Regulation of WASH-Dependent ActinPolymerization and Protein Traffickingby UbiquitinationYi-Heng Hao,1,7 Jennifer M. Doyle,1,7 Saumya Ramanathan,1 Timothy S. Gomez,6 Da Jia,2 Ming Xu,3 Zhijian J. Chen,3,4

Daniel D. Billadeau,6 Michael K. Rosen,2,4 and Patrick Ryan Potts1,5,*1Department of Physiology2Department of Biophysics3Department of Molecular Biology4Howard Hughes Medical Institute5Department of Pharmacology

UT Southwestern Dallas, TX 75390, USA6Department of Immunology and Division of Oncology Research, Mayo Clinic, Rochester, MN 55905, USA7These authors contributed equally

*Correspondence: ryan.potts@utsouthwestern.edu

http://dx.doi.org/10.1016/j.cell.2013.01.051

SUMMARY

Endosomal protein trafficking is an essential cellularprocess that is deregulated in several diseases andtargeted by pathogens. Here, we describe a role forubiquitination in this process. We find that the E3RING ubiquitin ligase, MAGE-L2-TRIM27, localizesto endosomes through interactions with the retromercomplex. Knockdown of MAGE-L2-TRIM27 or theUbe2O E2 ubiquitin-conjugating enzyme signifi-cantly impaired retromer-mediated transport. Wefurther demonstrate that MAGE-L2-TRIM27 ubiquitinligase activity is required for nucleation of endosomalF-actin by the WASH regulatory complex, a knownregulator of retromer-mediated transport. Mecha-nistic studies showed that MAGE-L2-TRIM27 facili-tates K63-linked ubiquitination of WASH K220.Significantly, disruption of WASH ubiquitinationimpaired endosomal F-actin nucleation and retro-mer-dependent transport. These findings providea cellular and molecular function for MAGE-L2-TRIM27 in retrograde transport, including an unap-preciated role of K63-linked ubiquitination andidentification of an activating signal of the WASHregulatory complex.

INTRODUCTION

Endosomal protein recycling pathways facilitate the transfer of

membrane proteins from early and late endosomes back to the

trans-Golgi network (TGN) or plasma membrane (Bonifacino

and Rojas, 2006). In doing so, these pathways generally function

to prevent lysosomal delivery and degradation of membrane

proteins. Endosome-to-Golgi transport, referred to as retro-

grade transport, is an important cellular process that facilitates

the recycling of a variety of proteins, including sorting receptors

(such as CI-M6PR), SNARE membrane fusion proteins, mem-

brane receptors, metabolite transporters, and several proteins

that undergo polarized localization/secretion (Bonifacino and

Rojas, 2006; Johannes and Popoff, 2008). Importantly, retro-

grade transport has been implicated in a number of different

human pathologies. Endosome-to-Golgi transport is essential

for cellular entry of pathogenic toxins, such as Shiga, cholera,

and ricin, as well as viral pathogens such as HIV (Brass et al.,

2008; Sandvig and van Deurs, 2005). Furthermore, components

of the retrograde transport pathway are downregulated in Alz-

heimer’s disease and upregulated in cancer (Scott et al., 2009;

Small, 2008).

Recent studies have started to define themolecular machinery

crucial for different aspects of this process, including cargo

recognition, endosomal membrane budding, tubulation and

scission, and vesicle transport, tethering, and fusion at the

TGN (Bonifacino and Rojas, 2006; Cullen and Korswagen,

2012). One critical component is the retromer protein complex

that consists of VPS26, VPS29, and VPS35 and functions to

recognize retrograde cargo on endosomes (Bonifacino and Hur-

ley, 2008). Another essential factor in retrograde transport is

WASH. WASH is a member of the Wiskott-Aldrich syndrome

protein (WASP) family consisting of WASP/N-WASP, WAVE,

WHAMM, JMY, and WASH (Campellone and Welch, 2010).

Like other WASP family members, WASH contains a carboxy-

terminal VCA (verprolin homologous or WH2, central hydro-

phobic, and acidic) motif that binds to actin and the Arp2/3

complex to stimulate actin filament nucleation (Derivery et al.,

2009; Duleh and Welch, 2010; Jia et al., 2010; Linardopoulou

et al., 2007; Liu et al., 2009). Recent studies have demonstrated

thatWASH is endosomal-associated, exists in amacromolecular

complex termed the WASH regulatory complex (SHRC) and

functions downstream of the retromer complex to facilitate

endosome-to-Golgi transport (Derivery et al., 2009; Duleh and

Welch, 2010; Gomez and Billadeau, 2009; Linardopoulou et al.,

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. 1051

2007). Importantly, retromer and SHRC have been implicated

in other endosomal recycling pathways, including endosome-

to-plasma membrane sorting of integrins (Zech et al., 2011).

SHRC consists of at least five core factors, CCDC53, FAM21,

SWIP, Strumpellin, and WASH (Derivery et al., 2009; Gomez

and Billadeau, 2009; Jia et al., 2010). The SHRC is recruited to

early endosomes by multivalent interactions between repeat

elements in FAM21 and the retromer subunit VPS35 (Harbour

et al., 2012; Jia et al., 2012). WASH is essential for Arp2/3-

induced F-actin accumulation on endosomes (Derivery et al.,

2009; Gomez et al., 2012). With WASH suppression, scission

of membrane tubules emanating from endosomes was sug-

gested to be impaired (Derivery et al., 2009; Gomez and Billa-

deau, 2009) and knockout of WASH in MEFs resulted in the

collapse of the early endosomal and lysosomal networks

(Gomez et al., 2012).

Nucleation of F-actin byWASP familymembers is regulated by

a variety of signaling events that converge on VCA motif activa-

tion/exposure (Padrick and Rosen, 2010). The VCA motifs of

WASP/N-WASP,WAVE, andWASHare autoinhibited through in-

tramolecular and intermolecular interactions (Chen et al., 2010;

Ismail et al., 2009; Jia et al., 2010; Kim et al., 2000; Miki et al.,

1998). Several signaling molecules have been identified to

promote the activation/exposure of WASP/N-WASP and WAVE

VCA motifs to allow timed and localized F-actin nucleation,

including small GTPases Cdc42 and Rac1, PIP2 and PIP3 phos-

pholipids, and phosphorylation by the Src, Abl, and Cdk family

of kinases (Eden et al., 2002; Ismail et al., 2009; Kim et al.,

2000; Lebensohn and Kirschner, 2009; Miki et al., 1998; Padrick

and Rosen, 2010). Contrary to WASP/N-WASP and WAVE,

the mechanisms regulating WASH activation have been more

elusive. The Rho GTPase has been genetically linked to activa-

tion of WASH in Drosophila (Liu et al., 2009) but is not sufficient

to directly activate human WASH in vitro (Jia et al., 2010).

Ubiquitination is a posttranslational modification in which a

small 76 amino acid protein is covalently attached to lysine resi-

dues in substrate proteins through a three step E1, E2, and E3

enzymatic cascade (Fang and Weissman, 2004). Ubiquitination

can have pleiotropic effects on its substrates depending on the

length and type of ubiquitin chains. K48-linked ubiquitin chains

typically target proteins for degradation by the 26S proteasome

(Bochtler et al., 1999). However, K63-linked ubiquitination typi-

cally acts as a signaling event tomodify function, such as altering

protein-protein interactions, protein conformations, or targeting

proteins for lysosomal delivery (Sun and Chen, 2004). Recently,

we identified a class of proteins that bind to and enhance the

activity of E3 RING ubiquitin ligases (Doyle et al., 2010). These

ubiquitin ligase enhancers are known as melanoma antigen

(MAGE) genes and comprise a family of over 50 unique human

genes (Chomez et al., 2001). Although the biochemical function

of MAGE proteins in ubiquitination has been elucidated, the

cellular processes in which specific MAGE proteins act are

unclear. Here, we investigate the cellular function of MAGE-L2,

a paternally imprinted gene that is abundantly expressed in the

brain, maps to the Prader-Willi syndrome deletion locus, and is

implicated in normal circadian rhythm and the hypothalamic-

endocrine axis (Bischof et al., 2007; Boccaccio et al., 1999; Ko-

zlov et al., 2007; Tennese and Wevrick, 2011).

1052 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

RESULTS

Identification of MAGE-L2 E3 RING Ubiquitin LigasePartner and Subcellular LocalizationTo determine the specific E3 RING ubiquitin ligase partner of

MAGE-L2 and gain insight into its cellular function, we exam-

ined MAGE-L2 binding partners by tandem affinity purification

(TAP) coupled to mass spectrometry. The E3 RING ubiquitin

ligase TRIM27 was identified as a major binding partner of

MAGE-L2 (Figure 1A). Binding of MAGE-L2 and TRIM27 was

confirmed by reciprocal coimmunoprecipitation experiments

(Figures 1B and S1C available online). Moreover, in vitro trans-

lated MAGE-L2 bound recombinant GST-TRIM27 but not GST

alone, indicating the interaction between the two proteins is

direct (Figure 1C). To confirm the association of MAGE-L2 and

TRIM27, we stably expressed GFP-MAGE-L2 and mCherry-

TRIM27 in U2OS cells and examined their colocalization by

live-cell microscopy. We found that the two proteins colocalized

in discrete cytoplasmic puncta (Figure 1D), as well as in a

smaller nuclear pool (data not shown). These findings suggest

that TRIM27 binds MAGE-L2 and that the MAGE-L2-TRIM27

ubiquitin ligase complex localizes to discrete cytoplasmic

structures.

MAGE-L2-TRIM27 Binds and Localizes to Retromer-Containing EndosomesTo determine the identity of the MAGE-L2-TRIM27 structures,

we reanalyzed our mass spectrometry data of MAGE-L2 inter-

acting proteins for clues. Interestingly, MAGE-L2 was identified

to interact with VPS35 and VPS26 (Figure 1A), two components

of the endosomal retromer complex. mCherry-TRIM27 colocal-

ized with GFP-tagged VPS35, VPS29, and VPS26, as well as

the retromer-associated SHRC proteins, WASH and FAM21

(Figure S1A). Furthermore, costaining for endogenous VPS35

and TRIM27 clearly identified TRIM27 cytoplasmic structures

as retromer-positive endosomes (Figure 1E). Unfortunately, we

were unable to observe endogenous MAGE-L2 due to the lack

of specific, high-quality antibodies (data not shown). More

detailed analysis of TRIM27 localization revealed its localization

to numerous tubular structures emanating from endosomes (Fig-

ure 1F), a property shared with the retromer complex (Arighi

et al., 2004; Bonifacino and Hurley, 2008). These findings sug-

gest that MAGE-L2-TRIM27 localizes to the retromer-positive

subset of endosomes.

Next, we confirmed the interaction between MAGE-L2-

TRIM27 and the retromer complex initially observed by mass

spectrometry. MAGE-L2 and TRIM27 coimmunoprecipitated

with all three components of the retromer complex (Figures

S1B and S1C and data not shown) and MAGE-L2 directly bound

recombinant GST-VPS35 in vitro through itsWH-Bmotif (Figures

S1D, S1G, and S1H). In addition, MAGE-L2 interaction with

VPS35 did not impair VPS35 binding to the SHRC component

FAM21 (Figure S1I). Furthermore, MAGE-L2-VPS35 interaction

is functionally important because knockdown of VPS35 dramat-

ically inhibited MAGE-L2 and TRIM27 endosomal localization

(Figures S1E and S1F and data not shown). These findings sug-

gest that VPS35 recruits MAGE-L2-TRIM27 to retromer-positive

endosomes by binding to MAGE-L2.

– VPS26

– VPS35– MAGE-L2– TRIM27

kDa250 –

150 –

100 –

75 –

50 –

37 –

25 –

20 –

TAP

-Vec

tor

TAP

-MA

GE

-L2A

Myc

-M

AG

E-L

2

Myc

-Ve

ctor

– Myc-MAGE-L2

EndogenousTRIM27

WC

L

B

IP: A

nti-Myc

20%

Inpu

t

GS

T

GS

T-TR

IM27

– Myc-MAGE-L2

C

D

mC

h-TR

IM27

GFP

-MA

GE

-L2

Ove

rlay

+ D

NA

TRIM27VPS35 Overlay + DNA

F

mCh-TRIM27

E

75 –

50 –

75 –

50 –

yz

xz

Rr = 0.7

Rr = 0.8

**

–

EndogenousTRIM27–

Figure 1. MAGE-L2 Binds TRIM27 and Localizes to Retromer-Positive Endosomes

(A) TAP-Vector or TAP-MAGE-L2 were isolated from HEK293 stable cell lines, separated by SDS-PAGE, Coomassie stained, and the identity of specific bands

was determined by LC-MS/MS. Asterisks indicate Usp7.

(B) The indicated proteins were expressed in cells for 48 hr, anti-Myc IP was performed, and proteins were detected by western blot (WB).

(C) Binding of Myc-MAGE-L2 to recombinant GST-TRIM27 or GST alone was determined by GST pull-down assays and anti-Myc immunoblotting.

(D) Stably expressed GFP-MAGE-L2 and mCherry-TRIM27 colocalize in specific cytoplasmic puncta.

(E) U2OS cells were stained for endogenous VPS35 (red), endogenous TRIM27 (green), and DNA (blue). XZ and YZ projection stacks are shown.

(F) Live-cell imaging of stably expressed mCherry-TRIM27 shows localization to tubule-like protrusions (yellow arrowheads) from endosomes.

Pearson’s correlation coefficients (Rr) are shown. Scale bars, 20 mm. See also Figure S1.

MAGE-L2-TRIM27 Is Required for Endosomal ProteinRecyclingWe next assessed whether MAGE-L2-TRIM27 participates in

endosome-to-Golgi transport. Two independent siRNAs tar-

geting MAGE-L2 or TRIM27 were identified that significantly

reduced their target protein’s levels but had no effects on the

levels of other known essential factors required for retrograde

transport (Figures S2A–S2C). Multiple siRNAs targeting MAGE-

L2 or TRIM27 resulted in impaired CI-M6PR and TGN46 traf-

ficking to a degree similar to VPS35-RNAi (Figures 2A–2C and

S2D). Importantly, the overall organization of the TGN was unaf-

fected in MAGE-L2- or TRIM27-RNAi cells (Figure S2E). The

steady-state defects in CI-M6PR localization were corroborated

by examining transport of a small pool of surface localized

CI-M6PR to the TGN in MAGE-L2- and TRIM27-RNAi cells.

MAGE-L2- or TRIM27-RNAi impaired transport of surface-

labeled CI-M6PR to the TGN (Figure 2D). Importantly, the dis-

persed steady-state or surface-labeled CI-M6PR in TRIM27- or

MAGE-L2-RNAi cells colocalized with the endosomal marker

EEA1 (Figures 2E and S2F). Furthermore, CI-M6PR protein

levels were reduced in MAGE-L2- and TRIM27-RNAi cells and

Cathepsin D processing and trafficking were impaired (Fig-

ure 2F). Finally, overexpression of MAGE-L2 in combination

with TRIM27 increased CI-M6PR endosomal retrieval kinetics

(Figure S2G). These results suggest that MAGE-L2-TRIM27 is

required for endosome-to-Golgi retrograde transport.

To extend our findings, we examined the requirement of

TRIM27 on the trafficking of two additional physiologically and

pathologically relevant substrates of the retromer and SHRC

complexes. First, knockdown of TRIM27 significantly impaired

trafficking of the retromer cargo cholera toxin subunit B (CTxB)

to the TGN (Figure 2G). However, the trafficking of the SHRC-

independent cargo Transferrin receptor to perinuclear recycling

endosomeswas unaltered in TRIM27-RNAi cells (Figure 2G) (Du-

leh and Welch, 2010; Gomez and Billadeau, 2009; Gomez et al.,

2012). In addition, the retromer and SHRC complexes have been

implicated in the recycling of endosomal proteins to the plasma

membrane, including integrins (Duleh and Welch, 2012; Zech

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. 1053

siControl siTRIM27 #2

siMAGE-L2 #3 siVPS35

CI-M

6PR

+ D

NA

A

siCo

ntro

l

siTR

IM27

#2

siM

AGE-

L2 #

3

siVP

S35

– pCatD

– APC2

F

CI-M

6PR

EEA

1O

verla

y +

DN

A

siControl siTRIM27 #2 siMAGE-L2 #3E

908070605040302010

0

siCo

ntro

l

siTR

IM27

#1

siTR

IM27

#2

siM

AGE-

L2 #

1

siM

AGE-

L2 #

3

siVP

S35

Cells

with

Dis

pers

ed

CI-M

6PR

(%)

B*

**

**

siCo

ntro

l

siTR

IM27

#2

siM

AGE-

L2 #

3

siVP

S35

Cells

with

Juxt

anuc

lear

Inte

rnal

ized

CI-M

6PR

(%)

D

0102030405060708090

100

*

*

*

siControl siTRIM27 #2

siMAGE-L2 #3 siVPS35

TGN

46 +

DN

A

0102030405060708090

100

0 20 40 60Cells

with

Per

inuc

lear

Loc

aliz

ed

CTx

B or

Tra

nsfe

rrin

(%)

Time (min)

siControl - CTxBsiTRIM27 - CTxBsiControl - TfsiTRIM27 - Tf

siCo

ntro

lsi

TRIM

27 #

2si

MAG

E-L2

#3

Integrin α5Integrin α5 + EEA1

+ DNA

Mock siTRIM27 #2siMAGE-L2 #3 siFAM21

86.8 ± 19.5 30.6 ± 4.656.1 ± 6.6 22.7 ± 6.7

***

*

C

GI

J

Surface Integrin α5 (RFU)

Cell

Num

ber

siControlsiMAGE-L2siTRIM27

H

Cell

Lysa

teM

edia

– pCatD– iCatD

– CI-M6PR

– GAPDH

– mCatD

yz

xzRr = 0.1 Rr = 0.7 Rr = 0.6

Figure 2. MAGE-L2-TRIM27 Is Required for Endosomal Protein Recycling

(A) Cells were treated with the indicated siRNAs for 72 hr and stained for CI-M6PR (green) and DNA (blue).

(B) Quantitation of cells shown in (A). Compact juxtanuclear or dispersed CI-M6PR was scored and the percentage of cells with dispersed CI-M6PR is shown.

(C) Cells were treated with the indicated siRNAs for 72 hr and stained for TGN46 (red) and DNA (blue).

(D) Cells were treated with the indicated siRNAs. Cell surface CI-M6PR was labeled with anti-CI-M6PR antibody for one hour before imaging the pool of

internalized cell surface-labeled CI-M6PR.

(E) Cells were treated with the indicated siRNAs and stained for CI-M6PR (green), EEA1 (red), and DNA (blue). XZ and YZ projection stacks and Pearson’s

correlation coefficients (Rr) are shown.

(F) Cells were treated with the indicated siRNAs for 72 hr and cell lysates andmedia were analyzed by immunoblotting. pCatD represents unprocessed pro-CatD,

iCatD indicates intermediate processed CatD, and mCatD represents fully matured CatD.

(G) Cells were treated with the indicated siRNAs for 72 hr, and transport of CTxB-488 or Tf-568 was determined at the indicated times.

(legend continued on next page)

1054 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

et al., 2011). Therefore, we examined whether MAGE-L2 and

TRIM27 function extends beyond endosome-to-Golgi trafficking

by examining their contribution to the recycling of integrin a5 to

the plasma membrane. Knockdown of MAGE-L2 or TRIM27

significantly decreased cell surface integrin a5 (Figure 2H) and

increased the intracellular pool of integrin a5 in EEA1-negative

endosomes (Figure 2I). Importantly, this decrease in cell surface

integrin a5 in MAGE-L2- or TRIM27-RNAi cells is functionally

significant because invasion of cells through matrigel, a process

regulated by integrins, is significantly impaired in MAGE-L2- or

TRIM27-RNAi cells (Figure 2J). These results suggest that

MAGE-L2-TRIM27 functions with the retromer and SHRC com-

plexes in both endosome-to-Golgi and endosome-to-plasma

membrane protein recycling.

Ube2OE2Ubiquitin-Conjugating Enzyme Is Required forRetrograde TransportNext, we determined whether the requirement for MAGE-L2-

TRIM27 in retrograde transport involved its ubiquitin ligase

activity by utilizing a TRIM27 RING mutant lacking E3 ubiquitin

ligase activity. Unlike wild-type TRIM27, expression of an RNAi

resistant TRIM27 RING mutant was unable to rescue retrograde

transport of CI-M6PR in TRIM27-RNAi cells (Figures 3A and

S3A). Importantly, the TRIM27 RING mutant localized similarly

to the wild-type protein (Figure S3B). These results suggest

that the ubiquitin ligase activity of MAGE-L2-TRIM27 is impor-

tant for proper retrograde transport.

Next, we examined which of the more than 35 different

E2 ubiquitin-conjugating enzymes functions with MAGE-L2-

TRIM27 to facilitate retrograde endosome-to-Golgi trafficking

of CI-M6PR. Of the 35 E2 enzymes examined, knockdown of

only Ube2O and Ube2L6 resulted in dramatic alteration of CI-

M6PR localization (Figures 3B and S3C, and data not shown).

Ube2O-RNAi showed similar penetrance as TRIM27-RNAi (Fig-

ures 3B and 3C), resulted in CI-M6PR dispersion to EEA1-

positive endosomes (Figure 3D), and reduced total CI-M6PR

(Figure 3E). However, Ube2L6-RNAi did not redistribute CI-

M6PR to EEA-positive endosomes (Figure 3D) or promote lyso-

somal degradation of CI-M6PR (Figure 3E) but rather disrupted

TGN organization (Figure S3D). Notably, Ube2O, but not

Ube2L6, was also identified in a yeast-2-hybrid screen for E2

enzymes interacting with TRIM27 (Markson et al., 2009). These

results suggest that the Ube2O E2 ubiquitin-conjugating enzyme

is the physiologically relevant E2 enzyme functioning with

MAGE-L2-TRIM27 in supporting retrograde transport.

Retrograde Transport Requires K63-Linked UbiquitinChainsWe next investigated whether retrograde transport was depen-

dent on ubiquitin, and if so, which type of polyubiquitin chain.

To do so, we utilized the previously developed system (Xu

et al., 2009) to inducibly knockdown ubiquitin by addition of

(H) Cells were treated with the indicated siRNAs for 72 hr and surface integrin a5

(I) Cells were treated with the indicated siRNAs for 72 hr and immunostained wit

(J) Invasive potential of cells treated with the indicated siRNAs was assayed on

crystal violet and quantitated.

All data are represented as the mean ±SD. Asterisk indicates p < 0.05. Scale ba

tetracycline and replace it with a ubiquitin variant that is unable

to produce a specific ubiquitin chain, namely ubiquitin in which

lysine 48 or lysine 63 has been mutated to arginine (K48R or

K63R, respectively; Figures S3E and S3F). We found that deple-

tion of ubiquitin from cells dramatically affected CI-M6PR

trafficking (Figures 3F and S3G). Addition of wild-type or K48R

ubiquitin completely rescued CI-M6PR localization (Figures 3F

and S3G), suggesting that ubiquitin is required for retrograde

trafficking, but K48-linked ubiquitin chains are not. In contrast,

depletion of K63-ubiquitin chains dramatically blocked CI-

M6PR TGN localization and relocalized it to EEA1-positive endo-

somes (Figures 3F and S3G and data not shown) but had no

significant effect on the overall organization of the TGN (Fig-

ure S3H). Similarly to the steady-state behavior, retrograde

transport of cell surface-labeled CI-M6PR to the TGN is also

dependent on K63-linked ubiquitination (Figures 3G and S3I).

Furthermore, depletion of K63-linked ubiquitin chains altered

trafficking of the lysosomal hydrolase Cathepsin D, resulting in

the reduction of mature Cathepsin D, accumulation of pro- and

intermediate-Cathepsin D, and the secretion of pro-Cathepsin

D into the cell culture media (Figure 3H). These results suggest

that K63-linked ubiquitination is important for proper retrograde

transport.

MAGE-L2-TRIM27, Ube2O, and K63-Linked UbiquitinChains Are Required for Efficient Endosomal F-ActinAssemblyNext, we assessed the precise mechanism by which MAGE-L2-

TRIM27, Ube2O, and K63-linked ubiquitination promotes retro-

grade transport. Retrograde transport requires several ordered

steps to facilitate endosome-to-Golgi transport, including endo-

somal localization of the retromer complex and localization and

activation of the SHRC to facilitate endosomal F-actin accumu-

lation by the Arp2/3 complex (Cullen and Korswagen, 2012). We

interrogated several of these specific steps to determine when

ubiquitination may be important. The localization of the retromer

complex to CI-M6PR substrate-containing endosomes was un-

affected by depletion of TRIM27 or K63-linked ubiquitin chains

(Figures S4A and S4B). In addition, the SHRC was still recruited

normally to retromer-positive endosomes in TRIM27-RNAi cells

(Figure S4C). However, knockdown of MAGE-L2 or TRIM27 re-

sulted in reduced endosomal F-actin to a degree similar to

WASH-RNAi (Figures 4A and 4B). Similarly, knockdown of the

physiologically relevant Ube2O E2 enzyme, but not Ube2L6, re-

sulted in a reduced endosomal F-actin (Figures 4A and 4B).

Consistent with reduced endosomal F-actin, knockdown of

MAGE-L2, TRIM27, or Ube2O, but not Ube2L6, significantly

reduced the accumulation of the Arp2/3 complex subunit,

ARPC5, on SHRC-positive endosomes (Figures 4C and 4D).

Notably, there was no defect in total ARPC5 or F-actin levels

(Figure 4D and data not shown). Likewise, depletion of K63-

linked ubiquitin chains resulted in the specific reduction of

was determined by flow cytometry. Grey curve indicates control IgG staining.

h integrin a5 (green), EEA1 (red), and DNA (blue).

matrigel-coated transwell invasion chambers. Invaded cells were stained with

rs, 20 mm. See also Figures S2 and S7.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. 1055

NegativeControls

Ubiquitin E2 Enzymes

0102030405060708090

Moc

k #1

Moc

k #2

siC

ontro

lsi

TRIM

27si

VP

S35

siW

AS

H1

siFA

M21

siU

be2B

siU

be2A

siU

BE

2Csi

Ube

2Q2

siU

be2Q

1si

UE

VLD

siB

IRC

6si

Ube

2Zsi

Ube

2Osi

AK

TIP

siU

be2J

2si

Ube

2J1

siU

be2U

siU

be2H

siU

be2S

siU

be2G

2si

Ube

2G1

siU

be2R

2si

Ube

2R1

siU

be2D

3si

Ube

2D2

siU

be2D

4si

Ube

2D1

siU

be2E

3si

Ube

2E2

siU

be2E

1si

Ube

2L6

siU

be2L

3si

Ube

2Tsi

Ube

2Ksi

Ube

2NL

siU

be2N

siU

be2V

2si

Ube

2V1

siU

be2W

Cel

l with

Dis

pers

ed C

I-M6P

R (%

)

100

PositiveControls

CI-M

6PR

EE

A1

Ove

rlay

+D

NA

siControl siUbe2OsiUbe2L6

Cel

ls w

ith D

ispe

rsed

CI-M

6PR

(%)

siU

be2O

siC

ontro

l

siU

be2L

6

siTR

IM27

#2

9080706050403020100

C D

*

*

*

B

E

0500

10001500200025003000350040004500

siU

be2O

siC

ontro

l

siU

be2L

6

siTR

IM27

#2

CI-M

6PR

Inte

nsity

(RFU

)

* *

80

70

60

50

40

30

20

10

0

siC

ontro

l+

Vect

or

Vect

or

TRIM

27 W

T

TRIM

27R

ING

Mut

.

siTRIM27 #2

Cel

ls w

ith D

ispe

rsed

CI-M

6PR

(%)

A

*

*

*

– pCatD– iCatD– mCatD

–pCatD

50 –

37 –

50 –

– APC2

– Tet + TetU2OS shUb-Ub(K63R)

50 –

Cell Lysate

Media

H

0102030405060708090

100

Cel

ls w

ith D

ispe

rsed

CI-M

6PR

(%)

shU

b

shU

b-U

b(W

T)

shU

b-U

b(K

48R

)

shU

b-U

b(K

63R

)

F

**

– Tet+ Tet

shU

b

shU

b-U

b(W

T)

shU

b-U

b(K

48R

)

shU

b-U

b(K

63R

)

G

*

0102030405060708090

100

*

Cel

ls w

ith J

uxta

nucl

ear

Inte

rnal

ized

CI-M

6PR

(%)

– GAPDH37 –

yz

xzRr = 0.1 Rr = 0.1 Rr = 0.7

Figure 3. Ube2O E2 and K63-Ubiquitin Chains Are Required for Retrograde Transport

(A) Cells were treated with control or TRIM27 siRNAs for 24 hr before transfection of RNAi-resistant wild-type or RING mutant TRIM27. Forty-eight hours after

transfection, cells were stained for CI-M6PR. The percentage of transfected cells with dispersed CI-M6PR is shown.

(B) Cells were treated with siRNAs targeting the indicated E2 enzymes for 72 hr, stained with anti-CI-M6PR, imaged, and quantitated. Dotted line denotes cutoff

that reproducibly indicates CI-M6PR trafficking defect.

(C) Cells were treated with the indicated siRNAs for 72 hr, stained with CI-M6PR, and imaged. The percentage of cells with dispersed CI-M6PR is shown.

(D) Cells were treatedwith the indicated siRNAs for 72 hr and stained for CI-M6PR (green), EEA1 (red), and DNA (blue). XZ and YZ projection stacks and Pearson’s

correlation coefficients (Rr) are shown.

(E) Cells were treated with the indicated siRNAs for 72 hr, stained for CI-M6PR, imaged, and CI-M6PR intensity was determined.

(F) Ubiquitin replacement cell lines were treated with or without tetracycline for 72–96 hr before steady-state CI-M6PR localization was determined by immu-

nostaining. The percentage of cells with dispersed CI-M6PR is shown.

(legend continued on next page)

1056 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

endosomal F-actin and Arp2/3 complex, without affecting total

F-actin, Arp2/3 complex, or SHRC levels (Figures 4E–4H, S3J,

and data not shown). Consistent with the previously described

role of SHRC and endosomal F-actin in membrane tubule

scission (Derivery et al., 2009; Gomez and Billadeau, 2009),

TRIM27-RNAi, shUb-Ub, and shUb-Ub(K63R) cells accumulated

endosomal tubules (Figures S4D and S4E). These results sug-

gest that K63-linked ubiquitination by MAGE-L2-TRIM27 and

Ube2O is required for localization of the Arp2/3 complex to the

SHRC and generation of endosomal F-actin.

Uninhibited Endosomal Actin Assembly RescuesEndosome-to-Golgi Transport Defects of TRIM27-RNAiand K63-Ubiquitin Depleted CellsWe next examined whether the defective endosome-to-Golgi

retrograde trafficking in TRIM27-RNAi or shUb-Ub(K63R) cells

is due to decreased endosomal F-actin. First, we designed a

fusion protein in which endosomal F-actin accumulation could

be induced by an uninhibited WASH VCA motif. WASH VCA

motif was fused to the C-terminal domain of FAM21 (D356N),

which binds VPS35 for endosomal targeting (Figure 5A) (Gomez

and Billadeau, 2009; Jia et al., 2012). Upon expression in cells,

this fusion protein localized to retromer-positive endosomes

(Figure 5A) and increased F-actin accumulation on endo-

somes (Figures 5A and 5B). We next determined whether this

FAM21-WASH-VCA fusion could rescue retrograde transport

in TRIM27-RNAi cells. Indeed, this was the case for both

Cathepsin D trafficking (Figure 5C) and CI-M6PR TGN localiza-

tion (Figures 5D and S4F). Furthermore, the FAM21-WASH-

VCA fusion rescued CI-M6PR trafficking in cells deficient for

K63-linked ubiquitin chains (Figures 5E and S4G). Therefore,

the primary requirement for TRIM27 and K63-linked ubiquitin

chains in retrograde trafficking appears to be for accumulation

of endosomal F-actin.

MAGE-L2-TRIM27UbiquitinatesWASHK220 toPromoteEndosomal F-Actin AssemblyGeneration of endosomal F-actin is facilitated by the SHRC,

which binds the Arp2/3 complex and actin through a conserved

VCA domain in WASH (Derivery et al., 2009; Gomez and Billa-

deau, 2009). However, the WASH VCA domain exists in an

autoinhibited state that must be relieved before nucleation of

F-actin by the Arp2/3 complex can be achieved (Jia et al.,

2010). Therefore, we speculated that MAGE-L2-TRIM27 may

facilitate WASH-VCA exposure, Arp2/3 complex and actin bind-

ing, and consequent F-actin nucleation on endosomes through

targeted ubiquitination of the SHRC. Based on the highly homol-

ogous WAVE regulatory complex, regulators of SHRC activation

are predicted to act on WASH itself or SWIP (Chen et al., 2010;

Jia et al., 2010). Although we were unable to detect any ubiquiti-

nation of endogenous SWIP (Figure S5A), endogenous WASH

(G) Ubiquitin replacement cells were treated with or without tetracycline for 72–96

hour and internalized cell surface-labeled CI-M6PR was imaged. CI-M6PR loc

internalized CI-M6PR is shown.

(H) Cells were treated with the indicated siRNAs for 72 hr before Cathepsin D pro

unprocessed pro-CatD, iCatD indicates intermediate processed CatD, and mCa

Data are represented as the mean ±SD. Asterisk indicates p < 0.05. Scale bars,

was highly ubiquitinated (Figure 6A). WASH ubiquitination was

TRIM27-dependent, as knockdown of TRIM27 dramatically

reduced endogenous WASH ubiquitination (Figure 6B). Further-

more, WASH ubiquitination was K63 linked, as WASH was

unable to be polyubiquitinated by K63R ubiquitin (Figure 6C).

These findings suggest that TRIM27 is required for K63-linked

ubiquitination of WASH.

Next we investigated the importance of WASH ubiquitination.

To do so, we first determined the specific lysine in WASH that is

conjugated to K63-linked ubiquitin chains. A single lysine, K220,

in WASH had been identified in global proteomics studies to be

ubiquitinated (Kim et al., 2011). Significantly, K220 is a highly

conserved residue in species that express TRIM27 orthologs

(Figure S5D). This residue is located in a region of WASH that

is analogous to the ‘‘meander’’ region of WAVE, which is known

to make contacts necessary for intracomplex inhibition in the

WAVE regulatory complex (Chen et al., 2010). Therefore, we

examined whether WASH ubiquitination was dependent on

K220. Endogenous WASH was knocked down by RNAi and

YFP-tagged wild-type or K220R WASH was re-expressed to

insure integration of the re-expressed WASH into the SHRC.

Unlike YFP-WASH wild-type, YFP-WASH K220R failed to be

ubiquitinated (Figure 6C), suggesting that WASH ubiquitination

is dependent on K220. We next examined whether WASH

ubiquitination is important for retrograde transport. Using our

re-expression system, YFP-WASH K220R, unlike wild-type

YFP-WASH, could not support proper retrograde transport

resulting in CI-M6PR dispersion (Figures 6D and S5B) and

degradation (Figures 6E and S5B) and Cathepsin D secretion

(Figure 6F). These results suggest that K63-linked ubiquitination

ofWASHK220 by TRIM27 is required forWASH function in retro-

grade transport.

We next examined whether WASH K220 ubiquitination may

act as a signal to relieve WASH autoinhibition and promote its

activity toward the Arp2/3 complex. As previously reported

(Derivery et al., 2009), knockdown of WASH resulted in reduced

endosomal Arp2/3 complex localization (Figures 6G and S5C).

Unlike wild-type YFP-WASH, re-expression of YFP-WASH

K220R was unable to rescue proper endosomal Arp2/3 complex

localization (Figures 6G and S5C). Importantly, YFP-WASH

K220R localized properly to endosomes indicating that it still

incorporated into the SHRC (Figure S5C). These results suggest

that ubiquitination of WASH K220 is required for WASH activa-

tion and Arp2/3 complex endosomal localization.

In Vitro Reconstitution of MAGE-L2-TRIM27 and K63-Ubiquitin-Dependent SHRC ActivityFinally, we developed an in vitro reconstitution system to directly

test whether ubiquitination of WASH is required for its actin

assembling activity. To do so, we reconstituted WASH knock-

out MEFs (Gomez et al., 2012) with stable expression of

hr. Cell surface CI-M6PR was then labeled with anti-CI-M6PR antibody for one

alization was determined and the percentage of cells showing juxtanuclear

cessing and secretion was determined by immunoblotting. pCatD represents

tD represents fully matured CatD.

20 mm. See also Figure S3.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. 1057

siControl siTRIM27 #2 siMAGE-L2 #3

AR

PC

5FA

M21

Ove

rlay

+ D

NA

-Tet +TetshUb-Ub(K63R)

AR

PC

5FA

M21

Ove

rlay

+ D

NA

siControl

End

osom

al

Tota

l

0200400600800

1000120014001600

-Tet +Tet

AR

PC

5 In

tens

ity (R

FU)

End

osom

al

Tota

l

End

osom

al

Tota

l

shUb-Ub(K63R)

F-A

ctin

Ove

rlay

+ D

NA

VP

S35

0

250

500

7501000

1250

1500

1750

2000

F-A

ctin

Inte

nsity

on

VP

S35

End

osom

es (R

FU)

siWASHsiControl siTRIM27 #2 siUbe2O

siC

ontro

l

siTR

IM27

#2

siU

be2L

6

siW

AS

H

siM

AG

E-L

2 #3

siU

be2O

0

200

400600

8001000

12001400

1600

AR

PC

5 In

tens

ity (R

FU)

siTRIM27#2

End

osom

al

Tota

l

siMAGE-L2 #3

End

osom

al

Tota

lsiUbe2O

End

osom

al

Tota

l

siUbe2L6

End

osom

al

Tota

l

siUbe2O

A B

C D

E F G

F-A

ctin

WA

SH

-Tet +TetshUb-Ub(K63R)

0

500

1000

1500

2000

2500

3000

F-A

ctin

Inte

nsity

on

WA

SH

End

osom

es (R

FU)

-Tet

+Tet

shUb-Ub(K63R)

H

*

*

**

*

* * * *

yz

xz

yz

xz

yz

xz

Ove

rlay

+ D

NA

yz

xz

Figure 4. MAGE-L2-TRIM27, Ube2O, and K63-Linked Ubiquitination Are Required for Endosomal F-Actin Assembly

(A) Cells were treated with the indicated siRNAs for 72 hr before staining VPS35 (green), F-actin (red), and DNA (blue). XZ and YZ projection stacks are shown.

(B) F-actin intensity on VPS35-positive endosomes of cells described in (A).

(C) Cells were treated with the indicated siRNAs for 72 hr before staining ARPC5 (green), FAM21 (red), and DNA (blue). XZ and YZ projection stacks are shown.

(D) Cells treated as in (C) were imaged and endosomal (solid bars) or total (open bars) ARPC5 levels were determined.

(E) shUb-Ub(K63R) ubiquitin replacement cells were treated with or without tetracycline for 96 hr before staining for WASH (green), F-actin (red), and DNA (blue).

XZ and YZ projection stacks are shown.

(F) Cells described in (E) were imaged, and F-actin intensity on retromer-positive endosomes was quantitated and is shown.

(legend continued on next page)

1058 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

0

500

1000

1500

2000

2500

3000

End

osom

al F

-Act

in In

tens

ity (R

FU)

GFP

GFP

-FA

M21

Δ35

6N

-WA

SH

-VC

A

GFP FAM21Δ356N WASHVCA357 1341 316 468

VPS35

FAM21Δ356N

-WASH-VCAF-Actin

Overlay + DNA

BA

EC

Endosomal Targeting ActinNucleation

GFP-FAM21Δ356N-WASH-VCAYFP-WASH

siTRIM27 #2

TRIM27 –

pCatD(Media)

–+–

–++

+–+

GFP-FAM21Δ356N

-WASH-VCA

–

+––

YFP-WASH –

0102030405060708090

GFP

-FA

M21

Δ35

6N

-WA

SH

-VC

A

GFP

-FA

M21

Δ35

6N

-WA

SH

-VC

A

GFP

GFP

– Tet+ Tet

Cel

ls w

ith D

ispe

rsed

CI-M

6PR

(%)

shUb-Ub(K63R)

*

D

0

10

20

30

40

50

60

70

80C

ells

with

Dis

pers

ed C

I-M6P

R (%

)

siControl siTRIM27 #2

GFP

-FA

M21

Δ35

6N

-WA

SH

-VC

A

GFP

-FA

M21

Δ35

6N

-WA

SH

-VC

A

GFP

GFP

*

–

*

GAPDH(Media) –

Rr=0.6

Figure 5. Uninhibited WASH-VCA Bypasses the Requirement for TRIM27 and K63-Linked Ubiquitin Chains in Retrograde Transport

(A) Schematic of uninhibited, endosomal localized WASH-VCA (top). Cells were transiently transfected for 48 hr before staining GFP-FAM21D356N-WASH-VCA

(green), F-actin (red), VPS35 (cyan), and DNA (blue). Scale bar, 20 mm. Pearson’s correlation coefficient (Rr) is shown.

(B) Cells were transfected with GFP-alone or GFP-FAM21D356N-WASH-VCA and VPS35-localized F-actin intensity was determined.

(C) Cells were treated with control or TRIM27 siRNAs for 24 hr before transfection with control YFP-WASH or constitutively active GFP-FAM21D356N-WASH-VCA.

Forty-eight hours later, the indicated proteins were examined by immunoblotting.

(D) Cells were treated with control or TRIM27 siRNAs for 24 hr before transfection with GFP alone or GFP-FAM21D356N-WASH-VCA. Forty-eight hours after

transfection, cells were immunostained for CI-M6PR and the percentage of transfected cells with dispersed CI-M6PR was determined.

(E) shUb-Ub(K63R) cells were treated with or without tetracycline and transfected with either GFP alone or GFP-FAM21D356N-WASH-VCA. Cells were immu-

nostained after 96 hr for CI-M6PR and the percentage of transfected cells with dispersed CI-M6PR was determined.

Data are represented as the mean ±SD. Asterisk indicates p < 0.05.

HA-GFP-WASH wild-type, DVCA, or K220R. The intact SHRC

was then purified from each of these cell lines by anti-HA chro-

matography to near homogeneity (Figures S6A and S6B). Under

these conditions, very little free WASH is present (Figure S6B).

The activity of the reconstituted SHRCs was first assayed by

examining their capacity to assemble F-actin on beads in cell

lysates (Cory et al., 2003). As expected, SHRC stimulated F-actin

accumulation on beads in amanner dependent on the VCAmotif

(G) shUb-Ub(K63R) ubiquitin replacement cells were treated with or without tetr

(blue). XZ and YZ projection stacks are shown.

(H) Cells treated as in (G) were imaged and endosomal (solid bars) or total (open

Data are represented as the mean ±SD. Asterisk indicates p < 0.05. Scale bars,

ofWASH andwas inhibited by cytochalasin D (Figure 7A). Impor-

tantly, WASH knockout MEFs reconstituted with WASH K220R

mutant did not support actin assembly on beads (Figure 7A). In

addition, the activity of SHRC was dependent on MAGE-L2-

TRIM27, as SHRC isolated from MAGE-L2- or TRIM27-RNAi

cells was significantly less active (Figure 7B). Furthermore, we

examined the activity of purified SHRC in Arp2/3 complex-

dependent pyrene-actin assembly assays. SHRC reconstituted

acycline for 96 hr before staining with ARPC5 (green), FAM21 (red), and DNA

bars) ARPC5 levels were determined.

20 mm. See also Figure S4.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. 1059

– –Myc-Ubiquitin

250 –

150 –

100 –

– –WASH-RNAi

IP: Anti- Myc-Ub

WB: WASH

WCL

– – Myc-Ubiquitin– – TRIM27-RNAi

+ ++ +

++

++BA

D

G

0

500

1000

1500

2000

Endo

som

al A

rpC5

Inte

nsity

(RFU

)

Con

trol

YFP

YFP

-WA

SH

WT

YFP

-WA

SH

K22

0R

shWASH

75 –

50 –

– WASH

– TRIM27

UbiquitinatedWASH

CWT WT Myc-UbiquitinK63R–WT K220R shWASH+YFP-WASHWTWT

250 –

150 –

100 –

– YFP-WASH

UbiquitinatedYFP-WASH

250 –

150 –100 –

75 –

50 –

Myc-Ubiquitin

F

*

*

YFP

-WA

SH

K22

0D

shWASH

WT

K22

0R

– YFP-WASH (WCL)

– pCatD (Media)

YFP-WASH K

220R

WT

– GAPDH (Media)

∆GG– –+

Con

trol

YFP

YFP

-WA

SH

WT

YFP

-WA

SH

K22

0R

shWASH

End

osom

al C

I-M6P

R In

tens

ity (R

FU) 4000

3500

3000

2500

2000

1500

1000

500

0

* *

YFP

-WA

SH

K22

0D

0

10

20

30

40

50

60

70

80

90

Cel

ls w

ith D

ispe

rsed

CI-M

6PR

(%)

*

*

Con

trol

YFP

YFP

-WA

SH

WT

YFP

-WA

SH

K22

0R

shWASH

YFP

-WA

SH

K22

0D

E

Figure 6. WASH K63-Linked Ubiquitination by TRIM27 Is Required for Endosomal F-Actin Nucleation and Retrograde Transport

(A and B) Cells were treated with the indicated siRNAs for 24 hr before transfection of the indicated vectors. Forty-eight hours after plasmid transfection, anti-Myc

IP was performed. Whole-cell lysates (WCL) or anti-Myc IP samples were immunoblotted for WASH and TRIM27.

(C) Cells were transfected with the indicated Myc-ubiquitin vectors and dual-knockdown/re-expression vectors to knockdown endogenous WASH and re-

express YFP-wild-type or K220RWASH. Seventy-two hours after transfection, anti-Myc IPwas performed.Whole-cell lysates (WCL) or anti-Myc IP samples were

immunoblotted for WASH and Myc-ubiquitin.

(D and E) Cells were transfected with the indicated dual-knockdown/re-expression vectors to knockdown endogenous WASH and re-express YFP-WASH

variants. Seventy-two hours after transfection cells were immunostained for CI-M6PR and dispersed CI-M6PR (D) and endosomal CI-M6PR abundance (E) was

determined in transfected cells.

(F) Cells were transfected with the indicated dual-knockdown/re-expression vectors as indicated. Seventy-two hours after transfection, Cathepsin D secretion

into the cell culture media was determined by immunoblotting. Two independent samples from each condition are shown.

(G) Cells described in (D) were immunostained for ARPC5 and the endosomal-localized pool of ARPC5 in transfected cells was quantitated.

Data are represented as the mean ±SD. Asterisk indicates p < 0.05. See also Figure S5.

1060 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

with wild-type, ubiquitinated WASH displayed increased activity

toward the Arp2/3 complex compared to SHRC reconstituted

with nonubiquitinated WASH K220R (Figure 7C). To further

examine the interplay of WASH ubiquitination and activity, we

treated purified wild-type SHRC with the K63-specific deubiqui-

tinating enzyme AMSH. This treatment deconjugated K63-

ubiquitin chains from WASH (Figure 7D) and significantly in-

hibited SHRC activity on beads (Figure 7E) and in pyrene-actin

assembly assays (Figure 7F). These results suggest that K63-

ubiquitination of WASH K220 by MAGE-L2-TRIM27 facilitates

the activation of SHRC in a reversible manner.

To determine if ubiquitination of WASH K220 may facilitate

SHRC activation by disrupting autoinhibitory contacts in the

meander region around WASH K220, we mutated WASH K220

to aspartic acid (K220D) to destabilize autoinhibitory contacts

in this region. Unlike inactive WASH K220R, WASH K220D is

active in vitro (Figures 7A and 7C) and facilitates endosome-

to-Golgi retrograde transport (Figures 6D–6F) and endosomal

Arp2/3 complex localization (Figure 6G) in cells. These results

suggest that destabilizing the WASH meander region around

the ubiquitination site can facilitate SHRC activity independent

of ubiquitination.

DISCUSSION

MAGE proteins are a family of proteins that contain a conserved

domain known as the MAGE homology domain. Recently, we

showed that MAGE proteins function biochemically to bind to

and enhance the activity of E3 RING ubiquitin ligases (Doyle

et al., 2010). In this study we investigated the cellular function

of one specific MAGE protein, MAGE-L2. Proteomic analysis re-

vealed that MAGE-L2 specifically bound the TRIM27 E3 RING

ubiquitin ligase. TRIM27 belongs to a large family of E3 RING

ubiquitin ligases known as tripartite motif (TRIM) proteins.

TRIM27 was originally identified and named Ret finger protein

(RFP), due to its discovery as a gene that undergoes a transloca-

tion event with the Ret tyrosine kinase receptor in thyroid car-

cinomas (Saenko et al., 2003; Takahashi and Cooper, 1987).

Subsequent work has implicated it in several processes, includ-

ing transcriptional regulation, NFkB signaling, CD4+ T cell

homeostasis, and as an oncogene (Cai et al., 2011; Krutzfeldt

et al., 2005; Shimono et al., 2000; Zha et al., 2006; Zoumpoulidou

et al., 2012). It will be of particular interest in the future to deter-

mine if any of these diverse functions of TRIM27 are attributed to

its regulation of WASH and vesicular transport.

Cellular studies revealed that MAGE-L2 and TRIM27 colocal-

ized on cytoplasmic structures that were determined to be

retromer-positive endosomes. Of note, TRIM27 was previously

shown to localize to similar structures that were unidentified at

the time (Harbers et al., 2001). Furthermore, the localization of

TRIM27 was regulated by PKC, JNK, and RAS signaling path-

ways. To determine if any of these signaling pathways may con-

tribute to the upstream regulation of MAGE-L2-TRIM27 and

endosome-to-Golgi retrograde transport, we examined the

effects of short-term inhibition of the PKC, JNK, MEK, and

PI3K signaling pathways on the kinetics of surface CI-M6PR

trafficking to the TGN. Specific inhibition of the JNK signaling

pathway blocked CI-M6PR retrieval to the TGN (Figure S7).

Future studies into the mechanism by which JNK signaling

regulates endosomal protein recycling will be of particular

interest.

Our mechanistic studies uncovered that K63-linked ubiquiti-

nation of WASH K220 by MAGE-L2-TRIM27 is required for

endosomal F-actin nucleation and retrograde transport. WASH

K220 is predicted to exist in an analogous region on the SHRC

as the meander region on the known WAVE regulatory complex

structure. The meander region makes contacts with WAVE

itself and SRA1 (analogous to SWIP in SHRC) and is essential

for maintaining WAVE in an inactive state (Chen et al., 2010).

Importantly, this region is subjected to regulation by phosphory-

lation (Padrick and Rosen, 2010). Thus, the proposed meander

region on both the WAVE and WASH regulatory complexes

may be regulated by posttranslational modifications, phosphor-

ylation, and ubiquitination, respectively. In addition, WASH K220

is highly conserved in vertebrates where TRIM27 is found. How-

ever, it is not conserved in invertebrates where TRIM27 orthologs

are not present (Boudinot et al., 2011). Thus, regulation of SHRC

by ubiquitination is likely highly conserved in vertebrates, but

additional forms of regulation are predicted in invertebrates.

Our results suggest that ubiquitination of WASH facilitates

activation by directly destabilizing autoinhibitory contacts in

the SHRC, thus allowing VCA exposure, Arp2/3 complex

binding, and subsequent F-actin assembly (Figure 7G). Con-

sistently, destabilizing these autoinhibitory contacts bypasses

the requirement for WASH ubiquitination. In the context of the

cell, other factors likely cooperate to further enhance activity,

perhaps by clustering the active SHRC. Another likely point

of regulation is in deactivating WASH after sufficient endo-

somal F-actin nucleation for retrograde trafficking. Indeed, we

find that ubiquitin-mediated activation of SHRC is reversible.

Future studies into relevant deubiquitinating enzymes will be of

interest.

Our findings also provide important insights into pathological

conditions associated with MAGE-L2, TRIM27, and retrograde

transport. Genes required for retrograde transport are fre-

quently amplified in melanomas, contribute to tumorigenesis,

and mediate trafficking of proteins required for tumor progres-

sion (Scott et al., 2009; Zech et al., 2011). Our findings extend

this work to show that the TRIM27 oncogene facilitates integrin

a5 recycling, an important event in tumorigenesis. Additionally,

retrograde transport has been implicated in Alzheimer’s disease,

where components of the pathway are downregulated and

mutated (Small, 2008). Similarly, MAGE-L2 has been reported

to be downregulated in the hippocampus of patients with incip-

ient Alzheimer’s disease (Blalock et al., 2004). Finally, retrograde

transport is an essential pathway in which many microbial toxins

and some viruses enter cells. Indeed, inhibition of TRIM27 blocks

trafficking of cholera toxin. Our results suggest a potential

strategy to combat these pathogens.

EXPERIMENTAL PROCEDURES

Cell Culture, Transfections, siRNAs, and Antibodies

Cells were cultured under standard conditions and transfected according

to manufacturer’s recommendation. Detailed descriptions of cell culture

conditions, transfection procedures, siRNA sequences, and antibodies are

described in the Extended Experimental Procedures.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. 1061

siMAGE-L2 siTRIM27siControl

GFP

F-A

ctin

Pha

se

GFP

F-A

ctin

Pha

seHA-GFP-WASH Complex

ΔVCA WT WT + CytoD K220R K220D

GFP

F-A

ctin

Pha

se

Buffer AMSH

Buf

fer

AM

SH

WASH –

K63

-Ub

WA

SH

–170

–130

–100

kDa

05

101520253035

0 500 1000 1500 2000 2500

Fluo

resc

ence

Inte

nsity

(AU

)

Time (s)

BufferWTK220RK220D

0

5

10

15

20

25

30

0 500 1000 1500 2000 2500

Fluo

resc

ence

Inte

nsity

(AU

)

Time (s)

WT + BufferWT + AMSH

F

G

EDC

BA

SWIP-STRUM

Active

EndosomalActin Nucleation

Inhibited

WASHComplex

K63-LinkedUbiquitination

K220

ProteinTrafficking

RetromerVPS3526 29

U

Ube2OTRIM27

FAM21CCDC53

WASH

MAGE-L2VCA

SWIP-STRUM

RetromerVPS3526 29

Ube2OTRIM27

FAM21CCDC53

WASH

MAGE-L2

VCA UUU

UUU

Arp2/3U

Figure 7. In Vitro Reconstitution of MAGE-L2-TRIM27 and K63-Ubiquitin-Dependent SHRC Activity

(A) Cell lysates from WASH knockout MEFs reconstituted with the indicated HA-GFP-WASH proteins were incubated with anti-HA agarose beads and F-actin

stained with phalloidin-568. WT+CytoD was treated with 10 mM cytochalasin D before addition of beads.

(B) WASH knockout MEFs reconstituted with wild-type HA-GFP-WASH were transfected with the indicated siRNAs for 96 hr and actin assembly was determined

as described in (A).

(C) The activity of purified SHRC fromWASH knockout MEFs reconstituted with the indicated HA-GFP-WASH variants was examined by pyrene-actin assembly

assays.

(legend continued on next page)

1062 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

Tandem Affinity Purification and Mass Spectrometry

TAP was performed using 293/TAP-Vector or 293/TAP-MAGE-L2 stable cell

lines as described previously (Doyle et al., 2010) and in the Extended Experi-

mental Procedures.

Immunoprecipitation, Immunoblotting, and Cathepsin D Secretion

Assay

Immunoprecipitation and immunoblotting were performed as described previ-

ously (Potts and Yu, 2005). Cathepsin D secretion assay details are described

in the Extended Experimental Procedures.

Protein Purification and In Vitro Binding Assays

Recombinant proteins were produced using standard procedures described in

the Extended Experimental Procedures. In vitro binding assays were per-

formed as described previously (Doyle et al., 2010) and specified in the

Extended Experimental Procedures.

Immunofluorescence, Microscopy, and Quantitative Measurements

Immunofluorescence was performed essentially as described previously

(Potts and Yu, 2007) and in Extended Experimental Procedures. Retrograde

transport of cell surface CI-M6PR was performed as described previously

(Gomez and Billadeau, 2009) and detailed in the Extended Experimental

Procedures.

Actin Assembly Assays and Purification of SHRC

WASH knockout MEFs (Gomez et al., 2012) were reconstituted with HA-GFP-

tagged WASH and the resulting SHRC variants were purified as described

in the Extended Experimental Procedures. Bead-based and pyrene-actin

assembly assays were performed as described previously (Cory et al., 2003;

Jia et al., 2010) and detailed in the Extended Experimental Procedures.

SUPPLEMENTAL INFORMATION

Supplemental Information includes Extended Experimental Procedures and

seven figures and can be found with this article online at http://dx.doi.org/

10.1016/j.cell.2013.01.051.

ACKNOWLEDGMENTS

We thank Mitsuaki Tabuchi (Kawasaki Medical School, Japan) for providing

the retromer construct. We also thank Potts lab members for helpful discus-

sions and critical reading of the manuscript. This work was supported by the

Cancer Prevention and Research Initiative of Texas R1117 (P.R.P.), Depart-

ment of Defense CA110261 (P.R.P.), Howard Hughes Medical Institute

(M.K.R.), Mayo Foundation (D.D.B.), Michael L. Rosenberg Scholar in Medical

Research fund (P.R.P.), NIH R01-AI065474 (D.D.B.), NIH R01-GM063692

(Z.J.C.), NIH R01-GM56322 (M.K.R.), Sara and Frank McKnight Fellowship

(P.R.P.), and Welch Foundation I–1544 (M.K.R.) and I-1389 (Z.J.C.).

Received: July 25, 2012

Revised: November 29, 2012

Accepted: January 24, 2013

Published: February 28, 2013

REFERENCES

Arighi, C.N., Hartnell, L.M., Aguilar, R.C., Haft, C.R., and Bonifacino, J.S.

(2004). Role of the mammalian retromer in sorting of the cation-independent

mannose 6-phosphate receptor. J. Cell Biol. 165, 123–133.

(D) Cell lysates from reconstituted wild-type HA-GFP-WASHMEFs were treated w

to anti-K63 ubiquitin chain IP, separated by SDS-PAGE, and immunoblotted for

(E) Cell lysates described in (D) were subjected to actin assembly assays on ant

(F) Purified wild-type SHRC was treated with buffer or 3.5 mM AMSH for 1 hr at 3

(G) Model of MAGE-L2-TRIM27, Ube2O, and K63-ubiquitin regulation of SHRC a

Bischof, J.M., Stewart, C.L., and Wevrick, R. (2007). Inactivation of the mouse

Magel2 gene results in growth abnormalities similar to Prader-Willi syndrome.

Hum. Mol. Genet. 16, 2713–2719.

Blalock, E.M., Geddes, J.W., Chen, K.C., Porter, N.M., Markesbery, W.R., and

Landfield, P.W. (2004). Incipient Alzheimer’s disease: microarray correlation

analyses reveal major transcriptional and tumor suppressor responses.

Proc. Natl. Acad. Sci. USA 101, 2173–2178.

Boccaccio, I., Glatt-Deeley, H., Watrin, F., Roeckel, N., Lalande, M., and Mus-

catelli, F. (1999). The human MAGEL2 gene and its mouse homologue are

paternally expressed andmapped to the Prader-Willi region. Hum.Mol. Genet.

8, 2497–2505.

Bochtler, M., Ditzel, L., Groll, M., Hartmann, C., and Huber, R. (1999). The

proteasome. Annu. Rev. Biophys. Biomol. Struct. 28, 295–317.

Bonifacino, J.S., and Hurley, J.H. (2008). Retromer. Curr. Opin. Cell Biol. 20,

427–436.

Bonifacino, J.S., and Rojas, R. (2006). Retrograde transport from endosomes

to the trans-Golgi network. Nat. Rev. Mol. Cell Biol. 7, 568–579.

Boudinot, P., van der Aa, L.M., Jouneau, L., Du Pasquier, L., Pontarotti, P.,

Briolat, V., Benmansour, A., and Levraud, J.P. (2011). Origin and evolution of

TRIM proteins: new insights from the complete TRIM repertoire of zebrafish

and pufferfish. PLoS ONE 6, e22022.

Brass, A.L., Dykxhoorn, D.M., Benita, Y., Yan, N., Engelman, A., Xavier, R.J.,

Lieberman, J., and Elledge, S.J. (2008). Identification of host proteins required

for HIV infection through a functional genomic screen. Science 319, 921–926.

Cai, X., Srivastava, S., Sun, Y., Li, Z., Wu, H., Zuvela-Jelaska, L., Li, J., Sala-

mon, R.S., Backer, J.M., and Skolnik, E.Y. (2011). Tripartite motif containing

protein 27 negatively regulates CD4 T cells by ubiquitinating and inhibiting

the class II PI3K-C2b. Proc. Natl. Acad. Sci. USA 108, 20072–20077.

Campellone, K.G., and Welch, M.D. (2010). A nucleator arms race: cellular

control of actin assembly. Nat. Rev. Mol. Cell Biol. 11, 237–251.

Chen, Z., Borek, D., Padrick, S.B., Gomez, T.S., Metlagel, Z., Ismail, A.M.,

Umetani, J., Billadeau, D.D., Otwinowski, Z., and Rosen, M.K. (2010). Struc-

ture and control of the actin regulatory WAVE complex. Nature 468, 533–538.

Chomez, P., De Backer, O., Bertrand, M., De Plaen, E., Boon, T., and Lucas, S.

(2001). An overview of the MAGE gene family with the identification of all

human members of the family. Cancer Res. 61, 5544–5551.

Cory, G.O., Cramer, R., Blanchoin, L., and Ridley, A.J. (2003). Phosphorylation

of the WASP-VCA domain increases its affinity for the Arp2/3 complex and

enhances actin polymerization by WASP. Mol. Cell 11, 1229–1239.

Cullen, P.J., and Korswagen, H.C. (2012). Sorting nexins provide diversity for

retromer-dependent trafficking events. Nat. Cell Biol. 14, 29–37.

Derivery, E., Sousa, C., Gautier, J.J., Lombard, B., Loew, D., and Gautreau, A.

(2009). The Arp2/3 activator WASH controls the fission of endosomes through

a large multiprotein complex. Dev. Cell 17, 712–723.

Doyle, J.M., Gao, J., Wang, J., Yang, M., and Potts, P.R. (2010). MAGE-RING

protein complexes comprise a family of E3 ubiquitin ligases. Mol. Cell 39,

963–974.

Duleh, S.N., and Welch, M.D. (2010). WASH and the Arp2/3 complex regulate

endosome shape and trafficking. Cytoskeleton (Hoboken) 67, 193–206.

Duleh, S.N., and Welch, M.D. (2012). Regulation of integrin trafficking, cell

adhesion, and cell migration by WASH and the Arp2/3 complex (Hoboken:

Cytoskeleton).

Eden, S., Rohatgi, R., Podtelejnikov, A.V., Mann, M., and Kirschner, M.W.

(2002). Mechanism of regulation of WAVE1-induced actin nucleation by

Rac1 and Nck. Nature 418, 790–793.

ith buffer or 3.5 mMAMSH deubiquitinating enzyme for 1 hr at 30�C, subjectedWASH (anti-HA).

i-HA beads as described in (A).

0�C and activity determined by pyrene-actin assembly assays.

ctivity and endosomal protein recycling. See also Figure S6.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. 1063

Fang, S., and Weissman, A.M. (2004). A field guide to ubiquitylation. Cell. Mol.

Life Sci. 61, 1546–1561.

Gomez, T.S., and Billadeau, D.D. (2009). A FAM21-containing WASH complex

regulates retromer-dependent sorting. Dev. Cell 17, 699–711.

Gomez, T.S., Gorman, J.A., de Narvajas, A.A., Koenig, A.O., and Billadeau,

D.D. (2012). Trafficking defects in WASH-knockout fibroblasts originate from

collapsed endosomal and lysosomal networks. Mol. Biol. Cell 23, 3215–3228.

Harbers, M., Nomura, T., Ohno, S., and Ishii, S. (2001). Intracellular localization

of the Ret finger protein depends on a functional nuclear export signal and

protein kinase C activation. J. Biol. Chem. 276, 48596–48607.

Harbour, M.E., Breusegem, S.Y., and Seaman, M.N. (2012). Recruitment of the

endosomal WASH complex is mediated by the extended ‘tail’ of Fam21

binding to the retromer protein Vps35. Biochem. J. 442, 209–220.

Ismail, A.M., Padrick, S.B., Chen, B., Umetani, J., and Rosen, M.K. (2009). The

WAVE regulatory complex is inhibited. Nat. Struct. Mol. Biol. 16, 561–563.

Jia, D., Gomez, T.S., Metlagel, Z., Umetani, J., Otwinowski, Z., Rosen, M.K.,

and Billadeau, D.D. (2010). WASH and WAVE actin regulators of the Wis-

kott-Aldrich syndrome protein (WASP) family are controlled by analogous

structurally related complexes. Proc. Natl. Acad. Sci. USA 107, 10442–10447.

Jia, D., Gomez, T.S., Billadeau, D.D., and Rosen, M.K. (2012). Multiple repeat

elements within the FAM21 tail link the WASH actin regulatory complex to the

retromer. Mol. Biol. Cell 23, 2352–2361.

Johannes, L., and Popoff, V. (2008). Tracing the retrograde route in protein traf-

ficking. Cell 135, 1175–1187.

Kim, A.S., Kakalis, L.T., Abdul-Manan, N., Liu, G.A., and Rosen, M.K. (2000).

Autoinhibition and activation mechanisms of the Wiskott-Aldrich syndrome

protein. Nature 404, 151–158.

Kim,W., Bennett, E.J., Huttlin, E.L., Guo, A., Li, J., Possemato, A., Sowa, M.E.,

Rad, R., Rush, J., Comb, M.J., et al. (2011). Systematic and quantitative

assessment of the ubiquitin-modified proteome. Mol. Cell 44, 325–340.

Kozlov, S.V., Bogenpohl, J.W., Howell, M.P., Wevrick, R., Panda, S., Hoge-

nesch, J.B., Muglia, L.J., Van Gelder, R.N., Herzog, E.D., and Stewart, C.L.

(2007). The imprinted gene Magel2 regulates normal circadian output. Nat.

Genet. 39, 1266–1272.

Krutzfeldt, M., Ellis, M., Weekes, D.B., Bull, J.J., Eilers, M., Vivanco, M.D.,

Sellers, W.R., and Mittnacht, S. (2005). Selective ablation of retinoblastoma

protein function by the RET finger protein. Mol. Cell 18, 213–224.

Lebensohn, A.M., and Kirschner, M.W. (2009). Activation of the WAVE

complex by coincident signals controls actin assembly. Mol. Cell 36, 512–524.

Linardopoulou, E.V., Parghi, S.S., Friedman, C., Osborn, G.E., Parkhurst, S.M.,

and Trask, B.J. (2007). Human subtelomeric WASH genes encode a new

subclass of the WASP family. PLoS Genet. 3, e237.

Liu, R., Abreu-Blanco, M.T., Barry, K.C., Linardopoulou, E.V., Osborn, G.E.,

and Parkhurst, S.M. (2009). Wash functions downstream of Rho and links

linear and branched actin nucleation factors. Development 136, 2849–2860.

Markson, G., Kiel, C., Hyde, R., Brown, S., Charalabous, P., Bremm, A.,

Semple, J., Woodsmith, J., Duley, S., Salehi-Ashtiani, K., et al. (2009). Analysis

1064 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

of the human E2 ubiquitin conjugating enzyme protein interaction network.

Genome Res. 19, 1905–1911.

Miki, H., Sasaki, T., Takai, Y., and Takenawa, T. (1998). Induction of filopodium

formation by a WASP-related actin-depolymerizing protein N-WASP. Nature

391, 93–96.

Padrick, S.B., and Rosen, M.K. (2010). Physical mechanisms of signal integra-

tion by WASP family proteins. Annu. Rev. Biochem. 79, 707–735.

Potts, P.R., and Yu, H. (2005). Human MMS21/NSE2 is a SUMO ligase

required for DNA repair. Mol. Cell. Biol. 25, 7021–7032.

Potts, P.R., and Yu, H. (2007). The SMC5/6 complexmaintains telomere length

in ALT cancer cells through SUMOylation of telomere-binding proteins. Nat.

Struct. Mol. Biol. 14, 581–590.

Saenko, V., Rogounovitch, T., Shimizu-Yoshida, Y., Abrosimov, A., Lushnikov,

E., Roumiantsev, P., Matsumoto, N., Nakashima, M., Meirmanov, S., Ohtsuru,

A., et al. (2003). Novel tumorigenic rearrangement, Delta rfp/ret, in a papillary

thyroid carcinoma from externally irradiated patient. Mutat. Res. 527, 81–90.

Sandvig, K., and van Deurs, B. (2005). Delivery into cells: lessons learned from

plant and bacterial toxins. Gene Ther. 12, 865–872.

Scott, K.L., Kabbarah, O., Liang, M.C., Ivanova, E., Anagnostou, V., Wu, J.,

Dhakal, S., Wu, M., Chen, S., Feinberg, T., et al. (2009). GOLPH3 modulates

mTOR signalling and rapamycin sensitivity in cancer. Nature 459, 1085–1090.

Shimono, Y., Murakami, H., Hasegawa, Y., and Takahashi, M. (2000). RET

finger protein is a transcriptional repressor and interacts with enhancer of

polycomb that has dual transcriptional functions. J. Biol. Chem. 275, 39411–

39419.

Small, S.A. (2008). Retromer sorting: a pathogenic pathway in late-onset Alz-

heimer disease. Arch. Neurol. 65, 323–328.

Sun, L., and Chen, Z.J. (2004). The novel functions of ubiquitination in

signaling. Curr. Opin. Cell Biol. 16, 119–126.

Takahashi, M., and Cooper, G.M. (1987). ret transforming gene encodes

a fusion protein homologous to tyrosine kinases. Mol. Cell. Biol. 7, 1378–1385.

Tennese, A.A., and Wevrick, R. (2011). Impaired hypothalamic regulation of

endocrine function and delayed counterregulatory response to hypoglycemia

in Magel2-null mice. Endocrinology 152, 967–978.

Xu, M., Skaug, B., Zeng, W., and Chen, Z.J. (2009). A ubiquitin replacement

strategy in human cells reveals distinct mechanisms of IKK activation by

TNFalpha and IL-1beta. Mol. Cell 36, 302–314.

Zech, T., Calaminus, S.D., Caswell, P., Spence, H.J., Carnell, M., Insall, R.H.,

Norman, J., and Machesky, L.M. (2011). The Arp2/3 activator WASH regulates

a5b1-integrin-mediated invasive migration. J. Cell Sci. 124, 3753–3759.

Zha, J., Han, K.J., Xu, L.G., He, W., Zhou, Q., Chen, D., Zhai, Z., and Shu, H.B.

(2006). The Ret finger protein inhibits signaling mediated by the noncanonical

and canonical IkappaB kinase family members. J. Immunol. 176, 1072–1080.

Zoumpoulidou, G., Broceno, C., Li, H., Bird, D., Thomas, G., and Mittnacht, S.

(2012). Role of the tripartite motif protein 27 in cancer development. J. Natl.

Cancer Inst. 104, 941–952.

Supplemental Information

EXTENDED EXPERIMENTAL PROCEDURES

Cell Culture and TransfectionsHEK293 (ATCC), HeLa Tet-ON (Clontech), or U2OS (ATCC) cells were grown in DMEM (Invitrogen) supplemented with 10% FBS (Hy-

clone), 2 mM L-Glutamine (Invitrogen), 100 units/ml penicillin (Invitrogen), 100 mg/ml streptomycin (Invitrogen), and 0.25 mg/ml

amphotericin B (Invitrogen). WASH knockout MEFs were cultured as previously described (Gomez et al., 2012). U2OS ubiquitin

replacement cells were grown under similar conditions. Cell lines in which ubiquitin knockdown and replacement with wild-type

or K63R ubiquitin were described previously (Xu et al., 2009). Similar methods were performed to construct K48R ubiquitin replace-

ment cells. Ubiquitin knockdown and replacement was induced by addition of 1 mg/ml tetracycline for 72 hr (shUb and shUb-Ub

[K48R]) or 96 hr (shUb-Ub[WT] and shUb-Ub[K63R]). Plasmid and siRNA transfection were performed for 48-96 hr with Effectene

(QIAGEN) or Lipofectamine RNAiMAX (Invitrogen), respectively, according to the manufacturer’s protocol.

siRNAssiRNAs (Thermo Dharmacon) used in this study are: FAM21 (50-GGUCUUAACUGGAUUACAA-30), MAGE-L2 #1 (50-UACCUAGA

GUACAGGCGAAdTdT-30), MAGE-L2 #3 (50-ACACUGAGCCCGCAGAGUAdTdT-30), TRIM27 #1 (50-CGGAGAGUCUAAAGCA

GUUdTdT-30), TRIM27 #2 (50-GAACCAGCUCGACCAUUUAdTdT-30), VPS35 (50-AUUUGGUGCGCCUCAGUCAdTdT-30), and

WASH (50-CGCCACUGUGUUCUUCUCUAdTdT-30). siRNAs targeting ubiquitin E2 enzymes, mouse MAGE-L2, and mouse

TRIM27 were purchased as SmartPools (Thermo Dharmacon). Negative control siRNAs (siControl) against an irrelevant gene

(LonRF1) were also purchased as a SmartPool (ThermoDharmacon). SimultaneousWASH knockdown and re-expression constructs

were described previously (Gomez and Billadeau, 2009).

AntibodiesAntibodies used in this study are as follows: anti-APC2 (Tang et al., 2001), anti-ARPC5 (305011; Synaptic Systems), anti-CapZa1

(AB6016; Millipore), anti-CapZB (AB6017; Millipore), anti-Cathepsin D (219361; Calbiochem), anti-CCDC53 (ABT69; Millipore),

anti-CD49e Integrin a5 (555616; BD PharMingen), anti-CI-M6PR (AB2733; Abcam), anti-EEA1 (3288; Cell Signaling Technology),

anti-FAM21 (ABT79; Millipore or as described previously [Gomez and Billadeau, 2009]), anti-GAPDH (2118; Cell Signaling Tech-

nology), anti-goat IgG-HRP (SC-2056; Santa Cruz), anti-Golgin-97 (A-21270; Invitrogen Molecular Probes), anti-hemagglutinin

(12CA5; Roche), anit-K48-linked ubiquitin (8081; Cell Signaling Technology), anti-K63-linked ubiquitin (05-1308; Millipore; or 5621;

Cell Signaling Technology), anti-mouse IgG-HRP (Amersham), anti-Myc (9E10; Roche), anti-rabbit IgG-HRP (Amersham), anti-Strum-

pellin (SC-87442; Santa Cruz), anti-SWIP (ABT70; Millipore), anti-TGN46 (T7576; Sigma), anti-TRIM27 (18791; IBL), anti-VPS26

(AB23892; Abcam), anti-VPS29 (GTX104768; GeneTex), anti-VPS35 (AB10099; Abcam), and anti-WASH (SAB4200373; Sigma or

as described previously [Gomez and Billadeau, 2009]).

Tandem Affinity Purification and Mass SpectrometryTandem affinity purification (TAP) was performed using 293/TAP-Vector or 293/TAP-MAGE-L2 stable cell lines as described previ-

ously (Doyle et al., 2010). In brief, ten 150mm2 dishes of cells were lysed, bound to IgGSepharose beads (GE Amersham), cleaved off

the beads with TEV protease, collected on Calmodulin Sepharose beads (GE Amersham), eluted with SDS sample buffer, separated

by SDS-PAGE, and stained with colloidal Coomassie blue (Pierce). Unique protein bands were excised, in-gel proteolyzed, and iden-

tified by LC/MS-MS.

Protein Purification and In Vitro Binding AssaysGST-tagged VPS35, GST-TRIM27, and His-AMSH were produced in BL21(DE3) cells by overnight induction at 16 �C with 0.5 mM

Isopropyl b-D-1-thiogalactopyranoside (IPTG) and 100 mM ZnCl2 in the case of GST-TRIM27. GST-VPS35 and GST-TRIM27 were

purified from bacterial lysates with glutathione Sepharose (GE Amersham) and eluted with 10 mM glutathione. His-AMSH was puri-

fied from bacterial lysates with Ni-NTA agarose (QIAGEN) and eluted with 250 mM imidazole. In vitro binding assays were performed

as described previously (Doyle et al., 2010). Briefly, 15 mg of purified GST-tagged proteins were bound to glutathione Sepharose

beads (Amersham) for 1 hr in binding buffer (25 mM Tris [pH 8.0], 2.7 mM KCl, 137 mM NaCl, 0.05% Tween-20, and 10 mM 2-mer-

captoethanol) and then blocked for 1 hr in 5%milk powder. In vitro translated proteins (Promega SP6-TNT Quick rabbit reticulocyte

lysate system) were then incubated with the bound beads for 1 hr, extensively washed, eluted with SDS-sample buffer, boiled, sub-

jected to SDS-PAGE, and immunoblotting.

Cathepsin D Secretion AssayCathepsin D secretion was detected as previously described (Rojas et al., 2008). In brief, 72 hr after RNAi or 96 hr after ubiquitin

replacement, cells were washed and incubated for 16 hr in OPTI-MEM without serum. Adherent cells were collected in SDS-sample

buffer for loading controls. The cell culture media were collected and centrifuged for 15 min at 16,000 3 g to pellet cell debris. The

clarified supernatant was treated with 0.02% sodium deoxycholate for 30 min on ice. Proteins were precipitated in 15% trichloro-

acetic acid overnight at 4 �C. Precipitated proteins were pelleted at 16,0003 g for 10 min at 4 �C, washed in ice-cold acetone twice,

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. S1

air-dried, resuspended in SDS-sample buffer, separated by SDS-PAGE, and immunoblotted. Equal loading was determined by pon-

ceau S staining proteins precipitated from the media and immunoblotting of cell lysates.

RT-QPCRRT-QPCR was performed as described previously (Doyle et al., 2010). Briefly, total RNA was extracted (Qiagen), DNase treated

(Invitrogen), converted to cDNA with SuperScript III (Invitrogen), and SYBR green QPCR (Bio-Rad) performed on an Abi-7900HT

instrument (Applied Biosystems).

Immunofluorescence, Microscopy, and Quantitative MeasurementsImmunofluorescencewas performed essentially as described previously (Potts and Yu, 2007). Cells were washed in 1X PBS, fixed for

15 min at room temperature in 3% paraformaldehyde, washed in 13 PBS twice, and incubated in blocking solution (PBS containing

3% bovine serum albumin [BSA] and 0.1% saponin) for 20 min at 4 �C. After blocking, cells were incubated in blocking solution con-

taining approximately 2 mg/ml of each primary antibody for 60 min at room temperature. Cells were then washed three times in PBS

containing 0.1% saponin and incubated for 30 min at room temperature in blocking solution containing 4 mg/ml Alexa-488, Alexa-

568, or Alexa-647 secondary antibodies (Invitrogen Molecular Probes). After incubation, cells were washed three times in PBS con-

taining 0.1% saponin, DNA stained in 1 mg/ml 4’,6-diamidino-2-phenylindole (DAPI) for 2 min, washed in PBS, and mounted. For

F-actin staining, 33 nMAlexa-568-conjugated Phalloidin (Invitrogen) was incubated for 30min at room temperature following primary

antibody staining.

Cells were imaged with a 633 or 1003 objective on a DeltaVision or Nikon TiU inverted fluorescence microscope. Images were

acquired with a CoolSnap HQ2 charge-coupled device camera (Photometrics) at 0.3 mm intervals, deconvolved using the nearest

neighbor algorithm, and stacked to better resolve endosome structures. Intensity measurements were performed using ImageJ Soft-

ware. For those experiments specifically quantitating the amount of endosomal localized F-actin or ARPC5, imageswere thresholded

using the endosomal markers FAM21, VPS35, or WASH and intensity measurements were performed only on the F-actin or ARPC5

signal colocalizing. Quantitation of CI-M6PR trafficking was performed by blind analysis of the percentage of cells showing compact

juxtanuclear (normal) or dispersed (abnormal) CI-M6PR staining. At least 50–100 cells were counted for each condition in each exper-

iment. Statistical analysis was performed using two-tailed unpaired Student’s t test.

For assaying retrograde transport of cell surface CI-M6PR, cells were incubated for 60min or the indicated times at 37�C in serum-

free media (DMEM containing 1% BSA, and 25 mMHEPES [pH 7.4]) containing 10 mg/ml of monoclonal antibody against the luminal

domain of CI-M6PR. Cells were washed in serum-free media three times to remove excess unbound anti-CI-M6PR antibody. Cells

were then fixed, permeabilized, and stained with secondary antibody (anti-mouse Alexa-488) only. For live-cell microscopy, cells

were grown in 4-well chambered coverglasses (Lab-Tek) and imaged on a DeltaVision inverted fluorescence microscope with envi-

ronmental chamber. Single plane images were acquired every 500 ms. Time-lapse images were processed and analyzed by ImageJ

Software.

Flow Cytometry AnalysisHeLa cells were transfected with the indicated siRNAs for 72 hr before staining with 10 mg/ml anti-CD49e (Integrin a5; BD

PharMingen) or IgG isotype control for 30 min on ice. Cells were washed twice and incubated with secondary anti-mouse Alexa-

488 antibody for 30 min on ice. Cells were washed three times and analyzed by flow cytometry.

Cell Invasion AssayHeLa cells were transfected with the indicated siRNAs for 48 hr and subsequently serum starved for 16 hr. One hundred thousand

cells were then plated in serum-free media on matrigel-coated 8 mm transwell invasion chambers (BD Biosciences) with a chemoat-

tractant media containing 2.5% FBS in the bottom chamber. Twenty-four hours after plating, noninvaded cells were scraped off and

invaded cells on the bottom of the transwell membrane were stained with DAPI and crystal violet. The numbers of cells in each field

were counted using a 10X objective.

Actin Assembly Assays and Purification of SHRCWASH knockout MEFs were prepared and characterized previously (Gomez et al., 2012). The WASH knockout MEFs were recon-

stituted with HA-GFP-tagged human WASH wild-type, DVCA, K220R, or K220D by retroviral infection and selection. Equal expres-

sion of mutants at levels comparable to endogenous was confirmed by immunoblotting (data not shown). In vitro actin assembly on

beads using cell lysates was performed as described previously (Cory et al., 2003). Briefly, one 15 cm2 plate ofWASH knockout MEFs

reconstituted with the indicated HA-GFP-tagged WASH proteins was lysed in 1 ml of NP-40 lysis buffer containing MgCl2 (50 mM

Tris-HCl [pH 7.7], 150 mM NaCl, 0.5% NP-40, 5 mM MgCl2, and 1X protease inhibitor cocktail). Lysate protein concentrations

were estimated using Bradford Assay (Bio-Rad) and normalized to 2 mg/ml. HA-GFP-WASH incorporated into SHRC was purified

using anti-HA agarose beads (Sigma) by overnight incubation. Beads were extensively washed in NP-40 lysis buffer containing

MgCl2 and fixed in 3% paraformaldehyde. F-actin on SHRC-captured beads was stained with 66 nM Alexa-568-conjugated Phalloi-

din and beads were squashed on slides and imaged. For examining the dependence of MAGE-L2-TRIM27, WASH knockout MEFs

reconstituted with HA-GFP-WASH were transfected with mouse MAGE-L2 or mouse TRIM27 siRNAs for 72 hr before purification of

S2 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

SHRC and assaying activity as described above. In the indicated samples, clarified cell lysates were treated with 3.5 mM AMSH for

1 hr at 30�C or with 10 mM cytochalasin D before incubation with anti-HA beads.

For in vitro pyrene-actin assembly assays, the tagged WASH SHRC complex was purified to near homogeneity from ten 15 cm2

plates of WASH knockout MEFs reconstituted with the indicated HA-GFP-tagged WASH proteins. Cell pellets were lysed in NP-40

lysis buffer (50 mM Tris-HCl, [pH 7.7], 150 mM NaCl, 0.5% NP-40, 1 mM DTT, and 1X protease inhibitor cocktail) and clarified by

centrifugation and filtration. Clarified lysates were bound to 1 ml of anti-HA agarose beads (Sigma) and washed extensively in lysis

buffer. The SHRCwas eluted six times with 100 mg/ml 3X HA peptide (Sigma). The elutions were pooled, concentrated, and assayed

for all SHRC components and purity by western blotting, Coomassie staining, and mass spectrometry. Note, HA-GFP-WASH was

present at stoichiometric levels to other SHRC proteins, suggesting very little free HA-GFP-WASH in our preparations.

Pyrene-actin assembly assays were performed as described previously (Leung et al., 2006). Briefly, actin was purified from rabbit

muscle and the Arp2/3 complex was purified from bovine thymus. Actin assembly assays contained 4 mM actin (5% pyrene labeled),

10 nMArp2/3 complex in KMEI-15Gbuffer (15% (w/v) glycerol, 50mMKCl, 1mMMgCl2, 1mMEGTA, and 10mM imidazole [pH 7.0]).

In the indicated samples, purified SHRC was treated with buffer or 3.5 mM AMSH for 1 hr at 30�C before pyrene-actin assembly

assays.

SUPPLEMENTAL REFERENCES

Leung, D.W., Morgan, D.M., and Rosen, M.K. (2006). Biochemical properties and inhibitors of (N-)WASP. Methods Enzymol. 406, 281–296.

Rojas, R., van Vlijmen, T., Mardones, G.A., Prabhu, Y., Rojas, A.L., Mohammed, S., Heck, A.J., Raposo, G., van der Sluijs, P., and Bonifacino, J.S. (2008). Regu-

lation of retromer recruitment to endosomes by sequential action of Rab5 and Rab7. J. Cell Biol. 183, 513–526.

Tang, Z., Li, B., Bharadwaj, R., Zhu, H., Ozkan, E., Hakala, K., Deisenhofer, J., and Yu, H. (2001). APC2 Cullin protein and APC11 RING protein comprise the

minimal ubiquitin ligase module of the anaphase-promoting complex. Mol. Biol. Cell 12, 3839–3851.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. S3

Myc

-Vec

tor

Myc

-VP

S35

HA-MAGE-L2

IP:Anti-Myc

WCL

– Myc-VPS35

– HA-MAGE-L2

– HA-MAGE-L2

IP:Anti-Myc

B

siVPS35

siControl

GFP-MAGE-L2

0

20

40

siControl siVPS35

# G

FP-M

AGE-

L2En

doso

mes

per

Cel

l

100 –

75 –

50 –

37 –

Myc

-Vec

tor

Myc

-VPS

35

Myc

-VPS

26

Myc

-MAG

E-L2

Myc

-VPS

29

HA-TRIM27

– HA-TRIM27

– HA-TRIM27

IP:Anti-Myc

IP: A

nti-M

yc

WCL

– Myc-VPS35

– Myc-MAGE-L2

Myc-VPS26

– Myc-VPS29

––

C

20%

Inpu

t

GST

GST

-VPS

35

– Myc-VPS26

– Myc-VPS29

– Myc-TRIM27

– Myc-MAGE-L2

D FE50

30

10

Overlay

mCh-TRIM27

GFP-VPS35

OverlayOverlay Overlay Overlay

mCh-TRIM27 mCh-TRIM27 mCh-TRIM27 mCh-TRIM27

GFP-VPS29 GFP-VPS26 GFP-WASH GFP-FAM21

Overlay

mCh-TRIM27

GFP-Vector

A

Rr = 0.7 Rr = 0.7 Rr = 0.8 Rr = 0.6 Rr = 0.6 Rr = 0.1

250 – 100 – 75 – 50 –37 –

25 –20 –

15 –

Inpu

t

GS

T

GS

T-V

PS

35In

put

GS

T

GS

T-V

PS

35In

put

GS

T

GS

T-V

PS

35In

put

GS

T

GS

T-V

PS

35

Inpu

t

GS

T

GS

T-V

PS

35

Inpu

t

GS

T

GS

T-V

PS

35In

put

GS

T

GS

T-V

PS

35In

put

GS

T

GS

T-V

PS

35

Inpu

t

GS

T

GS

T-V

PS

35In

put

GS

T

GS

T-V

PS

35

Inpu

t

GS

T

GS

T-V

PS

35

Inpu

t

GS

T

GS

T-V

PS

35

Myc-MAGE L2Full Length A B C D E F G H I J K

kDa

Inpu

t

GS

T

GS

T-V

PS

35In

put

GS

T

GS

T-V

PS

35In

put

GS

T

GS

T-V

PS

35

TRIM28 VPS26 VPS29

kDa250 – 100 – 75 – 50 –37 –

25 –

F

C

A

E

FLWH-A WH-B Ext

BWH-A WH-B Ext

DWH-A WH-B Ext

GWH-B ExtHWH-AIWH-BJWH-B ExtKWH-A WH-B Ext

+

+

+

+

+++

–

–

––

–

BindsVPS35

530

530

530

530

510510

464372

390390

298

298

315

315316

206207

100

10099

11

1

1

MAGE-L2

VPS35Binding

G

H

Input (20%) GST GST-VPS35

MAGE-L2 –

FAM21–

MAGE-L2FAM21

+–

–+

+–

–+

++

+–

–+

++

I

Figure S1. MAGE-L2-TRIM27 Localizes toRetromer-Positive Endosomes andMAGE-L2WH-BMotif Binds VPS35 In Vitro, Related to Figure 1

(A) mCherry-TRIM27 and the indicated GFP-tagged proteins were expressed in U2OS cells and representative images are shown. Pearson’s correlation

coefficients (Rr) are shown.

(B) Myc-VPS357 or Myc-Vector and HA-MAGE-L2 were expressed in cells. Forty-eight hours after transfection anti-Myc IP was performed and proteins were

detected by anti-HA and anti-Myc western blot (WB).

(C) TRIM27 associates with the retromer complex in cells. The indicated proteins were expressed for 48 hr before anti-Myc IP and western blotting with anti-Myc

and anti-HA antibodies.

(D) MAGE-L2, but not TRIM27, directly binds to VPS35 TRIM27 in vitro. Binding of the indicated Myc-tagged proteins to recombinant GST-VPS35 or GST alone

was determined by GST pull-down assays. Binding of the indicated Myc-tagged proteins was determined by anti-Myc immunoblotting. Myc-VPS26 and Myc-

VPS29 were used as positive controls.

(E) Localization of GFP-MAGE-L2 to endosomes requires VPS35. U2OS cells stably expressing GFP-MAGE-L2 were treated with control or VPS35 siRNAs for

72 hr before live-cell imaging of GFP-MAGE-L2.

(F) Quantitation of the number of GFP-MAGE-L2 endosomes in control- or VPS35-RNAi cells described in (E). Data are represented as the mean ± standard

deviation. Asterisk indicates p < 0.05.

(G) Mapping MAGE-L2 region interacting with VPS35. Binding of the indicated Myc-MAGE-L2 fragments with recombinant GST-VPS35 or GST alone was

determined by GST pull-down. Binding of theMyc-MAGE-L2 fragments was determined by anti-Myc immunoblotting. Myc-TRIM28 andMyc-TRIM27 were used

as negative controls, while Myc-VPS26 and Myc-VPS29 were used as positive controls.

(H) Summary of binding experiments shown in (G).

(I) MAGE-L2 and FAM21 do not compete for binding to VPS35. Binding of Myc-MAGE-L2 and Myc-FAM21 with recombinant GST-VPS35 or GST alone was

determined by GST pull-down followed by anti-Myc immunoblotting.

Scale bars, 20 mm.

S4 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

A

VPS35 –

TRIM27 –

WASH –

Strumpellin –

SWIP –*

FAM21 –

CCDC53 –

CapZα1 –

CapZβ –

APC2 –

–

–

Moc

k

siC

ontro

l

siTR

IM27

#2

siM

AG

E-L

2 #3

siV

PS

35

siW

AS

H

VPS26 –

B

0

0.2

0.4

0.6

0.8

1

1.2

siC

ontro

l

siM

AG

E-L

2 #1

siM

AG

E-L

2 #3

Rel

ativ

e M

AG

E-L

2/C

ycB

Tran

scrip

t Lev

el

**

siC

ontro

lsi

MA

GE

-L2

siTR

IM27

Overlay + DNASurface labeled

CI-M6PR EEA1F

siControl siMAGE-L2

siTRIM27 siVPS35

Egolgin-97 + DNA

0102030405060708090

100

10 min 20 min

Cel

ls w

ith P

erin

ucle

ar C

I-M6P

R (%

)

ControlMAGE-L2TRIM27MAGE-L2 + TRIM27

G

Anti-CI-M6PR Uptake Time

Rr = 0.1

Rr = 0.7

Rr = 0.6

C

– APC2

– Myc-MAGE-L2

siC

ontro

l

Poo

l

#1 #2 #3

MAGE-L2 siRNAs

D 9080706050403020100

siC

ontro

l+

Myc

-Vec

toor

siM

AG

E-L

2 #3

+ M

yc-V

ecto

r

Cel

ls w

ith D

ispe

rsed

C

I-M6P

R (%

)

siM

AG

E-L

2 #3

+ M

yc-M

AG

E-L

2

*

Figure S2. MAGE-L2 and TRIM27 Are Required for Retrograde Transport, but Not Overall TGN Architecture, Related to Figure 2

(A) Cells were treated with the indicated siRNAs for 72 hr and the indicated proteins were detected by immunoblotting. Asterisk indicates nonspecific band. Note

reduction in components of SHRC upon depletion of WASH and reduction in components of the retromer complex upon depletion of VPS35.

(B) Cells were treated with the indicated siRNAs for 72 hr and endogenous MAGE-L2 mRNA levels were determined by RT-QPCR. MAGE-L2 mRNA levels were

normalized to Cyclophilin B mRNA levels and the relative abundance compared to siControl samples are shown. Data are represented as the mean ± standard

deviation. Asterisk indicates p < 0.05.

(C) Cells were treated with the indicated siRNAs for 24 hr before transfection of Myc-MAGE-L2 expression vector. Forty-eight hours after plasmid transfection cell

lysates were collected, subjected to SDS-PAGE, and immunoblotted for Myc-MAGE-L2 (anti-Myc) or APC2 (loading control).

(D) Cells were treated with the indicated siRNAs for 24 hr before transfection of Myc-vector control or Myc-MAGE-L2 RNAi resistant expression vectors. Forty-

eight hours after plasmid transfection cells were stained with anti-Myc antibody to identify transfected cells and anti-CI-M6PR to assess retrograde transport. The

percentage of cells with dispersed CI-M6PR was hen determined. Data are represented as the mean ± standard deviation. Asterisk indicates p < 0.05.

(E) Cells were treated with the indicated siRNAs for 72 hr before golgin-97 (green) and DNA (blue) were stained. Representative images are shown.

(F) Cells were treated with the indicated siRNAs for 72 hr. Cell surface CI-M6PRwas labeled with anti-CI-M6PR antibody for one hour at 37 �C before imaging the

pool of internalized cell surface-labeled CI-M6PR (green), EEA1 (red), and DNA (blue). Pearson’s correlation coefficients (Rr) are shown.

(G) Cell surface CI-M6PR was labeled with mouse anti-CI-M6PR antibody for 10 or 20 min at 37 �C before the pool of internalized cell surface-labeled CI-M6PR

was imaged with anti-mouse Alexa-488. The percentage of cells with proper retrograde transport of CI-M6PR to perinuclear TGN region is shown.

Scale bars, 20 mm.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. S5

siUbe2A siUbe2D4

siUbe2J2 siUbe2K siUbe2L3 siUbe2L6

siUbe2O siUbe2Q1 siUbe2U siUbe2V1

siUbe2V2 siUbe2W siUbe2Z

siControl siFAM21

A

C

siAKTIP

siC

ontro

l +Ve

ctor

siTR

IM27

#2

+Ve

ctor

siTR

IM27

#2

+TR

IM27

WT

siTR

IM27

#2

+TR

IM27

RIN

G M

ut.

CI-M6PR CI-M6PR + DNAA B

Wild-Type

mC

h-TR

IM27

+ D

NA

RING Mutant

golgin-97 + DNAsiControl siUbe2O siUbe2L6

B

ETet-On

InducibleUbiquitin

KnockdownConstruct

IRES

Tet-On

TW-bUTW-bU a72S04L HA

IRES

Tet-On

Ub-K48R Ub-K48R a72S04L HA

BBBBBBBBBBBBBBBBBBBBBBBBBIRES

Tet-On

Ub-K63R Ub-K63R a72S04L HA

Inducible Rescue Constructs

shUb-Ub(WT)

shUb-Ub(K48R)

shUb-Ub(K63R)

4x shUb1 TargetingUbc and Uba52

6x shUb2 TargetingUbb and Rps27a

+

shUb-Ub(K63R)– + Tet

WASH –

Strumpellin –

SWIP –*

FAM21 –

CCDC53 –

CapZα1 –

CapZβ –

J

G

I

– Te

t+

Tet

shUb shUb-Ub(WT)

shUb-Ub(K48R)

shUb-Ub(K63R)

CI-M6PR+ DNACI-IIM6PRR+RRRRDNADNA

CI-M6PR + DNA

– Te

t+

Tet

Internalized CI-M6PR + DNA

shUb shUb-Ub(WT)

shUb-Ub(K48R)

shUb-Ub(K63R)

shUb-Ub(K48R) shUb-Ub(K63R)– + – +

Anti-K48 Ub Anti-K63 Ub Anti-poly Ub

shUb-Vector– + Tet

170 –130 –100 –

75 –

55 –

40 –35 –

kDa

– CapZα1

F

shUB-Ub(K63R) -Tet

shUb-UB(K63R) +Tet

golg

in-9

7 +

DN

A

H

D

Figure S3. MAGE-L2-TRIM27 Ubiquitin Ligase Activity, Ube2O E2, and K63-Linked Ubiquitination Required for Retrograde Transport,

Related to Figure 3

(A) TRIM27 ubiquitin ligase activity is required for retrograde transport of CI-M6PR. Cells were treated with control or TRIM27 siRNAs for 24 hr before transfection

of RNAi-resistant wild-type or RING mutant TRIM27. Forty-eight hours after transfection, cells were stained for CI-M6PR (green) localization and DNA (blue).

Transfected cells were marked by mCherry expression (data not shown).

(B) TRIM27 RING mutant localizes similar to wild-type TRIM27. mCherry-tagged wild-type or RING mutant TRIM27 were expressed in cells for 48 hr before

imaging.

(C) Cells were treated with siRNAs targeting the indicated E2 enzymes for 72 hr, stained with anti-CI-M6PR, and imaged. Representative images are shown for

a subset of enzymes screened.

(D) Knockdown of Ube2L6, but not Ube2O, results in impaired TGN organization. Cells were treated with the indicated siRNAs for 72 hr before golgin-97 (green)

and DNA (blue) were stained.

(E) Schematic of Ubiquitin replacement cell lines. (Top) Inducible shUbiquitin knockdown construct includes four hairpins targeting the Ubc and Uba52 ubiquitin

encoding genes and six hairpins targeting the Ubb and Rps27a ubiquitin encoding genes. (Bottom) Inducible ubiquitin replacement constructs. Either nothing or

wild-type (WT), K48R, or K63R ubiquitin was fused to Ribosomal L40 and Ribosomal S27A encoding genes to mimic the normal ubiquitin maturation process.

(F) Validation of ubiquitin depletion cell lines. shUb-Ub(K48R), shUb-Ub(K63R), or shUbwere treated with or without tetracycline for 72-96 hr and cell lysates were

immunoblotted with anti-K48-linked ubiquitin, anti-K63-linked ubiquitin, or anti-polyubiquitin antibodies, respectively. Anti-CapZa1 immunoblotting was per-

formed as a loading control.

(G) The indicated ubiquitin replacement cell lines were treated with or without tetracycline for 72-96 hr before steady-state CI-M6PR localization was determined

by immuostaining: CI-M6PR (green) and DNA (blue).

(H) shUb-Ub(K63R) cells were treated with or without tetracycline for 96 hr before golgin-97 (green) and DNA (blue) were stained.

(I) The indicated ubiquitin replacement cells were treated with or without tetracycline for 72-96 hr, followed by followed by serum starvation for 16 hr. Cell surface

CI-M6PR was then labeled with anti-CI-M6PR antibody for one hour before imaging the pool of internalized cell surface-labeled CI-M6PR. CI-M6PR (green) and

DNA (blue) are shown.

(J) No changes toWASH regulatory complex protein levels upon K63-linked ubiquitin depletion. shUb-Ub(K63R) cells were treated with or without tetracycline for

96 hr before cell lysates were collected and immunoblotted for the indicated proteins. Asterisk indicates nonspecific band.

Scale bars, 20 mm.

S6 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

VPS35FAM21 Overlay + DNA

siC

ontro

lB

VPS35 CI-M6PR Overlay + DNAsi

Con

trol

siTR

IM27

VPS35 CI-M6PR Overlay + DNA

shU

b-U

b(W

T)+T

etsh

Ub-

Ub(

K63

R)

+Tet

A

C

siTR

IM27

siControl siTRIM27

VP

S35

D

E

CI-M

6PR

shUb-Ub(K63R)+Tet

shUb-Ub+Tet

shUb-Ub(WT)+Tet

4% 25% 19%

2% 18%

1 μm

1 μm

Cells with Endosomal Tubules (%)

Cells with Endosomal Tubules (%)

F-Actin

CI-M6PR Overlay

F-Actin

CI-M6PR Overlay + DNA

FAM21Δ356N-WASH-VCA

FAM21Δ356N

-WASH-VCA

G shUb-Ub(K63R) +TetF siTRIM27 #2

*

*

* *

*

*

* *

Rr = 0.6

Rr = 0.6

Rr = 0.6

Rr = 0.6

Rr = 0.7

Rr = 0.7

Figure S4. MAGE-L2-TRIM27 Specifically Controls Retrograde Transport through Regulation of Endosomal Actin Dynamics, Related to

Figure 4

(A) Cells were treated with control or TRIM27 siRNAs for 72 hr before staining for VPS35 (green), CI-M6PR (red), and DNA (blue). Representative images are

shown. Scale bars, 20 mm.

(B) shUb-Ub(WT) or shUb-Ub(K63R) cells were treated with tetracycline for 96 hr before staining for VPS35 (green), CI-M6PR (red), and DNA (blue). Repre-

sentative images are shown. Scale bars indicate 20 mm.

(C) Cells were treated with control or TRIM27 siRNAs for 72 hr before staining for VPS35 (red), FAM21 (green), and DNA (blue). Representative images are shown.

Scale bars, 20 mm.

(D) Cells were treated with control or TRIM27 siRNAs for 72 hr before staining with anti-VPS35 antibody. Yellow arrowheads indicate endosomal tubules.

Percentage of cells with endosomal tubules is shown below. Scale bars, 1 mm.

(E) shUb-Ub(WT, shUb-Ub, and shUb-Ub(K63R) cells were treated with tetracycline for 72-96 hr before staining with anti-CI-M6PR antibody. Yellow arrowheads

indicate endosomal tubules. Percentage of cells with endosomal tubules is shown below. Scale bars, 1 mm.

(F) Endosomal localized, uninhibited WASH-VCA rescues CI-M6PR retrograde transport in TRIM27-RNAi cells. Cells were treated with TRIM27 siRNA for 24 hr

before transfection with GFP-FAM21D356N-WASH-VCA. Forty-eight hours after transfection, cells were stained for GFP-FAM21D356N-WASHVCA (red), CI-M6PR

(green), F-actin (cyan), and DNA (blue). Representative image is shown. Compare transfected (YFP-positive) and untransfected cells (asterisks). Scale bars,

20 mm.

(G) Endosomal localized, uninhibited WASH-VCA rescues CI-M6PR retrograde transport in K63-ubiquitin depleted cells. shUb-Ub(K63R) cells were treated with

tetracycline and transfected with GFP-FAM21D356N-WASH-VCA. After 96 hr, cells were stained for GFP-FAM21D356N-WASH-VCA (green), CI-M6PR (cyan), and

F-actin (red). Representative image is shown. Compare transfected (YFP-positive) and untransfected cells (asterisks). Scale bars, 20 mm.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. S7

B

C

YFPYFP-WASH

WTYFP-WASH

K220R

shWASH

AR

PC

5O

verla

y +

DN

AY

FP

D

Human WASH1 PLSISKREQLE 225Chimp WASH1 PLSISKREQLE 225Macaque WASH1 PLSISKREQLE 225Horse WASH PLSISKREQLE 225Dog WASH PLSISKREQLE 225Rat WASH PLSISKREQLE 224Mouse WASH PLSISKREQLE 224Opossum WASH PLSISKREQLE 221Xenopus WASH PLSITKREQLE 222Fugu WASH PLSIAKREQLE 223Zebrafish WASH PLSITKREQLE 225Mesquito WASH SSPIIRSRMKK 218Drosophila WASH PHSLAHGTTKL 236C. briggsae WASH ALSR-LSNDDV 243C. Elegans WASH ALSSTLMQEDS 242Dictystelium WASH PVTVKDGDSRI 225Entamoeba WASH SLS----ELNE 200

TRIM

27 Orthologue

UbiquitinatedLysine (K220)

Control YFPYFP-WASH

WT

***

YFP-WASHK220R

shWASH

CI-M

6PR

CI-M

6PR

+D

NA

No TR

IM27

A

250 –

150 –

100 –

IP: Anti-Myc (Ubiquitin)WB: Anti-SWIP

WCL InputWB: Anti-SWIP

TRIM27-RNAiMyc-Ubiquitin

++

+–

–+

––

kDa

*

*

**

**

*

*

*

**

*

*

**

*

**

*

Figure S5. Retrograde Transport and Endosomal Arp2/3 Localization Is Defective in Cells Expressing WASH K220R, Related to Figure 6

(A) No detectable ubiquitination of theWASH regulatory complex subunit, SWIP. Cells were treated with control or TRIM27 siRNAs for 24 hr before transfection of

Myc-empty or Myc-Ubiquitin expression vectors. Forty-eight hours after plasmid transfection, anti-Myc IP was performed. Whole cell lysates (WCL) or anti-Myc

IP samples were immunoblotted for SWIP.

(B) Retrograde transport is defective in WASH K220R cells. Cells were transfected with the indicated dual-knockdown/re-expression vectors to knockdown

endogenous WASH and re-express YFP-control, YFP-WASH WT, or YFP-WASH K220R. Seventy-two hours after transfection cells were stained for CI-M6PR

(red), YFP (data not shown), and DNA (blue). Note large difference in CI-M6PR localization and abundance in nontransfected cell (asterisk) in YFP-WASH K220R

condition.

(C) Endosomal Arp2/3 localization is defective in WASH K220R cells. Cells were transfected with the indicated dual-knockdown/re-expression vectors to

knockdown endogenous WASH and re-express YFP-control, YFP-WASHWT, or YFP-WASH K220R. Seventy-two hours after transfection cells were stained for

ARPC5 (red), YFP (green), and DNA (blue). Compare transfected (YFP-positive) and untransfected cells (asterisks). Note YFP-WASH K220R localizes to en-

dosomes normally.

(D) UbiquitinatedWASH (K220) is highly conserved in vertebrates in which TRIM27 is present, but is not conserved in invertebrates where TRIM27 is not present.

Sequence alignments around human WASH K220 are shown for the indicated organisms. Dotted line separates vertebrate (top) from invertebrate (bottom).

Scale bars, 20 mm.

S8 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.

HA-GFP-WASH WTW

CL

Dep

lete

d

1 2 3 4 5 6Elutions

WC

L

1 2 3 4 5 6Elutions

WC

L

1 2 3 4 5 6Elutions

– WASH

– FAM21

– STRUM

– SWIP

CCDC53––

WC

L

1 2 3 4 5 6Elutions

HA-GFP-WASH K220R HA-GFP-WASH K220D HA-GFP-WASH ΔVCA

–FAM21–SWIP–STRUM–HA-GFP

170 –130 –

100 –

70 –

55 –

40 –

35 –

25 –

kDa

CoomassieStained Gel

-WASH

Pur

ified

SH

RC

A

B

Figure S6. Purification of Intact SHRC Complex Reconstituted with HA-GFP-WASH Variants, Related to Figure 7

(A) Whole cell lysates (WCL) from WASH knockout MEFs stably expressing HA-GFP-WASH wild-type, K220R, K220D, or DVCA were subjected to anti-HA

chromatography and eluted with HA peptide six times (elutions 1-6). Samples fromWCL, depletedWCL after anti-HA chromatography, and eluted fractions were

separated by SDS-PAGE and immunoblotted with the indicated antibodies.

(B) Material described in (A) was concentrated, separated on SDS-PAGE, and stained with colloidal Coomassie blue. Purified SHRC reconstituted with HA-GFP-

WASH wild-type is shown as a representative example. Note CCDC53 was below the limit of Coomassie staining, but was conformed to be present by

immunoblotting.

Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc. S9

15 m

in30

min

45 m

in60

min

DMSOControl

BIM IPKC Inhibitor

LY294002PI3K Inhibitor

SP600215JNK Inhibitor

U0126MEK Inhibitor

A

B

0102030405060708090

100

0 15 30 45 60Cel

ls w

ith P

erin

ucle

ar C

I-M6P

R (%

)

Time (min)

DMSOBIM ILY294002SP600125U0126

Figure S7. Transport of Cell Surface-Labeled CI-M6PR to TGN Is Dependent on JNK Signaling, Related to Figure 2

(A) Cells were treated with DMSO (60 min), 4 mMPKC inhibitor Bisindolylmaleimide I (BIM I) (30 min), 50 mMPI3K inhibitor LY294002 (60 min), 50 mM JNK inhibitor

SP600215 (45 min), or 10 mMMEK1/2 inhibitor U0126 (60 min) prior to incubation with 10 mg/ml anti-CI-M6PR in serum-free media for 15, 30, 45, or 60 min. Cells

were washed three times in serum-free media, fixed, and internalized CI-M6PR was detected by anti-mouse-Alexa 488 secondary antibody staining. Scale bar,

10 mm.

(B) Cells described in in (A) were quantitated for the percentage of cells with perinuclear TGN localized surface-labeled CI-M6PR.

S10 Cell 152, 1051–1064, February 28, 2013 ª2013 Elsevier Inc.