Evidence for annexin A6-dependent plasma membrane remodelling of lipid domains

Abstract

Background and purpose: Annexin A6 (AnxA6) is a calcium-dependent phospholipid-binding protein that can be recruited to the plasma membrane to function as a scaffolding protein to regulate signal complex formation, endo- and exocytic pathways as well as distribution of cellular cholesterol. Here, we have investigated how AnxA6 influences the membrane order.

Experimental approach: We used Laurdan and di-4-ANEPPDHQ staining in (i) artificial membranes; (ii) live cells to investigate membrane packing and ordered lipid phases; and (iii) a super-resolution imaging (photoactivated localization microscopy, PALM) and Ripley's K second-order point pattern analysis approach to assess how AnxA6 regulates plasma membrane order domains and protein clustering.

Key results: In artificial membranes, purified AnxA6 induced a global increase in membrane order. However, confocal microscopy using di-4-ANEPPDHQ in live cells showed that cells expressing AnxA6, which reduces plasma membrane cholesterol levels and modifies the actin cytoskeleton meshwork, displayed a decrease in membrane order (∼15 and 30% in A431 and MEF cells respectively). PALM data from Lck10 and Src15 membrane raft/non-raft markers revealed that AnxA6 expression induced clustering of both raft and non-raft markers. Altered clustering of Lck10 and Src15 in cells expressing AnxA6 was also observed after cholesterol extraction with methyl-β-cyclodextrin or actin cytoskeleton disruption with latrunculin B.

Conclusions and implications: AnxA6-induced plasma membrane remodelling indicated that elevated AnxA6 expression decreased membrane order through the regulation of cellular cholesterol homeostasis and the actin cytoskeleton. This study provides the first evidence from live cells that support current models of annexins as membrane organizers.

Figure 1

AnxA6 modifies membrane order in LUVs. Normalized fluorescence spectra of Laurdan stained (A) PC/PS/PE, (B) PC/PS/PE/Chol, (C) PC/PIP₂/PE and (D) PC/PIP₂/PE/Chol LUVs in the absence or presence of purified porcine AnxA6 (5 μg). Mean and SD graphical representation of its corresponding GP values. See Methods for the preparation of LUV details. The presented spectra are representative of five independent experiments (three replicates per condition for each experiment), where Lo (440 nm) and Ld (490 nm) emission wavelengths are represented. The means of three replicates from five independent experiments were used for statistical analysis. Unpaired Student's t-test showed statistically significant differences. *P < 0.05, **P < 0.01, ***P < 0.001. Chol, cholesterol.

Figure 2

Membrane order of AnxA6-overexpressing A431 and AnxA6 knockout MEF cells. A431-WT, A431-A6, MEF-WT and MEF-A6ko cells were stained with 1.5 μg·mL⁻¹ di-4-ANEPPDHQ for 30 min and imaged with a confocal microscope. (A) Representative GP pseudocoloured images of A431 and MEF cells. Bar, 10 μm. GP value histogram graphical representations of di-4-ANEPPDHQ stained (B) A431 and (C) MEF images from (A). (D) Mean and SD of A431-WT versus A431-A6 and (E) MEF-WT versus MEF-A6ko ΔGP representation under normal conditions, 30 min 10 mM mβCD and 10 min 5 μM LatB treatments of di-4-ANEPPDHQ stained images. The mean GP values of five images from five independent experiments were used to generate the ΔGP values for the statistical analysis. Two-way anova tests were conducted on (C) and (D), and statistically significant interaction between AnxA6 levels and drug treatment (F(2, 84) = 8.907, P = 0.0003 in A431 cells; F(2, 66) = 11.20, P < 0.0001 in MEF cells) was determined. Bonferroni post-test analysis showed significant differences for drug treatment (#P < 0.05, ##P < 0.01, ###P < 0.001) and AnxA6 expression (*P < 0.05, **P < 0.01, ***P < 0.001).

Figure 3

Cluster analysis of Lck10-PS-CFP2 and Src15-PS-CFP2 in A431-WT and A431-A6 cells. A431-WT and A431-A6 cells were transfected with (A and B) Lck10-PS-CFP2 and (C and D) Src15-PS-CFP2 and fixed 20 min with 4% PFA. PALM images were acquired and cluster analysis of 35–50 non-overlapping 3 × 3 μm regions at the plasma membrane from 7 to 10 PALM images was performed as explained in Materials and Methods. (A) Graphical representation of mean Ripley's K functions of 35–50 non-overlapping regions of Lck10-PS-CFP2 in A431-WT and A431-A6 cells. It reports the degree of clustering relative to a random distribution (indicated by the 95% CI, grey dotted line). (B) Graphical representation of mean ± SEM of maximum L(r)-r at radius = 60 nm, molecules in cluster, cluster density and cluster radius of Lck10-PS-CFP2 in A431-WT and A431-A6 cells. (C) Mean Ripley's K functions of 35–50 non-overlapping regions of Src15-PS-CFP2 in both A431-WT and A431-A6 cells. Grey dotted line, 95% CI. (D) Graphical representation of mean ± SEM of maximum L(r)-r at radius = 60 nm, molecules in cluster, cluster density and cluster radius of Src15-PS-CFP2 in A431-WT and A431-A6 cells. Unpaired Student's t-test showed statistically significant differences in (D). **P < 0.01, ***P < 0.001.

Figure 4

Cluster analysis of Lck10-PS-CFP2 and Src15-PS-CFP2 in MEF-WT and MEF-A6ko cells. MEF-WT and MEF-A6ko cells were transfected with (A and B) Lck10-PS-CFP2 and (C and D) Src15-PS-CFP2 and fixed 20 min with 4% PFA. PALM images were acquired and cluster analysis of 35–50 non-overlapping 3 × 3 μm regions at the plasma membrane from 7 to 10 PALM images was performed as explained in Materials and Methods. (A) Graphical representation of mean Ripley's K functions of 35–50 non-overlapping regions of Lck10-PS-CFP2 in both MEF-WT and MEF-A6ko cells. Grey dotted line, 95% CI. (B) Graphical representation of mean ± SEM of maximum L(r)-r at radius = 60 nm, molecules in cluster, cluster density and cluster radius of Lck10-PS-CFP2 in MEF-WT and MEF-A6ko cells. (C) Mean Ripley's K functions of 35–50 non-overlapping regions of Src15-PS-CFP2 in both MEF-WT and MEF-A6ko cells. Grey dotted line, 95% CI. (D) Graphical representation of mean ± SEM of maximum L(r)-r at radius = 60 nm, molecules in cluster, cluster density and cluster radius of Src15-PS-CFP2 in MEF-WT and MEF-A6ko cells. Unpaired Student's t-test showed statistically significant differences in (B) and (D). *P < 0.05, ***P < 0.001.

Figure 5

Ripley's K function of Lck10-PS-CFP2 and Src15-PS-CFP2 in mβCD and LatB-treated A431 and MEF cells. Mean Ripley's K function graphical representations of 35–50 non-overlapping 3 × 3 μm regions at the plasma membrane from 7 to 10 PALM images of (A and B) A431-WT and A431-A6 and (C and D) MEF-WT and MEF-A6ko of (A and C) Lck10-PS-CFP2 and (B and D) Src15-PS-CFP2 transfected cells. Graphs show the mean of Ripley's K functions under normal conditions, 10 mM mβCD and 5 μM LatB treatments. Grey dotted line, 95% CIs.

Figure 6

Proposed model for AnxA6-induced membrane organization. The proposed model suggests that fibroblasts plasma membrane (A) has lipid raft confined into actin fibre corrals due to its cholesterol content and its cortical actin cytoskeleton. (B) When AnxA6 is knocked out, higher plasma membrane cholesterol content and a prominent cortical actin cytoskeleton can be observed, which could drive domain partitioning of both Lo and Ld phases. On the other hand, epithelial cells (C and D) might have non-rafts confined into corrals where cortical actin cytoskeleton determines corral size and cluster partitioning. (D) In this setting, AnxA6 expression may induce diminution of cortical actin cytoskeleton and plasma membrane cholesterol, allowing larger non-raft Ld domains.