Mapping bilayer thickness in the ER membrane

Abstract

In the plasma membrane and in synthetic membranes, resident lipids may laterally unmix to form domains of distinct biophysical properties. Whether lipids also drive the lateral organization of intracellular membranes is largely unknown. Here, we describe genetically encoded fluorescent reporters visualizing local variations in bilayer thickness. Using them, we demonstrate that long-chained ceramides promote the formation of discrete domains of increased bilayer thickness in the yeast ER, particularly in the future plane of cleavage and at ER-trans-Golgi contact sites. Thickening of the ER membrane in the cleavage plane contributed to the formation of lateral diffusion barriers, which restricted the passage of short, but not long, protein transmembrane domains between the mother and bud ER compartments. Together, our data establish that the ER membrane is laterally organized and that ceramides drive this process, and provide insights into the physical nature and biophysical mechanisms of the lateral diffusion barriers that compartmentalize the ER.

Fig. 1. Design of ER-targeted fluorescent reporters with a transmembrane domain varying in length.

(A) A collection of GFP-labeled WALP peptides is targeted and retained in the ER membrane through a Suc2 signal peptide sequence (SS) and a KKXX motif. The diagram depicts the various conformations WALP (reporter) and Sec66 (reference) TMDs can assume in an idealized bilayer containing shorter unsaturated phospholipids (thin bilayer) or longer saturated ceramides (thick bilayer). The TMD lengths are indicated in amino acids (AA) and in Å. (B) Images of cells coexpressing the reference marker Sec66-mCherry and the reporter proteins GFP-WALP21 and GFP-WALP29. The individual channels and the composite image are shown. Scale bar, 2 μm.

Fig. 2. Statistical analysis of the signal in ER pixels.

(A) Plot of the normalized GFP intensity of each ER pixel as a function of their normalized mCherry intensity. In each channel (GFP and mCherry), the median intensity is set to 1 in each individual cell (fig. S1 and Materials and Methods). ER pixels are identified by thresholding the sum of the GFP and mCherry signals (fig. S1 and Materials and Methods). Data are shown for 20 cells each of strains coexpressing Sec66-mCherry with GFP-WALP21 (red) and GFP-WALP29 (blue). (B) Close-up of the indicated region in (A). (C) Frequency distribution of the GFP:mCherry ratio (log₁₀) for all ER pixels shown in (A) for the GFP-WALP21–expressing cells. The Gaussian model of this distribution is indicated. (D) Same as (C) for the ER pixels of the GFP-WALP29–expressing cells. The Gaussian model of (C) (WALP21 model) and the mixed model for GFP-WALP29 are shown. Dark gray bars represent the residuals after applying the WALP21 model to the distribution. (E) Distribution of the pixels deviating significantly from the null hypothesis (WALP21 model) at threshold P values of P = 0.05 (magenta) and P = 0.01 (yellow). The same cells as in Fig. 1B are shown. Each channel is shown, as well as the ratiometric image of the ER pixels, onto which the significant pixels are mapped. Scale bars, 2 μm. The significant dots observed in the GFP-WALP29:Sec66-mCherry ratiometric image represent RITs.

Fig. 3. RITs localize to two distinct ER membrane domains.

(A) Ratiometric analysis of cells expressing either a short TMD (GFP-WALP21) or long TMD (GFP-WALP29) together with Sec66-mCherry. Close-up images of the indicated bud neck regions are shown. (B) Quantification of the bud neck WALP:Sec66 log ratio relative to the rest of the cortical ER (see fig. S1 and Materials and Methods) for the indicated GFP-WALP reporters in wild-type and mutant cells of indicated genotype. Medians and 95% confidence intervals are shown. Asterisks in black depict statistical significance compared to WALP21, while red asterisks depict whether the means are different from “0.” *P < 0.05, ****P < 0.0001; ns, not significant. n > 100 cells. (C) Intensity scan through RITs. The normalized GFP and mCherry intensity of line scans through 27 RITs is averaged and plotted after aligning all RITs relative to their center (right). The GFP/mCherry ratio (log) is also plotted and color-coded according to significance as in (A). The close-up images of the GFP, mCherry, and ratiometric channels are shown for a RIT example (left). (D) Relative localization of GFP-WALP29 (blue) and the TGN marker Sec7 (labeled with mCherry, red). The images shown were acquired by super-resolution microscopy (Airyscan) to enhance resolution. The individual and merged channels, as well as close-up of the indicated area (yellow box), are shown. (E) Quantification of the relative localization of the different markers in (D). Medians with corresponding SDs are depicted (n > 100 cells per clone, three independent clones). Scale bars, 2 μm.

Fig. 4. Ceramide length modulates the localization of WALP23 and WALP25 reporters.

(A and C) Ratiometric imaging of indicated GFP-WALP proteins coexpressed with Sec66-mCherry in cells of indicated genotype and producing C18:24/26 ceramide (A) or C18:16 ceramide (C, short ceramide). The individual channels, as well as the ratiometric images with significant pixels highlighted in magenta (P < 0.05) and yellow (P < 0.01), are shown. Scale bar, 2 μm. (B and D) Schematic representations of the various conformations W(AL)Ps can assume in a bilayer containing unsaturated phospholipids (left side of panels) or in a phospholipid/ceramide mixture (right side of the panels). (B) C18:26 ceramide; (D) C18:16 ceramide.

Fig. 5. Effect of ceramide length on the compartmentalization of the ER.

(A) Quantification of the relative bud neck WALP29:Sec66 log ratio in cells of indicated genotype. Median with 95% confidence intervals are shown. *P < 0.05, ****P < 0.0001; n > 100 cells. (B) Metaphase cells expressing Sec61-GFP were subjected to constant photobleaching (yellow box, bleached area); elapsed time is indicated in each movie frame (scale bar, 2 μm). The graph shows the average fluorescence decay kinetics of the monitored areas (mother, orange; bud, green; n = 15 cells). The Barrier index (B.I.₅₀) is defined as the ratio of the bud fluorescence half-life (t_bud) over the mother half-life (t_mom). (C) Means and SEM of the B.I.₅₀’s of cells expressing different lengths of GFP-WALP as a function of ceramide length (mean is from two to four biological replicate with n ≥ 10 cells per clone). WALP17 is not determined. (D) The means and SEM of the B.I.₅₀’s of cells expressing Sec61-GFP as reporter are plotted. Each circle corresponds to the mean of a biological replicate (n ≥ 10 cells per clone). *P < 0.05, ****P < 0.0001; ns, not significantly different to wild-type. (E) Schematic model of the yeast ER with thick, ceramide-rich domains (green, RITs) associated with the TGN (red) in the mother and bud but not at the bud neck.