ARTICLE

Thrombomodulin domain 1 ameliorates diabetic nephropathyin mice via anti-NF-κB/NLRP3 inflammasome-mediatedinflammation, enhancement of NRF2 antioxidant activityand inhibition of apoptosis

Shun-Min Yang & Shuk-Man Ka & Hua-Lin Wu & Yu-Chuan Yeh & Cheng-Hsiang Kuo &

Kuo-Feng Hua & Guey-Yueh Shi & Yi-Jen Hung & Fone-Ching Hsiao & Sung-Sen Yang &

Yi-Shing Shieh & Shih-Hua Lin &Chyou-WeiWei & Jeng-Shin Lee &Chu-Yi Yang &Ann Chen

Received: 6 August 2013 /Accepted: 29 October 2013 /Published online: 30 November 2013# Springer-Verlag Berlin Heidelberg 2013

AbstractAims/hypothesis Chronic inflammatory processes havebeen increasingly shown to be involved in the pathogen-esis of diabetes and diabetic nephropathy. Recently, wedemonstrated that a lectin-like domain of thrombomodulin(THBD), which is known as THBD domain 1 (THBDD1)

and which acts independently of protein C activation,neutralised an inflammatory response in a mouse modelof sepsis. Here, therapeutic effects of gene therapy withadeno-associated virus (AAV)-carried THBDD1 (AAV-THBDD1 ) were tested in a mouse model of type 2diabetic nephropathy.

Electronic supplementary material The online version of this article(doi:10.1007/s00125-013-3115-6) contains peer-reviewed but uneditedsupplementary material, which is available to authorised users.

S.<M. Yang :Y.<C. Yeh :A. Chen (*)Department of Pathology, Tri-Service General Hospital, NationalDefense Medical Center, No. 325, Sec. 2, Cheng-Gung Road, Taipei,Taiwan, Republic of Chinae-mail: doc31717@ndmctsgh.edu.tw

S.<M. Ka (*)Graduate Institute of Aerospace and Undersea Medicine, NationalDefenseMedical Center, No. 161, Sec. 6,Min-Quan E. Road, Taipei,Taiwan, Republic of Chinae-mail: mariaka@ndmctsgh.edu.tw

H.<L. Wu : C.<H. Kuo :G.<Y. ShiDepartment of Biochemistry andMolecular Biology, National ChengKung University, Tainan, Taiwan, Republic of China

H.<L. Wu : C.<H. Kuo :G.<Y. ShiCardiovascular Research Center, National Cheng Kung University,Tainan, Taiwan, Republic of China

H.<L. WuCenter for Bioscience and Biotechnology, National Cheng KungUniversity, Tainan, Taiwan, Republic of China

K.<F. HuaDepartment of Biotechnology and Animal Science, National IlanUniversity, Ilan, Taiwan, Republic of China

Y.<J. Hung : F.<C. HsiaoDivision of Endocrinology and Metabolism, Tri-Service GeneralHospital, National Defense Medical Center, Taipei, Taiwan,Republic of China

S.<S. Yang : S.<H. LinDivision of Nephrology, Tri-Service General Hospital, NationalDefense Medical Center, Taipei, Taiwan,Republic of China

Y.<S. ShiehDepartment of Dentistry, Tri-Service General Hospital, NationalDefense Medical Center, Taipei, Taiwan,Republic of China

C.<W. WeiDepartment of Nutrition, Master Program of Biomedical Nutrition,Hungkuang University, Taichung, Taiwan,Republic of China

J.<S. LeeHarvard Gene Therapy Initiative, Harvard Medical School, Boston,MA, USA

C.<Y. YangGraduate Institute of Life Sciences, National Defense MedicalCenter, Taipei, Taiwan, Republic of China

Diabetologia (2014) 57:424–434DOI 10.1007/s00125-013-3115-6

Methods To assess the therapeutic potential of THBDD1 andthe mechanisms involved, we delivered AAV-THBDD1 (1011

genome copies) into db/db mice and tested the effects ofrecombinant THBDD1 on conditionally immortalisedpodocytes.Results A single dose of AAV-THBDD1 improved albu-minuria, renal interstitial inflammation and glomerularsclerosis, as well as renal function in db/db mice. Theseeffects were closely associated with: (1) inhibited activa-tion of the nuclear factor κB (NF-κB) pathway and theNACHT, LRR and PYD domains-containing protein 3(NLRP3) inflammasome; (2) promotion of nuclear factor(erythroid-derived 2)-like 2 (NRF2) nuclear translocation;and (3) suppression of mitochondria-derived apoptosis inthe kidney of treated mice.Conclusions/interpretation AAV-THBDD1 gene therapy re-sulted in improvements in a model of diabetic nephropathy bysuppressing the NF-κB–NLRP3 inflammasome-mediated in-flammatory process, enhancing the NRF2 antioxidant path-way and inhibiting apoptosis in the kidney.

Keywords Adeno-associated virus . db/db mice . Diabeticnephropathy .Gene therapy .NLRP3 inflammasome .NRF2 .

Thrombomodulin domain 1

AbbreviationsAAV Adeno-associated virusAPC Activated protein CBAX BCL-2-associated X proteinBCL-2 B cell leukaemia/lymphoma 2BUN Blood urea nitrogenDN Diabetic nephropathyGPx Glutathione peroxidaseHG High glucoseHMGB1 High-mobility group box 1HO-1 Haem oxygenase-1MCP-1 Monocyte chemoattractant protein 1NF-κB Nuclear factor kappa-light-chain-enhancer of

activated B cellsNG Normal glucoseNLRP3 NACHT, LRR and PYD domains-containing

protein 3NRF2 Nuclear factor (erythroid-derived 2)-like 2PI Propidium iodideRLU Relative luminescence unitsROS Reactive oxygen speciesrTHBDD1 Recombinant THBDD1THBD ThrombomodulinTHBDD1 THBD domain 1VEGF Vascular endothelial growth factorWT Wild-type

Introduction

Type 2 diabetes is a leading cause of end-stage kidneydisease due to diabetic nephropathy (DN) [1, 2], which ischaracterised by progressive albuminuria and a gradualdecline in GFR, in association with progressive renalfibrosis or sclerosis, leading to ultimate chronic renalfailure [3]. Although many therapeutic remedies focusingon hyperglycaemia and high blood pressure have beenused, many patients still suffer progressive and severerenal injury [4], making the investigation of other pathogenicpathways and relevant therapeutic strategies worthwhile.Importantly, immune-mediated inflammatory processes [5],apoptosis [6, 7] and reactive oxygen species (ROS) [5, 8] inthe kidney are involved in the developmental and progressivestages of DN.

Thrombomodulin (THBD), a component of the anticoagu-lant pathway, binds to thrombin on the endothelial cell surfaceand catalyses activation of protein C, thus protecting cellsfrom thrombus formation [9, 10]. THBD consists of fivedomains: a lectin-like domain of THBD known as THBDdomain 1 (THBDD1), followed by a domain containing sixepidermal growth factor repeats, a serine- and threonine-richdomain, a transmembrane domain and a cytoplasmic domain.Previously, recombinant THBD has been shown to improvecertain types of acute renal damage, such as thrombotic glo-merulonephritis [11] and acute ischaemic renal injury [12].However, studies by our team [13] and others [14] haveshown that THBD also has a protein C- and thrombin-independent physiological function. In agreement with thisnotion, we recently demonstrated that recombinant THBDD1(rTHBDD1 ) can inhibit an inflammatory response byneutralising lipopolysaccharide through binding to Lewis Yantigen [15] and has anti-angiogenesis effects on culturedHUVECS [16]. We also demonstrated that cardiomyocytesare protected from apoptosis by THBD treatment as a result ofits effect on the balance of B cell leukaemia/lymphoma 2(BCL-2) and BCL-2-associated X protein (BAX) levels[17]. The anti-apoptotic effect of THBD has also been con-firmed in studies using endothelial cells and neurons [12].

In the present study, we delivered adeno-associated virus(AAV)-carried THBDD1 (AAV-THBDD1) into db/db mice,which had already developed hyperglycaemia. We found thatthese animals were protected from renal inflammation andfibrosis or sclerosis through the inhibition of inflammation,activation of the nuclear factor (erythroid-derived 2)-like 2(NRF2) antioxidant pathway and prevention of apoptosis.

Methods

Preparation of rTHBDD1 protein and AAV-THBDD1 Theproduction of rTHBDD1 and AAV serotype 2/8 vectors

Diabetologia (2014) 57:424–434 425

encoding human THBDD1 (amino acid 1-155) (AAV-THBDD1 ) has been described previously [16].

Animal experiments Male C57BLKs/J db/db mice and theirage-matched, normal, wild-type (WT) db/m littermates werepurchased from Jackson Laboratories (Bar Harbor, ME,USA). The mice were classified into three groups (n =12each): (1) AAVempty vector-treated db/db mice (vector-alonecontrol db/db ); (2) AAV-THBDD1-treated mice; and (3) WTmice as normal control. We injected the mice intravenouslywith a single dose of AAV or AAV-THBDD1 . This dose(1011 genome copies) was chosen according to our previousstudy because it caused no detectable pathological effects in themice [16] by the age of 32 weeks (week 0 after the AAV orAAV-THBDD1 injection), i.e. at a time point whenhyperglycaemia and hyperglycosuria had already developedand just before the onset of proteinuria. The db/db mice injectedwith the empty AAV vector served as disease controls. All themice were killed 18 weeks after the AAV or AAV-THBDD1injections and samples were collected for individual analyses asdescribed previously [7], with slight modifications. All animalexperiments were performed after approval by the InstitutionalAnimal Care and Use Committee of The National DefenseMedical Center, Taiwan, and are consistent with the NIHGuidefor the Care and Use of Laboratory Animals.

Clinical and histopathological evaluation The collection andassay of blood and urine samples were done as describedpreviously [7]. For histopathological evaluation, mesangialexpansion (increased matrix), and sclerosis and hyalinosis inthe glomerulus, as well as renal interstitial inflammationwere evaluated by light microscopy and scored as describedpreviously [18].

Immunofluorescence Frozen sections of the kidney weresubjected to immunofluorescence staining with goat anti-mouse IgG, IgM and C3 (Sigma, St Louis, MO, USA). Thesource of FITC-conjugated, labelled IgG, IgM and C3 hasbeen described previously [19].

Confocal microscopy of podocyte injury in vivo Frozen sectionswere stained with guinea pig anti-nephrin antibody (AcrisAntibodies, Herford, Germany) and rabbit antibody againstvascular endothelial growth factor (VEGF) (Santa CruzBiotechnology, Dallas, TX, USA), followed by AlexaFluor 488-labelled rabbit anti-guinea pig IgG antibody(Invitrogen, Grand Island, NY, USA) or phycoerythrin-labelledgoat anti-rabbit IgG antibody (Jackson ImmunoResearch, WestGrove, PA, USA) as described previously [7].

Western blot analysis Rabbit antibodies against mouseNACHT, LRR and PYD domains-containing protein 3(NLRP3) (Zymed, South San Francisco, CA, USA),

caspase-3, caspase-8 or caspase-9 (Cell Signaling Technology,Frankfurt amMain, Germany), and BCL-2 or BAX, as well asgoat antibodies against mouse NRF2 or β-actin (Santa CruzBiotechnology), were used as described previously [18].

ELISA Serum levels of IL-1β and monocyte chemoattractantprotein 1 (MCP-1) (eBioscience, San Diego, CA, USA), andrenal haem oxygenase-1 (HO-1) levels (Enzo Life Sciences,Farmingdale, NY, USA) were measured using commercialELISA kits according to the manufacturers’ instructions.

Enzyme activity assay Renal glutathione peroxidase (GPx)activity (Cayman, Ann Arbor, MI, USA), and caspase-1,caspase-3 and caspase-9 activity (R&D, Minneapolis, MN,USA)weremeasured using commercial ELISA kits accordingto the manufacturers’ instructions. Renal phosphorylated nu-clear factor κB (NF-κB) p65 activity was measured using anELISA assay kit (TransAM; Active Motif, Carlsbad, CA,USA) according to the manufacturer’s instructions and amulti-mode ELISA plate reader (Bio-Tek, Winooski, VT,USA).

Measurement of ROS production ROS levels in serum andrenal tissue were measured as described previously [18].ROS activity was expressed as relative luminescenceunits (RLU) per 15 min per mg of organ dry weight, i.e.RLU/mg or RLU/ml.

In vitro analyses of podocytes Conditionally immortalisedmouse podocytes were obtained from S. J. Shankland(Department of Medicine, University of Washington, Seattle,WA, USA) and differentiation induced by culturing the cellsfor 13 to 15 days at 37°C in the absence of IFN-γ (Roche,Nutley, NJ, USA), as described previously [20].

Apoptosis Podocytes (5×104) were incubated for 1 h with orwithout rTHBDD1 (5, 17 or 50μg/ml), and then incubated for12 or 24 h with medium containing normal glucose (NG)(5 mmol/l D-glucose) as a control or high glucose (HG)(30 mmol/l D-glucose). The cells were then collected andanalysed for (1) caspase-3 protein abundance by western blotanalysis and (2) apoptosis by flow cytometry, as describedpreviously [21].

Protein content of high-mobility group box 1 and MCP-1 inpodocytes Podocytes (1×104) were incubated with or withoutrTHBDD1 (50 μg/ml), and with NG (5 mmol/l D-glucose),HG (30 mmol/l D-glucose) or NG (5 mmol/l D-glucose)+mannitol (25 mmol/l; used as an osmolality control). Forimmunofluorescence, the cells were incubated with rabbitanti-high-mobility group box 1 (HMGB1) (Abcam, Cambridge,UK) antibody, followed by FITC-labelled goat anti-rabbit IgGantibody (Jackson ImmunoResearch) and counterstaining with

426 Diabetologia (2014) 57:424–434

DAPI (Sigma). Cells were then measured by fluorescence mi-croscopy at amagnification of ×400. The levels ofMCP-1 in theculture medium were measured by ELISA (eBioscience) ac-cording to the manufacturer’s instructions.

Statistical analysis Values are presented as the mean ± SEM.Comparisons of serum levels of THBD and sugar, and urinelevels of albumin and glucose between groups were madewith one-way ANOVA. Comparison between two groupswas performed using Student’s t test. A value of p <0.05was considered statistically significant.

Results

Expression of Thbdd1 mRNA in liver tissue and THBDD1protein in serum mRNA levels of Thbdd1 were significantlyelevated in liver tissue of AAV-THBDD1 -treated db/db mice(Electronic supplementary material [ESM] Fig. 1a, methodsare provided in the ESM Methods) compared with vector-alone control db/db mice (AAV-injected only) and controldb/m (WT) mice at week 18 when the animals were killed.At weeks 1 and 2, AAV-THBDD1 -treated db/db mice hadincreased serum THBDD1 levels compared with those ofvector-alone control db/db mice and WT mice, respectively,

with levels peaking in week 4 after AAV-THBDD1administration and remaining persistently high until week 18(p <0.005) (ESM Fig. 1b, methods are provided in the ESMMethods). These findings indicate that systemic THBDD1expression was present in the circulation of animals thatreceived the gene therapy.

THBDD1 improves albuminuria and renal function, but has noeffect on the treatment of hyperglycaemia or hyperglycosuria Asshown in Fig. 1a, vector-alone control db/db mice firstshowed a significantly higher urinary albumin:creatinine ratiothan WT mice at week 3 after injection, with this ratio con-tinuing to increase until animals were killed. This effect wasmarkedly decreased in AAV-THBDD1 -treated db/db mice(p <0.01). At week 18 after injection, AAV-THBDD1 -treateddb/db mice had significantly reduced serum levels of bloodurea nitrogen (BUN) (Fig. 1b) and creatinine (Fig. 1c)(p <0.05 for both) compared with vector-alone control db/dbmice. However, there was no significant difference in serumor urinary glucose levels between AAV-THBDD1 -treated andvector-alone control db/db mice (serum glucose AAV-THBDD1 -treated db/db 32±1.7 mmol/l, vector-alone controldb/db 33.8±1.9 mmol/l; urinary glucose AAV-THBDD1 -treated db/db 5+ (>111.1 mmol/l) vs vector-alone controldb/db 5+ (>111.1 mmol/l), although both groups had signif-icantly higher serum and urinary glucose levels thanWTmice

gWT AAV-THBDD1 + db/db

HE

d

**†

ND

†

Glo

mer

ulos

cleo

sis

(%)

a

Uri

nary

alb

umin

/Cr

(mg/

mm

ol)

Weeks after THBDD1administration

****

†

1813830

80

60

40

20

0-26 -24

AAV-THBDD1injection

b

Seru

m B

UN

(m

mol

/l) †*

**

c

Seru

m C

r (m

mol

/l) ** *NS

fVector-alone control db/db

e

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

0.10

0.08

0.06

0.04

0.02

0.00

100

80

60

40

20

0

10

15

0

5

Fig. 1 Proteinuria, renal function and renal pathology. (a) Time course ofappearance of proteinuria. Black circles, WT (db/m ); white circles,vector-alone control db/db ; black triangles, AAV-THBDD1 -treateddb/db ; Cr, creatinine. The arrow indicates the time of injection ofAAV-THBDD1 or vector. (b ) Serum BUN levels at week 18 after

injection. (c ) Serum creatinine levels at week 18 after injection.(d–f ) Haematoxylin and eosin (HE) staining. Arrows indicate zonesof fibrosis: original magnification ×400. (g ) Scoring of glomerularsclerosis. ND, not detectable. *p <0.05, **p <0.01 and †p <0.005

Diabetologia (2014) 57:424–434 427

(serum glucose 6.5±0.7 mmol/l, urinary glucose 0+; p <0.005for both). There was no difference in physical activity betweenAAV-THBDD1-treated and vector-alone control db/db mice,although both groups of mice were obviously overweightcompared with WT mice. No evidence of spontaneous inter-nal bleeding was seen in AAV-THBDD1 -treated db/db mice.

THBDD1 improves renal pathological lesions Light micros-copy revealed that vector-alone control db/db mice had dif-fuse mesangial expansion (increased matrix), and scatterednodular sclerosis and hyalinosis in the glomeruli (Fig. 1d–g),associated with disperse mononuclear leucocyte infiltration ofthe renal interstitium, although the renal tubules were mainlyintact. However, these changes were substantially improved inAAV-THBDD1-injected db/db mice. As shown by immuno-histochemistry, levels of collagen IV (ESM Fig. 2, methodsare provided in the ESM Methods) were significantly higherin vector-alone control db/db mice than in WT mice andsubstantially lower in AAV-THBDD1 -treated db/db mice(p <0.05). Immunofluorescence showed that IgG, IgM andC3 were diffusely present in the glomerulus of vector-alonecontrol and AAV-THBDD1 -treated db/db mice compared

with the negative staining seen in the kidney of WT mice.There was no significant difference in IgG, IgM or C3deposits between vector-alone control and AAV-THBDD1 -treated db/db mice (data not shown).

THBDD1 decreases serum and renal ROS levels, andincreases renal NRF2 nuclear translocation Vector-alonecontrol db/db mice had significantly higher serum (Fig. 2a)and renal (Fig. 2b) levels of superoxide anions thanWTmice,this effect being significantly reduced in AAV-THBDD1 -treated db/db mice (p <0.005 for both). Consistent with this,renal AAV-THBDD1 -treated db/db mice had fewerdihydroethidium-positive cells than vector-alone controldb/db mice (ESM Fig. 3, methods are provided in the ESMMethods).Wemeasured protein levels of NRF2 in the nucleus(active NRF2), and GPx activity and HO-1 levels (both in-volved in the NRF2 pathway) in renal tissues. Compared withvector-alone control db/db mice, AAV-THBDD1 -treateddb/db mice had significantly higher nuclear NRF2 proteinlevels (Fig. 2c, d). Vector-alone control db/db mice also hadmuch lower GPx activity than WT mice, an effect preventedby AAV-THBDD1 treatment (Fig. 2e), while HO-1 content

a

Seru

m s

uper

oxid

e an

ions

(RL

U [

15 m

in]-

1 ml-1

)

Kid

ney

supe

roxi

de a

nion

s(R

LU

[15

min

]-1 [

mg

dryw

eigh

t]-1

)bNS

† †

NS† †

cWT

70 200

160

120

80

40

0

60

50

40

30

20

10

0

Vector-alonecontrol db/db

AAV-THBDD1db/db

NRF2

kDa

55

45

NR

F2:β

-act

in r

atio

dNS

** †

β-actin

1.6

1.2

0.8

0.4

0.0

GPx

act

ivity

(nm

ol m

in-1

mg-1

)

e †

NS †

HO

-1 (

ng/µ

g)

fNS

**

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

500 30

20

10

0

400

300

200

100

0

Fig. 2 ROS levels and NRF2antioxidant signalling pathway inthe kidney. (a) ROS levels inserum or (b) renal tissues. (c)Representative western blots fornuclear NRF2 with β-actin as theloading control and (d) semi-quantitative analysis of theNRF2:β-actin ratio from thewestern blot data. (e) Renalcytosolic GPx activity and (f)HO-1 levels. *p <0.05, **p <0.01and †p <0.005

428 Diabetologia (2014) 57:424–434

was significantly higher in AAV-THBDD1 -treated db/dbmice than in vector-alone control db/db or WT mice (Fig. 2f).

THBDD1 inhibits NF-κB activation, MCP-1 expression andinfiltration of macrophages in the kidney As shown in Fig. 3a,vector-alone control db/db mice had significantly greaterrenal nuclear pNF-κB p65 activity than WT mice, thiseffect being completely suppressed in AAV-THBDD1 -treated db/db mice (p <0.005). Vector-alone control db/dbmice also had significantly higher MCP-1 levels in serumthan WT mice as assessed by ELISA (Fig. 3b). This effectwas markedly inhibited in AAV-THBDD1 -treated db/dbmice. Furthermore, significant renal interstitial infiltration ofmacrophages (F4/80-positive) was seen in vector-alonecontrol db/db mice compared with WT mice and wasalso markedly inhibited in AAV-THBDD1 -treated db/dbmice (Fig. 3c–e). In addition to the above, we also performedan in vitro experiment. As shown in Fig. 3f, rTHBDD1significantly suppressed protein levels of MCP-1 in HG-treated podocytes at early (12 h) and late (24 h) time points.There was no significant difference in MCP-1 protein levelsbetween osmolality control (NG + mannitol), osmolality

control + rTHBDD1 or NG + rTHBDD1 compared with NGalone conditions at 12 and 24 h, respectively.

THBDD1 inhibits NLRP3 inflammasome activation andHMGB1 production Vector-alone control db/db mice hadmarkedly higher serum levels of IL-1β (Fig. 4a) and IL-18(Fig. 4b) than WT mice; levels of both were significantlydecreased in AAV-THBDD1 -treated db/db mice. RenalNLRP3 levels were significantly increased in vector-alonecontrol db/db mice compared with WT mice, while AAV-THBDD1 treatment resulted in a significant reduction ofNLRP3 levels (Fig. 4c, d). AAV-THBDD1-treated mice alsohad significantly decreased renal caspase-1 activity comparedwith vector-alone control db/db mice (Fig. 4e). Immunohis-tochemistry studies showed that renal IL-1β expression invector-alone control db/db mice was significantly higher thanin WT mice and that this effect was markedly inhibited inAAV-THBDD1-treated db/db mice (ESM Fig. 4, methods areprovided in the ESM Methods).

Immunofluorescence studies showed that cytosolicHMGB1 protein content was significantly increased insaline-treated podocytes cultured in HG medium compared

d e

Glo

mer

ular

F4/

80 (

cells

/gcs

)

Inte

rstit

ial F

480

(cel

ls/f

ield

)

f

MC

P-1

(pg/

ml)

12 h 24 h

*

b

Seru

m M

CP-

1 (p

g/m

l)

a

NF-

κB p

65 a

ctiv

ity (

pg/m

l)

†NS

†**

NS**

F4/8

0

c WT

10 50120

100

80

60

40

20

0

40

30

20

10

0

8

6

4

2

0

100

80

60

40

20

0

1000

800

600

400

200

0

Vector-alone control db/db AAV-THBDD1 db/db

NSNS NS NS

*

†† NS

NS

NSNS

NS

++++++++++++

+++

++

- ---- ---- ---

- ---- - ++ - ---- -++++- ---

- - - - ----- + +- - - - +- + +- - - -

- - - --- +- - - - ---+- - - - ---

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

Glu. 5 mmolGlu. 30 mmol

Man. 25 mmolrTHBDD1 5 µg/ml

rTHBDD1 17 µg/mlrTHBDD1 50 µg/ml

Fig. 3 Renal NF-κB activationand related cytokines. (a)Renal pNF-κB p65 activityand (b) serum MCP-1 levelsmeasured by ELISA. (c)Immunohistochemistry study ofrenal F4/80 (macrophage)expression. Arrows indicateF4/80-positive cells; originalmagnification ×400. (d)Percentage of positive staining forF4/80 in the glomerulus and (e)tubulointerstitial compartment.gcs (d), glomerular cross section.(f) MCP-1 in podocytes culturedin medium, measured by ELISA.Glu., glucose; Man., mannitol.*p <0.05, **p<0.01 and†p <0.005

Diabetologia (2014) 57:424–434 429

with NG medium and that this HG effect was significantlyinhibited in cells treated with rTHBDD1 (Fig. 4f). There wasno significant difference in cytosolic HMGB1 protein contentin osmolality control (NG + 25 mmol/l mannitol) comparedwith NG conditions.

THBDD1 prevents loss of nephrin and inhibits VEGF expressionin the glomerulus As shown in Fig. 5, vector-alone controldb/db mice had less nephrin in the glomeruli than WT mice(22.7±0.2% of the total glomerular area positive for nephrin,compared with 39.9±1.4% in WT; p <0.05), while AAV-THBDD1 -treated db/db mice had significantly increasednephrin content in the glomerulus (32.8±1.5%) comparedwith vector-alone control db/db mice (p <0.05). Vector-alonecontrol db/db mice also had significantly increased glomer-ular VEGF content in podocytes (19.7±0.5% of the totalglomerular area positive for VEGF) compared with WTmice (3.2±1.8%) (p <0.01), whereas AAV-THBDD1 -treateddb/db mice had only faint VEGF staining (5.6±0.2%;p <0.01 compared with vector-alone control db/db mice).

THBDD1 inhibits apoptosis in the glomerulus and renaltubule As shown by TUNEL staining, vector-alone controldb/db mice had greatly increased renal apoptosis mainly inpodocytes compared with WT mice, an effect that was signif-icantly inhibited (p< 0.05) in AAV-THBDD1 -treated db/dbanimals (ESM Fig. 5, methods are provided in the ESM

Methods). Examination of renal levels of activated caspase-8and caspase-9 revealed no evidence for caspase-8 activation invector-alone control db/db or AAV-THBDD1-treated db/dbmice compared with WT mice (data not shown), but levels ofthe mature form (p35 fragment) of caspase-9 were greatlyincreased in vector-alone control db/db mice compared withWT mice, this effect being suppressed in AAV-THBDD1 -treated db/db mice (p< 0.005) (Fig. 6a, b). In addition, themature form (p17 fragments) of renal caspase-3 was increased

HG (30 mmol/l)NG (5 mmol/l) HG (30 mmol/l) + rTHBDD1NG (5 mmol/l)

+ mannitol (25 mmol/l)

ckDa

10645

WTVector-alonecontrol db/db

AAV-THBDD1db/db

β-ActinNLRP3

f

HM

GB

1

Seru

m I

L-1

β (p

g/m

l)

Seru

m I

L-1

8 (p

g/m

l)

NL

RP3

: β-a

ctin

rat

io

a dNS

NS** **

NS † †

NSb

† †

e

Cas

pase

-1 a

ctiv

ity(a

bsor

banc

e/m

g pr

otei

n)

† **

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

160 10000.25 16

12

8

4

0

0.20

0.15

0.10

0.05

0.00

800

600

400

200

0

120

80

40

0

Fig. 4 NLRP3 inflammasome activation. (a) Serum levels of IL-1β or(b) IL-18 measured by ELISA. (c) Representative western blots for renalcytosolic NLRP3 with β-actin as loading control and (d ) semi-quantitative analysis of the NLRP3: β-actin ratio from the western blot

data. (e) Renal cytosolic caspase-1 activity by ELISA. (f) Immunofluo-rescence staining of HMGB1 in cultured podocytes incubated in mediumas indicated; original magnification ×400. Arrows indicate positivestaining. **p <0.01 and †p <0.005

Nep

hrin

AAV-THBDD1db/dbWT

Mer

geV

EG

F

Vector-alonecontrol db/db

Fig. 5 Expression of nephrin and VEGF in the glomerulus. Immunoflu-orescence staining of the glomerulus using anti-nephrin antibody (green)or anti-VEGF antibody (red) with confocal laser scanning microscopy.Arrows indicate double positive staining; original magnification ×800

430 Diabetologia (2014) 57:424–434

in vector-alone control db/db mice, but this effect was greatlysuppressed in AAV-THBDD1 -treated db/db mice (p <0.005)(Fig. 6a, c). We also measured renal levels of BAX and BCL-2, finding that the ratio was increased in vector-alone controldb/db mice compared with WT mice, an effect that wassignificantly decreased (p <0.01) in AAV-THBDD1 -treateddb/db mice (Fig. 6a, d). AAV-THBDD1-treated mice alsohad significantly decreased renal caspase-9 and caspase-3activity compared with vector-alone control db/db mice(Fig. 6e, f). These findings suggest that the THBDD1-induced reduction in renal apoptosis is due to suppression ofthe intrinsic pathway of apoptosis and reduction of the BAX:BCL-2 ratio.

We examined the time course and dose-dependent effectsof rTHBDD1 treatment on cultured podocytes, which havebeen shown to undergo apoptosis in HG medium [22]. Asshown in Fig. 7a, b, at 24 h the percentage of annexin V- andpropidium iodide (PI)-positive cells was significantly increasedin HG-treated compared with NG-treated podocytes, this HGeffect being significantly inhibited in cells treated withrTHBDD1 at 50 μg/ml (p <0.005). This finding was confirmedby the observation that the increased levels of the mature form(p17) of caspase-3 seen in HG-treated podocytes werecompletely suppressed in rTHBDD1 -treated podocytes at24 h (p <0.005) (Fig. 7c, d). No difference in the percentage

of apoptotic cells was observed between NG-treated, HG-treated and HG + rTHBDD1-treated podocytes at 12 h (datanot shown). In addition, only 50 μg/ml of rTHBDD1 wascapable of inhibiting apoptosis in HG-treated podocytes, asdemonstrated by flow cytometry and western blot analysis.

Discussion

In a recent study, we showed that administration of AAVserotype 2/8 carrying THBDD1, which was also used in thepresent study, can cause hepatic expression of THBDD1 andrelease of the protein into blood, thereby inhibiting angiogene-sis in rats [16]. Other authors [23] have shown that recombinantAAV serotype 2/8 vectors are a promising tool for long-termtissue protein expression in vivo. Consistent with this, weshowed significantly increased expression of Thbdd1 mRNAin liver tissue of AAV-THBDD1-treated db/db mice, with theTHBDD1 protein released into blood exerting a renoprotectiveeffect on renal lesions of the treated mice.

Our data provide the first demonstration, to our knowledge,that AAV-THBDD1 gene therapy exerts its beneficial effectsin db/db mice as a result of anti-inflammatory effects causedby suppressing the activation of NF-κB and the NLRP3

aWT

kDa

Casp3 p1717

Casp9 p3535

BCL-226

BAX23

β−Actin45

b c d

Cas

p9 p

35:β

-act

in r

atio

Cas

p3 p

17:β

-act

in r

atio

BA

X:B

CL

-2 r

atio

e f

Cas

pase

-9 a

ctiv

ity(a

bsor

banc

e/m

g pr

otei

n)

Cas

pase

-3 a

ctiv

ity(a

bsor

banc

e/m

g pr

otei

n)

Vector-alonecontrol db/db

AAV-THBDD1db/db

NS NSNS0.4 0.25 1.8

1.5

1.2

0.9

0.6

0.3

0.0

0.20

0.15

0.10

0.05

0.00

0.3

0.2

0.1

0.0

3.0 12

8

4

0

2.5

2.0

1.5

1.0

0.5

0.0

† † † † * **

NS † †NS

† **

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

WT

Vector

-alon

e

contr

ol db/db

AAV-THBDD1

db/db

Fig. 6 Western blot analysis forapoptosis. (a) Representativewestern blots for renal caspase-9p35, caspase-3 p17, BCL-2 andBAX, with semi-quantitativeanalysis of the (b) active caspase-9 p35:β-actin, (c) caspase-3p17:β-actin and (d) BAX:BCL-2ratios from the western blot data.(e) Renal cytosolic caspase-9 and(f) caspase-3 activity by ELISA.Cultured podocytes wereincubated in NG medium(5 mmol/l), HG medium(30 mmol/l) or HG plusrTHBDD1 (5, 17 or 50 μg/ml).*p <0.05, **p<0.01 and†p <0.005

Diabetologia (2014) 57:424–434 431

inflammasome, enhancing the NRF2 antioxidant pathway andinhibiting mitochondria-derived apoptosis in the kidney.

First, AAV-THBDD1 administration resulted in inhibitionof NF-κB activation, F4/80-positive macrophage infiltrationand NLRP3 inflammasome activation in the kidney comparedwith vector-alone control db/db mice. These results show that,although intact THBD can activate protein C and activatedprotein C (APC) has anti-inflammatory and anticoagulationproperties, the effect of THBDD1 in AAV-THBDD1 -treateddb/db mice was probably independent of APC. This finding isconfirmed by our recent observation, in a mouse model ofsepsis, that THBDD1 has APC-independent anti-inflammatoryactivity in vivo and inhibits neutrophil infiltration into thekidney [15]. Collectively, ours and several lines of research havesuggested possible mechanisms for the direct anti-inflammatoryeffects of THBDD1, including: (1) interference with the bindingof lipopolysaccharide to its receptor and inhibition ofinflammatory mediator production by macrophages [15];(2) suppression of adhesionmolecule expression by neutrophilsacting via the NF-κB andMAPK pathways [24]; (3) preventionof leucocyte activation by sequestration of HMGB1 protein[25]; and (4) interference with complement activation (C3aand C5a) via the classical and lectin pathways [26]. In addition,our previous study showed that anti-inflammatory effects cansignificantly ameliorate DN in db/db mice [7] by blocking the

NF-κB-mediated inflammatory pathway. We did not identifythe sequence of events in vivo, which include reduced inflam-mation, reduced oxidative stress, increased antioxidant capacityand reduced apoptosis as the possible basis for the protectiveeffects of THBDD1. Nevertheless, reduced inflammationmightwell be the key mechanism involved. Very recently, Wang et alalso demonstrated that THBD controls DN through its lectin-like domain by regulating complement independently of APC[27]. In this regard, our data show that there is no significantdifference in C3 deposits between vector-alone control db/dbmice and AAV-THBDD1 -treated db/db mice. It is likelythat THBDD1 achieved its preventive effects in the modelof DN without improving glomerular localisation of C3 inthe treated mice.

NLRP3 inflammasome has been shown to play a role in thepathogenesis of type 2 diabetes and DN in humans [28, 29],and in a model of type 2 diabetes [30]. Our data suggest thatmodulation of the NLRP3 inflammasome pathway by AAV-THBDD1 administration helps prevent renal inflammationand fibrosis in our mouse model of DN. Additional evidencesupporting the anti-inflammatory effects of THBDD1 comesfrom the decreased HMGB1 protein levels in HG medium-cultured podocytes that were treated with rTHBDD1 . Thisobservation is consistent with findings that HMGB1 and recep-tors for AGE are blocked by THBDD1, thereby inhibiting

a NG (5 mmol/l)104

104

103

103

102

102

101

101100

100

104

104

103

103

102

102

101

101100

100

104

104

103

103

102

102

101

PI

101100

100

8

NSNS

†

NSNS

†

†

†A

nnex

in V

+PI

+ (

%)

Cas

p3 p

17:β

-act

in r

atio

76543210

5_ _ 5 17 50

Glucose (mmol/l)rTHBDD1 (μg/ml)

30 30 30 30

5_ _ 5 17 50

Casp3kDa 1745

0.8

0.6

0.4

0.2

0.0

β-Actin

Glucose (mmol/l)rTHBDD1 (μg/ml)

30 30 30 30

5_ _ 5 17 50

Glucose (mmol/l)rTHBDD1 (μg/ml)

30 30 30 30

Annexin VHG (30 mmol/l) + rTHBDD1 (5 μg/ml) HG (30 mmol/l) + rTHBDD1 (17 μg/ml) HG (30 mmol/l) + rTHBDD1 (50 μg/ml)

104

104

103

103

102

102

101

101100

100

104

104

103

103

102

102

101

101100

100

HG (30 mmol/l)

b c

d

Fig. 7 Flow cytometry forapoptosis. (a) Flow cytometryapoptosis assay of annexin V andPI positivity (encircled area bycontinuous lines indicatesapoptotic population) with (b)annexin V- and PI-positivefindings in per cent. (c)Representative western blots forcaspase-3 p17 in culturedpodocytes and (d) semi-quantitative analysis of theactivated caspase-3 p17:β-actinratio. Cultured podocytes wereincubated in NG (5 mmol/l),HG (30 mmol/l) or HG plusrTHBDD1 (5, 17 or 50 μg/ml)medium. †p <0.005

432 Diabetologia (2014) 57:424–434

HMGB1-mediated inflammatory reactions [25]. Increased se-rum levels of THBD domains have been observed in patientswith diabetes [31–33] and some specific forms of THBDdomains were differentially expressed in patients with DNand non-diabetic nephritis [33]; however, the role of THBD inthese renal conditions was not determined. In the present study,the beneficial effects of THBDD1 on the db/db mice treatedsuggest that THBD may be protective to diabetes-associatedrenal damage, although further investigation on this point iswarranted.

Excessive ROS production has been shown to increase theexpression of extracellular matrix proteins involved in thedevelopment of renal fibrosis [18], and to play a major path-ogenic role in the development and complications of diabetes[8]. Blocking of cellular signalling pathways leading to ROSgeneration has been shown to blunt DN [34]. Interestingly,NRF2 has been shown to exert its beneficial effects on DN[29] and other renal conditions [18] by countering oxidativestress. Moreover, the data [18, 27] on renal HO-1 in AAV-THBDD1 -treated animals are not consistent with the entiretyof the other data in our study, where renal HO-1 levels weregreater than inWTmice and vector-alone control db/db mice.This issue needs to be addressed. It is established thatexpression levels or the activity of HO-1 and GPx areinvolved in the elimination or inactivation of ROS [35],and that this effect is mediated by activation of NRF2 [36].Although normal tissues express extremely low basallevels of HO-1 protein [37, 38], diabetic rats receivingantioxidant treatment have significantly increased HO-1protein levels compared with normal controls [39]. Together,our data suggest that the expression of HO-1 protein may be akey factor in preventing mice from developing renal lesionsin DN.

Podocyte loss has been implicated in the progression of DN[6, 7], and this has been confirmed in mouse models of type 2diabetes, which show apoptosis in various compartments ofthe kidney [6, 8]. Our in vitro data showed that rTHBDD1 hada significant inhibitory effect on podocyte apoptosis inducedin a HG environment and that AAV-THBDD1 gene deliveryinhibited apoptosis in renal tissues via a caspase-9 intrinsicpathway. However, the concentration of THBDD1 usedin vitro was relatively high, and the effects demonstratedcould therefore have been due to a charge effect alone of theTHBDD1 used. To address this issue, we performed anexperiment with HG-treated podocytes, in which BSA, whichbears approximately the same net charge as rTHBDD1 (itsnet charge is 4.9), was added to the cell culture in a dose-dependent and time-dependent manner. No detectableinhibitory effect on apoptosis of the cells was observed byflow cytometry with annexin V- and PI-positive staining (datanot shown).

THBD activates protein C and APC has an anticoagulanteffect, with the associated risk of bleeding [40]. No evidence

of systemic side effects, including bleeding, was seen in AAV-THBDD1 -treated db/db mice, suggesting that THBDD1 isrelatively safe for use in vivo for at least 18 weeks at the doseused in this study.

In the present study, albuminuria and histological changeswere approximately halved in treated mice, yet most otherendpoints addressing intermediates appeared to be normal-ised. The in vivo microenvironment is very complicated andchanges in each variable may vary according to differentanalytical techniques. The question of whether this effect, orother causes, can explain the observed discrepancy requiresfurther investigation.

Funding This study was supported by grants from the Department ofHealth, Executive Yuan (DOH97-TD-I-11-TM006), the Ministry ofEconomic Affairs (100-EC-17-A-19-S1-161) and the National ScienceCouncil (NSC 101-2321-B-016-003) of Taiwan, Republic of China.

Duality of interest The authors declare that there is no duality ofinterest associated with this manuscript.

Contribution statement SMYacquired and analysed data, and draftedthe manuscript. SMK performed the experiments using culturedpodocytes, analysed the data and edited the manuscript. HLW analyseddata and edited the manuscript. YCY engineered the AAV-THBDD1,analysed data and revised the manuscript. CHK generated the rTHBDD1 ,analysed data and revised the manuscript. KFH, GYS, YJH, FCH, SSY,YSS, SHL, CWW, YJH and CYY analysed the data and revised themanuscript. JSL designed, constructed and produced the AAV vectors,analysed data and revised the manuscript. AC designed the study andfinalised the manuscript. All authors approved the final version of themanuscript. SMK and AC are the guarantors of this work and, as such,had full access to all the data in the study and take responsibility for theintegrity of the data and the accuracy of the data analysis.

References

1. Booth GL, Kapral MK, Fung K, Tu JV (2006) Relation between ageand cardiovascular disease in men and women with diabetescompared with non-diabetic people: a population-based retro-spective cohort study. Lancet 368:29–36

2. Wolf G, Ritz E (2003) Diabetic nephropathy in type 2 diabetesprevention and patient management. J Am Soc Nephrol 14:1396–1405

3. Chow F, Ozols E, Nikolic-Paterson DJ, Atkins RC, Tesch GH (2004)Macrophages in mouse type 2 diabetic nephropathy: correlation withdiabetic state and progressive renal injury. Kidney Int 65:116–128

4. Stolar M (2010) Glycemic control and complications in type 2diabetes mellitus. Am J Med 123:S3–S11

5. Elmarakby AA, Sullivan JC (2012) Relationship between oxidativestress and inflammatory cytokines in diabetic nephropathy.Cardiovasc Ther 30:49–59

6. Isermann B, Vinnikov IA, Madhusudhan T et al (2007) Activatedprotein C protects against diabetic nephropathy by inhibiting endo-thelial and podocyte apoptosis. Nat Med 13:1349–1358

7. Ka SM, Yeh YC, Huang XR et al (2012) Kidney-targeting Smad7gene transfer inhibits renal TGF-beta/MAD homologue (SMAD) and

Diabetologia (2014) 57:424–434 433

nuclear factor kappaB (NF-kappaB) signalling pathways, and im-proves diabetic nephropathy in mice. Diabetologia 55:509–519

8. Brezniceanu ML, Liu F, Wei CC et al (2008) Attenuation ofinterstitial fibrosis and tubular apoptosis in db/db transgenic miceoverexpressing catalase in renal proximal tubular cells. Diabetes 57:451–459

9. Morser J (2012) Thrombomodulin links coagulation to inflammationand immunity. Curr Drug Targets 13:421–431

10. Ohsawa I, Ohi H, Fujita T, Kanmatsuse K (1996) Elevation of plasmathrombomodulin level in primary glomerulonephritis with heavyproteinuria. Nihon Jinzo Gakkai Shi 38:300–304

11. Ikeguchi H, Maruyama S, Morita Y et al (2002) Effects of humansoluble thrombomodulin on experimental glomerulonephritis.Kidney Int 61:490–501

12. Ozaki T, Anas C, Maruyama S et al (2008) Intrarenal administrationof recombinant human soluble thrombomodulin amelioratesischaemic acute renal failure. Nephrol Dial Transplant Off Publ EurDial Transplant Assoc Eur Ren Assoc 23:110–119

13. Li YH, Shi GY, Wu HL (2006) The role of thrombomodulin in athero-sclerosis: from bench to bedside. Cardiovasc Hematol AgentsMedChem4:183–187

14. Conway EM (2012) Thrombomodulin and its role in inflammation.Semin Immunopathol 34:107–125

15. Shi CS, Shi GY, Hsiao SM et al (2008) Lectin-like domain ofthrombomodulin binds to its specific ligand Lewis Y antigen andneutralizes lipopolysaccharide-induced inflammatory response.Blood 112:3661–3670

16. Kuo CH, Chen PK, Chang BI et al (2012) The recombinant lectin-likedomain of thrombomodulin inhibits angiogenesis through interactionwith Lewis Y antigen. Blood 119:1302–1313

17. Li YH, Chung HC, Luo CYet al (2010) Thrombomodulin is upreg-ulated in cardiomyocytes during cardiac hypertrophy and preventsthe progression of contractile dysfunction. J Card Fail 16:980–990

18. Tsai PY, Ka SM, Chang JM et al (2012) Antroquinonol differentiallymodulates T cell activity and reduces interleukin-18 production, butenhances Nrf2 activation, in murine accelerated severe lupus nephritis.Arthritis Rheum 64:232–242

19. Chao TK, Rifai A, Ka SM et al (2006) The endogenous immuneresponse modulates the course of IgA-immune complex mediatednephropathy. Kidney Int 70:283–297

20. Ohse T, Pippin JW, Vaughan MR, Brinkkoetter PT, Krofft RD,Shankland SJ (2008) Establishment of conditionally immortalizedmouseglomerular parietal epithelial cells in culture. J Am Soc Nephrol 19:1879–1890

21. Justo P, Sanz AB, Egido J, Ortiz A (2005) 3,4-Dideoxyglucosone-3-ene induces apoptosis in renal tubular epithelial cells. Diabetes 54:2424–2429

22. Susztak K, Raff AC, Schiffer M, Bottinger EP (2006) Glucose-induced reactive oxygen species cause apoptosis of podocytes andpodocyte depletion at the onset of diabetic nephropathy. Diabetes 55:225–233

23. Heilbronn R, Weger S (2010) Viral vectors for gene transfer: currentstatus of gene therapeutics. Handb Exp Pharmacol 197:143–170

24. Conway EM, Van de Wouwer M, Pollefeyt S et al (2002) Thelectin-like domain of thrombomodulin confers protection fromneutrophil-mediated tissue damage by suppressing adhesion moleculeexpression via nuclear factor kappaB and mitogen-activated proteinkinase pathways. J Exp Med 196:565–577

25. Li YH,Kuo CH, Shi GY,WuHL (2012) The role of thrombomodulinlectin-like domain in inflammation. J Biomed Sci 19:34

26. Van de Wouwer M, Plaisance S, De Vriese A et al (2006) The lectin-like domain of thrombomodulin interferes with complement activa-tion and protects against arthritis. J Thromb Haemost 4:1813–1824

27. Wang H, Vinnikov I, Shahzad K et al (2012) The lectin-like domainof thrombomodulin ameliorates diabetic glomerulopathy via comple-ment inhibition. Thromb Haemost 108:1141–1153

28. Donath MY, Shoelson SE (2011) Type 2 diabetes as an inflammatorydisease. Nat Rev Immunol 11:98–107

29. Vandanmagsar B, Youm YH, Ravussin A et al (2011) The NLRP3inflammasome instigates obesity-induced inflammation and insulinresistance. Nat Med 17:179–188

30. Masters SL, Dunne A, Subramanian SL et al (2010) Activation of theNLRP3 inflammasome by islet amyloid polypeptide provides amechanism for enhanced IL-1beta in type 2 diabetes. Nat Immunol11:897–904

31. Rustom R, Leggat H, Tomura HR, Hay CR, Bone JM (1998) Plasmathrombomodulin in renal disease: effects of renal function andproteinuria. Clin Nephrol 50:337–341

32. Aso Y, Fujiwara Y, Tayama K, Takebayashi K, Inukai T,Takemura Y (2000) Relationship between soluble thrombomodulinin plasma and coagulation or fibrinolysis in type 2 diabetes.Clin Chim Acta Int J Clin Chem 301:135–145

33. Uehara S, Gotoh K, Handa H (2001) Separation and characterizationof the molecular species of thrombomodulin in the plasma of diabeticpatients. Thromb Res 104:325–332

34. Kashihara N, Haruna Y, Kondeti VK, Kanwar YS (2010) Oxidativestress in diabetic nephropathy. Curr Med Chem 17:4256–4269

35. Na HK, Surh YJ (2008) Modulation of Nrf2-mediated antioxidant anddetoxifying enzyme induction by the green tea polyphenol EGCG.Food Chem Toxicol Int J Publ Br Ind Biol Res Assoc 46:1271–1278

36. Tsai PY, Ka SM, Chang JM et al (2011) Epigallocatechin-3-gallateprevents lupus nephritis development in mice via enhancing the Nrf2antioxidant pathway and inhibiting NLRP3 inflammasome activation.Free Radic Biol Med 51:744–754

37. da Silva JL, ZandBA,Yang LM, SabaawyHE, Lianos E, AbrahamNG(2001) Heme oxygenase isoform-specific expression and distribution inthe rat kidney. Kidney Int 59:1448–1457

38. Hung SY, Liou HC, Kang KH, Wu RM, Wen CC, Fu WM(2008) Overexpression of heme oxygenase-1 protects dopaminergicneurons against 1-methyl-4-phenylpyridinium-induced neurotoxicity.Mol Pharmacol 74:1564–1575

39. Ptilovanciv EO, Fernandes GS, Teixeira LC et al (2013) Hemeoxygenase 1 improves glucoses metabolism and kidney histologicalalterations in diabetic rats. Diabetol Metab Syndr 5:3

40. Gilbert RE, Marsden PA (2008) Activated protein C and diabeticnephropathy. N Engl J Med 358:1628–1630

434 Diabetologia (2014) 57:424–434