vascularized and functional human liver from an ipsc-derived organ bud transplant

7/24/2019 Vascularized and Functional Human Liver From an IPSC-Derived Organ Bud Transplant

1/5

LETTER doi:10.1038/nature12271

Vascularized and functional human liver from aniPSC-derived organ bud transplantTakanori Takebe 1,2 , Keisuke Sekine 1, Masahiro Enomura 1, Hiroyuki Koike 1, Masaki Kimura 1, Takunori Ogaeri 1, Ran-Ran Zhang 1,Yasuharu Ueno 1, Yun-Wen Zheng 1, Naoto Koike 1,3 , Shinsuke Aoyama 4 , Yasuhisa Adachi 4 & Hideki Taniguchi 1,2

A criticalshortage ofdonor organsfortreatingend-stageorganfailurehighlightsthe urgentneed forgeneratingorgansfromhuman inducedpluripotentstem cells(iPSCs) 1 . Despitemanyreportsdescribingfunc-tionalcell differentiation 24 , nostudieshave succeeded ingeneratinga three-dimensional vascularizedorgansuch as liver.Here we show thegeneration of vascularized and functional human liver from humaniPSCs by transplantation of liver buds created in vitro (iPSC-LBs).Specified hepatic cells (immature endodermal cells destined to track the hepatic cell fate) self-organized into three-dimensional iPSC-LBsby recapitulating organogenetic interactions between endothelial andmesenchymal cells 5 . Immunostaining and gene-expression analysesrevealed a resemblance between in vitro grown iPSC-LBs and in vivo liver buds. Human vasculatures in iPSC-LB transplants became func-tional by connecting to the host vessels within 48 hours. The forma-tion offunctionalvasculaturesstimulatedthe maturationof iPSC-LBsinto tissue resembling the adult liver. Highly metabolic iPSC-derivedtissue performed liver-specific functions such as protein productionand human-specific drug metabolism without recipient liver replace-ment 6 . Furthermore,mesenteric transplantationof iPSC-LBs rescuedthe drug-induced lethal liver failure model. To our knowledge, this isthefirst report demonstrating thegenerationof a functionalhumanorgan from pluripotent stem cells. Although efforts must ensue totranslate these techniques to treatments for patients, this proof-of-concept demonstration of organ-bud transplantation provides a promising new approach to study regenerative medicine.

Since the discovery of embryonic stem cells in 1981, decades of laboratorystudieshave failed to generate a complex vascularizedorgansuch as liver from pluripotent stem cells, giving rise to the prevailing belief that in vitro recapitulation of the complex interactions among cells and tissues during organogenesis is essentially impractical7. Herewe challenge this idea by focusing on the earliest process of organo-genesis, that is, cellular interactions during organ-bud development.

During early liver organogenesis, newly specified hepatic cells delami-nate fromtheforegutendodermal sheetandforma liverbud 8,acondensedtissue mass that is soon vascularized. Such large-scale morphogeneticchanges depend on the exquisite orchestration of signals between endo-dermal epithelial,mesenchymal and endothelial progenitorsbefore bloodperfusion5. These observations led us to propose that three-dimensionalliver-bud formation can be recapitulated in vitro by culturing hepaticendoderm cells with endothelial and mesenchymal lineages (Fig. 1a). Toexamine this hypothesis, we first prepared hepatic endoderm cells fromhuman iPSCs(iPSC-HEs)by directed differentiation, producingapproxi-mately 80% of the treated cells expressed the hepatic marker HNF4A,which is involved in cell fate determination (Supplementary Fig. 1).

Next, to recapitulate early organogenesis, human iPSC-HEs were cul-tivated with stromal cell populations; human umbilical vein endothelialcells (HUVECs) and human mesenchymal stem cells (MSCs) unless sta-ted otherwise, because of their primitive nature (Supplementary Fig. 2aand Supplementary Discussion). Notably, although cells were plated in

two-dimensional conditions, human iPSC-HEs self-organized intomacroscopically visible three-dimensional cell clusters by an intrinsicorganizing capacity up to 48 h after seeding (Fig. 1 b, c, Supplementary

1 Department ofRegenerativeMedicine,YokohamaCity UniversityGraduate School ofMedicine,3-9 Fukuura, Kanazawa-ku,Yokohama,Kanagawa236-0004,Japan. 2 AdvancedMedicalResearch Center,Yokohama City University, Yokohama, Kanagawa 236-0004, Japan. 3 Department of Surgery, Seirei Sakura Citizen Hospital, 2-36-2 Ebaradai, Sakura, Chiba 285-8765, Japan. 4 ADME & Tox. ResearchInstitute, Sekisui Medical Company Ltd., Tokai, Ibaraki 319-1182, Japan.

H u m a n

i P S C - H

E o n

l y

H u m a n

i P S C - H

E H U V E C

h u m a n

M S C

a

e

d

Self-organization intothree-dimensional liver bud

Recapitulation oforganogenetic interactions

Human MSCs

HUVECs

Hepatic speci cationHuman

iPSCs

Ectopic transplantationinto immunode cient mice

CraniumMesenterySubcapsular

46-day culture Bloodperfusion

c

b

H u m a n

i P S C - H

E H U V E C h u m a n

M S C

H u m a n

i P S C - H

E H U V E C h u m a n

M S C

0 h 12 h 24 h 36 h 72 h

0.1

1

10

AFP

0.1

1

10

100

1,000 ALB

0.01

0.1

1

10

iPSC-HEs

Control iPSC-LBs

iPSC-HEs

Control iPSC-LBs

iPSC-HEs

Control iPSC-LBs

iPSC-HEs

Control iPSC-LBs

iPSC-HEs

Control iPSC-LBs

iPSC-HEs

Control iPSC-LBs

iPSC-HEs

Control iPSC-LBs

iPSC-HEs

Control iPSC-LBs

HNF4

0.1

1

10

100 RBP4

0.1

1

10

100 CYP3A7

0.1

1

10

100 G6PC

0.1

1

10

100 FOXA2

0.1

1

10

100 TTR

R e l a t

i v e e x p r e s s

i o n s

( f o l d c h a n g e

)

HumaniPSC-HEs

Figure 1 | Generationof human liver buds from human iPSCs. a , Schematicrepresentation of our strategy. b, Self-organization of three-dimensional humaniPSC-LBs in co-cultures of human iPSC-HEs with HUVECsandhuman MSCs (seealso Supplementary Video 1).Thetime-lapse fluorescence imaging of human iPSC-HEs is shown here. Scale bar, 5mm. c, Gross observation of human iPSC-LBs (toppanel)andconventional two-dimensional cultures (bottom panel). Scale bar,1 mm.d, Presence of human iPSC-HEs and nascent endothelial networks inside humaniPSC-LBs.Green,humaniPSC-HE orHUVEC;red, HUVEC or human MSC. Scalebars, 100mm. e, Quantitative PCR analysis of hepatic marker gene expression inhuman iPSC-LBs at day 6 of culture. Control samples were iPSC-HE:HUVEC:humanMSC cells that had been grown as separate cell types and not culturedtogether,thenforcontrolgeneexpressionanalysis thecellswere mixedtogetherat thesame ratio of cell types as in human iPSC-LBs. Results representmean 6 s.d.,n 5 8.

2 5 J U LY 2 0 1 3 | V O L 4 9 9 | N AT U R E | 4 8 1Macmillan Publishers Limited. All rights reserved201 3
http://www.nature.com/doifinder/10.1038/nature12271http://www.nature.com/doifinder/10.1038/nature12271


2/5

Fig. 2b, c and Supplementary Video 1). The presumed human iPSC-derived liver buds (iPSC-LBs) were mechanically stable and could bemanipulatedphysically. Wevisualizeda formationofendothelialnetwork and homogenously distributed human iPSC-HEs by using fluorescent-protein-labelledcells(Fig.1d).Althoughhuman iPSC-LBsarea tri-lineagemixed tissue and difficult to compare directly with human iPSC-HEs,quantitative polymerase chain reaction (PCR) analysis revealed that cellsin human iPSC-LB hadsignificantly increased expression of early hepatic

marker genes such as alpha-fetoprotein ( AFP ), retinol binding protein 4(RBP4), transthyretin ( TTR) and albumin ( ALB) (Fig. 1e)4. Microarray profilingshowedthat FGFandBMPpathwayswereupregulated highlyinhuman iPSC-HEs when co-cultured with stromal cells, suggesting theinvolvement of stromal-cell-dependent factors in liver-bud formation(Supplementary Fig. 3a). Consistent with this, loss- and gain-of-functionexperiments suggested that, in addition to the direct cell-to-cell interac-tions, stromal-cell-dependent paracrine support is essential for three-dimensional liver-bud formation through the activating FGF and BMPpathways (Supplementary Fig. 3bi and Supplementary Discussion) 9.

Human liver-bud formation is initiated on thethird or fourth week of gestation, and this corresponds to embryonic day 9.5 (E9.5) to E10.5 formouseliver-bud formation 10 . Similar toE10.5mouse liverbud, immuno-histochemistryshowed thathumaniPSC-LBis composedofproliferating

AFP-positive hepatoblasts11

as well as mesenchymal and endothelial

progenitors (Fig. 2a). Hepatic cells in human iPSC-LBs were as prolif-erative as E10.5 mouse liver buds (Fig. 2b). Flow cytometric character-ization revealed that 42.7 6 7.5% (n 5 6) of cells in human iPSC-LBswere identified as iPSC-HEs using fluorescence-labelled HUVECs andMSCs.Amongthese, approximately71.9 6 7.3%(n 5 6)ofhuman iPSC-HEs expressed ALB and 29.96 2.8% expressed AFP, and 19.36 4.5%were positive for both ALB and AFP (Fig. 2c).

To characterize the expression profiles of human iPSC-LBs and to

compare withthoseof a correspondingdevelopmental stage, we carriedoutmicroarrayanalysisof 83selectedgenesthatare seriallyupregulatedduring liver development. Hierarchical clustering analyses suggestedthat the expression profiles of human iPSC-LBs at day4 of cultureresembled those of mouse E10.5 and E11.5 liver buds rather thanadvanced fetal or adult livers (Fig. 2d). These expression profiles wererelatively similar to those of human fetal liver cell-derived liver buds(FLC-LBs) (Fig. 2d), which also have an ability to form LBs (Sup-plementary Figs 2a and 4). The 83-gene expression profile of humaniPSC-LBs showed closer signatures to more advanced human liver

19.3 4.5%(6.1 1.4%)

46.1 6.3%(14.5 2.0%)

ALB Log APC intensity

A F P L o g

F l T C i n t e n s

i t y

a

d

Human iPSC-derived liver budCK8/18 AFP DAPI CK8/18 CD31 DAPI

CK8/18 Desmin DAPI AFP PCNA DAPI

CK8/18 AFP DAPI CK8/18 Flk1 DAPI

CK8/18 Desmin DAPI AFP BrdU DAPI

E10.5 murine liver bud

29.9 2.8%

(9.4 1.9%)

AFP Log APC intensity

ALB Log APC intensityGFP-HUVEC

60

40

20

0 K u s a b

i r a O r a n g e

h u m a n

M S C

46.4 8.0%

10.9 0.9%

42. 7 7.5%

cb

P r o

l i f e r a t

i n g

h e p a t o b

l a s t n u m

b e r

( % )

i P S C - L

B E 1

0 . 5 E 1

5 . 5 E 1

7 . 5 8 w

Mouse E9.5 LT

Mouse E10.5 LT

Human iPSC-LB

Human FLC-LB

Mouse E11.5 LT

Mouse E13.5 LT

Mouse E15.5 LT

Mouse E17.5 LT

Mouse E19.5 LT

Mouse P0 LTMouse P3 LT

Mouse P8 w LT

Human ALT (30 yr)

71.9 7.3%1.9 7.3(20.4 1.9%)20.4 1.9 )

71.9 7.3%(20.4 1.9%)

28.3 9.1%8.3 9.1(8.9 2.9%)8.9 2.9 )28.3 9.1%(8.9 2.9%)

6.3 2.2%.3 2.2(2.0 0.7%)2.0 0.7 )6.3 2.2%(2.0 0.7%)

Figure 2 | In vitro characterization of human iPSC-LBs. a , Immunostaining of CK8/18, AFP, PECAM1 (CD31), FLK1, desmin, PCNA and BrdU(5-bromodeoxyuridine). Scale bars, 100 mm. b, Proportions of proliferating hepatic cells, as assessed by dividing the number of PCNA-positive orBrdU-positive cells by the number of CK8/18-positive cells (shown as apercentage).Results represent means 6 s.d.,n 5 3.c, Representativeflow cytometry profile showing the average number of AFP-positive and/or ALB-positive humaniPSC-HEs at day 4 of culture in six independent differentiation experiments.Human iPSC-HEs were separated from stromal populations by the use of fluorescence labelled cells. Averagepercentagesof the total cells ands.e.m.aregivenin brackets. d, Comparison of liver developmental gene signatures among humaniPSC-LB,humanFLC-LB, human adult (30yearsold)livertissue(ALT)and mouselivertissue (LT) of various developmentalstages (fromE9.5 to 8 weeksafter birth).

3 d 7 d 14 d

V a s c u

l a r a r e a

( % )

g

b

a

0 dHuman iPSC-HE HUVEC human MSC

ihDextran

Dextran

ed f

*

F u n c

t i o n a

l v e s s e

l l e n g t

h ( m m

)

H u m a n v e s s e l

l e n g

t h ( m

)

5,000 60

H U V E C

M S C

H u m a

n i P S C - L

B T x

H U V E C M

S C

H u m a n

i P S C - L B T

x A d u l t l i

v e r

7

6

5

80

70

60

50

40

30

20

10

0

4

3

2

1

0

40

20

10

0

4,000

3,000

2,000

1,000

0 d 3 d 7 d

1 5 d

4 5 d

6 0 d

0

B r an

ch

p oi nt s

~~

*

NS

0 d 2 d 3 d

Human iPSC-HEHUVEC

human MSC Dextran

c

Human iPSC-HEHUVEC human MSC

mouse CD31

Human iPSC-HEHUVEC

human MSC

Human iPSC-HEHUVEC

human MSC

H u m a n

i P S C - L

B T x

M o u s e a d u l

t

l i v e r

Figure 3 | Generation of human liver with functional vascular networks in vivo. a , Macroscopic observation of transplanted human iPSC-LBs, showing perfusion of human blood vessels. Dotted area indicates the transplanted humaniPSC-LBs. b, Intravital tracking of human iPSC-LBs, showing in vivo dynamics of vascularization. c, Dextran infusion showing the functional human vesselformation at day3. Scale bar, 500mm. d, Visualization of the connections (arrows)amongHUVECs(green)andhost vessels(blue).Scalebar,250 mm. e,f ,LocalizationofhumanMSCs orhuman iPSC-derived cellsat day15.Scale bars,100 and250 mm.g , Quantification of human vessels over time (mean 6 s.e.m., n 5 3). Error barsattached to the bars relate to the left axis (vessel length), error bars attached tosquares relate to the right axis (branch points). h, Comparison of functional vessellength between human iPSC-LB and HUVEC human MSC transplants (Tx)(mean 6 s.e.m.,n 5 5,*P , 0.01). i, Vascularnetworksof human iPSC-LB-derivedtissue is similar to that of mouse adult livers (mean 6 s.e.m., n 5 5, *P , 0.01).

RESEARCH LETTER

4 8 2 | N AT U R E | V O L 4 9 9 | 2 5 J U LY 2 0 1 3Macmillan Publishers Limited. All rights reserved201 3


3/5

tissues derived from fetuses and adults 22 to 40 weeks into gestationthan did human iPSC-derived mature hepatocyte-like cells (humaniPSC-MHs) induced by published protocols 4. This reflectsthe efficienthepatic commitment in liver buds as they mimicked ontogeneticinteractions consistent with higher albumin production capacity afterextended culture (Supplementary Figs 5 and 6).

Haemodynamic stimulation is essential for liver-bud maturation 12 . Totest whether human iPSC-LBs were capable of generating completely functional liver, we used a cranial window model because of the opticalaccess13 . We first tested the feasibility of this model by using mouse liver-bud-derived cells (Supplementary Discussion andSupplementary Fig. 7).To track the in vivo fate of transplanted cells, we performed repeated liveimaging ofliver-budtransplant-derivedtissueover time.Notably, invitro -derived human iPSC-LBs connectedquickly withhost vasculature within48h of transplantation (Fig. 3a, b, Supplementary Fig. 8 and Supplemen-tary Videos 2 and 3). By infusing fluorescein-conjugated dextran and

mouse CD31 antibody, we showed that human blood vessels within thetransplantbecame patent(unobstructed) byconnectinghost vessels at theedge ofthetransplant (Fig. 3c,d), andthis wasconfirmed bywhole-mount

immunostaining of the explants (Supplementary Figs 9 and 10a, b, andSupplementary Video 4). Functional vessel formation was essential forthe successful engraftment because human iPSC-HEs transplanted with-out endothelial cells failed to vascularize and engraft (Supplementary Fig.10cf). Human vessels were long-lastingas they were stabilized byhumanMSC-derived perivascular cells (Fig. 3eg)13,14 . Interestingly, the humaniPSC-LBtransplantvascularnetworks were similar indensityand morph-ology to those of adult livers, whereas the vasculature in only HUVECsand human MSC transplants were much less dense; functional vesselswere ofsimilar diameter (approximately12 mm)inbothsettings (iniPSC-LB transplants, and in HUVEC and human MSC transplants) (Fig. 3h, i,andSupplementary Figs 11 and12). Animal studies suggest that endothe-lial cellsnotonlyform passive conduitstodelivernutrientsandoxygen butalso establishan instructive vascular niche,which stimulates liver organo-

genesis and regeneration through elaboration of paracrine trophogens5,15

.Ourtransplantationapproach provided a unique intravital monitoring system for evaluating human iPSC-derived organ maturation and dif-ferentiation throughout the organogenesis, enabling us to dissect thepreviously uncharacterized roles of stromal cell types in human organdevelopment.

The liver-bud transplants were examined histologically at multipletime points. Hepatic cells in human iPSC-LBs proliferated rigorously,particularly until2 monthsafter transplantation(SupplementaryFig. 13).Similar to human FLC-LBs, human iPSC-LB transplants consisted of hepatic cord-like structures characteristic of adult liver after 60 days(Fig. 4a), composed of cells expressing the tight junction protein zonaoccludens 1 (ZO1), ALB and CK8/18 (Fig. 4b), asialoglycoproteinreceptor 1 (ASGR1) and collagen IV (Supplementary Fig. 14), whichwere normally found along the entire length of the sinusoid 16 . AFP-negative transplant-derived hepatic cells had the ultrastructural fea-tures of mature hepatocytes (Supplementary Fig. 15ae). Quantitativefluorescence-activatedcellsorting(FACS)profilingof humaniPSC-LBtransplants showed that 32.9 6 6.4% (n 5 8) of the total human cellswere positive for just ALB, whereas 3.86 1.5% were positive for bothALB and AFP, as seen in human FLC-LB transplants, suggesting thematuration of human iPSC-HEs (Fig. 4c and Supplementary Fig. 16).

To evaluate the functional maturation of human iPSC-LB ectopictransplants compared with that of human adult hepatocytes, humanserum albuminwas trackedby ELISA.Fourtransplanted human iPSC-LBs (consisting 1.83 105 immature hepatic cells) began producing albumin at approximately day 10 and produced up to 1,983 ng ml 2 1

by day 45, and hepatic maturation was observed at iPSC-LBs trans-plantedatvarious ectopicsites,whereas 4 3 106 humanadulthepatocytescould not keep producing high levels of albumin and showed a peak at

0.05

0.10

0.15

0 10 20 30 400

500

1,000

1,500

2,000

Human iPSC-LB (mesentery)

Human iPSC-LB (capsular)Human iPSC-LB (cranium)

Human AH (capsular)

Human iPSC-MH (capsular)

Days post transplantation

S e r u m

h u m a n a l

b u m

i n ( n g m

l 1 )

0

16

14

12

10

8

6

4

2

0

200

400

600

800

b

S e r u m

h u m a n

A A T ( n g

m l 1 )

Cranium Mesentery Capsular

0 10 20 300

20

40

60

80

100

Sham, n = 9

Human iPSC-LB, n = 17

Human FLC-LB, n = 7Human AH, n = 7

e

f

c

a

g

Nativehuman iPSCs (A)

Human adultliver (C)

66

108 2047

1 23

Human iPSC-liver (B)

Pathway Metabolites: A (B C)Glycolysis/ Gluconeogenesis Phosphoenolpyruvic acid

TCA cycle Succinic acidTCA cycle Fumaric acidTCA cycle 6-Phosphogluconic acidPPP Ribose 5-phosphatePPP Ribulose 5-phosphatePPP Sedoheptulose 7-phosphate

AA Gly, Ser, Cys 2-Hydroxybutyric acid AA Gly, Ser, Cys CDP-choline AA Gly, Ser, Cys N ,N -Dimethylglycine AA Gly, Ser, Cys Sarcosine AA Gly, Ser, Cys Taurine AA Gly, Ser, Cys Taurocholic acid

*PPP, pentose phosphate pathway*AA, amino acid synthesis/metabolism

Human ALB DAPI CK8/18 DAPI

H u m a n

i P S C - L

B T x

H u m a n

F L C - L

B T x

ZO1 DAPI

Human ALB DAPI CK8/18 DAPI ZO1 DAPI

i

P e r c e n t s u r v

i v a l

GCV 50 mg kg 1 i.p.(day 7 and 10)

Days

K T P / M 1 f o l d c h a n g e

*

h

*

*

Human FLC-LB Txmouse LB TxHUVEC human MSC Tx Human iPSC-LB Tx


A F P L o g F

I T C i n t e n s

i t y

32.9 6.4%

3.8 1.5%1.9 0.8%


AFP Log APC intensity36.7 7.3%

5.6 2.8%

62.0%

H2Kd Log PE intensity

d

M e t a b o l

i c r a

t i o ( A U C

0 8 h

o f 4 O H D B / D B )

S h a m

C W T x

M e s e n

t e r y T x

C o n t r o l H u

m a n

F L C - L B

T x H u m a n

i P S C - L

B T x

C o n t r

o l L B

x 2 L B

x 4 L B

x 1 2 L B

x 1 2

64.3 6.3%4.3 6.364.3 6.3%

Figure 4 | Functional characterization of human iPSC-LB derived liver.a , b, Haematoxylin and eosin, and immunofluorescence staining of tissues atday60. Dotted line indicates the border on the brain. Scale bars, 200 mm (a , toprow); 100 mm (a , bottom row and b, left and middle images on top and bottomrows) and 25 mm (b, right images, top and bottom rows). c, Representative FACSprofile for AFP- and/or ALB-positive human cells at day60 (n 5 8). Human cellswere identified as a murine H2Kd-negative population. Results are shown ascontour maps coloured by density of the scatter data. ALB2 /AFP2 population isgrey; ALB2 /AFP1 populationis green;ALB1 /AFP2 population isblue; andALB 1 /

AFP1

population is purple. d, e, ELISA showing levels of human serum ALB andAAT over time at various ectopic sites (cranium, mesentery and kidney subcapsule), compared with that of hAH (mean 6 s.e.m., n 5 4 in d, n 5 3 ine). Seealso Supplementary Tables 2 and3. f , g , Human-specific ketoprofen ( f ) anddebrisoquine (g ) metabolite formations (mean 6 s.e.m., n 5 3, *P , 0.05).Metabolic ratios were determined by dividing the AUC0-8h (the area under thecurve from time 0 until 8 h; detailed explanation is described in Supplementary Methods) ratio of 4-hyroxydebrisoqune (4OHDB) to debrisoquine (DB). h, Venndiagram shows the metabolic profilesof human iPSC-LB transplantsmeasured by CE-TOFMS (capillary electrophoresis time-of-flight mass spectrometry)(Supplementary Fig. 18). Representative metabolites are shown. i, KaplanMeiersurvivalcurves ofTK-NOGmice after liverinjury( P 5 0.0120).i.p., intraperitoneal.

LETTER RESEARCH

2 5 J U LY 2 0 1 3 | V O L 4 9 9 | N AT U R E | 4 8 3Macmillan Publishers Limited. All rights reserved201 3


4/5

day15 (Fig. 4d). Notably, transplanted conventional human iPSC-MHs(1 3 107 cells per mouse) produced much less albumin, suggesting theimportance of three-dimensional and vascularized tissue formation forsuccessful engraftment and maturation at ectopic sites. The amount of human serum alpha-1-antitrypsin (AAT) was in proportionto thenum-berof liver-budtransplants(Fig.4e).Gene-expression profilingofhumaniPSC-LBs explanted 60days (n 5 8) after transplantation demonstratedthe significant hepatic maturation compared with conventional human

iPSC-MHs (n5

8)4

(Supplementary Fig. 17).To analyse the drug metabolism activity that is a major function of theliver, themice were challenged with ketoprofen 17 or debrisoquine 18 , whichare known to be metabolized differently by mice and humans. After thedrug exposure, formation of human-specific metabolites was demon-strated in urine andserum samples collected from mice transplantedwithhuman iPSC-LBs(Fig.4f,g andSupplementaryDiscussion).Furthermore,to characterize the profiles of small-molecule metabolites such as the pro-ducts of sugar, amino acid and nucleotide metabolism, metabolome ana-lysis of human iPSC-LB transplants was performed. This showed222metabolites, including liver-specific metabolites such as taurocholicacid (Fig. 4h). This profile, with a high number of metabolites, wassimilarto that ofhuman adult liver but not to that oforiginalhuman iPSCs (Sup-plementary Fig.18).These results demonstrate thepotential forpredicting drug metabolite profiles of humans, with drug metabolite profiles of micetransplantedwithhuman iPSC-LBsmimicking invivo humanphysiology.This is particularlystriking,as successfuldetectionofhumandrug respon-sivenessreliesontheuseofhigh-qualityhepatocytesandseverelydamagedhost liver using conventional chimaeric mouse models 19,20 .

Withan aimto furtherour understanding ofclinicaltransplantationand its future use, we evaluated the possibility of a minimally invasivemesenteric transplantation model, which would be a more realistictarget site. Macroscopic observation confirmed the successful engraft-ment of transplanted LBs on mesentery (Supplementary Fig. 19a, b).Transplantation of 12 human iPSC-LBs improved the survival of TK-NOGmice21 after the gancyclovir-induced liver failure compared withthe sham-operated, human FLC-LB- and human adult-hepatocyte-transplanted mice (Fig. 4i and Supplementary Discussion). Thus, wesuccessfully generated vascularized and functional human liver by transplantation of human iPSCs-LBs.

ManipulationofiPSCsholdsgreat promise forregenerativemedicine.However, clinical trials of cell transplantation, currently an importanttarget of the stem-cell-based approach, have presented unsatisfactory results22 . Our study demonstrates a proof-of-concept that organ-budtransplantation offers an alternative approach to the generation of athree-dimensional, vascularized organ. These results highlight theenormous therapeutic potential using in vitro-grown organ-bud trans-plantation for treating organ failure.

METHODS SUMMARYHepatic differentiation of human iPSCs was achieved based on a protocol reportedpreviously 4. HUVECs and human MSCs (Lonza) were maintained in endothelialgrowth medium (Lonza) or MSC growth medium (Lonza) at 37 u C in a humidified

5%CO2 incubator. To generate human liver buds in vitro, 13

106

human iPSC-derived hepatic cells, 0.83 106 to 13 106 HUVECs and 2 3 105 human MSCs wereresuspended in the mixture of EGM and hepatocyte culture medium (Cambrex)containing withdexamethasone(0.1 mM, Sigma-Aldrich),oncostatin M (10ngml 2 1,R&D System), hepatocyte growth factor (HGF; 20ngml

2 1, PromoKine) andSingleQuots (Lonza) and plated on pre-solidified 3 2 Matrigel diluted with EGM(BD Biosciences) in a 24-well plate. After approximately 4to 6 days of culture, self-organized human iPSC-LBs were detached, collected and transplanted into a pre-formed cranial window 13 of an immunodeficient mouse.

Full Methods and any associated references are available in the online version ofthe paper.

Received 18 April 2012; accepted 7 May 2013.Published online 3 July 2013.

1. Klein,A. S.et al. Organ donation and utilization in the United States, 19992008. Am. J. Transplant. 10, 973986 (2010).

2. Kriks,S. et al. Dopamine neurons derived from human ES cells efficientlyengraft in animal models of Parkinsons disease. Nature 480, 547551(2011).

3. Kroon,E. et al. Pancreatic endoderm derived from human embryonic stem cellsgenerates glucose-responsiveinsulin-secreting cells in vivo.Nature Biotechnol. 26,443452 (2008).

4. Si-Tayeb, K. et al. Highly efficient generation of human hepatocyte-like cells frominduced pluripotent stem cells. Hepatology 51, 297305 (2010).

5. Matsumoto, K., Yoshitomi,H., Rossant, J. & Zaret, K. S. Liver organogenesispromoted by endothelial cells prior to vascular function. Science 294, 559563(2001).

6. Espejel, S. et al. Induced pluripotent stem cell-derived hepatocytes have thefunctional and proliferative capabilities needed for liver regeneration in mice. J. Clin. Invest. 120, 31203126 (2010).

7. Kobayashi,T. etal. Generationof ratpancreas in mouse by interspecific blastocystinjection of pluripotent stem cells. Cell 142, 787799 (2010).

8. Zhao, R. & Duncan, S. A. Embryonic development of the liver. Hepatology 41,956967 (2005).

9. Si-Tayeb,K., Lemaigre,F. P.& Duncan,S.A. Organogenesisanddevelopmentof theliver. Dev. Cell 18, 175189 (2010).

10. Gouysse, G. et al. Relationship between vascular development and vasculardifferentiation during liver organogenesis in humans. J. Hepatol. 37, 730740(2002).

11. Jung, J., Zheng, M., Goldfarb, M. & Zaret, K. S. Initiation of mammalian liverdevelopment from endoderm by fibroblast growth factors. Science 284,19982003 (1999).

12. Korzh,S. et al. Requirement of vasculogenesis and bloodcirculation in latestagesof liver growth in zebrafish. BMC Dev. Biol. 8, 84 (2008).

13. Koike,N. et al. Tissue engineering: creation of long-lasting blood vessels. Nature428, 138139 (2004).

14. Takebe,T. etal. Generation of functional humanvascular network. Transplant.Proc.44, 11301133 (2012).

15. Ding, B. S. et al. Inductive angiocrine signals from sinusoidal endothelium arerequired for liver regeneration. Nature 468, 310315 (2010).

16. Martinez-Hernandez, A. The hepatic extracellular matrix. I. Electronimmunohistochemical studies in normal rat liver. Lab. Invest. 51, 5774(1984).

17. Ishizaki, T. et al. Pharmacokinetics of ketoprofenfollowing single oral,intramuscular and rectal doses and after repeatedoral administration. Eur. J. Clin.Pharmacol. 18, 407414 (1980).

18. Yu, A. M., Idle, J. R. & Gonzalez, F. J. Polymorphic cytochrome P450 2D6:humanized mouse model and endogenous substrates. Drug Metab. Rev. 36,243277 (2004).

19. Katoh, M. et al. In vivodrug metabolism model for human cytochrome P450enzyme using chimeric mice with humanized liver. J. Pharm. Sci. 96, 428437(2007).

20. Kamimura, H. etal. Assessmentof chimeric micewith humanizedliver as a toolforpredicting circulating human metabolites. Drug Metab. Pharmacokinet. 25,223235 (2010).

21. Hasegawa, M. et al. The reconstituted humanized liver in TK-NOG mice ismature and functional. Biochem. Biophys. Res. Commun. 405, 405410(2011).

22. Dudley, S. C. Jr. Beware of cells bearing gifts: cell replacement therapy andarrhythmic risk. Circ. Res. 97, 99101 (2005).

Supplementary Information is available in the online version of the paper .

Acknowledgements We thank F. Kawamata, E. Yoshizawa, Y. Suzuki, S. Nakai,Y. Takahashi, N. Tsuchida and N. Sasaki for kindly providing technical support; J. Nakabayashi, K. Yasumura, R. Fujiwara,T. Amiya, A. Nakano, Y. Mitsuhashi and all ofthe members of our laboratory for help with several experiments and comments. Weare also grateful to D. Fukumura, Y. Goshima, T.Hirose, M. Ichino, U. Yokoyama,T. Ogawa and R. K. Jain for critical evaluation of the manuscript. This work wassupported by grants to H. Taniguchi from the Strategic Promotion of InnovativeResearchand Development (S-innovation, 62890004) of the Japan Science andTechnology Agency (JST). This work was also supported by the Grants-in-Aid of theMinistry of Education, Culture, Sports, Science, and Technology of Japan to T. Takebe(no. 24106510, 24689052), N. Koike (no. 22390260) and H. Taniguchi(no. 21249071, 25253079); by the Specified ResearchGrant of the Takeda ScienceFoundation and a grant from the Japan IDDM network to H. Taniguchi; andby a grant of the YokohamaFoundation for Advanced Medical Science toT. Takebe.

Author Contributions T.T. conceived the study, performed the experiments, collectedandanalysed thedata anddraftedthe manuscript.K.S.,M.E., H.K., M.K., T.O., R-R.Z. andS.A. performed the experiments with the technical guidance and expertise of K.S.,Y.-W.Z., Y.U. and T.T. K.S., Y.-W.Z., N.K., Y.A. and H.T. provided critical advice on theresearchstrategy and design.

Author Information Microarray data, including that of human iPSC-LBs, humanFLC-LBs, human adult liver tissues (ALT) and mouse liver tissue of variousdevelopmental stages, have been deposited in the Gene Expression Omnibus underaccession number GSE46631 . Reprints and permissions information is available atwww.nature.com/reprints . The authors declare no competing financial interests.Readersare welcome to commenton the online versionof thepaper . Correspondenceand requests for materials should be addressed to T.T. ([email protected])or H.T. ([email protected]) .

RESEARCH LETTER

4 8 4 | N AT U R E | V O L 4 9 9 | 2 5 J U LY 2 0 1 3Macmillan Publishers Limited. All rights reserved201 3
http://www.nature.com/doifinder/10.1038/nature12271http://www.nature.com/doifinder/10.1038/nature12271http://www.nature.com/doifinder/10.1038/nature12271http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE46631http://www.nature.com/reprintshttp://www.nature.com/doifinder/10.1038/nature12271mailto:[email protected]:[email protected]:[email protected]:[email protected]://www.nature.com/doifinder/10.1038/nature12271http://www.nature.com/reprintshttp://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE46631http://www.nature.com/doifinder/10.1038/nature12271http://www.nature.com/doifinder/10.1038/nature12271http://www.nature.com/doifinder/10.1038/nature12271


5/5

METHODSCell culture and differentiation. TkDA3 human iPSC clone used in this study was kindly provided by K. Eto and H. Nakauchi. Undifferentiated human iPSCswere maintained onthe mouseembryonic fibroblast cellsas feeder cells in humaniPSC medium. For endodermal differentiation, human iPSCs were seeded on aMatrigel-coated dish and the medium was changed to RPMI1640 with 1% B27without insulin and human activin A (100ng ml

2 1) for 5 to 6 days. For hepaticspecification, human iPSC-derived endodermal cells were treated further withRPMI1640 with 1% B27and humanbasicFGF (10ng ml

2 1), human BMP4 (20ng ml

21) for3 to 4 days.Recombinanthuman activin A waskindlyprovided byY. Eto

at Ajinomoto Co. Human FLCs (CS-ABI-3716; Applied Cell Biology ResearchInstitute) were plated on collagen IV-coated 6-well plates (BD Biosciences) andcultured in our standard medium (1:1 mixture of DMEM and F-12 (SigmaAldrich) supplemented with 10% FBS (lot 7219F; ICN Biochemical), 50 mmoll

2 1

HEPES (Wako Pure Chemical Industries), 2 mmoll2 1

L-glutamine (Life Tech-nologies Corporation), 50 mmol l

2 1 2-mercaptoethanol (Sigma), 1 3 penicillin-streptomycin (Life Technologies),10 mmoll

2 1 nicotinamide(Sigma), 1 3 102 4 M

Dexamethasone (Sigma) and 1 mg ml2 1 insulin (Wako)). Human recombinantHGF (50ngml

2 1) and EGF (20 ngml) (Sigma) were added before cultivation.HUVECs and human MSCs (Lonza) were maintained in endothelial growthmedium or MSC growth medium (Lonza) at 37 u C in a humidified 5% CO2incubator. Cryopreserved human adult hepatocytes (lots 582, 737, 746 andHC2-8) werepurchased fromXenoTech(Lenexa).Human adult hepatocytes werethawed, and the number of viable cells was counted according to the manufac-turers instructions and used for transplantation experiments. In vitro albuminproduction tests were performed on viable cells just after first 24 h of culture.Retroviral transduction. For live imaging, cells were infected with retrovirusesexpressing EGFP or human Kusabira-Orange ( KO1) as described13 . In brief, aretrovirus vector pGCDNsam IRESEGFP or KOFP was transfected into 293gag/pol (gp) and 293gpg (gp and VSV-G) packaging cells (kindly provided by M. Onodera) in which viral particle production is induced using a tetracycline-inducible system. Culture supernatants of retrovirus-infected cells were passedthrough a 0.45-mm filter (Whatman, GE Healthcare) and used immediately forinfection. KOFP displays a major absorption wavelength maximum at 548 nmwith a slight shoulder at 515 nm and emits a bright orange fluorescence, with apeak at 561nm.Transplantation. In vitro-generated liver buds were detached, collected andtransplanted into a pre-formed cranial window or indicated sites of a non-obesediabetic/severe combined immunodeficient (NOD/SCID) mouse (Sankyo Lab.Co.). The in vivo fate of transplanted cells was monitored by intravital imaging

using a fluorescence microscope (model BZ-9000; Keyence) or the Leica TCS SP5confocal microscope (Leica Microsystems). For survival curves, TK-NOG mice(bodyweights ofapproximately20 to 30g) were used inthis study(suppliedby theCentral Institute for Experimental Animals) 21 . Ganciclovir (50 mg kg

2 1, intraper-itoneal), a drug that is not toxic to human or mouse tissues, was administered toinduce tissue-specific ablation of transgenic liver parenchymal cells at day 7 and10 after a dozen human iPSC-LBswere transplantedon the mesentery. Forcontrolexperiments, we thawed and isolated the viable human adult hepatocytes, andresuspended the viable cells(3 3 106 to 4 3 106) incold mediumandcoldMatrigel(BD) and transplanted them into the kidney subcapsular space. The mice werebred and maintained according to the Yokohama City University institutionalguidelines for the use of laboratory animals.Intravital imaging. Tail vein injections of 1% tetramethylrhodamine-conjugated dex-tran(2,000,000 MW),fluorescein-isothiocyanate-conjugateddextran(2,000,000MW)and Texas-Red-conjugated dextran (70,000MW, neutral) were used to identify vessel lumens (all from Invitrogen). Host endothelial cells were visualized using

Alexa647-conjugated mouse-specific CD31 (BD), injected intravitally. Confocalimage stacks were acquired for the implanted vessels and dextran. Image projec-tionswere processedusingMetaMorphAngiogenesisModule software(MolecularDevices). Total tubule length, the percentage of tubules per field (area of tubulesdivided by the area of the transplants that were imaged) and tube diameter werethen logged automatically into an Excel spreadsheet.

Gene-expression analysis. Quantitative PCR analyses were conducted asdescribed previously 23 . Total RNA of human fetal liver (lot no. A601605) andhuman adult liver (lot no. B308121) were obtained from Biochain Institute(Hayward).Microarray and data analysis. Total RNA was prepared from human iPSC-derived cells or tissues (human iPSC-LBs, human FLC-LBs), human adult livertissue and CD45-negative and Ter119-negative murine liver cells at various dif-ferent developmental stages (E9.5, E10.5, E11.5, E13.5, E15.5, E17.5, E19.5, post-natal day 0 (P0), P3 and postnatal week 8) using an RNeasy Mini Kit (Qiagen).

RNA for gene-expression profiling was hybridized on a Whole Human GenomeAgilent4 3 44Kv2 Oligonucleotide Microarray or WholeMouse Genome Agilent4 3 44K v2 Oligonucleotide Microarray (Agilent Technologies) according to themanufacturers instructions. To perform cross-species comparison of expressionprofiles,26,153expression dataat genelevel werecross-referenced to otherspeciesusing the HomoloGene IDs in the MGI curated data set of humanmouseorthol-ogy with Phenotype Annotations ( http://www.informatics.jax.org ). To generatethe heat map, we used a hierarchical clustering method with Euclidean distancecomplete linkage on GeneSpring11.5.1. to analyse the 83 originally selected liversignature gene-expression profiles. Of all genes, 83 genes were selected as liversignature genes because they increased continuously during both murine andhuman liver development.ELISA. Blood samples were allowed to clot in a centrifuge tube (approximately 5 min) atroom temperature (24 u C), loosenedfrom the sidesof thetubeandkeptat4 u C (meltingice)for 20min.Clottedbloodwascentrifugedfor10to 15minat 400 g ,4 u C and the serum fraction was removed, taking care to exclude erythrocytes orclotted materials. Human ALB and AAT in the mouse serum samples was mea-sured using a Human Albumin ELISA Quantitation Kit (Bethyl Laboratories) andhuman alpha 1-antitrypsin ELISA Quantitation Kit (GenWay Biotech) according to the manufacturers instructions.Whole mount immunostaining. Mice were perfusedwith 4% paraformaldehydein phosphate buffer solution (PBS) through cardiac puncture. The cover-glassforming the cranial window was removed, and the transplants (approximately 300-mm thick)were extracted andplacedin 4% paraformaldehyde in PBSfor 1.5hon ice. For immunostaining, fixed collagen gels were washed three times in PBS(10min each), blocked with 3%BSA or 0.1% Triton X-100 for 1h, incubated withprimary antibodies at 4 u C overnight, followed by three 10-min washes in PBS or0.1% Triton X-100. The sample was incubated with secondary antibodies at 4 u Covernight, followed by three 10-min washes in PBS or 0.1% Triton X-100. Tissuesamples werecounterstainedwith DAPIand mounted on glass slides in mounting media (Vector Laboratories),under a coverslip. The followingprimaryantibodieswereused:mouse anti-human ZO1,mouseanti-human CD31andrat anti-mouseCD31 (BD Biosciences), rabbit anti-mouse collagen IV (Millipore) and desmin(Dako Corporation). Immunostaining was analysed using the Leica TCS SP5confocal microscope.Tissue processing and immunostaining. Tissues were fixed overnight at 4 u C in4% paraformaldehyde, processed, and embedded in paraffin. Transverse sections(4 mm) were placed on MAS-coated slides (Matsunami) for immunostaining orstandard histological staining with haematoxylin and eosin. Immunostaining waspreceded by autoclave antigen retrieval in citrate buffer (pH 6.0). The primary antibodies were anti-human: CD31, smooth muscle actin, AFP, CK8/18 (all fromDako Corporation) and ALB (BD Biosciences). Tissue sections were incubatedwith secondary antibody Alexa Fluor (Life Technologies) for 1 h at room temper-ature, followedby DAPI(Sigma)nuclearstaining. The images wereacquiredusing the LSM510 laser scanning microscope (Carl Zeiss).Statistical analysis. Data are expressed as the means 6 s.d. or s.e.m. from at leastthree independent experiments. Comparisons between three or four groups were

analysed using the Wilcoxon statistical analyses or KruskalWallis test by ranks,and post-hoc comparisons were made using the MannWhitney U -test withBonferroni correction. Two-tailed P values of , 0.05 were considered significant.

23. Kobayashi, S. etal. Reconstructionof human elasticcartilageby a CD44 1 CD90 1

stemcell in theear perichondrium. Proc. NatlAcad.Sci. USA 108, 1447914484(2011).

LETTER RESEARCH

ll bl h d ll h d201
http://www.informatics.jax.org/http://www.informatics.jax.org/

vascularized and functional human liver from an ipsc-derived organ bud transplant

Documents