MHC class II distribution in dendritic cells and B cellsis determined by ubiquitin chain lengthJessica K. Maa,b,1, Mia Y. Platta,b,1, Jeffrey Eastham-Andersonb, Jeoung-Sook Shinc,2, and Ira Mellmanb,2,3

aDepartment of Cell Biology, Yale University School of Medicine, New Haven, CT 06520; bGenentech, South San Francisco, CA 94080; and cDepartment ofMicrobiology and Immunology, Sandler Asthma Basic Research Center, University of California, San Francisco, CA 94143

This contribution is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2011.

Contributed by Ira Mellman, February 19, 2012 (sent for review January 28, 2012)

Dendritic cells (DCs) and B cells present antigen-derived peptidesbound to MHC class II (MHC II) molecules for recognition by CD4-positive T lymphocytes. DCs control the intracellular traffic ofpeptide–MHC II complexes by regulating the ubiquitination ofMHC II. In resting or “immature”DCs, ubiquitinatedMHC II moleculesare targeted to lysosomes, but upon pathogen-induced “matura-tion,” ubiquitination is down-regulated and MHC II can accumulateon the plasma membrane of mature DCs. Although B cells constitu-tively ubiquitinate their MHC II, it unexpectedly remains at the sur-face. We find that DCs and B cells differ in MHC II-conjugatedubiquitin (Ub) chain length: four to six Ub in immature DCs vs. twoto three in B cells. In both cell types, experimentally increasing Ubchain length led to efficient lysosomal transport of MHC II, whereasMHC II with fewer than two Ubs did not reach lysosomes. Thus, Ubchain length plays a crucial role in regulating the intracellular fateand function of MHC II in DCs and B cells.

Dendritic cells (DCs) and B lymphocytes are professional an-tigen-presenting cells (APCs) capable of stimulating efficient

T-cell responses (1, 2). However, their approaches to antigenpresentation differ in important respects. Whereas DCs are highlyendocytic and internalize a wide variety of antigens, B cells takeup and process only the single antigen recognized by their B-cellreceptor. DCs are also distinguished by their ability to regulateantigen processing and presentation by “maturation” (3, 4). Im-mature DCs, found in peripheral tissues, are adept at endocyticuptake of antigen but do not efficiently generate peptide–MHCclass II (MHC II) complexes or express them stably on the cellsurface. In part, this is because MHC II in immature DCs is ubiq-uitinated on a single conserved lysine in the cytoplasmic domain ofthe β-chain (5, 6) by E3 ligases of the membrane-associated RING-CH (MARCH) family (7, 8). Like other ubiquitinated membraneproteins (9), ubiquitinated MHC II molecules are targeted to andsequestered in multivesicular late endosomes and lysosomes. Uponreceiving a maturation stimulus (e.g., Toll-like receptor agonist),however, ubiquitination ceases (5, 6) and peptide–MHC II com-plexes are translocated to and accumulate at the plasma membrane(10–13). In B cells,MHC II surface expression is always high despitealso being ubiquitinated by MARCH ligases in naïve B cells (8).Internalization and down-regulation of receptor tyrosine kinases

by ubiquitination is well known. Ligand binding activates the ki-nase, resulting in autophosophorylation and subsequent recruit-ment of soluble E3 ligases (e.g., Cbl) that ubiquitinate one or moreacceptor lysines. The ubiquitin (Ub) moieties are recognized byUb-interaction motif (UIM)-containing adapter molecules (e.g.,epsins, eps15) that associate with clathrin-coated pits, leading toreceptor internalization (14–18). Upon delivery to early endo-somes, Ub is recognized by members of the endosomal sortingcomplex required for transport (ESCRT) complexes 0–III, whichprevent receptor recycling by facilitating entry of ubiquitinatedcargo into nascent invaginations of the endosomal membrane(19). It is not known whether clathrin-coated pits and the ESCRTmachinery recognize Ub similarly, or whether recognition requiresa single Ub added to a single lysine, multiple lysines, or chains ofUb added to one or more sites (20–24). Nor is it known why

ubiquitinated MHC II in naïve B cells remains on the surface,whereas in immature DCs it is sequestered in late endocyticcompartments. Here, we show that differences in MHC II traf-ficking between DCs and B cells are a consequence of differencesin Ub chain length, not cell type.

ResultsMHC II Ubiquitination, Localization, and Endocytosis Differ BetweenDCs and B Cells.Given the different fates of ubiquitinated MHC IIin DCs and B cells, we first asked whether the two cell typesexhibited quantitative or qualitative differences in ubiquitination.Remarkably, Ub chain lengths were quite different, with up to sixUb monomers conjugated to MHC II in primary mouse bonemarrow-derived DCs (BMDCs) but only two to three in splenic Bcells (Fig. 1A). The difference in Ub chain length correlated witha difference in MHC II localization. In immature DCs, MHC IIwas predominantly in late endosomes and lysosomes, whereas inresting B cells it was primarily on the cell surface (Fig. 1B).The difference also correlated with the ability of both cell types

to internalize MHC II. We monitored endocytosis by bindingfluorescent anti-MHC II monoclonal antibody to DCs or B cellson ice, then warming the cells to 37 °C for various times. Sur-face-bound antibody was removed by incubation at pH 3.0, andacid-resistant, internalized antibody was quantified by flow cy-tometry. DCs internalized antibody more efficiently than B cells,even from the first time point (7 min) (Fig. 1C). By 40 min, DCshad internalized ∼20% of surface MHC II, whereas B cellsinternalized <5%.

MHC II Trafficking Is Altered by Variations in Ubiquitination in DCs.Because the difference between DCs and B cells might reflectcell type-specific differences in endocytosis rather than Ub chainlength, we first reduced the number of Ubs added to MHC II inDCs. For this purpose, we constructed a ubiquitination-incom-petent MHC II β-chain mutant whose single cytoplasmic lysinewas converted to arginine (KR), and to which was fused a cas-sette encoding a single Ub at the C terminus of the β-chain. TheUb fusion provided two opportunities for further ubiquitination:(i) as a Ub donor via an isopeptide bond between its two C-terminal glycines and a substrate’s lysine, and (ii) as a Ub ac-ceptor for E3 ligases via its own internal lysine residues (25). Wetherefore prepared two additional KR fusions using mutant UbcDNAs in which the di-glycine motif was deleted alone or withK > R mutations of all seven internal Ub lysines (Fig. 2A). The

Author contributions: J.K.M., M.Y.P., J.-S.S., and I.M. designed research; J.K.M., M.Y.P.,J.E.-A., and J.-S.S. performed research; J.E.-A. contributed new reagents/analytic tools;J.K.M., M.Y.P., J.E.-A., J.-S.S., and I.M. analyzed data; and J.K.M., M.Y.P., J.-S.S., and I.M.wrote the paper.

The authors declare no conflict of interest.

See Profile on page 8790.1J.K.M. and M.Y.P. contributed equally to this work.2J.-S.S. and I.M. contributed equally to this work.3To whom correspondence should be addressed. E-mail: mellman.ira@gene.com.

8820–8827 | PNAS | June 5, 2012 | vol. 109 | no. 23 www.pnas.org/cgi/doi/10.1073/pnas.1202977109

Dow

nloa

ded

by g

uest

on

Mar

ch 9

, 202

1

control and mutant MHC II β-chain–Ub (KRUb) constructswere retrovirally expressed in DCs from MHC II β-chain−/−mice, as described previously (5, 12).Western blot for Ub after an immunoprecipitation of MHC II

revealed two (red arrowhead) to six Ubs on control MHC II (Fig.2B, Left). KR fused to wild-type Ub not only exhibited the two tosix Ub species but also the 37-kDa monoubiquitinated MHC IIinput and higher-order poly-ubiquitinated species (designatedKRUb-poly) (Fig. 2B, Right, right lane). Although the predom-inant species in KRUb constructs fused to the lysine and glycineUb mutants (KRUb1, KRUb2) was the mono-Ub β-chain input,the KRUb with only the di-glycine deletion (KRUb2) also con-tained a small amount of di-ubiquitinated β-chain [Fig. 2B,KRUb2 (red arrowhead)]. Although imperfect, this panel ofconstructs enabled expression of MHC II bearing one, two, ormultiple Ubs: KRUb1, KRUb2, and KRUb-poly. An entirelyUb-deficient MHC II was provided by expressing the KR mutantwithout fused Ub: KR(0).As expected, wild-type MHC II was found intracellularly with

relatively little surface expression, whereas KR(0) was largely atthe plasma membrane (Fig. 2C). Surprisingly, KRUb1 localizationwas barely distinguishable from KR(0), suggesting that a single Ubis not a sufficient signal for MHC II internalization and/or lyso-somal targeting,much like what has been seen in experiments usingCD4, which internalizes at the same rate unmodified as mono-ubiquitinated (26). With one additional Ub in KRUb2, a smallamount of MHC II accumulation in H2-M–positive late endocyticcompartments was detected. In fact, KRUb2 expression in DCswas reminiscent of endogenousMHC II localization in B cells (Fig.

2C vs. Fig. 1B). Localization of KRUb-poly resembled that of en-dogenous oligoubiquitinated MHC II in immature DCs.The surface expression of MHC II for each transductant was

quantified by flow cytometry (Fig. 2D). The reduction in surfaceMHC II in KRUb2 or KRUb-poly was dramatic, with both ex-hibiting low levels of surface expression comparable to that foundin cells expressing wild-type MHC II.We next determined whether Ub chain length also affected

internalization and lysosomal targeting of surface MHC II bymonitoring the continuous uptake of anti-MHC II antibody (1 hat 37 °C). Similar to the results obtained for MHC II distributionat steady state (Fig. 2C), KR(0)- and KRUb1-expressing DCsbound but failed to internalize appreciable amounts of antibody(Fig. 2E). Significant internalization, however, was exhibited byKRUb2- and KRUb-poly–expressing DCs. These results weresimilar to antibody uptake by DCs expressing wild-type MHC II(Fig. 2E), although it was unclear whether antibody internalizedby KRUb2-expressing cells was delivered efficiently to H2-M–

positive late endosomes and lysosomes. We presume the non-lysosomal structures represent early endosomes, but staining withantibodies to endogenous early endosome markers (e.g., EEA1)was too inefficient to test colocalization reliably.We next used flow cytometry to monitor the internalization of

antibody prebound at 0 °C. As shown in Fig. 2F, relatively littleendocytosis was observed for cells expressing KR(0) or KRUb1.KRUb2-expressing DCs did internalize a significantly greateramount of antibody at all time points, albeit less than in cellsexpressing wild-type MHC II. This may reflect enhanced effi-ciency of the internalization step or greater antibody accumula-tion by lysosomal targeting. Antibody uptake by KRUb-poly–expressing cells could not be detected owing to the low levels ofantibody bound at 0 °C to surface MHC II in most cells (Fig. 2D)and a possible partial impairment of internalization of antibodybound at 37 °C in others (Fig. 2E). Similar data were obtainedusing monovalent anti-MHC II Fab fragments, indicating thatthe results were not influenced by cross-linking.Taken together, these data demonstrate that monoubiquitination

is insufficient for internalization and lysosomal delivery of surfaceMHC II in DCs. Each additional Ub facilitated increased MHC IIendocytosis or lysosomal accumulation, although efficient lyso-somal accumulation was favored with chain lengths greater thantwo Ubs. Most importantly, reducing theMHC II Ub chain lengthin immature DCs produced a localization pattern reminiscent ofthat in B lymphocytes: predominantly cell surface.

Ubiquitinated MHC II in B Cells Behaves Similarly to DCs. Using thesame panel of KRUb constructs, we next asked whether increasingUb chain length would enhanceMHC II endocytosis and lysosomaldelivery in B cells. We retrovirally infected MHC II β-chain−/−B cells activated by overnight incubation with LPS to stimulateproliferation, a necessary step to prepare cells for retroviral in-fection. We observed similar MHC II ubiquitination patterns as inDCs, althoughKRUb-poly produced fewer di- and triubiquitinatedβ-chains in favor of additional poly-ubiquitinated forms (Fig. 3A).By immunofluorescence, the steady-state distribution of KR(0)

and KRUb1 was restricted to the plasma membrane. In KRUb2-expressing cells, most MHC II was on the surface, but a fractioncould be detected intracellularly. KRUb-poly–expressing B cells,however, looked similar to immature DCs, with the bulk of MHCII in lysosomes and relatively little on the plasma membrane (Fig.3B). These results were supported by flow cytometry, which showedthat increasing Ub chain length to greater than two Ub decreasedsurface expression of MHC II (Fig. 3C).Internalization of anti-MHC II antibody at 37 °C recapitulated

the steady-state distribution pattern. Little, if any, internalizedantibody was observed for KR(0) and KRUb1, a small amountfor KRUb2, and a considerable amount for KRUb-poly (Fig.3D). The antibody uptake results were quantified by determining

0 5 10 15 20 25 30 35 40

25

20

15

10

5

0

Time (min):

% e

ntr

y

DC

B

A

C

BIP: MHC II

DC B#Ub

6

5

4

3

2

1

10075

50

37Ub

MHC II

MHC II

LAMP-2

DC B

Fig. 1. Differences in Ub chain length on MHC II of splenic DCs and splenic Bcells correlate with differences in MHC II endocytosis. (A) Western blot ofsplenic DC and splenic B-cell (B) MHC II immunoprecipitates. MHC II wasimmunoprecipitated (IP) with TIB120 antibody; IPs were immunoblotted (IB)with anti-Ub antibody P4D1 andMHC II β-chain antibody Thorax. (B) Confocalmicroscopy of wild-type splenic DCs and splenic B cells. Cells were bound tocoverslips, fixed in paraformaldehyde (PFA) and labeled with anti-MHC IIantibody TIB120 (red) and lysosomal marker LAMP-2 (green). (C) Endocytosisof surface MHC II in splenic DCs and B cells. Cells were bound to anti-MHC IIantibody TIB120 on ice and washed. This was followed by incubating the cellsat 37 °C for the various times indicated to allow internalization of antibody-bound surface MHC II. An acidic wash (pH 3.0) was then used to strip thesurface of remaining antibody. Internalized MHC II was protected from theacid stripping and detected by flow cytometry. Mean fluorescence intensity(MFI) values of MHC II and SEM were determined, and internalization isexpressed as percentages of control MFI levels. Control cells were treatedidentically except for substituting a PBS wash for the acidic wash.

Ma et al. PNAS | June 5, 2012 | vol. 109 | no. 23 | 8821

CELL

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Mar

ch 9

, 202

1

the ratio of each cell’s interior to surface fluorescence intensity ata defined z axis section (5). KRUb-poly was consistently found tohave the highest ratio, whereas KRUb2 exhibited a small butstatistically significant increase in fractional internalization rel-ative to KR(0) and KRUb1 (Fig. 3E). By expressing an MHC IIβ-chain conjugated to greater than two Ub in B cells, it waspossible to enhance MHC II endocytosis and late endocyticdelivery, producing a phenotype reminiscent of immature DCs.Thus, Ub chain length, not cell type, is responsible for the in-tracellular fate of MHC II.

Surface Expression of MHC II Decreases with Increasing Ub ChainLength in DCs and B Cells. Any conclusions concerning the precisenumber of Ubs required for lysosomal transport were limited bythe fact that the Ub fused to MHC II β-chain was subject tofurther modification by Ub ligases in the cell, yielding chains ofindeterminate length, conjugation site, and branching. Ub ad-dition was also likely to be dynamic, with a small fraction of totalMHC II additionally ubiquitinated and for variable lengths oftime. We therefore synthesized a series of KR molecules fused totandem lysine-mutant Ub chains. Linear chains occur naturallyand are synthesized by the LUBAC E3 ligase. They have been

shown to interact with Ub-binding proteins (e.g., NEMO) withaffinities comparable to K63-linked Ub chains, and can directproteasomal degradation (27). Although there is no evidencethat tandem repeats play a role in endocytosis, they are struc-turally similar to lysine 63-linked Ub chains (28). Thus, weconstructed MHC II K > R fusions (KR-Ub) encoding stablechains of one, two, three, four, or six Ub monomers (Fig. 4A).Expression of the KR-Ub tandem repeats in DCs and B cells

clearly illustrated a relationship between Ub chain length andMHC II trafficking. Consistent with the KRUb fusion proteins(Figs. 2 and 3), surface expression of KR-Ub in MHC IIβ-chain−/− DCs and B cells decreased with increasing Ub mon-omers (Fig. 4B). Similarly, steady-state localization of MHC II inboth DCs and B cells correlated with Ub chain length. WhereasUb1 was largely restricted to the plasma membrane, both surfaceand intracellular MHC II were observed in Ub2-expressing cells,and Ub4 and Ub6 were detected largely in intracellular com-partments (Fig. 4 C and D).Surprisingly, despite its low surface expression of MHC II by

FACS, Ub3 by immunofluorescence exists as a heterogeneouspopulation at steady state, resembling Ub2 more than Ub4 andUb6. DCs and B cells can be found with all surface MHC II, all

K R

a bWT KR(0) KRUb1 KRUb2 KRUb-poly

Wild-type ubiquitin:

7Lys, C-ter GG

R

R

R R

R R

R R

DGG

Ub

R

K

K K

K K

K K

DGG

Ub

R

K

K K

K K

K K

GG

Ub

K

K K

K K

K K

GG

Ub

0

5

10

15

20

25

0 5 10 15 20 25 30 35 40 45

WT

KR(0)

KRUb1

KRUb2

Time (min):

% e

ntr

y

A B

C

0

50

100

150

200

250

WT KR(0) KRUb1 KRUb2 KRUb-

poly

MH

C II

MF

I

D

E F

IP:

MHC II WCL

75

50

37

IB: Ub IB: Ub

IP: MHC II Whole Cell Lysate

KR

Ub

1

KR

Ub

2

KR

Ub

-p

oly

KR

Ub

1

KR

Ub

2

KR

Ub

-p

oly

75

50

37

100

WT KR(0) KRUb1 KRUb2 KRUb-poly

MHC II H2-M

WT KR(0) KRUb1 KRUb2 KRUb-poly

MHC II H2-M

Fig. 2. Expression and endocytosis of MHC II–Ub fusion proteins in DCs differ depending on Ub extension. (A) Diagram of MHC II: wild-type and Ub fusionconstructs. Mutations (red) include a K > R mutation in MHC II β-chain and variations in fused monoubiquitin, including C-terminal di-glycine deletion andseven K > R mutations. (B) Western blots of MHC II IPs and whole-cell lysate (WCL). MHC II β-chain−/− BMDC cultures were retrovirally transduced to expresswild-type MHC II (Left) or MHC II–Ub fusion constructs (Right). MHC II was immunoprecipitated from cell lysates with anti-MHC II antibody TIB120. IPs andWCLs were immunoblotted with anti-Ub antibody P4D1. Arrowheads approximate band migration of di-ubiquitinated MHC II. (C) Confocal microscopy oftransduced BMDCs. Cells were bound to coverslips, fixed in PFA, and labeled with TIB120 (red) and lysosomal marker H2-M (blue). (D) MFI of surface-boundMHC II. Data are expressed as mean and SEM. (E) Confocal microscopy of surface MHC II uptake. Transduced BMDCs were incubated with TIB120 for 1 h at37 °C for uptake, extensively washed, fixed, and stained to detect TIB120 (red) and H2-M (blue). (F) Endocytosis of surface MHC II. Acid wash protocol was asdescribed in Fig. 1C.

8822 | www.pnas.org/cgi/doi/10.1073/pnas.1202977109 Ma et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 9

, 202

1

internal MHC II, and both surface and internal MHC II (Fig. 4 Eand F). To more accurately measure the effect of Ub chainlength on MHC II localization, DCs and B cells expressing Ub1,Ub2, Ub3, and Ub4 (comparable to Ub6) were scored accordingto the cellular localization of MHC II: all surface, all internal, orintermediate. The data reveal that with each additional Ub at-tached to MHC II, an increasing fraction of total MHC II can befound intracellularly in both DCs and B cells (Fig. 4G). FromUb2 to Ub3, there are fewer “all surface” and “intermediate”cells, in favor of more “all internal” cells.

MHC II Internalization Efficiency Increases with Increasing Ub ChainLength in DCs and B Cells. Endocytosis of anti-MHC II antibodyrecapitulated the steady-state distribution. Whereas Ub1 waspoorly internalized in DCs and B cells, Ub4 and Ub6 mediatedefficient transport to late endosomes and lysosomes, and Ub2exhibited an intermediate degree of internalization (Fig. 5A).Again, Ub3 displayed heterogeneity, with no clearly represen-tative degree of MHC II internalization. In all KR-Ub–express-ing cells, although we would expect the ratio of internal-to-external MHC II to mirror that seen in Fig. 3E, with the ratioinversely proportional to Ub number, this value could not bereliably measured, because total protein levels were also in-versely proportional to Ub number: although immunofluores-cence clearly shows that chain lengths of four or more Ub enableefficient MHC II internalization, it also follows that those speciesare efficiently degraded.The extent of internalization of the Ub2 and Ub3 tandem

constructs seemed greater than observed for endogenously gen-erated di- and triubiquitinated MHC II molecules in B cells. Thisis not surprising given that these tandem repeats are stably expressed,whereas endogenous chains are transient and comprise only a smallfraction of total MHC II at any one time. There are likely moreopportunities for internalization of even poorly recognized butuniformly expressed di- and triubiquitinated molecules. Never-theless, the internalization data also suggested that the stablyexpressed Ub1 was no better at mediating antibody uptake thannonubiquitinated MHC II (KR(0); Figs. 2E and 3D). To examinethis possibility quantitatively, as well as to quantify MHC II in-ternalization in Ub3-expressing cells, we scored anti-MHC IIantibody internalization in DCs expressing KR, Ub1, Ub2, Ub3,and Ub4 by scoring the number of internal vesicles labeled after

15 min at 37 °C: zero, one to five, or more than five MHC II-positive vesicles. Given the low frequency (<10%) of productivelyinfected cells, biochemical measurements were not possible.Even though roughly half of Ub1-expressing DCs had inter-

nalized antibody after 15 min, all these cells displayed fewerthan five vesicles per cell (Fig. 5B). These values were indis-tinguishable from antibody uptake in KR-expressing cells, con-firming that a single Ub does not detectably enhance MHC IIendocytosis. Increasing the Ub tandem repeat from one to twomarkedly increased internalization efficiency: 87% of DCs in-ternalized Ub2 in 15 min, with 77% of those containing morethan five vesicles. Ub3 was similar to Ub2, with a slight increase inMHC II internalization efficiency: 93% of DCs internalized Ub3in 15 min, with 75% of those containing more than five vesicles.As expected, 100% of Ub4-expressing DCs internalized MHC IIantibody, with the vast majority (93%) containing more than fivevesicles. These data suggest that DCs use a mechanism for in-efficient, ubiquitin-independent internalization of MHC II atsteady state, but upon the addition of at least two Ubs, in-ternalization is rendered more efficient, and each subsequent Ubenhances internalization further still, similar to wild-type MHCII in DCs (Figs. 4C and 5A).

Enhanced Ubiquitination of MHC II in B Cells Results in MoreIntracellular MHC II. Finally, we used a strategy to increase Ubchain length in B cells without relying on synthetic Ub constructs.MARCH-I has been shown to ubiquitinate a variety of membraneproteins, notablyMHC II inAPCs (7, 8). Normally expressed at lowlevels in B cells and DCs, MARCH-I overexpression has beenshown to decrease the surface expression of MHC II in B cells (8);Ub chain length was not evaluated in these studies, however.As for the above experiments, splenic B cells were activated

by overnight LPS incubation, then retrovirally transduced witha MARCH-1 construct. Although MHC II in freshly isolatedsplenic B cells was found to be di- and triubiquitinated (Fig. 1A), Bblasts activated by LPS stimulation lost this ubiquitination (Fig. 6A,GFP−),much like inDCs.Nevertheless, ubiquitination was restoredby overexpression of MARCH-I (Fig. 6A, GFP+). Furthermore,rather than reproducing the typical B-cell di- and triubiquitination,the MHC II–Ub pattern resembled that normally found in imma-ture DCs, with up to seven distinct monomers (Fig. 6A).

A B C

D0.5

0.4

0.3

0.2

0.1

0KR(0) KRUb1 KRUb2 KRUb-poly

EKR(0) KRUb1 KRUb2 KRUb-poly

MHC II H2-M

MHC II H2-M

KR(0) KRUb1 KRUb2 KRUb-polyWhole Cell Lysate

KR

(0)

KR

Ub

1

KR

Ub

2

KR

Ub

-po

ly

191

97

64

51

39

KR

(0)

KR

Ub

1

KR

Ub

2

KR

Ub

-po

ly

IP: MHC II

IB: Ub

0100200300400500600700

KR(0) KRUb1 KRUb2 KRUb-poly

MH

C II

MF

I

Fig. 3. Expression and endocytosis of MHC II–Ub fusion proteins in splenic B blasts differ depending on Ub extension. (A) Western blots of MHC II IPs andWCLs of MHC II β-chain−/− splenic B blasts retrovirally transduced to express MHC II–Ub fusion constructs. MHC II was immunoprecipitated from cell lysateswith anti-MHC II antibody Y3P. IPs and WCLs were immunoblotted with anti-Ub antibody (Biomol). Arrowhead approximates band migration of di-ubiq-uitinated MHC II. (B) Confocal microscopy of transduced B blasts. Cells were bound to coverslips, fixed in PFA, and labeled with TIB120 (red) and H2-M (blue).(C) MFI of surface-bound MHC II in transduced B blasts. Data are expressed as mean and SEM. (D) Confocal microscopy of surface MHC II uptake. Transduced Bblasts were incubated with TIB120 for 30 min at 37 °C for uptake, extensively washed, fixed, and stained to detect TIB120 (red) and H2-M (blue). (E)Quantification of MHC II antibody internalization. The cell interior and exterior were identified on z axis sections at 2.0 ⌈m above the coverslip, and MHC IIMFI was quantified for each region. Relative intensity of interior to exterior regions is shown as the mean and SEM.

Ma et al. PNAS | June 5, 2012 | vol. 109 | no. 23 | 8823

CELL

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Mar

ch 9

, 202

1

Further characterization ofMHC II inMARCH-I–expressing Bcells revealed that MHC II localized to both the cell surface andlysosomal compartments at steady state (Fig. 6B). Consistent withthis observation, surface MHC II steadily decreased with in-creasing MARCH-I expression (indicated by increasing GFP in-tensity) (Fig. 6C). Increased MHC II internalization was againseen as correlating with the efficiency of MARCH-I expression(Fig. 6 D and E).By overexpressing MARCH-I, we not only restored but also

enhanced MHC II ubiquitination in wild-type B blasts and ob-served the predicted changes in MHC II localization and in-ternalization. As a final quantitative measure, an acid-wash assayshowed that the endocytic rate of MHC II in MARCH-I–ex-pressing B cells was significantly increased relative to nonexpress-ing cells (Fig. 6F).

DiscussionAlthough the difference in Ub chain length between DCs and Bcells does not seem profound, by manipulating Ub addition ina physiological setting we found that chain lengths of two tothree (B cells) vs. four to six (DCs) were wholly responsible fordetermining the localization of MHC II in both cell types. Assuch, two important questions are answered.

First, the sole reason immature DCs accumulate MHC II inlate endosomes and lysosomes, whereas B cells retain MHC II atthe plasma membrane, can be attributed to the longer Ub chainsconjugated to MHC II in DCs. The ability of DCs to regulatetightly the intracellular distribution of MHC II is likely to reflectthe specialized features of DCs as APCs. Immature DCs surveyperipheral tissues, where they encounter incoming pathogens orself-antigens and capture them by nonspecific endocytosis. DCsthen process and load internalized antigens onto MHC II mol-ecules during their 24- to 48-h trek to lymphoid organs. Havingantigen and MHC II colocalized in antigen processing com-partments likely contributes to the exquisite efficiency thatcharacterizes DCs as APCs (1). B cells, on the other hand, canachieve efficient antigen presentation by using a highly specificreceptor-mediated mechanism of antigen uptake via the B-cellreceptor (2). Although the data clearly demonstrate the impor-tance of degree of MHC II ubiquitination, they do not indicatehow each cell type regulates chain length—by ligase, deubiqui-tinase, or other factors.Second, it is now clear that Ub chain length plays a key role in

determining the intracellular fate of ubiquitinated membraneproteins. To date, the significance of oligo-Ub chains vs. Ubmonomers (conjugated to a single or multiple lysine acceptors)has been unclear (20–24), although the K63 linkage is well

BA

WT Ub1 Ub2 Ub4 Ub6

DC

B cell

MHC II LAMP-2

C

D

MHC II Steady StateLocalization

Ub1 Ub2 Ub3 Ub4 Ub1 Ub2 Ub3 Ub40

25

50

75

100

125surface

intermediate

internal

DC B cell

GE

F

B cell

MHC II H2-M

DC

surface intermediate internal

K

a b

R

Wild-type ubiquitin:

7Lys, C-ter GG

WT KR Ub1 Ub2 Ub3 Ub4 Ub6

R

R

R R

R R

R R

GV

Ub

R

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

K

K K

K K

K K

GG

Ub

R

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

R

R R

R R

R R

GV

Ub

B cell

Ub1 Ub2 Ub3 Ub4 Ub60

1000

2000

3000

4000

DC

Ub1 Ub2 Ub3 Ub4 Ub60

1000

2000

3000

4000

% c

ells

wit

h M

HC

II

loca

liza

tio

n

Fig. 4. Surface expression and localization of MHC II synthetic Ub chains in DCs and splenic B blasts decrease with increasing Ub chain length. (A) Diagram ofMHC II: wild-type and Ub chain conjugates. Mutations (red) include a K > R mutation in MHC II β-chain, seven K > R mutations in Ub, and a C-terminal G > Vmutation in Ub. (B) MFI of surface-bound MHC II in transduced MHC II β-chain−/− BMDCs (Left) and splenic B blasts (Right). Data are expressed as mean andSEM. (C–F) Confocal microscopy of transduced BMDCs (C and E) and B blasts (D and F). (C and D) Steady-state localization of MHC II in BMDCs and B cellsexpressing wild type, Ub1, Ub2, Ub4, and Ub6. (E and F) Ub3-expressing BMDCs and B cells exhibit a heterogeneous population of MHC II localization: allsurface, all internal, and intermediate between the two extremes. Cells were bound to coverslips, fixed in PFA, and labeled with anti-MHC II antibody TIB120(red) and lysosomal marker LAMP-1 or H2-M (blue). (G) Quantification of MHC II localization. Transduced BMDCs and B cells were stained with TIB120 andH2-M, then scored for MHC II localization: all surface, both surface and internal, or all internal. n = 50.

8824 | www.pnas.org/cgi/doi/10.1073/pnas.1202977109 Ma et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 9

, 202

1

known to play a role in multivesicular body (MVB) sorting andlysosomal targeting (29, 30). Although previous studies haveattributed a role for Ub number in the regulation of chimericproteins [e.g., CD4 (26)], the use of a biologically relevant ex-ample here clearly demonstrates the physiological importance ofUb chain length as a critical determinant of intracellular fate.One possible explanation for this feature is facilitating lysosomalsorting efficiency. Because the affinity of a single Ub for a singleUIM is low (31), longer chains would exert greater avidity forsuch interactions, as recently shown in vitro for the MVB sortingadapter Hrs, which, incidentally, contains two UIMs (32). This isfurther evidenced by the observations here that each additionalUb conjugated to MHC II subsequently promotes more endo-cytosis and intracellular localization of MHC II. These obser-vations, then, reveal lysosomal sorting to be a nonstochastic,probabilistic event, dependent upon the strength of interactionsbetween ubiquitinated cargo and sorting machinery. This isconsistent with our previous finding that even in immature DCs,only ∼10% of the total MHC II is ubiquitinated at any one time(5). Because Ubs are typically removed from membrane proteinsupon sequestration into MVBs, the transient presence of Ubchains of different length is sufficient to control the final fate ofMHC II molecules.Although endocytosis was not observed for monoubiquitinated

MHC II, we cannot eliminate the possibility of continuous in-ternalization and recycling of a small pool of MHC II, whichcannot be easily detected by antibody uptake and even less so bycell surface biotinylation. Nor do the data distinguish betweenMHC II ubiquitination preceding or succeeding plasma mem-brane trafficking. Overexpression studies of MARCH-1 localizethe ligase to endosomal membranes (33, 34), where it may bepoised to ubiquitinate MHC II during its translocation to the cellsurface or along the endocytic route. The data also suggest thatUb conjugated to MHC II may not exert its function until at leastMHC II reaches the cell surface, or later. The MHC II antibodyinternalization assay in Fig. 5 required MHC II delivery to thecell surface, and both Ub4 and Ub6 were clearly tagged by ex-tracellularly added antibody, despite being largely intracellular atthe steady state. Thus, at least a fraction of the oligoubiquiti-nated MHC II can still reach the surface, after which point Ubchains signal MHC II internalization.Despite extensive progress in understanding the relationship

between ubiquitination and membrane traffic, the role of oli-

goubiquitination has remained murky both because most mem-brane proteins have multiple potential lysine conjugation sitesand because it has been difficult to control the length of Ubchains. By identifying two physiologically significant examples inwhich chain length on a single acceptor lysine was shown to reg-ulate cargo transport to lysosomes, our data demonstrate thatchain length is indeed an important factor in determining the fateof ubiquitinated membrane proteins.

Materials and MethodsCells. BMDCs from C57/BL6 mice (Jackson Laboratory) and MHC II β-chain−/−

mice (Taconic) were prepared by depleting bone marrow cells of ery-throcytes, T cells, B cells, granulocytes, and MHC II-positive cells, andsubsequently cultured in DC growth medium [RPMI 1640 (Invitrogen)supplemented with 5% (vol/vol) FBS, 2 mM L-glutamine, 20 μg/mL genta-micin, 50 μM β-mercaptoethanol, and recombinant mouse GM-CSF (pro-duced as culture supernatant from J558L cells transfected with mouse GM-CSF cDNA)]. Cells were incubated at 37 °C in 7% CO2, and medium waschanged every 2 d. Splenic DCs from C57/BL6 mice were isolated using CD11c+ cell isolation beads (Miltenyi), and the purity was determined by FACSanalysis (80–95%). Splenic B cells were isolated from C57/BL6 mice and MHCII β-chain−/− mice using a B-cell Isolation Kit (Miltenyi). B cells were culturedin B-cell medium [IMDM (Gibco) supplemented with 10% FBS, 4 mM L-glu-tamine, and 50 μM β-mercaptoethanol].

Plasmids and Retroviral Transduction. The cDNA of wild type MHC II IAb wascloned into the LZRS-pBMN plasmid (a gift from Gary Nolan, Stanford Uni-versity, Palo Alto, CA) using the HindIII and NotI sites. The KR(0) mutantconstruct was generated by site-directed mutagenesis. The KRUb-poly con-struct was generated by PCR, eliminating the stop codon of the KR(0) mu-tant and fusing it in-frame with cDNA encoding Ub, a gift from Pietro DeCamilli [Yale University, New Haven, CT (27)]. The KRUb2 mutant was gen-erated by eliminating the two terminal glycine residues of the fused Ub inthe KRUb-poly construct. The KRUb1 mutant was generated by site-directedmutagenesis of the seven internal lysines of the Ub of the KRUb2 mutant.The cDNA of wild-type murine MARCH-I was cloned into the LZRS-pBMNplasmid using the HindIII and NotI sites. IRES-EGFP, obtained by PCR usingpIRES-EGFP2 (Clontech) as a template, was inserted using the NotI sites ofthe LZRS vectors. All KR-Ub tandem repeat plasmids were synthesized byBlue Heron. Retrovirus was generated by transfection of plasmid vectors intophoenix-ecotropic cells using Fugene (Roche). Stable transfectants were se-lected in puromycin, and virus was collected in DC growth medium or B-cellmedium supplemented with polybrene (American Bioanalytical). Virus wasadded to DC or B-cell cultures, and tissue culture plates were spun in a ta-bletop centrifuge at 32 °C, 1,200 × g for 2 h. Virus was removed, and freshmedium was added. Expression was assayed 24 h (B cells) or 48 h (DCs) aftertransduction. In each experiment, viral transduction efficiency was monitored

B cell

DC

WT Ub1 Ub4 Ub6Ub2

MHC II LAMP-2

A

BInternal MHC II

KR Ub1 Ub2 Ub3 Ub40

25

50

75

100

125 0 vesicles1-5>5

Pulse 15 min

Pe

rce

nt

cells

wit

h

MH

C II

+ v

esi

cle

s

Fig. 5. MHC II internalization is enhanced with increasingnumbers of Ub. (A) Confocal microscopy of surface MHC II up-take. Transduced MHC II β-chain−/− BMDCs (Top) and B blasts(Bottom) were incubated with TIB120 for 1 h at 37°C for con-tinuous uptake, extensively washed, fixed, and stained to detectTIB120 (red) and lysosomal marker LAMP-1 (blue). (B) Quantifi-cation of MHC II antibody uptake. Transduced BMDCs were in-cubated with TIB120 at 37 °C for 15 min, washed extensively, andfixed immediately. Surface TIB120 was blocked, cells were per-meabilized, internalized TIB120 was stained, and cells werescored by number of internal MHC II vesicles. n = 30.

Ma et al. PNAS | June 5, 2012 | vol. 109 | no. 23 | 8825

CELL

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Mar

ch 9

, 202

1

by measuring GFP expression by flow cytometry in each experiment; thisapproach was preferable to measuring MHC II β-chains because their rates ofdegradation would be expected to vary as a function of the numbers ofubiquitins attached. Moreover, for each MHC II–Ub construct used, the rel-ative amount of MHC II on the surface was independent of levels of GFPexpressed, as was the steady-state intracellular distribution.

Western Blot. MHC II expression was determined by Western blot. KR-Ubtransductants were sorted on GFP and CD11c (for DCs) or B220 (for B cells)

double-positive cells. Sorted cells were lysed, denatured in sample buffer, andrun for SDS/PAGE. MHC II expression was detected using Bett anti-MHCII antibody.

Antibodies. “Bett” or “Thorax” (rabbit polyclonal antibodies against IAb) wasused for Western blot. TIB120 (anti-MHC II) (BD Biosciences) was used forimmunoprecipitation, immunofluorescence microscopy, and FACS analysis.Y3P (anti-MHC II) (ATCC) was used for immunoprecipitation. CD11c mAb,biotin-CD86 mAb, FITC-CD86 mAb, PerCP/Cy5.5-CD86 mAb, APC-CD86 mAb,PE-CD80 mAb, PE-TIB120 mAb, and Pacific Blue-B220 mAb were purchasedfrom BD Biosciences and used for FACS analysis. P4D1 (anti-Ub mAb; Covance)or rabbit polyclonal anti-Ub antibody (Biomol) was used for Western blot.“Ulm” [rabbit anti-H2-M antibody (28)], anti-LAMP-2 (grown from hybrid-omas in the laboratory), and anti-LAMP-1 (BioLegend) were used for im-munofluorescence microscopy.

Immunoprecipitation. BMDCs, splenic DCs, or B cells (1 × 107) were solubilizedin Triton-lysis buffer [1% Triton X-100, 10 mM Tris·HCl (pH 7.6), 100 mMNaCl, and 5 mM MgCl2] supplemented with 20 mM N-ethylmaleimide andprotease inhibitor mixtures. The soluble cell lysates were mixed with 5 μg ofMHC II antibody (TIB120 or Y3P) and rocked for 1 h. Protein G-Sepharosebeads were added and rocked for 1 more hour. The beads were washedthree times with Triton lysis buffer and run for SDS/PAGE.

Endocytosis of MHC II and Quantitation. BMDCs cultured fromMHC II β-chain−/−

mice were retrovirally transduced on day 2 to express wild-type or MHC II–Ubfusion proteins. On day 5, cells were seeded on Alcian blue-coated coverslipsand treated with MHC II antibody (20 μg/mL, TIB120) for 30 min to 1 h at 37 °C.Subsequently, the cells were washed several times and fixed with 4% para-formaldehyde. Cells were then permeabilized and stained to detect cell-as-sociated anti-MHC II antibody and lysosomal H2-M. For quantitation ofinternalized MHC II antibody, DCs were treated with antibody for 15 min at37 °C, washed, and fixed with 4% paraformaldehyde. Surface antibody wasblocked with unconjugated anti-rat IgG (Jackson ImmunoResearch), cellswere permealized, internalized antibody stained with fluorescent anti-ratIgG, and antibody-positive vesicles counted. Splenic B blasts generated fromMHC II β-chain−/− mice were retrovirally transduced to express wild-typeMHC II, MHC II–Ub fusion proteins, or MARCH-I after overnight culture with25 μg/mL of LPS (Sigma). The day after transduction, B cells were seeded oncoverslips and treated as described above for the BMDCs. For quantitation, Bcells were imaged on a Leica SP5 confocal microscope. Images of z-sectionstaken 2.0 mm from the coverslip were used for quantitative analysis. Todistinguish surface-bound from internalized anti-MHC II antibody, we de-fined the outer edge of each cell by applying a global intensity threshold tothe MHC II image, then smoothening the resulting binary mask to removeany small protrusions. The “interior” was defined by eroding the outer edgeof each cell a distance of seven pixels. The “surface” was defined as theregion between the interior and the outer edge of each cell. These ma-nipulations and the integrated red or green fluorescence intensity of eacharea were analyzed in the Metamorph software package (MDS AnalyticalTechnologies).

Cell Sorting. After retroviral transduction with MARCH-1, B cells were washedand resuspended in PBS supplemented with 0.5% BSA and 2 mM EDTA. Cellswere passed through a 70-μm cell strainer and sorted using a BD FACSAria.Transduced B cells were gated on B220+GFP+ cells. Nontransduced B cellswere gated on B220+GFP− cells.

ACKNOWLEDGMENTS. We thank members of the I.M. laboratory for theirsupport, Vishva Dixit for excellent advice and discussion, and Ivan Dikic forkindly supplying cDNA constructs. This paper is dedicated to our dear friendand mentor Ralph Steinman who passed just too early to learn he had beenawarded the Nobel Prize for discoveries on which this paper, and ourcareers, are based.

1. Trombetta ES, Mellman I (2005) Cell biology of antigen processing in vitro and in vivo.Annu Rev Immunol 23:975–1028.

2. Vascotto F, et al. (2007) Antigen presentation by B lymphocytes: How receptor sig-naling directs membrane trafficking. Curr Opin Immunol 19:93–98.

3. Mellman I, Steinman RM (2001) Dendritic cells: Specialized and regulated antigenprocessing machines. Cell 106:255–258.

4. Reis e Sousa C (2006) Dendritic cells in a mature age. Nat Rev Immunol 6:476–483.

5. Shin JS, et al. (2006) Surface expression of MHC class II in dendritic cells is controlled byregulated ubiquitination. Nature 444:115–118.

6. van Niel G, et al. (2006) Dendritic cells regulate exposure of MHC class II at theirplasma membrane by oligoubiquitination. Immunity 25:885–894.

7. De Gassart A, et al. (2008) MHC class II stabilization at the surface of human dendriticcells is the result of maturation-dependent MARCH I down-regulation. Proc Natl AcadSci USA 105:3491–3496.

2.5

2.0

1.5

1.0

0.5

0GFP: low mid high

Re

lati

ve fl

uo

r. In

t/E

xt

1 10 100 1000 10000

1

10

100

1000

10000

GFP

MH

C II

M1CTL

Time (min): 0 15 40

10

8

6

4

2

0

% e

ntr

y

Uninfected M1-expressing

MHC II H2-M

Uninfected M1-expressing

MHC II H2-M

IP: MHC IIWCL

97

64

51

39

GFP: + - + -A B

C D

E F

Fig. 6. Increased efficiency of MHC II endocytosis in MARCH-I–over-expressing B cells. (A) Western blot of WCL and MHC II IPs of wild-typesplenic B blasts retrovirally transduced to express MARCH-I (M1) Ub ligase.GFP(+)-transduced cells were sorted, with GFP(-) cells as control. MHC II wasimmunoprecipitated from cell lysates with anti-MHC II TIB120 antibody. IPsand WCLs were immunoblotted with anti-Ub antibody P4D1. Arrowheadapproximates band migration of di-ubiquitinated MHC II. (B) Confocal mi-croscopy of M1-expressing B blasts. Cells were bound to coverslips, fixed inPFA, and labeled with TIB120 (red) and lysosomal marker H2-M (blue). (C)FACS analysis of surface-bound MHC in unsorted M1-expressing B blasts.Cells were stained with PE-conjugated TIB120. (D) Confocal microscopy ofsurface MHC II uptake in M1-expressing B blasts. B blasts were incubatedwith TIB120 for 30 min at 37 °C for uptake, extensively washed, fixed, andstained to detect TIB120 (red) and H2-M (blue). (E) Quantification of TIB120internalization as described in Fig. 3E for cells with low-, mid-, and high-intensity GFP. Data are expressed as mean and SEM. (F) Endocytosis of sur-face MHC II in M1-expressing B cells. Acid wash protocol on sorted GFP(+)(“M1”) and GFP(-) (“CTL”) cells as described in Fig. 1C.

8826 | www.pnas.org/cgi/doi/10.1073/pnas.1202977109 Ma et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 9

, 202

1

8. Matsuki Y, et al. (2007) Novel regulation of MHC class II function in B cells. EMBO J 26:846–854.

9. Mukhopadhyay D, Riezman H (2007) Proteasome-independent functions of ubiquitinin endocytosis and signaling. Science 315:201–205.

10. Boes M, et al. (2002) T-cell engagement of dendritic cells rapidly rearranges MHC classII transport. Nature 418:983–988.

11. Cella M, Engering A, Pinet V, Pieters J, Lanzavecchia A (1997) Inflammatory stimuliinduce accumulation of MHC class II complexes on dendritic cells. Nature 388:782–787.

12. Chow A, Toomre D, Garrett W, Mellman I (2002) Dendritic cell maturation triggersretrograde MHC class II transport from lysosomes to the plasma membrane. Nature418:988–994.

13. Pierre P, et al. (1997) Developmental regulation of MHC class II transport in mousedendritic cells. Nature 388:787–792.

14. Chen H, et al. (1998) Epsin is an EH-domain-binding protein implicated in clathrin-mediated endocytosis. Nature 394:793–797.

15. Polo S, et al. (2002) A single motif responsible for ubiquitin recognition and mono-ubiquitination in endocytic proteins. Nature 416:451–455.

16. Confalonieri S, Di Fiore PP (2002) The Eps15 homology (EH) domain. FEBS Lett 513:24–29.

17. Hurley JH, Wendland B (2002) Endocytosis: Driving membranes around the bend. Cell111:143–146.

18. Traub LM (2009) Tickets to ride: Selecting cargo for clathrin-regulated internalization.Nat Rev Mol Cell Biol 10:583–596.

19. Hurley JH, Emr SD (2006) The ESCRT complexes: structure and mechanism of a mem-brane-trafficking network. Annu Rev Biophys Biomol Struct 35:277–298.

20. Barriere H, et al. (2006) Molecular basis of oligoubiquitin-dependent internalizationof membrane proteins in Mammalian cells. Traffic 7:282–297.

21. Haglund K, et al. (2003) Multiple monoubiquitination of RTKs is sufficient for theirendocytosis and degradation. Nat Cell Biol 5:461–466.

22. Hawryluk MJ, et al. (2006) Epsin 1 is a polyubiquitin-selective clathrin-associatedsorting protein. Traffic 7:262–281.

23. Huang F, Goh LK, Sorkin A (2007) EGF receptor ubiquitination is not necessary for itsinternalization. Proc Natl Acad Sci USA 104:16904–16909.

24. Huang F, Kirkpatrick D, Jiang X, Gygi S, Sorkin A (2006) Differential regulation of EGFreceptor internalization and degradation by multiubiquitination within the kinasedomain. Mol Cell 21:737–748.

25. Pickart CM, Fushman D (2004) Polyubiquitin chains: Polymeric protein signals. CurrOpin Chem Biol 8:610–616.

26. Barriere H, Nemes C, Du K, Lukacs GL (2007) Plasticity of polyubiquitin recognition aslysosomal targeting signals by the endosomal sorting machinery. Mol Biol Cell 18:3952–3965.

27. Rahighi S, et al. (2009) Specific recognition of linear ubiquitin chains by NEMO isimportant for NF-kappaB activation. Cell 136:1098–1109.

28. Komander D, et al. (2009) Molecular discrimination of structurally equivalent Lys 63-linked and linear polyubiquitin chains. EMBO Rep 10:466–473.

29. Lauwers E, Jacob C, André B (2009) K63-linked ubiquitin chains as a specific signal forprotein sorting into the multivesicular body pathway. J Cell Biol 185:493–502.

30. Duncan LM, et al. (2006) Lysine-63-linked ubiquitination is required for endolysoso-mal degradation of class I molecules. EMBO J 25:1635–1645.

31. Dikic I, Wakatsuki S, Walters KJ (2009) Ubiquitin-binding domains—from structures tofunctions. Nat Rev Mol Cell Biol 10:659–671.

32. Ren X, Hurley JH (2010) VHS domains of ESCRT-0 cooperate in high-avidity binding topolyubiquitinated cargo. EMBO J 29:1045–1054.

33. Bartee E, Mansouri M, Hovey Nerenberg BT, Gouveia K, Früh K (2004) Down-regulation of major histocompatibility complex class I by human ubiquitin ligasesrelated to viral immune evasion proteins. J Virol 78:1109–1120.

34. Jabbour M, Campbell EM, Fares H, Lybarger L (2009) Discrete domains of MARCH1mediate its localization, functional interactions, and posttranscriptional control ofexpression. J Immunol 183:6500–6512.

Ma et al. PNAS | June 5, 2012 | vol. 109 | no. 23 | 8827

CELL

BIOLO

GY

INAUGURA

LART

ICLE

Dow

nloa

ded

by g

uest

on

Mar

ch 9

, 202

1