Rab GTPases and Their Interacting Proteins in Health and Disease
Quantitative image analysis approaches for
probing Rab GTPase localization and function
in mammalian cells
Vasanth R. Singan*†, Kenan Handzic* and Jeremy C. Simpson*1
*School of Biology and Environmental Science & Conway Institute of Biomolecular and Biomedical Research, University College Dublin, Belfield, Dublin 4,
Ireland, and †School of Medicine and Medical Science, University College Dublin, Belfield, Dublin 4, Ireland
Abstract
Membrane traffic pathways play an essential role in cells, providing a mechanism for organelles of
the endomembrane system to communicate and exchange material between each other. A significant
number of infections and diseases are associated with trafficking pathways, and as such gaining a greater
understanding of their regulation is essential. Fluorescence-based imaging techniques are widely used to
probe the trafficking machinery within cells, and many of these methods have the potential to be applied
in a quantitative manner. In the present mini-review, we highlight several recent examples of how image
intensity, kinetic measurements, co-localization and texture feature analysis have been used to study the
function of one key family of membrane traffic regulators, the Rab GTPases. We give specific emphasis to
the importance of the quantitative nature of these recent studies and comment on their potential
applicability to a high-throughput format.
The dynamics of membrane traffic
The compartmentalization of eukaryotic cells into distinct
organelles enables them to perform a multitude of diverse
functions in parallel. With this complexity arises the need
for cellular machinery to actively transport the required
materials between compartments. The regulation of this
process of membrane traffic not only is of academic interest,
but also has relevance to drug delivery, infection and a
number of diseases [1]. By their inherent nature, membrane
traffic events are highly dynamic, involving the continuous
recycling of regulatory proteins between the cytosol and
appropriate membranes, and the fission and transport of
membrane-bounded transport carriers between organelles.
The regulation of membrane traffic is therefore complex,
requiring participation of a wide variety of molecule classes,
including soluble, peripheral and transmembrane proteins,
and lipids, particularly of the phosphoinositide family. One
key set of proteins well established in membrane traffic
regulation is the Rab family of small GTPases. In human
cells there are approximately 60 members, playing roles in all
stages of trafficking events [2]. Their importance is further
underlined by the ever increasing number of diseases being
attributed to Rab protein dysfunction [3]. One hallmark
of Rabs is that they constantly cycle between a soluble
cytosolic (inactive) pool and a membrane-associated (active)
Key words: image analysis, image texture feature, membrane traffic, quantitative co-localization,
Rab GTPase.
Abbreviations used: ERC, endosomal recycling compartment; FIP3, Rab11 family-interacting
protein 3; FRAP, fluorescence recovery after photobleaching; GFP, green fluorescent protein;
GM130, cis-Golgi matrix protein of 130 kDa; LAMP1, lysosome-associated membrane protein 1;
RWC, rank-weighted co-localization; SPT, single-particle tracking.
1
To whom correspondence should be addressed (email [email protected]).
Biochem. Soc. Trans. (2012) 40, 1389–1393; doi:10.1042/BST20120145
pool. Many of them are also present on the transport
intermediates themselves, and as such show high levels of
dynamics. In recent years, there has been an increased move
towards studying membrane traffic events in live cells, as this
provides access to both their spatial and temporal resolution,
which is particularly relevant to understanding Rab function
in both healthy and diseased cell states. Instrumental to
this has also been the advancement of fluorescent labelling
technologies and range of fluorescent proteins available, in
addition to the enhanced capabilities of modern microscopy
systems. However, the generation of image data is only
the first step in using microscopy approaches to study
membrane traffic. Although visual analysis of images is useful,
it is becoming increasingly accepted that all image data
need to be quantitative, not only to ensure robustness, but
also to maximize the information obtained. In the present
mini-review, we highlight some of the key quantitative image
analysis tools available to identify novel membrane traffic
machinery and study their localization and dynamics, with
particular focus on recent examples applied to the Rab family
of small GTPases.
Image intensity-based approaches
Membrane traffic processes have both a spatial and temporal
dimension. Trafficking events are most easily visualized
by taking a molecule of interest and labelling it with a
fluorophore that can be excited at a specific illumination
wavelength. Careful choice of fluorophore, particularly when
using fluorescent proteins, is important as properties such
as size, tendency to dimerize, brightness and photostability
can all influence subsequent quantitative analysis. The
measurement of fluorescence intensity from particular
C The
C 2012 Biochemical Society
Authors Journal compilation 1389
1390
Biochemical Society Transactions (2012) Volume 40, part 6
regions within an image is perhaps one of the simplest metrics
to quantify and has been used in a variety of ways in the
study of membrane traffic. In living cells, changes in image
intensity in perturbed systems can help to identify regulators
and define the roles of the molecules under study. One of the
most established quantitative methods in this respect is FRAP
(fluorescence recovery after photobleaching), a technique that
makes use of the fact that highly focused and intense pulses
of light can irreversibly destroy the targeted fluorophores.
Subsequent imaging is then used to monitor the return of
non-bleached fluorophores into the bleached area. Although
there are potential pitfalls in using this technique [4], it can
provide highly quantitative measurements of rates of protein
association and dissociation with membranes and flux rates
between organelles.
Recently, photobleaching approaches have been applied to
shed further light on the function of Rabs associated with
the Golgi apparatus, in particular the poorly characterized
Rab30 protein [5]. FRAP experiments in HeLa cells revealed
that GFP (green fluorescent protein)–Rab30 continuously
recycles on and off Golgi membranes, with kinetics similar to
those seen for another Golgi-localizing Rab (Rab6A), with
recovery of greater than 60% of the fluorescence intensity
over 6 min in the Golgi area. Although a number of GFP–
Rab30 vesicles were also observed in these experiments,
time-lapse imaging revealed that they were rarely seen to
be motile, thereby suggesting that, unlike Rab6A, Rab30 is
not a component of Golgi-derived transport carriers, but
rather plays a role in the maintenance of Golgi structure.
The Golgi apparatus is known to harbour a number of Rab
proteins, in particular Rab1, Rab2, Rab6, Rab30, Rab33B
and Rab43 [6]; however, the co-ordinated role of these in
Golgi morphology and trafficking through this organelle
is still far from clear. Recent quantitative imaging data
suggest a functional link between Rab6 and Rab33B in
retrograde traffic to the ER (endoplasmic reticulum) [7],
providing the intriguing possibility of the existence of Rab
cascades on this organelle, whereby the activity of one
Rab would determine the activity and localization of a second
Rab and so on. This theory is relatively well advanced in
the context of dynamics of the endosomal system [8], but
remains to be demonstrated for the Golgi. Presley and coworkers reported a novel quantitative approach based on
confocal line scan images across the Golgi followed by
intensity measurements in order to determine the relative
distribution of molecules across the various cisternae [9].
When applied to Rabs, such an approach undoubtedly offers
greater insight into their co-ordination on this organelle
[6], and in the future it would be pertinent to conduct
such experiments in the context of understanding their
mechanism of localization. Similarly, systematic FRAPbased characterization of membrane association/dissociation
rates for all Rabs may also yield important information.
However, such experiments are technically challenging from
a quantitative perspective as the compact nature of the Golgi
cisternae provides only limited resolution possibilities in
fluorescence light microscopy studies, and reliable kinetic
C The
C 2012 Biochemical Society
Authors Journal compilation quantification from FRAP experiments is highly dependent
on the volume of membrane occupied by the Rab and the
area bleached. Nevertheless, such experiments carried out
in cellular backgrounds depleted of putative Rab accessory
proteins, for example GEFs (guanine-nucleotide-exchange
factors), could provide critical data towards our greater
understanding of the mechanisms by which Rabs associate
with membranes.
Photobleaching studies have also been recently applied
to the study of Rabs in endosomal function within cells.
Gidon et al. [10] used photobleaching, combined with energytransfer measurements, to show the kinetics of the late steps
of Rab11-dependent recycling from the ERC (endosomal
recycling compartment) to the plasma membrane. Specifically
they demonstrated that a myosin Vb, Rab11A and Rab11–
FIP2 platform not only spatially regulates the trafficking
of the lectin langerin at sorting sites in the ERC, but also
regulates the late stages of docking/tethering and fusion
of langerin recycling vesicles at the plasma membrane in
a temporal manner. The detailed quantification of these
events afforded by such image intensity-based approaches
would not have been possible using biochemical methods.
A complementary technique to FRAP is photoactivation of
fluorophores. This approach utilizes a modified GFP, which
on excitation with 405 nm light converts from a dark state
into a visible state, allowing a population of the molecules to
be tracked over time. This technique was recently used
to highlight the role of Rab25 in integrin trafficking and
cancer progression. In cells expressing fluorescently labelled
Rab25, photoactivation of a distinct population of integrins
revealed that, during migration, rather than being degraded,
these molecules were in fact transported and recycled to the
plasma membrane, in a process dependent on the chloride
intracellular channel protein CLIC3 [11].
Kinetic-based methods
Live-cell video microscopy to provide analysis of protein
distribution in real time is also widely used in the study
of the dynamics of membrane traffic processes. Time-lapse
experiments are incredibly data-rich, not only allowing
visible access to discrete and subtle processes occurring in
cells, but also providing quantitative information relating
to speed, distance and trajectory of subcellular structure
movement over time. Within the context of Rabs, a recently
reported time-lapse study has provided further insight into
the mechanism of cytokinesis, specifically examining the
role of Rab11 and its effector FIP3 in delivering membrane
to the cleavage furrow [12]. Careful time-lapse imaging
revealed that although both GFP-tagged FIP3 (Rab11 familyinteracting protein 3) and mCherry–Rab11 accumulated near
the intercellular bridge during cytokinesis, at abscission
only FIP3 became localized to the Flemming body of the
midbody structure, at a site pre-populated by the GTPase
Arf6 (ADP-ribosylation factor 6). This study provides an
excellent illustration of how dual-colour time-lapse imaging
can provide in-depth detail about the kinetics of Rab function.
Rab GTPases and Their Interacting Proteins in Health and Disease
Organelle tracking has also been extensively employed,
particularly for quantitative studies, with a recently reported
application to study melanosome motility [13]. In this
work, the authors combined phase-contrast information from
highly pigmented melanosomes, with GFP–Rab information
from the fluorescence channel. In this way, they were
able to record and quantify, in a semi-automated manner,
several thousand melanosome movements from individual
cells [13]. Using this technique, they showed that Rab27a was
associated with slow-moving melanosomes, whereas Rab32
and its closely related isoform Rab38 were associated with
fast-moving melanosomes. This study demonstrated further
that Rab27a and its effectors have the ability to switch
melanosomes from microtubule- to actin-based transport
systems [13].
SPT (single-particle tracking) is increasingly being used as
a technique to study dynamic vesicle trafficking, with one
recent example addressing the important issue of how to
differentiate the dynamics of highly co-localizing proteins:
Szymanski et al. [14] used SPT of fluorescently tagged
Rab7 and LAMP1 (lysosome-associated membrane protein
1), both classical markers of late endosomes and lysosomes,
in order to establish their membrane distribution. Although
standard co-localization analysis (discussed below) revealed
high coefficients (∼0.8) between these two markers, it was
only when SPT was applied that the presence of Rab7 and
LAMP1 on distinct populations of dynamic membranes
was revealed. Subsequent SPT experiments using dextran as
an endocytic probe revealed that this cargo shifts between
different Rab7- and LAMP1-positive membranes over time,
but ultimately accumulates in a compartment labelled with
both markers [15].
Kinetic applications not only are of use for understanding
the function of Rab proteins and other trafficking machinery,
but also can be applied in the context of developing new
therapeutics. For example, using fluorescently labelled Rab
proteins (Rab5, Rab7, Rab9 and Rab11) as markers of the
endocytic pathway, a combined tracking and co-localization
approach was used to quantitatively follow the internalization of a pulse of synthetic nanoparticles into cells [16].
The real-time quantification of these potential drug-delivery
agents with various Rabs provides important insight into how
internalized molecules gradually change their intracellular
membrane environment over time. Potentially, this type of
approach, as with SPT, could be applied for any cargo and
trafficking step of interest in order to reveal the precise
kinetics of the transfer of material between organelles. This
is of particular relevance for elucidating Rab membrane
dynamics throughout the endocytic system, where the
complement of Rab proteins employed is relatively complex.
Quantitative co-localization and image
texture approaches
The majority of the quantitative imaging methods and
examples discussed so far provide in-depth information about
the function of a particular protein under investigation, but
cannot easily be applied in a systematic or high-throughput
manner. Nevertheless this is relevant, as recent genomewide membrane traffic studies have now provided us with
lists of uncharacterized proteins associated with endocytic
and secretory pathways [17,18]. At the very basic level, we
need to be able to assign precise subcellular localizations to
these proteins, and then determine their interaction networks.
To establish localization, a commonly used approach is to
co-localize the protein of interest with another marker of
known location, in so-called co-localization studies. A significant number of membrane traffic and Rab protein studies
have used this approach, typically immunolabelling organelles with specific antibodies and comparing their patterns
with either antibody staining or GFP-tagged versions of the
protein of interest. However, simply showing a merged image
of the two colour channels does not necessarily prove colocalization, and furthermore inconsistencies during image
acquisition can severely distort the degree of co-localization
presented. Appropriate quantification is therefore needed to
ensure that biologically relevant measurements are obtained.
A wide variety of algorithms exist for quantifying colocalization (reviewed in [19]); however, all of these methods
use either pixel intensity correlation or pixel co-occurrence
to generate a co-localization coefficient, and each of these
approaches has its own drawbacks. We recently developed
a novel method, the RWC (rank-weighted co-localization)
algorithm, which uses both pixel intensity correlation and
co-occurrence information to generate a co-localization
coefficient [20]. One key advantage of this method is that
it eliminates bias introduced at the stage of image analysis,
meaning that it can be applied systematically on largescale image datasets produced by automated microscopy.
Our preliminary studies indicate that when using the
cis-Golgi protein GM130 (cis-Golgi matrix protein of
130 kDa) as a reference channel this method is sufficiently
sensitive to be able to efficiently discriminate Rab proteins
localized to distinct compartments (see Figure 1A for an
example). Therefore, in principle, it could be systematically
applied to quantitatively study the localization profiles of
the entire Rab family across the endomembrane system.
Although, in this example, fixed cells were used, it could
be envisaged that a similar approach may also be possible in
live cells allowing, for example, Rab cascades to be studied
in a spatiotemporal manner.
Additional information to classify protein distribution and
function can also come from the so-called image texture
analysis. Texture features from individual cells effectively
provide a quantitative description of the fluorescence pattern
observed [21,22] (Figure 1B). Again, one potential advantage
of this technique is that it is applicable to high-throughput
image data generated from automated microscopes. To date,
this type of analysis has been used to cluster and classify
proteins with relatively distinct localization patterns [23,24].
In the example shown in Figure 1(C), the robustness of
using texture features for the identification and clustering
of distinct subcellular localization patterns can be seen.
Accurate clustering and classification of proteins within
C The
C 2012 Biochemical Society
Authors Journal compilation 1391
1392
Biochemical Society Transactions (2012) Volume 40, part 6
Figure 1 Use of quantitative co-localization and image texture features for systematic characterization of Rab proteins
(A) HeLa cells expressing either mCherry–Rab1A or mCherry–Rab5A were fixed and immunostained for the cis-Golgi marker
GM130. RWC coefficients for each Rab against GM130 were then calculated from 30 cells for each condition. Black lines in box
plots indicate median values. (B) HeLa cells expressing mCherry–Rab1A, mCherry–Rab5A or mCherry–Rab30 were imaged
and then individual cells were subjected to texture feature analysis. A total of 24 features were extracted from each cell
and the profiles of their scores are shown in the corresponding graphs. (C) HeLa cells were immunostained for six different
organelles as indicated, then image texture features extracted from five cells for each localization. PCA (principal component
analysis) and clustering was used to cluster the cells according to their features. All cells were correctly grouped. (D) Image
texture analysis and PCA clustering on the cells shown in (B). Cells expressing the two Golgi-associated Rabs (Rab1A and
Rab30) failed to be completely separated by the PCA and clustering, indicating a limitation of this technique. Scale bars,
10 μm.
the endomembrane system of cells is more challenging
for a number of reasons, including the fact that these
organelles tend to show a high degree of heterogeneity in
their morphology, and also because proteins associated with
membrane traffic function do not tend to localize to a single
compartment. Within the context of Rab proteins, this is
C The
C 2012 Biochemical Society
Authors Journal compilation particularly true, complicated further by the fact that at
steady state they exist in both a soluble and a membraneassociated pool. Automated texture analysis and clustering
can partially resolve the localization of different Rab proteins
(Figure 1D), but further refinements to this technique will
be needed if it is to be fully effective in resolving proteins
Rab GTPases and Their Interacting Proteins in Health and Disease
with overlapping distributions. Combining texture feature
analysis with quantitative co-localization information has
been shown to improve the clustering and classification of
a subset of Rab proteins localizing to the juxtanuclear area
of cells [25], but its applicability to larger datasets, such as
the entire 60 human Rab GTPases and each of their effectors,
remains to be demonstrated.
Concluding remarks
Imaging-based approaches continue to be a primary tool
towards furthering our understanding of cell function,
particularly in the context of Rab proteins and membrane
traffic. In recent years, there has been a steady move towards
improving and extending the quantification of our image data,
in order to extract a deeper level of knowledge from what we
observe. The methodologies now available to us are both
extensive and powerful, but one immediate challenge is to
adapt many of these techniques into a format compatible
with high-throughput imaging. This combination has the
potential to provide us with new and exciting opportunities
to understand the complexity of membrane traffic regulation
in a truly global and quantitative manner.
Funding
V.R.S. is supported by a graduate Ph.D. scholarship in Bioinformatics
and Systems Biology from the Irish Research Council for Science,
Engineering and Technology (IRCSET), and the J.C.S. laboratory is
supported by a Principal Investigator (PI) award [grant number
09/IN.1/B2604] from Science Foundation Ireland (SFI).
References
1 Howell, G.J., Holloway, Z.G., Cobbold, C., Monaco, A.P. and
Ponnambalam, S. (2006) Cell biology of membrane trafficking in human
disease. Int. Rev. Cytol. 252, 1–69
2 Stenmark, H. (2009) Rab GTPases as coordinators of vesicle traffic. Nat.
Rev. Mol. Cell Biol. 10, 513–525
3 Li, G. (2011) Rab GTPases, membrane trafficking and diseases. Curr. Drug
Targets 12, 1188–1193
4 Weiss, M. (2004) Challenges and artifacts in quantitative photobleaching
experiments. Traffic 5, 662–671
5 Kelly, E.E., Giordano, F., Horgan, C.P., Jollivet, F., Raposo, G. and
McCaffrey, M.W. (2012) Rab30 is required for the morphological integrity
of the Golgi apparatus. Biol. Cell 104, 84–101
6 Dejgaard, S.Y., Murshid, A., Erman, A., Kizilay, O., Verbich, D., Lodge, R.,
Dejgaard, K., Ly-Hartig, T.B., Pepperkok, R., Simpson, J.C. and Presley, J.F.
(2008) Rab18 and Rab43 have key roles in ER–Golgi trafficking. J. Cell Sci.
121, 2768–2781
7 Starr, T., Sun, Y., Wilkins, N. and Storrie, B. (2010) Rab33b and Rab6 are
functionally overlapping regulators of Golgi homeostasis and trafficking.
Traffic 11, 626–636
8 Del Conte-Zerial, P., Brusch, L., Rink, J.C., Kalaidzidis, Y., Zerial, M. and
Deutsch, A. (2008) Membrane identity and GTPase cascades regulated
by toggle and cut-out switches. Mol. Syst. Biol. 4, 206
9 Dejgaard, S.Y., Murshid, A., Dee, K.M. and Presley, J.F. (2007) Confocal
microscopy-based linescan methodologies for intra-Golgi localization of
proteins. J. Histochem. Cytochem. 55, 709–719
10 Gidon, A., Bardin, S., Cinquin, B., Boulanger, J., Waharte, F., Heliot, L., de
la Salle, H., Hanau, D., Kervrann, C., Goud, B. and Salamero, J. (2012) A
Rab11A/myosin Vb/Rab11-FIP2 complex frames two late recycling steps
of Langerin from the ERC to the plasma membrane. Traffic 13, 815–833
11 Dozynkiewicz, M.A., Jamieson, N.B., MacPherson, I., Grindlay, J., van den
Berghe, P.V., von Thun, A., Morton, J.P., Gourley, C., Timpson, P., Nixon,
C. et al. (2011) Rab25 and CLIC3 collaborate to promote integrin
recycling from late endosomes/lysosomes and drive cancer progression.
Dev. Cell 22, 131–145
12 Takahashi, S., Takei, T., Koga, H., Takatsu, H., Shin, H.W. and Nakayama,
K. (2011) Distinct roles of Rab11 and Arf6 in the regulation of
Rab11-FIP3/arfophilin-1 localization in mitotic cells. Genes Cells 16,
938–950
13 Hume, A.N., Wilson, M.S., Ushakov, D.S., Ferenczi, M.A. and Seabra, M.C.
(2011) Semi-automated analysis of organelle movement and
membrane content: understanding Rab–Motor complex transport
function. Traffic 12, 1686–1701
14 Szymanski, C.J., Humphries, IV, W.H. and Payne, C.K. (2011) Single
particle tracking as a method to resolve differences in highly colocalized
proteins. Analyst 136, 3527–3533
15 Humphries, I.V., W.H., Szymanski, C.J. and Payne, C.K. (2011)
Endo-lysosomal vesicles positive for Rab7 and LAMP1 are terminal
vesicles for the transport of dextran. PLoS ONE 6, e26626
16 Sandin, P., Fitzpatrick, L.W., Simpson, J.C. and Dawson, K.A. (2012)
High-speed imaging of Rab family small GTPases reveals rare events in
nanoparticle trafficking in living cells. ACS Nano 6, 1513–1521
17 Collinet, C., Stöter, M., Bradshaw, C.R., Samusik, N., Rink, J.C., Kenski, D.,
Habermann, B., Buchholz, F., Henschel, R., Mueller, M.S. et al. (2010)
Systems survey of endocytosis by multiparametric image analysis.
Nature 464, 243–249
18 Simpson, J.C., Joggerst, B., Laketa, V., Verissimo, F., Cetin, C., Erfle, H.,
Bexiga, M.G., Singan, V.R., Hériché, J.K., Neumann, B. et al. (2012)
Genome-wide RNAi screening identifies human proteins with a
regulatory function in the early secretory pathway. Nat. Cell Biol. 14,
764–774
19 Bolte, S. and Cordelières, F.P. (2006) A guided tour into subcellular
colocalization analysis in light microscopy. J. Microsc. 224, 213–232
20 Singan, V.R., Jones, T.R., Curran, K.M. and Simpson, J.C. (2011) Dual
channel rank-based intensity weighting for quantitative co-localization of
microscopy images. BMC Bioinf. 12, 407
21 Boland, M.V., Markey, M.K. and Murphy, R.F. (1998) Automated
recognition of patterns characteristic of subcellular structures in
fluorescence microscopy images. Cytometry 33, 366–375
22 Newberg, J. and Murphy, R.F. (2008) A framework for the automated
analysis of subcellular patterns in human protein atlas images. J.
Proteome Res. 7, 2300–2308
23 Glory, E. and Murphy, R.F. (2007) Automated subcellular location
determination and high-throughput microscopy. Dev. Cell 12, 7–16
24 Hamilton, N.A. and Teasdale, R.D. (2008) Visualizing and clustering high
throughput sub-cellular localization imaging. BMC Bioinf. 9, 81
25 Singan, V.R., Handzic, K., Curran, K.M. and Simpson, J.C. (2012) A method
for improved clustering and classification of microscopy images using
quantitative co-localization coefficients. BMC Res. Notes 5, 281
Received 9 July 2012
doi:10.1042/BST20120145
C The
C 2012 Biochemical Society
Authors Journal compilation 1393