A test of current models for the mechanism of milk-lipid droplet secretion

Abstract

Milk lipid is secreted by a unique process, during which triacylglycerol droplets bud from mammary cells coated with an outer bilayer of apical membrane. In all current schemes, the integral protein butyrophilin 1A1 (BTN) is postulated to serve as a transmembrane scaffold, which interacts either with itself or with the peripheral proteins, xanthine oxidoreductase (XOR) and possibly perilipin-2 (PLIN2), to form an immobile bridging complex between the droplet and apical surface. In one such scheme, BTN on the surface of cytoplasmic lipid droplets interacts directly with BTN in the apical membrane without binding to either XOR or PLIN2. We tested these models using both biochemical and morphological approaches. BTN was concentrated in the apical membrane in all species examined and contained mature N-linked glycans. We found no evidence for the association of unprocessed BTN with intracellular lipid droplets. BTN-enhanced green fluorescent protein was highly mobile in areas of mouse milk-lipid droplets that had not undergone post-secretion changes, and endogenous mouse BTN comprised only 0.5-0.7% (w/w) of the total protein, i.e. over 50-fold less than in the milk-lipid droplets of cow and other species. These data are incompatible with models of milk-lipid secretion in which BTN is the major component of an immobile global adhesive complex and suggest that interactions between BTN and other proteins at the time of secretion are more transient than previously predicted. The high mobility of BTN in lipid droplets marks it as a potential mobile signaling molecule in milk.

Keywords: butyrophilin; exocrine biology; lactation; milk-lipid secretion; mouse; perilipin-2; xanthine oxidoreductase.

Figure 1. Proposed mechanisms of milk-lipid secretion

(A) Electron micrograph of a lipid droplet at the point of secretion (lactating guinea-pig mammary gland), Bar 1.0 μm. (B) Postulated topology of BTN, XOR, and PLIN2 in secreted lipid droplets. BTN is a type 1 glycoprotein with two Ig domains (IgI and IgC1, green rectangles) in the exoplasmic (exo) domain. The two N glycosylation sites discussed in the text are indicated by black triangles. The B30.2 domain (black rectangle) in the cytoplasmic (cyto) region binds to XOR (grey ball). PLIN2 (blue rectangle), a member of the PAT protein family, binds to the phospholipid monolayer (light grey) on the lipid droplet surface (orange). (C) Proteins of bovine MLGM separated in one dimension by SDS-polyacrylamide gel electrophoresis. The major protein bands of XOR, BTN and PLIN2 and their apparent M_rs are indicated. Note that most of the 66 kDa protein band comprises BTN, unlike mouse MLGM. (D) Current molecular models for the secretion of milk-lipid droplets. (i) BTN (red bars) in the apical plasma membrane (APM) self associates and binds to XOR (green triangles) to form dimers or aggregates of higher order, which bind to proteins (black dots) on the surface of cytoplasmic lipid droplets (LD), possibly including PLIN2 or XOR (1). Formation of this macromolecular complex drives the expulsion of lipid droplets from the cell. (ii) PLIN2 (blue dots) on the surface of cytoplasmic lipid droplets directly binds to the phospholipid bilayer of the apical plasma membrane through a four-helix bundle motif at the C terminus. Binding induces membrane curvature and BTN and XOR are subsequently recruited into the budding lipid droplet (16). (iii) BTN on the surface of cytoplasmic lipid droplets binds to BTN in the apical plasma membrane to form a network of adhesive molecules encircling the droplet. XOR, PLIN2, or other proteins are not required for this interaction (18).

Figure 2. Distribution of BTN in established cell lines

Cells were co-transfected with pECFP-BTN, and pPLIN2-EYFP and the unfixed cells examined by microscopy. (A–C) Epifluorescent images of bovine MAC-T cells. (D–F) Confocal images of HEK 293T cells, which were treated with digitonin to remove excess cytoplasmic PLIN2-EYFP. Note that in each case, ECFP-BTN (green) is targeted to the plasma membrane (arrowheads) and intracellular aggregates (arrows), but does not associate with lipid droplets (red), which were identified by PLIN2-EYFP. Bars 10 μm.

Figure 3. Distribution of BTN in HC 11 cells

HC 11 cells were treated with insulin, dexamethasone and prolactin to induce XOR expression, and then transduced with Adv-BTN-EGFP. Cultures were subsequently stained with BODIPY 665 to localize lipid droplets and fixed with 4% (w/v) paraformaldehyde. Three-dimensional images of cells were constructed using Imaris software from serial confocal sections. (A,B) Example from a stack of 25 0.32 μm confocal sections. (A) BTN-EGFP; (B) BODIPY 665; (C–E) Confocal section 1.3 μm from the base of the cell, showing separate, (C) green (BTN-EGFP), and (D) red (BODIPY 665) channels in grey scale, and (E) a colored merged image; (F,G) y/z sections 1 and 2, as indicated in (E) showing separate green (G) and red (R) channels in grey scale, and colored merged images (M). Note BTN associates with large lipid droplets in the periphery of the cell (arrowheads in A,B,E,G) (section 2) but not with smaller droplets towards the center of the cell (section 1). Bars 5 μm.

Figure 4. Distribution of BTN in lactating mammary gland

(A–D) Mouse mammary gland transduced in vivo with Adv-BTN-EYFP, fixed in 4% paraformaldehyde and frozen in O.C.T. compound. (A) merged image of BTN-EYFP (green), nile-red stained lipid droplets (red), and DAPI-stained nuclei (blue); (B) BTN-EYFP (green), (C) nile-red stained lipid (red); (D) merged image of B and C; (E,F) Frozen sections of lactating mouse mammary gland labeled with (E) anti-peptide antibody to mouse BTN, and (F) mouse PLIN2, followed in both cases with goat anti-(rabbit IgG)-Alexa Fluor™-488. (G,H) Resin-embedded lactating bovine mammary gland sections labeled with anti-peptide antibody to (G) BTN, or (H) PLIN2, followed in each case with goat anti-(rabbit IgG)-colloidal gold (4 nm), and enhanced with silver. The same two alveoli in adjacent sections in G and H are labeled X and Y. Note in E and G that BTN is largely associated with the apical plasma membrane (single arrowheads), compared with PLIN2 in F and H which associates with cytoplasmic lipid droplets (CLD) (arrows). AL, alveolar lumen; N, nucleus; Bars (A–D) 5 μm; (E–H) 25 μm.

Figure 5. Digestion of mouse or bovine MLGM proteins with endo-H, or N- glycanase

Lipid droplets or MLGM samples were digested with either endo-H, or N-glycanase and the samples analyzed by immunoblotting after SDS-PAGE. (A) bovine lipid droplets; (B) bovine MLGM; (C) CD1 lipid droplets; (D) C57/Bl6 lipid droplets. Individual samples from five animals digested with endo-H are shown for each species or strain to the left, and time courses for the digestion of single samples with N-glycanase analyzed on the same blot are to the right as indicated. Densitometric analysis of the blots in (A–D) is shown in Table 2.

Figure 6. FRAP analysis of the mobility of BTN-EGFP in milk-lipid droplets and HC 11 cells

(A) Unfixed milk-lipid droplets containing BTN-EGFP showing examples of droplets with a “continuous” membrane (yellow arrowheads), and “condensed” membrane (yellow arrows); (B,C) Examples of FRAP analysis of the mobility of BTNEGFP in lipid droplets showing time-dependent recovery of fluorescence in bleached regions in, (B) a “continuous” area of membrane, and, (C) a “condensed” area of membrane; (D) Example of FRAP analysis of the mobility of BTNEGFP in the plasma membrane of HC 11 cells; (E) Summary of FRAP analysis of the mobility of BTN-EGFP in lipid droplets and HC 11 cell plasma membrane, as indicated. Numbers to the right of the graph indicate the number of estimates in each case. Bleached areas in B–D are indicated by dotted circles before photobleaching and arrowheads after photobleaching; Bars, (A) 50 μm, (B–D) 5 μm.

Figure 7. Comparison of the protein composition of milk-lipid droplets from Btn^+/+ and Btn^−/− mice by oneand two-dimensional gel electrophoresis and immunoblot

Separation of the proteins associated with unfractionated lipid droplets from, A (a–c) Btn^+/+, and B (d–f) Btn^−/− mice. Gels were stained with Coomassie blue. The isoelectric focusing (IEF) pH gradients increase from left to right, and M_rs of proteins separated by SDS-PAGE decrease from top to bottom. One– dimensional SDS gels are shown to the left of each figure and the 66 kDa protein band containing BTN is indicated with an asterisk (*). Detailed analyses of the 66 kDa band by two-dimensional gel electrophoresis and immunoblot are shown in (a–c) for Btn^+/+ mice and (d–f) for Btn^−/− mice. (a,d) The 66 kDa region stained with Coomassie blue (CB). (b,e) Immunoblots of the same regions in a,d stained with anti-peptide antibody to mouse BTN. (c,f) Immunoblots of the same regions in a and d stained with a commercial antibody to Mfg-e8. Note that the BTN variants, indicated by a “fence” in a,b, are minor components of this M_r region of the gel and that variants of Mfg-e8, indicated by a bracket in a,c,d,f are major components.

Figure 8. Analysis of the morphology of mouse MLGM by electron microscopy

Mouse MLGM samples were prepared for standard transmission electron microscopy. (A) Survey micrograph at low magnification; (B–D) Examples of the heterogeneous nature of mouse MLGM at higher magnification. Note that the mouse membrane lacks the extensive protein coat seen on the cytoplasmic face of bovine MLGM (6), and that the membrane forms sheets in B, and numerous vesicles in C,D, some of which contain electron dense material (example in D); (E) MLGM after treatment with papain. Note that the morphology is similar to untreated membrane (compare C and E), unlike bovine MLGM, which is depleted of the associated protein coat after treatment with protease [see Figure 2 of ref (30) for an example]; (F) MLGM of Btn^−/− mice is similar to that of wild-type mice. Bars, (A) 1 μm; (B–F) 200 nm.