested whether this detachment was attributed to Ca 2+ -induced cell contraction. The peak Ca 2+ release was significantly increased by shBest3-I upon caffeine stimulation (Fig. 5E). Endogenous Best3V2 may tune the amplitude of Ca 2+ release from intracellular Ca 2+ stores, such as from the perinuclear ER or lysosomes, which may prevent excessive fluctuations of the free Ca 2+ concentrations in differentiating myoblasts. Unfortunately, we did not observe a visible alteration in C2C12 myoblast differentiation following Best3 knockdown, i.e., morphology and cell growth (data not shown). For differentiating myoblasts, we postulate that premature excitation and contraction may be objectively reduced by Best3V2 at certain differentiation stages. In addition, it may facilitate more stable attachment to the extracellular matrix during elongation. However, further detailed investigations are required to determine its exact physiological roles in myoblasts and mature muscles.
In summary, Best3V2 without the ER retention signal still exerts an impact on the ER in both NIH3T3 cells and C2C12 myoblasts. In C2C12 myoblasts, the endogenous Best3 splice variants inhibited excitation and contraction by regulating the cytosolic Ca 2+ levels.

Discussion
Our imaging data showed that Best3V2 alters the H + gradient in lysosomes, which may be accompanied by alterations in the Ca 2+ levels. These alterations impair lysosome-endosome hybrid reformation. For example, the luminal Ca 2+ and V-type H + ATPases have been suggested to be required for the condensation of the content and retrieval of components from endosome-lysosome hybrids 36 . Dysregulation of lysosomal channels results in the pathogenesis of many lysosome storage diseases 37,38 . Loss-of-function mutations in the lysosomal Cl − /H + exchanger ClC-7 or its monotopic accessory protein Ostm1 result in osteopetrosis 39,40 . ClC-7 plays a prominent role in V-ATPase-mediated acid secretion 41 . Lysosomal Ca 2+ dysfunction caused by an NPC1 mutation has been linked with human lysosome vacuolization 42,43 . Lysosome swelling has also been observed following the mutation of non-selective cation channel mucolipin-1 in lysosome, which results in a failure to reform lysosomes from hybrids and thereby causes the autosomal recessive disease mucolipidosis type IV, a lysosomal channelopathy leading to lipid storage 44,45 .
The ER and lysosome are the two main intracellular Ca 2+ stores. The well-known NAADP-induced Ca 2+ signals are from the lysosome Ca 2+ store. The ER or SR is the major intracellular neutral Ca 2+ store. The local Ca 2+ signal from the plasma membrane or acidic Ca 2+ stores are amplified by neighboring IP 3 and RYRs on the ER by Ca 2+ -induced Ca 2+ release (CICR) to mediate a larger global Ca 2+ signal 46,47 . However, the Ca 2+ release from the lysosome is also affected by the ER or SR. For example, the Ca 2+ response to NAADP in the lysosome is suppressed by interfering with ER Ca 2+ sequestration 48 . Therefore, the targeting of Best3V2 to both Ca 2+ stores directly placed the 2 important intracellular Ca 2+ stores under its control. Its physiological significance in myoblasts and muscles is still unknown. However, two putative consequences can be postulated. One is the inhibition of motility or contraction by reducing Ca 2+ release from the SR. The other is the regulation of the Ca 2+ levels Scientific RepoRts | 6:27332 | DOI: 10.1038/srep27332 within certain intracellular organelles. For the former, when mononuclear myoblasts merge into multinuclear myoblasts, attach to mature muscle fibers and elongate into long muscles, premature contraction may not be necessary for certain differentiation steps. For the latter, attention must be paid to determine how excess Ca 2+ levels can be prevented in certain organelles, such as the nucleus, upon muscle contraction. Evidence shows that the nuclear calcium level is finely regulated and involves the outer nuclear envelope [49][50][51] . A considerable number of studies have shown that the same signals that regulate "excitation-contraction coupling" (EC-coupling) are also capable of regulating certain gene expression programs, a new paradigm coined "excitation-transcription coupling" (ET-coupling) [52][53][54] . Dysregulation of ET coupling led to problems with cell growth and gene expression. For example, in the adult heart, neurohormonal/mechanical stress enhances ET coupling, resulting in hypertrophy, the reexpression of the fetal gene program, and alterations in ion channel and transporter expression (see review 55 ). In this context, the abundant distribution of Best3V2 on the nuclear envelope may form a barrier or cushion around the nuclei to reduce excess, local Ca 2+ fluctuations upon excitation and contraction. Best3-deficient animal models are needed to verify this hypothesis. Additionally, the role of Best3V2 in muscle contraction (exercise intensity, duration, and frequency) should also be investigated further. In vivo, contraction may be finely regulated by strictly controlling the level of endogenous Best3V2. Under native conditions, Best3V2 promotes Ca 2+ release from the ER at an appropriate level.
In the ER, the distribution of Ca 2+ channels in the ER membrane is discontinuous, with clustering in certain locations and different Ca 2+ concentrations in sub-compartments [56][57][58] . However, the ER luminal Ca 2+ movements show that this organelle system is one continuous Ca 2+ store with a Ca 2+ tunnel effect 59,60 . The location of Best3V2 on the nuclear envelope may alter the Ca 2+ levels in the adjacent ER or even throughout the ER.
However, our imaging data imply that the Best3V2-induced phenomena in the lysosome and ER cannot be explained by Best3V2 having a role as an anionic channel. For example, because Best3V2 works as an anion channel, Best3V2 overexpression should theoretically increase the H + levels in the lysosome and Ca 2+ levels in the ER, which in turn will render the lysosome more acidic and initiate higher peak Ca 2+ release in the ER. Our results show the opposite trend. This observation can be explained by 2 possible roles for Best3V2. One explanation is that, unlike full-length Best3, Best3V2 works as a cation channel, which is similar to the cation conductance of the bacterial Bestrophin. The other possible explanation is that Best3V2 loses channel activity, but it indirectly influences other cation channels in the lysosomes and ER. However, additional studies are required to characterize its ion conductance within intracellular organelles or in artificial lipid bilayers.
In conclusion, this manuscript provides evidence that a rare splice variant of Best3 targets both the lysosome and the ER around the nuclear envelope. The splice variant is mainly expressed in myoblasts and muscles, and functions to regulate Ca 2+ release from cytosolic stores. Its inhibition of Ca 2+ release may help prevent premature contraction in myoblasts or prevent Ca 2+ overloading to the nucleus in mature muscles. However, its exact physiological roles in vivo require further verification in animal models. The expression of Best3V2 in myoblasts and muscle cells suggests that the Ca 2+ regulation in these cells is complicated, and additional novel channels are involved in its fine control.
Fluorescence Microscopy. For the immunofluorescence analysis, the cells were plated onto coverslips coated with 0.1% gelatin (Sigma Aldrich) in 12-well plates and fixed with 3% paraformaldehyde (PFA) (Songon, Shanghai) at room temperature. The fixed cells were permeabilized with cold ethanol, phosphate-buffered saline (PBS) containing 0.2% Triton X-100, or PBS containing 0.5% saponin. Nonspecific binding sites were blocked by incubating the coverslips with 3% BSA for 20 min. The fixed cells were incubated with primary antibodies for 1-2 h and with the appropriate Alexa Fluor-conjugated secondary antibodies for 1 h before mounting. The slides were visualized under a confocal or fluorescence microscope. The images were further analyzed with the FV10-ASW 3.0 viewer software.
For live cell imaging, the cells were grown in 35-mm glass-bottom dishes (MatTek or Nest) and staged into a heated incubation chamber with CO 2 . The cells were imaged using an Olympus FV-10 confocal microscope or a fluorescent microscope, and the images were acquired with the FV10-ASW 3.0 viewer (Olympus) or Metamorph (Molecular Devices) software. The image stacks or movie files were further processed with ImageJ software. A phase microscope was used for vacuole imaging and counting. The vacuole sizes were calculated with blind method in digitally enlarged images on a computer screen using a scale ruler. Biotinylation of the proteins on the plasma membrane surface. Plasma membrane targeting was monitored via cell surface protein biotinylation and isolation with Cell Surface Protein Isolation Kit (Pierce, Thermo Fisher Scientific, USA), according to the manufacturer's instructions. Four days after the cells reached confluency, the C2C12 cells were exposed to membrane-impermeable Sulfo-NHS-LC biotin for 30 min on ice. The biotinylated membrane proteins were solubilized and affinity-purified with streptavidin beads. The purified proteins were subjected to Western blotting with the Best3 antibody, followed by the horseradish peroxidase-conjugated secondary antibody. The protein concentration was assayed by the Bradford protein assay and biotinylated membrane proteins for myoblat and mature skeletal muscles were equally loaded for Western blotting.

Cell lysis and Western blots.
For the immunoblot, the cells were placed on ice, washed twice in cold PBS, and lysed in mild RIPA buffer [50 mM Tris, pH 7.4, 150 mM NaCl, 0.5 mM DTT, 2mM EDTA, and 1% NP-40] with protease inhibitors. The Western blots were performed by separating the samples on SDS-PAGE gels, which were transferred to PVDF membrane for blotting. The concentration was assayed by the Bradford protein assay method.
Satellite cell isolation from mouse skeletal muscles. C57BL/6 mice were purchased from Vital River Laboratories (Beijing, China). All experimental protocols were approved by the Institutional Animal Care and Use Committee of Qingdao Agricultural University. All experiments were performed according to the guidelines of the Institutional Animal Care and Use Committee of Qingdao Agricultural University and Tianjin University of Science and Technology. Muscle tissues from 6-week-old mice were minced into small pieces and incubated in MEM with 10% FBS and 0.2% collagenase (type I, Sigma) at 37 °C in a water-saturated incubator with 5% CO 2 for approximately 2.5 h. The muscles were transferred with wide-mouth Pasteur pipette into MEM, teased with Scientific RepoRts | 6:27332 | DOI: 10.1038/srep27332 forceps, and triturated to release single fibers. The cells were cultured in MEM with 10% FBS, and the medium was changed every 3 days. Approximately 10-20 single fibers per well were grown on laminin-coated cover glasses in a 12-well plate. The bottom of the 35-mm plate or cover glass was coated with 10 μ g/ml of laminin (Invitrogen) at 37 °C for 2 h. The laminin coating solution was removed immediately before the myocytes were plated. ER Ca 2+ release assay. NIH3T3 cells were seeded at low density on glass coverslips and used the following day for the ratiometric imaging measurements of Ca 2+ release from the ER. The cells were loaded in tissue culture medium with fura-2-AM (5 μ M) for 20-30 min at room temperature. The experiments were performed in Ringer's solution containing (in mM) 121 NaCl, 2.4 K 2 HPO 4 , 0.4 KH 2 PO 4 , 1.2 CaCl 2 , 1.2 MgCl 2 , 5.5 glucose, and 10 Hepes/NaOH, pH 7.2. The cells were continuously superfused on a heated microscope stage in an open Leiden chamber. Coverslips with dye-loaded cells were mounted into a heated metal flow-through perfusion chamber placed on the stage of an inverted Olympus fluorescence microscope. A one-half volume of bradykinin in Ringer's solution was added to the chamber to a final concentration of 10 nM with a pipette within seconds. The fluorescence emitted from the cells was measured in response to alternating excitation at 340 and 380 nm using a computer-controlled four-place sliding filter holder. Time lapse images were recorded at 1 s intervals after stimulation with 10 nM bradykinin. The images at 0 s and 17 s were shown. DsredN1 was co-transfected with Best3V2 at a 1:3 ratio as an indicator of the positively transfected cells. All measurements were automatically corrected for background. The ratio of light emitted from fura-2 at the two excitation wavelengths (340/380) provides a measure of the ionized cytoplasmic Ca 2+ concentrations.
For the C2C12 myoblast assay, C2C12 cells were grown in DMEM containing 10% FBS and maintained in a humidified incubator at 37 °C in the presence of 5% CO 2 /95% air until the cells were completely confluent, which was designated day 0. Then, the cells were transferred to DMEM with 1% horse serum to initiate differentiation into myoblasts. For lentivirus infection, virus particles were added before the cells reached confluency to obtain a better infection. The cells were also treated with puromycin to select the positively infected cells 24 hours after infection. On day 2 after confluency, the myoblasts were assayed in the intracellular Ca 2+ release assay. Fura 2-AM was applied to the C2C12 myoblasts for the Ca 2+ release assay, similar to the NIH3T3 cells. The images were further analyzed using ImageJ software.
shRNAi. For shRNAi manipulation, the HeLa cells were infected with a lentivirus encoding shRab7a or Scramble (Scr) for 2 days. Then, the cells were treated with 1 μ g/ml puromycin for 24 h after infection, and then these cells were further transfected with Best3V2_HA or Clc3s, respectively. For RNAi in the C2C12 myoblasts, the C2C12 cells were seeded on laminin-coated coverslips and infected with shBest3-I, shBest3-II, or scrambled control lentiviruses. Validated lentiviral hairpin constructs of shRab7a were obtained from Sigma-Aldrich.
The sequence targeting shRab7a was: 5′ -GGCTAGTCACAATGCAGATAT-3′ . The sequence targeting shBest3-I was: 5′ -GACAGGCTGATGCTCCTCATT-3′ . The sequence targeting shBest3-II was: 5′ -GCAGATGAGAGGAAATTATTC-3′ . Cell detachment assay of differentiating C2C12. The C2C12 cells were seeded on laminin-coated coverslips and infected with a lentivirus before they reached confluency. On day 5 after confluency, C2C12 myoblasts grown on glass coverslips were subjected to a detachment assay by treating them with caffeine (0.1 mM) for 10 min at 37 °C. The long-shaped cells were usually bound together to form a thin cell layer on the coverslips. When the cells detached, the cells still bound together very closely, and thus the thin cell layer can be seen by eye. The number of coverslips with a detached cell layer was calculated.

Blocking of ion
Statistical analysis. Statistically significant differences were determined using the two-tailed Student's t-test in Excel. Where applicable, the data were expressed as the means + S.D. Unless indicated otherwise, the experiments were repeated 3 times.