Fig 2: Proposed mechanisms by which endosome lumen acidification regulates Rab7 activity and endosome maturation, a process mediated by proton efflux through the Christianson Syndrome protein NHE6 and the pH-dependent Rab7 GAP TBC1D5.During endosome maturation, the v-ATPase pumps protons into the endosomal lumen. Early in endosome maturation, as the proton concentration is relatively lower (higher pH) i), proton efflux through NHE6 is limited, and TBC1D5 is highly active, accelerating Rab7-GTP hydrolysis. This promotes conversion to inactive Rab7-GDP. Consequently, late endosomes are not efficiently fused with lysosomes. As endosomes mature, the v-ATPase more actively pumps protons into the endosomal lumen. ii), During this phase, as proton concentration is higher (lower pH), NHE6 leaks protons from late endosomes. As late endosomes gradually acidify, the cytoplasmic microenvironment adjacent to NHE6 on late endosomes is progressively acidified. This change in local pH inactivates TBC1D5 by protonation of His207 in the GAP domain. The GAP activity of TBC1D5 on Rab7 is thereby inhibited, and as a result, Rab7 is converted to the active GTP-bound form, driving a molecular cascade leading late endosomes to fuse with lysosomes.

Fig 3: Regulation of Rab7 activity requires NHE6-mediated proton efflux.a, Schematic of the luminal pH measurement. Primary neurons are loaded with pH-sensitive, FITC-conjugated transferrin (Tfn) and pH-insensitive Alexa Fluor546-Tfn. More orange color indicates Tfn-positive endosomes with lower luminal pH.b, Representative images of the neurites and soma from WT and NHE6-ED neurons loaded with FITC-Tfn and Alexa Fluor546-Tfn. The decreased pH of endosomes in the neurites and soma of NHE6-ED neurons. Scale bar: 5 μm.c, Lower luminal pH of transferrin-positive endosomes in the neurites and soma of NHE6-ED neurons. The calibration curve was used to determine the luminal pH of transferrin-positive endosomes. Each dot indicates each pup (WT = 8, NHE6-ED = 9 pups, n = 10,000–14,000 neurons imaged from each pup). Two-tailed unpaired t-test with Welch’s correction.d, Representative kymographs show the motility of mEmerald-Rab7 endosomes in primary WT and NHE-ED neurons. Retrograde direction is indicated. Scale bar, 2 μm. Increased numbers of stationary mEmerald-Rab7 endosomes along with decreased retrograde numbers were detected in primary NHE6-ED neurons. Means from each animal (big dots, WT = 3, NHE6-ED = 3 pups) are overlay the entire dataset (small dots, n = 19–28 neurons from three independent experiments) and used for statistical analysis. Animals from same litters are color-coded. Ordinary one-way ANOVA with Tukey’s HSD.e, Decreased RILP-bound Rab7 (GTP-Rab7 pull-down) in NHE6-ED mouse brains at 2 months. Lysates from WT and NHE6-ED mouse brains were incubated with His-RILP recombinant proteins or His-Myc (negative control) for the pull-down assay. After elution, Rab7 was detected by western blot. The densitometry of GTP-Rab7 was divided by the total Rab7 (3 brains for each genotype). Two-tailed unpaired t-test with Welch’s correction.f, Slower FRAP recovery of mEmerald-Rab7 in NHE6-ED neurons. Yellow circles indicate where photobleaching occurs and the fluorescence intensity was measured (WT = 3, Null = 4 pups, n = 30–47 neurons from three independent experiments). Non-linear mixed model was conducted. Scale bar, 5 μm. Scale bar for time course images, 0.1 μm.g, Increased co-localization of Rab7 (green) and TBC1D5 (magenta) in NHE6-ED neurons. Quantification shows the Mander’s coefficient (M1: Rab7 overlapping TBC1D5, M2: TBC1D5 overlapping Rab7) from high-content imaging. Each dot indicates a mean of each animal (WT = 4, Null = 4 pups, n = 10,000–14,000 neurons imaged from each pup). Animals from same litters are color-coded. Two-tailed unpaired t-test with Welch’s correction. Scale bar, 5 μm. Data are represented as mean ± SEM. ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.

Fig 4: A conserved histidine in the GAP domain of TBC1D5 is critical for the pH sensing function.a, The co-localization of NHE6 (cyan), TBC1D5 (magenta), and Rab7 (yellow) in primary neurons at DIV7. Triple labelling is shown in white. An orange box indicates the inset. On average, 30.98 % (± 3.03%) of Rab7-punctae were co-stained with TBC1D5 and NHE6. Means from each animal (big dots, 5 WT pups) overlay the entire dataset (small dots, n = 40 neurons from five independent experiments). Intensity line profile shows the fluorescence intensity of each protein along a dashed line within the inset. The images were acquired using Nikon NSPARC. Data are represented as mean ± SEM. Scale bar, 5 μm. Scale bar for the inset, 1 μm.b, AlphaFold 370 model showing a TBC1D5 dimer (upper top, green and turquoise) directly bound to the C-terminus of a NHE6 dimer (cyan and purple), which is embedded in a lipid bilayer. GTP-Rab7 (gray) is correctly positioned on the catalytic domain of TBC1D5 in this model. Note location of an interfacial histidine (H207) proximal to the invariant Arginine finger/aspartate and Glutamine finger/tyrosine. The AlphaFold 3 model was generated using the full-length NHE6 sequence. Note that N-terminal signal peptides was not present in the cryo-EM structure71 of NHE6 due to cleavage after Leu32 and also disordered structure between residues 32–73.c, Alignment of representative TBC domain orthologues and paralogues from the three major phylogenetic supergroups105. An interfacial histidine of TBC1D5 (H207) is conserved in orthologues (C.elegans RBG-3 and Saccharomyces cerevisiae Gyp6) and in one clade of paralogues (S. cerevisiae Gyp8 and human TBC1D20) but not in other paralogues.d, A conserved histidine located in the TBC domain is critical for the pH sensing function of TBC1D5. Catalytic efficiency (kcat/KM) of the C.elegans TBC1D5 orthologue is strongly dependent on pH. H166 of C.elegans TBC1D5 corresponds to H207 of human TBC1D5. The solid line in the graph represents a maximum likelihood-fitted model describing the pH dependence of one or more titratable groups. The data is normalized to the maximum value of the fitted model. In the WT, catalytic efficiency is strongly reduced with decreasing pH, whereas the magnitude of the pH effect is substantially diminished in the mutants. For WT, the fitted pKa is 6.0 ± 0.11. Two mutants, H166N, and H166S, modestly increased pKa values (H166N: 6.4 ± 0.31; H166S: 6.5 ± 0.57). WT reaches a maximum kcat/KM of ~ 8700 ± 1100 M−1 s−1 with a large dynamic range of pH response, reflected in a maximum fractional difference (Max - Min / Max) ~100 %. H166N retains high maximal GAP activity (~ 4100 ± 300 M−1 s−1); however, the magnitude of the pH response is substantially diminished with a maximum fractional difference (Max - Min / Max) = ~56 %. H166S has substantially reduced maximum GAP activity (~ 880 ± 110 M−1 s−1) and the magnitude of the pH response is also substantially diminished with a maximum fractional difference (Max - Min / Max) = ~39%. (For WT, n = 2–5 experiments; for H166N and H166S, n = 4 experiments).

Fig 5: Knock-down of the Rab7 GAP TBC1D5 rescues endosomal phenotypes in NHE6-null rat neurons.a, Reduced protein expression of TBC1D5 in primary littermate WT and NHE6-null rat neurons, with lentiviral transductions of anti-TBC1D5 shRNA, as compared to scrambled shRNA control transductions The expression of NHE6 and Rab7 was not affected by the shRNA transduction (three independent experiments). Two-way ANOVA with Fisher’s LSD test was performed.b, Representative kymographs from primary littermate WT and NHE6-null neurons with scrambled or TBC1D5 shRNA show the motility of mEmerald-Rab7 endosomes. Retrograde direction is indicated. Scale bar, 1 μm. The number of retrograde Rab7 endosomes in primary NHE6-null neurons transduced with TBC1D5 shRNA was increased compared to NHE6-null neurons transduced with scrambled shRNA. Also, the number of stationary Rab7 endosomes was decreased in primary NHE6-null neurons transduced with TBC1D5 shRNA. Means from each animal (big dots, WT = 3, Null = 3 pups) are overlay the entire dataset (small dots, n = 21–31 neurons from three independent experiments) and used for statistical analysis. Animals from same litters are color-coded. Ordinary one-way ANOVA with Tukey’s HSD was performed.c, The FRAP recovery rate of mEmerald-Rab7 in primary NHE6-null neurons transduced with TBC1D5 shRNA exhibited significant increase relative to NHE6-null neurons with scrambled shRNA. The recovery rate of mEmerald-Rab7 in NHE6-null neurons transduced with TBC1D5 shRNA was comparable to WT neurons (6 pups in total for each genotype, n = 20–35 neurons from 3 experiments). Non-linear mixed model was conducted.d, The increased endosomes-lysosomes fusion in primary NHE6-null neurons transduced with TBC1D5 shRNA. Shown are still images at different time points from live imaging of the endosome-lysosome fusion assay on primary littermate WT and NHE6-null neurons transduced with scrambled or TBC1D5 shRNA at different time points. The level of endosome-lysosome fusion is presented as the percentage fold change in overlap between the endosome label (magenta) and the lysosome label (green), from time point 0 for the same animal. Yellow arrows indicate the fusion events. (WT = 5, Null = 5 pups, n = 10,000–14,000 neurons imaged from each pup). Linear mixed model was conducted. Scale bar, 5 μm. Data are represented as mean ± SEM. ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05.