# 2006 The Authors

Journal compilation# 2006 Blackwell Publishing Ltd

doi: 10.1111/j.1600-0854.2006.00506.xTraffic 2007; 8: 47–60Blackwell Munksgaard

Rab10 is Involved in Basolateral Transport in PolarizedMadin–Darby Canine Kidney Cells

Sebastian Schuck1,†, Mathias J. Gerl1,†,

Agnes Ang2, Aki Manninen3, Patrick Keller4,

Ira Mellman2 and Kai Simons1,*

1Max Planck Institute of Molecular Cell Biology andGenetics, 01307 Dresden, Germany2Department of Cell Biology, Yale University School ofMedicine, New Haven, CT 06520, USA3Biocenter Oulu, University of Oulu, 90220 Oulu, Finland4Meso Scale Discovery, Gaithersburg, MD 20877, USA*Corresponding author: Kai Simons, simons@mpi-cbg.de†These authors contributed equally to this work.

The sorting of newly synthesized membrane proteins to

the cell surface is an important mechanism of cell polar-

ity. To identifymore of themolecular machinery involved,

we investigated the function of the small GTPase Rab10

in polarized epithelial Madin–Darby canine kidney cells.

We find that GFP-tagged Rab10 localizes primarily to

the Golgi during early cell polarization. Expression of an

activated Rab10 mutant inhibits biosynthetic transport

from the Golgi and missorts basolateral cargo to the

apical membrane. Depletion of Rab10 by RNA interfer-

ence has only mild effects on biosynthetic transport

and epithelial polarization, but simultaneous inhibition

of Rab10 and Rab8a more strongly impairs basolateral

sorting. These results indicate that Rab10 functions

in trafficking from the Golgi at early stages of epithelial

polarization, is involved in biosynthetic transport to

the basolateral membrane and may co-operate with

Rab8.

Key words: basolateral membrane, biosynthetic trans-

port, epithelial polarity, Golgi, Rab GTPases

Received 2 August 2006; revised and accepted for publi-

cation 10 October 2006; published online 21 November

2006

Eukaryotic cells use an elaborate transport system to

control the distribution of their proteins and lipids. Once

synthesized, these molecules are sent to many different

destinations, including the endoplasmic reticulum (ER), the

Golgi, endosomes and the plasma membrane. Additional

complexity is found in polarized cells, which typically

possess distinct surface domains. Epithelial cells, for

example, have apical and basolateral membranes,

whereas neurons have axonal and somatodendritic do-

mains. To generate and maintain such asymmetry, polar-

ized cells sort newly synthesized cargo to different parts of

their plasma membrane. Biosynthetic sorting therefore is

an important determinant of cell polarity (1–3).

The trafficking routes and sorting sites for biosynthetic

cargo in polarized cells are still being defined. Even in the

well-studied epithelial Madin–Darby canine kidney (MDCK)

cell line, it is not entirely clear which apical and basolateral

proteins travel from the Golgi to the plasma membrane by

direct pathways, which proteins reach the surface via

endosomes and which proteins use a transcytotic mode

of delivery (4–6). In addition, it has long been debated

where apical and basolateral cargo is separated, and both

the trans Golgi network (TGN) and recycling endosomes

have been proposed as major biosynthetic sorting sites

(7,8). Although parts of the molecular machinery for polar-

ized surface transport are already known, more of the

missing players need to be identified to resolve these issues.

Rab GTPases are key regulators of membrane trafficking

(9,10). Rabs cycle between an inactive GDP-bound and

an active GTP-bound state. They are recruited from the

cytosol onto cell membranes in the inactive form, are

activated by guanine nucleotide exchange factors and then

bind a large number of Rab effector proteins. In this way,

Rabs co-ordinate the assembly of effector complexes and

generate functional membrane domains. Following inacti-

vation by GTP hydrolysis, Rabs release their effectors and

become available for the next round of activation and

effector binding. Rab effector proteins are functionally

diverse, which allows Rabs to co-ordinate multiple aspects

of membrane trafficking, including transport vesicle for-

mation, motility, docking and fusion. Many Rabs have been

specifically linked to particular transport pathways and are

therefore useful landmarks to map out the intracellular

trafficking network.

The Rab GTPase Sec4p is essential for post-Golgi traffick-

ing in yeast (11). The closest Sec4p homologues in higher

organisms are Rab8, Rab10 and Rab13 (12). Rab8 is

involved in transport to the basolateral membrane in

MDCK cells (13), and Rab13 functions in tight junction

assembly (14). The role of Rab10 is less clear. Rab10 has

a single isoform, is conserved throughout metazoan evo-

lution and is ubiquitously expressed in mouse and human

tissues (12,15). The mammalian Rab10 has been found at

the late Golgi in fibroblasts (16). More recently, two studies

have provided evidence that interfering with Rab10 func-

tion leads to defects in endocytosis in polarized cells. In

Caenorhabditis elegans, lack of Rab10 impairs endocytic

recycling in the intestine (17). In MDCK cells, mutant vari-

ants of Rab10 affect early endocytic events (18). Neverthe-

less, the homology to Sec4p suggests that Rab10may also

play a role in exocytic trafficking in MDCK cells. Here, we

test this hypothesis to gain more insight into the molecular

www.traffic.dk 47

machinery for biosynthetic transport in polarized epithelial

cells.

Results

Rab10 localizes primarily to the Golgi in MDCK cells

Rab proteins usually localize to the sites at which they

regulate membrane trafficking events. Mutants of a given

Rab that prefer the inactive GDP-bound state additionally

indicate where membrane recruitment takes place,

whereas mutants that prefer the active GTP-bound state

show where effector binding can occur [e.g. (19,20)]. We

therefore determined the distribution of green fluorescent

protein (GFP)-tagged versions of wild-type canine Rab10,

inactivated Rab10T23N and activated Rab10Q68L in

MDCK–TfR cells. These cells produce the human trans-

ferrin receptor, which facilitates labeling of the recycling

endosomes with fluorescent transferrin. The different

Rab10 constructs were introduced into MDCK–TfR cells

on coverslips by transient transfection and analyzed by

immunofluorescence microscopy. Weakly expressing

cells were chosen for imaging to avoid mislocalization

due to high levels of expression.

GFP-Rab10 was present in perinuclear tubular structures

that overlapped with those positive for giantin, a marker for

the Golgi (Figure 1A). GFP-Rab10 also showed substantial

co-localization with furin, a marker for the TGN, but only

minor overlap with endocytosed transferrin that had been

accumulated in recycling endosomes (Figure 1B). In addi-

tion, no co-localization was observed with the early endo-

some marker EEA1 (data not shown). Activated GFP-

Rab10Q68L co-localized well with furin, but not with

endocytosed transferrin (Figure 1C). Inactivated GFP-

Rab10T23N was in two pools, one that was dispersed

throughout the cytosol and one that coincided with giantin

as well as furin (Figure 1D and data not shown).

These observations indicate that Rab10 localizes primarily

to the Golgi, including the TGN, at least in incompletely

polarizedMDCK cells grown on coverslips. The distribution

of GFP-Rab10T23N and GFP-Rab10Q68L further indicates

that Rab10 is recruited onto Golgi membranes and is also

active at the Golgi. Thus, the localization of the different

Rab10 variants suggests that Rab10 is involved in mem-

brane trafficking events that take place at, or emanate

from, the Golgi.

Activated Rab10 inhibits biosynthetic transport from

the Golgi to plasma membrane

The normal nucleotide cycle of Rabs can often be inhibited

by the expression of inactivated or activated mutant

variants. At sufficiently high levels, inactivated GDP-bound

mutants are thought to sequester guanine nucleotide

exchange factors, so that the endogenous Rab can no

longer be activated (21,22). The effects of activated GTP-

bound mutants may depend on their expression levels.

At low levels, they can support the activity of the endog-

enous Rab (21). At high levels, they can act as inhibitors

(13,23), presumably by sequestering effector proteins and

thus interrupting the cycle of effector binding and release.

To explore the possible role of Rab10 in biosynthetic

transport, we introduced Rab10T23N and Rab10Q68L into

MDCK cells using recombinant adenoviruses to ensure

strong expression. We then analyzed the trafficking of

yellow fluorescent protein (YFP)-tagged temperature-

sensitive vesicular stomatitis virus glycoprotein (VSV-G)

in these cells. VSV-G is a transmembrane protein that can

be reversibly blocked in the ER at 39.58C and is sorted

mostly to the basolateral domain in polarized MDCK cells

at 328C, the permissive temperature for VSV-G transport

(24). Cells grown to confluence on coverslips were in-

fected with adenoviruses encoding red fluorescent protein

(RFP)-tagged Rab10T23N or Rab10Q68L, using conditions

that achieved infection rates close to 100%. Twelve hours

later, they were infected with a second adenovirus,

encoding VSV-G, and incubated at 39.58C for 6 h to

accumulate newly synthesized VSV-G in the ER. Protein

synthesis was blocked with cycloheximide and the tem-

perature was shifted to 328C, thus chasing VSV-G through

the biosynthetic pathway as a synchronous wave (25).

Cells were fixed after various times and analyzed by

immunofluorescence microscopy.

In control cells, VSV-G was found throughout the cytosol

immediately after the shift to 328C, exhibiting a pattern

characteristic of the ER and distinct from the compact

perinuclear distribution of giantin (Figure 2A, top panel).

After 25 min, VSV-G had moved to the Golgi, as indicated

by co-localization with giantin (Figure 2A, middle panel).

After 60 min, the bulk of VSV-G had reached the plasma

membrane (Figure 2A, bottom panel). In contrast, VSV-G

transport was strongly delayed in cells expressing acti-

vated Rab10Q68L. VSV-G accumulated in the ER and was

transported to the Golgi essentially as in control cells

(Figure 2B, top and middle panel). However, it then

persisted in perinuclear giantin-positive structures (Fig-

ure 2B, bottom panel). In addition, the giantin staining often

appeared more fragmented, possibly because the structure

of the Golgi is disturbed by the accumulation of VSV-G as

well as endogenous cargo.

Inactivated Rab10T23N caused no obvious delay of VSV-G

transport. Given the qualitative nature of the microscopic

assay, however, we cannot exclude that a slight delay of

VSV-G transport was missed. In addition, we noticed that

the expression of Rab10T23N, while it was quite strong

and easily detectable in almost all cells by the fluorescent

signal of the RFP tag, was much weaker than that of

Rab10Q68L. A comparison of the mRNA and protein levels

of the two Rab10 variants determined by quantitative

reverse transcriptase polymerase chain reaction (RT–PCR)

and immunoblotting with an anti-RFP antibody indicated

that RabT23N is less stable than Rab10Q68L (data not

48 Traffic 2007; 8: 47–60

Schuck et al.

Figure 1: Rab10 localizes primarily to

the Golgi. (A and D) GFP-Rab10 and GFP-

Rab10T23N were expressed in MDCK–

TfR cells by nucleofection. Both Rab10

variants co-localize with giantin, a marker

for theGolgi. (B andC)GFP-Rab10 andGFP-

Rab10Q68L were expressed in MDCK–TfR

cells by microinjection. Fluorophore-con-

jugated transferrin was bound to the cell

surface on ice for 30 min and accumu-

lated in recycling endosomes by incuba-

tion at 378C in the absence of fluorescent

transferrin for 22 min. Both Rab10 var-

iants co-localize well with furin, a marker

for the TGN. There is little co-localization

with internalized transferrin. Scale bars ¼10 mm.

Traffic 2007; 8: 47–60 49

Rab10 in Basolateral Transport

shown). It is possible, therefore, that we did not reach

levels of Rab10T23N sufficient to delay VSV-G transport.

Finally, Rab10 may be involved in a trafficking step that is

not rate limiting for VSV-G transport and that can still

support overall cargo flow through the biosynthetic path-

way without Rab10. Only blocking this step by sequester-

ing multiple effectors with activated Rab10Q68L may lead

to transport inhibition. Similar cases, in which an activated

Rabmutant caused clear alterationswhereas a correspond-

ing inactivated mutant had subtle or no effects, have been

reported before (13,14,26,27).

To confirm the inhibition of VSV-G transport by Rab10Q68L

biochemically, VSV-G delivery to the plasma membrane

was measured using surface biotinylation. As before,

confluent coverslip-grown control and Rab10Q68L cells

were infected with VSV-G adenovirus, the cargo was

accumulated in the ER at 39.58C and chased to the plasma

membrane by shifting the temperature to 328C. After thechase, tight junctions were opened with ethylenediamine-

tetraacetic acid (EDTA) and the whole cell surface was

biotinylated. Total and biotinylated VSV-G were quantified

by electrochemiluminescence. Following release from the

Figure 2: Activated Rab10 inhibits transport from theGolgi to the plasmamembrane. Coverslip-grown control and Rab10Q68L cells

were infected with VSV-G adenovirus. VSV-G was accumulated in the ER at 39.58C for 6 h. A) Control cells. VSV-G was chased out of the

ER at 328C for 0, 25 or 60min. VSV-G transiently co-localizes with giantin as it moves to the Golgi and reaches the plasmamembrane. Scale

bars¼ 10mm. B) Rab10Q68L cells. VSV-Gwas chased out of the ER as above. VSV-Gmoves to the Golgi as in control cells, but then fails to

reach the plasmamembrane and remains in the Golgi. Scale bars¼ 10 mm. C) VSV-Gwas chased out of the ER at 328C for 0, 30, 40, 50, 60

or 80 min. To measure arrival at the plasma membrane, the cell surface was biotinylated, and surface and total VSV-G were quantified by

electrochemiluminescence. ER to plasma membrane transport is strongly inhibited in Rab10Q68L cells. Data are mean � SEM from an

experiment performed in quadruplicate. D) VSV-Gwas accumulated in the Golgi at 19.58C for 90min and chased out of the Golgi at 328C for

0, 15, 30, 45, 60 or 75 min. Arrival at the plasma membrane was measured as above. Golgi to plasma membrane transport is strongly

inhibited in Rab10Q68L cells. Data are mean � SEM from an experiment performed in quadruplicate.

50 Traffic 2007; 8: 47–60

Schuck et al.

ER and a certain lag period, VSV-G steadily accumulated at

the plasma membrane in control cells, reaching maximum

levels after about 60 min. Surface arrival was strongly in-

hibited in cells expressing Rab10Q68L (Figure 2C). To test

if this was due to a delay in ER to Golgi transport, we made

use of the fact that Golgi exit but not ER exit of VSV-G is

blocked at 19.58C. VSV-G was accumulated in the ER at

39.58C, protein biosynthesis was shut off and the temper-

ature was shifted to 19.58C for 90 min. This led to an

accumulation of VSV-G in the Golgi in both control and

Rab10Q68L cells (Figure S1). When cells were then

incubated at 328C, VSV-G rapidly appeared on the surface

in control cells, but still did so only slowly in the presence

of Rab10Q68L (Figure 2D). Hence, the inhibition of VSV-G

surface delivery is not due to impaired ER to Golgi trans-

port, but more likely reflects an inhibition of a later trans-

port step.

Activated Rab10 retains biosynthetic cargo

in the TGN

Rab10Q68L could accumulate VSV-G in the Golgi by

inhibiting intra-Golgi transport, by delaying Golgi exit or

by interfering with post-Golgi trafficking. To narrow down

which transport step is most severely affected by

Rab10Q68L, we monitored the glycosylation of VSV-G

during surface delivery. VSV-G is modified in the ER by the

addition of two N-linked core glycans, which are converted

to complex type glycans in the Golgi (28). The glycans are

sensitive to cleavage by endoglycosidase H (endoH) until

they are rendered endoH resistant bymannosidase II in the

medial Golgi. EndoH resistance therefore indicates that

VSV-G has passed through the early Golgi. Control and

Rab10Q68L cells were infected with VSV-G adenovirus,

newly synthesized VSV-G was accumulated in the ER at

39.58C, chased to the Golgi at 19.58C and finally chased to

the plasma membrane at 328C. At each stage, cells were

lysed, treated with endoH and subjected to immunoblot-

ting with an anti-VSV-G antibody, or fixed and processed for

immunofluorescence microscopy.

In control cells, VSV-G was entirely endoH sensitive when

accumulated in the ER, as indicated by a shift in its

electrophoretic mobility upon endoH treatment (Figure 3A,

lane 1 and 2). Incubation at 19.58C for 120 min led to trans-

port from the ER to the Golgi as judged by co-localization

with giantin (see Figure S1). At this point, about half of the

VSV-G had become endoH resistant (Figure 3A, lane 3).

This incomplete conversion into the endoH-resistant form

has been observed before (28) and likely reflects the high

demand on the machinery for protein transport and glyco-

sylation during release of the accumulated VSV-G from the

ER. When cells were chased for another 60 min at 328C,VSV-G reached the plasma membrane (see Figure 2A and

D) and became fully endoH resistant (Figure 3A, lane 4). In

the presence of Rab10Q68L, VSV-G stayed endoH sensi-

tive during the incubation at 19.58C (Figure 3A, lane 5–7),

although it had reached the Golgi (see Figure S1). This

observation is consistent with a defect in transport through

the early Golgi. However, it is equally consistent with an

inhibition of Golgi exit, which could cause cargo to build up

in the TGN and subsequently lead to an accumulation of

incoming cargo also in the early Golgi. It is possible that

endogenous cargo had already obstructed the late as well

as medial Golgi at the timewhen VSV-G was released from

the ER, thus preventing VSV-G from becoming endoH re-

sistant.When the temperaturewas shifted to 328C for 60min,

most of the VSV-G acquired endoH resistance (Figure 3A,

lane 8). In contrast to control cells, however, VSV-G re-

mained intracellular and exhibited a distribution similar to

that of giantin (Figure 3B, top panel). These observations

show that VSV-G had encountered mannosidase II at this

stage, but was unable to exit the Golgi. Nevertheless, VSV-G

and giantin showed little overlap. Instead, they were

typically found in close apposition, with the more tubular

giantin-positive structures surrounding those positive for

VSV-G. When the experiment was repeated with MDCK–

TGN38 cells, which express the human TGN38 as amarker

for the TGN, extensive co-localization between VSV-G

and TGN38 was observed (Figure 3B, bottom panel).

These results show that VSV-G is able to pass through

the cis Golgi in Rab10Q68L cells, but is then retained in

the TGN. Hence, the step most severely affected by

Rab10Q68L is downstream of transport through the early

Golgi. These effects of Rab10Q68L suggest that Rab10

functions in transport from the Golgi, although we cannot

rule out that the inhibition of Golgi exit is caused indirectly

by a block at a later step.

Activated Rab10 selectively disrupts

basolateral sorting

At long chase times, VSV-G eventually reached the plasma

membrane in Rab10Q68L cells. However, instead of being

sorted mostly to the basolateral side as in control cells, it

was partially mistargeted to the apical surface. This was

already obvious in confluent cells on coverslips, but is best

appreciated in MDCK cells grown on filters to generate

a well-polarized epithelial monolayer. When VSV-G was

expressed in control cells, accumulated in the ER and

chased to the plasma membrane at 328C for 90 min, it

was mainly at the basolateral domain, and there was

little overlap with gp135/podocalyxin, a marker for the

apical membrane (Figure 4A, top panel). The VSV-G visible

underneath the apical membrane was still in the subapical

Golgi region, as judged by co-localization with giantin (data

not shown). In contrast, a large fraction of VSV-G was

transported to the apical membrane in cells expressing

Rab10Q68L (Figure 4A, bottom panel). Rab10Q68L cells

were somewhat less well polarized than control cells, pre-

sumably as a result of the disturbed biosynthetic sorting,

but overall polarity of the epithelial monolayer was main-

tained. The intracellular retention of VSV-G by Rab10Q68L

was less pronounced than in coverslip-grown cells (see

Figure 2), likely because infection with the Rab10Q68L

Traffic 2007; 8: 47–60 51

Rab10 in Basolateral Transport

adenovirus was less efficient in filter-grown cells and

yielded lower Rab10Q68L expression levels.

To quantify the mistargeting caused by Rab10Q68L, VSV-G

was expressed in filter-grown MDCK cells, radioactively

labeled during accumulation in the ER at 39.58C and chased

to the plasma membrane at 328C for 90 min. After the

chase, cells were placed on ice and treated with trypsin

from the apical or basolateral side. Low trypsin concentra-

tions were used that preserve the separation of apical and

basolateral domains (29). This treatment resulted in partial

cleavage of the VSV-G ectodomain. Full-length and cleaved

VSV-G were immunoprecipitated with an antibody against

the cytoplasmic YFP tag and the fraction of VSV-G cleavable

from the apical or basolateral sidewas quantified (Figure S2).

In control cells, approximately 70% of the surface VSV-G

were basolateral (Figure 4B), in agreement with previous

data obtained by domain-selective surface biotinylation

(24). In Rab10Q68L cells, however, the majority of the

surface VSV-G was found at the apical membrane, and only

40% were basolateral. As the basolateral domain is larger

than the apical, this distribution does not reflect random

delivery, but preferential apical transport. In addition, only

about 80% of the cells were infected by the Rab10Q68L

adenovirus, so that the effect of the activated Rab10mutant

is probably underestimated due to the presence of cells

that express VSV-G but not Rab10Q68L. These results

confirm that activated Rab10 causes apical mistargeting of

VSV-G and suggest that Rab10 is involved in the polarized

sorting of newly synthesized membrane proteins.

To test if Rab10Q68L also affects the polarity of other

basolateral or apical cargo, we analyzed the targeting of

five additional transmembrane proteins. Besides VSV-G,

we used YFP-tagged amyloid precursor protein (YFP-APP)

and LYFPGT46 as basolateral markers. Both VSV-G and

YFP-APP are sorted basolaterally by means of tyrosine-

based sorting motifs in their cytoplasmic tails. LYFPGT46

is a chimeric fusion protein that contains the atypical FTSL

basolateral sorting motif at the extreme C-terminus of its

cytoplasmic part (30). Apical markers were HA-M2-YFP

(based on influenza virus hemagglutinin), p75NTR (the

Figure 3: Activated Rab10 retains VSV-G in the TGN. A) Coverslip-grown control and Rab10Q68L cells were infected with VSV-G

adenovirus. VSV-Gwas accumulated in the ER at 39.58C for 6 h, accumulated in the Golgi at 19.58C for 120min and chased out of the Golgi at

328C for 60min.After eachstage,cell lysateswere treatedwithendoHandVSV-Gwasanalyzedby immunoblotting.Transport beyond theearly

Golgi renders VSV-G resistant to degradation to its faster migrating de-glycosylated form (VSV-Ge). In control cells, VSV-G starts to acquire

endoH resistance during Golgi accumulation and becomes entirely endoH resistant during the chase at 328C. In Rab10Q68L cells, VSV-G

remains endoH sensitive during Golgi accumulation, but is also mostly endoH resistant after the chase at 328C, showing that it has passed

the early Golgi. B) Rab10Q68L was expressed in MDCK or MDCK–TGN38 cells (upper and lower panel, respectively). VSV-G was accu-

mulated in the Golgi and chased at 328C for 60 min as above. After the chase at 328C, VSV-G is found in apposition to giantin and co-

localizes with TGN38, indicating that it has reached the TGN but is unable to leave the Golgi. Scale bars are 10 mm in the merge and 2 mm in

the zoom.

52 Traffic 2007; 8: 47–60

Schuck et al.

human neurotrophin receptor) and an apical variant of

VSV-G, in which the tyrosine-based motif is masked.

MDCK cells grown to confluence on coverslips were

infected with Rab10Q68L adenovirus and, 12 h later, with

a cargo adenovirus encoding one of the marker proteins.

To focus on exocytic sorting, we aimed to generate a short

pulse of cargo. The cargo adenoviruses were used at high

multiplicities of infection so that the onset of expression

was rapid. In each case, the cargo was chased to the cell

surface with cycloheximide as soon as enough fluorescent

signals had accumulated to allow microscopic analysis,

and the chase was stopped as soon as the majority of the

cargo had reached the plasma membrane. Other marker

proteins were also tested, but only for those listed above

could the chase be started before the cargo had reached

the cell surface and terminated before substantial endo-

cytosis had occurred. In this way, the polarity of the initial

biosynthetic sorting was captured and contributions from

resorting after endocytosis were largely avoided.

The basolateral sorting of VSV-G and its partial apical

mistargeting by Rab10Q68L were readily observed also

in this setup with coverslip-grown cells (Figure S3A).

Similarly, YFP-APP was sorted mainly to the basolateral

membrane in control cells, but showed apical mistargeting

in the presence of Rab10Q68L (Figure 5A). The same was

true for LYFPGT46 (Figure 5B). The sorting of HA-M2-YFP

was essentially unchanged by Rab10Q68L (Figure 5C), as

was that of p75NTR and apical VSV-G (Figure S3B and C).

Using a trypsin-based biochemical assay (29), we con-

firmed that the apical sorting of influenza virus hemag-

glutinin was only slightly affected by Rab10Q68L also in

filter-grown cells (Figure S4).

These observations suggest that Rab10Q68L generally

causes apical mistargeting of basolateral cargo with tyro-

sine-based motifs, many of which are sorted by the AP-1B

clathrin adaptor complex (31). It is unknown if LYFPGT46 is

sorted by the same machinery, so that it remains to be

Figure 4: Activated Rab10 causes apical missorting of VSV-G. Filter-grown control and Rab10Q68L cells were infected with VSV-G

adenovirus. A) VSV-G was accumulated in the ER at 39.58C for 12 h and chased to the plasma membrane at 328C for 90 min. VSV-G is

transported mainly to the basolateral membrane in control cells. In Rab10Q68L cells, a substantial fraction of VSV-G is missorted to the

apical membrane, where it co-localizes with gp135/podocalyxin. Some VSV-G is still visible in the subapical Golgi region. B) VSV-G was

radioactively labeled while accumulating in the ER at 39.58C for 9 h and chased to the plasma membrane at 328C for 90 min. Cells were

treatedwith cold trypsin from the apical or basolateral side. VSV-Gwas immunoprecipitated, full-length and cleaved VSV-Gwere quantified

by phosphorimager analysis, and the fraction of apical and basolateral surface VSV-G was calculated. More than 70% of the surface VSV-G

were found at the basolateral membrane in control cells, but approximately 60%were sorted to the apical membrane in Rab10Q68L cells.

Data are mean � SEM from three independent experiments.

Traffic 2007; 8: 47–60 53

Rab10 in Basolateral Transport

determined if the effect of Rab10Q68L is restricted to AP-

1B-dependent cargo. The finding that the polarity of three

apical markers is essentially unaffected by Rab10Q68L

argues that interfering with Rab10 function specifically

disrupts basolateral sorting. Missorting of basolateral but

not apical cargo has also been found with activated Rab8.

Activated Rab8 does not, however, inhibit exit from the

Golgi (13).

Inhibition of Rab10 and Rab8a disturb basolateral

sorting in an additive manner

An activated Rab mutant can sequester multiple effector

proteins and thus have pleiotropic effects. To more faith-

fully assess the role of Rab10, we used RNA interference

(RNAi) to deplete the endogenous protein in MDCK cells.

Retrovirus-mediated expression of small hairpin RNAs

followed by elimination of non-transduced cells by anti-

biotic selection reduced Rab10 mRNA levels by 95 � 1%

[mean � standard error of the mean (SEM), n ¼ 5]. We

could not measure Rab10 protein levels directly as no

suitable antibody was available. Yet, expression of trans-

fected GFP-Rab10 was almost completely suppressed in

Rab10 knockdown cells (Figure 6A), suggesting that the

endogenous Rab10 had largely been removed.

Rab10 knockdown cells displayed no obvious delay in

biosynthetic transport of VSV-G and established a normal

epithelial monolayer when seeded onto filters (data not

shown). In addition, they sorted VSV-G with only slightly

lower accuracy than control cells (Figure 6B). This finding

suggests that Rab10 is not essential for basolateral sort-

ing, although we cannot exclude that the residual Rab10

was sufficient to maintain its function.

One reason for the mild sorting phenotype in Rab10

knockdown cells could be that the function of Rab10 is

partially redundant with that of another Rab protein,

particularly Rab8. MDCK cells express two Rab8 isoforms,

Rab8a and Rab8b, of which Rab8a is the major form as

judged by quantitative RT–PCR (data not shown). Retrovirus-

mediated depletion of Rab8a mRNA by 91 � 1% (mean �SEM, n ¼ 3) caused minor missorting of VSV-G, indicating

that also depletion of Rab8a alone does not disrupt baso-

lateral sorting (Figure 6C). When we targeted both Rab10

Figure 5: Activated Rab10 causes missort-

ing of basolateral but not apical cargo.

Coverslip-grown control and Rab10Q68L cells

were infected with A) YFP-APP, B) LYFPGT46

or C) HA-M2-YFP adenovirus. Cargo was accu-

mulated for the minimum time necessary to

allow microscopic analysis and chased to the

plasma membrane for 90 min. An apical and

a basal optical section are shown for each

condition. YFP-APP and LYFPGT46 are trans-

ported mostly to the basolateral membrane in

control cells, but are partially missorted to the

apical membrane in Rab10Q68L cells. The

polarity of HA-M2-YFP is essentially unchanged

in Rab10Q68L cells. YFP-APP consists of the

YFP tag (yellow) and the full-length APP (gray)

with the YTSI sorting motif (red). LYFPGT46 is

YFP (yellow) fused to the LDL receptor trans-

membrane domain (gray) and the CD46 cyto-

plasmic tail (black), including the FTSL sorting

motif (green). HA-M2-YFP is hemagglutinin

(gray) fused to the cytoplasmic tail of influenza

virus M2 (black) and YFP (yellow). Scale bars ¼10 mm.

54 Traffic 2007; 8: 47–60

Schuck et al.

and Rab8a using two different retroviruses, the fraction of

apical surface VSV-G increased from 30% to slightly over

40% (data not shown). Thus, the combined depletion of

Rab10 and Rab8a affects VSV-G sorting somewhat more

strongly than the depletion of either of the two Rabs alone.

However, we were unable to simultaneously reduce the

mRNA levels of both Rab10 and Rab8a by more than 90%,

presumably because cells with a strong depletion of both

Rabs are rapidly eliminated from the cell population. The

observed VSV-G missorting in Rab10/Rab8a double knock-

down cells is therefore likely an underestimate of the true

effect of removing both Rabs.

We reasoned that the expression of inactivated Rab10T23N

might be equivalent to Rab10 RNAi as it sequesters the

guanine nucleotide exchange factor necessary for the con-

version of Rab10 to the active GTP-bound form. Consistent

with this idea, adenovirally expressed Rab10T23N mis-

sorted VSV-G to a similar extent as Rab10 RNAi (Figure 6B

and C). We then combined Rab8a RNAi with adenoviral

expression of Rab10T23N. In this way, we aimed at

interfering with the function of Rab10 more acutely than

is possible with RNAi. This treatment increased the

fraction of surface VSV-G at the apical membrane from

30 to 45% (Figure 6C). Interference with both Rab8a and

Rab10 thus augments the effects of inhibiting only one of

the two Rabs and causes a more pronounced apical mis-

sorting of VSV-G. This effect suggests that Rab8a and

Rab10 are functionally related, either co-operating along

the same pathway or acting in parallel pathways.

Depletion of Rab10 impairs polarization of

MDCK cysts

One reason for the absence of an obvious morphological

phenotype in filter-grown Rab10 knockdown cells could be

that cell polarization is very robust in the presence of the

strong spatial cue provided by the filter support. It has been

noted before that growth in a three-dimensional matrix,

in which MDCK cells develop into hollow spheres of cells

called cysts, places greater demands on the machinery for

epithelial polarization than the two-dimensional filter sys-

tem and provides a more sensitive assay for cell polarity

(32). When single MDCK cells were cultured in collagen

gels, Rab10 knockdown cells indeed exhibited polarization

defects not seen on filters. After 10 days of culture, control

cells had formed well-organized cysts with a smooth apical

membrane facing the cyst lumen. Rab10 knockdown cysts

often had irregular shapes and their apical surface was

frequently bulged out (Figure 7). This difference between

control and Rab10 knockdown cells persisted up to at least

13 days of culture, the longest time tested.

More work is needed to understand the immediate

cause for the observed morphological aberrations and

potentially link them to a role of Rab10 in polarized sorting.

Nevertheless, these results show that Rab10 has a non-

redundant function in MDCK cells that is required for

proper epithelial polarization in cysts.

Discussion

The aim of this study was to better understand the

molecular machinery for biosynthetic transport in polarized

Figure 6: Depletion of Rab10 or Rab8a by RNAi causes minor

missorting of VSV-G, but simultaneous inhibition of both

Rabs has a clear effect. Knockdown cells were generated by

retrovirus-mediated RNAi. MDCK cells transduced with empty

retrovirus served as control. A) Rab10 knockdown cells suppress

expression of GFP-Rab10. Control and Rab10 knockdown cells

were transfected with GFP-Rab10 by nucleofection. Equivalent

amounts of cell lysate were separated by SDS–PAGE and ana-

lyzed by immunoblotting. Lysate from untransfected cells (MDCK)

served as background control. B) Rab10 knockdown only slightly

affects VSV-G sorting. VSV-G transport was assayed in filter-

grown control and Rab10 knockdown cells as in Figure 4B. Data

are mean � SEM from four independent experiments. C) Rab8a

knockdown and expression of Rab10T23N only slightly affect VSV-

G sorting, but combined treatment more strongly impairs baso-

lateral targeting. VSV-G transport was assayed in filter-grown

control and Rab8a knockdown cells that expressed Rab10T23N

or not as in Figure 4B. Data are mean � SEM from three

independent experiments. KD, knockdown.

Traffic 2007; 8: 47–60 55

Rab10 in Basolateral Transport

epithelial cells. We focused on Rab proteins, which are key

regulators of the intracellular membrane trafficking net-

work. We analyzed Rab10 because it is homologous to

Sec4p, a central component of post-Golgi trafficking in

yeast. Our results indicate that Rab10 regulates transport

from the Golgi, at least at early stages of MDCK cell

polarization, and is involved in basolateral sorting. How-

ever, Rab10 is not strictly required for biosynthetic trans-

port, as depletion of Rab10 yields only mild sorting and

polarity phenotypes. Depletion of Rab8a also causes only

minor missorting of basolateral cargo, but simultaneous

inhibition of the two Rabs results in a more pronounced

sorting defect. These findings suggest that Rab10 and

Rab8 are functionally related and point to considerable

robustness built into post-Golgi trafficking in polarized cells.

Where might Rab10 act? We observed GFP-tagged wild-

type Rab10, activated Rab10Q68L and inactivated

Rab10T23N mostly at the Golgi in coverslip-grown MDCK

cells. A minor population of GFP-Rab10 and GFP-

Rab10Q68L was found on recycling endosomes. Babbey

et al. recently reported that GFP-Rab10T23N localizes to

the Golgi also in filter-grown, fully polarized MDCK cells.

However, they observed GFP-Rab10 and GFP-Rab10Q68L

mostly on recycling endosomes under these conditions

(18). We have confirmed that transiently transfected GFP-

Rab10 is on recycling endosomes in more fully polarized

MDCK cells grown on filters (our unpublished data). It

therefore appears that Rab10 is at the Golgi at early stages

of MDCK cell polarization, but then shifts towards recy-

cling endosomes. This view is consistent with the other

previous studies of Rab10 localization, which showed that

HA-tagged Rab10 is at the late Golgi in non-polarized cells

(16) and that GFP-Rab10 is distributed between the Golgi

and endosomes in polarized intestinal cells in C. elegans

(17). Similar changes in the localization of Rab proteins

during cell polarization have been observed before. Rab11,

for instance, is found at recycling endosomes and the Golgi

in non-polarized cells, but is restricted to recycling endo-

somes in fully polarized MDCK cells [(33) and references

therein].

In the presence of Rab10Q68L, newly synthesized VSV-G

accumulates in the Golgi of incompletely polarized MDCK

cells, indicating a role of Rab10 in anterograde trafficking

through the secretory pathway. The step most severely

affected by Rab10Q68L is downstream of transport through

the early Golgi. It is difficult to determine if Rab10Q68L

retains VSV-G in the Golgi directly, by slowing down export

from the TGN, or indirectly, by inhibiting a later trafficking

step. However, as we found Rab10 primarily at the Golgi,

the most straightforward explanation for the effect of

Rab10Q68L is that Rab10 functions in transport from the

Golgi at early stages of MDCK cell polarization.

There is evidence that biosynthetic cargo such as VSV-G

reaches the basolateral membrane of MDCK cells via

recycling endosomes (25,34). As Rab10Q68L causes

apical missorting of VSV-G, one possibility is that Rab10

functions in trafficking from the Golgi to recycling endo-

somes during early epithelial polarization. Disrupting this

pathway by the expression of Rab10Q68L could force

basolateral cargo into a direct route from the Golgi to the

apical membrane. Alternatively, Rab10 could function in

trafficking from the Golgi to the apical membrane and

activated Rab10Q68L could draw basolateral cargo into

this pathway. While we cannot rule out this explanation,

we think it less likely as it does not account for the

observed Golgi retention of basolateral cargo. In addition,

the delivery of apical cargo is slowed down rather than

accelerated by Rab10Q68L (our unpublished data), pos-

sibly as a secondary consequence of an accumulation

of endogenous basolateral cargo in the Golgi. In either

scenario, the decision of apical versus basolateral delivery

would be associated with Golgi exit, arguing that the Golgi

is an important biosynthetic sorting station.

We observed the missorting of basolateral but not

apical cargo by Rab10Q68L in both coverslip-grown and

Figure 7: Depletion of Rab10 by RNAi disturbs epithelial

polarization. Control cells transduced with empty retrovirus and

Rab10 knockdown cells were grown in collagen gels to form

cysts. A) Apical and basolateral membranes were stained with

antibodies against podocalyxin (red) and E-cadherin (green),

respectively, and cell nuclei with DAPI (blue). Rab10 knockdown

cysts are typically less well organized and have curved apical

membranes. B) Cysts with flat or curved apical membranes were

counted. Data are mean � SEM from three independent experi-

ments, at least 50 cysts per condition were counted in each

experiment. Scale bars ¼ 30 mm. KD, knockdown.

56 Traffic 2007; 8: 47–60

Schuck et al.

filter-grown cells, suggesting that Rab10 plays a similar

role in basolateral sorting at early as well as late stages of

polarization. However, given the change in its localization

during polarization, Rab10 may be able to function at

different stations of the biosynthetic pathway. Epithelial

cells clearly make differential use of their repertoire of

trafficking pathways depending on their polarization state

(6). Biosynthetic sorting may likewise be accomplished at

different trafficking stations during polarization and the

involved sorting machinery may be relocated accordingly.

We therefore speculate that the sorting events involving

Rab10 occur at the Golgi during early polarization, but shift

towards recycling endosomes when MDCK cells achieve

full epithelial polarity.

We did not analyze possible effects of Rab10Q68L on

endocytosis. However, as Rab10Q68L missorts certain

biosynthetic cargo, interfering with Rab10 function could

disrupt the supply of components of the endocytic machin-

ery and in this way indirectly inhibit endocytosis and

recycling. This view is compatible with the work in C.

elegans, which showed that lack of Rab10 causes shrink-

age of recycling endosomes and swelling of early endo-

somes in intestinal cells (17). Rab10 is normally distributed

between the Golgi and endosomes in these cells, so that

the decrease in recycling endosome size may result from

an inhibition of Golgi to endosome transport. Internalized

cargo could subsequently accumulate in early endosomes

because transport through recycling endosomes is impaired.

Such indirect effects may also explain why the only other

study on Rab10 in MDCK cells reported that Rab10T23N is

at the Golgi and affects endocytic recycling from early

endosomes (18). We conclude that Rab10 functions in

exocytic transport, but we do not rule out that it may also

be needed, directly or indirectly, for endocytic trafficking.

The effects of Rab10Q68L indicate that Rab10 is involved

in basolateral sorting in MDCK cells, but do not answer the

question if Rab10 is also required for this process. At high

expression levels, activated Rab mutants are likely to

sequester and thus inactivate a number of effector pro-

teins. This may generally be more disruptive than interfer-

ing with the co-ordination of effector proteins by inhibiting

the corresponding endogenous Rab, be it with an inacti-

vated mutant or by RNAi. Inactivated Rab10T23N and

Rab10 RNAi had only minor effects on basolateral sorting,

suggesting that Rab10 is not essential for polarized trans-

port in MDCK cells. Inhibition of Rab10 may be tolerated

due to redundancy with other Rabs. One Rab that likely has

a related function is Rab8, as suggested by the more

pronounced missorting of basolateral cargo after simulta-

neous inhibition of both Rab10 and Rab8a. Accordingly,

expression of either inactivated Rab8 or inactivated Rab10

in the Drosophila wing epithelium cause relatively mild

problems in wing morphogenesis, whereas combined

expression of the two mutant Rabs has much more severe

effects (E.Marois andS. Eaton,MPI-CBG,Dresden,Germany,

personal communication). It therefore appears that Rab10,

although it is ubiquitously expressed (15), is indispensable

only in particular cell types, such as C. elegans intestinal

cells (17), during certain developmental processes, such as

cyst formation in MDCK cells, or after inhibition of func-

tionally related Rabs, such as Rab8.

The interplay of Rab10 and Rab8 deserves further analysis.

It is possible that the two Rabs can replace each other or

that they act in parallel basolateral pathways. A third

possibility is that they co-operate along the same pathway.

As discussed above, Rab10 likely functions transport

between the Golgi and recycling endosomes, whereas

Rab8 is associated with recycling endosomes (13). An

attractive hypothesis is that Rab10 and Rab8 co-operate

during basolateral transport as part of a Rab cascade, as

proposed for Rab5 and Rab4 as well as Rab5 and Rab7 in

the endocytic pathway in mammalian cells (35,36) and

for Ypt31/32p and Sec4p in the exocytic pathway in

yeast (37). A further speculation is that the multiple func-

tions of Sec4p in yeast are carried out by separate Rabs

in higher organisms, so that Rab8, Rab10 and possibly

Rab13 could represent a split version of Sec4p. Several

other Rab proteins will ultimately have to be integrated

into this model to obtain a more complete map of the

biosynthetic transport routes in polarized cells. These

include Rab11a and Rab11b, at least one of which seems

to function in basolateral transport (19,23), and Rab14,

which also could have a role in biosynthetic post-Golgi

trafficking (20).

Functional similarity and redundancy between Rab pro-

teins is a recurrent theme. Different isoforms of the same

Rab often have the same function andmay be interchange-

able (38–40). Distinct Rabs sometimes have related func-

tions and may be able to fulfill the same task by similar

means (26,41). Thus, it is likely that the intracellular

trafficking network has a number of redundant elements.

The resulting robustness may allow cells to tolerate the

inhibition of certain parts of their trafficking machinery

without overall disruption of sorting and transport. The Rab

family has expanded substantially during evolution (12).

This expansion presumably reflects the diversification of

membrane trafficking pathways in higher organisms, but

the benefits of increased robustness afforded by redun-

dancy may also have been a driving force.

In summary, this work defines Rab10 as a new component

of the machinery for polarized biosynthetic transport. Under-

standing the precise function of Rab10 and how it is

connected with other parts of the transport machinery, such

as cargo adaptors, molecular motors, vesicle fusion proteins

and other Rab GTPases, will require the identification of

Rab10 effector proteins. Given the complex morphology

of the post-Golgi membrane system, it will be a continuing

challenge to map transport and sorting events onto in-

tracellular locations. In addition, we consider it likely that

sorting decisions are made in different places depending on

cell type and developmental stage. To come to grips

Traffic 2007; 8: 47–60 57

Rab10 in Basolateral Transport

with this complexity, we need a more detailed understand-

ing of the molecular sorting machinery and its modular

construction.

Materials and Methods

AntibodiesMouse monoclonals were anti-gp135/podocalyxin (G. Ojakian, SUNY

Downstate Medical Center, Brooklyn, NY, USA), anti-giantin (H. P. Hauri,

University of Basel, Basel, Switzerland), anti-E-cadherin (Simons lab,

MPI-CBG, Dresden, Germany), anti-acetylated b-tubulin (Sigma, Munich,

Germany), anti-p75NTR and anti-myc (both from Santa Cruz Biotechnology,

Santa Cruz, CA, USA). Polyclonals were rabbit anti-mRFP (M. Zerial, MPI-

CBG), goat anti-GFP (D. Drechsel, MPI-CBG), rabbit anti-VSVG (Simons lab)

and rabbit anti-furin (Dianova, Hamburg, Germany). The rabbit anti-podoca-

lyxin antibody was raised against the CDNLAKDDLDEEEDTHL epitope.

Plasmids and adenovirusesCanine Rab10 cDNA M. Zerial was cloned into pEGFP-C1 (Clontech,

Moutain View, CA, USA) to generate pGFP-Rab10. Mutations yielding

pGFP-Rab10T23N and pGFP-Rab10Q68L were introduced using the Quik-

Change kit (Stratagene, La Jolla, CA, USA). RFP-tagged variants were

constructed by replacing the GFP with monomeric RFP1. They were

subcloned into pAdEasy-1 (Qbiogene, Heidelberg, Germany) and adenovi-

ruses were generated according to the Qbiogene manual. The p75NTR

adenovirus was from E. Rodriguez-Boulan (Cornell University, New York,

NY, USA). All other adenoviruses have been described before (24,42–44).

Cell cultureMDCK cells, MDCK–TfR cells expressing the human transferrin receptor

and MDCK–TGN38 cells expressing myc-tagged human TGN38 were

cultured as described (13). For transport assays, cells were seeded at

1.5 � 105 per 24 well containing a glass coverslip or at 5 � 105 per 12-mm

Transwell filter (Corning Life Sciences, Corning, NY, USA). After 24 h, cells

were infected or not with Rab10Q68L adenovirus in OptiMEM at 378C for

1 h and grown for another 12 h.

Transfection and immunofluorescence microscopyGFP-Rab10 constructs were introduced into MDCK–TfR cells by nucleo-

fection or microinjection. Nucleofection, an electroporation-based method,

was done as recommended by the manufacturer (Amaxa, Cologne,

Germany), cells were grown on coverslips for 20 h and immunostained.

Low expression levels of GFP-Rab10 and especially GFP-Rab10Q68L were

difficult to achieve by this method and the proteins were often mainly

cytosolic. Microinjection was therefore preferred to more tightly control

their levels of expression. Coverslip-grown cells were microinjected, grown

for 2 h, labeled with fluorescent transferrin and immunostained as

described (13). Confocal microscopy was performed with a laser scanning

microscope (LSM510; Carl Zeiss, Jena, Germany) using 40� water immer-

sion (n¼ 1.5) or 63� oil immersion (n¼ 1.4) objectives at 258C. Imageswere

processed with Adobe Photoshop software (Adobe, San Jose, CA, USA).

Microscopic transport assays on coverslipsTo assay VSV-G transport, coverslip-grown control and Rab10Q68L cells

were infected with VSV-G adenovirus in OptiMEM at 378C for 2 h. For

microscopic analysis, cells were incubated in normal medium at 39.58C for

6 h, chased in medium containing 40 mg/mL cycloheximide at 328C for

90 min and immunostained. To monitor its glycosylation, VSV-G was accu-

mulated in the ER, chased with cycloheximide at 19.58C for 120 min

and allowed to exit the Golgi at 328C for 60 min. Cells were collected at the

end of each stage and treated with endoH as described (28). The same

conditions as above were applied for apical VSV-G. For YFP-APP,

LYFPGT46, HA-M2-YFP and p75NTR, cells were infected with the respec-

tive adenovirus for 1–3 h, incubated at 378C for 2–9 h, chased with

cycloheximide for 90 min and immunostained. In each case, expression of

the marker was allowed for the minimum time necessary to produce a

sufficient amount for microscopic analysis, and the chase was started

when no marker protein was yet visible at the cell surface. Endocytosis

of these markers was relatively slow. This allowed most of the cargo

to reach theplasmamembranebeforesubstantial endocytosis hadoccurred.

Biochemical VSV-G transport assay on coverslipsVSV-G was accumulated in the ER of coverslip-grown control and

Rab10Q68L cells as described above. To assay ER to plasma membrane

transport, cells were chased in medium containing 40 mg/mL cycloheximide

at 328C for 0, 30, 40, 50, 60 and 80 min. To assay Golgi to plasma

membrane transport, cells were incubated in medium with cycloheximide

at 19.58C for 90 min to accumulate VSV-G in the Golgi and chased at 328Cfor 0, 15, 30, 45, 60 and 75min. Coverslips were washed with cold PBS and

incubated with 1 mM EDTA/PBS on ice for 10 min to open the tight

junctions and make the entire cell surface biochemically accessible. Cells

were surface biotinylated with 1 mg/mL Sulfo-NHS-LC-biotin (Pierce,

Rockford, IL, USA) on ice for 2 � 20 min. Unreacted biotin was quenched

with cold 0.3% BSA/0.1 M glycine for 2 � 10 min and cells were lysed with

PBS containing 2% Nonidet P-40 (NP-40), 0.2% sodium dodecyl sulphate

(SDS) and protease inhibitors. Total and biotinylated VSV-G were quantified

using electrochemiluminescence technology from Meso Scale Discovery

(Gaithersburg, MD, USA). To determine total VSV-G, MA6000 96-well

plates were coated with 10 ng affinity-purified rabbit anti-GFP antibody per

well, quantification of biotinylated VSV-G was done on avidin-coated

MA6000 high bind 96-well plates. Wells were incubated with 3% BSA in

PBS/0.2% NP-40 for 1 h to prevent non-specific binding. Lysates were

diluted in PBS containing 0.06% NP-40, 0.006% SDS and 100 mM HEPES

and added to the wells for 1 h to allow capture of the VSV-G by means of

the YFP tag or the biotin moiety. Wells were washed with PBS/0.2%

NP-40, and 20 ng rabbit anti-GFP antibody labeled with MSD Sulfo-Tag

were added for detection of the immobilized VSV-G. After washing, MSD

Read Buffer T with surfactant was added, and the electrochemilumines-

cence signal was detected using a Sector Imager 6000. Values for surface

VSV-G were normalized for total VSV-G to eliminate variations in expression

levels.

Transport assays on filtersTo measure the polarity of VSV-G surface transport, filter-grown control and

Rab10Q68L cells were infected with VSV-G adenovirus in OptiMEM at

378C for 2 h. For microscopic analysis, cells were incubated at 39.58C for

12 h, chased with cycloheximide at 328C for 90 min and immunostained.

For biochemical analysis, VSV-G was radioactively labeled and accumulated

in the ER by incubation at 39.58C for 9 h in labeling medium (methionine/

cysteine-free MEM, 1.5 mg/mL methionine, 2.5% fetal calf serum, 20 mM

HEPES pH 7.2) with 50 mCi 35S-methionine added to the basolateral side.

Cells were washed with PBS containing Mg2þ and Ca2þ (PBSþ) and

incubated in chase medium containing excess non-radioactive methionine

(methionine/cysteine-free MEM, 150 mg/mL methionine, 20 mM HEPES

pH 7.2) at 328C for 90 min. Cells were washed 3 � 5 min with cold PBSþand treated with 100 mg/mL trypsin (Worthington, Lakewood, NJ, USA) in

PBSþ from the apical or basolateral side on ice for 30 min. Tight junctions

remain intact under these conditions and restrict trypsin to the side to

which it has been added (29). Cells were washed 3� 5 min with 0.1 mg/mL

soy bean trypsin inhibitor (Worthington) and full-length and cleaved VSV-G

were immunoprecipitated with an anti-GFP antibody against the cytosolic

YFP tag that is protected from cleavage. Immunoprecipitates were

separated by SDS–PAGE and full-length and cleaved VSV-G were quantified

by phosphorimager analysis. The ratio of apical to basolateral surface VSV-G

was calculated from the fraction of VSV-G cleaved by trypsin from the apical

or basolateral side. The polarity of influenza virus hemagglutinin surface

transport was measured as described (29). Briefly, filter-grown MDCK cells

were infected with influenza virus and incubated for 3 h to allow host

protein shut-off to occur. Newly synthesized virus proteins were labeled

with 35S-methionine for 10 min and chased to the plasma membrane for

30 min. Cells were trypsinized as above, cell lysates were separated by

58 Traffic 2007; 8: 47–60

Schuck et al.

SDS–PAGE and the fraction of HA cleaved from the apical or basolateral side

was quantified by phosphorimager analysis.

RNAiThe Rab10 and Rab8a target sequences were GCTGAAGATATCCTTC-

GAAAG and AAGACAAGTTTCCAAGGAACG, respectively. Retrovirus-

mediated RNAi and quantitative RT–PCR were done as before (45).

Culture of MDCK cystsHydrated collagen I solution (2 mg/mL) was prepared by mixing 16 volumes

ice-cold VitrogenTM (3mg/mL collagen I; Cohesion, Palo Alto, CA, USA) with

2 volumes chilled 10� DMEM (Invitrogen, Karlsruhe, Germany), 2 volumes

0.1 M NaOH, 1 volume 7.5% NaHCO3 and 1 volume 1 M HEPES pH 7.2.

Fetal calf serum was added to a final concentration of 1% (v/v) and the

mixture was kept on ice until use. Subconfluent MDCK cells were trypsi-

nized and a single cell suspension containing 5� 106 cells/mLwas prepared

in PBS. Cells were pipeted into the collagen solution to yield a mixture

containing 2 � 105 cells/mL. The mixture was pipeted onto 24-well plates

and incubated at 378C for 30–45 minutes to allow the collagen to solidify.

Medium was added and the incubation was continued for 10 days. Medium

was exchanged every other day.

Immunostaining of MDCK cystsCollagen gels were rinsed twice with PBSþ and incubated in 0.05%

collagenase A in PBSþ for 5 min at 378C. After one wash with PBSþ, cells

were fixed in 4% paraformaldehyde for 30 minutes and excess aldehyde

was quenched with 200 mM glycine in PBS. Cells were permeabilized for

30 min in PBS containing 0.1% Triton-X-100 and non-specific binding was

blocked by incubation in PBS containing 0.5% BSA, 0.2% fish skin gelatin

and 0.01% Triton-X-100 (blocking solution) for 2 h. Cells were incubated

overnight with primary antibodies diluted in blocking solution. Dilutions

were 1:100 for rabbit anti-podocalyxin, and 1:50 for mouse anti-E-Cadherin

and mouse anti-acetylated b-tubulin, 4’,6-diamidino-2-phenylindole (DAPI,

Sigma) was diluted 1:10 000 and TRITC-Phalloidin (Sigma) 1:1000. Gels

were washed extensively with blocking solution, followed by overnight

incubation with fluorophore-conjugated secondary antibodies diluted 1:300

in blocking solution. Gels were washed with PBS and mounted onto slides

with Mowiol. Images were acquired with an OLYMPUS FluoView-1000

laser scanning confocal microscope (Olympus, Hamburg, Germany) using

a 60� PlanApo oil objective (n ¼ 1.1) at 258C.

Acknowledgments

We thank George Ojakian, Hans-Peter Hauri, Enrique Rodriguez-Boulan and

Marino Zerial for antibodies, adenoviruses and plasmids. We are grateful to

Marino Zerial for discussion and comments on the manuscript, and to Eric

Marois and Suzanne Eaton for communicating unpublished results. This

work was supported by EU FP5 contract no. HPRN-CT-2002-00259 and

Transregio SFB-TR13-TPA1.

Supplementary Materials

Figure S1: Accumulation of VSV-G in the Golgi is unaffected by

Rab10Q68L. Coverslip-grown control and Rab10Q68L cells were infected

with VSV-G adenovirus. VSV-G was accumulated in the ER at 39.58C for

6 h and chased out of the ER at 19.58C for 90 min. VSV-G co-localizes with

giantin in control and Rab10Q68L cells (upper and lower panel, respec-

tively), showing that it has left the ER and reached the Golgi. No change

occurred when the incubation at 19.58C was prolonged to 120 min. Scale

bars ¼ 20 mm.

Figure S2: Trypsin-based assay for biosynthetic sorting of VSV-G.

Filter-grown control and Rab10Q68L cells were infected with VSV-G

adenovirus. VSV-G was radioactively labeled for 9 h while accumulating in

the ER at 39.58C and chased to the plasma membrane at 328C for 90 min.

Cells were treated with trypsin from the apical or basolateral side at 48C for

30 min. Full-length VSV-G (VSV-G0) and cleavage products (VSV-GC) were

precipitated with an antibody against the cytoplasmic YFP tag, separated by

SDS–PAGE and quantified by phosphorimager analysis. The ratio of cleaved

to full-length VSV-G was calculated for each sample, so that a measure for

the fraction of VSV-G accessible by trypsin from the apical and basolateral

side was obtained. Samples not treated with trypsin served as background

controls.

Figure S3: Activated Rab10 causes missorting of basolateral but not

apical cargo. Coverslip-grown control and Rab10Q68L cells were infected

with A) VSV-G, B) p75NTR or C) apical VSV-G adenovirus. Cargo was

accumulated for theminimum time necessary to allowmicroscopic analysis

and chased to the plasma membrane at 328C (A, C) or 378C (B) for 90 min.

Immunostaining for podocalyxin and, in (B), neurotrophin receptor. An

apical and a basal optical section are shown for each condition. VSV-G is

missorted to the apical membrane in the presence of Rab10Q68L, but the

polarity of p75NTR and apical VSV-G is largely unchanged. VSV-G is VSV-G

ts045 (gray) with the YTDI sorting motif (red), fused to a spacer (black) and

YFP (yellow). p75NTR is the human neurotrophin receptor (gray). Apical

VSV-G is VSV-G ts045 (gray) fused to YFP (yellow) without a spacer, so that

the basolateral sorting motif (red) is masked. Scale bars ¼ 10 mm.

Figure S4: Apical sorting of influenza virus HA is essentially normal in

the presence of activated Rab10. Filter-grown control and Rab10Q68L

cells were infected with influenza virus and the polarity of HA surface

transport was determined. The accuracy of HA targeting to the apical

membrane is slightly reduced in cells expressing Rab10Q68L. This effect

is minor compared with the severe missorting observed for VSV-G (see

Figure 4B). Data are from an experiment performed in duplicate.

Supplemental materials are available as part of the online article at http://

www.blackwell-synergy.com

References

1. Mostov K, Su T, ter Beest M. Polarized epithelial membrane traffic:

conservation and plasticity. Nat Cell Biol 2003;5:287–293.

2. Horton AC, Ehlers MD. Neuronal polarity and trafficking. Neuron

2003;40:277–295.

3. Schuck S, Simons K. Polarized sorting in epithelial cells: raft clustering

and the biogenesis of the apical membrane. J Cell Sci 2004;117:

5955–5964.

4. Traub LM, Apodaca G. AP-1B: polarized sorting at the endosome.

Nat Cell Biol 2003;5:1045–1047.

5. Rodriguez-Boulan E, Kreitzer G, Musch A. Organization of vesicular

trafficking in epithelia. Nat Rev Mol Cell Biol 2005;6:233–247.

6. Schuck S, Simons K. Controversy fuels trafficking of GPI-anchored

proteins. J Cell Biol 2006;172:963–965.

7. Griffiths G, Simons K. The trans Golgi network: sorting at the exit site of

the Golgi complex. Science 1986;234:438–443.

8. Matter K, Mellman I. Mechanisms of cell polarity: sorting and transport

in epithelial cells. Curr Opin Cell Biol 1994;6:545–554.

9. Zerial M, McBride H. Rab proteins as membrane organizers. Nat Rev

Mol Cell Biol 2001;2:107–117.

10. Grosshans BL, Ortiz D, Novick P. Rabs and their effectors: achieving

specificity in membrane traffic. Proc Natl Acad Sci U S A 2006;103:

11821–11827.

Traffic 2007; 8: 47–60 59

Rab10 in Basolateral Transport

11. Salminen A, Novick PJ. A ras-like protein is required for a post-Golgi

event in yeast secretion. Cell 1987;49:527–538.

12. Pereira-Leal JB, Seabra MC. Evolution of the Rab family of small

GTP-binding proteins. J Mol Biol 2001;313:889–901.

13. AngAL, Folsch H, Koivisto UM, PypaertM,Mellman I. The Rab8GTPase

selectively regulates AP-1B-dependent basolateral transport in polarized

Madin-Darby canine kidney cells. J Cell Biol 2003;163:339–350.

14. Kohler K, Louvard D, Zahraoui A. Rab13 regulates PKA signaling during

tight junction assembly. J Cell Biol 2004;165:175–180.

15. Gurkan C, Lapp H, Alory C, Su AI, Hogenesch JB, Balch WE. Large-

scale profiling of Rab GTPase trafficking networks: the membrome.

Mol Biol Cell 2005;16:3847–3864.

16. Chen YT, Holcomb C, Moore HP. Expression and localization of two

low molecular weight GTP-binding proteins, Rab8 and Rab10, by

epitope tag. Proc Natl Acad Sci U S A 1993;90:6508–6512.

17. Chen CC, Schweinsberg PJ, Vashist S, Mareiniss DP, Lambie EJ, Grant

BD. RAB-10 is required for endocytic recycling in the C. elegans

intestine. Mol Biol Cell 2006;17:1286–1297.

18. Babbey CM, Ahktar N, Wang E, Chen CC, Grant B, Dunn KW. Rab10

regulates membrane transport through early endosomes of polarized

Madin-Darby canine kidney cells. Mol Biol Cell 2006;17:3156–3175.

19. Chen W, Feng Y, Chen D, Wandinger-Ness A. Rab11 is required for

trans-Golgi network-to-plasma membrane transport and a preferential

target for GDP dissociation inhibitor. Mol Biol Cell 1998;9:3241–3257.

20. Junutula JR, de Maziere AM, Peden AA, Ervin KE, Advani RJ, van Dijk

SM, Klumperman J, Scheller RH. Rab14 is involved in membrane

trafficking between the Golgi complex and endosomes. Mol Biol Cell

2004;15:2218–2229.

21. Stenmark H, Parton RG, Steele-Mortimer O, Lutcke A, Gruenberg J,

Zerial M. Inhibition of rab5 GTPase activity stimulates membrane fusion

in endocytosis. EMBO J 1994;13:1287–1296.

22. Riederer MA, Soldati T, Shapiro AD, Lin J, Pfeffer SR. Lysosome

biogenesis requires Rab9 function and receptor recycling from endo-

somes to the trans-Golgi network. J Cell Biol 1994;125:573–582.

23. Lock JG, Stow JL. Rab11 in recycling endosomes regulates the sorting

and basolateral transport of E-cadherin.Mol Biol Cell 2005;16:1744–1755.

24. Keller P, Toomre D, Diaz E, White J, Simons K. Multicolour imaging of

post-Golgi sorting and trafficking in live cells. Nat Cell Biol 2001;3:140–149.

25. Ang AL, Taguchi T, Francis S, Folsch H, Murrells LJ, Pypaert M,Warren

G, Mellman I. Recycling endosomes can serve as intermediates during

transport from the Golgi to the plasmamembrane of MDCK cells. J Cell

Biol 2004;167:531–543.

26. Kauppi M, Simonsen A, Bremnes B, Vieira A, Callaghan J, Stenmark H,

Olkkonen VM. The small GTPase Rab22 interacts with EEA1 and

controls endosomalmembrane trafficking. J Cell Sci 2002;115:899–911.

27. Yi Z, Yokota H, Torii S, Aoki T, Hosaka M, Zhao S, Takata K, Takeuchi T,

Izumi T. The Rab27a/granuphilin complex regulates the exocytosis of

insulin-containing dense-core granules. Mol Cell Biol 2002;22:

1858–1867.

28. Griffiths G, Pfeiffer S, Simons K, Matlin K. Exit of newly synthesized

membrane proteins from the trans cisterna of the Golgi complex to the

plasma membrane. J Cell Biol 1985;101:949–964.

29. Matlin KS, Simons K. Sorting of an apical plasma membrane glycopro-

tein occurs before it reaches the cell surface in cultured epithelial cells.

J Cell Biol 1984;99:2131–2139.

30. Maisner A, Zimmer G, Liszewski MK, Lublin DM, Atkinson JP,

Herrler G. Membrane cofactor protein (CD46) is a basolateral protein

that is not endocytosed. Importance of the tetrapeptide FTSL at the

carboxyl terminus. J Biol Chem 1997;272:20793–20799.

31. Folsch H, Ohno H, Bonifacino JS, Mellman I. A novel clathrin adaptor

complex mediates basolateral targeting in polarized epithelial cells. Cell

1999;99:189–198.

32. Roh MH, Fan S, Liu CJ, Margolis B. The Crumbs3-Pals1 complex

participates in the establishment of polarity in mammalian epithelial

cells. J Cell Sci 2003;116:2895–2906.

33. Lapierre LA, Dorn MC, Zimmerman CF, Navarre J, Burnette JO,

Goldenring JR. Rab11b resides in a vesicular compartment distinct

from Rab11a in parietal cells and other epithelial cells. Exp Cell Res

2003;290:322–331.

34. Futter CE, Connolly CN, Cutler DF, Hopkins CR. Newly synthesized

transferrin receptors can be detected in the endosome before they

appear on the cell surface. J Biol Chem 1995;270:10999–11003.

35. Vitale G, Rybin V, Christoforidis S, Thornqvist P, McCaffrey M,

Stenmark H, Zerial M. Distinct Rab-binding domains mediate the inter-

action of Rabaptin-5 with GTP-bound Rab4 and Rab5. EMBO J 1998;

17:1941–1951.

36. Rink J, Ghigo E, Kalaidzidis Y, Zerial M. Rab conversion as a mecha-

nism of progression from early to late endosomes. Cell 2005;122:

735–749.

37. Ortiz D, Medkova M, Walch-Solimena C, Novick P. Ypt32 recruits the

Sec4p guanine nucleotide exchange factor, Sec2p, to secretory

vesicles; evidence for a Rab cascade in yeast. J Cell Biol 2002;157:

1005–1015.

38. Nuoffer C, Davidson HW, Matteson J, Meinkoth J, Balch WE. A

GDP-bound of rab1 inhibits protein export from the endoplasmic

reticulum and transport between Golgi compartments. J Cell Biol 1994;

125:225–237.

39. Barral DC, Ramalho JS, Anders R, Hume AN, Knapton HJ, Tolmachova

T, Collinson LM, Goulding D, Authi KS, Seabra MC. Functional redun-

dancy of Rab27 proteins and the pathogenesis of Griscelli syndrome.

J Clin Invest 2002;110:247–257.

40. Schluter OM, Schmitz F, Jahn R, Rosenmund C, Sudhof TC. A

complete genetic analysis of neuronal Rab3 function. J Neurosci 2004;

24:6629–6637.

41. Wang X, Kumar R, Navarre J, Casanova JE, Goldenring JR. Regulation

of vesicle trafficking in madin-darby canine kidney cells by Rab11a and

Rab25. J Biol Chem 2000;275:29138–29146.

42. Pralle A, Keller P, Florin EL, Simons K, Horber JK. Sphingolipid-cholesterol

rafts diffuse as small entities in the plasma membrane of mammalian

cells. J Cell Biol 2000;148:997–1008.

43. Ehehalt R, Keller P, Haass C, Thiele C, Simons K. Amyloidogenic

processing of the Alzheimer beta-amyloid precursor protein depends

on lipid rafts. J Cell Biol 2003;160:113–123.

44. Meder D, Moreno MJ, Verkade P, Vaz WL, Simons K. Phase coexis-

tence and connectivity in the apical membrane of polarized epithelial

cells. Proc Natl Acad Sci U S A 2006;103:329–334.

45. Schuck S, Manninen A, Honsho M, Fullekrug J, Simons K. Generation

of single and double knockdowns in polarized epithelial cells by

retrovirus-mediated RNA interference. Proc Natl Acad Sci U S A 2004;

101:4912–4917.

60 Traffic 2007; 8: 47–60

Schuck et al.