UBR box N-recognin-4 (UBR4), an N-recognin of theN-end rule pathway, and its role in yolk sac vasculardevelopment and autophagyTakafumi Tasakia,1,2, Sung Tae Kima,2, Adriana Zakrzewskaa, Bo Eun Leea, Min Jueng Kangb, Young Dong Yoob,Hyun Joo Cha-Molstadc, Joonsung Hwangc, Nak Kyun Soungc, Ki Sa Sunga, Su-Hyeon Kima,c, Minh Dang Nguyend,Ming Sune, Eugene C. Yib, Bo Yeon Kimc,3, and Yong Tae Kwona,b,3

aCenter for Pharmacogenetics and Department of Pharmaceutical Sciences, School of Pharmacy, University of Pittsburgh, Pittsburgh, PA 15261; bWorld ClassUniversity Program, Department of Molecular Medicine and Biopharmaceutical Sciences, Graduate School of Convergence Science and Technology andCollege of Medicine, Seoul National University, Seoul 110-799, Korea; cWorld Class Institute, Korea Research Institute of Bioscience and Biotechnology,Ochang, Cheongwon 363-883, Korea; dDepartment of Clinical Neurosciences, Hotchkiss Brain Institute, University of Calgary, Calgary, AB, Canada T2N 4N1;and eDepartment of Cell Biology, Center for Biologic Imaging, University of Pittsburgh, Pittsburgh, PA 15261

Edited by Peter M. Howley, Harvard Medical School, Boston, MA, and approved January 23, 2013 (received for review October 8, 2012)

The N-end rule pathway is a proteolytic system in which destabiliz-ing N-terminal residues of short-lived proteins act as degradationdeterminants (N-degrons). Substrates carrying N-degrons are recog-nized by N-recognins that mediate ubiquitylation-dependent selec-tive proteolysis through the proteasome. Our previous studiesidentified the mammalian N-recognin family consisting of UBR1/E3α, UBR2, UBR4/p600, and UBR5, which recognize destabilizingN-terminal residues through the UBR box. In the current study, weaddressed the physiological function of a poorly characterizedN-recognin, 570-kDa UBR4, in mammalian development. UBR4-defi-cient mice die during embryogenesis and exhibit pleiotropic abnor-malities, including impaired vascular development in the yolk sac(YS). Vascular development in UBR4-deficient YS normally advancesthrough vasculogenesis but is arrested during angiogenic remodel-ing of primary capillary plexus associated with accumulation ofautophagic vacuoles. In the YS, UBR4 marks endoderm-derived,autophagy-enriched cells that coordinate differentiation of meso-derm-derived vascular cells and supply autophagy-generated aminoacids during early embryogenesis. UBR4 of the YS endoderm is as-sociated with a tissue-specific autophagic pathway that mediatesbulk lysosomal proteolysis of endocytosed maternal proteins intoamino acids. In cultured cells, UBR4 subpopulation is degraded byautophagy through its starvation-induced association with cellularcargoes destined to autophagic double membrane structures. UBR4loss results in multiple misregulations in autophagic induction andflux, including synthesis and lipidation/activation of the ubiquitin-likeprotein LC3 and formation of autophagic double membrane struc-tures. Our results suggest that UBR4 plays an important role in mam-malian development, such as angiogenesis in the YS, in part throughregulation of bulk degradation by lysosomal hydrolases.

cardiovascular system | ubiquitin ligase

The N-end rule pathway is a proteolytic system that controlsthe half-lives of proteins based on destabilizing activities ofN-terminal residues (1–3). In mammals, known N-recognins bindto destabilizing N-terminal residues to induce selective pro-teolysis through the ubiquitin (Ub)-proteasome system (UPS).To date, N-recognins have not been implicated in bulk pro-teolysis by lysosomal hydrolases. Substrates of the N-end rulepathway include cytosolic short-lived proteins carrying type-1(Arg, Lys, and His; positively charged) and type-2 (Phe, Tyr, Trp,Leu, and Ile; bulky hydrophobic) N-degrons (4–7). N-degronscan be exposed through N-terminal Met excision or proteolyticcleavage of otherwise stable proteins. The principal degron Argcan also be generated through posttranslational arginylation ofAsp and Glu (and Cys in mammals as well) by ATE1-encodedArgtRNA-protein transferases (R-transferases) (8–11). In addi-tion, the arginylation-permissive pro–N-degrons Asp and Glucan be generated through deamidation of the secondary pro–N-

degrons Asn and Glu (12). As the classical (type 1/2) N-end rulelists 13 of 20 amino acids of the genetic code, a vast number ofcellular proteins or polypeptide fragments have potential to ac-quire N-degrons through posttranslational modifications of N-terminal residues. In addition to type-1 and type-2 N-degrons, arecent study reported that acetylated N-terminal residues, occur-ring at 80%of human proteins, can function as N-degrons (13). Thefunctions of the N-end rule pathway in degradation of short-livedproteins carrying N-degrons have been characterized in variousprocesses (2, 3). One outstanding question in the N-end rulepathway concerns differential functions of N-recognins that sharebinding specificity to N-terminal residues. Our previous studiesidentified the mammalian N-recognin family consisting of 200-kDaUBR1/E3α, 200-kDa UBR2, 570-kDa UBR4/p600, and 300-kDaUBR5 (4, 5, 7) (Fig. S1). These N-recognins share the UBR box,a ∼70-residue zinc finger domain that acts as a substrate recogni-tion domain (14). Whereas the canonical N-recognins UBR1 andUBR2 have been characterized in various mammalian processes,functions and mechanisms of two noncanonical N-recognins,UBR4 and UBR5, in the N-end rule pathway remain poorly un-derstood. Different from other N-recognins, UBR4 does not havea known ubiquitylation domain (7), suggesting that the actionmechanism of UBR4 in N-end rule degradation could be differentfrom those of canonical N-recognins. Recent studies using culturedcells suggest that UBR4 plays a role in membrane morphogenesisand extracellular matrix-mediated cell survival and signaling (15),human papillomavirus (HPV) E7-mediated cellular immortaliza-tion (16–18), and migration and development of neurons (19).The autophagy-lysosome system (hereafter autophagy) mediates

bulk degradation of cytoplasmic constituents such as proteins andorganelles. Autophagy is induced by starvation or other cellularstresses to promote lysosomal digestion of nonessential cellularmaterials, providing fuels and amino acids and, thus, maintainingessential cellular processes. Substrates of autophagy are seques-tered and delivered to autophagosomes whose fusion with lyso-somes generates autolysosomes wherein the content is degradedby lysosomal hydrolases. The formation of autophagosomes

Author contributions: T.T., S.T.K., A.Z., Y.D.Y., H.J.C.-M., J.H., K.S.S., S.-H.K., M.S., E.C.Y., B.Y.K.,and Y.T.K. designed research; T.T., S.T.K., A.Z., B.E.L., M.J.K., Y.D.Y., J.H., N.K.S., and M.S.performed research; M.D.N. and B.Y.K. contributed new reagents/analytic tools; T.T., S.T.K.,A.Z., B.E.L., M.J.K., Y.D.Y., H.J.C.-M., J.H., N.K.S., K.S.S., S.-H.K., M.S., E.C.Y., B.Y.K., and Y.T.K.analyzed data; and T.T., S.T.K., K.S.S., S.-H.K., B.Y.K., and Y.T.K. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Present address: Medical Research Institute, Kanazawa Medical University, Ishikawa 920-0293, Japan.

2T.T. and S.T.K. contributed equally to this work.3To whom correspondence may be addressed. E-mail: yok5@pitt.edu or bykim@kribb.re.kr.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental.

3800–3805 | PNAS | March 5, 2013 | vol. 110 | no. 10 www.pnas.org/cgi/doi/10.1073/pnas.1217358110

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF1mailto:yok5@pitt.edumailto:bykim@kribb.re.krmailto:bykim@kribb.re.krhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplementalhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplementalwww.pnas.org/cgi/doi/10.1073/pnas.1217358110

involves Autophagy-related protein 12 (ATG12) recruitment tophagophores, which is conjugated to its substrate, ATG5. TheATG5–ATG12 conjugate promotes ligation of the ubiquitin-like protein LC3-I with phospholipid phosphotidylethanolamine(PE) to generate LC3-II whose PE moiety is anchored tophagophore membranes.Studies in the late 70s by New and others demonstrated that the

yolk sac (YS) is a primary supplier of amino acids for the embryo atearly stages [until approximately embryonic day 9.5 (E9.5) in mice]and that these amino acids are generated through lysosomal di-gestion of endocytosed proteins (20, 21). However, the importanceof the YS as a source of amino acids has received little attentionsince then. As illustrated in Fig. S2, the YS of mouse embryos atE8.5 can be divided into two developmentally distinct layers. Theinner layer, derived frommesodermal cells, is a circulation organ inwhich blood vessels are developed through intercellular signaling ofendothelial cells with vascular smooth muscle cells (SMCs) andsurrounding mesenchymal cells. The outer layer, derived fromendodermal cells, contains a highly specialized proteolytic systemin which extracellular maternal proteins (e.g., albumins) in the YScavity are internalized via endocytosis and delivered to multi-vesicular bodies (MVBs)/late endosomes (22) (Fig. S2B). Cargo-loaded MVBs are fused with autophagosomes which, in turn, aredigested by hydrolases of professionally enlarged lysosomes (al-ternatively called vacuoles) of the YS endoderm. The resultinglysosome-derived amino acids with amaternal original are essentialfor protein translation during vascular development and otherdevelopmental processes in the YS and embryo. Because the pla-centa that transports maternal nutrients and gases to the embryos isnot functional at E9.5, the embryo at this stage strictly depends onlysosome-generated amino acids from the YS endoderm. There-fore, in definition, the YS endoderm contains a tissue-specificautophagic system that mediates constitutive degradation of ex-tracellular proteins as a major type of cargoes.In this study, we show that UBR4 plays an important role in

developmental processes during embryogenesis, in part throughits function in autophagy. Our previous and current studies to-gether reveal a dual function of UBR4 in selective proteolysis bythe proteasome and bulk degradation by the lysosome.

ResultsMice LackingUBR4Die During Embryogenesis.To determine the role ofUBR4 in mammalian development, we constructed mice lackingUBR4. As the UBR box is a general substrate recognition domainfor destabilizing N-terminal residues, a UBR-box–containing regionspanning exons 36 through 42 was replaced with internal ribosomeentry site (IRES)-translated tau-lacZ and a floxed tACE-Cre-Neocassette in embryonic stem (ES) cells (Fig. S3). The selectionmarker Neo flanked by loxP sites was removed by the Cre recom-binase expressed from the angiotensin-converting enzyme (tACE)promoter in spermatogonia of germ-line–transmitted F1 pups.Southern blot and reverse transcription-PCR (RT-PCR) analysesconfirmed that UBR4 was properly inactivated (Fig. S4). Geno-typing F2 C57BL/6J:129/Ola hybrid offspring from heterozygousparents yielded no viable UBR4−/− mice (Table S1), indicating thatUBR4−/− mice die as embryos or neonates. By E9.0–9.5, UBR4−/−

embryos were retrieved at Mendelian ratios but growth arrested atthe 14- to 18-somite stage, some of which were found dead (TableS1). No live mutants were retrieved at E11.5 and afterward. Thus,mouse UBR4 is indispensable for survival during embryogenesis.

UBR4-Deficient Embryos Are Impaired in YS Vascular Development.One prominent and invariable phenotype of UBR4−/− embryoswas defective vascular development in the YS. UBR4-deficientYSs were apparently normal at E8.5 but became pale by E9.5(Fig. 1A, arrow). The vascular defect was severer in the YScompared with the embryo proper, and the severity of YS vas-cular defect correlated to death and developmental arrest of theembryo. UBR4−/− YS showed wild-type rates in cell proliferationas determined by bromodeoxyuridine (BrdU) incorporation as-say of pregnant females (Fig. 2L) and apoptosis as determined byterminal deoxynucleotidyl transferase dUTP nick end labeling(TUNEL) assay (Fig. 2M and Fig. S5A). As further discussed inSI Results, these results suggest that YS vascular defects are theprimary cause of death in UBR4-deficient embryos.

Vascular Development in UBR4-Deficient YS Is Arrested DuringAngiogenic Remodeling of Primary Capillary Plexus. To character-ize vascular development in +/+ and UBR4−/− embryos, we ex-amined expression of platelet endothelial cell adhesion molecule

Fig. 1. UBR4-deficient embryos die during embryogenesis with angiogenic arrest of primary capillary plexus in the YS. (A) Appearance of control and UBR4−/−

embryos at E9.5 with the YS intact. Arrowheads, blood vessels. (B and C) Whole mount PECAM-1 staining of control and UBR4−/− YS at E8.5 (B) and E9.5 (C).(Scale bars, 100 μm.) (D) Cross-sections of +/+ and UBR4−/− YS at E9.5 stained for PECAM-1. (Scale bars, 50 μm.) *, vessels; ee, extraembryonic endodermal layer;mes, mesodermal layer. (E and F) Cross-sections of α–SMA-stained +/+ and UBR4−/− YS at E8.5 (E) and E9.5 (F). (Scale bars, 25 μm.) (G and H) Cross-sectionsof control and UBR4−/− embryos stained for PECAM-1 (G) and α-SMA (H). *, aorta. (Scale bars, 25 μm.) (I) Cross-sections of control and UBR4−/− embryonichearts stained for α-SMA. a, atrium; v, ventricle. (Scale bar, 100 μm.) (J) Real-time PCR analysis of control and UBR4−/− YS and embryos at E9.5.

Tasaki et al. PNAS | March 5, 2013 | vol. 110 | no. 10 | 3801

BIOCH

EMISTR

Y

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF3http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF4http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=STXT

1 (PECAM-1), a marker of differentiated endothelial cells.The mRNA expression of PECAM-1 and α-fetoprotein (AFP),a major plasma protein produced by the YS, was normal at E9.5UBR4−/− YS (Fig. S5D). When visualized by whole-mountimmunostaining, E8.5 UBR4−/− YS contained a wild-type level ofPECAM-1+ cells around primary vessels, forming a network ofPECAM-1+ primary vessels (Fig. 1B). Thus, UBR4 is not criticalfor vasculogenesis during which angioblasts proliferate and dif-ferentiate to endothelial cells. However, the same staining showedthat the vascular network of UBR4−/− YS rapidly loses its in-tegrity over a period of 24 h from E8.5 through E9.5 (Fig. 1 Cand D and Fig. S5B). We therefore examined expression ofα-smooth muscle actin (α-SMA), a marker of vascular and car-diac SMCs, inUBR4+/− andUBR4−/−YS at E8.5–E9.5 to monitorreorganization of primary capillary plexus. During angiogenicremodeling, endothelial cells recruit and induce the differentia-tion of mesenchymal cells to vascular SMCs, establishing cell-to-cell adhesion with each other and endothelial cells. Semi-quantitative real-time PCR analysis showed that the mRNA levelof α-SMA in UBR4−/− YS was markedly down-regulated by E9.5(Fig. 1J). Immunostaining of UBR4−/− YS showed that α-SMA+cells normally located in capillary vessels at E8.5 (Fig. 1E) butwere markedly down-regulated in large vessels at E9.5 and, toa lesser degree, in small vessels (Fig. 1F). Down-regulation ofα-SMA+ cells was also observed in arteries (Fig. 1 H vs. G), butnot the hearts (Fig. 1I), of UBR4−/− embryos, which correlated toa moderately affected vascular network on the surface of themutants (Fig. S5C). Thus, YS vascular development is arrestedduring angiogenic remodeling of primary capillary plexus.

In the YS, UBR4 Marks Endoderm-Derived, Autophagy-Enriched CellsThat Coordinate Differentiation of Mesoderm-Derived Vascular Cells.The expression of UBR4 at E8.5–E9.5 was prominent in the YSrelative to the embryo (Fig. 2A, arrow), as determined by whole-mount staining of β-galactosidase (β-gal) derived from tau-lacZ of

UBR4−allele.Whereas β-gal staining inUBR4+/−YSatE9.5 revealeda vascular network (Fig. 2C), this pattern was disrupted in UBR4−/−YS(Fig.2C), similarly toadisorganizedPECAM-1+vascularnetwork(Fig. 1C andD). This prompted us to examine UBR4 expression onthe cross-section of the YS comprising two developmentally distinctlayers, the outer layer derived from the endoderm and the innerlayer from the mesoderm (Fig. S2B). The β-gal staining was notreadily detected in endothelial cells, vascular SMCs, or nearby un-differentiated mesenchymal cells (precursors of vascular SMCs) inthemesoderm (Fig. 2B). Instead, the stainingwas largely restricted toautophagy-enriched endodermal cells (Fig. 2B) that coordinate dif-ferentiationofmesoderm-derived vascular cells and supply lysosome-produced amino acids during early embryogenesis.

UBR4 Is Associated with a Tissue-Specific Constitutive AutophagicPathway in the YS Endoderm. To determine the function of UBR4in the YS endoderm, we immunostained endogenous UBR4 onYS cross-sections at E9.5. In contrast to typical cytosolic proteinsof the UPS that would appear to be diffusive, the stainingappeared as cytosolic punctate signals (Fig. 2 E and F). Althoughmorphologically heterogeneous, some had a round shape withdiameters of 0.1–0.3 or ∼1.0 μm and codistributed with pro-fessionally enlarged lysosomes in a cytosolic compartment be-tween the microvilli and nucleus (Fig. 2F). A significant portionof UBR4 puncta were at least partially associated with punctapositive for LC3, a hallmark of double membrane structures(Fig. 2 I–K). Despite colocalization, the shapes and sizes ofUBR4 puncta were distinct from those of LC3 puncta (Fig. 2K).These results suggest that UBR4 is not a constitutive componentof autophagic machinery but associated with autophagic cargoes.Next, we asked whether UBR4 loss affects autophagic activity.Immunostaining of LC3 on YS cross-sections from +/+ andUBR4−/− embryos at E9.5 indicated that UBR4−/− YS containeda markedly increased level of LC3 puncta in number and signalintensity (Fig. 2G), without significant changes in cell proliferation

Fig. 2. Expression of UBR4 in endoderm-derived, autophagy-enriched YS cells and its association with a constitutive autophagic pathway of the YS endo-derm. (A) X-gal staining of E9.5 UBR4−/− embryo and YS. (B) Cross-section of X-gal–stained UBR4+/− YS. L, lysosome; ee, extraembryonic endoderm; mes,mesodermal layer; bv, blood vessel. (Scale bar, 10 μm.) (C) Whole-mount X-gal staining of UBR4+/− and UBR4−/− YS. Arrowhead, a relatively intense β-gal signalsurrounding a blood vessel. (Scale bar, 25 μm.) (D) EM of HEK293–UBR4V5 cells labeled for UBR4 (12 nm; arrowhead) and LC3 (18 nm). (Scale bar, 200 nm.) (E)Immunostaining of UBR4 on cross-sections of +/+ and UBR4−/− YS at E9.5. (Scale bar, 10 μm.) (F) Enlarged views of areas indicated by rectangles in E. (Scale bar,5 μm.) (G and H) Cross-sections of +/+ and UBR4−/− YS stained for LC3 (G) or ATG12 (H). (Scale bar, 10 μm.) *, lysosome. (I) Cross-sections of the YS costained forUBR4 and LC3. Arrowhead, lysosome; v, blood vessel. (Scale bar, 10 μm.) (J) Enlarged views of area indicated by rectangles in I. (Scale bar, 2 μm.) (K) Enlargedviews of areas indicated by rectangles in J. (L) BrdU incorporation assay of control and UBR4−/− embryos at E9.5. S-phase indices were determined to be 43.5%and 41.7% for +/+ and UBR4−/− YS, respectively. (M) TUNEL assay of +/+ and UBR4−/− YS at E9.5.

3802 | www.pnas.org/cgi/doi/10.1073/pnas.1217358110 Tasaki et al.

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF5http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF2www.pnas.org/cgi/doi/10.1073/pnas.1217358110

and death (Fig. 2 L and M). ATG12 puncta, a hallmark of phag-ophores, were also significantly induced in UBR4−/− YS, althoughless striking compared with LC3 (Fig. 2H). Given that MVBs car-rying maternal proteins from the YS cavity are major autophagiccargoes in the YS endoderm, we asked whether UBR4 is associatedwith the MVB using the transmission electron microscopy (EM).The EM of UBR4 and LC3 in HEK293–UBR4V5 cells identifiedone or a few UBR4 dots (not as a cluster of aggregates) within theMVB or in the proximity of endosome-like cellular materials andLC3 at the “entrance” of the MVB (Fig. 2D). These results suggestthat UBR4 plays a role in a tissue-specific constitutive autophagicpathway of the YS endoderm, through its association with auto-phagic cargoes. Given that the YS endoderm coordinates vasculardevelopment and supplies lysosome-produced amino acids, mis-regulation of autophagy in theYS endoderm is likely to contribute tothe vascular failure in the YS mesoderm (SI Results).

UBR4 Plays a Role in Delivery of Autophagic Cargoes to Phagophores.To test whether UBR4 has a general function in autophagy, weperformed anti-UBR4 immunostaining in HEK293 cells express-ing 570-kDa UBR4 with C-terminal V5 tag (UBR4V5). Thestaining revealed cytosolic punctate signals with diameters of0.1–0.3 or∼1.0 μm, typical for phagophores and autophagosomes,respectively (Fig. 3 A–F). Most UBR4 puncta colocalized at leastpartially with LC3 puncta (Fig. 3 A and D), a hallmark of phag-ophores and autophagosomes. UBR4 puncta with sizes of 0.1–0.3 μm also showed similar colocalization with ATG12 (Fig. 3 Cand F) and ATG5 (Fig. 3 B and E) that mark phagophores.However, UBR4 puncta had shapes and sizes distinct from thoseof ATG12, ATG5, and LC3 (Fig. 3D–F, arrowheads). In addition,a confocal distance measurement indicated that the center ofUBR4 puncta was separated from that of LC3 puncta by ∼0.15μm. To exclude the possibility that UBR4 in HEK293–UBR4V5cells is misfolded and degraded by autophagy as aggregates, wefractionated the cells growing in normal or starvation media.The majority of UBR4 was retrieved from cytosolic and ER

membrane fraction in hypotonic buffer and could be solubilizedwith a nonionic detergent, 1%Nonidet P-40 (Fig. S6). To directlydetermine the association ofUBR4with autophagic vacuoles, weperformed the EM of UBR4V5 in HEK293–UBR4V5 cells. Theanalysis identified one or a few individual UBR4V5 dots (not asa cluster of aggregates) in association with cargo-like materialswith a diameter of ∼0.1 μm and in the proximity of ATG12 onthe surface or lumen of autophagic vacuoles (Fig. 3 G and H).As further discussed in SI Results, these results collectivelysuggest that UBR4 is specifically associated with (unidentified)cellular materials to form UBR4–cargo complexes which, in turn,are delivered to LC3+ autophagic vacuoles. However, we do notexclude the possibility that a fraction of overexpressed UBR4molecules aremisfoldedand, thus, deposited toautophagic vacuoles.Finally, we tested whether recombinant expression of UBR4V5

affects formation of autophagic double membrane structures.Compared with parental HEK293 cells, HEK293–UBR4V5 cellsgenerated increased numbers of ATG12, ATG5, and LC3 puncta(Fig. S7 A–C), suggesting that the level of UBR4 may need to betightly regulated to maintain the homeostasis in UBR4-de-pendent autophagy and other processes.

UBR4 Subpopulation Is Degraded by Autophagy Through Starvation-Enhanced Association with Autophagic Core Machinery.We establishedmouse embryonic fibroblasts (MEFs) from +/+ and UBR4−/− em-bryos at E8.5. The steady-state level of UBR4 was reduced whenMEFs were starved in Hanks’ balanced salt solution (HBSS) andrecovered when supplemented with 10% FBS (Fig. 3M). The down-regulated level ofUBR4 in nutrient-deprivedMEFs (Fig. 3L, lane 13vs. 1) was readily restored by serum resupplementation (Fig. 3L, lane15 vs. 13). The level of UBR4 was increased by the treatment ofbafilomycin A1, which inhibits the fusion of autophagosomes withlysosomes (Fig. 3L, lane 3 vs. 1) or 3-methyladenine that inhibits classIII PI3K (Fig. 3L, lane 5 vs. 1). To examine whether the metabolicstability of UBR4 is sensitive to nutrient states, we monitoredthe decay of UBR4 in +/+ and UBR4−/− MEFs treated with

Fig. 3. UBR4 plays a role in delivery of autophagic cargoes to phagophores, and UBR4 subpopulation is degraded by autophagy through starvation-enhancedassociation with autophagic core machinery. (A–C) Costaining of UBR4 in comparison with LC3 (A), ATG5 (B), and ATG12 (C) in HEK293–UBR4V5 cells. (D–F) Enlargedviews of areas indicated by rectangles inA–C. (G andH) EM of HEK293–UBR4V5 cells labeled for UBR4 (12 nm) and ATG12 (18 nm). (Scale bar, 200 nm.) (I–K) PLA assayofUBR4 in comparisonwithATG12 (I), ATG5 (J), and LC3 (K) inHEK293–UBR4V5 cells. (Scale bar, 1 μm.) (L) Immunoblottingof+/+ (WT) andUBR4−/− (KO)MEFs growingin normal or starvationmedium for 90min in the presence or absence of 0.2 μMbafilomycin A1 or 10mM 3-methyladenine. (M) Immunoblotting ofMEFs cultured innormal media or HBSS for 90 min, with or without being supplemented by 10% (vol/vol) FBS. (N) Cyclohexamide-chase assay of UBR4 in MEFs. Cells were cultured innormal or starvation medium for 6 h in the presence of cycloheximide. Shown is a quantitation of a representative blot image from three independent experiments.

Tasaki et al. PNAS | March 5, 2013 | vol. 110 | no. 10 | 3803

BIOCH

EMISTR

Y

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF6http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF7

cyclohexamide, a translation inhibitor. UBR4 was a relatively stableprotein in untreated cells over 6 h (Fig. 3N). However, whenautophagy was induced using starvation media, UBR4 became sig-nificantly short-lived with a half-life of ∼3.5 h (Fig. 3N). Starvation-induced decay was moderately, but significantly, inhibited by bafilo-mycin A1 (Fig. 3N and Fig. S7D). The relatively modest stabilizationby autophagic inhibition may be because cyclohexamide partiallyinhibits autophagic flux. To examine whether nutrient deprivationpromotes UBR4 association with autophagic core machinery, weused proximity ligation assay (PLA), which visualizes two proteinslocated in the proximity (

UBR4−/− cells showed a comparable level of postpulse degradationwith a half-life of ∼90 min. These results, together with the findingthat the level of LC3 inUBR4−/− cells is sensitive to bafilomycin A1(Fig. 4B, lanes 4 vs. 2 and 10 vs. 8), suggesting that UBR4−/− cellscontain a largely normal range of autophagic flux.

DiscussionSince the discovery in 1986 (1), the N-end rule pathway has beencharacterized as a proteolytic system that controls the half-livesof short-lived proteins through selective proteolysis by the 26Sproteasome. In this study we demonstrate that mice lackingUBR4, a newly identified N-recognin, die during embryogenesisassociated with severe YS vascular defects. UBR4−/− YSs nor-mally generate a honeycomb-like network of primary vascularplexus. However, angiogenic remodeling of the primary vesselsare arrested over a period of 24 h from E8.5 through E9.5 asvascular SMCs fail to reorganize into mature vessels. As furtherdiscussed in SI Results, several lines of evidence suggest that YSvascular defects are the primary cause of the embryonic death.Despite severe defects in vascular development of mesodermal

cells, the expression of UBR4 is not readily detected in the YSmesoderm, a site of vascular development, but is prominent ina neighboring layer derived from the endoderm. The YS endodermis a digestive organ whereinmaternal proteins from theYS cavity areuptaken and digested into amino acids through the MVB–auto-phagosome–lysosome pathway (22). The results presented in thispaper together suggest that UBR4 subpopulation in the YS endo-derm is specifically associated with cellular cargoes to form UBR4–cargo complexes destined to autophagosomes and professionallyenlarged lysosomes. Given that embryonic growth and vascular de-velopment in the YS mesoderm strictly requires lysosome-derivedamino acids from the YS endoderm, it is likely that misregulation ofautophagic pathways in the UBR4−/− YS endoderm significantlycontributes to vascular failure in the mesoderm. One question thatremains to be answered is whether UBR4 puncta in the YS endo-derm carry maternal proteins as autophagic substrates. Anotherquestion is themechanism by which anN-recognin, known to bind tosubstrates of selective proteasomal degradation, is also associatedwith substrates of bulk lysosomal degradation. UBR4 does not havea known ubiquitylation domain but does contain the UBR box,which is conserved in theUBRN-recognin family and binds to type-1destabilizing N-terminal residues, including the primary degron Arg.It is therefore tantalizing to speculate that UBR4 is associated withcellular cargoes through binding of the UBR box to the N-terminalArg exposed from the substrates.Our results show that UBR4−/− YS and MEFs contain an in-

creased level of LC3+ puncta compared with controls. This suggeststhat multiple misregulations of autophagy and other UBR4-

dependent processes in UBR4−/− cells result in cellular stress (e.g.,nutrient insufficiency, and/or accumulation of damaged organelles),which, in turn, stimulates the formation of autophagic vacuoles. In-triguingly, whereas UBR4-loss induces autophagy, UBR4 over-expression also results in autophagic induction, as determined byenhanced formation of UBR4+LC3+ puncta in HEK293–UBR4V5 cells (Fig. S7). This suggests that the level of UBR4mayneed to be tightly regulated by regulated proteolysis to maintainthe homeostasis of UBR4-dependent autophagy and other pro-cesses. A mutually nonexclusive, alternative possibility is that theoverexpression of UBR4 for a long period may generate excessiveUBR4+ cargoes destined to autophagy, which, in turn, inducesformation of phagophores. Further investigation is needed tounderstand the mechanism by which UBR4 plays a dual role inselective degradation of short-lived proteins via the proteasomeand bulk degradation of cellular materials via the lysosome.

Materials and MethodsFordescriptionsofmaterialsandmethods, includingantibodiesandPCRprimers,see SI Results, SIMaterials andMethods, and Tables S1 and S2. Immunostaining,quantitative RT-PCR, transmission electronmicroscopy, proximity ligation assay,immunoblotting, and mouse knockout were performed using standard tech-niques, which are described in SI Results and SI Materials and Methods. Wecloned UBR4 from a bacterial artificial chromosome (BAC) library (CITB; Invi-trogen) derived from the mouse ES cell line CJ7 using a mouse UBR4 cDNAfragment (nucleotides 4231–5430) as a probe. The targeting vector pUBR4KO-tauLacZ was constructed using gene recombineering (recombination-mediatedgenetic engineering), linearized with FseI, and electroporated into E14 ES cells(a gift from Peter Mombaerts, Rockefeller University, New York). The clone2F12 ES cells were injected into C57/BL6J blastocysts to generate chimeric micewhich, in turn,were used to obtain germ-line transmission of theUBR4−mutantallele. Animal studies were conducted according to the Guide for the Care andUse of Laboratory Animals published by the National Institutes of Health (NIHpublication no. 85–23, revised in 1996) and the protocols (0812811-A1) ap-proved by the Institutional Animal Care and Use Committee at the University ofPittsburgh. Euthanization involves inhalant anesthetic, isoflurane, followed byi.p. injection of xylazine (10 mg/kg) and ketamine (100 mg/kg) mixture.

ACKNOWLEDGMENTS. We are grateful to David Rubinsztein at the Univer-sity of Cambridge for critical comments on the manuscript; Donna Stolz atthe University of Pittsburgh for help with transmission electron microscopy;and the Y.T.K. and B.Y.K. research team members at the University of Pitts-burgh, Seoul National University, and Korea Research Institute of Bioscienceand Biotechnology for helpful discussions. This work was supported in partby NIH Grant HL083365 (to Y.T.K.), in part by a Canadian Institute of HealthResearch Operating Grant (to M.D.N.), and in part by World Class UniversityGrant R31-2008-000-10103-0 (to Y.T.K. and E.C.Y.) and World Class InstituteGrant WCI 2009-002 (to B.Y.K.) through the National Research Foundationfunded by the Ministry of Education, Science, and Technology, Korea.

1. Bachmair A, Finley D, Varshavsky A (1986) In vivo half-life of a protein is a function of

its amino-terminal residue. Science 234(4773):179–186.2. Sriram SM, Kim BY, Kwon YT (2011) The N-end rule pathway: Emerging functions

and molecular principles of substrate recognition. Nat Rev Mol Cell Biol 12(11):

735–747.3. Tasaki T, Sriram SM, Park KS, Kwon YT (2012) The N-end rule pathway. Annu Rev

Biochem 81:261–289.4. Kwon YT, et al. (1998) The mouse and human genes encoding the recognition com-

ponent of the N-end rule pathway. Proc Natl Acad Sci USA 95(14):7898–7903.5. Kwon YT, et al. (2003) Female lethality and apoptosis of spermatocytes in mice lacking

the UBR2 ubiquitin ligase of the N-end rule pathway. Mol Cell Biol 23(22):8255–8271.6. Kwon YT, Xia Z, Davydov IV, Lecker SH, Varshavsky A (2001) Construction and analysis

of mouse strains lacking the ubiquitin ligase UBR1 (E3α) of the N-end rule pathway.Mol Cell Biol 21(23):8007–8021.

7. Tasaki T, et al. (2005) A family of mammalian E3 ubiquitin ligases that contain the

UBR box motif and recognize N-degrons. Mol Cell Biol 25(16):7120–7136.8. Hu RG, et al. (2005) The N-end rule pathway as a nitric oxide sensor controlling the

levels of multiple regulators. Nature 437(7061):981–986.9. Kwon YT, et al. (2002) An essential role of N-terminal arginylation in cardiovascular

development. Science 297(5578):96–99.10. Kwon YT, Kashina AS, Varshavsky A (1999) Alternative splicing results in differential

expression, activity, and localization of the two forms of arginyl-tRNA-protein trans-

ferase, a component of the N-end rule pathway. Mol Cell Biol 19(1):182–193.11. Lee MJ, et al. (2005) RGS4 and RGS5 are in vivo substrates of the N-end rule pathway.

Proc Natl Acad Sci USA 102(42):15030–15035.

12. Kwon YT, et al. (2000) Altered activity, social behavior, and spatial memory in micelacking the NTAN1p amidase and the asparagine branch of the N-end rule pathway.Mol Cell Biol 20(11):4135–4148.

13. Hwang CS, Shemorry A, Varshavsky A (2010) N-terminal acetylation of cellular pro-teins creates specific degradation signals. Science 327(5968):973–977.

14. Tasaki T, et al. (2009) The substrate recognition domains of the N-end rule pathway.J Biol Chem 284(3):1884–1895.

15. Nakatani Y, et al. (2005) p600, a unique protein required for membrane morpho-genesis and cell survival. Proc Natl Acad Sci USA 102(42):15093–15098.

16. DeMasi J, Huh KW, Nakatani Y, Münger K, Howley PM (2005) Bovine papillomavirusE7 transformation function correlates with cellular p600 protein binding. Proc NatlAcad Sci USA 102(32):11486–11491.

17. Huh KW, et al. (2005) Association of the human papillomavirus type 16 E7 onco-protein with the 600-kDa retinoblastoma protein-associated factor, p600. Proc NatlAcad Sci USA 102(32):11492–11497.

18. White EA, et al. (2012) Systematic identification of interactions between host cellproteins and E7 oncoproteins from diverse human papillomaviruses. Proc Natl AcadSci USA 109(5):E260–E267.

19. Shim SY, et al. (2008) Protein 600 is a microtubule/endoplasmic reticulum-associatedprotein in CNS neurons. J Neurosci 28(14):3604–3614.

20. Freeman SJ, Lloyd JB (1983) Evidence that protein ingested by the rat visceral yolk sacyields amino acids for synthesis of embryonic protein. J Embryol ExpMorphol 73:307–315.

21. New DA (1978) Whole-embryo culture and the study of mammalian embryos duringorganogenesis. Biol Rev Camb Philos Soc 53(1):81–122.

22. Fader CM, Colombo MI (2009) Autophagy and multivesicular bodies: Two closelyrelated partners. Cell Death Differ 16(1):70–78.

Tasaki et al. PNAS | March 5, 2013 | vol. 110 | no. 10 | 3805

BIOCH

EMISTR

Y

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=SF7http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=ST2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1217358110/-/DCSupplemental/pnas.201217358SI.pdf?targetid=nameddest=STXT

Supporting InformationTasaki et al. 10.1073/pnas.1217358110SI ResultsYolk Sac Vascular Defects Are the Primary Cause of Death in UBR4-Deficient Mouse Embryos. It is challenging to dissect phenotypes ofgenetically engineered mouse embryos that die during embryo-genesis as multiple misregulations may interfere with each other.We suggest that UBR4-deficient mouse embryos die primarilydue to vascular defects in the yolk sac (YS) based on the followingreasons. First, embryonic death at approximately embryonic day9.5 (E9.5)–E10.5 is typically caused by cardiovascular defectsbecause most of noncardiovascular organs at this stage, such asliver, brain, kidney, and pancreas, are not functional yet and,thus, are not required for survival at E9.5–E10.5. Second, defectsin cardiac development do not cause lethality at E9.5–E10.5because the heart is also not functional at this stage. Third, be-cause the placenta that provides maternal nutrients and gases tothe embryo is not functional at E9.5–E10.5, the embryo must besupplied with nutrients, amino acids, and gases exclusively fromthe YS. As such, the death of mouse embryos at this stage istypically caused by circulation defects, either in the YS or em-bryo. Fourth, our results suggest that vascular defects are moresevere in the yolk sac compared with the embryonic body. Fifth,the severity of yolk sac vascular defects tightly correlates to thetiming of embryonic death and the severity of developmentalarrest. Sixth, UBR4 is more prominently expressed in the yolksac relative to the embryonic body. Seventh, UBR4-deficientyolk sac shows normal rates of cell proliferation (Fig. 2L) anddeath (Fig. 2M).

Misregulation of Autophagy in the YS Endoderm Contributes to theVascular Failure in the YS Mesoderm. As illustrated in Fig. S2, theYS endoderm has a specialized proteolytic system that inter-nalizes, delivers (via multivesicular bodies, MVBs), and digestsmaternal proteins using hydrolases of professionally enlargedlysosomes. The resulting lysosome-derived amino acids are: (i)used to synthesize grow factors to orchestrate vascular devel-opment in the YS mesoderm, (ii) supplied to the underlying YSmesodermal layer to synthesize protein in vascular development,and (iii) supplied to the embryos for protein synthesis duringembryogenesis. Our results show that: (i) the level of UBR4 in theYS mesoderm is nondetectible, (ii) UBR4 marks the YS endo-derm, (iii) UBR4 forms cytosolic punctate structures that coloc-alize with LC3 puncta, (iv) UBR4 puncta have size and shapedistinct from those of LC3 puncta, (v) UBR4-deficient YSs showan increased level of puncta of LC3 and ATG12, and (vi) UBR4 isfound within the MVB in cultured cells. Taken together, it is likelythat misregulation in autophagic degradation of cargoes (includingmaternal proteins) causes stress in the YS endoderm, which inturn affects vascular development in the YS mesoderm.

UBR4 Subpopulation Is Degraded by Autophagy Through Its Associationwith Autophagic Cargoes.As UBR4 is a huge protein with a size of570 kDa, a portion of UBR4 stably expressed in HEK293 cellsmay be deposited to autophagic vacuoles as misfolded aggregates.Nonetheless, several lines of evidence suggest that the majorityof UBR4 molecules are first associated with (unidentified) cel-lular cargoes to form UBR4–cargo complexes destined to au-tophagic vacuoles. First, the majority of UBR4 in HEK293–UBR4V5 cells is retrieved from cytosolic and ER membranefraction and can be solubilized with a nonionic detergent (Fig.S6). Second, the EM staining study of UBR4 invariably revealedone or a few individually isolated dots (not as aggregates) as-sociated with cellular cargoes or autophagic vacuoles (Fig. 3 G

and H). Third, transient expression of UBR4V5 resulted in theformation of UBR4V5+LC3− puncta both in mouse embryonicfibroblasts (MEFs) (Fig. S8) and HEK293 cells (Fig. S9), whichare likely to represent UBR4V5+ cellular cargoes that have notbeen delivered to autophagic vacuoles. Fourth, cytosolic punctapositive for “endogenous” UBR4 in the visceral YS show shapesand sizes distinct from those of ATG12, ATG5, and LC3 (Fig. 3D–F, arrowheads). In addition, a confocal distance measurementindicated that the center of UBR4 puncta is separated from thatof LC3 puncta by ∼0.15 μm. Fifth, endogenous UBR4 is a rela-tively stable protein in MEFs (Fig. 3N), and its autophagic deg-radation is accelerated by nutrient deprivation (Fig. 3N). A proteinprone to misfolding is expected to be constitutively degraded.

SI Materials and MethodsConstruction of Mice Lacking UBR4.We cloned the UBR4 gene froma bacterial artificial chromosome (BAC) library (CITB; In-vitrogen) derived from the mouse embryonic stem (ES) cell lineCJ7 using a mouse UBR4 cDNA fragment (nucleotides 4231–5430) as a probe. The interval ribosome entry site (IRES)-taulacZ-pA-ACNF plasmid, derived from the IRES-taulacZ-pAcassette (1) and the ACNF cassette (2), is a gift from PeterMombaerts (Rockefeller University, New York). The targetingvector pUBR4KO-tauLacZ was constructed using gene re-combineering (recombination-mediated genetic engineering) asdescribed (3, 4), linearized with FseI, and electroporated intoE14 ES cells (a gift from Peter Mombaerts). Targeted ES cellclones were analyzed using Southern blotting and PCR analysis.The clone 2F12 ES cells were injected into C57/BL6J blastocyststo generate chimeric mice, which, in turn, were used to obtaingerm-line transmission of the UBR4− mutant allele.

Antibodies. Rabbit polyclonal anti-human UBR4 antibody (1:400),raised against residues 3755–4160 (5), was used for immunostainingof the YS. Rabbit polyclonal anti-UBR4 antibody (1:300, IHC-00640; Bethyl Laboratories) was used for immunostaining of cul-tured cells. Other primary antibodies are rabbit polyclonal anti-V5(1:500, V8137; Sigma), goat polyclonal anti-ATG12 (1:100, SC-70128; Santa Cruz Biotechnology), goat polyclonal anti-ATG5(1:200, SC-8666; Santa Cruz Biotechnology), rat anti-BrdU (1: 100,H9367; Accurate Chemical & Scientific), mouse monoclonal anti–α-SMA (1:400, A2547; Sigma), rat monoclonal anti–PECAM-1(1:300, MEC13.3; Pharmingen), and goat polyclonal anti-LC3(1:100, SC-16755; Santa Cruz Biotechnology). Antibodies for im-munoblotting analysis are anti-LC3 (1:2,000, L7543; Sigma), p62(1:5,000, GP92-C; Progen), GABARAP (OSG00009W; ThermoScientific, 1:1,000), V5 (1:2,000; Invitrogen), and β-actin (1:100,000,A1978; Sigma). The following secondary antibodies are from In-vitrogen: Alexa Fluor 488 goat anti-rabbit IgG (1:200, A11034),Alexa Fluor 555 donkey anti-goat IgG (1:200,A21432), AlexaFluor555 goat anti-mouse IgG (1:200, A21424), and anti-rat IgG-HRP(1: 500; Jackson ImmunoResearch).

Reverse Transcriptase-PCR Analysis. Total RNA was prepared froma pool of five E9.5 embryos for each of +/+ and UBR4−/− gen-otypes using TRIzol (Invitrogen) and RNeasy kit (Qiagen). Thefirst cDNA strand was synthesized using SuperScript III reversetranscriptionase (Invitrogen) and hexamer random oligos. Re-verse transcription for quantitative PCR was performed using theiScript cDNA synthesis kit (Bio-Rad). Quantitative PCR was per-formed in duplicates by using Platinum SYBR Green (Invitrogen)

Tasaki et al. www.pnas.org/cgi/content/short/1217358110 1 of 7

www.pnas.org/cgi/content/short/1217358110

with ABI 7300/7500 real-time PCR system and primer sequenceslisted in Table S2.

Histology and β-Galactosidase Staining. For histological analysis,embryos were fixed overnight at 4 °C in 4% paraformaldehyde inPBS. Embryos were treated with 70% ethanol, dehydrated,embedded in paraffin wax, and sectioned transversely or sagit-tally with 7 μm thickness, followed by staining with hematoxylinand eosin (H&E). To detect the activity of β-galactosidase onsections, embryos or tissues were fixed in 0.25% glutaraldehydefor 5 min, rinsed in PBS three times, and stained overnight at37 °C in X-gal solution (1.3 mg/mL potassium ferrocyanide, 1 mg/mL potassium ferricyanide, 0.3% Triton X-100, 1 mMMgCl2, 150mM NaCl, and 1 mg/mL 4-choloro-5-bromo-3-indolyl-β-galacto-side (X-gal; Roche Applied Science) in PBS (pH 7.4), followedby postfixation.

Immunohistochemistry. For immunohistochemistry of animal tis-sues, paraffin-embedded slides were treated with xylene, followedby gradual dehydration in EtOH (100%, 90%, 80%, and 70%; 6min each) and rehydration in water for 20 min. The slides weretreated with blocking solution (1% BSA and 0.2% Triton X-100in PBS) for 1 h and incubated with primary antibodies and sub-sequently secondary antibodies. Confocal images were taken witha Fluoview FV1000 confocal laser scanning microscope equippedwith Olympus Plan-Apo 60× (1.42 NA) oil immersion lens andanalyzed using Fluoview software (Olympus). Two ImageJ (1)plugins, colocalization finder (by Christophe Laummonerie andJerome Mutterer) and JACoP (Just Another Colocalization Plu-gin, 2), were used to evaluate colocalization of fluorescent images.

Immunoelectron Microscopy. HEK293–UBR4V5 cells were fixedin 2% PFA in PBS, pH 7.4, for 1 h and washed with PBS. Fixedcells were collected by scraping and resuspended in 3% gelatin.The gelatin was solidified on ice, and cells embedded in gelatinwere fixed again with 2% PFA for 15 min. After cryoprotectedwith 2.3 M sucrose with PVP solution overnight, samples werefrozen in liquid nitrogen and trimmed into 0.5-mm cubes. Thefrozen cell blocks were sectioned by cryomicrotome (Leica EMCrion) in 70-nm sections and collected on 200-mesh grids. Pri-mary antibodies were diluted in 0.1 M PBS with 0.5% BSA and0.15% glycine. Donkey anti-rabbit 12 nm gold beads and donkeyanti-goat 18 nm gold beads were diluted to 1:25 before use. Afterimmunostaining, sections were incubated with 2.5% glutalaldehydefor 10min andwith 2%neutral UA acetate for 7min. Sections werefurther processed with 4% uranyl acetate and methyl cellulose forcontrasting and drying, respectively. After drying, samples wererecorded using JEOL JEM1011 TEM with a high-resolution AMTdigital camera.

Establishment of UBR4−/− MEFs. Primary MEFs were establishedfrom E8.5 UBR4−/− embryos and their +/+ littermates. The em-bryos were minced by pipetting in Iscove’s modified Dulbecco’smedium (IMDM) (31980–022; Invitrogen), 15% FBS (HyClone),

0.1 mM nonessential amino acids (Invitrogen), 0.1 mM β-mer-captoethanol, and 100 units/mL penicillin/streptomycin (In-vitrogen). The cells were seeded on the gelatinized 35-mm culturedish. Permanent cell lines were established from primary MEFsthrough crisis-mediated immortalization over 10 passages.

Proximity Ligation Assay. The proximity ligation assay (PLA) wasused to visualize protein–protein interactions in UBR4V5 stableHEK293 cells using a reagent kit (Olink Bioscience). The cellswere fixed with 4% paraformaldehyde for 0.5 h and blocked for1 h in 10% goat serum (Invitrogen), followed by incubationwith UBR4 and test antibodies (ATG12 or LC3) in antibodysolution (0.3% Triton X-100 and 5% BSA in PBS) overnight at4 °C. The cells were incubated with two oligonucleotide-taggedprobe antibodies against primary antibodies. Annealing of theprobes occurs when the target proteins are within 400 Å fromeach other, which initiates the amplification of green fluo-rophore. Following a 1-h incubation with the PLA probes, thecells were subject to hybridization, ligation, and DNA amplifi-cation in accordance with the manufacturer’s instruction. Az-stack image of the green PLA signals was obtained by com-bining 20 focal planes (0.53 μm/slice) using a confocal micro-scope (Olympus; FV1000).

In Vivo Proliferation Assay. In vivo proliferation assay was per-formed as described. Pregnant mice were intraperitoneallyinjected (150 mg/g) with 5-bromo-2-deoxyuridine (BrdU) in 250μL saline. After 2 h postinjection, embryos were subjected toparaffin sectioning and immunostaining BrdU (S phase marker).

Knockdown of UBR4. HEK293 cells were transfected with control(4390843) or UBR4 siRNA (4392420, ID 23628) from LifeTechnologies at a final concentration of 10 nM using Lipofect-amine RNAiMAX reagent (13778150). After 48 h post-siRNAtransfection, the cells were transfected with pcDNA6.2-cluv-hUBR4V5 and/or vector plasmid using Lipofectamine LTX(15338100). After 24 h postplasmid transfection (total 72 h aftersiRNA treatment), the cells were subjected to immunocyto-chemistry (see below) or immunoblotting using RIPA buffer (150mM NaCl, 50 mM Tris·HCl, pH 8.0, 1% Nonidet P-40, 0.5%sodium deoxycholate, and 0.1% SDS) supplied with a proteaseinhibitor mixture (1:100; Sigma; P8340).

Immunocytochemistry of Cultured Cells.Coverslips (22 mm2) placedin a six-well plate were incubated with diluted poly-L-lysine (1:10in sterile deionized water) for 30 min at room temperature. Afterwashing, the plate was exposed under a UV radiation lampovernight in a hood. Cells were cultured and fixed in 4% para-formaldehyde in PBS (pH 7.4) for 0.5 h at room temperature.After washing twice with PBS, the cells were treated withblocking solution (5% FBS and 0.3% Triton X-100 in 0.1 MPBS) for 1 h, followed by incubation with primary and, sub-sequently, secondary antibodies. To induce starvation, the cellswere washed twice with PBS and incubated in HBSS.

1. Mombaerts P, et al. (1996) Visualizing an olfactory sensory map. Cell 87(4):675–686.2. Bunting M, Bernstein KE, Greer JM, Capecchi MR, Thomas KR (1999) Targeting genes

for self-excision in the germ line. Genes Dev 13(12):1524–1528.3. Liu P, Jenkins NA, Copeland NG (2003) A highly efficient recombineering-based

method for generating conditional knockout mutations. Genome Res 13(3):476–484.

4. Warming S, Costantino N, Court DL, Jenkins NA, Copeland NG (2005) Simpleand highly efficient BAC recombineering using galK selection. Nucleic Acids Res33(4):e36.

5. Shim SY, et al. (2008) Protein 600 is a microtubule/endoplasmic reticulum-associatedprotein in CNS neurons. J Neurosci 28(14):3604–3614.

Tasaki et al. www.pnas.org/cgi/content/short/1217358110 2 of 7

www.pnas.org/cgi/content/short/1217358110

Fig. S1. The UBR box N-recognin family of the mammalian N-end rule pathway. AI, autoinhibitory domain; CRD, cystein-rich domain; Fbox, F-box; HECT,homologous to the E6-AP carboxyl terminus; N, N-domain; PABC, poly(A) binding protein C-terminal domain; PHD, plant homeodomain finger; RING, RINGfinger; UBA, ubiquitin association domain; UBR, UBR box.

Fig. S2. Structure and function of the visceral yolk sac of murine embryos. (A) Developing structure of mouse embryo at E8.5. The visceral yolk sac functions asan early placenta, which exchanges nutrient and waste between E5.0 and E10 in mice. The yolk sac is also the first site where precursors of the erythroid celllineage develop. (B) Illustration of layers of the visceral yolk sac (VYS) in mice. Maternal proteins are absorbed via microvilli on the apical surface of theendodermal cell layer of the VYS. Endocytosed maternal proteins are transferred to the lysosome through the early and late endosomes. In the lysosome,the maternal proteins are digested and secreted into the blood vessels in the mesenchymal cell layer. Digested amino acids are delivered to the embryo via thevitelline circulation. B, blood vessel; H, hematopoietic cell; L, lysosome; M, mesenchymal cell; N, nucleus.

Tasaki et al. www.pnas.org/cgi/content/short/1217358110 3 of 7

www.pnas.org/cgi/content/short/1217358110

Fig. S3. Targeting strategy of the UBR4 gene using recombineering. (A) Restriction map of the ∼7.5-kb 5′-proximal region of the ∼108-kb mouse UBR4 genespanning exons 27 through 53. (B) Targeting vector. (C) Targeted UBR4− allele in ES cells. (D) Targeted UBR4− allele in F1 pups. Exons are denoted by solidvertical rectangles. Directions of transcription of the neomycin (neo) and the thymidine kinase (tk) genes are indicated. Homologous recombination resulted inthe replacement of the UBR4 exons 36–42 with the neo cassette. Exons 36 and 37, marked by an asterisk, contain Gly and Asp that are conserved in the UBR boxN-recognin family and are essential for binding to type-1 destabilizing N-terminal residues in Saccharomyces cerevisiae Ubr1. Primers used for PCR genotypingare indicated by solid rectangles.

Fig. S4. Construction of UBR4-deficient mice. (A) Southern analysis of BamHI-cut DNA from the ES cell clones 1D11, 2F12, and 1D2. The 5′-probe detects 16.7kb (wild-type) and 8.1 kb (UBR4− allele) diagnostic fragments. The 3′-probe detects 16.7 kb (wild-type) and 5.0 kb (UBR4− allele) diagnostic fragments. (B) PCRanalysis of mouse tail DNA. (C) Semiquantitative PCR analysis of UBR4 mRNA in +/+ (WT) and UBR4−/− (KO) embryos. Total RNA was isolated from embryonicbodies and YS.

Tasaki et al. www.pnas.org/cgi/content/short/1217358110 4 of 7

www.pnas.org/cgi/content/short/1217358110

Fig. S5. Defective YS vascular development of UBR4-deficient embryos. (A) TUNEL assay of +/+ and UBR4−/− YS at E9.5. ee, extraembryonic endoderm; mes,mesodermal layer. Arrowhead, TUNEL+ cells in a positive control experiment. (B) Whole mount PECAM-1 staining of control and UBR4−/− YS at E9.5.Arrowheads indicate some of the differences in the branching patterns, relative to wild-type embryos. (Scale bars, 100 μm.) (C) Whole mount PECAM-1 staining ofheads of +/+ and UBR4−/− embryos at E9.5. Arrowhead, the intracranial artery. (D) Real-time PCR analysis of PECAM-1, α-SMA, and AFP in +/+ andUBR4−/− YS at E9.5.

Fig. S6. UBR4 is a soluble protein. (A) Fractionation strategy. HEK293–UBR4V5 cells were harvested after starvation or control treatment and proteins andcellular proteins were fractionated by a differential centrifugation as described. (B) Western blot analysis. After fractionation, solubilized proteins wereseparated by SDS/PAGE and specific proteins were detected by anti-V5, anti-UBR4, and antimitochondrial proteins. (C) Extraordinarily large-sized UBR4 proteincan be degraded through autophagy upon starvation without forming aggregates. Immunoblotting of soluble and insoluble proteins from MEFs in normalmedia or HBSS for 90 min. Cells were lysed using 1% Nonidet P-40 and insoluble proteins were separated by centrifugation. Results of 1% Nonidet P-40 solubleare also shown in Fig. 3M. Nonspecific (ns) signals were taken from anti-UBR4 blots. Cells were lysed by adding cold 1% Nonidet P-40 containing buffer withprotease inhibitors and total lysate was collected by scraping. Insoluble fraction was separated by 12,000 × g centrifugation and resuspended with SDS-samplebuffer before denaturation.

Tasaki et al. www.pnas.org/cgi/content/short/1217358110 5 of 7

www.pnas.org/cgi/content/short/1217358110

Fig. S7. The function of UBR4 in autophagy of cultured cells and its turnover by autophagy. (A–C) Stably expressed UBR4V5 induces the formation of punctapositive for ATG5, ATG12, and LC3. HEK293 and HEK293–UBR4V5 cells were immunostained for ATG5 (A), ATG12 (B), and LC3 (C). (Scale bars, 5 μm.) (D)Cyclohexamide-chase assay of UBR4 in MEFs. Cells were cultured in normal or starvation medium for 6 h in the presence of cycloheximide. Shown is a rep-resentative blot image from three independent experiments.

Fig. S8. Expression and conversion of LC3 in +/+ and UBR4-deficient MEFs transiently expressing UBR4V5. (A) Immunofluorescent analysis of V5 (UBR4V5) andLC3 in +/+ and UBR4-deficient MEFs transiently transfected with a plasmid encoding UBR4V5. Wild-type and UBR4-deficient MEFs were transfected witha plasmid (1.0 μg) encoding UBR4V5. One day after transfection, both transfected (TF) and nontransfected (NTF) cells were trypsinized, mixed, and replated onsix-well dishes. This procedure minimizes individual variations between transfected and nontransfected cells. Note that LC3-I is accumulated in UBR4−/− MEFsgrowing in normal media, compared with +/+ MEFs, and that transient expression of UBR4V5 in UBR4−/− MEFs only marginally reduces the synthesis of LC3-Iand the formation of LC3 foci. This marginal effect is consistent with a possibility that autophagic induction in UBR4−/− cells is caused by cellular stress and/orimpairment of organelles that cannot be readily rescued by transient supplement of UBR4 for less than 24 h. Also note that transient expression of UBR4V5 inwild-type MEFs results in the formation of UBR4V5+LC3− puncta, which are likely to represent UBR4V5+ cellular cargoes that have not been delivered toautophagic vacuoles. (Scale bar, 20 μm.) (B) Enlarged views of the cellular areas indicated by a–d. (a) Nontransfected wild-type cell. (b) Transfected wild-typecell. (c) Nontransfected UBR4-deficient cell. (d) Transfected UBR4-deficient cell. (Scale bar, 4 μm.) (C) Immunoblotting analysis of V5 (UBR4V5) and LC3 in +/+and UBR4-deficient MEFs transiently transfected with the indicated amounts of a plasmid encoding UBR4V5.

Tasaki et al. www.pnas.org/cgi/content/short/1217358110 6 of 7

www.pnas.org/cgi/content/short/1217358110

Fig. S9. Knockdown of UBR4 induces the formation of LC3 puncta in HEK293 cells. (A) HEK293 cells were transfected with control or UBR4 siRNA. After 48 hpost-siRNA transfection, the cells were transfected with 1 μg pcDNA6.2-cluvhUBR4V5. After 24 h postplasmid transfection, the cells were subjected to im-munocytochemistry of V5 and LC3. (Scale bar, 5 μm.) (B) Immunoblotting analysis of HEK293 cells that had been treated with UBR4 siRNA for 48 h, followed bytransient transfection of the plasmid pcDNA6.2-cluvhUBR4V5. Note that overexpression of UBR4V5 in UBR4 knockdown cells results in disappearance ofendogenous human UBR4 and accumulation of recombinant human UBR4V5.

Table S1. Genotyping of embryos from UBR4+/− intercrosses

Stage +/+ +/− −/− ND No. of litters Averge litter size

E8.5 27 39 22 10 8.8E9.0 13 23 12* 5 9.6E9.5 43 107 40* 3 23 8.4E10.5 7 17 6* 3 4 8.3E11.5 6 8, 2† 3† 2 6.0E12.5 3 5 0 1 8.0E15.5 1 2 0 1 3.0P21 65 138 0 35 5.8

ND, not determined.*Morphologically abnormal.†Found dead.

Table S2. Primers used for PCR analysis

Gene Primer Sequence

PECAM-1 PECAM-1–F GAGCCCAATCACGTTTCAGTTTPECAM-1–R TCCTTCCTGCTTCTTGCTAGCT

β-Actin β-ACTIN–F TGTGATGGTGGGAATGGGTCAGAAβ-ACTIN–R TGTGGTGCCAGATCTTCTCCATGT

AFP Afp–F GGAAGCATGTTAAATGAGCATGAfp–R AGGGCCAGCTTCTGAAT

α-SMA ACTA2–F ATTGTGCTGGACTCTGGAGATGGTACTA2–R TGATGTCACGGACAATCTCACGCT

Tasaki et al. www.pnas.org/cgi/content/short/1217358110 7 of 7

www.pnas.org/cgi/content/short/1217358110