|Home | About | Journals | Submit | Contact Us | Français|
A key regulator of receptor-mediated endocytosis, Rab5, plays a pivotal role in cargo receptor internalization, endosomal maturation, and in transduction and degradation of internalized signaling molecules and recycling cargo receptor. Stressful conditions within cells lead to increased Rab5 activation, and increasing evidence correlates Rab5 activity abnormalities with certain diseases. Current antibody-based imaging methods cannot distinguish active Rab5 from total Rab5 population and provide dynamic information on magnitude and duration of Rab5 activation in cellular events and pathogenesis. We report here novel molecular imaging probes that specifically target GTP-bound Rab5 associated with the early endosome membrane in live cells and fixed mouse brain tissues. Our Rab5 activity fluorescent biosensor (RAFB) contains the Rab5 binding domain of Rab5 effector, Rabaptin 5, a fluorophore (a quantum dot or fluorescent dye) and a cell penetrating peptide for live-cell delivery. The quantum dot conjugated RAFB was able to image the elevated Rab5 activity in both the cortex and hippocampi tissues of a Ts65Dn mouse. A pre-quenched RAFB based on fluorescence resonance energy transfer (FRET) can image cytosolic active Rab5 in single live cells. This novel method should enable imaging of the biological process in which Rab5 activity is regulated in various cellular systems.
Eukaryotic cells contain a highly dynamic vesicular trafficking network in which membrane and soluble contents of vesicles are frequently exchanged among intracellular organelles and with the extracellular environment (1). Rab GTPases are key regulatory proteins marking functionally distinct endosomes and orchestrating dynamic trafficking and fusion events at specific cellular locations (2).
Rab GTPases belong to the Ras-like small (21-25 kDa) GTPase superfamily. There are over sixty Rabs identified in mammals including Rab5, a key regulator of receptor-mediated endocytosis, and many are ubiquitously expressed (3). Fusion of Rab5-bound endosomes requires Rab5 effectors including rabaptin-5, the rate-limiting effector for endosome fusion. Physiologically rabaptin-5 binds two copies of its only known binding partner, active Rab5 (4).
A majority of intracellular Rab5 is inactive. Diseased cells show increased Rab5 activation, presumably because endocytosis is mis-regulated by demands for clearance of aggregated peptides. Increasing evidence correlates Rab5-mediated endosome dysfunctions with many diseases including amyotrophic lateral sclerosis (ALS) (5), Alzheimer’s disease (AD) (6, 7), Down syndrome (DS) (8), Parkinson’s disease (PD) (9), and Huntington’s diseases (HD) (10, 11). Also, rabaptin-5 is down regulated in hypoxic tumors, presumably decreasing Rab5-mediated endocytosis of receptor tyrosine kinases, which are critical in driving oncogenesis (12).
Although Rab5 regulated receptor-mediated signal transduction has been investigated for many years, little progress has been made toward understanding the detailed mechanisms surrounding how various extracellular signals are processed, transported and terminated through receptor-mediated endocytosis and inner membrane fusion/fission. The conventional endogenous Rab5 imaging protocol uses antibodies against Rab5, and reveals little dynamic information on the magnitude and duration of Rab5 activation in terms of endocytic and autophagic involvement in pathogenesis of diseases described above. Furthermore, the antibody-based method only measures the total population of Rab5, not active Rab5. Recently a genetically encoded fluorescent protein based sensor has been reported for live-cell imaging of Rab5 activity (13), but it does not image endogenous active Rab5. Therefore, an imaging probe that specifically images the spatio-temporal dynamics of endogenous Rab5 activation is much needed, and will accelerate the studies of protein signal transduction, vesicular trafficking, toxin and viral entry and cellular response to stress, which are all mediated by Rab5 through endocytosis (14).
We report here the development and validation of novel Rab5 biosensors for detecting endogenous Rab5 activity in fixed tissues and single live cells. Our Rab5 activity fluorescent biosensor (RAFB) uses the Rab5 binding domain of Rab5 effector --Rabaptin 5 (124 amino acid in length) as the binding moiety specific to active Rab5 and a cell penetrating peptide for live-cell delivery. This peptide moiety is conjugated to a fluorophore that can be quenched through fluorescence resonance energy transfer (FRET), so it can enter live cells and stay quenched until reaching the cytoplasm. Our RAFB biosensor exclusively targets the GTP-bound Rab5 associated with the early endosome membrane, with minimum background noise compared with conventional fluorophore-conjugated antibodies, allowing imaging of active Rab5 in live cells and fixed tissues. This new biosensor for imaging Rab5 activity will facilitate studies of pathophysiological changes in disease related to up-regulated Rab5 activity.
Rab5 presented on the inner leaflet of the plasma membrane, early and intermediate endosomes or macropinosome is important for both clatherin-dependent, receptor-mediated endocytosis and clatherin-independent, fluid-phase uptake. Rab5 GTPase is activated by guanine nucleotide exchange factors (GEFs) and deactivated by GTPase activating proteins (GAPs). The activation of Rab5 GTPase accelerates the rates of nucleotide exchange in Rab5 GTP binding site and the Rab5-bound GTP hydrolysis (15). The translocation of Rab5 GTPase between cytosol and inner membranes is facilitated by GDP dissociation inhibitor (GDI) and GDI displacement factors (GDFs). Targeting of Rab5 GTPases to specific organelles also depends on GEFs, effectors and GAPs (16) (Figure 1, panel a).
We chose the Rab5 binding domain (R5BD) of rabaptin 5 (Rab5 effector) as the active Rab5 recognition moiety of the Rab5 activity fluorescent biosensor (RAFB). The cDNA of the 124 amino acid peptide sequence (17) (Figure 1, panel b) was subcloned into the GST-fusion protein expression vector pGEX-4T-2, resulting in a GST-R5BD fusion protein construct with the GST sequence at the 5′ end (Figure 1, panel b). The six amino acids thrombin cleavage sequence (LVPRGS) is derived from the vector, and the 9 amino acids polyarginine sequence (RRRRRRRRR) was introduced at the N-terminus of the R5BD by PCR to increase the RAFB cytoplasmic membrane entering efficiency (18). An 11 amino acids (PCHPQFPRCYA) biotin mimetic sequence (BMS) (19) was genetically fused to the carboxyl terminus of the R5BD for subsequent fluorophore conjugation through binding to streptavidin. Quantum dot (QD), a nanocrystal widely utilized in fluorescence imaging due to its superior fluorescent properties, was chosen to prepare the first probe (20). The 45 KD GST-R5BD fusion protein was pulled down in the cell lysate by the GSH-coated Sepharose 4B beads (Supplementary Figure 1). The GST-R5BD-BMS fusion protein was directly conjugated with streptavidin-coated QD525. The final probe was generated by thrombin cleavage and the QD-conjugated RAFB was recovered from the supernatant and quantified by the BCA assay (Figure 1, panel c).
We first tested the RAFB-QD525 for imaging Rab5 activity on the tissues of a Ts65Dn mouse, the most widely used mouse model for Down syndrome (DS) (21). Elevated expression of amyloid precursor protein has been linked to neurodegenerative processes in Ts65Dn mice (22). Additionally, abnormally high levels of Rab5 activity have been observed in Ts65Dn mouse tissues possibly due to the mis-regulated endocytotic pathway (23).
50 μm-thick cryostat cut sections of the 9 month-old DS mouse brains along with the wild-type (2N) controls were permeablized using 0.3% Triton X-100, and immunostained using RAFB-QD525. Polyclonal antibody against Rab5 was used to label the total Rab5 in mouse brain cells, and the fluorescence intensities generated by the immunostaining were used as the index to evaluate the relative amount of total Rab5 in the brain tissues of 2N and DS mice (Figure 2, panel a-d). The ratio of the average immunostaining fluorescence intensity of over 200 neuron cells between DS and 2N mice in cortex is 1.217 ± 0.024 (Figure 2, panel e), and 0.882 ± 0.018 in hippocampus (Figure 2, panel g), indicating the difference in the total Rab5 between DS and 2N neurons in both cortex and hippocampus is relatively small (p<0.0001) and the difference between the cortex and hippocampus is not statistically significant (P=0.6533). However, the level of active Rab5 (Rab5-GTP) in DS mouse brain is much higher than the 2N mice, as revealed by the RAFB-QD525 probe. Imaging analysis of single cortical and hippocampal neurons rendered the same observation: the elevated Rab5 activity in the DS mouse model, but similar total Rab5 level in both genotypes (Figure 2, panel c and d). The active Rab5 levels represented by fluorescence intensity from RAFB-QD525 in 2N and DS mouse brain samples were similarly quantified by analyzing over 200 RAFB-stained cells. We found that in the DS mouse model (i.e. Ts65Dn), the average level of GTP-bound active Rab5 was increased by 4.5-fold in the cortices (Figure 2, panel f) and by 1.8-fold in the hippocampi (Figure 2, panel h), compared with the 2N mice. We have further generated the 3D simulated graphic images to depict the different Rab5 active level in cortices and hippocampi for both DS and 2N mice (Figure 2, panel i).
In order to examine the general tissue structure within the cortical and hippocampal areas, we used 0.5% thionin to stain the paraffin-embedded tissue sections through the cortical and hippocampal regions of the brain in Ts65Dn and 2N mice. Thionin stains DNA and Nissl substance, which is primarily ribosomal RNA. The staining is widely used as an indicator of general status of brain cells. No apparent abnormalities were found in both cortical and hippocampal regions for DS and 2N mouse brains, respectively, from the thionin staining (Figure 2, panel j). We have also shown before that Ts65Dn mice brain is devoid of any plaques and tangles, and no increased levels of amyloid has been found in these mice (22). Together with the RAFB imaging data, these results suggest that the abnormal active Rab5 level may occur earlier before the appearance of other pathophysiological hallmarks of AD.
Rab5 has been reported to regulate endocytosis and set the pace of membrane fission and fusion between plasma membrane and inner membranes (24, 25). Abnormally high Rab5 activity perturbs the homeostasis between membrane fission and fusion, thus compromises vesicle trafficking and endosome-associated signal transduction, creating pathogenic conditions in the cell (26). We next developed the RAFB to image active Rab5 in single live cells in real time.
To apply the RAFB for live-cell imaging, we have designed a pre-quenching module to keep the RAFB from emission during the endocytosis process after internalization The pre-quenched RAFB will regain fluorescence emission after it leaves the endosome and enters the cytoplasm (27). To have RAFB pre-quenched during the membrane docking and endocytosis steps before it reaches the cytoplasm and binds the active GTP-bound Rab5 at the cytosolic surface of the endosome (Figure 3, panel a) will presumably minimize the imaging background, because it is often difficult to wash out excessive amount of RAFB under the live-cell imaging conditions once RAFB gets internalized inside the cell. Furthermore, many pathophysiological conditions lead to compromised endocytotic pathways which prevent endosome sorting and maturation, hence stalling the endosomal cargo contents inside the abnormal endocytotic compartments instead of releasing them into the cytosol (28, 29). In this case the Rab5 activity imaging could be inaccurate due to possible false positive fluorescence from the aggregated inner membranes, and the stuck RAFB within the aggregates.
To keep the RAFB quenched before reaching the cytoplasm from endosomes, we introduced a dabcyl quencher moiety in the sensor. We chose fluorescein isothiocyanate (FITC) as the fluorophore in order to maximize the dabcyl quenching efficiency meanwhile maintain the optimal fluorescence quantum yield at the neutral pH environment in the cytoplasm where Rab5 is activated. The dabcyl quencher was connected via the disulfide bond through the side chain of cysteine, so once in the cytoplasm the abundant glutathione would reduce the disulfide bond to release the dabcyl quencher (Figure 3, panel b). The addition of biotin to the FITC-Cys-Dabcyl complex through lysine enables its conjugation to the R5B-BMS fusion peptide through the binding to streptavidin.
The biotinylated Lys(Dde)-Cys(Trt)-OH was synthesized by conventional solid phase peptide synthesis using Fmoc-Cys(Trt)-OH, Fmoc-Lys(Dde)-OH and biotin as the building blocks subsequently. After the removal of the Dde protecting group, FITC was introduced to the free ε-amine of Lysine. Then the resulting peptide was cleaved from the resin and purified by reverse-phase high performance liquid chromatograph (HPLC) (Figure 3, panel c). The product was confirmed by the MALDI-TOF mass spectrometer (Supplementary Figure 2, panel d).
We have first examined the quenching and activation efficiency of the FITC-Dabcyl unit after the synthesis. For in vitro test, series dilution of the Biotin-Lys(FITC)-Cys(S-S-Dabcyl) (from 0.1 μM to 5 μM) were treated with 5 mM of reduced L-glutathione at 37 °C for 5 min to simulate the cytosolic condition. The fluorescence emission spectra (from 510 nm to 650 nm) were collected, and the Biotin-Lys(FITC)-Cys(S-S-Dabcyl) has little emission at all concentrations. Addition of glutathione (GSH) produced a large (up to 2,500 fold) increase in its emission (Figure 3, panel d). For in vivo test, we measured the fluorescence emission at 520 nm of HeLa cells incubated with different concentrations of Biotin-Lys(FITC)-Cys(S-S-Dabcyl) (0, 0.1, 0.5, 1, 2, 5 μM). DMEM culture medium was used as a negative control, and all samples were incubated at 37 °C for 15 min with 5% CO2 before measurement. Strong fluorescence emission was detected in HeLa cell samples (F-D/cell, orange columns, Figure 3, panel e) in comparison to the samples incubated in DMEM culture medium without cells (F-D/medium, green columns, Figure 3, panel e). The fluorescent intensity increases up to 400 fold after activation in cells. These results suggest that the pre-quenched Biotin-Lys(FITC)-Cys(S-S-Dabcyl) can be efficiently activated by intracellular glutathione.
Biotin-Lys(FITC)-Cys(S-S-Dabcyl) was then conjugated with the recombinant R5BD-BMS peptide to generate the pre-quenched RAFB for intracellular Rab5 activity imaging. The GST-R5BD-BMS fusion protein was pull-down by GSH coated Sepharose 4B beads from expressing E. coli lysate, and streptavidin (SA) was added at 1:1 molar ratio to the beads slurry after 3 times PBS wash. Equivalent moles of Biotin-Lys(FITC)-Cys(S-S-Dabcyl) was then added to bind with SA, and the R5BD-BMS-SA-Biotin-Lys(FITC)-Cys(S-S-Dabcyl) biosensor (pre-quenched RAFB) was cleaved out by thrombin from the beads and recovered in the aqueous phase (Figure 4, panel a).
A control RAFB was generated by mutating the valine 822 into aspartic acid in the R5BD domain of the RAFB using site-directed mutagenesis to produce a V822D isoform (R5BD(V822D)-BMS) that does not bind to GTP-bound active Rab5 (17). Similar conjugation of this mutant with Biotin-Lys(FITC)-Cys(S-S-Dabcyl) through SA afforded the negative probe RAFB-Mut as a control for live-cell imaging of Rab5 activity.
At the basal level in normal conditions, the active form of Rab5 in vivo accounts for just a small fraction of total Rab5 (15). Diseased cells show increased Rab5 activation, presumably because activation and suppression of Rab5 cycle is disturbed by perturbations of the homeostasis of Rab5-GTP binding, hydrolysis, and recycling between endosomal membrane and cytoplasm (30). We simulated the diseased states in mammalian cell cultures by overexpressing different forms of Rab5 to create a set of high and low levels of active Rab5. Two Rab5 mutants were fused with mCherry (a transfection marker) to make the mCherry-fused, constitutively active Rab5Q79L-mC and dominantly negative Rab5S34N-mC (31). The Q79L point mutation abolished the GTPase activity of Rab5, and therefore this mutant could not hydrolyze its bound GTP, leading to constitutive activation of Rab5. Contrary to the Q79L mutation, the Rab5S34N construct has its serine 34 replaced by asparagine, which stops the recharging process of GTP back to Rab5 once it is hydrolyzed and keeps Rab5 staying at the GDP-bound inactive state.
HeLa cells were transiently transfected with Rab5Q79L-mC or Rab5S34N-mC, and incubated with 100 nM of pre-quenched RAFB 18 hours after transfection. As an additional control, Rab5Q79L-mC expressing cells were incubated with 100 nM of pre-quenched non-binding RAFB-Mut. Cells were imaged live without washing after 30 min incubation at 37 °C (Figure 4, panel b). Strong punctuated green fluorescence signal (corresponding to the Rab5 activity) was observed in red fluorescent cells (indicating expression of constitutively active Rab5) when stained with RAFB (Figure 4, panel b, top row). In contrast, minimal green fluorescent signal was detected in constitutively active Rab5 expressing cells when stained with the negative control RAFB-Mut (Figure 4, panel b, bottom row). The average fluorescence intensity in cells stained with the RAFB is 65-fold higher than that in cells stained with the RAFB-Mut (Figure 4, panel c). This result confirms that the point mutation of V822D in the R5BD effectively disabled the binding between the RAFB and the active Rab5 and the observed fluorescence from the pre-quenched RAFB correlates with the level of active Rab5 in cells.
A very modest level of fluorescence signal was detected from the dominantly negative Rab5 expressing cell line when stained with the pre-quenched RAFB (Figure 4, panel b, middle row). The average fluorescence intensity in Rab5S34N-mC expressing cells is 11-fold lower than that in cells expressing constitutively active Rab5 (Rab5Q79L-mC) (Figure 4, panel c). The low RAFB signal in Rab5S34N-mC expressing cells is consistent with the fact that no exogenous active Rab5 is expressed and the observed weak, evenly distributed fluorescence is likely due to the basal level of endogenous Rab5 activation.
In Rab5Q79L-mC expressing cells, the strong RAFB signals are punctuated, suggesting that they were with aggregated endosomes with active Rab5 associated on the surface. In Rab5S34N-mC expressing cells, the pre-quenched RAFB reaches the cytosol but cannot associate with endosomes because few active Rab5 at the cytosolic side of endosomes is available to bind. Consequently, the fluorescence of RAFB will be activated by the glutathione reduction in the cytosol, but will be weak, evenly distributed. These results have demonstrated that the pre-quenched RAFB can specifically image active Rab5 in live cells.
In summary, we report here the synthesis and validation of novel Rab5 activity biosensors for the detection of active Rab5 in both fixed tissues and live cells. Taking advantage of the Rab5 binding domain and the pre-quenching strategy, our biosensors are able to image the active form of small G protein Rab5 located on the cytosolic side of the early endosome membrane with minimum background noise (32). Since both endocytotic and autophagic pathways have been reported to be dysfunctional in different disease models, we expect a wide spectrum of applications for the Rab5 activity-imaging probe. Furthermore, it opens the door to label many signaling pathway regulators during pathogenetic cellular events. An array of downstream effectors of important small G proteins such as Ras, Rho, Rap1, Rac1, RalA, Cdc42, Arf1 and Arf6, could also be analyzed in a similar fashion as with Rab5 by exploiting their corresponding specific effector-binding domains to the active forms of these small G proteins.
The 372 bp cDNA sequence encoding Rabaptin-5 C-terminal binding domain (R5BD) specific to GTP-bound active Rab5 was synthesized and cloned into E. coli expression vector pGEX-4T-2 vector (GE Healthcare), with 5′-polyarginine and 3′-BMS DNA coding sequence addition. Site-directed mutagenesis was conducted to mutate the valine 822 into aspartic acid (V822D) in the R5BD, and the non-binding R5BD mutant was used to make the negative control RAFB-Mut. Details of DNA sequences, cloning and mutagenesis steps are described in Supporting Information.
The pGEX-4T-2-R5BD-BMS and the non-binding mutant (pGEX-4T-2-R5BD(V822D)-BMS) plasmids were transformed into BL21 (DE3) competent cells (Invitrogen) and the recombinant peptides were expressed by induction of 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). GST-R5BD fusion proteins were pulled down by glutathione-coated Sepharose 4B beads (GE Healthcare), and conjugated with biotin-Lys(FITC)-Cys(S-S-Dabcyl) through streptavidin. Assembled RAFBs were cleaved by thrombin from the GST domain and recovered by centrifugation. More details of the RAFB production steps are described in Supporting Information.
Detailed synthetic procedures and characterizations of the Biotin-Lys(FITC)-Cys(S-S-Dabcyl) and Biotin-Lys(FITC)-Cys fluorophores are described in Supporting Information.
HeLa cells were purchased from the American Tissue Culture Collection (ATCC) and maintained at 37 °C with 5% CO2 in DMEM supplemented with 100 unit/ml penicillin-streptomycin (GIBCO), and 10% (v/v) fetal bovine serum (GIBCO). Cells were split every 3 days at 1:5 to 1:10 dilutions and cultured to over 80% confluence before DNA plasmid transfection. Transfected cells were incubated at 37 °C with 5% CO2 for 18 hours, and then trypsinized and seeded on 35 mm glass bottom culture dishes with 14 mm microwells (MatTek) at a density of 5×104 cells per well for 24 hours culturing before washed by phenol red-free DMEM medium (GIBCO). Cells were treated with RAFB and the non-binding mutant (RAFB-Mut), for different time points, and subjected to microscopic imaging. Detailed imaging procedures are described in Supporting Information.
Ts65Dn (DS) mice and their littermates (2N) were saline perfused as previously described (22). Mouse brains were extracted and fixed in 4% (v/v) paraformaldehyde (PFA) and dehydrated by 30% (w/v) sucrose in PBS. A set of mouse brains were then embedded in optimal cutting temperature compound (Tissue-Tek) and μm cut coronal into 50 sections. Sections were then permeablized by 20% (v/v) goat serum (Bio-Rad) plus 0.3% (v/v) Tween-20 (Fisher Scientific) and immunostained by Rab5b polyclonal antibody (Santa Cruz, 1:1000) and 5 nM RAFB simultaneously at 4°C overnight. Secondary antibody Alexa Fluor 568 goat anti-rabbit IgG (H+L) (Molecular Probes) was added to visualize total Rab5 staining. Another set of brains was embedded in paraffin and cut into 6-μm-thick sections for histochemical staining. These sections were stained using 0.5% thionin (Sigma-Aldrich) and mounted on microscope glass slides (VWR) with xylene based mounting medium.
Fluorescence images for fixed brain tissues were captured using a Zeiss LSM 510 confocal laser scanning microscope. Thionin-stained images were captured using a Nikon Eclipse microscope and a Nikon DX1200 digital camera. RAFB live-cell detections of Rab5 activity were performed using an Olympus IX81-ZDC focus drift-compensating microscope in a closed environment chamber (Precision Control) with a Hamamatsu C1060010B CCD camera. Details of live-cell imaging process are described in Supporting Information. Image-Pro Plus (Mediacy) software was used for RAFB/RAFB-Mut live-cell imaging fluorescent intensity quantification, and Rab5 antibody/RAFB brain tissue staining image analysis and production of surface imaging. The surface imaging utility allows us to convert optical density generated by staining into changes in Z direction facilitating the analysis of the amount and location of staining.
We wish to thank other members of the Rao and Salehi laboratories for their assistance, especially Sarah Moghadam, and extend our gratitude to the Center for Research and Treatment of Down Syndrome and the Cell Sciences Imaging Facility at Stanford Medical School for providing mouse brain tissues and imaging facilities. This work has been supported by a Young Investigator Award from Human Science Frontier Program, and research grants from NIGMS (1R01GM086196-01) and DSRTF (to AS).
Supporting information Available: This material is available free of charge via the Internet at http://pubs.acs.org