Page 1 of 49

Beta cell-derived angiopoietin-1 regulates secretion and glucose

homeostasis by stabilizing islet microenvironment

- Running title: Ang-1 contributes to glucose homeostasis -

Ho Seon Park,1,2,3 Hak Zoo Kim,2,3 Jong Suk Park,1,2,3 Junyeop Lee,4 Seung-Pyo Lee,5 Hail Kim,6

Chul Woo Ahn,1,2,3 Yoshikazu Nakaoka,7 Gou Young Koh,6,8 Shinae Kang1,2,3

1Department of Internal Medicine, 2Gangnam Severance Hospital, 3Severance Institute for

Vascular and Metabolic Research, Yonsei University College of Medicine, Yonsei University,

Seoul, South Korea; 4Department of Ophthalmology, Yeungnam University College of Medicine,

Daegu, South Korea; 5Department of Internal Medicine, Seoul National University Hospital, Seoul,

South Korea; 6Graduate School of Medical Science and Engineering, Korea Advanced Institute of

Science and Technology, Daejon, South Korea; 7Department of Vascular Physiology, National

Cerebral and Cardiovascular Center Research Institute, Osaka, Japan 8Center for Vascular

Research, Institute for Basic Science, Daejon, South Korea.

Corresponding author:

Gangnam Severance Hospital, 211 Eonju-ro, Gangnam-gu, Seoul, 135-710 Korea

Phone: +82-2-2019-3335

Fax: +82-2-3463-3882

E-mail: [email protected]

For Peer Review Only Diabetes Publish Ahead of Print, published online February 6, 2019 Diabetes Page 2 of 49

Word count: 3999

Number of tables: 0

Number of figures: 7

Number of supplementary tables: 1

Number of supplementary figures: 9

For Peer Review Only Page 3 of 49 Diabetes

Abstract

Islets are highly vascularized for prompt insulin secretion. Although angiopoietin-1 (Ang1) is a

well-known angiogenic factor, its role in glucose homeostasis remains largely unknown. The

objective of this study was to investigate whether and how Ang1 contributes to glucose

homeostasis in response to metabolic challenge. We used inducible systemic Ang1-knockout

(Ang1sys-/-) and β-cell specific Ang1-knockout (Ang1β-cell-/-) mice fed a high-fat diet for 24 weeks.

Although the degree of insulin sensitivity did not differ between Ang1sys-/- versus Ang1sys+/+ mice,

serum insulin levels were lower in Ang1sys-/- mice, resulting in significant glucose intolerance.

Similar results were observed in Ang1β-cell-/- mice, suggesting a critical role of β-cell-derived Ang1

in glucose homeostasis. There were no differences in β-cell area nor vasculature density, but

glucose-stimulated insulin secretion was significantly decreased and PDX-1 expression and

GLUT2 localization were altered in Ang1β-cell-/- compared to Ang1β-cell+/+ mice. These effects were

associated with less pericyte-coverage, disorganized endothelial cell ultrastructure, and also,

enhanced infiltration of inflammatory cells and upregulation of adhesion molecules in the islets of

Ang1β-cell-/- mice. In conclusion, β-cell-derived Ang1 regulates insulin secretion and glucose

homeostasis by stabilizing the blood vessels in the islet and may be a novel therapeutic target for

diabetes treatment in the future.

For Peer Review Only Diabetes Page 4 of 49

Introduction

The islet is composed of endocrine cells and its microenvironment which contains several types of cells such as endothelial and immune cells (1). During development, premature endocrine cells produce several angiogenic factors that attract endothelial cells into the premature islet cluster.

The recruited endothelial cells induce further β-cell differentiation and maturation (2, 3), suggesting the importance of close cross-talk between endocrine cells and their microenvironment

(1, 4).

Several studies have reported that vascular endothelial (VEGF)-A, a well- known angiogenic factor mainly produced from islet cells (3), acts as a crucial factor in islet development by orchestrating the communication between islet cells and nearby endothelial cells

(5-7). Another potent angiogenic factor, angiopoietin-1 (Ang1), is mainly produced from β-cells

(3). Ang1 exerts its signal via the Tie2 receptor on endothelial cells and mainly contributes to vascular maturity, stability, and integrity (8, 9). Systemic Ang1 (8) or Tie2 (10, 11) mutant/null mice are embryonically lethal because of defects in systemic vascular development. More specifically, the vessels display loose connective tissue and are barely covered with pericytes, indicating that Ang1 plays a role in connecting the blood vessels with the surrounding microenvironment (8, 9, 12).

With the development of time- and/or tissue-specific Ang1 deletion rodent models, Ang1 was predicted to be dispensable in the adult phase, at least under quiescent conditions, but the role of Ang1 is critical in adults when specific perturbations occur (13). Such context-dependent action of Ang1 makes it difficult to determine how Ang1 contributes to the function of a specific tissue/organ and when or under what circumstances Ang1 exerts its action. Although Ang1 has

For Peer Review Only Page 5 of 49 Diabetes

been demonstrated to enhance the efficacy of islet transplantation in diabetes research (14, 15), the

Tet-on inducible systemic Ang1 deletion or β-cell specific Ang1 overexpression showed no

dramatic metabolic changes under quiescent conditions (5).

However, because β-cells express relatively higher Ang1 compared to the surrounding

tissues in the adult stage (3) and precise communication between endocrine cells and vessels is

indispensable for β-cell function (4, 16), we hypothesized that Ang1 from the islets contributes

significantly to glucose homeostasis under metabolically stressed conditions such as high-fat diet

(HFD). The objective of the current study was to investigate the role of islet-derived Ang1 using

inducible systemic Ang1-knockout (Ang1sys-/-) and β-cell specific Ang1-knockout (Ang1β-cell-/-)

mice after 24 weeks of HFD, a rodent model that mimics obesity-induced type 2 diabetes mellitus

in humans.

For Peer Review Only Diabetes Page 6 of 49

Research Design and Methods

Animals

Ang1flox/flox mice (Acc.No.CDB0627K, RBRC09330,

www2.clst.riken.jp/arg/mutant%20mice%20list.html) were kindly provided by Dr. G.Y. Koh

which were originally developed and supplied by Dr. Y. Nakaoka and Riken BioResource Center

through the National Bio-Resource Project of the MEXT, Japan. The mice were generated by

insertion of the loxP allele into the two introns flanking exon1 of Ang1 (17, 18). Ang1flox/flox mice were crossed with Rosa26-CreERT2 and maintained as Rosa26-CreERT2;Ang1flox/wt in a C57BL/6N

genetic background to study inducible systemic Ang1 deletion. Rip-Cre (B6.Cg-Tg(Ins2-

cre)25Mgn/J, Jackson Laboratory, Bar Harbor, ME, USA) mice were purchased and crossed with

Ang1flox/flox mice to generate a Rip-Cre;Ang1flox/wt mice in a C57BL/6N genetic background. The

Rip-Cre;Ang1flox/wt mice were crossed to generate either the Rip-Cre;Ang1flox/flox (Ang1β-cell-/-) as

β-cell specific Ang1-knockout and Rip-Cre;Ang1wt/wt (Ang1β-cell+/+) as corresponding control mice

respectively throughout the entire experiments. To generate inducible systemic Ang1-knockout

mice (Ang1sys-/-) and their control comparator (Ang1sys+/+), Rosa26-CreERT2;Ang1flox/flox and their control Rosa26-CreERT2;Ang1wt/wt mice were intraperitoneally injected with 3mg of tamoxifen

(Sigma Aldrich, St. Louis, MO, USA) every other day for 6 days at 7 weeks of age. The mice were

fed with a standard normal diet (ND) or HFD with 60% of the total calories from fat (Research

Diets Inc., New Brunswick, NJ, USA), from 8 weeks of age for Ang1sys-/- and Ang1sys+/+ mice or

from 5 weeks of age for Ang1β-cell-/- and Ang1β-cell+/+ mice for a total 24 weeks. All animals were

maintained in a specific pathogen-free animal facility with 12-h light and dark cycles at the

Gangnam Biomedical Research Institute of Yonsei University College of Medicine with access to

For Peer Review Only Page 7 of 49 Diabetes

food and water ad libitum. All animal experiments were approved by the Yonsei University Health

System Institutional Animal Care and Use Committee (YUHS-IACUC).

Metabolic phenotyping

To evaluate glucose intolerance or insulin sensitivity, an intraperitoneal glucose tolerance

or insulin tolerance test was performed. Briefly, after 16 or 8-h of fasting, D-glucose (1.5g/kg for

Ang1sys+/+ and Ang1sys-/-, 0.5g/kg for Ang1β-cell+/+ and Ang1β-cell-/-) or insulin (1.5U/kg) was injected

into the abdominal cavity and then whole blood was collected from the tail vein. Blood glucose

level was measured using the ACCU-CHEK Performa glucometer (Roche Diagnostics Operations,

Inc., Basel, Switzerland) and serum insulin level with the Insulin ELISA (ALPCO, Salem, NH,

USA).

In vitro assay of glucose homeostasis

Mouse islets were isolated as previously described (19). For glucose-stimulated insulin

secretion (GSIS) analysis, islets were cultured overnight in RPMI-1640 (Thermo Fisher Scientific,

Waltham, MA, USA) with 10% fetal bovine serum (Thermo Fisher Scientific) and 30 evenly sized

islets were picked per group. These islets were incubated for 1-h with no glucose in Krebs Ringer

Bicarbonate Buffer (KRBB) and then stimulated with 5.6, 16.7, or 20 mM of glucose each for 1-

h. The supernatant from each incubated buffer was collected and the amount of secreted insulin

was measured. The same set of islets was used to measure the amount of total insulin contents in

the islets. Briefly, the islets were incubated for 24-h in 75% of acidic-ethanol containing 0.2M HCl

at 4℃ and the supernatant processed for analysis. The degree of GSIS was presented as % release

calculated by dividing the amount of secreted insulin in the KRBB by the total insulin content in

the same batch of islets.

For Peer Review Only Diabetes Page 8 of 49

To measure total insulin content per pancreas, the entire pancreas was dissected, weight measured, and then homogenized completely in 5mL of 0.2M acetic acid. The homogenate was transferred and boiled at 100℃ for 15-min. After cooling, the supernatant was taken and the insulin concentration was measured using ELISA. Total insulin content per pancreas was defined as the total amount of insulin per gram weight of the total pancreas.

For ex vivo islet culture experiment, high fat high sucrose diet (HFSD) with 58, 25, 17% calories from fat, carbohydrate, and (Research Diets) was fed for 4 weeks to 30 weeks-old

Ang1β-cell-/- and Ang1β-cell+/+ mice. For experiments using exogenous Ang1 supply, COMP-Ang1 protein (17, 20) was provided by Dr. G. Y. Koh.

Histologic analysis

Histological evaluation including functional vessel was performed according to previous report (19). Anti-insulin, anti-glucagon (Dako & Cell Signaling Technology Inc., Danvers, MA), anti-Ang1, anti-GLUT2 (Santa Cruz Biotechnology, Inc, Dallas, TX), anti-CD31, anti-CD45 (BD

Biosciences, San Diego, CA), anti-PDX-1 (Abcam, Cambridge, MA), anti-NG2, anti-collagen IV

(EMD Millipore Cooperation, Temecula, CA), anti-VEGFR2, anti-Tie2 (R&D Systems,

Minneapolis, MN), anti-PDGFRβ, anti-F4/80, anti-CD3e, and anti-B220 (eBioscience Life

Technologies , San Diego, CA) were used as primary Abs. 4',6-diamidino-2-phenylindole (DAPI)

(Thermo, Fisher Scientic) was used for nuclear staining. For whole pancreas imaging, the pancreas was scanned using a slide scanner (Leica SCN400F, Leica Microsystems GmbH, Wetzlar,

Germany) and the slide images were captured using the image viewer (SCN400image viewer version 2.2, Leica Microsystems CMS GmbH). The images were analyzed with the ImageJ program (version 1.49v, http://imagej.nih.gov/ij/, NIH, Bethesda, MD, USA). Electron microscopic image analysis was performed according to the previous report (21)

For Peer Review Only Page 9 of 49 Diabetes

Western blot

Western blot was performed as previously described (21) with anti-PDX-1, anti-GLUT2,

and anti-GAPDH (Sigma Aldrich) Abs.

Quantitative PCR

Total RNA was extracted using an RNA isolation kit (PicoPure, 12204-01, Thermo Fisher

Scientific). The cDNA was prepared (SuperScript III Reverse Transcriptase, 18080-044, Thermo

Fisher Scientific) and the amount of RNA analyzed by quantitative RT-PCR (StepOnePlus™ Real-

Time PCR system, Applied BiosystemsTM, Foster City, CA, USA) with the designated Taqman

primer and probes (Applied BiosystemsTM). See supplementary table for all primer information.

Statistical analysis

The statistical significance was tested with the independent t-test using SPSS 21.0 (SPSS

Inc., Chicago, IL, USA) and a p-value <0.05 was regarded as significant.

For Peer Review Only Diabetes Page 10 of 49

Results

Inducible systemic deletion of Ang1 results in glucose intolerance with lower serum insulin level

after 24 weeks of HFD

To evaluate the effect of Ang1 on glucose homeostasis in adulthood, we deleted Ang1 from

7 weeks of age and fed Ang1sys-/- and the Ang1sys+/+ mice a ND or HFD for 24 weeks after 8 weeks

of age (Fig. 1A). Deletion of Ang1 from the genomic DNA, the corresponding transcripts and

protein after tamoxifen injection was confirmed (Fig. 1B and C, Supplementary Fig. 1). After 12 weeks of HFD, there was no difference in body weight nor blood glucose levels after glucose challenge between Ang1sys+/+ and Ang1sys-/- (Fig. 1D and E). However, after feeding HFD for 24 weeks, Ang1sys-/- mice developed glucose intolerance compared to Ang1sys+/+ (Fig. 1G), without significant differences in body weight and insulin sensitivity (Fig. 1F and H). Therefore, we measured serum insulin levels to investigate the cause of glucose intolerance by Ang1 deletion.

Although there was no change in the serum insulin level between Ang1sys-/- and Ang1sys+/+ mice fed with ND, HFD-fed Ang1sys-/- mice showed lower serum insulin levels after glucose challenge compared to HFD-fed Ang1sys+/+ mice (Fig. 1I). These findings suggest that the impaired glucose clearance from the blood in the Ang1sys-/- mice results from defective insulin secretion from β-cell.

β-cell-derived Ang1 is critical for glucose homeostasis after 24 weeks of HFD

Next, we generated β-cell specific Ang1-knockout mice (Ang1β-cell-/-) and fed them with

ND or HFD for 24 weeks starting from 5 weeks of age (Fig. 2A). Deletion of Ang1 from the genomic DNA was confirmed in the isolated islets (Fig. 2B) and depletion of Ang1 protein was also validated by immunostaining of β-cells (Fig. 2C). After 12 weeks of HFD, Ang1β-cell-/- showed

For Peer Review Only Page 11 of 49 Diabetes

a trend towards glucose intolerance without any difference in body weight (Fig. 2D and E). After

24 weeks of HFD, although there was no differences in both body weight and insulin sensitivity

in the Ang1β-cell-/- mice compared to the Ang1β-cell+/+ mice (Fig. 2F and G), the HFD-fed Ang1β-

cell-/- mice developed significant glucose intolerance compared to the HFD-fed Ang1β-cell+/+ mice

(Fig. 2H and I). Similar to the Ang1sys-/- mice, the serum insulin level of HFD-fed Ang1β-cell-/- mice

was lower compared to the HFD-fed control mice after glucose challenge (Fig. 2J). These findings

suggest that β-cell-derived Ang1 is crucial for maintaining glucose homeostasis under

metabolically challenged conditions, possibly by controlling the blood level of insulin.

No morphological and compositional difference in endocrine cells and vascular density in islets

by Ang1 deletion

Next, we investigated the mechanism underlying the impaired insulin secretion of Ang1β-

cell-/- mice fed with HFD. Because glucose intolerance was observed in Ang1β-cell-/- mice fed with

HFD without any difference in insulin sensitivity, we considered several possibilities such as

decreased β-cell mass resulting in decreased insulin production, hampered systemic insulin

circulation by problems in the intra-islet vasculature, or depressed insulin secretion from β-cells.

First, we evaluated whether changes occurred in endocrine cell composition or mass. However,

there was no difference in insulin-positive nor glucagon-positive area per whole pancreas or per

islets between Ang1β-cell+/+ and Ang1β-cell-/- mice (Fig. 3A–E). Neither the total insulin content per

pancreas weight nor the whole pancreas weight differed (Fig. 3F, Supplementary Fig. 2),

suggesting that the lower serum insulin concentration in the Ang1β-cell-/- mice compared to the

Ang1β-cell+/+ mice was not because of any difference in β-cell mass nor insulin production. Because

Ang1 is a potent angiogenic factor (8), we next evaluated whether decreased vascular density in

For Peer Review Only Diabetes Page 12 of 49

the islets decreased the level of systemic insulin, even with appropriate production of insulin from

islets. However, there were no differences in the density of CD31-positive vessels (Fig. 3G and

H), the fluorescein isothiocyanate-lectin perfused functional vessel (Fig. 3G and I) nor the

VEGFR2-positive blood vessels (Supplementary Fig. 3), suggesting that the lower serum insulin

level after glucose challenge by Ang1 deletion is not because of the decrease of functional blood

vessels in the islets. These findings demonstrate that glucose intolerance in the Ang1β-cell-/- mice is

not a result of a defect in insulin production nor a change in intra-islet vascular density.

Impaired insulin secretion and defects in pancreatic and duodenal homeobox 1 (PDX-1) and

glucose transporter 2 (GLUT2) pathway by Ang1 deletion

Because there was no difference in endocrine mass nor vascular density in Ang1β-cell-/- mice

fed with HFD (Fig. 3), we next evaluated whether the decreased serum insulin level after glucose

challenge in vivo was because of a defect in insulin secretion from the islets. The islets were

isolated and the GSIS was tested under ex vivo culture. Although there was no difference in the

amount of insulin-2 transcript between the Ang1β-cell-/- and the Ang1β-cell+/+ mice (Fig. 4A), the

Ang1β-cell-/- islets in HFD showed significantly impaired insulin secretion for both low and high

glucose levels, suggesting that glucose intolerance in Ang1β-cell-/- mice occurred because of a defect

in insulin secretion (Fig. 4B). To further understand the mechanism related to impaired insulin

secretion in Ang1β-cell-/- islets, we evaluated various genes associated with β-cell maturation and function. The expression pattern of PDX-1, a major transcription factor involved in β-cell differentiation/maturation/GSIS (22, 23), was generally impaired in old Ang1β-cell+/+ mice

compared to young Ang1β-cell+/+ mice (Supplementary Fig 4). In Ang1β-cell-/- islets, the mRNA

(Supplementary Fig 5A) and protein (Fig. 4C and D) expression level of PDX-1, and the degree of

For Peer Review Only Page 13 of 49 Diabetes

nuclear localization (Fig. 4E) and the signal intensity of PDX-1 per islet (Fig. 4F) was significantly

decreased compared to Ang1β-cell+/+ with HFD. We next evaluated GLUT2, a molecule critical for

insulin secretion by sensing the change in serum glucose levels and known to be downregulated

under PDX-1-defective conditions (24-26). The degree of GLUT2 membrane localization (Fig. 4G)

and the signal intensity per islet (Fig. 4H) was significantly reduced in Ang1β-cell-/- mice, although

the mRNA and protein level did not differ (Supplementary Fig. 5B-E). Next, we tested the effect

of exogenous Ang1 (COMP-Ang1 protein, a well-known stable variant of Ang1 that mimics the

effect of Ang1 (20, 27) in the islets isolated from Ang1β-cell-/- and the Ang1β-cell+/+ mice fed with

HFSD. The decreased mRNA level of PDX-1 at day 0 recovered after 48-h even without COMP-

Ang1 treatment and the COMP-Ang1 treated islets showed higher PDX-1 level compared to the

COMP-Ang1 non-treated group (Supplementary Fig. 6A). The GLUT2 expression also increased

by COMP-Ang1 treatment (Supplementary Fig. 6B). Collectively, these results clearly

demonstrate that hyperglycemia in Ang1β-cell-/- mice occurred because of deterioration in the

insulin secretory function of the islets and that the PDX-1 and GLUT2 pathway is impaired when

β-cells are deprived of Ang1.

Loss of pericytes from the vessels in Ang1-deleted islets

Because Ang1 coordinates vessel maturation and integrity, we investigated whether there

were changes in the coverage of pericytes, an important cell supporting the (28).

Generally, most islet endothelial cells were surrounded by NG2-positive pericytes, but the

proportion of NG2-positive cells per CD31-positive vessel area in the Ang1β-cell-/- islets was

significantly reduced compared to that of the control mice fed HFD (Fig. 5A and B). This

morphologic data was corroborated by the transcript level analysis (Fig. 5C). The trend was similar

For Peer Review Only Diabetes Page 14 of 49

when pericytes were visualized by PDGFRβ immunostaining (Supplementary Fig. 7). Because pericytes are the key players in maintaining the vessel integrity, we examined whether the vascular ultrastructure was affected by Ang1 deletion, despite no change in the gross density of vessels. In electron microscopic examination, Ang1β-cell-/- mice fed with HFD showed a distinct pattern of disorganized fenestration and caveolae (Fig. 5D). These data demonstrate that β-cell-derived Ang1 is critical for maintaining normal vascular structures in the islets.

Higher degree of inflammation in Ang1-deleted islets

We also examined the inflammatory cells in islets, as these cells are important components of the islet microenvironment. The degree of inflammatory cell infiltration is generally increased by HFD (29, 30) and inflamed islets have defects in insulin secretion (29-32). Because Ang1 is known as an anti-inflammatory molecule under various conditions (9, 14, 15, 33), we evaluated the degree of CD45-positive cell infiltration into the islets in Ang1β-cell-/- mice with or without

HFD. Although 24 weeks of HFD itself did not increase the number of CD45-positive cells per islet in Ang1β-cell+/+ mice, Ang1 deletion together with 24 weeks of HFD induced significant infiltration of CD45-positive cells compared to their comparator mice (Fig. 6A and B). When we examined the subpopulation of the immune cells infiltrating the islets, the majority of the immune cells were F4/80+ suggesting that the are the major proportion of the CD45+ cells

(Supplementary Fig. 8A and B). Next, we investigated whether the impaired GSIS could be recovered when this insulitis phenotype was rescued. When the islets from 30 weeks old mice fed with HFSD were isolated, there were significantly higher number of CD45+ cells in the islets from the Ang1β-cell-/- compared to the Ang1β-cell+/+ at day 0, the difference of which almost disappeared after 7 days of culture (Supplementary Fig. 9A) (34). At day 0 of isolation, the Ang1β-cell-/- mice

For Peer Review Only Page 15 of 49 Diabetes

showed impaired insulin secretion compared to the Ang1β-cell+/+ mice at 20mM of glucose

stimulation. However, after 7 days of culture, there was no difference of GSIS between the Ang1β-

cell+/+ and the Ang1β-cell-/- both in the 5.6mM and the 20mM glucose stimulation. This means that

the impaired insulin secretory function in the Ang1β-cell-/- islets recover after 7 days of culture,

along with the disappearance of inflammatory cell infiltration (Supplementary Fig. 9B). The

inflamed islets were also associated with higher amounts of ICAM-1 and VCAM-1 production,

the main adhesion molecules that attract leukocytes from the blood vessel into the tissue (Fig. 6C

and D) (35). Furthermore, the expression of vitronectin, a matrix protein responsible for providing

a barrier to leukocyte infiltration and mediating Ang1 signaling via (12, 36-38), was

significantly reduced in Ang1β-cell-/- mice fed a HFD, a pattern that differed from that in the Ang1β-

cell+/+ mice (Fig. 6E). This suggests that leukocytes transmigrate more easily into the inflamed islets.

Collectively, these data suggest that β-cell-derived Ang1, in response to chronic metabolic

challenges such as HFD, helps suppress inflammation in the islets and maintain insulin secretory

function.

For Peer Review Only Diabetes Page 16 of 49

Discussion

In this study, we demonstrated that β-cell-derived Ang1 contributes to glucose homeostasis

by coordinating insulin secretion from islets. Mechanistically, Ang1 helps support the perivascular

ultrastructure of the intra-islet vessel and protect islets from inflammation, thereby enabling insulin

secretory machinery in the islets to function adequately after long-term HFD (Fig. 7). This is the

first study revealing that Ang1 can act as a crucial regulator of the interaction between endocrine

cells and its microenvironment such as blood vessels and inflammatory cells in the islets.

In our study, Ang1-deprived β-cells developed glucose intolerance and impaired GSIS after

24 weeks of HFD. In contrast to our initial hypothesis that the absence of Ang1 would cause severe

vascular/endocrine damage, Ang1-knockout mice did not present any gross abnormalities in intra-

islet vasculature nor islet endocrine cell morphology nor composition. This led us to investigate

the secretory function of insulin from the islets. The isolated islets of Ang1β-cell-/- mice showed

reduced insulin secretion with no change in the total amount of insulin production compared to

that of the control mice, indicating that Ang1 deletion compromises the secretory machinery. In

support of this, the degree of nuclear localization of PDX-1, a crucial factor for GSIS (22, 23), was

lower in Ang1β-cell-/- mice fed the HFD (Fig. 4E and F) and the cell surface translocation of GLUT2, another key factor in GSIS (24-26), was also decreased in Ang1β-cell-/- β-cells both under ND and

HFD (Fig. 4G and H). This clearly suggests that the lower insulin secretion in the Ang1β-cell-/- mice is associated with a defect in the glucose sensing/insulin secretory machinery in β-cells, albeit no effect on β-cell mass nor vasculature density.

The mechanism by which Ang-1 is related to the GSIS of β-cells was intriguing. Because

Ang1 signaling is known to contribute to vessel maturation, stability, and integrity mainly by

For Peer Review Only Page 17 of 49 Diabetes

orchestrating the interaction of endothelial cells with nearby cells or connective tissue (9, 12, 39),

we evaluated whether Ang-1 exerts its effect by changing the coverage of the vessels by pericytes,

the major vascular components supporting the vascular integrity (28). Although additional studies

are needed to examine the role of pericytes, most previous studies found that pericytes are crucial

players in guiding the proper differentiation and function of adjacent cells by communicating with

endothelial cells (28, 40, 41). In line with previous studies showing that Ang1 deletion mice

develop poor perivascular coverage of the vessels by pericytes (8, 42), the NG2-positive pericyte

covered area per CD31-positive area decreased significantly in Ang1β-cell-/- mice (Fig. 5A and B).

Considering the known role of pericytes in tissue fibrosis, inflammation (28, 41), and maintenance

of vascular ultrastrucure (40, 43), it appears that the depletion of pericytes would lead to impaired

β-cell maturation or function (43-46). Indeed, our findings in Ang1β-cell-/- mice demonstrated the

role of pericytes in maintaining glucose homeostasis, which stresses that a ‘healthy’ interaction

between the endothelial cells and the pericytes is an important cornerstone for maintenance of a

‘healthy’ islet function. Mechanistically, pericyte defect may hamper GSIS because of the defect

in Tcf7L2/BMP4 through PDX-1/GLUT2 pathway (46) or may lead to insufficient production of

the basement membrane protein, ultimately leading to β-cell dysfunction (47).

In addition to defects in pericytes and the ultrastructural breakdown of the intra-islet

vasculature, we observed significant inflammatory cell infiltration in the islets of Ang1β-cell-/- mice

(Fig. 6A and B). Ang1 was originally found to be a pro-survival and anti-inflammatory factor (33,

39), and overexpression of Ang1 in rodent islets showed anti-apoptotic and anti-inflammatory

effects (15). We also observed similar actions of Ang1, as Ang1 knockout developed more

infiltration of CD45-positive cells into the islets. Upregulation of adhesion molecules, such as

ICAM-1 and VCAM-1, in Ang1-knockout islet may attract leukocytes into the inflamed area, a

For Peer Review Only Diabetes Page 18 of 49

finding supported by previous studies of Ang1 deletion (33, 48). Vitronectin is a candidate molecule that mediates Ang1 signaling between cells and connective tissue and thus, maintains vascular integrity (36, 38). Thus, the increase in vitronectin may be a defense mechanism of the islets to overcome metabolic stress under the HFD (37, 48). In contrast, the impaired vitronectin pathway in Ang1β-cell-/- mice fed the HFD may accelerate the breakdown of the islet microenvironment by inflammation. A HFD itself or inflammation also appears to inhibit PDX-1 signaling (29, 30), GLUT2 membrane localization (24, 49), which was supported by our series of experiments as well (Fig. 4), finally leading to impaired GSIS in Ang1β-cell-/- mice.

Although it has been consistently shown that Ang1 governs vascular development resulting in embryonic lethality when defective from the developmental stage (8, 10, 42), Ang1 does not appear to play a crucial role in maintaining life, at least under the quiescent conditions of the postnatal stage (13). In our study, Ang1 depletion impaired GSIS after long-term HFD challenge in both β-cell-specific knockout from the developmental stage or systemic knockout from the adult stage. Thus, the action of Ang1 may be context-dependent, such as development or severe defect by systemic deletion, together with destabilized/inflamed/injured conditions by long-term HFD.

This may explain why VEGF-A-defective mice developed glucose intolerance only when VEGF-

A was deleted in the developmental stage (50). The effect of Ang1 on vasculature formation in the islets may not be as strong as VEGF-A. β-cell-specific Ang1-knockout adult mice in our study showed no significant changes in vascular density nor endocrine mass. However, Ang1 is a very delicate and elaborate factor that orchestrates the vascular maturity, integrity, and stability in terms of vascular health, particularly under stressed conditions. Under conditions of balance breakdown by chronic metabolic stress, Ang1 may help maintain intra-islet vascular health, a role that was not evident in quiescent state of metabolism.

For Peer Review Only Page 19 of 49 Diabetes

In conclusion, a defect in β-cell-derived Ang1, inducing islet inflammation and disrupting

the normal intra-islet vascular ultrastructure, impairs the insulin secretory mechanism in response

to metabolic challenge. This is the first report demonstrating that Ang1 governs the

microenvironment of the islets under chronic stress conditions, which may contribute to adequate

glucose control. More detailed studies of the downstream target of Ang1 in the islets is warranted

and may lead to the discovery of novel therapeutic targets for diabetes in the future.

For Peer Review Only Diabetes Page 20 of 49

Acknowledgments

We appreciate the wonderful help from Professor Bum Jin Lim on analyzing the electron microscopy image.

Author Contributions

H.S.P., H.Z.K., S.K., designed the study. H.S.P, H.Z.K. S.K. conducted the research. H.S.P.,

H.Z.K., S.K. analyzed data. H.S.P., S.K. wrote the manuscript. J.L., S.-P.L., H. K., G.Y.K. gave constructive comments regarding the study concept. J.S.P., C.W.A., S.-P.L., H. K., Y.N., G.Y.K. reviewed and edited the manuscript.

Guarantor Statement

S.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Conflict of Interest

No potential conflicts of interest relevant to this article were reported

Funding

This research was supported by the Basic Science Research Program through the National

Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and

Technology (A1061486) and by the Bio & Medical Technology Development Program of the NRF funded by the Korean government, MSIP (NRF-2013M3A9D5072550).

For Peer Review Only Page 21 of 49 Diabetes

Prior Presentation

Parts of this study were presented in an abstract form at the 77th Scientific Sessions of the

American Diabetes Association, San Diego, CA, 9–13 June 2017 and at the 2015 International

Conference on Diabetes and Metabolism of the Korean Diabetes Association, Seoul, Korea, 15–

17 October 2015.

Data and Resource Availability

The data and resource generated during and/or analyzed during the current study are available from

the corresponding author on reasonable request.

For Peer Review Only Diabetes Page 22 of 49

References

`1. Peiris H, Bonder CS, Coates PT, Keating DJ, Jessup CF: The beta-cell/EC axis: how do islet cells talk to each other? Diabetes 2014;63:3-11 2. Lammert E, Cleaver O, Melton D: Induction of pancreatic differentiation by signals from blood vessels. Science 2001;294:564-567 3. Brissova M, Shostak A, Shiota M, Wiebe PO, Poffenberger G, Kantz J, Chen Z, Carr C, Jerome WG, Chen J, Baldwin HS, Nicholson W, Bader DM, Jetton T, Gannon M, Powers AC: Pancreatic islet production of vascular endothelial growth factor--a is essential for islet vascularization, revascularization, and function. Diabetes 2006;55:2974-2985 4. Eberhard D, Kragl M, Lammert E: 'Giving and taking': endothelial and beta-cells in the islets of Langerhans. Trends Endocrinol Metab 2010;21:457-463 5. Cai Q, Brissova M, Reinert RB, Pan FC, Brahmachary P, Jeansson M, Shostak A, Radhika A, Poffenberger G, Quaggin SE, Jerome WG, Dumont DJ, Powers AC: Enhanced expression of VEGF-A in beta cells increases endothelial cell number but impairs islet morphogenesis and beta cell proliferation. Dev Biol 2012;367:40- 54 6. Lammert E, Gu G, McLaughlin M, Brown D, Brekken R, Murtaugh LC, Gerber HP, Ferrara N, Melton DA: Role of VEGF-A in vascularization of pancreatic islets. Curr Biol 2003;13:1070-1074 7. Brissova M, Aamodt K, Brahmachary P, Prasad N, Hong JY, Dai C, Mellati M, Shostak A, Poffenberger G, Aramandla R, Levy SE, Powers AC: Islet microenvironment, modulated by vascular endothelial growth factor-A signaling, promotes beta cell regeneration. Cell Metab 2014;19:498-511 8. Suri C, Jones PF, Patan S, Bartunkova S, Maisonpierre PC, Davis S, Sato TN, Yancopoulos GD: Requisite role of angiopoietin-1, a ligand for the TIE2 receptor, during embryonic . Cell 1996;87:1171- 1180 9. Augustin HG, Koh GY, Thurston G, Alitalo K: Control of vascular morphogenesis and homeostasis through the angiopoietin-Tie system. Nat Rev Mol Cell Biol 2009;10:165-177 10. Dumont DJ, Gradwohl G, Fong GH, Puri MC, Gertsenstein M, Auerbach A, Breitman ML: Dominant- negative and targeted null mutations in the endothelial , tek, reveal a critical role in vasculogenesis of the embryo. Genes Dev 1994;8:1897-1909 11. Sato TN, Tozawa Y, Deutsch U, Wolburg-Buchholz K, Fujiwara Y, Gendron-Maguire M, Gridley T, Wolburg H, Risau W, Qin Y: Distinct roles of the receptor tyrosine kinases Tie-1 and Tie-2 in blood vessel formation. Nature 1995;376:70-74 12. Fukuhara S, Sako K, Minami T, Noda K, Kim HZ, Kodama T, Shibuya M, Takakura N, Koh GY, Mochizuki N: Differential function of Tie2 at cell-cell contacts and cell-substratum contacts regulated by angiopoietin-1. Nat Cell Biol 2008;10:513-526 13. Jeansson M, Gawlik A, Anderson G, Li C, Kerjaschki D, Henkelman M, Quaggin SE: Angiopoietin-1 is essential in mouse vasculature during development and in response to injury. J Clin Invest 2011;121:2278- 2289 14. Su D, Zhang N, He J, Qu S, Slusher S, Bottino R, Bertera S, Bromberg J, Dong HH: Angiopoietin-1 production in islets improves islet engraftment and protects islets from -induced apoptosis. Diabetes 2007;56:2274-2283 15. Wang D, Jiang L, Liang Y, Hao X, Chen C, Xia W, Zhuang R, Su Y, Guo S: COMP-Ang1 promotes long- term survival of allogeneic islet grafts in a bioinert perforated chamber by inhibiting inflammation via inhibition of the TLR4 signaling pathway. Biotechnol Lett 2016;38:1033-1042 16. Hogan MF, Hull RL: The islet endothelial cell: a novel contributor to beta cell secretory dysfunction in diabetes. Diabetologia 2017;60:952-959

For Peer Review Only Page 23 of 49 Diabetes

17. Lee J, Kim KE, Choi DK, Jang JY, Jung JJ, Kiyonari H, Shioi G, Chang W, Suda T, Mochizuki N, Nakaoka Y, Komuro I, Yoo OJ, Koh GY: Angiopoietin-1 guides directional angiogenesis through alphavbeta5 signaling for recovery of ischemic retinopathy. Sci Transl Med 2013;5:203ra127 18. Arita Y, Nakaoka Y, Matsunaga T, Kidoya H, Yamamizu K, Arima Y, Kataoka-Hashimoto T, Ikeoka K, Yasui T, Masaki T, Yamamoto K, Higuchi K, Park JS, Shirai M, Nishiyama K, Yamagishi H, Otsu K, Kurihara H, Minami T, Yamauchi-Takihara K, Koh GY, Mochizuki N, Takakura N, Sakata Y, Yamashita JK, Komuro I: Myocardium-derived angiopoietin-1 is essential for coronary vein formation in the developing heart. Nat Commun 2014;5:4552 19. Kang S, Park HS, Jo A, Hong SH, Lee HN, Lee YY, Park JS, Jung HS, Chung SS, Park KS: Endothelial progenitor cell cotransplantation enhances islet engraftment by rapid revascularization. Diabetes 2012;61:866-876 20. Cho CH, Kammerer RA, Lee HJ, Steinmetz MO, Ryu YS, Lee SH, Yasunaga K, Kim KT, Kim I, Choi HH, Kim W, Kim SH, Park SK, Lee GM, Koh GY: COMP-Ang1: a designed angiopoietin-1 variant with nonleaky angiogenic activity. Proc Natl Acad Sci U S A 2004;101:5547-5552 21. Choi JW, Jo A, Kim M, Park HS, Chung SS, Kang S, Park KS: BNIP3 is essential for mitochondrial bioenergetics during adipocyte remodelling in mice. Diabetologia 2016;59:571-581 22. Ahlgren U, Jonsson J, Jonsson L, Simu K, Edlund H: beta-cell-specific inactivation of the mouse Ipf1/Pdx1 gene results in loss of the beta-cell phenotype and maturity onset diabetes. Genes Dev 1998;12:1763-1768 23. Brissova M, Shiota M, Nicholson WE, Gannon M, Knobel SM, Piston DW, Wright CV, Powers AC: Reduction in pancreatic transcription factor PDX-1 impairs glucose-stimulated insulin secretion. J Biol Chem 2002;277:11225-11232 24. Ohtsubo K, Takamatsu S, Minowa MT, Yoshida A, Takeuchi M, Marth JD: Dietary and genetic control of glucose transporter 2 glycosylation promotes insulin secretion in suppressing diabetes. Cell 2005;123:1307-1321 25. Guillam MT, Dupraz P, Thorens B: Glucose uptake, utilization, and signaling in GLUT2-null islets. Diabetes 2000;49:1485-1491 26. Thorens B: GLUT2, glucose sensing and glucose homeostasis. Diabetologia 2015;58:221-232 27. Koh GY: Orchestral actions of angiopoietin-1 in vascular regeneration. Trends Mol Med 2013;19:31-39 28. Armulik A, Genove G, Betsholtz C: Pericytes: developmental, physiological, and pathological perspectives, problems, and promises. Dev Cell 2011;21:193-215 29. Donath MY, Storling J, Maedler K, Mandrup-Poulsen T: Inflammatory mediators and islet beta-cell failure: a link between type 1 and type 2 diabetes. J Mol Med (Berl) 2003;81:455-470 30. Eguchi K, Nagai R: Islet inflammation in type 2 diabetes and physiology. J Clin Invest 2017;127:14-23 31. McDaniel ML, Kwon G, Hill JR, Marshall CA, Corbett JA: and nitric oxide in islet inflammation and diabetes. Proc Soc Exp Biol Med 1996;211:24-32 32. Donath MY, Schumann DM, Faulenbach M, Ellingsgaard H, Perren A, Ehses JA: Islet inflammation in type 2 diabetes: from metabolic stress to therapy. Diabetes Care 2008;31 Suppl 2:S161-164 33. Kim I, Moon SO, Park SK, Chae SW, Koh GY: Angiopoietin-1 reduces VEGF-stimulated leukocyte adhesion to endothelial cells by reducing ICAM-1, VCAM-1, and E-selectin expression. Circ Res 2001;89:477-479 34. Strandell E, Eizirik DL, Sandler S: Reversal of beta-cell suppression in vitro in pancreatic islets isolated from nonobese diabetic mice during the phase preceding insulin-dependent diabetes mellitus. J Clin Invest 1990;85:1944-1950 35. Albelda SM, Buck CA: Integrins and other cell adhesion molecules. FASEB J 1990;4:2868-2880 36. Carlson TR, Feng Y, Maisonpierre PC, Mrksich M, Morla AO: Direct cell adhesion to the angiopoietins mediated by integrins. J Biol Chem 2001;276:26516-26525

For Peer Review Only Diabetes Page 24 of 49

37. Korpos E, Kadri N, Kappelhoff R, Wegner J, Overall CM, Weber E, Holmberg D, Cardell S, Sorokin L: The peri-islet basement membrane, a barrier to infiltrating leukocytes in type 1 diabetes in mouse and human. Diabetes 2013;62:531-542 38. Leavesley DI, Kashyap AS, Croll T, Sivaramakrishnan M, Shokoohmand A, Hollier BG, Upton Z: Vitronectin--master controller or micromanager? IUBMB Life 2013;65:807-818 39. Brindle NP, Saharinen P, Alitalo K: Signaling and functions of angiopoietin-1 in vascular protection. Circ Res 2006;98:1014-1023 40. Hellstrom M, Gerhardt H, Kalen M, Li X, Eriksson U, Wolburg H, Betsholtz C: Lack of pericytes leads to endothelial hyperplasia and abnormal vascular morphogenesis. J Cell Biol 2001;153:543-553 41. Chatterjee S, Naik UP: Pericyte-endothelial cell interaction: a survival mechanism for the tumor vasculature. Cell Adh Migr 2012;6:157-159 42. Vikkula M, Boon LM, Carraway KL, 3rd, Calvert JT, Diamonti AJ, Goumnerov B, Pasyk KA, Marchuk DA, Warman ML, Cantley LC, Mulliken JB, Olsen BR: Vascular dysmorphogenesis caused by an activating mutation in the receptor tyrosine kinase TIE2. Cell 1996;87:1181-1190 43. Hayden MR, Karuparthi PR, Habibi J, Lastra G, Patel K, Wasekar C, Manrique CM, Ozerdem U, Stas S, Sowers JR: Ultrastructure of islet microcirculation, pericytes and the islet exocrine interface in the HIP rat model of diabetes. Exp Biol Med (Maywood) 2008;233:1109-1123 44. Nakamura M, Kitamura H, Konishi S, Nishimura M, Ono J, Ina K, Shimada T, Takaki R: The endocrine pancreas of spontaneously diabetic db/db mice: microangiopathy as revealed by transmission electron microscopy. Diabetes Res Clin Pract 1995;30:89-100 45. Sasson A, Rachi E, Sakhneny L, Baer D, Lisnyansky M, Epshtein A, Landsman L: Islet Pericytes Are Required for beta-Cell Maturity. Diabetes 2016;65:3008-3014 46. Sakhneny L, Rachi E, Epshtein A, Guez HC, Wald-Altman S, Lisnyansky M, Khalifa-Malka L, Hazan A, Baer D, Priel A, Weil M, Landsman L: Pancreatic Pericytes Support beta-Cell Function in a Tcf7l2- Dependent Manner. Diabetes 2018;67:437-447 47. Nikolova G, Jabs N, Konstantinova I, Domogatskaya A, Tryggvason K, Sorokin L, Fassler R, Gu G, Gerber HP, Ferrara N, Melton DA, Lammert E: The vascular basement membrane: a niche for insulin gene expression and Beta cell proliferation. Dev Cell 2006;10:397-405 48. Imai Y, Patel HR, Doliba NM, Matschinsky FM, Tobias JW, Ahima RS: Analysis of gene expression in pancreatic islets from diet-induced obese mice. Physiol Genomics 2008;36:43-51 49. Wu Y, Wu T, Wu J, Zhao L, Li Q, Varghese Z, Moorhead JF, Powis SH, Chen Y, Ruan XZ: Chronic inflammation exacerbates glucose metabolism disorders in C57BL/6J mice fed with high-fat diet. J Endocrinol 2013;219:195-204 50. Reinert RB, Brissova M, Shostak A, Pan FC, Poffenberger G, Cai Q, Hundemer GL, Kantz J, Thompson CS, Dai C, McGuinness OP, Powers AC: Vascular endothelial growth factor-a and islet vascularization are necessary in developing, but not adult, pancreatic islets. Diabetes 2013;62:4154-4164

For Peer Review Only Page 25 of 49 Diabetes

Figure Legends

Figure 1. Inducible systemic deletion of Ang1 results in glucose intolerance with low serum

insulin level after high fat diet.

A: Schematic diagram of the animal experiment for inducible systemic Ang1 deletion. Conditional

knockout of Ang1 was induced with tamoxifen at 7 weeks of age and the mice were fed with

normal diet (ND) or high fat diet (HFD) starting at 8 weeks of age for 24 weeks. B: Ang1 deletion

was confirmed at 1 or 6 weeks after tamoxifen administration. Genomic DNA extracted from the

tail was used for PCR with primers detecting Cre, Ang1flox, Ang1wt or the deleted Ang1. C: Ang1

deletion was confirmed by qRT-PCR of the mRNA transcripts from the indicated tissues after

tamoxifen administration. The amount of Ang1 transcript was presented as a relative value after

normalization to the amount of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcript

and compared to the value of Ang1sys+/+ set at 1 in each tissue (n=4~6 per group). D: Body weight

after 12 weeks of the indicated diet (n=6~7 per group). E: Intraperitoneal glucose tolerance test

(IPGTT) after 12 weeks of the indicated diet (n=5~7 per group). F: Body weight after 24 weeks of

the indicated diet (n=5~7 per group). G: IPGTT after 24 weeks of the indicated diet (n=8~10 per

group). H: Intraperitoneal insulin tolerance test after 24 weeks of the indicated diet (n=6~7 per

group). I: Serum insulin concentration was measured at baseline and at 15 min after IPGTT

(n=8~12 per group). All data are presented as mean±SE. E, G and H: Closed circle, Ang1sys+/+ with

ND; open circle, Ang1sys-/- with ND; closed triangle, Ang1sys+/+ with HFD; open triangle, Ang1sys-/-

with HFD. *, p<0.05 Ang1sys-/- versus the Ang1sys+/+ in each diet group at each time point. N.S.,

Not significant. Rosa26-CreERT2;Ang1sys+/+, Ang1sys+/+. Rosa26-CreERT2;Ang1sys-/-, Ang1sys-/-.

For Peer Review Only Diabetes Page 26 of 49

Figure 2. β-cell derived Ang1 is critical for glucose homeostasis after high fat diet.

A: Schematic diagram of the animal experiment for β-cell specific deletion of Ang1. Ang1β-cell+/+

and Ang1β-cell-/- mice were fed with normal diet (ND) or high fat diet (HFD) starting at 5 weeks of

age for 24 weeks. B: The genomic DNA extracted from the islets was used for PCR with primers detecting Cre, Ang1flox, Ang1wt or the deleted Ang1 to confirm the β-cell specific deletion of Ang1.

C: β-cell specific deletion of Ang1 was confirmed using the immunofluorescent staining for Ang1.

Scale bars, 100µm. D: Body weight after 12 weeks of the indicated diet (n=10 per group). E:

Intraperitoneal glucose tolerance test (IPGTT) after 12 weeks of the indicated diet (n=16~23 per

group). F: Body weight after 24 weeks of the indicated diet (n=10 per group). G: Intraperitoneal

insulin tolerance test after 24 weeks of the indicated diet (n=5~8 per group). H: IPGTT after 24

weeks of the indicated diet (n=13~23 per group). I: Area under the curve (AUC) of blood glucose level during IPGTT (n=13~23 per group). J: Serum insulin concentration was measured at baseline and at 15 min after IPGTT (n=5~7 per group). All data are presented as mean±SE. *, p<0.05

Ang1β-cell-/- versus Ang1β-cell+/+ in each diet group at each time point. E, G and H: Closed circle,

Ang1β-cell+/+ with ND; open circle, Ang1β-cell-/- with ND; closed triangle, Ang1β-cell+/+ with HFD;

open triangle, Ang1β-cell-/- with HFD. N.S., Not Significant. Rip-Cre;Ang1β-cell+/+, Ang1β-cell+/+.

Rip-Cre;Ang1β-cell-/-, Ang1β-cell-/-.

Figure 3. No morphological and compositional difference of endocrine cells and vascular

density by Ang1 deletion.

Total pancreas was harvested after 24 weeks of HFD. A: Representative immunofluorescent

staining of insulin, glucagon and the blood vessels in the Ang1β-cell-/- or the Ang1β-cell+/+ mice fed either normal diet (ND) or high fat diet (HFD). B, C: Insulin-positive (B) or glucagon-positive (C)

For Peer Review Only Page 27 of 49 Diabetes

area per entire area of the pancreas (n=5~6 per group). D, E: Insulin-positive (D) or glucagon-

positive (E) area per entire area of the islet (n=5~6 per group). F: Total insulin amount per

pancreatic weight (n=4 per group). G: Representative immunofluorescent staining of the blood

vessels by CD31 and the perfused vessels by lectin in the Ang1β-cell-/- or the Ang1β-cell+/+ fed either

ND or HFD. H, I: Area of the blood vessels (CD31-positive area) (H) or perfused blood vessels

(lectin-positive area) (I) per insulin-positive area in the islet (n=4~6 per group). All data are

presented as mean±SE. N.S., Not Significant. Scale bars, 100µm. Rip-Cre;Ang1β-cell+/+, Ang1β-

cell+/+. Rip-Cre;Ang1β-cell-/-, Ang1β-cell-/-.

Figure 4. Impairment of insulin secretion together with the defect in PDX-1 and GLUT2

pathway by Ang1 deletion.

A: Islets were isolated after the indicated diet and the amount of Ins2 transcript analyzed. The

relative amount of Ins2 transcript are presented as a relative value after normalization to GAPDH

compared to the value of Ang1β-cell+/+ with normal diet (ND) set at 1 (n=5~6 per group). B: Cultured

islets from the indicated mice and diet were incubated in either 5.6 or 16.7 mM glucose for 1-hr.

The secreted amount of insulin from each islet was measured from the supernatant and the

percentage of insulin release was calculated by dividing the concentration of insulin in the

supernatant with the total amount of insulin per islet. Insulin content per group was

2314.9±121.13ng (Ang1β-cell+/+) and 2108.3±119.12ng (Ang1β-cell-/-) under ND, and

5163.6±701.96ng (Ang1β-cell+/+) and 4494.4±636.09ng (Ang1β-cell-/-) under HFD (n=4~7 per group).

C: Representative image of PDX-1 Western blot. D: The signal intensity of PDX-1 in Western blot

was divided by that of GAPDH, and then presented as a relative value of Ang1β-cell+/+ with ND set

at 1 (n=4 per group). E: Representative immunostaining of PDX-1. F: The signal intensity of PDX-

For Peer Review Only Diabetes Page 28 of 49

1 per insulin-positive area presented as a relative value of Ang1β-cell+/+ with ND set at 100 (n=4 per

group). G: Representative immunostaining of GLUT-2. H: The signal intensity of GLUT-2 per

insulin-positive area presented as a relative value of Ang1β-cell+/+ with ND set at 100. (n=4 per

group). All data are presented as mean±SE. †, p<0.05 Ang1β-cell-/- versus Ang1β-cell+/+ in each diet

group at each glucose concentration. #, p<0.05 16.7mM versus 5.6mM in the respective diet group

in Ang1β-cell+/+. *, p<0.05 Ang1β-cell-/- versus Ang1β-cell+/+ in each diet group. N.S., Not Significant.

Scale bars, 100µm. Rip-Cre;Ang1β-cell+/+, Ang1β-cell+/+. Rip-Cre;Ang1β-cell-/-, Ang1β-cell-/-.

Figure 5. Loss of pericytes from the vessels in the Ang1 deleted islets.

A: Representative immunostaining of the pericytes (NG2) and the blood vessels (CD31) in the islet

of the Ang1β-cell-/- or the Ang1β-cell+/+ fed either normal diet (ND) or high fat diet (HFD). Scale bar,

100µm. B: The area of blood vessels covered by the pericytes were presented as a percentage by

analyzing the CD31-positive area costained with NG2 (n=4~6 per group). C: Islets were isolated

and the amount of NG2 transcript analyzed. The relative amount of NG2 transcript are presented

as a relative value after normalization to GAPDH compared to the value of Ang1β-cell+/+ with ND

set at 1 (n=4~6 per group). D: The section of pancreas was visualized in the vicinity of the β-cell

and the blood vessel with transmission emission microscopy. The inlet of the image of the lower

panel is a high magnification view. Arrow, disorganized fenestration; Arrow head, caveolae; Lm,

blood vessel lumen; Scale bars, 1µm. All data are presented as mean±SE. *, p<0.05 Ang1β-cell-/-

versus Ang1β-cell+/+ in each diet group. N.S., Not Significant. Rip-Cre;Ang1β-cell+/+, Ang1β-cell+/+.

Rip-Cre;Ang1β-cell-/-, Ang1β-cell-/-.

Figure 6. Higher degree of inflammation in the Ang1 deleted islets.

For Peer Review Only Page 29 of 49 Diabetes

A: Representative immunostaining to visualize the CD45-positive cells, together with collagen IV

in the islet of the Ang1β-cell-/- or the Ang1β-cell+/+ fed either normal diet (ND) or high fat diet (HFD).

Scale bars, 100µm. B: Number of CD45-positive cells per islets (n=4~5 per group). C, D and E:

Islets were isolated and the amount of ICAM-1 (C), VCAM-1 (D) and vitronectin (E) transcript

analyzed. The relative amount of each transcript are presented as a relative value after

normalization to GAPDH compared to the value of Ang1β-cell+/+ with ND set at 1 (n=4~6 per group).

All data was presented as mean±SE. *, p<0.05 Ang1β-cell-/- versus Ang1β-cell+/+ in each diet group.

N.S., Not Significant. Rip-Cre;Ang1β-cell+/+, Ang1β-cell+/+. Rip-Cre;Ang1β-cell-/-, Ang1β-cell-/-.

Figure 7. Schematic diagram of the current investigation.

For Peer Review Only Diabetes Page 30 of 49

209x297mm (300 x 300 DPI)

For Peer Review Only Page 31 of 49 Diabetes

297x210mm (300 x 300 DPI)

For Peer Review Only Diabetes Page 32 of 49

209x296mm (300 x 300 DPI)

For Peer Review Only Page 33 of 49 Diabetes

209x296mm (300 x 300 DPI)

For Peer Review Only Diabetes Page 34 of 49

209x296mm (300 x 300 DPI)

For Peer Review Only Page 35 of 49 Diabetes

209x296mm (300 x 300 DPI)

For Peer Review Only Diabetes Page 36 of 49

209x297mm (300 x 300 DPI)

For Peer Review Only Page 37 of 49 Diabetes

Supplementary Table

Table 1. Primer sequence

F: Forword Gene name Sequence (5’ to 3’) R: Reverse

F CCGGATTCAACATGGGCAATGTGCC Ang1 R CAGTCAAAATGCCTAAGATAAAC

Ang1 F GCAGCCATAGCAATGCCAGAGGT Genotyping (deleted allele) R TCCCATGGCAACTCACAAAACTCC

F AGGTTCGTTCACTCATGGA Cre R TCGACCAGTTTAGTTACCC

F ATACCGGAGATCATGCAAGC RipCre R TGTGCCAAAGGGATTTTAGG

Gene name Taqman assay ID

Ang1 Mm00456503_m1

Ins2 Mm00731595_gH

Gene NG2 Mm00507257_m1

expression ICAM-1 Mm00516023_m1

VCAM-1 Mm01320970_m1

Vitronectin Mm00495976_m1

GAPDH Mm99999915_g1

For Peer Review Only Diabetes Page 38 of 49

Supplementary Figure Legends

Supplementary Figure 1. Deletion of Ang1 in the islets of the Ang1sys-/- mice.

Total pancreas was harvested and immunostained with Ang1 and insulin to demonstrate the effective deletion of Ang1 in the Ang1sys-/- mouse. Representative images are shown. Scale bars,

100µm.

Supplementary Figure 2. No difference in the total pancreas weight between the Ang1β-cell+/+ and the Ang1β-cell-/- mice.

Experimental groups were composed of the Ang1β-cell+/+ and Ang1β-cell-/- mice, each fed with either the ND or the HFD. The entire pancreas was harvested and the weight measured (n=4 per group).

Note that there was no statistical difference in the total pancreas weight between the Ang1β-cell+/+ and the Ang1β-cell-/- mice. All data was presented as mean±SE. N.S., Not Significant.

Supplementary Figure 3. No difference in the VEGFR2-positive blood vessels between the

Ang1β-cell+/+ and Ang1β-cell-/- mice.

Experimental groups are composed of the Ang1β-cell+/+ and Ang1β-cell-/- mice, each fed with either the ND or the HFD. VEGFR2 were visualized together with CD31 and insulin to demonstrate that there was no difference in the VEGFR2-positive blood vessels between the Ang1β-cell+/+ and Ang1β- cell-/- mice. Representative images of the islets are shown for each experimental group. Scale bars,

100µm. Ins, insulin.

Supplementary Figure 4. Impaired PDX-1 expression in old mice compared to young mice

For Peer Review Only Page 39 of 49 Diabetes

PDX-1 was immunostained in 7 and 30 weeks of control mice (Rip-Cre;Ang1β-cell+/+) and a

representative image was presented. Scale bars, 100µm.

Supplementary Figure 5. Difference in the expression level of the PDX-1 and the GLUT2

between the Ang1β-cell+/+ and Ang1β-cell-/- mice.

Experimental groups are composed of the Ang1β-cell+/+ and Ang1β-cell-/- mice, each fed with either

the ND or the HFD (n=4 per group). A: Comparison of the PDX-1 mRNA. B: Comparison of the

GLUT2 mRNA level C: Western blot of the GLUT2 expression. D: Comparison of the GLUT2

protein level. All data was presented as mean±SE. *, p<0.05 Ang1β-cell-/- versus Ang1β-cell+/+ in each

diet group. N.S., Not significant.

Supplementary Figure 6. Recovery of the PDX-1 and GLUT2 mRNA level with exogenous

Ang1 supplementation.

Experimental groups are composed of 0 weeks old Ang1β-cell+/+ and Ang1β-cell-/- mice, each fed with

high fat, high sucrose diet for 4 weeks. The isolated islets were treated with 200ng/ml of COMP-

Ang1 or vehicle for 48hr (n=4 per group). A: PDX-1 transcription by exogenous Ang1 treatment.

Note the significant enhancement of the PDX-1 transcription when Ang1 was supplied

exogenously. B: GLUT2 transcription by exogenous Ang1 treatment. Note the significant

enhancement of the GLUT2 transcription when Ang1 was supplied exogenously. All data was

presented as mean±SE. *, p<0.05 Ang1β-cell-/- versus Ang1β-cell+/+ at 0 hr. *, p<0.05 COMP-Ang1-

treated-group versus vehicle-treated-group in each genotype mice at 48 hr after COMP-Ang1

treatment. N.S., Not significant.

For Peer Review Only Diabetes Page 40 of 49

Supplementary Figure 7. Loss of pericyte coverage from the vessels in the HFD fed Ang1β-

cell-/- mice.

Experimental groups are composed of the Ang1β-cell+/+ and Ang1β-cell-/- mice, each fed with either

the ND or the HFD. PDGFRβ was visualized together with CD31 and insulin to demonstrate the

degree of pericyte coverage in the endothelial cells. Note the loss of PDGFRβ+ pericytes in the

HFD fed Ang1β-cell-/- mice. Scale bars, 100µm.

Supplementary Figure 8. Predominance of F4/80+ immune cell subpopulation in the HFD

induced insulitis.

Experimental groups are composed of the Ang1β-cell+/+ and Ang1β-cell-/- mice, each fed with either

the ND or the HFD. Lymph node was stained as positive control for CD3e and B220. A: Abundant

F4/80+ cells in the islets in the HFD fed Ang1β-cell-/- mice. B: Scarce CD3e+ or B220+ cells in the

islets. Red arrow head, CD3e; Green arrow head, B220. Scale bars, 100µm.

Supplementary Figure 9. Rescue of impaired glucose-stimulated insulin secretion with the

improvement of insulitis in the ex vivo cultured islets.

Experimental groups are composed of 30 weeks old Ang1β-cell+/+ and Ang1β-cell-/- mice, each fed with high fat high sucrose diet for 4 weeks (n=4 per group). A: Recovery of the insulitis phenotype with 7 days of ex vivo culture of the islets in the Ang1β-cell-/- mice. Scale bars, 100µm. B: A significant recovery of the impaired glucose-stimulated insulin secretion was noted along with the improvement of the insulitis. All data was presented as mean±SE. *, p<0.05 Ang1β-cell-/- versus

Ang1β-cell+/+. N.S., Not significant.

For Peer Review Only Page 41 of 49 Diabetes

209x297mm (300 x 300 DPI)

For Peer Review Only Diabetes Page 42 of 49

209x297mm (300 x 300 DPI)

For Peer Review Only Page 43 of 49 Diabetes

209x297mm (300 x 300 DPI)

For Peer Review Only Diabetes Page 44 of 49

209x297mm (300 x 300 DPI)

For Peer Review Only Page 45 of 49 Diabetes

209x297mm (300 x 300 DPI)

For Peer Review Only Diabetes Page 46 of 49

209x297mm (300 x 300 DPI)

For Peer Review Only Page 47 of 49 Diabetes

209x297mm (300 x 300 DPI)

For Peer Review Only Diabetes Page 48 of 49

209x297mm (300 x 300 DPI)

For Peer Review Only Page 49 of 49 Diabetes

209x297mm (300 x 300 DPI)

For Peer Review Only