LO INMEDIATO Y LO MEDIATO, INMEDIATEZ Y MEDIATEZ

In agreement with earlier findings, B0AT1, ACE2 and APN were significantly enriched in BBMV compared to homogenised intestinal mucosa (Fig. 3.1A). The renal trafficking facilitator of B0AT1 collectrin, was not detected in BBMVs, but was detected in the intestinal mucosa homogenates, indicating expression outside the apical membrane. APN function was demonstrated using the peptidomimetic substrates L-leucine-p-nitroanilide or L-alanine-4-

107 nitroanilide (Fig. 3.1B). APN was estimated to comprise about 2.7 ± 0.3 % of total protein in BBMV and 0.5 ± 0.1 % of total protein from the intestinal homogenate based on relative enzymatic activity using a purified kidney porcine APN as a standard (Sigma-Aldrich).

Solubilisation of BBMV membrane proteins with two different detergents, followed by Blue- native gel electrophoresis revealed the presence of large protein complexes, containing B0AT1, APN and/or ACE2 (Fig. 3.2A). An overview of the complexes and their protein content is presented in Table 3.1. Solubilisation with digitonin (0.5% or 1% w/v) resulted in two large complexes (1 and 2) containing B0AT1 and ACE2 at 611 kDa and 558 kDa, respectively. A slightly smaller complex (3) at 488 kDa appeared to contain all three proteins and a smaller complex containing ACE2 only was detected at 376 kDa (4). The presence of the protein monomer for APN and ACE2 was confirmed by solubilisation in the non-ionic detergent Triton X-100 (5, 7). The size of B0AT1 (121 kDa) solubilised by Triton X-100 (6) suggested the presence of a homodimer. The absence of β-actin in every sample confirmed that solubilisation isolated lipid-soluble complexes (data not shown). The size of complexes suggests that B0AT1, ACE2 and APN could interact with additional brush-border membrane proteins (section 3.3).

Detection of membrane proteins in complexes of the same molecular weight provides evidence for protein-protein interaction, but the similarity of the molecular weight could be incidental. To provide further evidence of specific B0AT1 and APN complexes co- immunoprecipitation was used. Following solubilisation of BBMV, pull-down of B0AT1 revealed co-immunoprecipitation of APN and the pull-down of APN revealed the co- immunoprecipitation of B0AT1(Fig. 3.2B).

109 Figure 3.1 Isolation of intestinal brush-border membrane vesicles and APN activity

Murine intestinal BBMV were prepared using MgCl2 precipitation and centrifugation (section 2.2.2). A) Total BBMV or homogenate (homo) protein (20 µg) was separated by SDS-PAGE. Following semi-dry transfer onto a nitrocellulose membrane and blocking, individual proteins were detected and visualised using immunoblot analysis. B) APN activity in BBMV and intestinal homogenate was measured by a colorimetric assay using 6.5 mM L-alanine-4- nitroanilide. Each data point represents the mean ± S.D. of 8 to 10 individual time-courses for each experimental condition.

111 Figure 3.2 Detection of intestinal brush-border protein complexes

Murine intestinal BBMV were prepared using MgCl2 precipitation and centrifugation (section 2.2.2). A) BBMVs (50 µg/sample) at 1 mg/ml were solubilised with percentage v/v or w/v of detergent as indicated and mixed with Coomassie Brilliant Blue G-250 loading dye before separation. Samples were separated by native PAGE. Following semi-dry transfer onto a PVDF membrane, individual proteins were detected by immunoblot analysis. To allow a direct comparison of band positions the blot was stripped and re-probed for the proteins indicated. Protein complexes and individual proteins are numbered 1 to 7 (Table 3.1). B)

Murine BBMV were solubilised in co-immunoprecipitation buffer, followed by incubation with primary antibodies against either B0AT1 or APN and isolation using protein A. Following overnight incubation BBMV samples, cleared lysate, and intestinal homogenate were separated by SDS-PAGE. After semi-dry transfer onto nitrocellulose membranes, blots were pre-stripped and blocked o/n to reduce background signal. Individual proteins were detected as indicated above the blots.

112 Table 3.1: Molecular weight isolated from murine intestinal brush-border.

Protein complexes were identified using 4-16% non-continuous blue native PAGE (Fig. 3.2A). The protein standards, thyroglobulin (669 kDa), ferritin (440 kDa), and BSA (134 kDa and 69 kDa), were used to create a standard curve for calculating mean M.W. ± S.D. (n = 3).

Complex Solubilisation Conditions Proteins Detected M.W. (kDa ± S.D.)

1 digitonin 0.5%/1% w/v B0AT1, ACE2 611 ± 11 2 digitonin 0.5%/1% w/v B0AT1, ACE2 558 ± 5 3 all conditions B0AT1, ACE2, APN 488 ± 3 4 digitonin 0.5%/1% w/v ACE2 376 ± 6 5 Triton X100 0.5%/1% w/v APN 195 ± 3 6 Triton X100 0.5%/1% w/v B0AT1 126 ± 4 7 Triton X100 0.5%/1% w/v ACE2 91 ± 54

113 To examine if the isolated complexes were contiguous with detergent-resistant membrane micro-domains, BBMV were treated with Triton X-100 at 0 °C to 4 °C and the suspension separated by density equilibrium centrifugation on a linear sucrose gradient (section 2.5.2). All three proteins co-segregated to the Triton X-100 resistant fractions 7 and 8 (Fig. 3.3A). The majority of ACE2 and APN protein, however, remained in the detergent soluble protein fractions of the gradient bed (fractions 1-5, 40 % (w/v) sucrose), while a larger proportion of B0AT1 was retained in the floating lower density fractions. Interestingly, the peptidases, but not B0AT1, also appeared in the last two sucrose gradient fractions (14-15), corresponding to the lowest density area of the sucrose gradient. Caveolin-1, a lipid-raft marker from other tissues but not intestinal epithelial cells [493], was not detected in either the soluble or detergent-resistant fraction (Fig. 3.3B). Detergent resistant membranes are expected to appear in the lower density sucrose gradient fractions, corresponding to fractions 6 to 15, or 30% (w/v) to 5 % (w/v) of the linear sucrose gradient [494].

To further investigate the specificity of the APN/B0AT1 interaction we used a protein complementation assay in which both APN and B0AT1 were fused to halves of the enhanced GFP (eGFP) protein (Fig. 3.4A). Venus1 represents eGFP residues 1-157 and was fused to the intracellular N-terminus of APN (Venus1-APN), while Venus2 represents eGFPresidues 158-238 fused to either the intracellular N-terminus (Venus2-B0AT1) or C-terminus (B0AT1- Venus2) of B0AT1. If protein–protein interactions bring the two halves of eGFP closely together, fluorescence can be observed [495]. Transfection of B0AT1-Venus2 or Venus2- B0AT1 with Venus1-APN constructs in HEK293 cells, resulted in a significant green fluorescent signal above background, suggesting that both halves could recombine. When HEK 293 cells were co-transfected with Venus1-B0AT1 and Venus1-APN vectors, containing the same half of the eGFPprotein (residues 1 to 157), no fluorescence was

115 Figure 3.3 Isolation of detergent-resistant membranes from BBMV

A) BBMV (2 ml at 1 to 2 mg/ml) were treated with 1% (v/v) Triton X-100 at 0-4 °C and centrifuged with a 5-30% (w/v) linear sucrose gradient for 24 hrs. The gradient was fractionated and samples analysed by SDS-PAGE. Following semi-dry transfer onto a nitrocellulose membrane, individual proteins were visualised using immunoblot analysis as indicated next to each blot. Sucrose gradient fractions (1 ml) are numbered from highest (1) to lowest (15) density (e = 4). B) 30 µg of protein from the homogenates of listed organs or tissue samples were loaded onto gels and analysed using an anti-human caveolin antibody. Homo = intestinal homogenate, BBMV were isolated from mouse small intestine.

116

117 Figure 3.4 Co-localisation of B0AT1 and APN

HEK293 cells (> 90% confluent) were transfected with plasmid DNA in 8-well microscope slide dishes. The eGFP fluorescence was visualised with a Leica SP5 confocal system and processed with LAS AF software, the white scale bars = 25 µm. All images were taken with the ×63 magnification manual lens, with the right hand panel of each experimental group digitally magnified. A) Cartoons (left) represent the various paired B0AT1 and APN Venus Fly Trap constructs transfected and visualised in the confocal micrographs (right), including the non-B0AT1 and APN fusion constructs Venus1-zip and zip-Venus2. The + or – above the panels indicates whether or not cells were co-transfected with a pcDNA3.1+ expression vector encoding collectrin. B) Single B0AT1 and APN Venus Fly Trap controls (overleaf), including non-Venus Fly Trap eGFP expression at lower (left) and higher (right) digital magnification.

119 observed. In the absence of collectrin, the kidney-specific B0AT1-trafficking subunit (left panel), the fluorescence induced by the Venus2-B0AT1/Venus1-APN and B0AT1- Venus2/Venus1-APN vector pairs was observed largely inside the cells and not at the plasma membrane. In contrast, when the same vector pairs were co-transfected with collectrin (right panel), co-localisation of B0AT1/APN predominated at the plasma membrane. The transfection of individual eGFPhalves showed only background fluorescence (Fig. 3.4B).