De novo formation of an early endosome during Rab5 to Rab7 transition

Rab5 and Rab7a are the main determinants of early and late endosomes and are important regulators of endosomal progression. The transport from early endosomes to late endosome seems to be regulated through an endosomal maturation switch where Rab5 is exchanged with Rab7a on the same endosome. Here we provide new insight into the mechanism of endosomal maturation where we have discovered a stepwise Rab5 detachment, sequentially regulated by Rab7a. The initial detachment of Rab5 is Rab7a independent and demonstrate a diffusion-like exchange between cytosol and endosomal membrane, and the second phase is slower where Rab5 converges into a specific domain that specifically detaches as a Rab5 indigenous endosome. Consequently, we show that early endosomal maturation regulated through the Rab5 to Rab7a switch induce the formation of a new fully functional early endosome. Progression through a stepwise early endosomal maturation regulates the direction of the transport and concomitantly regulates the homeostasis of early endosomes.

The RabGTPase cycle has become the standard to define endosomal progression and maturation (Rink et al., 2005). However, performing live cell imaging, we were able to provide new data on endosomal maturation and progression through Rab5 to Rab7a exchange. Rab5 detachment from the maturing endosome occurs in a stepwise manner where the initial detachment is independent of Rab7a, whereas the completion of the Rab5 detachment depends on endosomal Rab7a recruitment (Chotard et al., 2010;Li et al., 2009)

Results
The Rab5 detachment occurs in two phases.
Endosomal progression, from early to late endosomes, is carefully regulated by the transition from a Rab5 to a Rab7a positive endosome (Poteryaev et al., 2010;Rink et al., 2005). To better understand this pivotal transition, we performed a live cell study of this particular conversion on endosomes in MDCK-Ii cells and Hela cells (Nordeng et al., 2002;Stang and Bakke, 1997).
In MDCK-Ii cells with enlarged endosomes, mCh-Rab5 remains on the endosomal membrane for a longer time (Landsverk et al., 2011). This prolongs the lifetime of an early endosome without changing the actual transition from Rab5-positive EEs to Rab7a-positive LEs (Landsverk et al., 2011). MDCK-Ii cells were transfected with mCh-Rab5 alone (control) or together with EGFP-Rab7a wild type (WT) or the constitutively active mutant EGFP-Rab7aQ67L ( Figure 1). During early endosomal progression we observed numerous homotypic fusion events (Skjeldal et al., 2012) where the endosomes grew larger, preparing for the transition from an early to late endosome. Before the transition from Rab5 positive to Rab7a positive endosomes we observed that the early endosomal homotypic fusion events paused.
Once the early endosomal homotypic fusion events ceased, we could initiate the single endosomal analysis of the mCh-Rab5 coat detachment and concomitant acquisition of EGFP-Rab7a ( Fig 1A, B, C).
In control cells (MDCK-Ii transfected with mCh-Rab5), we followed single mCh-Rab5 positive endosomes and plotted the detachment curve ( Fig 1A, Movie 1A). Similarly to previously observed Rab5 to Rab7a conversion (Rink et al., 2005) the coat gradually detached from the maturing endosome and became mCh-Rab5 negative ( Figure 1A). In order to better understand the mCh-Rab5 coat dynamics during maturation we measured the total time of detachment and calculated the halftime coat detachment (D1/2) (see materials and methods). In the control cells ( Figure 1A), we measured the D1/2 for mCh-Rab5 to be 7.0 1.2 minutes ( Figure 1D). To further analyze if a higher expression level of Rab7a could affect the endosomal Rab5 detachment, we double-transfected MDCK-Ii cells with mCh-Rab5 and EGFP-Rab7aWT ( Figure 1B, Movie 1B). As previously shown (Rink et al., 2005) we could observe a gradual exchange between Rab5 and Rab7a ( Figure 1). Analysis further showed that a higher expression of Rab7a induced a faster D1/2 for mCh-Rab5, 4.2 1.7 minutes ( Figure 1B). We could additionally measure a similar effect on the mCh-Rab5 D1/2 when we expressed the GTP bound form of Rab7a, EGFP-Rab7aQ67L, the mCh-Rab5 D1/2 was 3.9 1.9 minutes (Figure 1 C, Movie 1C), indicating that Rab7a recruited to the endosomal membrane might regulate Rab5 detachment during endosomal conversion. Comparing the D1/2 between the three experiments we could measure a significant difference in the D1/2 for mCh-Rab5 when the cells were transfected with either of the Rab7a constructs ( Figure 1D).
Further analysis of the characteristics of the detachment graph in the control experiment revealed a new pattern of endosomal coat dynamics. We could clearly measure two different phases for the mCh-Rab5 detachment; an initial fast detachment the first 2-3 minutes, (red box, Figure 1D) followed by a slower phase, here termed the completion phase, where the curve is less steep (green box Figure 1D). In another set of experiments where the cells were transfected with EGFP-Rab7a WT we could measure a lag period before Rab7a got recruited to the endosome (blue box figure 1D) coinciding with the Rab5 fast and initial detachment. In fact, 50% of Rab5 detached before any Rab7a was detected on the endosomes. However, when the cells were expressing the GTP bound form of Rab7a, EGFP-Rab7aQ67L, the lag period was not detected.
These experiments show us that the mCh-Rab5 detachment dynamics can be divided in two phases, an initial fast phase and a slower, completion phase. In addition, these experiments suggest that the recruitment of Rab7a to the maturing endosome occurs only after the initial detachment of Rab5, and that Rab7a may be important for the slower completion phase, as illustrated by the fact that expression of Rab7a increase the D1/2 Rab5 detachment rate.

Initial Rab5 detachment is Rab7a independent.
We find that the endosomal maturation starts with a fast detachment of Rab5 before any Rab7a is detected on the membrane (Figure 1). Our analysis showed that the onset of Rab7a attachment to the endosomes is occurring after Rab5 has started to detach (red and blue box fig 1D) at a time where ~50% of Rab5 has already left the endosome. To further address the significance of Rab7a recruitment for the dissociation of mCh-Rab5 from the endosome, we co-transfected mCh-Rab5 with the dominant negative mutant of Rab7a (EGFP-Rab7aT22N) or depleted Rab7a with specific siRNA (siRab7a) in HeLa and MDCK-Ii cells ( Figure 2). Rab7aT22N is locked in the GDP state and thus localized in the cytosol, acting as a dominant negative mutant of Rab7a (Bucci et al., 2000;Feng et al., 1995).
In MDCK-Ii cells co-transfected with mCh-Rab5 and EGFP-Rab7aT22N ( Fig (Figure 2 B) The frequency seemed to be quite stable for the endosomes in each cell and the amplitude between the maximum mCh-Rab5 intensity varied between 2-4 minutes (2B graph). Furthermore, depletion of Rab7a using RNAi in Hela cells showed a similar partial mCh-Rab5 detachment and the reattachment ( Figure 2C, D). We could observe from the graphs that the dissociation of mCh-Rab5 was fast and seemed to be similar to the fast-initial phase of the mCh-Rab5 detachment ( Figure 1D, red box). Careful analysis of the endosomal membrane we could observe mCh-Rab5 positive domains as an intermediate state between the maximum intensity during the intensity oscillations ( Figure 2B, white arrows).
This shows that the mCh-Rab5 positive endosomes fail to reach the second phase, the completion phase, when no Rab7 is recruited to the endosome in transition.
To search for a role of the Rab5 GDP/GTP cycle in this process we transfected Hela cells with the constitutively active mutant of Rab5 (EGFP-Rab5Q79L), which is unable to hydrolyze GTP and remains membrane-bound (Stenmark et al., 1994), together with DsRed-Rab7aT22N. Hela cells expressing EGFP-Rab5Q79L induced enlarged endosomal structures (Wegner et al., 2010) that had a uniform EGFP-Rab5Q79L positive membrane for more than one hour ( Figure   2E). Under these conditions, no cyclic variation was seen for the EGFP-Rab5Q79L positive membrane intensity ( Figure 2E). As the Rab7a dominant negative mutant did not induce detachment/reattachment of EGFP-Rab5Q79L, this indicate that the initial fast detachment phase of Rab5 is regulated by the rapid GDP/GTP cycle.
These above results demonstrate that Rab5 has the ability to transiently leave the maturing endosome independently of membrane associated Rab7a, but Rab7a has to be recruited to the endosome membrane for Rab5 to complete its detachment. Overall, these results show that mCh-Rab5 detachment occurs in two phases: the initial phase, which is Rab7a independent; and the completion phase, which is Rab7a dependent.

Rab5 domains converge and leaves the maturing endosome.
During the analysis of single maturation events on Ii-enlarged endosomes in MDCK-Ii cells We then asked if this mechanism of maturation was unique for Rab5. In order to better understand this, we co-transfected cells with a well-known tethering protein EEA1-GFP wt (Bergeland et al., 2008;Murray et al., 2016;Simonsen et al., 1998) together with mCh-Rab5 and analyzed the coat detachment of the respective membrane associated proteins. The detachment of the two membrane associated proteins occurred simultaneously and the convergence towards a local endosomal hotspot was observed with both EEA1-GFP wt and mCh-Rab5. This indicates that both EEA1 and Rab5 follows a similar endosomal detachment pattern ( Figure 3C, white box, Movie 3C). Further analysis of the Rab5-EEA1 positive vesicles showed that after their release from the newly matured endosome, they instantly engaged in early homotypic fusion ( Figure 3C). EEA1 and Rab5 are kept on the membrane as a fully functional fusion machinery to ensure that the newly released endosomes are instantly fully functional to engage in homotypic fusion. Early endosomes work as sorting stations for protein destined for degradation or recycling (Huotari and Helenius, 2011). In order to better understand the function of the newly formed mCh-Rab5 endosome we added EGF-Alexa647 to the cells. This experiment showed us that the newly formed mCh-Rab5 positive endosome could recruit EGF-Alexa647 containing vesicles showing that the newly formed early endosome is primed for another round of sorting of EGF/EGFR ( Figure 3D). vesicles seem to carry the signal for the maturing endosome to proceed from the initial phase to the convergence part of maturation since in the Rab7aT22N transfected cells the endosomes never reached the completion phase.

Rab5 converged domains consist of the membrane bound immobile fraction.
We have shown that Rab5 detachment occurs in two phases: an initial rapid and a slower convergence phase after Rab7a is recruited to the endosome. Rab GTPases alternates between a membrane bound and a cytosolic state and we can measure the immobile and the mobile fraction present on the endosomes (Sprague and McNally, 2005). We have previously described that endosomal associated proteins can specifically alter the ratio between the mobile and the long lived immobile fraction both as a function of signaling and between interphase and mitosis (Bergeland et al., 2008;Haugen et al., 2017). As this is believed to reflect a functional state, we wanted to investigate whether these two fractions of Rab5 on the endosomal membrane varied through to the two detachment phases of Rab5.
To analyze the binding kinetics of Rab5 before and during conversion, we performed single endosome FRAP experiments on endosomes prior to transition and on converged domains. In these experiments, we used the MDCK-Ii cells expressing enlarged endosomes transfected with EGFP-Rab5. Analysis of bleaching experiments on single endosomes positive for EGFP-Rab5 showed us that EGFP-Rab5 has a mobile fraction of 22% and an immobile fraction of 78% ( Figure 4), similar to previously observed values (Haugen et al., 2017). In order to compare the initial Rab5 detachment with the slower convergence maturation we had to specifically bleach the converging domains before the transfer to the newly formed Rab5 positive endosome.
Enlarged endosomes together with a fast bleaching module (Andor Dragonfly mosaic) made this technically possible ( Figure 4A). Bleaching experiments and analysis of the converging domains revealed an altered binding dynamic of EGFP-Rab5 in the converged domains compared to the binding dynamics prior to domain formation ( Figure 4A). We could measure a significant change in the immobile fraction of EGFP-Rab5 in the converging domains, which increased by a factor of 4 ( Figure 4B). This result shows a gradual increase of the immobile fraction during the convergence phase prior to a complete detachment. We could further show that this immobile fraction consisted of the GTP bound form of Rab5 by transfecting the cells with GFP-Rab5Q79L and performing the same FRAP experiment ( Figure 4B). Comparing the membrane fractions of the converging domain and the GFP-Rab5Q79L we could show that they are exactly the same, strongly indicating that the converging domains and the Rab5 leaving the endosome to form a new endosome consists of the immobile fraction. These experiments proved that the immobile fraction is 80% in the converging domains and that this is the GTP bound form of Rab5 (Figure 4).
These results show that Rab5 detaches in two phases; the initial phase may be regulated by the GDP-GTP cycle and pertains to the mobile fraction of Rab5, while the slower phase is a concentration of Rab5 immobile fraction to the converging domains prior to a transfer from the maturing endosome to prime the formation a new early endosome.

Discussion
Endosomal trafficking is carefully regulated through organelle specific Rab GTPases (Bhuin and Roy, 2014;Zerial and McBride, 2001). These GTPases identify and regulate the progression of endocytosed macromolecules and plasma membrane receptors (Langemeyer et al., 2018). Progression and direction from the early to the late endosome have been shown to be regulated by a Rab5 to Rab7a switch (Cabrera et al., 2014;Huotari and Helenius, 2011;Nordmann et al., 2010;Poteryaev et al., 2010;Rink et al., 2005). This switch/conversion have previously been described entirely as an exchange between the cytosolic pool and the membrane bound fraction of Rab5 and Rab7a, specifically regulated through the Rab GTP/GDP cycle (Pfeffer, 2001;Pfeffer, 2017;Stenmark and Olkkonen, 2001).
In this paper we have shown that enlarged early endosomes in MDCK-Ii cells (Landsverk et al., 2011;Skjeldal et al., 2012;Stang and Bakke, 1997) and early endosomes in Hela cells, mature into late endosomes by a stepwise Rab5 detachment, controlled but not initiated by Rab7a. We show that the detachment of Rab5 follows a two-step mechanism where the initial detachment is fast and diffusion-like, as expected for the GTP/GDP cycle, and the second part is slower and regulated by converging Rab5 domains (Figure 1). When Rab5 were cotransfected with Rab7a, either wt or Q67L mutant, we could measure a faster D1/2 of Rab5, indicating a Rab7a regulatory role of Rab5 detachment, specifically in the converging phase of Rab5 detachment. Furthermore, Rab7a was not recruited to the maturing endosome before the initial phase was completed, as shown in the Rab7a lag phase ( Figure 1D, blue box). This indicates that the Rab5 detachment has a Rab7a independent phase and a Rab7a dependent phase, which was confirmed after co-transfection with Rab7aT22N ( be an alternate mechanism additional to the Rab-GDI mediated exchange (Rink et al., 2005).
Here we provide evidence for a direct Rab5 transfer from the maturing endosome to form a novel Rab5 positive endosome.
We could furthermore show that the Rab5 effector protein EEA1 did also follow Rab5 during convergence and transferred together with Rab5 onto a newly formed endosome. The released endosome engaged in early endosomal fusion immediately post formation and could recruit endocytosed EGF, showing that it is a fully functional early endosome ready for another round of maturation.
To better understand the binding dynamics of Rab5 during the convergence phase we performed FRAP experiments on Rab5 positive endosomes before convergence and correspondingly bleached the converging domains. We could show that a Rab5 positive endosome, before the conversion, had 20% immobile fraction and 80% mobile fraction as previously published (Sann et al., 2012). However, in converged domains we measured a total change in the fractions, where we found 80% to be immobile and 20% to be mobile. In the control experiment where we bleached the dominant active mutant of Rab5 (EGFP-Rab5Q79L) we found the same distribution between immobile fraction and mobile fraction as in the converged Rab5 positive domain strongly suggesting that the fraction of Rab5 that detaches from the maturing endosome and gives rise to a new endosome is transferred as a membrane bound GTP form.
Both EEA1 and Rab5 are fusogenic membrane associated proteins that will have to go into a quiescent state during the early endosome to late endosome transition. This may be explained with our FRAP experiment where we measure a redistribution of the immobile and mobile fractions of Rab5. The first initial detachment proved to be the mobile fraction that detaches through the regular GDI mediated GTP/GDP cycle. Inactivating Rab5 fusogenic properties through a redistribution of the mobile fraction to immobile fraction may also change the activity of EEA1 inhibiting the entropic collapse which is necessary for tethering prior to fusion (Murray et al., 2016). We hypothesize that the Rab5 fractions on the newly formed endosome will be redistributed back to mobile fraction 80% and immobile fraction 20% in order to reactivate the fusion machinery. A redistribution of the immobile and mobile fraction of Hrs and Eps15 has previously been published to be important for EGFR degradation (Haugen et al., 2017). Similar redistribution of Rab5 was shown in this study which may indicate that a local redistribution of the mobile and immobile fraction on endosomes is a general mechanism to regulate endosomal maturation and progression.
In this paper we describe a new mechanism on the dynamics of Rab5 and Rab7a during endosomal maturation. The Rab5 to Rab7a switch is definitely not an exchange where Rab5 is directly replaced by Rab7a, the Rab5/Rab7a exchange is regulated in distinct steps. The initial diffusive phase of Rab5 detachment may act as a switch in endosomal transition activating the SAND-1/Mon-1/HOPS cascade (Poteryaev et al., 2010). This is to actively recruit the Rab7a positive endosome necessary for completing convergence phase of Rab5 detachment.
However, the switch does not seem to control the entire transition but act as a signal to activate the second phase of transition. We find that Rab7a recruitment to the endosome in transition most likely occur through vesicular interactions providing Rab7a to the maturing endosome. We were not able to observe whether the incoming Rab7a positive endosomes fused with the endosome in transition or was transferred f e e ic e he he b i a d (Duclos et al., 2003).
Both the shuttling model and the maturation model have contradictory characteristics, especially regarding endosomal homeostasis (Helenius et al., 1983). The shuttle model predicts a set of pre-existing endosomal compartments, shuttling carrier vesicles between the early and late compartments (Griffiths and Gruenberg, 1991). T he e-existing endosomes in to context we have to compare the condition before and after mitosis. During cell division the mitotic machinery ensures that the number of endosomal vesicles in the two daughter cells have approximately half the copy number each (Bergeland et al., 2001) and among the sustained vesicles after mitosis we would find the pre-existing endosomes. In this study we found that, after mitosis the number of early endosomal vesicles increase the first three hours of the G1 phase and subsequently we observed an increase in the number late endocytic compartments, indicating that the early endosome pool mature to late endosomes (Bergeland et al., 2001). This indicate that all the early endosomes in the daughter cells are pre-existing (Kamentseva et al., 2020) and will mature to produce new late endosomes. Consequently, Rab5 convergence induce formation of early endosomes to maintain the homeostasis between early and late endosomes.
We here show that early endosomes through Rab5 to Rab7a switch, continuously give rise to new functional endosomes, a crucial step to maintain the homeostasis of early endosomes.
This maturation regulated by transition from early to late endosomes provides the pool of early endosomes that exert the role as a sorting organelle. Furthermore, the maturation through conversion gives a direction for the endosomal progression. With these novel findings we here present a crucial step in understanding how endosomal trafficking is regulated to maintain endosomal homeostasis and how membrane associated proteins could be recycled during endosomal maturation (See graphical model figure 5).

Cell culture
Madine-Darby canine kidney strain II (MDCK) and HeLa cells were grown in an incubator at 37 C a d 5 % CO2 in complete DMEM medium (Bio Whittaker), supplemented with 10% fetal calf serum (FCS, Integro), 2 mM L-Glutamine and 25 U/ml penicillin (all from Bio Whittaker).
MDCK cells, stably transfected with invariant chain (Ii) in the pMep4 plasmid, where grown in the same conditions, with additionally 25 µg/ml Hygromycin (Bio Whittaker).

RNA interference
RNAi interference was used to knockdown Rab7a in Hela cells. Cells were plated one day before transfection in complete DMEM. Transfection was done with OligofectamineTM

Western blotting and antibodies
Western blotting was used to determine the efficiency of the Rab7a knockdown. The silenced and control cells were lysed in a buffer containing 25 mmol Hepes, 125 mmol potassium acetate, 2,5 mmol magnesium acetate, and 5 mmol EGTA. Right before lysis, the buffer was complemented with 1 mM DTT, 0,5% NP-40 and a protease inhibitor cocktail (Roche). 20 µg of protein were loaded on precast Tris-Hepes gels (NuSep 12%) and transferred to Immobilon® -P PVDF membranes (Millipore). The PVDF membranes were blotted with specific antibodies di ed i a 2% i i , e igh a 4 C. The ba d e e i a i ed af e i c ba i with ECL chemiluminescence kit (Amersham, GE Healthcare). An anti-tubulin antibody was used as a loading control (monoclonal mouse anti--Tubuline, Zymed, Thermofisher). The primary Rab7a antibody was purchased from Cell Signaling Technology (#2094S), and used 1:500. HRP-linked donkey anti-rabbit (NA934 GE Healthcare LifeSciences) were used as secondary antibody. Relative protein levels were measure with ImageJ.

Live Confocal Microscopy
Live cell Imaging and FRAP experiments. Obtained data from FRAP was normalized and corrected for bleaching (Pelkmans et al., 2001) and fitted by nonlinear regression to a function that assumes a single diffusion coefficient (Yguerabide et al., 1982); The values for F(0), F( ) and T1/2 were calculated using GraphPad Prism 8 (http://www.graphpad.com/scientific-software/prism/) and the immobile fractions (IF) were calculated as described in Lippincott-Schwartz et al (Lippincott-Schwartz et al., 2001).