Klotho is associated with VEGF receptor-2 and thetransient receptor potential canonical-1 Ca2+ channelto maintain endothelial integrityTetsuro Kusabaa,1, Mitsuhiko Okigakia,2, Akihiro Matuia,1, Manabu Murakamib, Kazuhiko Ishikawac, Taikou Kimuraa,Kazuhiro Sonomuraa, Yasushi Adachid, Masabumi Shibuyae, Takeshi Shirayamaa, Shuji Tandaa, Tsuguru Hattaa,Susumu Sasakia, Yasukiyo Moria, and Hiroaki Matsubaraa

aDepartment of Cardiovascular and Renal Medicine, Kyoto Prefectural University of Medicine, Kyoto 602-8566, Japan; bDepartment of Pharmacology, AkitaUniversity School of Medicine, Akita 010-0852, Japan; cDepartment of Geriatric Medicine, Osaka University Graduate School of Medicine, Osaka 565-0871,Japan; dDepartment of Pathology I, Kansai Medical University, Osaka 570-8506, Japan; and eDepartment of Molecular Oncology, Tokyo Medical and DentalUniversity, Tokyo 113-8519, Japan

Edited by Eric N. Olson, University of Texas Southwestern Medical School, Dallas, TX, and approved September 13, 2010 (received for review June 23, 2010)

Klotho is a circulating protein, and Klotho deficiency disturbs en-dothelial integrity, but the molecular mechanism is not fully clar-ified. We report that vascular endothelium in Klotho-deficientmice showed hyperpermeability with increased apoptosis anddown-regulation of vascular endothelial (VE)-cadherin becauseof an increase in VEGF-mediated internal calcium concentration([Ca2+]i) influx and hyperactivation of Ca2+-dependent proteases.Immunohistochemical analysis, the pull-down assay using Klotho-fixed agarose, and FRET confocal imaging confirmed that Klothoprotein binds directly to VEGF receptor 2 (VEGFR-2) and endothe-lial, transient-receptor potential canonical Ca2+ channel 1 (TRPC-1)and strengthens the association to promote their cointernaliza-tion. An in vitro mutagenesis study revealed that the second hy-drolase domain of Klotho interacts with sixth and seventh Igdomains of VEGFR-2 and the third extracellular loop of TRPC-1.In Klotho-deficient endothelial cells, VEGF-mediated internaliza-tion of the VEGFR-2/TRPC-1 complex was impaired, and surfaceTRPC-1 expression increased 2.2-fold; these effects were reversedby supplementation of Klotho protein. VEGF-mediated eleva-tion of [Ca2+]i was sustained at higher levels in an extracellularCa2+-dependent manner, and normalization of TRCP-1 expressionrestored the abnormal [Ca2+]i handling. These findings provideevidence that Klotho protein is associated with VEGFR-2/TRPC-1in causing cointernalization, thus regulating TRPC-1–mediatedCa2+ entry to maintain endothelial integrity.

endothelial cell | vascular calcification

The Klotho gene was identified by insertion mutagenesis inmice (1). The homogenous mutant mouse exhibited a pheno-

type of accelerated aging, including extensive vascular calcifica-tion with hyperphosphatemia. Klotho is abundantly expressed inthe kidney and, to a lesser extent, in other organs, including theaorta (1). Interestingly, patients with chronic renal failure havelow concentrations of Klotho protein in the serum (2) and vas-cular calcification with hyperphosphatemia, suggesting that a de-crease in circulating Klotho may contribute to vascular lesions inthe patients with chronic renal failure.We previously reported that Klotho expression is induced by

statin and attenuated by angiotensin II through the regulation ofRas homolog gene family, member A (3). VEGF-mediated an-giogenesis is impaired in Klotho-deficient mice, in which reducedrelease of endothelial NO was reported (4). Klotho gene deliverywas shown to improve endothelial dysfunction through an NO-dependent pathway (5) and to extend the survival of rats withglomerulonephritis (6) or angiotensin II-induced renal failure (7).Thus,Klotho is likely tobe akidney-derived vasoprotectiveprotein.The molecular function of Klotho has been partly deciphered.

Klotho can act as a glycosidase (8) that hydrolyses sugar residueson the epithelial Ca2+ channel, the transient receptor potential

(TRP) vanilloid receptor-related channel TRPV5 (9). Further-more, Klotho binds directly to multiple FGF receptors (FGFRs),to FGF23 (10), and also to insulin receptor (11) but not to EGF orPDGF receptors (11), and the Klotho/FGF23 complex binds toFGFR1with higher affinity thanFGF23and significantly enhancesthe ability of FGF23 to induce FGFR phosphorylation (10).However, the action of Klotho on VEGF receptor (VEGFR)-mediated signals remains to be determined. Ca2+ signals regulatevarious biological functions in endothelial cells (ECs), such asproliferation, migration, and apoptosis (12). Upon ligand bindingto receptors on ECs, the internal calcium concentration ([Ca2+]i)increases via inositol triphosphate-mediated Ca2+ release fromthe endoplasmic reticulum (ER), causing subsequent Ca2+ entrytriggered by the depletion of Ca2+ stores in the ER (12). One ofthe types of ion channel participating in store depletion-operatedCa2+ entry is the TRP family, which is divided further intofour subfamilies, TRPV, TRPC (canonical or classical), TRPM(melastatin-related), and TRPP (protein kinase D subtype) (13).The TRPC family consists of seven isoforms, each of which showsspecific cellular distribution and functions (13, 14). TRPC-1,TRPC-3, TRPC-4, and TRPC-6 are expressed in the ECs, andthese subtypes are closely involved in various vascular functions:TRPC-1 in VEGF-mediated Ca2+ entry (13–15), TRPC-4 inhypoxia-induced vascular remodeling, and TRPC-3 and TRPC-4in oxidative stress-induced responses (16). The association withKlotho protein has not yet been studied.In the present study, we examined whether Klotho protein

maintains endothelial integrity in association with VEGFR andendothelial TRPC-1. We report that vascular endothelium inKlotho-deficient mice is hyperpermeable because of enhancedapoptosis and decreased surface expression of vascular endo-thelial (VE)-cadherin, and Ca2+-dependent calpain/caspase-3are hyperactivated. Klotho binds directly to both VEGFR-2 andTRPC-1, but not to other TRPC subtypes, and the complex isinternalized in response to VEGF stimulation, thus regulatingVEGFR/TRPC-1–mediated Ca2+ influx to maintain endothelialbiological homeostasis.

Author contributions: T. Kusaba, M.O., T.S., S.S., and H.M. designed research; T. Kusaba,M.O., A.M., T. Kimura, K.S., Y.A., S.T., T.H., and Y.M. performed research; M.M., K.I., Y.A.,M.S., and S.S. contributed new reagents/analytic tools; T. Kusaba, M.O., A.M., Y.A., andH.M. analyzed data; and T. Kusaba and M.O. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1T. Kusaba and A.M. contributed equally to this work.2To whom correspondence should be addressed. E-mail: okigakim@koto.kpu-m.ac.jp.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1008544107/-/DCSupplemental.

19308–19313 | PNAS | November 9, 2010 | vol. 107 | no. 45 www.pnas.org/cgi/doi/10.1073/pnas.1008544107

ResultsVascular Hyperpermeability and Increased Apoptosis in Klotho-Deficient Aorta. To study the endothelial function in Klotho-deficient mice, we first evaluated the vascular apoptosis. TUNELstaining of the aorta showed that CD31+ ECs and alpha-smoothmuscle actin (αSMA)-positive vascular smooth muscle cells(VSMCs) already were TUNEL-positive in 4-wk-old Klotho-deficient mice in which vascular calcification was not apparentbut were barely detectable in WT littermates (Fig. 1A).Serum concentrations of Ca2+ and phosphate are elevated in

Klotho-deficient mice (1). We hypothesized that Klotho de-ficiency may cause endothelial hyperpermeability, leading toexudation of Ca2+/phosphate-rich plasma into the vessel wall. Toevaluate vascular permeability, Evans Blue dye was administeredto 2-wk-old mice 10 min before they were killed. In the aorta ofWT mice, dark blue-stained hyperpermeable lesions were barelydetectable on the luminal surface of longitudinally excised aorta(Fig. 1B, Left) and in the intimal region (Fig.1B, Right); suchlesions were patchily detected in the aorta of Klotho-deficient mice.

Increase in Calpain/Caspase-3 Activities and Acceleration of Apo-ptosis. Hyperactivation of calpain causes damage to renal epi-thelial cells in Klotho-deficient mice (17). We observed thatcalpain and caspase-3 activities against exogenous substrate inthe aorta from Klotho-deficient mice were 130% and 35%higher, respectively, than in WT mice (Fig. 2A). Calpain activi-ties were evaluated further by determining the endogenous levelof αII-spectrin, a cleaved substrate fragment that is highly sen-sitive to calpain (17). αII-spectrin is cleaved at particular siteby calpain, yielding a 136-kDa fragment (N-terminal cleavedproducts) and a 148-kDa fragment (C-terminal cleaved prod-ucts). Because αII-spectrin is a plasma membrane-bound cyto-skeletal protein, membrane fraction was isolated from the aortaand analyzed by SDS/PAGE and Western blotting using an anti-αII-spectrin antibody reactive to the 148-kDa cleaved fragmentand to full-length αII-spectrin. The relative intensity of the 148-kDa cleaved fragment to full-length αII spectrin was 4.1-foldhigher in the Klotho-deficient aorta than in WT aorta (P < 0.005;n = 5) (Fig. 2B).Although basal calpain and caspase-3 activities did not differ

significantly between WT and Klotho-deficient ECs, the activitiesafter VEGF stimulation weremarkedly higher in Klotho-deficient

ECs (1.8-fold and 3.2-fold, respectively) than in WT ECs (Fig.2C). In Ca2+-free medium, no significant increase was observedin VEGF-induced calpain and caspase-3 activities in eitherWT orKlotho-deficient ECs (Fig. 2C).Calpain is activated by an increase in [Ca2+]i, suggesting that

regulation of [Ca2+]i may be altered in Klotho-deficient cells.We therefore examined Ca2+-mediated apoptosis in Klotho-deficient ECs (Fig. 2D). Basal numbers of TUNEL-positive ECsdid not differ significantly between WT and Klotho-deficient ECs.The number of apoptotic cells increased markedly in Klotho-deficient ECs (5.3-fold, P < 0.05) 3 h after VEGF stimulation,whereas no change was observed in WT ECs. In Ca2+-free me-dium, no significant increase was observed in either group.

Hyperactivated Calpain Degrades p120 Catenin, Leading to In-ternalization and Proteolysis of VE-Cadherin. VE-cadherin playsa crucial role in maintaining endothelial integrity (18). We ob-served that VEGF stimulation significantly reduced the VE-cadherin level in Klotho-deficient ECs (Fig. 3A). VE-cadherinwas localized on the plasma membrane in both groups, but theexpression level was significantly lower in Klotho-deficient ECsafter VEGF stimulation (Fig. 3A). We therefore examined themechanism for VEGF-mediated down-regulation of VE-cadherin.Treatment of Klotho-deficient ECs with the μ-calpain inhibitoracetyl-leucyl-leucyl-norleucinal (ALLN) (20 μM) blocked theVEGF-mediated decrease in VE-cadherin expression towardthe WT level (Fig. 3A), suggesting that μ-calpain is involved inthe mechanism down-regulating VE-cadherin. Therefore, weimmunoprecipitated VE-cadherin from WT EC lysates and in-cubated it with purified μ-calpain; however, VE-cadherin proteinwas not cleaved (Fig. 3B). It has been reported that p120-catenin(p120ctn) associated with VE-cadherin prevents internalizationand subsequent lysosomal degradation of VE-cadherin (19). Weconfirmed that p120ctn is associated with VE-cadherin in WTECs (Fig. 3B) and found that incubation of immunoprecipitatedp120ctn with μ-calpain markedly cleaved the p120ctn (76% re-duction) (Fig. 3B). Although basal p120ctn levels are similar inWT and Klotho-deficient ECs, VEGF treatment caused a sig-nificant decrease in p120ctn expression in Klotho-deficient ECs(52% lower than in WT ECs; P < 0.01), but pretreatment withALLN blocked the VEGF-mediated decrease (Fig. 3C). Theexpression of p120ctn was localized mainly on the cell surface inboth WT and Klotho-deficient ECs. VEGF stimulation markedly

A

Dar

k bl

ue-s

tain

ed in

timal

are

a in

tota

l lum

inal

leng

th (%

)

Kl-/-WT

Dar

k bl

ue-s

tain

ed le

sion

in th

e to

tal l

umin

al a

rea

(%)

WT0

10

20

Kl-/-

p<0.005

0

10

20

WT Kl-/-

p<0.005

WT Kl-/-B

CD31TUNEL TUNEL/CD31

WT

Kl-/-

WT

ECTUNEL αSMA TUNEL/ αSMA

VSMC

WT

Kl-/- Kl-/-

WT

Kl-/-Kl-/-

Lumen

Lumen

Lumen

Lumen

Fig. 1. Endothelial apoptosis and hyperpermeability. (A)Immunostaining of the aorta with TUNEL, anti-CD31, andanti-αSMA antibodies. Double-immunofluorescent cellsare indicated by arrows. Images are representative ofsimilar results observed in 4-wk-old mice (n = 7). (B) EvansBlue dye was injected 10 min before 2-wk-old mice werekilled (n = 7). The aorta was excised longitudinally (Left),and cross-sectioned aorta was stained with H&E (Right).Hyperpermeabilized areas (dark blue, arrows) weremeasured with computed morphometry (n = 7). (Scalebars: 500 μm.) K−/−, Klotho-deficient mice.

Kusaba et al. PNAS | November 9, 2010 | vol. 107 | no. 45 | 19309

CELL

BIOLO

GY

decreased surface expression of p120ctn in Klotho-deficient ECsbut not in WT ECs (Fig. 3C).To evaluate p120ctn/VE-cadherin expressions on plasma mem-

brane quantitatively, we isolated plasma membrane fractions con-taining no early endosomal antigen 1 (EEA), the early endosomalmarker protein, or mannose-6-phospate receptor (MPR), the lateendosomal marker proteins (Fig. 3E, Right). The expression levelsof VE-cadherin and 120ctn after VEGF simulation were 60% and76% lower, respectively, in Klotho-deficient ECs than in WT ECs,although baseline levels in the two groups were similar (Fig. 3D).Pretreating Klotho-deficient ECs with ALLN blocked the VEGF-mediated decrease in VE-cadherin and p120ctn expression. An in-hibitor of internalization (chlorpromazine, 5 μg/mL) also inhibitedthe VEGF-mediated decrease in VE-cadherin but did not affectp120ctn (Fig. 3D). Taken together, these data suggest that hyper-activated calpain in Klotho-deficient ECs causes p120ctn degrada-tion, leading to enhanced internalization of VE-cadherin. TheassociationbetweenVE-cadherinandVEGFR-2/Klothowasbarelydetectable (Fig. 3E), suggesting that VE-cadherin is not cointer-nalized with VEGFR-2/Klotho. Furthermore, siRNA-mediatedTRPC-1 knockdown in Klotho-deficient ECs blocked VEGF-mediated decreases in p120ctn and VE-cadherin expression (Fig.S1B), indicating that TRPC-1–mediated Ca2+ entry is involved inVEGF-mediated down-regulation of p120ctn/VE-cadherin.

Direct Interactions Among Klotho, VEGFR-2, and TRPC-1. The TRPC-1, TRPC-3, TRPC-4, and TRPC-6members of the TRP family areexpressed in ECs (13, 14), and TRPC-1 promotes VEGF-medi-atedCa2+ entry, leading to increase vascular permeability (14). Toevaluate interactions among Klotho, VEGFR-2, and TRPCs, weoverexpressed Klotho by Klotho gene recombinant adenovirus(3 × 107 pfu) in human umbilical vein endothelial cell (HUVECs)(Fig. S2A) and then stimulated the HUVECs with VEGF for 30min. (We used HUVECs, rather thanWTECs, in this experimentto achieve more efficient infection.) As shown in Fig. 4A, Klothobound constitutively to TRPC-1 and VEGFR-2, and VEGF stim-ulation did not affect the association, whereas Klotho was notassociatedwithTRPC-3, TRPC-4, orTRPC-6, althoughHomer-1,a positive control interactor (20), bound to all TRPCs (Fig. S2A).To confirm the interactions among Klotho, VEGFR-2, and

TRPC-1, we overexpressed histidine (His)-tagged Klotho protein

in COS cells and extracted the Klotho protein from cell lysatesby fixing it to Ni-nitrilotriacetate (Ni-NTA)-affinity agarosebeads (Fig. S2B). Then lysates fromHUVEC were incubated withKlotho-fixed agarose. Protein bound to this agarose was analyzedby Western blotting. The result showed that both TRPC-1 andVEGFR-2 were specifically detectable in theKlotho-fixed agarose(Fig. 4B).FRET microscopy is an ideal technique to highlight in vivo

protein–protein interactions. To obtain the FRET signal betweenKlotho and VEGFR-2 or TRPC-1, EYFP was fused to the C ter-minus of Klotho, and ECFP was fused to the C terminus ofVEGFR-2 and TRPC-1. Recombinant cDNA plasmids codingKlotho-EYFP/VEGFR-2-ECFP or Klotho-EYFP/TRPC-1-ECFPwere coexpressed in 293 HEK cells. Klotho was colocalized withVEGFR-2 or TRPC-1, and FRET signals were obtained betweenEYFP-Klotho and ECFP-VEGFR-2 or ECFP-TRPC-1, whereascoexpression of ECFP and EYFP-Klotho protein did not producea FRET signal (Fig. 3C). Furthermore, no FRET signal wasdetected in cells in which only ECFP-TRPC-1 or only ECFP-VEGFR-2 was overexpressed (Fig. 3D), indicating that Klothoactually binds to TRPC-1 or VEGFR-2 in the living cells.

Klotho Strengthens Association Between VEGFR-2 and TRPC-1 andPromotes VEGF-Mediated Internalization of the VEGFR-2/TRPC-1Complex. We studied the association between VEGFR-2 andTRPC-1 by an immunoprecipitation experiment. The resultshowed that there is a very weak association between VEGFR-2and TRPC-1 in the WT ECs, and this association was attenuatedin Klotho-deficient ECs (Fig. 5A). Addition of Klotho markedlypromoted the association of VEGFR-2 with TRPC-1 in both WTand Klotho-deficient ECs (Fig. 5A). Because Klotho mRNA isdetectable in the steady state of WT ECs (Fig. 5B), it is likelythat endogenously released Klotho induces the basal weak as-sociation of VEGFR-2 and TRPC-1 in the WT ECs. Thesefindings suggested that Klotho induces a cross-link betweenVEGFR-2 and TRPC-1, forming a ternary complex. Because thesize of the intracellular region of Klotho is too small to interactwith other proteins (1), the extracellular domain of Klotho (se-creted form) may bind to VEGFR-2/TRPC-1. We obtained thesecreted form of Klotho from conditioned medium (CM) of NIH3T3-cells infected with Klotho gene recombinant adenovirus (3 ×

WT Kl-/- 0mM 1.5mM

TUNEL

1.5mM

trans- parent

D

(+VEGF)

[Ca 2+ ] o

EC

C

Cas

pase

-3

A

Cal

pain

Calpain Caspase-3

p<0.05

1.0

WT

1

2

3

0 WT Kl-/- Kl-/-

0.5

0

1.5

p<0.01 2.0

WT Kl-/-

+ - + + -

1.5 1.5 -

0

2.5

(mM) VEGF [Ca 2+ ]o

TUNEL + EC (%)

0 5

10

15

20

25

30

+ - + - 1.5 0

+ - + -

1.5 0 WT Kl-/-

Aorta EC

+ -

0 0

Rel

ativ

e va

lues

to W

T

Rel

ativ

e va

lues

to W

T-V

EG

F(-)

2.0

1.5

1.0

0.5

0

(mM)**

**

1

2

3

4 **

* Cleaved (148kDa)

IB anti- αII spectrin

Full length (288 kDa)

*

1

2

0

Aorta

B

WT Kl-/-

Rel

ativ

e in

tens

ity

( cle

aved

tofu

ll le

ngth

)

VEGF [Ca 2+ ]o

Fig. 2. VEGF-mediated apo-ptosis of Klotho-deficient ECsthrough Ca2+/calpain/caspase-3. (A) Calpain and caspase-3 ac-tivity was measured in the aor-tas of 4-wk-old mice (n = 5). (B)Membrane fractionof the aortaof 4-wk-old mice was subjectedto Western blot analysis usinganti-αII-spectrin antibody thatrecognizes the C-terminal le-sion of the calpain-cleaved (148kDa) and full-length (288 kDa)αII-spectrin. The relative in-tensity of the cleaved to thefull-length αII-spectrin wasevaluated (n = 5, *P < 0.005 vs.WT). (C and D) ECs were stimu-latedwith VEGF (100 ng/mL) for30 min in medium containing1.5 mM or 0 mM Ca2+. Calpain/caspase-3 activities (C) andTUNEL-positive apoptotic EC(arrowheads) (D) were evalu-ated. *P < 0.05 and **P < 0.01vs. baseline (n = 10). (Scale bar:50 μm.)

19310 | www.pnas.org/cgi/doi/10.1073/pnas.1008544107 Kusaba et al.

107pfu) (Fig. S2C). The concentration of secreted Klotho in theCM was ∼250 ng/mL, and it was added to the EC with 25% vol-ume (vol/vol). The addition of Klotho protein-rich CM to VEGF-stimulated ECs markedly strengthened the association (3.3-fold,P < 0.01) in both WT and Klotho-deficient ECs (Fig. 5A).Furthermore, TRPC-1 and VEGFR-2 were colocalized on the

cell surface ofWTECs. The addition of VEGF toWTECs causedthe marked internalization of TRPC-1 and VEGFR-2, whereassuch internalization was barely detected in Klotho-deficientECs (Fig. 5C). Total protein contents of TRPC-1 and VEGFR-2were not changed in ECs of either genotype at baseline or afterVEGF stimulation (Fig. S4), indicating a lack of internalization of

TRPC-1 and VEGFR-2 in Klotho-deficient ECs. The addition ofKlotho protein-rich CM to VEGF-stimulated, Klotho-deficientECs restored the lack of VEGF-mediated internalization ofTRPC1 and VEGFR-2 to the normal level (Fig. 5C), but theaddition of Klotho protein-rich CM to VEGF-stimulated WTECs did not affect the internalization patterns of TRPC-1 andVEGFR-2 (Fig. 5C).

Determination of the Binding Site of Klotho and VEGFR-2/TRPC-1.Wedetermined the domains responsible for interactions betweenKlotho and VEGFR-2 or TRPC-1. Recombinant cDNA plasmidsof Klotho mutants deleting hydrolase domains 1 or 2 (ΔHy1 or

BA

0

1 p120 ctn

VE-cad

μ-calpain (0.1 U/ml)

0

ALLN

0 1

1

**

VEGF

p120 ctn

α-tubulin

30 0 0 WT

30 Kl-/-

IP: VE-cad WT Kl -/-

VE-cad

p120 ctn

Blot

(min)

0

1

ALLN

- +

p120

ctn

VE

-cad

+

100 75

50

150

100 75

D

p120 ctn

VEGF

VE-cadherin

30 0 0 WT Kl-/-

ALLN

30 CPZ inhibitors

(min)

β1 integrin

PM

VE

-cad

1 *

0

*

0

1

p120

ctn

- - - -

p120 ctn

EEA

MPR

T C L PM

fract

ion

WT-V

E G F(0)

p120 ctn

VE-cad

p120 ctn VE-cadherin

+VEGF

Interna- lization

Lysozomal degradation

+CPZ +ALLN +VEGF

Calpain-induced p120 ctn proteolysis

Isolation of PM protein

30 0 0 WT

30 Kl-/-

+ (min) VEGF

- - -

1 **

0

Control

+VEGF

WTVE-Cadherin

WT

WT

C

WT Kl -/-

-

Control

+VEGF

- - - - VE-cad

0

α-tubulin

ALLN

- - +

+

*

*

VEGF(-)

Kl-/-

WT

Kl-/-

Kl-/-

p120

ctn

(rel

ativ

e to

WT

(0))

VE

-cad

(r

elat

ive

to W

T (0

))

p120

ctn

Kl-/-

IP:VE-cad WT Kl -/-

VE-cad

VEGF VEGF-R2

TRPC-1

+ + - -

E

Blot

Fig. 3. Calpain-mediated proteolysis of p120ctn induces internalization and degradation of VE-cadherin. (A, C, and E) ECs were stimulated with VEGF (100 ng/mL) for 30 min in the presence or absence of calpain inhibitor (ALLN, 20 μm). Cell lysates were analyzed by Western blotting with antibodies against VE-cadherin, p120ctn, or α-tubulin. Cells also were immunostained with anti-VE-cadherin or anti-p120ctn antibodies (arrowheads). (Scale bars: 5 μm.) Cell lysateswere immunoprecipitated and analyzed byWestern blotting with anti-p120ctn or anti-VE-cadherin antibodies (C) and VE-cadherin, VEGFR-2m, and TRPC-1 (E).*P < 0.01 vs. WT baseline. (B) VE-cadherin and p120ctn were immunoprecipitated from EC lysates. Immunocomplexes were incubated with purified μ-calpain(Calbiochem), final concentration 0.1 U/mL, under 100 μM of Ca2+ for 60 min at 37 °C and then were analyzed by Western blotting. *P < 0.01 and **P < 0.005vs. baseline. (D) ECs were incubated with VEGF, ALLN, or chlorpromazine, an inhibitor for internalization (5 μg/mL). The plasma membrane fraction wasisolated as described in SI Materials and Methods and analyzed by Western blotting with antibodies against VE-cadherin, p120ctn, β1-integrin, EEA-1 (the earlyendosomal marker), or MPR (the late endosomal marker). *P < 0.01 vs. WT baseline.

AVEGFR-2

His-Klothoagarose

agarose bound protein

Lysate

TRPC-1BlotControlagarose

His-Klothoagarose

Controlagarose

B

VEGFR-2

IP:TRPC-1

Ad-Klotho Ad-LacZVEGF

TRPC-1

Klotho

KlothoIP:VEGFR-2

Blot

Blot

+- +-Fig. 4. Klotho binds to VEGFR-2 and TRPC1. (A) HUVECswere infected with the Klotho gene or LacZ generecombinant adenovirus (3 × 107 pfu). Forty-eight hoursafter infection, cellswere stimulatedwithVEGF (100ng/mL)for 30 min, and the lysates were immunoprecipitated, fol-lowed byWestern blotting with antibodies against VEGFR-2 or Klotho. (B) His-tag–fused, Klotho-bound agarose wasincubated with HUVEC-derived lysates, followed by West-ern blotting with anti-VEGFR-2 or TRPC-1 antibody. HUVEClysates also were analyzed by Western blotting.

Kusaba et al. PNAS | November 9, 2010 | vol. 107 | no. 45 | 19311

CELL

BIOLO

GY

ΔHy2) were coexpressed with full-length cDNA plasmids ofVEGF-R2 or TRPC-1 to HEK293 cells. We found that the ΔHy2deletion mutant failed to bind to TRPC-1 or VEGFR-2 (Fig. 5A).Also, cDNA plasmids of VEGF-R2 mutants deleting Ig domains(ΔIg1–7, ΔIg1–4, ΔIg5–7, ΔIg5-6, ΔIg6-7, ΔIg7) (Fig. 5B) orTRPC-1 mutant plasmids deleting N-terminal loop (L) domains(ΔL 1-2, ΔL 1–3, ΔL 1–4, ΔL 1–5) (Fig. 5C) were cotransfectedwith full-length Klotho cDNA. VEGFR-2 mutants lacking thesixth and seventh domains of Ig (ΔIg1–7, ΔIg5–7, ΔIg6-7) orTRPC-1 mutants lacking the third extracellular loop (ΔL1–5) didnot bind to Klotho. To confirm that these domains are responsiblefor association, the cDNA plasmid of Klotho-Hy-2 (Klotho-Hy2)was overexpressed in HEK293 cells, and cell lysates were in-cubated with GST-fused VEGF-R2-Ig6/7, with GST-fusedTRPC-1-L5-domains, or with GST control fixed to glutathioneSepharose beads. Protein bound to the beads was analyzed byWestern blotting. The result showed that Klotho-Hy2 bound di-rectly to both GST-fused proteins but did not bind to controlGST protein (Fig. 5D). Furthermore, we found that in 293 HEKcells coexpressing the EYFP-fused Klotho-Hy2 and ECFP-fusedVEGF-R2-Ig6/7 or ECFP-fused TRPC-1-L5 domains, Klotho-Hy-2 was colocalized with VEGFR-2-Ig6/7 and TRPC-1-L5, andFRET signals were obtained between EYFP-Klotho-Hy2 andECFP-VEGFR-2-Ig6/7 or ECFP-TRPC-1-L5 (Fig. S6 A and B).These findings indicate that the second hydrolase domain ofKlotho interacts directly with the sixth/seventh Ig domains of

VEGFR-2 or the fifth loop (third extracellular loop) domain ofTRPC-1 (Fig. S5E).

Sustained [Ca2+]i Elevation After EGF Stimulation in Klotho-DeficientCells. Enhanced expression of TRPC-1 on the plasma membranebecause of the lack of internalization may augment Ca2+ influxin Klotho-deficient ECs. We therefore evaluated [Ca2+]i inFura-2–loaded aortic ECs. One minute after stimulationwith VEGF in buffer containing 1.5 mM Ca2+, [Ca2+]i increasedto a peak (2.1-fold elevation from the baseline) in WT ECs, fol-lowed by a prompt decrease toward the baseline. By contrast, inKlotho-deficient ECs, after reaching a peak comparable to thatin WT ECs, the elevation of [Ca2+]i was more sustained and didnot return to the baseline level even after 10 min. However, thesustained elevation of [Ca2+]i in Klotho-deficient ECs wasabolished in Ca2+-free buffer (Fig. 6A).Sustained [Ca2+]i elevation after a sharp peak is mediated

through a store-operated Ca2+ channel including TRPCs (15).Therefore, ECswere treatedwith thapsigargin (1 μM) inCa2+-freemedium to deplete the intracellularCa2+ store and thereafter wereexposed to 1.5 mM Ca2+ to induce store-operated Ca2+ current.We observed that the elevation of [Ca2+]i apparently was moreprolonged in Klotho-deficient ECs than in WT ECs (Fig. 6B). Toconfirm that the exaggeratedCa2+ influx inKlotho-deficient cells ismediated by TRPC-1, we knocked down TRPC-1 by siRNA(78% reduction in protein level). In most siRNA-treated cells,TRPC-1 expression on the plasma membrane was barely detect-

-WT Kl-/-

+ + ++- - - -

-VEGFR-2

TRPC-1

Merge

KlothoVEGF

C

A

83kDaTRPC-1

83kDa

195kDa

195kDa

VEGF

KlothoKl-/-

IP:VEGFR-2 TRPC-1

VEGFR-2Blot

VEGFR-2Blot

IP:TRPC-1

WT

++++

++

- -- -

- -

IP:VEGFR

TRPC-1

VEGFR-2

WT(+VEGF)

TRPC-1

VEGFR-2

WT(+VEGF)

IP:TRPC-1

Blot

BKlotho-mRNA

0

1

*

Blot

Fig. 5. Klotho enhances VEGF-mediated internalization ofTRPC-1 and VEGFR-2. (A and C)ECs were incubated with VEGF(100 g/mL) for 30 min. In somesamples, Klotho-rich CM (finalconcentration of Klotho protein∼250 ng/mL) was added 12 h be-fore VEGF treatment. (A) Celllysates were analyzed with im-munoprecipitation, followed byWestern blotting with anti-TRPC-1 or anti-VEGFR-2 antibodies. (C)Cells were immunostained withanti-VEGFR-2 antibody, followedby FITC-conjugated secondaryantibody and Alexa Fluor 568-preconjugated anti-TRPC-1 anti-body. White arrowheads andyellow arrows indicate VEGFR-2/TRPC-1 localized on the cell surface and in the cytoplasm, respectively. (B) siRNA for Klotho or nonsilencing (Cont.) RNA (50 nM final concentration) were in-troduced into ECs, and then cells were stimulated with VEGF (100 g/mL) for 30 min. Total RNA was analyzed by real-time PCR to measure the level of KlothomRNA. ECs also were immunoprecipitated and followed by Western blotting with anti-VEGFR-2 or anti-TRPC-1. *P < 0.005.

A

[Ca2+

]i

0

1.5mM [Ca2+]o

100

200

300

0200 400 600

WTVEGF

Kl-/-

(nM)

Kl-/-WT

*

*

** *

**

0 600 400 200

*

B

100

200

500 1000 15000

0 500 1000 15000

Kl-/-

WT

Thapsigargin (1 mM)

0 mM Ca2+ 1.5 mM Ca2+

*

**

* ****

C

siRNA plus fluorescent RNA-incorporated ECs

Kl-/- EC

[Ca2+

]i

200 400 600 800 1000

200

300

400(nM)

0 200200 400 600 800 100000

(sec)

100

Kl-/-WT

Thapsigargin (1 mM)

0 mM Ca2+ 1.5 mM Ca2+

0mM [Ca2+]o

Kl-/-

WT

100

200

300

00 200 400 600

Kl-/-WT

0 600 400 200

VEGF

(sec)

(sec)

[Ca2+

]i

Fig. 6. Sustained Ca2+ elevation through TRPC-1 inKlotho-deficient ECs. (A) Fura-2–loaded ECs were stimu-lated with VEGF (100 ng/mL) in buffer containing 1.5 mMCa2+ (1.5 mM) or Ca2+-free buffer. (B) Fura-2–loaded ECswere incubated with thapsigargin in Ca2+-free bufferand then were exposed to 1.5 mM Ca2+. Averaged [Ca2+]iis presented in the graph. *P < 0.05 vs. WT ECs at thesame time points. (C) Klotho−/− ECs were introduced withsiRNA against TRPC-1 or nonsilencing (control) RNA.SiRNA-incorporated cells were identified by cotransfec-tion with Alexa Fluor red fluorescent oligo-RNA (Left,arrowheads). They were fura-2–loaded, and the VEGF(100 ng/mL)-induced increase in [Ca2+]i was evaluated.Representative data are shown (Right) (n = 6).

19312 | www.pnas.org/cgi/doi/10.1073/pnas.1008544107 Kusaba et al.

able (Fig. S1A). We labeled siRNA-transfected cells with AlexaFluor red fluorescent oligo-RNA to test the effect of TRPC-1 in-hibition onVEGF-mediatedCa2+ influx (Fig. 6C).VEGF-inducedpeak [Ca2+]i levels were markedly reduced (62± 2% reduction vs.control), and the subsequent [Ca2+]i elevation was restored to thebaseline level, whereas ATP (1 mM)-induced [Ca2+]i levels weresimilar in siRNA and control cells. Furthermore, siRNA-mediatedTRPC-1 knockdown in Klotho-deficient ECs restored the VEGF-mediated decrease in p120ctn/VE-cadherin expression to the WTlevel (Fig. S1B) and significantly decreased the activities of caspase3 and calpain (51%and57%reduction, respectively;P< 0.01) (Fig.S1C) and the number of apoptotic cells (69% reduction; P < 0.01)(Fig. S1D). Thus, enhanced Ca2+ influx via TRPC-1 activatedcalpain/caspase-3/apoptosis in Klotho-deficient ECs.

DiscussionThe present study demonstrates that the disruption of endo-thelial integrity in Klotho-deficient mice results from endothelialhyperpermeability caused by abnormal Ca2+ handling. We alsoprovide evidence that VEGF-mediated Ca2+ signaling is tightlyregulated by interaction with Klotho and that Klotho plays animportant role in VEGF-mediated vascular action as well as inthe maintenance of endothelial function. Furthermore, Klothostrengthens the association between VEGFR-2 and TRPC-1 andcauses cointernalization of VEGFR-2 and TRPC-1, thus regu-lating the expression level of TRPC-1 on the plasma membrane.In contrast, a lack of Klotho results in the impairment of TRPC-1internalization, leading to increased expression of TRPC-1 thatenhances Ca2+ influx. We also found that calpain hyperactivatedby increased Ca2+ influx cleaves p120ctn, leading to subsequentdown-regulation of VE-cadherin in Klotho-deficient ECs. Thesefindings suggest that Klotho deficiency causes a sustained in-crease in [Ca2+]i and hyperactivity of Ca2+-dependent proteases,resulting in vascular hyperpermeability caused by endothelialdamage and, eventually, extensive vascular calcification.Klotho was shown to bind to receptor-type tyrosine kinases,

including FGFRs (10), although no interaction was reported inreceptors for EGF or PDGF (11). Klotho protein also was as-sociated with an epithelial TRP family Ca2+ channel, TRPV5 (9)and with Na+/K+-ATPase (21). The present study demonstratesthe direct association of Klotho with both VEGFR-2 and anendothelial major TRP family Ca2+ channel, TRPC-1. To trans-mit the VEGF-mediated Ca2+ signals effectively, VEGFR-2 andTRPC-1 need to be associated in the plasma membrane. Indeed,our mutagenesis analyses showed that the second hydrolase do-main of Klotho interacts directly with the extracellular sixth/seventh Ig domains of VEGFR-2 and the fifth loop (third ex-tracellular loop) domain of TRPC-1 (Fig. 6E). ECs endoge-

nously released Klotho, which constitutively induced a basalcross-link between VEGFR-2 and TRPC-1, forming a ternarycomplex (Fig. 5), suggesting that Klotho is constitutively associ-ated with TRPC-1 and VEGFR-2 and effectively transmits thephysiological VEGF signal to TRPC-1 that causes Ca2+ influx.We found that VSMCs as well as ECs in the aorta from Klotho-

deficient mice showed apoptotic changes, although VEGFR-2 isexpressed predominantly in endothelium. The mechanism(s) bywhich Klotho promotes VSMC apoptosis through VEGFR-2 re-main to be determined. It has been reported that endothelium-derived NO has an antiproliferative influence on VSMCs, and therelease of endothelial NO is impaired inKlotho-deficientmice (5).Given that VSMCs in proliferative condition are more vulnerableto apoptosis (22), a decrease in endothelialNO inKlotho-deficientECs might cause VSMC apoptosis.In conclusion, the present study presents evidence that Klotho

interacts directly with the extracellular domain of VEGFR-2 andTRPC-1 to regulate VEGF-mediated Ca2+ entry and the hyper-activity of Ca2+-dependent proteases, thus maintaining endo-thelial integrity. Klotho deficiency causes a sustained increase inintracellular [Ca2+]i, resulting in vascular hyperpermeabilitycaused by endothelial apoptosis or loss of endothelial integrityand, eventually, extensive vascular calcification. Klotho expres-sion is reduced in patients with diabetes mellitus or renal failure(2), and the circulating level of Klotho might be involved in thedevelopment of vascular damage in patients with these diseases.Klotho might be a therapeutic target for preventing cardiovas-cular disease that complicates various diseases such as chronickidney disease.

Materials and MethodsMaterials, animal experiments, primary culture of endothelial cells, mea-surement of intracellular calcium concentration, RT-PCR, TUNEL stainingassay, measurements of caspase-3 and calpain activity, in vitro proteolysis ofVE-cadherin and p120ctn by purified calpain, permeability of thoracic aorta,immunocytochemistry, adenovirus-mediated gene transfer, Western blot-ting, knockdown of TRPC-1 and Klotho, construct of deletion mutants,binding assays, FRET analysis, and statistical data analysis are described in SIMaterials and Methods. Four-week-old male Klotho homozygous mutantmice and their wild-type littermates (C57Bl/6 strain) were purchased fromNihon Clea Co., Ltd (Shizuoka, Japan).

ACKNOWLEDGMENTS. We thank Dr. Jun-ichi Nakai (Saitama University,Saitama, Japan) for thoughtful discussions about the FRET experiment andDrs. Takaomi Saido (RIKEN Brain Science Institute, Saitama, Japan) andHiroshi Manya (Tokyo Metropolitan Geriatric Hospital, Tokyo, Japan) forgenerously sharing anti-αII spectrin antibody with us. This work was sup-ported by Grants 15590778 and 18590822 from the Ministry of Education,Culture, Sports, Science and Technology of Japan (to M.O.).

1. Kuro-o M, et al. (1997) Mutation of the mouse klotho gene leads to a syndrome

resembling ageing. Nature 390:45–51.2. Koh N, et al. (2001) Severely reduced production of klotho in human chronic renal

failure kidney. Biochem Biophys Res Commun 280:1015–1020.3. Narumiya H, et al. (2004) HMG-CoA reductase inhibitors up-regulate anti-aging

klotho mRNA via RhoA inactivation in IMCD3 cells. Cardiovasc Res 64:331–336.4. Shimada T, et al. (2004) Angiogenesis and vasculogenesis are impaired in the

precocious-aging klotho mouse. Circulation 110:1148–1155.5. Saito Y, et al. (2000) In vivo klotho gene delivery protects against endothelial dysfunction

in multiple risk factor syndrome. Biochem Biophys Res Commun 276:767–772.6. Haruna Y, et al. (2007) Amelioration of progressive renal injury by genetic

manipulation of Klotho gene. Proc Natl Acad Sci USA 104:2331–2336.7. Mitani H, et al. (2002) In vivo klotho gene transfer ameliorates angiotensin II-induced

renal damage. Hypertension 39:838–843.8. Tohyama O, et al. (2004) Klotho is a novel beta-glucuronidase capable of hydrolyzing

steroid beta-glucuronides. J Biol Chem 279:9777–9784.9. Chang Q, et al. (2005) The beta-glucuronidase klotho hydrolyzes and activates the

TRPV5 channel. Science 310:490–493.10. Urakawa I, et al. (2006) Klotho converts canonical FGF receptor into a specific receptor

for FGF23. Nature 444:770–774.11. Kurosu H, et al. (2005) Suppression of aging in mice by the hormone Klotho. Science

309:1829–1833.

12. Nilius B, Droogmans G (2001) Ion channels and their functional role in vascularendothelium. Physiol Rev 81:1415–1459.

13. Inoue R, et al. (2006) Transient receptor potential channels in cardiovascular functionand disease. Circ Res 99:119–131.

14. Ahmmed GU, Malik AB (2005) Functional role of TRPC channels in the regulation ofendothelial permeability. Pflugers Arch 451:131–142.

15. Jho D, et al. (2005) Angiopoietin-1 opposes VEGF-induced increase in endothelialpermeability by inhibiting TRPC1-dependent Ca2+ influx. Circ Res 96:1282–1290.

16. Yao X, Garland CJ (2005) Recent developments in vascular endothelial cell transientreceptor potential channels. Circ Res 97:853–863.

17. Manya H, et al. (2002) Klotho protein deficiency leads to overactivation of μ-calpain.J Biol Chem 277:35503–35508.

18. Tiruppathi C, Minshall RD, Paria BC, Vogel SM, Malik AB (2002) Role of Ca2+ signalingin the regulation of endothelial permeability. Vascul Pharmacol 39:173–185.

19. Xiao K, et al. (2003) Cellular levels of p120 catenin function as a set point for cadherinexpression levels in microvascular endothelial cells. J Cell Biol 163:437–440.

20. Yuan JP, et al. (2003) Homer binds TRPC family channels and is required for gating ofTRPC1 by IP3 receptors. Cell 114:777–789.

21. Imura A, et al. (2007) alpha-Klotho as a regulator of calcium homeostasis. Science 316:1615–1618.

22. Matsuda A, Suzuki Y, Kondo K, Ikeda Y, Umemura K (2002) Hypercholesterolemiainduces regression in neointimal thickening due to apoptosis of vascular smoothmuscle cells in the hamster endothelial injury model. Cardiovasc Res 53:512–523.

Kusaba et al. PNAS | November 9, 2010 | vol. 107 | no. 45 | 19313

CELL

BIOLO

GY