Clinical or ATPase domain mutations in ABCD4 disrupt the interaction between the vitamin B12-trafficking proteins ABCD4 and LMBD1

Abstract

Vitamin B₁₂ (cobalamin (Cbl)), in the cofactor forms methyl-Cbl and adenosyl-Cbl, is required for the function of the essential enzymes methionine synthase and methylmalonyl-CoA mutase, respectively. Cbl enters mammalian cells by receptor-mediated endocytosis of protein-bound Cbl followed by lysosomal export of free Cbl to the cytosol and further processing to these cofactor forms. The integral membrane proteins LMBD1 and ABCD4 are required for lysosomal release of Cbl, and mutations in the genes LMBRD1 and ABCD4 result in the cobalamin metabolism disorders cblF and cblJ. We report a new (fifth) patient with the cblJ disorder who presented at 7 days of age with poor feeding, hypotonia, methylmalonic aciduria, and elevated plasma homocysteine and harbored the mutations c.1667_1668delAG [p.Glu556Glyfs*27] and c.1295G>A [p.Arg432Gln] in the ABCD4 gene. Cbl cofactor forms are decreased in fibroblasts from this patient but could be rescued by overexpression of either ABCD4 or, unexpectedly, LMBD1. Using a sensitive live-cell FRET assay, we demonstrated selective interaction between ABCD4 and LMBD1 and decreased interaction when ABCD4 harbored the patient mutations p.Arg432Gln or p.Asn141Lys or when artificial mutations disrupted the ATPase domain. Finally, we showed that ABCD4 lysosomal targeting depends on co-expression of, and interaction with, LMBD1. These data broaden the patient and mutation spectrum of cblJ deficiency, establish a sensitive live-cell assay to detect the LMBD1-ABCD4 interaction, and confirm the importance of this interaction for proper intracellular targeting of ABCD4 and cobalamin cofactor synthesis.

Keywords: ABC transporter; ABCD4; LMBD1; cblF; cblJ; fluorescence resonance energy transfer (FRET); homology modeling; inborn error of metabolism; protein-protein interaction; vitamin B12.

Figure 1.

Cobalamin cofactor synthesis rescue in cblF and cblJ patient fibroblasts. A, transfection of LMBRD1 (cblF) and ABCD4 (cblJ) WT alleles in immortalized fibroblasts of the two original patients with cblJ defect (cblJ01 and cblJ02), cblJ05, and a patient with the cblF defect. Transfections with empty vector (vector only) were used as negative controls. Mean plus S.D. is shown. B, cblF patient fibroblasts were transfected with DNA coding for fluorescent protein only (GFP), untagged LMBD1 in pTracer-CMV2 (LMBD1), or fluorescently tagged-LMBD1 (LMBD1-GFP). cblJ patient fibroblasts were transfected with DNA coding for fluorescent protein only (fRFP only), untagged ABCD4 in pTracer-CMV2 (ABCD4), or fluorescently tagged-ABCD4 (ABCD4-fRFP) ± LMBD1-GFP. Immortalized control fibroblasts without transfection (n = 23) are shown for comparison. Bars represent the mean and error bars S.D. from at least three separate experiments.

Figure 2.

Colocalization of ABCD4-fRFP and LMBD1-GFP with endogenous LAMP1. A, confocal microscopy images of immortalized control fibroblasts co-transfected with LMBD1-GFP and ABCD4-fRFP and stained with LAMP1 antibody to localize lysosomes. B, LMBD1-GFP overexpression compared with LAMP1 staining. Regions in which both proteins co-localize appear yellow in the merged image. C, ABCD4-fRFP overexpression compared with LAMP1 staining. The absence of yellow in the merged image indicates a lack of colocalization. For A–C, the white scale bar is 10 μm. Arrowheads depict examples of overlap between all three markers.

Figure 3.

Analysis of protein interactions in immortalized fibroblasts using flow cytometry-based FRET. A, histogram view of GFP intensity (% GFP) and fRFP intensity (% fRFP) for each (co)transfection. Cells were considered GFP+ or fRFP+ if intensity was >10² in each respective channel. From cells that were GFP+ and fRFP+, the MI of signal in the FRET channel was calculated. B, FRET intensity plotted against GFP intensity, where FRET+ cells are defined by a stick-shaped gate encompassing signal from conjugated LMBD1–ABCD4 above the intensity threshold of 10² for both GFP and FRET (left panel). Pictures depict a representative experiment, and the number in purple represents the average percentage of all live cells that are FRET+ for each condition (n ≥ 3). C, bar graph of the percent FRET+ positive cells from those cells that were GFP+ and fRFP+ for each condition. Data represent at least three separate experiments. The error bar represents S.D. D, table summarizing the data presented in panels B and C.

Figure 4.

Missense mutations from patients and in the ATPase domain of ABCD4 can disrupt interaction with LMBD1. A, left, FRET intensity plotted against GFP intensity, where FRET+ cells are defined as in Fig. 3B. The number in purple represents the percentage of all live cells that are FRET+ for each condition. Right, bar graph of the percent FRET+-positive cells from those cells that were GFP+ and fRFP+ for each condition. For each mutation % FRET+ cells are given as percent wild-type ABCD4. Data represent at least three separate experiments. The error bar represents S.D. B, merged confocal images of fibroblasts co-transfected with combinations of mutant ABCD4-fRFP and wild-type LMBD1-GFP. Regions in which both proteins co-localize appear yellow. The white square indicates the zoomed-in region below. The scale bar is 10 μm. Numbers indicate Pearson correlation coefficient. C, rescue of AdoCbl and MeCbl synthesis in immortalized fibroblasts of an ABCD4-deficient (cblJ02) patient after transfection with mutant ABCD4-fRFP proteins. The error bar represents S.D. Data represent a single experiment performed in triplicate.

Figure 5.

Homology modeling of the ATPase domain of ABCD4. A, schematic of ABCD4 protein sequence (top) and topology (bottom) with location of mutations used in the study indicated. Sequence and topology as described in Coelho et al. (4). The ATP domain highlighted in gray is the region used to generate the homology model (see also supplemental Fig. S9). B, two views of the ATPase domain model showing secondary structures (gray ribbon), an ADP molecule docked to the nucleotide-binding site, and the side chain of the residue (Arg-432) affected by the missense CblJ patient-five mutation.