A novel live cell imaging assay reveals regulation of endosome maturation

Cell-cell communication is an essential process in life, with endosomes acting as key organelles for regulating uptake and secretion of signaling molecules. Endocytosed material is accepted by the sorting endosome where it either is sorted for recycling or remains in the endosome as it matures to be degraded in the lysosome. Investigation of the endosome maturation process has been hampered by the small size and rapid movement of endosomes in most cellular systems. Here, we report an easy versatile live-cell imaging assay to monitor endosome maturation kinetics, which can be applied to a variety of mammalian cell types. Acute ionophore treatment led to enlarged early endosomal compartments that matured into late endosomes and fused with lysosomes to form endolysosomes. Rab5-to-Rab7 conversion and PI(3)P formation and turn over were recapitulated with this assay and could be observed with a standard widefield microscope. We used this approach to show that Snx1- and Rab11-dependent endosomal recycling occurred throughout endosome maturation and was uncoupled from Rab conversion. In contrast, efficient endosomal acidification was dependent on Rab conversion. The assay provides a powerful tool to further unravel various aspects of endosome maturation.


19
Endosomes are central organelles in orchestrating cell interactions with the extracellular 20 environment, whether by regulating the composition of signalling molecules at the plasma 21 membrane or by facilitating uptake and digestion of certain nutrients or degrading toxic or 22 no longer needed material. Their wide range of functions is accomplished through a 23 sequential and highly regulated process known as endosome maturation (Huotari and 24 Helenius, 2011; Podinovskaia and Spang, 2018;Spang, 2016). Early endosomes accept 25 incoming cargo from the endocytic vesicles and undergo extensive sorting to package 26 selected cargo into recycling vesicles for the return to the cell surface or to the Golgi, whereas 27 membrane cargo destined for removal is internalised into intraluminal vesicles (ILVs) for its 28 subsequent degradation in endolysosomes. As these sorting endosomes mature into late 29 endosomes, now mainly containing cargo destined for degradation, they no longer accept 30 cargo from the cell surface and acquire properties necessary for their interaction with 31 lysosomes. Upon fusion with lysosomes, late endosomes form endolysosomes, whose highly 32 acidic and hydrolytic milieu facilitates degradation of the remaining cargo and regeneration 33 acidification is another essential change that must take place for endosomes to mature, with 50 pH ~6.5, 5.5 and 4.5 characterising early endosomes, late endosomes and lysosomes, 51 respectively (Casey et al., 2010). These changes in GTPase recruitment, PIP composition and 52 acidification status, among others, must be tightly coordinated to ensure unidirectional and 53 aligned adjustments to endosome identity and purpose for the endocytic system to operate 54 properly. However, coordination of such processes during endosome maturation is poorly 55 understood. 56 A major setback in understanding the kinetics of endosome maturation is lack of 57 experimental systems, which would allow us to monitor endosomes at individual endosome 58 level over prolonged periods of time. The small size of the endosomes and their rapid 59 movement within the cell makes it impractical to track maturing endosomes as they rapidly, 60 within seconds or minutes, move out of field of focus or get lost among other vesicles in the 61 dense perinuclear space. Phagosomes have provided a unique way of studying certain aspects 62 of endosome maturation, allowing for uniform size and synchronisation (Naufer et al., 2018; 63   Nigericin was added to HeLa cells at 10 µM for 20 min and washed away, and cells were imaged by �me-lapse microscopy. Recovery �mes are specified rela�ve to removal of nigericin. 7 these compartments. Electron microscopy images revealed that the enlarged compartments 147 originate at the trans face of the Golgi (Fig 2A), in line with previous reports of ionophore 148 treatment leading to the swelling of the trans-Golgi leaflet (Ledger et al., 1980;Morre et al., 149 1983). We ruled out contribution from the autophagy pathway by staining mApple-Rab5 150 expressing cells with LC3b antibody and showing no detectable autophagy induction or LC3b 151 presence at the enlarged endosomes at 60 min post nigericin treatment (Fig.2, figure  152 supplement 1A). To determine whether the swollen TGN membranes would enter the 153 endosomal pathway, we performed immuno-electron microscopy with HeLa cells stably 154 expressing trans-Golgi marker GalT-GFP after acute nigericin treatment. The micrographs 155 demonstrate the presence of GalT in the enlarged trans-Golgi network (TGN) compartments 156 and in ILVs of multivesicular bodies at later time points (Fig 2B). Therefore, the membranes 157 that acquire Rab5 and convert to Rab7 positive endosomes are probably derived from the 158 TGN. This swelling of the TGN is likely a transient response to the acute stress because after 159 48 h the Golgi had recovered from the treatment (Fig 2C). Consistent with this notion, we 160 occasionally observed swollen Golgi leaflets also in untreated cells signifying a process that 161 occurs naturally in the cell, which we are uniquely amplifying with acute perturbation (Fig 2A). 162 Indeed, the cells continued to grow and divide (Fig.2, figure supplement 1B), and after an 163 initial slow start, the nigericin-treated cells recover their doubling rate within 24 h (Fig.2,  164 figure supplement 1C). In line with previous reports (Merion and Sly, 1983), our findings 165 indicate that short nigericin treatment induces reversible changes and has minimal impact on 166 cell health. 167 To corroborate our results, we monitored HeLa cells stably expressing GalT-GFP by 168 fluorescence microscopy following acute nigericin treatment. Golgi vesiculation was observed 169 within 15 min of nigericin washout (Fig 3A, video Fig 3A supplement 1). We observed similar 170 results when we used monensin as ionophore (Fig.2, figure supplement 1D). A large fraction 171 of these vesicles would adopt early endosomal identity because individual GalT-positive 172 structures acquired Rab5 over time (Fig 3A), as also observed by immuno-electron microscopy 173 ( Fig 3B). Moreover, similar to the transiently transfected GalT-GFP, endogenous GalT 174 persisted in the endosomes (Fig 3C), consistent with the observations of its subsequent 175 internalisation into ILVs (Fig 2B and 10D), and Golgi morphology was fully recovered within 176 48h (Fig.2, figure supplement 1E). The contribution of cargo from the endocytic pathway to 177 the enlarged compartments was evidenced by the addition of Dextran-AF488 for 1 h to the Nigericin was added to HeLa cells for 20 min and washed away, and cells were processed for electron microscopy (A-C), imaged by �me-lapse microscopy (D) or harvested for coun�ng (E) at specified �mes a�er the wash.
(A) Cells stained with osmium tetroxide and potassium hexacyanoferrate reveal large spherical compartments (cyan arrows) origina�ng at the trans-face of the Golgi (magenta arrows) in nigericin-treated cells and, occasionally, in untreated cells. Nigericin was added to HeLa cells for 20 min and washed away, and cells were imaged by �me-lapse microscopy (A), processed for electron microscopy (B) or for immunofluorescence (C) at specified �mes a�er the wash. (B) Immuno-EM image of cells at 2 h post recovery, with 12 nm Gold labelled GFP (green arrows) and 5 nm Gold labelled mApple (red arrows) present at the enlarged compartments. Scale bar = 500 nm.
(C) Images of cells stained with an�-GalT to reveal endogenous GalT presence at the enlarged compartments.   (Fig 4D; Fig. 4, 204 figure supplement 1B). Occasionally, Rab5 produced multiple peaks, with Rab7 plateauing off 205 after the latest Rab5 peak (Fig. 4, figure supplement 1C and D). Such Rab5 behaviour may 206 indicate the reversible nature of endosome maturation and existence of checkpoints to 207 ensure alignment of parallel processes. Our results closely agree with Rab conversion kinetics 208 in other systems and further refine Rab conversion kinetics in human cells. Empowered by 209 this strict sequential kinetics of Rab5 and Rab7 in maturing endosomes, we next explored the   nigericin treatment, and �me-lapse microscopy during the recovery phase, with subsequent quan�fica�on of mean fluorescence intensity (MFI) of the chosen markers at the rim of the enlarged endosomes, and the resul�ng kine�c plots of background-subtracted MFI normalised for maximum and minimum values over the en�re �me course of the endosome. Since endosome matura�on is asynchronous, rela�ve �me is calculated by using Rab5 peak as a reference for Rab conversion and set to t=0. The plot shown in the scheme represents the kine�c of the images in Figure 1C (marker 1 as Rab5 and marker 2 as Rab7).
(B,C,D) HeLa cells, stably expressing mApple-Rab5 and GFP-Rab7 were treated for 20 min with nigericin, washed and imaged over a 3 h period.
(B) Time-lapse images of a representa�ve endosome to show transient Rab5 recruitment and its subsequent displacement by Rab7.
(C) Corresponding graph of MFI of Rab5 and Rab7 at the rim of the endosome in (B) during and around the �me of Rab conversion. Numerical data for all analyzed endosomes is available in Figure 4D -Source Data 1. HeLa cells, stably expressing mApple-Rab5 and transiently transfected with the PI(3)P marker, GFP-FYVE, were treated for 20 min with nigericin, washed and imaged over 3 h, as described in Figure 4A. (C) Averaged Rab5 and PI(3)P kine�cs of 19 endosomes. Error bars represent standard devia�on. Representa-�ve graph of three independent experiments. Numerical data for all analyzed endosomes is available in Figure   5C -Source Data 1.
kinetics of other mediators of endosome maturation relative to either Rab5 or Rab7 211 recruitment, using the maximum peak of Rab5 or the 50% of the maximal fluorescence 212 intensity of Rab7 as reference point for Rab conversion. 213 214 PI(3)P levels peak concomitantly with Rab5 levels HeLa cells, stably expressing mApple-Rab5 and transiently transfected with Snx1-GFP, were treated for 20 min with nigericin, washed and imaged over 3 h, as described in Figure 4A. HeLa cells, stably expressing mApple-Rab7 and transiently transfected with Snx1-GFP, were treated for 20 min with nigericin, washed and imaged over 3 h, as described in Figure 4A. (E,F,G) ) Averaged Rab7 and Snx1 kine�cs binned into the three pa�erns of Snx1 recruitment as observed in (B,C,D), represen�ng 19, 21 and 20 endosomes for the single peak, double peak and con�nuing presence of Snx1, respec�vely.
Error bars represent standard devia�on. Three independent experiments were performed, and data pooled. Numerical data for all analyzed endosomes is available in Figure 7 -Source Data 1. endosomes (Fig 6D and E). Moreover, Snx1 levels either declined during Rab conversion or 243 persisted for a while. Our data indicate that Snx1 recruitment on early endosomes occurs 244 simultaneously with Rab5, but that Snx1 microdomains could either co-exist with or exist 245 independently of Rab5, suggesting that Rab5 may promote Snx1 recruitment but is not 246 essential for its maintenance or dynamics at endosomes. 247 To corroborate the apparent lack of strict coordination between Rab5 removal and 248 Snx1 persistence at the endosomes, we co-expressed Snx1-GFP and mApple-Rab7 (Fig 7A,  249 video Fig 7A supplement 1). Consistent with Snx1 presence during the Rab5 phase, Snx1 250 recruitment peaked during early stages of Rab7 recruitment, when the endosome is expected 251 to be Rab5-positive (Fig 7B-D). In about one third of all Rab7-positive endosomes analysed, 252 Snx1 recruitment was transient and was no longer present after Rab7 peaked or levelled off 253 to indicate completed Rab conversion (Fig 7B and E). In another third of analysed endosomes, 254 Snx1 initially displayed the same kinetics, but was recruited back again to the late Rab7-255 positive endosome (Fig 7C and F), suggesting that Rab5 may be dispensable for Snx1 256 recruitment to late endosomes. In the remaining subset of endosomes, Snx1 peaked and 257 persisted throughout endosome maturation (Fig 7D and G). Analysis of Rab7 and Snx1 258 domains at the endosome again revealed weak correlation of Rab7 and Snx1 domains (  Rab11

HeLa cells, stably expressing mApple-Rab5 (A-D) or mApple-Rab7 (E-H) and transiently transfected with
GFP-Rab11, were treated for 20 min with nigericin, washed and imaged over 3 h, as described in Figure 4A.  Figure   8D and S8 -Source Data 1 and Figure 8H and S8 -Source data 1. GalT-pHlemon sensor displayed a characteristic Golgi-ribbon appearance in both YFP and CFP 294 channels as well as punctate appearance of CFP signal alone, indicative of highly acidified 295 lysosomes or endolysosomes ( Fig 9A). We could reliably detect YFP/CFP ratios over the pH 296 4.0-7.5 range (Fig 9B, Fig. 9, figure supplement 1), allowing for accurate pH measurements of 297 the entire endolysosomal pathway. Our sensor designated pH 6.2 for the Golgi-ribbon 298 structures and pH 4.0-5.7 for the lysosomes and endolysosomes, as detected by the CFP 299 puncta, in untreated cells ( Fig 9C). Most importantly, our sensor located to the nigericin-300 induced enlarged endosomes and indicated a pH range between 5.5 and 6.6 at 50 min 301 washout, reflective of the different stages of maturation ( Fig 9D, video Fig 9D supplement 1). 302 Therefore, GalT-pHlemon is a useful tool to read-out pH in the endosomal system. 303 304 Endosomal acidification is most pronounced during Rab conversion  HeLa cells, stably expressing mApple-Rab5 (A-C) or mApple-Rab7 (D-F) and transiently transfected with GalT-pHlemon, were treated for 20 min with nigericin, washed and imaged over 3 h, as described in Figure 4A. (C,F) Averaged Rab5 (C) or Rab7 (F) and GalT-pH kine�cs of 15 and 18 endosomes, respec�vely. Error bars represent standard devia�on. Pooled data from two independent experiments. Numerical data for all analysed endosomes is available in Figure   10C -Source Data 1 and Figure 10F -Source data 1.
Equipped with a sensor locating to endosomes and responding to endosomal pH changes, we 306 investigated the kinetics of endosomal acidification relative to Rab5 and Rab7 recruitment, 307 using cells stably expressing mApple-Rab5 or mApple-Rab7 and transiently transfected with 308 GalT-pHlemon. The nigericin-induced enlarged endosomes showed a dramatic decrease in 309 YFP signal, which always coincided with Rab5-positive stages of endosome maturation ( Fig 1C). These perturbations 330 could be efficiently rescued by expression of wild-type Ccz1 in Ccz1 knock-out cell lines (Fig.  331 11A, videos Fig 11A supplement 1-3). To ensure that in rescue experiments we selected for 332 analysis only the cells expressing Ccz1, and not untransfected cells, we appended a far-red 333 fluorophore mNeptune2 via the T2A peptide linker to Ccz1, resulting in expression of the two 334 separate proteins in the transfected cells (Fig. 11, figure supplement 2A). The mNeptune2 was 335 tagged with NLS, targeting it to the nucleus, to minimise interference with the mApple signal 336 at the endosomes (Fig. 11, figure supplement 2B). Hence, we have generated Ccz1 knock-out

Rab5
GalT -pH Nigericin was added to cells for 20 min and washed away, and cells were imaged by �me-lapse microscopy, as described in Figure 4A.  cell lines that showed impaired Rab conversion and could be efficiently rescued with the Ccz1 338 rescue construct. 339 To test our hypothesis that impaired Rab conversion compromises endosomal 340 acidification, we expressed mApple-Rab5 and GalT-pHlemon in Ccz1-KO and control cells, and 341 monitored the YFP/CFP ratio kinetics of the pH sensor on Rab5-positive endosomes (Fig 11B,  342   Fig. 11, figure supplement 1B). While in the control and Ccz1 rescue cells, acidification 343 occurred with similar kinetics as observed above (Fig 10B and C), in Ccz1 KO cells the 344 acidification was strongly delayed (Fig 11B-D). Nevertheless, acidification occurred eventually 345 after a long delay in Ccz1 KO cells. In line with this conclusion, Ccz1 KO cells have grossly 346 enlarged CFP-filled puncta and compartments, reflective of their enlarged lysosomes and 347 acidified hybrid compartments as also observed with Lysotracker staining of Rab5-positive 348 compartments ( Fig 12A, Fig 12, figure supplement 1). The CFP-positive compartments in 349 untreated cells showed no differences in pH between Ccz1-replete and Ccz1-deficient cells, 350 ranging from pH 5.7 to pH 4.0, indicative of endolysosomes and lysosomes, respectively (Fig  351   12B). Furthermore, following disruption of pH by nigericin treatment and washout, wild-type 352 cells restored their lysosomal pH, while Ccz1 KO cells displayed a wide range of pH in 353 pHlemon-filled compartments, ranging from pH 6.4 to pH 4.0 (Fig 12C and D). Taken together, 354 our data suggest that Rab conversion is driving efficient endosomal acidification. concomitantly with Rab5 to maturing endosomes. However, even though in about one third 394 of the endosomes, there seemed to be temporal coordination between SNX1 and Rab5 395 removal from the endosome, sorting persisted in the remaining Rab7-positive endosomes. In 396 addition, we did not observe any spatial coordination on the endosomal membrane as the 397 SNX1 and Rab5 or Rab7 domains appeared to move independently. Moreover, Rab11 398 contacted maturing endosomes irrespective of whether they were Rab5 or Rab7 positive. 399 Therefore, our data suggest that the onset of recycling is coordinated with the arrival of Rab5, 400 at least for Snx1, but the process itself is independent of Rab conversion, as previously suggested (Rojas et al., 2008). Consistently, we have shown previously that when Rab 402 conversion is blocked, Rab11 localization and Rab11-dependent recycling are not affected in 403 C. elegans oocytes (Poteryaev et al., 2010;Poteryaev et al., 2007). 404 Although acidification of endosomes is required for endosome maturation 405 (Podinovskaia and Spang, 2018), how this process is regulated remains poorly understood. 406 Since Rab conversion is a major driver of endosome maturation, we asked whether Rab 407 conversion regulates endosomal acidification. Unfortunately, all pH sensor probes we tried 408 turned out not to be useful because they were mostly stuck in the ER. While in neurons, in 409 which the probes have mainly been applied, this might be less of an issue, in our system this 410 has prevented any meaningful analysis. We, hence, developed a new probe based on the 411 ratiometric pHlemon and GalT, which localises to Golgi but enters also the endolysosomal 412 pathway. With this new probe, we showed that Rab conversion is required for efficient 413 acidification. Over extended times, acidification of endosomes still occurred in absence of Rab 414 conversion and we speculate that this acidification can help drive fusion with lysosomes.

Single endosome analysis and quantification 562
Using the time-lapse images, endosomes were selected for analysis with the following 563 criteria. For mApple-Rab5 expressing cells, endosomes initially devoid of Rab5 and later 564 acquiring and subsequently losing Rab5 were identified. For mApple-Rab7 expressing cells, 565 endosomes initially devoid of Rab7 and subsequently acquiring Rab7 and stabilising its 566 expression were identified. This ensured that the entire Rab conversion event was captured 567 in the kinetic. 568 To quantify the recruitment of markers to the endosome, an oval selection tool in Fiji 569 was used to draw a circular region-of-interest (ROI) closely following the rim of the endosome 570 in the channel for the most visible marker or by predicting the location of the rim if cases 571 where the enlarged endosome was negative for both markers and appeared as a dark circle 572 ( Fig 4A). For less circular endosomes and endolysosomes, the ROI was adjusted using the 573 elliptical or a free-hand selection tool. ROIs were adjusted for every time point where the rim 574 of the endosome could be unambiguously identified. Mean fluorescence intensity (MFI) of a 575 two-pixel wide rim at the ROI was recorded in all channels. A larger two-pixel wide rim three 576 pixels away from the endosome was generated with a macro based on the original ROI, and 577 MFI was calculated as a measure of background for each time point (Fig 4A). We found that 578 adjusting the MFI at the rim of endosome for this background MFI minimised noise and 579 produced data reflective of visual assessment of marker presence at the endosome. For 580 intraluminal pHlemon measurements, the circular ROI at the rim of the endosome was 581 reduced by 1 pixel and the total MFI of the reduced ROI was calculated in both YFP and CFP 582 channels. A ROI in a field where no cells were present was measured to obtain background 583 values. This approach was found to produce pH measurements as accurate as the modified 584 rim measurements, in which select pixels were removed to exclude interference from the 585 highly acidified vesicles interacting with the enlarged endosome. For the subdomain 586 measurements, two-pixel thick segmented lines with spline fit were drawn around the full 587 perimeter of the endosome starting at the top, and histogram measurements were obtained 588 of fluorescence intensity along the length of the line. 589 Endosomal recruitment marker measurements were background-subtracted and 590 adjusted for the minimum and maximum values of the entire measured kinetic, to represent 591 a range from 0 to 1. The pHlemon measurements were kept as background-adjusted YFP/CFP 592 ratios. For averaging, kinetics were aligned for Rab5 peak or for Rab7 at 50% of its final maximum value, representing the point of Rab conversion (Fig 4C-D). At least 10-20 594 endosomes from at least 3-10 cells were used in analysis and each experiment was repeated 595 HeLa CCL2 cells were grown in 10 cm dishes, treated for 20 min with nigericin and left to 614 recover. At specified times, cells were fixed in DMEM containing 2.5% glutaraldehyde and 3% 615 formaldehyde for 2 h at room temperature. Cells were washed with PBS and cell membranes 616 stained with 2% osmium tetroxide and 1% potassium hexacyanoferrate in H2O for 1 h at 4˚C. 617 Following a wash with water, cells were further stained for proteins and nucleic acids with 2% 618 uranyl acetate in H2O overnight at 4˚C. Samples were subsequently dehydrated in 619 acetone/H2O in stepwise increases in acetone concentration of 20%, 50%, 70%, 90% and 3x 620 100%. The samples were infiltrated with 50% epon embedding medium in acetone for 1 h at 621 room temperature, and subsequently with 100% epon resin overnight. Next day, fresh epon 622 resin was added and samples were polymerised for 24 h at 60˚C. Sections of 60-70 nm were 623 collected on carbon-coated Formvar-Ni-grids and were viewed with a Phillips CM100 electron 624 microscope To prepare cells for immunolabeling, HeLa cells stably expressing GalT-GFP were 626 prepared as previously described (Beuret et al., 2017). Sections were stained sequentially 627 with rabbit anti-GFP (1:100; Abcam 6556) and goat anti-rabbit coupled to 10 nm gold particles 628 (BB International). For dual labelling, HeLa cells stably expressing mApple-Rab5 were 629 transiently transfected with GalT-EGFP, prepared for immunolabelling as above, and stained 630 sequentially for GalT-GFP and mApple-Rab5. The sections were blocked with PBST 631 (PBS+0,05% Tween20) supplemented with 2% BSA for 20 min, incubated overnight at 4°C with    (B) Images to show cells dividing shortly a�er nigericin treatment, with recovery �mes specified. Scale bar = 10 µm.
(C) Cells were plated in 12-well plates at 104 cells per well 24 h prior to nigericin treatment, trypsinised and counted at specified recovery �mes. Triplicate wells were counted per �me point per condi�on. Actual cells numbers and doubling �mes are presented. Representa�ve graph of three independent experiments. Source data is available in Figure S4C -Source Data 1.
(D Images to reveal extensive forma�on of GalT-posi�ve enlarged compartments upon either treatment following 2 h recovery. (E) Images to show Golgi vesicula�on and return to ribbon morphology within 48 h of recovery from nigericin treatment.
(F) Images of cells incubated with 0.5mg/mL Dextran-AF488 for 65min a�er nigericin washout to reveal endocytosed dextran presence in the enlarged Rab5-posi�ve endosomes.  HeLa cells, stably expressing mApple-Rab5 and transiently transfected with the PI(3)P marker, GFP-FYVE, were treated for 20 min with nigericin, washed and imaged over 3 h, as described in Figure 4A.  Nigericin was added to HeLa cells for 20 min and washed away, and cells were imaged by �me-lapse microscopy, as described in Figure 4A. (B) Averaged Rab5 and Snx1 kine�cs of 14 endosomes. Error bars represent standard devia�on. Representa�ve graph of three independent experiments. Numerical data for all quan�fied endosomes is available in Figure 6C -Source Data 1.
(C) Images and corresponding line profiles of normalised fluorescence intensity of Rab5 and Snx1 along the rim of the maturing endosome at consecu�ve �me points. Rab5 was adjusted for a single maximum and minimum values during the recorded kine�c to highlight its overall signal increase. Snx1 was adjusted for max and min values for each �me point to highlight the dynamic nature of Snx1 subdomains. Scale bar = 2 µm. Numerical data for all quan�fied endosomes is available in Figure S7C -Source Data 1.
(D) Images of Rab7 and Snx1 at an enlarged endosome and a corresponding line profile of normalised fluorescence intensity along the rim to show co-existence as well as independence of subdomains of the two markers. Scale bar = 1 µm.
(E) Correla�on plot of normalised fluorescence intensity of Rab7 and Snx1 as measured in (E) for 3 endosomes for a total of 96 �me points, and a corresponding regression line. Pearson's correla�on r=0.43. Numerical data for all quan�fied endosomes is available in Figure S7E -Source Data 1. HeLa cells, stably expressing mApple-Rab5 (A,C) or mApple-Rab7 (B) and transiently transfected with GFP-Rab11, were treated for 20 min with nigericin, washed and imaged over 3 h, as described in Figure 4A. (C) Correla�on plot of normalised fluorescence intensity of Rab5 and Rab11 as measured in Figure 7D for 14 endosomes for a total of 193 �me points, and a corresponding regression line. Pearson's correla�on r=0.28. Numerical data for all analysed endosomes is available in Figure 8D and S8 -Source Data 1 and Figure 8H and S8 -Source Data 1.  Time-lapse images of representa�ve endosomes, corresponding to those quan�fied in Figure 11B, to show acidifica-�on in endosomes recrui�ng Rab5 in the three cell types. (C) Images of untreated Ccz1 KO cells stained with Lysotracker Red (LTR) to reveal large, acidified compartments.