Plasma Membrane-associated Annexin A6 Reduces Ca2+ Entry by Stabilizing the Cortical Actin Cytoskeleton*

The annexins are a family of Ca2+- and phospholipid-binding proteins, which interact with membranes upon increase of [Ca2+]i or during cytoplasmic acidification. The transient nature of the membrane binding of annexins complicates the study of their influence on intracellular processes. To address the function of annexins at the plasma membrane (PM), we fused fluorescent protein-tagged annexins A6, A1, and A2 with H- and K-Ras membrane anchors. Stable PM localization of membrane-anchored annexin A6 significantly decreased the store-operated Ca2+ entry (SOCE), but did not influence the rates of Ca2+ extrusion. This attenuation was specific for annexin A6 because PM-anchored annexins A1 and A2 did not alter SOCE. Membrane association of annexin A6 was necessary for a measurable decrease of SOCE, because cytoplasmic annexin A6 had no effect on Ca2+ entry as long as [Ca2+]i was below the threshold of annexin A6-membrane translocation. However, when [Ca2+]i reached the levels necessary for the Ca2+-dependent PM association of ectopically expressed wild-type annexin A6, SOCE was also inhibited. Conversely, knockdown of the endogenous annexin A6 in HEK293 cells resulted in an elevated Ca2+ entry. Constitutive PM localization of annexin A6 caused a rearrangement and accumulation of F-actin at the PM, indicating a stabilized cortical cytoskeleton. Consistent with these findings, disruption of the actin cytoskeleton using latrunculin A abolished the inhibitory effect of PM-anchored annexin A6 on SOCE. In agreement with the inhibitory effect of annexin A6 on SOCE, constitutive PM localization of annexin A6 inhibited cell proliferation. Taken together, our results implicate annexin A6 in the actin-dependent regulation of Ca2+ entry, with consequences for the rates of cell proliferation.

including annexin A6 (15)(16)(17)(18)(19); some of which point to the existence of pH-dependent binding mechanisms (20 -22). Given the fact that several annexins are present within any one cell, it is likely that they form a [Ca 2ϩ ] and pH sensing system, with a regulatory influence on other signaling pathways.
The role of annexins as regulators of ion channel activity has been addressed previously (23)(24)(25). In particular, annexin A6 has been implicated in regulation of the sarcoplasmic reticulum ryanodine-sensitive Ca 2ϩ channel (25), the neuronal K ϩ and Ca 2ϩ channels (26), and the cardiac Na ϩ /Ca 2ϩ exchanger (27). Cardiac-specific overexpression of annexin A6 resulted in lower basal [Ca 2ϩ ], a depression of [Ca 2ϩ ] i transients and impaired cardiomyocyte contractility (28). In contrast, the cardiomyocytes from the annexin A6 null-mutant mice showed increased contractility and accelerated Ca 2ϩ clearance (29). Consistent with its role in mediating the intracellular Ca 2ϩ signals, especially Ca 2ϩ influx, ectopic overexpression of annexin A6 in A431 cells, which lack endogenous annexin A6, resulted in inhibition of EGF-dependent Ca 2ϩ entry (30).
The difficulty of investigating the influence of annexins on signaling events occurring at the PM lies in the transient and reversible nature of their Ca 2ϩ and pH-dependent lipid binding. Although the intracellular Ca 2ϩ increase following receptor activation or Ca 2ϩ influx promotes the association of the Ca 2ϩ -sensitive annexins A2 and A6 with the PM, the proteins quickly resume their cytoplasmic localization upon restoration of the basal [Ca 2ϩ ] i (14). Therefore, to investigate the effects of membrane-associated annexins on Ca 2ϩ homeostasis and the cell signaling machinery, we aimed to develop a model system allowing for a constitutive membrane association of annexins.
Here we used the PM-anchoring sequences of the H-and K-Ras proteins to target annexins A6 and A1 to the PM independently of [Ca 2ϩ ]. The Ras GTPases are resident at the inner leaflet of the PM and function as molecular switches (31). The C-terminal 9 amino acids of H-and N-Ras and the C-terminal 14 amino acids of K-Ras comprise the signal sequences for membrane anchoring of Ras isoforms (32). Although the palmitoylation and farnesylation of the C terminus of H-Ras (tH) serves as a targeting signal for predominantly cholesterol-rich membrane microdomains at the PM (lipid rafts/caveolae) (33), the polybasic group and the lipid anchor of K-Ras (tK) ensures the association of K-Ras with cholesterol-poor PM membrane domains. Importantly, these minimal C-terminal amino acid sequences are sufficient to target heterologous proteins, for example GFP, to different microdomains at the PM and influence their trafficking (34).
In the present study we fused annexins A6, A2, and A1 with fluorescent proteins and introduced the PM-anchoring sequences of either H-Ras (annexin-tH) or K-Ras (annexin-tK) at the C termini of the fusion constructs. We demonstrate that the constitutive PM localization of annexin A6 results in downregulation of store-operated Ca 2ϩ entry. Expression of membrane-anchored annexin A6 causes an accumulation of the cortical F-actin, and cytoskeletal destabilization with latrunculin A abolishes the inhibitory effect of PM-anchored annexin A6 on SOCE. Taken together, our results implicate annexin A6 in the maintenance of intracellular Ca 2ϩ homeostasis via actin-dependent regulation of Ca 2ϩ entry.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Monoclonal antibodies against annexins were purchased from BD Transduction (Lexington, KY) and Santa Cruz Biotechnology (Santa Cruz, CA). The anti-GFP monoclonal antibody (JL-8) was from Clontech. Chemicals and purified bovine calmodulin were from Sigma. Restriction endonucleases, Taq polymerase, and T4 DNA ligase were purchased from New England Biolabs. Fluo-3/AM and jasplakinolide were from Molecular Probes, Inc. Latrunculin A was from Calbiochem.
Expression of Annexins as Fusions with Fluorescent Proteins and Ras PM Anchor-The pC1-based GFP-tH and GFP-tK plasmids were a kind gift from Dr. J. Hancock (IMB, University of Queensland, Brisbane, Australia). These vectors were modified to replace GFP with the fluorescent proteins of choice (YFP or mCherry) and allow insertion of annexin sequences upstream of the fluorescent proteins' coding sequences. For this purpose, the coding sequences of YFP and mCherry proteins were amplified by PCR using the forward primer, 5Ј-ATAAAGCTAGCAAGCTTCTCGAGCGAGATCTCACCA-TGGTGAGCAAGGGCGAGGAG-3Ј, corresponding to the fluorescent protein cDNA and containing NheI, HindIII, XhoI, and BglII cloning sites (underlined); the reverse primer, 5Ј-ATAAAGAATTCGACTTGTACAGCTCGTC-CATGCCG-3Ј, containing the EcoRI cloning site. The PCR products were ligated with a larger fragment of the NheI-EcoRI-digested GFP-tH or GFP-tH vectors, resulting in the replacement of the GFP coding sequence and introduction of unique cloning sites for insertion of annexins. Annexins A1 and A6 were cloned into pYFP-tH, -tK, pmCherry-tH and-tK, and pN1-mCherry (Clontech) as described previously (20). Annexin A6-GFP-tK protein was created by cloning annexin A6 coding sequence into the NheI site located upstream of the GFP sequence of the original pC1-GFP-tK vector. The pEYFP-␤-actin expression vector was from Clontech.
Cell Culture and Transfections-HEK293 were maintained in Dulbecco's modified Eagle's medium containing 2 mM glutamine, 100 units of penicillin/ml, 100 g of streptomycin/ml, 10% fetal calf serum. All cells were grown in 5% CO 2 at 37°C in a humidified incubator. HEK293 cells were transiently transfected with plasmids using electroporation (Bio-Rad) and analyzed after incubating at 37°C for 48 h. Stable HEK293 cell lines expressing annexins A6YFP-tH, A6YFP-tH, and mCherry-tH were established using G-418 selection following the standard procedures. For cell proliferation analysis, cells were plated at equal density, harvested by trypsinization at selected time points, and counted following trypan blue extraction.
Live Cell Imaging and Intracellular Calcium Measurements-Live cell imaging and calcium imaging recordings were performed as described (14). Cells were observed under an Axiovert 200M microscope with laser scanning module LSM 510 META (Zeiss) using a ϫ63 or a ϫ100 oil immersion lens. Images were acquired at a scanning speed of 2 s, with 1-s intervals between frames. Images of the Fluo-3/AM-loaded cells were recorded and analyzed using the "Physiology Evaluation" software package (Zeiss).
Isolation of HEK293 Membranes; SDS-PAGE and Western Blotting Analysis-Unless otherwise stated, all procedures were performed at 4°C or on ice. Membrane preparation from HEK293 cells was performed using a detergent-free protocol (35), and Triton X-100 extraction of post-nuclear cell lysates followed by sucrose gradient centrifugation was done as described (36). Membrane-bound annexins were separated from the unbound proteins by low speed centrifugation (20 min, 12,000 rpm at 4°C). The membrane-bound proteins in the pellet and the soluble proteins in the supernatant were analyzed by SDS-PAGE followed by Western blotting with anti-annexin antibodies. Image analysis to estimate the protein content of the individual bands following SDS-PAGE and Western blotting was performed using PowerLook 1120 scanner and Image-Quant TL (v2003) software from Amersham Biosciences Europe GmbH.
Ca 2ϩ -ATPase Activity Assay-The enzymatic activity of plasma membrane Ca 2ϩ -ATPase (PMCA) was determined essentially as described (37). In brief, gradient-purified membranes (10 g of protein) were incubated in an assay system containing, in a final volume of 200 l, 25 mM Tris-HCl, pH 7.4, 50 mM KCl, 1 mM MgCl 2 , 200 M EGTA, and CaCl 2 to give 10 M free Ca 2ϩ , determined using MaxChelator software. The reaction mixture was supplemented with 100 nM thapsigargin to inhibit sarco-endoplasmic reticulum Ca 2ϩ -ATPase (SERCA). Where specified, 2 g (80 units) of purified bovine calmodulin were added. After 5-min preincubation at 37°C, reaction was started by addition of 1 mM ATP, continued for 30 min, and stopped by addition of 500 l of staining solution containing 14 mg/ml ascorbic acid, 0.3% ammonium molybdate, and 0.35 N H 2 SO 4 . After 20 min at room temperature, A 820 was determined by spectrophotometry (DU 530, Beckman Coulter). The data were expressed as arbitrary units (optical density per g of protein ϫ 10).
Statistical Analysis-Statistically significant differences were determined with a Mann-Whitney U test with an ␣ set to 0.05 for values not displaying a normal distribution, and with a twotailed Student t test, preceded by a Levene test, with an ␣ set to 0.05 for values with a normal distribution. Normality studies were done with Kolmogorov-Smirnov and Shapiro-Wilk tests (p Ͻ 0.05). All the studies were carried out with the SPSS program (version 15.0).

H-and K-Ras Membrane
Anchor Sequences Ca 2ϩ -independently Target Annexin A6 to the PM-We fused the coding sequences of annexin A6 to the fluorescent proteins (GFP or YFP) and the minimal PM targeting sequences of H-and K-Ras, and expressed the resulting proteins following transient transfection of HEK293 cells. The chimeric proteins had an approximate molecular mass of 95 kDa, and were detected by Western blotting with both anti-GFP and antiannexin A6 antibodies (Fig. 1A). The proteins constitutively localized to the PM of transfected cells under resting conditions (basal [Ca 2ϩ ] i , Fig. 1B for A6GFP-tK), with some additional Golgi labeling of A6YFP-tH (not shown). We then studied the effects of [Ca 2ϩ ] on the PM association of A6YFP-tK and -tH proteins. Membrane association of the FIGURE 1. H-and K-Ras anchors target annexin A6 to the PM independent of Ca 2؉ . A, annexin A6YFP-tH, annexin A6YFP-tK, GFP-tK, and annexin A1YFP were expressed in HEK293 cells. 48 h post-transfection the cells were collected, lysed, and analyzed by SDS-PAGE followed by Western blotting with monoclonal antibodies against GFP or human annexin A6. Lanes 1 and 5 show A6YFP-tH, lanes 2 and 6 A6YFP-tK, lanes 3 and 7 GFP-tK, and lanes 4 and 8 annexin A1YFP. The positions of endogenous annexin A6 (wt A6) and A6-fusions are indicated by arrows. B, live HEK 293 cells expressing annexin A6GFP-tK were examined at the confocal microscope under resting conditions (low basal [Ca 2ϩ ] i ). Cells are shown in orthogonal projection, bar ϭ 5 m. C, membranes from HEK293 cells expressing annexin A6YFP-tH, or annexin A6YFP-tK, were isolated in the presence of 0.2 mM Ca 2ϩ . The purified membranes were resuspended in Na ϩ -Tyrode buffer containing 0.2 mM Ca 2ϩ , or 1 mM EGTA, as indicated. Membranebound annexins were separated from the unbound proteins by centrifugation (20 min, 12,000 rpm at 4°C). The membrane-bound proteins in the pellet (P) and the soluble proteins in the supernatant (S) were analyzed by SDS-PAGE followed by Western blotting with monoclonal antibodies against annexin A6. Position of endogenous annexin A6 and A6YFP-tH and -tK fusions is indicated by arrows.
tK-and tH-tagged annexin A6 was compared with that of endogenous annexin A6 following low speed centrifugation in presence of either 0.2 mM Ca 2ϩ , or 1 mM EGTA (Fig. 1C). Endogenous annexin A6 detached from the membranes in the presence of EGTA and was detected in the soluble fraction, whereas both tKand tH-tagged annexin A6 fusion proteins remained PM-associated, indicating that the Ras membrane anchors were sufficient for Ca 2ϩindependent binding.
We then analyzed whether the annexin A6 fusion proteins retained the properties of endogenous annexin A6 by disrupting the Ras tag-mediated membrane anchoring using 0.5% Triton X-100 extraction (33) (supplemental Fig. S1). Membranes prepared from a mixture of A6YFP-tH-and GFP-tH-expressing HEK293 cells were extracted with 0.5% Triton X-100 in the presence of 0.2 mM Ca 2ϩ , and the distributions of A6YFP-tH, GFP-tH, endogenous annexin A6, and RhoA in sucrose gradient fractions were analyzed by Western blotting. Under these conditions, endogenous Hand K-Ras are found in soluble fractions (33), and consistent with these findings, most of the GFP-tH was extracted with Triton X-100 and found in high density gradient fractions (fractions 5-8), similar to RhoA (supplemental Fig. S1, A and B). Endogenous annexin A6 was associated predominantly with low density fractions (fractions 3 and 4) in the presence of Ca 2ϩ (supplemental Fig. S1B). Importantly, Rastagged A6YFP-tH displayed the same distribution as endogenous annexin A6 within the gradient and was found almost exclusively in fractions 3-5 in the presence of Ca 2ϩ (supplemental Fig. S1, A and B). Similar results were obtained when using a mixture of A6YFP-tK and GFP-tK membranes (supplemental Fig. S1C). These data strongly suggest that Triton X-100 extraction disrupts the membrane association of Ras-anchored annexin A6 and annexin A6 fusion proteins then behave like endogenous annexin A6 in their ability to interact with membranes in a Ca 2ϩdependent manner.
PM-targeted Annexin A6 Attenuates Store-operated Ca 2ϩ Entry-Earlier observations implicated annexin A6 in the regulation of intracellular Ca 2ϩ homeostasis (28 -30), therefore we examined whether its constitutive association with the PM had an effect on Ca 2ϩ transients or Ca 2ϩ entry into the cells. In all experiments, cells expressing mCherry protein, tagged to the PM with tH or tK membrane anchors served as a control for the nonspecific effects of transfection on cellular Ca 2ϩ responses. Both mCherry-tK and A6GFP-tK-expressing cells were co-cultured on the same coverslip and loaded with Fluo-3/AM, and Ca 2ϩ responses were recorded simultaneously ( Fig.  2A). Fig. 2B shows the result from a representative experiment. The intracellular Ca 2ϩ release was examined by applying 10 M CCh in Ca 2ϩ -free Na ϩ -Tyrode buffer. Both A6GFP-tK and mCherry-tK cells produced a similar Ca 2ϩ transient, indicating no differences in the filling of intracellular Ca 2ϩ stores. When the SOCE was stimulated by addition of 1 M thapsigargin (TG) followed by 2 mM or 5 mM [Ca 2ϩ ] ex , an attenuation of Ca 2ϩ entry was observed in the cells expressing A6GFP-tK (Fig. 2B). Statistical evaluation of data from five independent experiments (n ϭ 22 recordings), showed that the SOCE in A6GFP-tK-expressing cells was on average 60% of the control (p Ͻ 0.05) (Fig. 2D).
Given the different subcellular microlocalization of the tH-and tK membrane anchors (32), we next analyzed the effects of annexin A6, FIGURE 2. PM-associated annexin A6 reduces store-operated Ca 2؉ entry. A, HEK293 cells expressing annexin A6GFP-tK or mCherry-tK were co-cultured on the same coverslip, loaded with Fluo-3/AM, and examined at the confocal microscope. An arbitrary field containing both cell types was selected, and changes in [Ca 2ϩ ] i upon stimulation of intracellular Ca 2ϩ release and store-operated Ca 2ϩ entry examined, selecting an intracellular regions of interest to avoid the contribution of PM-associated GFP signal to Fluo-3 fluorescence changes observed in A6GFP-tK cells. B, cells were kept in Ca 2ϩ -free Tyrode buffer, stimulated with 10 M CCh, followed by SERCA inhibition with 1 M TG. SOCE was stimulated by adding 2 mM and 5 mM [Ca 2ϩ ] ex , and responses in mCherry-tK-expressing cells (control, black line) were compared with annexin A6GFP-tK cells (gray line). C, shows a similar experiment as in B, performed with annexin A6YFP-tH and mCherry-tH cells. D, basal [Ca 2ϩ ] i levels, CCh-and TG-stimulated ER release, and Ca 2ϩ entry in tH-and tK-tagged annexin A6 cells were expressed relative to the responses of the control, Ras-tagged mCherry cells. The graph shows an average of five independent experiments (n ϭ 22 recordings) Ϯ S.E. The SOCE reduction to ϳ60% of control levels shown in both annexin A6 constructs was statistically significant (p Ͻ 0.01). There was also a statistically significant (*, p Ͻ 0.05) reduction of basal [Ca 2ϩ ] i and in the amplitude of CCh-induced Ca 2ϩ release in A6YFP-tH cells compared with A6YFP-tK cells.

FIGURE 3. A constitutive PM localization of annexin A6 is necessary for SOCE attenuation.
A, HEK293 cells expressing annexin A6YFP-tH, wild-type cytoplasmic annexin A6-mCherry or mCherry-tH were co-cultured on the same coverslip, loaded with Fluo-3/AM, and examined in the confocal microscope. An arbitrary field containing all three cell types was selected, and changes in [Ca 2ϩ ] i upon stimulation of intracellular Ca 2ϩ release and store-operated Ca 2ϩ entry examined by calcium imaging, selecting intracellular regions of interests to avoid the contribution of the YFP signal into Fluo-3 fluorescence changes observed in A6YFP-tH cells. B, cells were kept in Ca 2ϩ -free Tyrode buffer, stimulated with 10 M CCh, followed by SERCA inhibition with 1 M TG. SOCE was stimulated by adding 2 mM and 5 mM [Ca 2ϩ ] ex , and responses in mCherry-tH-expressing cells (control, solid black line) were compared with annexin A6YFP-tK cells (gray line) and A6-mCherry (dotted black line). C, basal [Ca 2ϩ ] i levels, CCh-and TG-stimulated ER release and Ca 2ϩ entry in annexin A6-mCherry, and A6YFP-tH cells were expressed relative to responses of the control, mCherry-tH cells. The graph shows an average of three independent experiments (n ϭ 9 arbitrary fields measured) Ϯ S.E. SOCE was reduced to ϳ60% of control levels in annexin A6YFP-tH construct (*, p Ͻ 0.05), but there was no statistically significant difference between the responses recorded in A6-mCherry cells and control, provided that annexin A6-mCherry remains in the cytoplasm. D, live HEK293 cells expressing annexin A6-mCherry (arrows) and mCherry were observed in the confocal microscope and treated with 1 M TG followed by 2, 5, 7, and 10 mM [Ca 2ϩ ] ex . Localization of annexin A6-mCherry after adding the indicated [Ca 2ϩ ] ex is shown. There was no translocation of annexin A6-mCherry to PM at 2 or 5 mM [Ca 2ϩ ] ex . tagged with tH, on SOCE, using the same experimental set-up as described above for A6GFP-tK, with mCherry-tH serving as a control (Fig. 2C). The average of five experiments and a comparison of the effects of A6GFP-tK and A6YFP-tH on SOCE (Fig. 2D) demonstrates that, in addition to the attenuation of SOCE, observed both in A6GFP-tK-and A6YFP-tH-expressing cells (60% of control), there was a statistically significant (p Ͻ 0.05) reduction of basal [Ca 2ϩ ] i , and the amplitude of CChmediated transient in A6YFP-tH cells, compared with cells expressing tK-tagged A6 (Fig. 2D).
Previously annexin A6 has been implicated in the regulation of EGF-stimulated Ca 2ϩ entry into A431 cells (30). We performed similar experiments using A6YFP-tH-and mCherry-tH-transfected A431 cells and identified a down-regulation of EGF-stimulated Ca 2ϩ entry (supplemental Fig. S2). Taken together, these results indicate that the constitutive PM association of annexin A6 caused a decrease of Ca 2ϩ entry.

PM Localization of Annexin A6 Is Necessary for SOCE Attenuation-
To asses the effects of cytoplasmic versus PM-targeted annexin A6 on SOCE, we co-cultured HEK293 cells, expressing cytoplasmic annexin A6 fused with mCherry (A6-mCherry), A6YFP-tH, and mCherry-tH (Fig. 3A), and treated them with CCh, followed by SOCE stimulation at 2 and 5 mM Ca 2ϩ (Fig.  3B). In the control experiment, at these [Ca 2ϩ ] ex levels untagged A6-mCherry remained in the cytoplasm (Fig. 3D). There was no significant difference in basal [Ca 2ϩ ] i , the amplitude of CCh-induced intracellular Ca 2ϩ release, or SOCE between the cells expressing mCherry-tH (control) and cytoplasmic A6-mCherry, whereas both the store contents and SOCE were decreased in A6YFP-tH-expressing cells ( Fig. 3B for a representative recording, Fig. 3C for the average of n ϭ 9 recordings). These results indicate that a PM association of annexin A6 was required to elicit a distinct effect on the amplitude of SOCE.
Previously we showed that ectopically expressed annexin A6 translocated to the PM in live cells treated with ionomycin or TG upon elevation of [Ca 2ϩ ] i (14). In the set of experiments described above (Fig. 3, A-C), by choosing the appropriate [Ca 2ϩ ] ex. , we kept [Ca 2ϩ ] i below the threshold levels required for annexin A6 PM translocation. However, when HEK293 cells were treated with TG, followed by addition of 7 mM or 10 mM [Ca 2ϩ ] ex , large amounts of annexin A6-mCherry associated with the PM (Fig. 3D). Similarly, endogenous annexin A6 associated with the PM of Ca 2ϩ -ionomicin or TG-treated cells as detected by immunofluorescence (Fig.  4A). To assess the effects of annexin A6-PM translocation on Ca 2ϩ entry we co-cultured HEK293 cells expressing annexin A6-mCherry or mCherry alone, treated them with TG followed by SOCE stimulation with gradually increasing [Ca 2ϩ ] ex up to 10 mM. As shown in Fig. 4B, at [Ca 2ϩ ] ex below 5 mM, SOCE was similar in A6-overexpressing and control cells. When [Ca 2ϩ ] ex was increased above 7 mM, annexin A6-mCherry fusion protein translocated to the PM (Fig. 4B). PM translocation of A6-mCherry under these conditions correlates with ϳ20% decline in the rates of Ca 2ϩ entry, indicating that PM translocation of annexin A6 attenuated SOCE (Fig. 4, C and D). Knockdown of Endogenous Annexin A6 Causes Elevation of Ca 2ϩ Entry-To establish a role for endogenous annexin A6 in the regulation of SOCE, we performed knockdown experiments using transient transfections of HEK293 cells with annexin A6 shRNA plasmids targeting human annexin A6, together with mCherry, or scrambled shRNA (control) together with mCherry-tH. The levels of inhibition of annexin A6 expression were evaluated by immunofluorescence, and no endogenous annexin A6 protein was detected in mCherry-positive cells 48 h post-transfection (Fig. 5A). Cells were co-cultured on coverslips, and levels of SOCE were investigated after adding 1 M TG followed by increasing [Ca 2ϩ ] ex from 0.5 mM to 10 mM (Fig. 5B). Consistent with annexin A6 overexpression inhibiting SOCE, there was a statistically significant elevation of Ca 2ϩ entry caused by application of 7 mM [Ca 2ϩ ] ex in annexin A6 knockdown HEK293 cells compared with the controls (n ϭ 23 recordings, Fig. 5C).
Plasma Membrane Targeting of Annexins A1 and A2 Does Not Influence SOCE-All annexins interact with cellular membranes upon elevation of [Ca 2ϩ ] i , albeit with different [Ca 2ϩ ] i sensitivities (10,14). To test for the specificity of the annexin A6-induced SOCE attenuation within the annexin family, we performed identical assays with annexin A1, anchored to the PM with the tK sequence. We expressed annexin A1YFP-tK, and first examined its intracellular localization. Similar to membrane-targeted annexin A6, A1YFP-tK was evenly distributed at the PM of non-stimulated cells. However, upon elevation of [Ca 2ϩ ] i in ionomycin-treated cells, we observed a redistribution and punctuated PM staining of tagged annexin A1, consistent with its association with ceramide platforms, as described previously for cytoplasmic annexin A1YFP (38) (Fig.  6A). A1YFP-tK-expressing cells were stimulated with CCh, followed by treatment with TG. Fig. 6B shows a representative recording with five cells each imaged in the same field, and the data are summarized in Fig. 6C. Most importantly, activation of SOCE in cells expressing A1YFP-tK did not result in any significant differences compared with control cells.
Similar to annexin A1, annexin A2 is also known to interact with the PM in a Ca 2ϩ -dependent manner (14). We therefore also analyzed the effects of overexpression of membrane-targeted annexin A2 on SOCE and obtained results comparable to the data shown for annexin A1. H-Ras-anchored annexin A2 remained PM-associated in low and high [Ca 2ϩ ] i but did not show the punctuated distribution typical for annexin A1 (supplemental Fig. S3A). However, despite the differences in subcellular localization, there was no significant decrease of SOCE compared with the control (supplemental Fig. S3, B and C) (n ϭ 25 recordings). In summary, these data indicate that attenuation of Ca 2ϩ entry was annexin A6-specific.
Attenuated Ca 2ϩ Entry in A6YFP-tH Cells Leads to a Decrease in ER Store Contents-Annexin A6, constitutively attached to the PM by two different Ras membrane anchors tH and tK, caused similar decreases in SOCE. However, tH-tagged A6 also reduced the amounts of Ca 2ϩ stored in the ER as judged by the amplitude of CCh-mediated Ca 2ϩ transients (Fig. 2D). In non-stimulated A6YFP-tH cells, the levels of basal Fluo-3 fluorescence were often decreased, indicating a reduction of basal [Ca 2ϩ ] i levels (not shown). To rule out effects of tH-tagged annexin A6 on intracellular compartments, such as the ER, we performed additional control experiments. Similar to the results described in Fig. 2C, cells were transfected with A6YFP- tH, loaded with Fluo-3/AM in the absence of Ca 2ϩ , and treated with TG followed by addition of Ca 2ϩ to induce SOCE. Under these conditions, and consistent with the data shown in Fig. 2 (C and D), there was a marked decrease in the amplitude of Ca 2ϩ release from the ER, as evident after TG application (supplemental Fig. S4A). However, when the cells were re-loaded with 3 mM [Ca 2ϩ ] ex for 2 h after Fluo-3/AM incubation, prior to SOCE stimulation, there was no difference in the amplitude of intracellular Ca 2ϩ release between A6YFP-tH and mCherry-tH control cells, whereas the inhibitory effect on SOCE was not significantly affected (supplemental Fig. S4B). Thus Ca 2ϩ reloading before stimulation restored the basal [Ca 2ϩ ] i levels and the Ca 2ϩ store contents in the ER to the control levels, indicating that cells expressing tH-tagged annexin A6 did not develop deficiencies in ER loading or an excessive ER leak. These findings indicate that tH-tagged annexin A6 possibly induced a stronger inhibition of Ca 2ϩ entry across the PM compared with tK-tagged annexin A6, causing the decrease of ER store contents under Ca 2ϩ -free conditions. In view of these observations, in all subsequent experiments with A6YFP-tH cells, the cells were preincubated in Ca 2ϩ -Tyrode and perfused with Ca 2ϩ -free buffer shortly before application of TG.

The PM-localization of Annexin A6 Does Not Affect Ca 2ϩ Extrusion
Rates across the Membrane-To test whether A6YFP-tH-and -tKmediated down-regulation of SOCE was the result of an increased Ca 2ϩ extrusion, we measured the PM Ca 2ϩ -ATPase (PMCA) activity in membrane fractions, containing bound annexin A6, as well as the speed of Ca 2ϩ clearance from A6YFP-tH cells. The PMCA activity was similar in membrane preparations from the A6GFP-tK-expressing cells and in GFP-tK control cells (Fig. 7A). Likewise, the addition of purified recombinant annexin A6 to membrane preparations of HEK 293 cells in the presence of 10 mM Ca 2ϩ had no effect on PMCA activity, whereas addition of purified calmodulin, a known modulator of PMCA (37), resulted in an increase of PMCA activity (Fig. 7B).
We then measured Ca 2ϩ entry after PMCA inactivation upon ATP depletion in A6YFP-tH cells and mCherry-tH control (Fig. 7C). Cells were preincubated in 20 mM NaN 3 and 20 mM 2-deoxy-D-glucose, and Ca 2ϩ entry was stimulated by addition of 5 mM [Ca 2ϩ ] ex . Despite inactivation of PMCA by ATP depletion, a significant decrease in the rates of Ca 2ϩ entry in A6YFP-tH cells was observed, compared with the mCherry-tH control. Finally, we determined the effectiveness of Ca 2ϩ clearance in cells expressing A6YFP-tH and mCherry-tH (Fig. 7D). In line with the results described above, there was no difference in the Ca 2ϩ extrusion speed when control and A6YFP-tH cells restored their basal [Ca 2ϩ ] i after SOCE stimulation. Taken together, these results suggest that membrane-tagged annexin A6 acts on SOCE independent of PMCA and does not affect the speed of Ca 2ϩ extrusion.
The Ca 2ϩ Independent Interactions of A6YFP-tH and A6YFP-tK with Membranes Are Mediated by Actin-Stabilization of cortical cytoskeleton with jasplakinolide reduces SOCE in most cells (8), possibly by interfering with STIM1 function (39). Annexin A6 has been reported to bind actin filaments and stabilize cytoskeleton of contractile cells (40 -42). Confirming the previously published observations, purified endogenous annexin A6 bound actin in a Ca 2ϩ -dependent manner in vitro (supplemental Fig. S5). We then examined the association of membrane-anchored annexin A6 with cortical actin. Membrane fractions, which were gradient-purified in the presence of 1 mM EGTA, contained significant amounts of F-actin and caveolin (Fig. 8, A and B). Lipid extraction of A6YFP-tH-containing membranes with 1% Triton X-100 resulted in the solu-FIGURE 6. Annexin A1 does not influence Ca 2؉ entry in A1YFP-tK cells. A, annexin A1 was expressed as a fusion with YFP and the K-Ras anchor sequence. Live HEK293 cells producing A1YFP-tK were analyzed by confocal microscopy under resting conditions, and after stimulation with 2 mM Ca 2ϩ -ionomycin. tK-tagged annexin A1 was targeted to the PM at rest, and redistributed into a punctuated PM staining as shown previously (38) due to association with ceramide platforms at high Ca 2ϩ . Bar ϭ 5 m. B, annexin A1YFP-tK-expressing cells were co-cultured with mCherry-tK control, loaded with Fluo-3/AM, and stimulated with 10 M CCh, followed by SERCA inhibition with 1 M TG, and activation of SOCE with 2 mM and 5 mM [Ca 2ϩ ] ex . In a representative experiment, responses in mCherry-tK-expressing cells (control, black line) were compared with annexin A1YFP-tK cells (gray line). C, the graph shows an average of five independent experiments (n ϭ 21 recordings) Ϯ S.E. There was no statistically significant difference between the responses in annexin A1YFP-tK cells compared with the mCherry-tK control.
bilization of both tH-tagged annexin and caveolin, indicating that a proportion of annexin A6YFP-tH was lipid-associated (Fig. 8A). Analogous treatment of A6YFP-tK-containing membranes produced a different result: in contrast to caveolin, no solubilization of tK-tagged annexin was achieved, and it remained in the pellet (Fig. 8B). The fact that membrane-tagged annexin A6 co-purified with actin-rich and TX-insoluble membrane fraction is indicative of an interaction of annexin A6 with the actin cytoskeleton at the PM.
The enrichment of a protein in a particular fraction of detergent-resistant membranes separated on sucrose gradients is influenced by its interactions with both the lipid bilayer and membrane cytoskeleton (35). Because results described above pointed to an interaction of A6YFP-tH and A6YFP-tK with actin, we aimed to address the importance of the actin cytoskeleton for the Triton X-100-solubility of membrane-tagged annexin A6. Cortical actin cytoskeleton was destabilized by incubating HEK293 cells with 5 M latrunculin A (Lat A) for 2 h, and membranes isolated after extraction with 0.5% Triton X-100. In untreated A6YFP-tH controls, the majority of tHtagged annexin A6 is found in fractions 2-4 (detergent-resistant membranes) and 7, with ϳ10% of A6YFP-tH in the soluble fraction of the gradient (fraction 8) and small amounts present in the insoluble pellet (fraction 9). It should be noted that, under these conditions, the Ras lipid anchors are soluble when being extracted with Triton X-100 (see supplemental Fig.  S1), therefore the observed membrane binding most probably reflects the interaction of annexin A6 with its membrane-associated partner (lipid or protein). Lat A treatment resulted in redistribution and proportional increase of A6YFP-tH content in the soluble protein fraction of the gradient (fraction 8, 20% total) (Fig. 8C). Whereas in untreated cells, A6YFP-tK also accumulated in fractions 2, 3, and 7, large amounts of A6-YFP-tK (30% total) were found in the insoluble pellet (fraction 9) (Fig. 8D). Indeed, Lat A had a stronger effect on A6YFP-tK distribution as 30% of the protein was solubilized and relocated to fraction 8 of the gradient after Lat A incubation, compared with 5% of control. Furthermore, after Lat A treatment, the relative amount of insoluble A6YFP-tK was reduced (fraction 9) (Fig. 8D). Taken together, these data suggest that the Lat A-induced redistributions reflect the association of PM-targeted annexin A6 with actin. The Constitutive PM Localization of Annexin A6 Stabilizes the Cortical Actin Cytoskeleton-We then addressed the question whether PM localization of annexin A6 and its interaction with actin manifested themselves as changes in the distribution of F-actin filaments under the PM. We co-expressed YFP-␤actin with A6GFP-tK and compared the distribution of F-actin in these cells with the control, expressing YFP-␤-actin alone. The control cells generally lacked a well developed cortical actin ring; however, in the cells co-expressing membrane-associated annexin A6, a strong accumulation of YFP-␤-actin under the PM was observed (Fig. 9A, arrows). In agreement with the biochemical analysis ( Fig. 8 and supplemental Fig. S5) and further supporting F-actin-annexin A6 interaction, we observed a partial co-localization of annexin A6GFP-tK and YFP-␤-actin at the PM (Fig. 9B). Destabilization of the cortical actin with Lat A caused a re-distribution of YFP-␤-actin, and a rapid disappearance of the cortical ␤-actin aggregates, without affecting the PM localization of annexin A6GFP-tK (Fig. 9C). Similar to the sub-plasmalemmal aggregates of F-actin, observed in the annexin A6GFP-tK-expressing cells, the F-actin-stabilizing agent jasplakinolide also caused the formation of a pronounced cortical actin ring under the PM of YFP-␤-actin-expressing HEK293 cells (Fig. 9D). These results indicate that the constitutive PM localization of annexin A6 promoted its interaction with F-actin, resulting in a re-modeling of the actin cytoskeleton, and stabilization of cortical actin.
Destabilization of the Actin Cytoskeleton Abolishes the Annexin A6YFP-tH-mediated Decrease of Ca 2ϩ Entry-To investigate an involvement of the actin cytoskeleton in the inhibitory effects of membrane-tagged annexin A6 on SOCE, we examined Lat A-treated HEK293 cells, expressing annexin A6YFP-tH and mCherry-tH by confocal microscopy (Fig. 10A). In line with previous observations, Lat A-treated cells rounded up due to cytoskeleton destabilization. Yet, there was no noticeable intracellular redistribution of either annexin A6YFP-tH or mCherry-tH. A6YFP-tK behaved similarly (not shown). These data indicate that tH-or tK-membrane anchors facilitate an association of annexin A6 with the PM independent of the cortical actin cytoskeleton. On the other hand, given the results of the biochemical and microscopic studies in the presence of Lat A (Figs. 8 and 9D), once the H-and K-Ras anchor sequences have directed annexin A6 to the PM, they appear to facilitate an association of annexin A6 with F-actin.
Finally, we examined the effects of cytoskeletal destabilization on SOCE in annexin A6YFP-tH-expressing cells. In the untreated cells, similar to the results described in Figs. 2, 3, and 5, there was an ϳ40% attenuation of Ca 2ϩ entry in A6YFP-tH cells compared with the control. However, when SOCE was stimulated in cells, pre-treated with 5 M Lat A for 1 h, there was no difference in the amplitude of Ca 2ϩ entry compared with the mCherry-tH controls (Fig. 10B). The quantification of three independent experiments (n ϭ 10 recordings each for untreated and Lat A-treated cells) is summarized in Fig. 10C, demonstrating that Lat A treatment abolished the inhibitory effect of A6YFP-tH on SOCE (*, p Ͻ 0.05). In summary, these results suggest that membrane-targeted annexin A6 interacts with the cortical actin cytoskeleton to alter store-operated Ca 2ϩ entry and homeostasis.
Constitutive PM Targeting of Annexin A6 Reduces Cell Proliferation-In many cell types, SOCE is believed to be vital for cell proliferation (43,44). We investigated whether the reduction of SOCE observed in HEK293 cells ectopically expressing PM-anchored annexin A6 has an effect on cell proliferation rates. We generated HEK293 cells lines, stably expressing annexin A6YFP-tH, A6YFP-tK, and mCherry-tH control, and assessed their growth by counting viable cells at selected time points (Fig. 11A). Although differences in cellular growth were not apparent in the initial 18 h upon plating, we were able to determine a considerably slower proliferation of HEK-A6YFP-tK at 24 h and thereafter (Fig. 11B). The control HEK-mCherry-tH cells were growing similar to the wild-type HEK293. At 48 h post plating, the HEK293 cells stably expressing PM-anchored A6YFP-tH and A6YFP-tK showed reduced proliferation rates as their numbers reached ϳ70% of the control. It can be concluded that both tH-and tK-anchored annexin A6 proteins inhibit cell proliferation when constitutively expressed.

DISCUSSION
Intracellular Ca 2ϩ homeostasis is important for many cellular functions, and any drastic alteration of the Ca 2ϩ influx or extrusion can have far-reaching and often negative consequences. Therefore cells are equipped with a plethora of proteins controlling [Ca 2ϩ ] i during signaling events and at rest. Annexin A6 knock-in and knock-out experiments (28 -30) suggested a function of this Ca 2ϩ -and lipid-binding protein in the regulation of intracellular Ca 2ϩ homeostasis. Here we used constitutive PM targeting of annexin A6 to study its role in Ca 2ϩ entry into the cell. We fused fluorescent protein-labeled annexin A6 with PM anchor sequences of H-and K-Ras (tH and tK, respectively) and showed that the resulting proteins localized to the PM in resting cells (basal [Ca 2ϩ ] i ) and remained PM-associated during membrane isolation in Ca 2ϩ -free conditions. We induced SOCE in cells expressing tH-and tK-tagged annexin A6 and showed that both constructs caused a decrease of the SOCE.
PM association was necessary to observe a clear attenuation of SOCE, because ectopic expression of wild-type annexin A6, when localized in the cytoplasm, had no effect on Ca 2ϩ entry under the same experimental conditions. Previously, we showed that cytoplasmic annexin A6 was able to bind to the PM upon elevation of [Ca 2ϩ ] i in ionomycin-and TG-treated cells (14,45). It should be noted, that for clarity in the comparative experiments shown here, the level of [Ca 2ϩ ] i reached during SOCE in TG-treated cells did not promote the PM translocation of the untagged annexin A6-mCherry. However, when the TG-treated cells were perfused with higher [Ca 2ϩ ] ex , resulting in further increased [Ca 2ϩ ] i levels, we observed the PM translocation of both the endogenous and ectopically expressed annexin A6. Under these conditions, the wild-type annexin A6 also caused attenuation of SOCE, similar to its PM-targeted counterpart.
These data were further confirmed by knockdown experiments using annexin A6-specific shRNA. The annexin A6-negative HEK293 cells displayed elevated Ca 2ϩ entry when SOCE was stimulated with TG. Our results are in accordance with previously published observations, where annexin A6 null-mutant cardiomyocytes showed increased Ca 2ϩ transients and higher contractility (29), and A431 cells, lacking the endogenous annexin A6 showed higher EGF-dependent Ca 2ϩ entry than A431 cells "rescued" by introduction of ectopic annexin A6 (30).
Most annexins associate with the PM at high [Ca 2ϩ ] i , and thus could potentially influence Ca 2ϩ entry upon PM binding. To test whether the observed effects were annexin A6-specific, we expressed annexins A1 and A2 as fusions with the YFP and tK-and tH-targeting sequences, respectively. Annexin A1YFP-tK failed to reduce SOCE in stimulated cells, indicating the specificity of the annexin A6-mediated response. Similarly, Annexin A2YFP-tH did not significantly alter SOCE upon overexpression in HEK293 cells. These findings are somewhat surprising, taking into account the high Ca 2ϩ sensitivity of annexin A2 and its ability to bind actin (46). However, unlike annexin A6, annexin A2 appears to bind predominantly at the sites of active actin polymerization, which might not directly be linked to SOCE regulation (40,47). Additionally, it should be noted that annexin A2 exists as a heterotetrameric complex with S100A10 (p11) protein, and this interaction is essential for most of its functions. One could envisage that PM tethering of monomeric annexin A2 using the lipid anchors of Ras proteins does not allow heterotetramer formation with S100A10 (p11) and thus interferes with the function of annexin A2. In-depth evaluation of the effects of constitutive PM localization of this and other annexins awaits further study.
Interestingly, when annexin A6 was targeted to the PM by either tH, which predominantly associated with lipid rafts, or tK, which is primarily found in non-rafts, both constructs exhibited a similar inhibitory effect on SOCE. Apart from additional labeling of the Golgi FIGURE 9. The constitutive localization of annexin A6 at the plasma membrane promotes the stabilization of cortical actin. A, HEK293 cells, cotransfected with plasmids expressing annexin A6GFP-tK and YFP-␤-actin, or with YFP-␤-actin alone, were co-cultured for 48 h, fixed, and examined in the confocal microscope under the settings, which completely separated the GFP and YFP emission signals. Any possible bleed-through was controlled by imaging the cells, expressing either A6GFP-tK or YFP-␤-actin alone (not shown). The cells co-expressing YFP-␤-actin with A6GFP-tK accumulated F-actin under the PM (arrows). Bar ϭ 5 m. B, co-localization of annexin A6GFP-tK (here in green) and YFP-␤-actin (here in red) under the plasma membrane. The graph shows an intensity profile of the green and red signals, corresponding to A6GFP-tK and b-actin, along an arbitrary axis (arrow). There is a co-localization of the signals at the PM. Bar ϭ 5 m. C, live HEK293 cells co-expressing A6GFP-tK and YFP-␤-actin in Ca 2ϩ -Tyrode buffer were examined in the confocal microscope. Images of the same cells were taken before (control) and 2 min after addition of 5 M Lat A (ϩ Lat A). Lat A treatment quickly dispersed the patches of PM-associated F-actin (arrows). Bar ϭ 5 m. D, cells expressing YFP-␤-actin were left untreated (control), or exposed to 1 M jasplakinolide (Jasp) for 1 h, before fixation and examining in the confocal microscope. Bar ϭ 5 m. apparatus, which is characteristic for tH-tagged proteins (34), there was no visible difference in the intracellular distribution of the two proteins. The stronger attenuation of Ca 2ϩ entry by tH-tagged A6 was not statistically significant, but manifested itself in significantly reduced amounts of Ca 2ϩ stored in the ER, when the cells were maintained in low [Ca 2ϩ ] ex . These effects were reversed by Ca 2ϩ supplementation, indicating that tHtagged annexin A6 did not have a direct effect on ER loading or cause an excessive ER leak.
Calcium clearance counterbalances Ca 2ϩ entry to achieve a precise regulation of Ca 2ϩ homeostasis. In non-excitable HEK293 cells, the PMCA is the principal pump responsible for Ca 2ϩ efflux, and its activity has been shown to be stimulated by calmodulin and influenced by phosphorylation, hormones, and reactive oxygen species (48). Previously published studies indirectly implicated annexin A6 in the enhancement of neuronal Na ϩ /Ca 2ϩ exchanger (26). Here we tested the effect of the PMtargeted or recombinant purified annexin A6 on the activation of PMCA, and showed that there was no enhancement of the pump activity. A6YFP-tH-expressing cells showed the same rates of Ca 2ϩ clearance as the control, and inhibition of the PMCA in ATP-depleted cells did not abolish A6YFP-tH-mediated down-regulation of Ca 2ϩ entry.
PM-targeted annexin A6 attenuated both TG-induced SOCE, and EGF receptor-mediated Ca 2ϩ entry. The process of Ca 2ϩ entry is influenced by a multitude of proteins and lipids, modulating the coupling between ER and PM at the sites of SOCE stimulation, or acting directly on the Ca 2ϩ channel components. To clarify the role of annexin A6 in the down-regulation of Ca 2ϩ entry, we investigated the factors, which determine its PM binding and localization in addition to the H-and K-Ras lipid anchors. We used membrane fractionation in the presence of EGTA and Triton X-100, when both the endogenous annexin A6 and tH-or tK-tagged GFP had lost their ability to associate with membranes, and showed that A6YFP-tH and A6YFP-tK retained their membrane binding, probably due to the interaction of annexin A6 with the actin cytoskeleton. Cytoplasmic annexin A6 associates with the cortical cytoskeleton in a Ca 2ϩ -dependent and Ca 2ϩ -independent manner (41,49), and its complex with actin has been suggested to stabilize cardiomyocyte sarcolemma during cell stimulation (50). Results presented here demonstrate that annexin A6 binds actin in vitro and suggest that actin is a binding partner of both tH-and tK-tagged annexin A6 at the PM. Microscopic studies further strengthen this hypothesis, because A6GFP-tK-expressing cells exhibited a noticeable redistribution of F-actin filaments, and an increased sub-plasmalemmal cortical actin ring.
The actin cytoskeleton plays a major role in the regulation of SOCE (7)(8)(9). Agents, acting on cortical actin, are well known to interfere with the activation of SOCE: stabilization of the cortical cytoskeleton with jasplakinolide reduces SOCE in most cells (8), possibly by interfering with STIM1 function (39), whereas disruption of cytoskeleton with latrunculin A or cytochalasin D activates Ca 2ϩ entry (51). In agreement with membrane-anchored annexin A6 inhibiting SOCE via actin reorganization, we showed that Lat A treatment abolished the annexin A6YFP-tH-mediated attenuation of SOCE. These findings implicate the interaction of annexin A6 with cortical actin in the regulation of Ca 2ϩ entry into the cells. Although the exact mechanism by which annexin A6 exerts its effects on cytoskeletal organization and the concomitant activity of the SOCE machinery awaits further clarification, it is possible that, similar to jasplakinolide, the role of annexin A6 lies in PM recruitment and stabilization of cortical actin. Our results indicate that annexin A6 is a key player linking the actin cytoskeleton with SOCE, and its interaction with PM and subsequent stabilization of cortical actin might contribute to attenuation of Ca 2ϩ entry in vivo.
Many details of SOCE, including the functional STIM-Orai1 protein complex remain to be elucidated. Caveolae have been suggested as important Ca 2ϩ entry sites, and recently it has been shown that alterations of caveolin expression modulated SOCE (52). Likewise, it has been reported that the disruption of cholesterol-rich microdomains reduces Ca 2ϩ entry (53). We have previously shown that the membrane binding of annexin A6 is modulated by cholesterol, which is necessary for EGTAresistant association of annexin A6 with endosomal mem- FIGURE 11. The constitutive PM targeting of annexin A6 reduces cell proliferation. A, three HEK293 cell lines, stably expressing annexin A6YFP-tH, A6YFP-tK, and mCherry-tH, and the wild-type HEK293 cells were seeded in 12-well plates at 2.5 ϫ 10 5 cells/well, in triplicates. At times indicated, cells were detached by trypsinization, and the total number of cells per well counted after trypan blue exclusion. The graph shows the increase in absolute cell numbers observed in two independent proliferation assays, each time point in triplicate Ϯ SDEV. At 48 h there was a statistically significant reduction of cells per well in HEK-A6YFP-tH, and HEK-A6YFP-tK compared to the control (*, p Ͻ 0.05). B, the graph shows data from A for HEK-A6YFP-tH and HEK-A6YFP-tK relative to the number of HEK-mCherry-tH control cells at 18, 24, and 48 h post plating.