Beta cell specific probing with fluorescent exendin-4 is progressively reduced in type 2 diabetic mouse models

Abstract

Probes based on GLP-1R agonist exendin-4 have shown promise as in vivo β cell tracers. However, questions remain regarding the β cell specificity of exendin-4 probes, and it is unclear if the expression levels of the GLP-1R are affected in a type 2 diabetic state. Using in vivo probing followed by ex vivo imaging we found fluorescent exendin-4 probes to distinctly label the pancreatic islets in mice in a Glp-1r dependent manner. Furthermore, a co-localization study revealed a near 100 percent β cell specificity with less than one percent probing in other analyzed cell types. We then tested if probing was affected in models of type 2 diabetes using the Lepr^db/db (db/db) and the Diet-Induced Obese (DIO) mouse. Although nearly all β cells continued to be probed, we observed a progressive decline in probing intensity in both models with the most dramatic reduction seen in db/db mice. This was paralleled by a progressive decrease in Glp-1r protein expression levels. These data confirm β cell specificity for exendin-4 based probes in mice. Furthermore, they also suggest that GLP-1R targeting probes may provide a tool to monitor β cell function rather than mass in type 2 diabetic mouse models.

Abbreviations

BHK

baby hamster kidney

Cck/Gst

cholecystokinin/gastrin

DIO

diet-induced obese

Fmoc

fluorenylmethyloxycarbonyl chloride

GLP-1R

glucagon-like peptide-1 receptor

HEK

human embryonic kidney

HPLC

high-performance liquid chromatography

LC-MS

liquid chromatography-mass spectrometry

PP

pancreatic polypeptide

T2D

type 2 diabetes

wt

wildtype.

Introduction

Currently, there is a lack of fundamental knowledge concerning the dynamics of β cell mass and function, and its relation to other metabolic factors connected to the etiology of type 2 diabetes (T2D). Significant attention has therefore been given to the development of non-invasive in vivo β cell imaging techniques.^1,2 Such techniques will enable longitudinal studies of the functional β cell mass and are important for the development of novel therapeutic approaches aiming at preserving or restoring the functional β cell mass. The GLP-1R agonist exendin-4,^3-6 enhances glucose-stimulated insulin secretion via activation of the GLP-1R expressed on β cells. Due to these properties, exendin-4-based imaging probes have been explored for in vivo imaging of β cells and insulinomas in several imaging modalities.^7-13 Although GLP-1R agonists are well studied for their pharmacological effects on β cells, less is known concerning the specificity and dynamics of GLP-1R expression in the pancreas. This is mainly due to a lack of specific antibodies against the GLP-1R, and species differences in GLP-1R expression.^14,15 Moreover, the applicability of exendin-4 probes as β cell probes in T2D conditions has not been comprehensively investigated. Fluorescently labeled exendin-4 injected in vivo enables the visualization of cells expressing the GLP-1R and also the quantification of probe uptake in mice, as the probe will internalize upon GLP-1R binding.¹⁶

Here, we synthesized 2 fluorescently labeled exendin-4 probes; Ex4-Cy3 and Ex4-Cy5, and used these to determine probe specificity in wt and Glp-1r^−/− mouse pancreata using histological analyses ex vivo, and to evaluate the application of Ex4-Cy3 as a β cell probe in T2D-like conditions using 2 mouse models; the Lepr^db/db(db/db) and the Diet-Induced Obese (DIO) mouse.

Materials and Methods

Ethics statement

Mice were maintained under the regulations of the Danish acts of animal protection and animal experiments, and Novo Nordisk A/S guidelines.

Animals and measurements

C57Bl/6J (wt), C57Bl/Ks db/db and db/+ mice were acquired from Taconic Biosciences, Denmark. Mice (db/+) for OPT experiments were acquired from Charles River Laboratories, USA. C57Bl/6 Glp-1r^−/− mice were derived from a previously described line,¹⁷ bred for Novo Nordisk A/S at Taconic. Diet-induced Obese and lean control C57Bl/6J mice were acquired from the Jackson Laboratory, USA. Wt, Glp-1r^−/−, db/db and db/+ mice were fed Altromin maintenance diet 1320 (Altromin Spezialfutter GmbH, Lage, Germany). DIO mice were kept on a 60 % kcal % fat diet (D12492); lean littermates received a 10 % kcal % fat control diet (D12450B) (Research Diets Inc., New Brunswick, NJ, USA) during the course of the study. Blood glucose was measured on a Biosen 5040 S-line analyzer (EKF Diagnostics, Magdeburg, Germany). HbA_1c samples were measured on a Cobas 6000 analyzer (Roche Diagnostics, Indianapolis, IN, USA). Plasma insulin content was measured using a semi-heterogeneous LOCI assay.¹⁸

Synthesis of Ex4-Cy3 and Ex4-Cy5

[Leu14, Leu28]-exendin4-Cys-amide was prepared on a Rink Amide resin using standard Fmoc chemistry and purified by reverse phase-HPLC. 0.7 µmol [Leu14,Leu28]-exendin4-Cys-amide and 1 mg Cy3-maleimide or Cy5-maleimide (GE Healthcare, Little Chalfont, United Kingdom) were dissolved in a solution containing 100 µl DMSO and 2 µl di-isopropyl ethylamine. After 20 min the reaction was complete and the mixture of product with slight excess of Cy3-maleimide or Cy5-maleimide was applied to a PD10 column (GE Healthcare), equilibrated and running in 20 mM ammonium bicarbonate. Identity and purity of the probes; [Leu14, Leu28]-exendin4-Cys-(Cy3-Mal)-amide and [Leu14, Leu28]-exendin4-Cys-(Cy5-Mal)-amide were confirmed by LC-MS and the probe was lyophilized. Receptor activity of the Cy3 and Cy5 probes were tested in a binding assay; the C-terminal fluorophore labeling maintained near-100 % receptor affinity (data not shown), in accordance with previous reports.¹⁹ See suppl. Figure 2 for probe design. Probes were dissolved in 1 X Dulbecco's PBS (Life Technologies, Grand Island, NY, USA).

Figure 2.

Ex4-Cy3 targets β cells in the pancreas. Representative images of islets from Ex4-Cy3-probed C57Bl/6 (wt) mice immunostained with markers for β cells (insulin, A), duct cells (Dolichos Biflorus Agglutinin, B and C), α cells (glucagon, D,) δ cells (somatostatin, E) and PP cells (pancreatic polypeptide, F). (G and H) Optical projection tomography (OPT) generated rendering of an Ex4-Cy5-probed db/+ mouse pancreas (head section). The anatomy in (G) was rendered using tissue autofluorescence. Arrows in (A-F) point at the location of the inserts. Scale bars: (F) = 25 μm for (A-F). (F, bottom insert) = 50 μm for inserts in (A-F). (H) = 1.5 mm for (G and H)

Generation of Glp-1r antibody 7F38A2

Glp-1r^−/− mice¹⁷ were immunized with Glp-1r-transfected Baby Hamster Kidney (BHK) cells and antibodies were generated using hybridoma technology. Glp-1r specific monoclonal antibodies were identified by image-based screening (ImageXpress, Molecular Devices, CA, USA) using the BHK-Glp-1r cells used for immunization and a mock-transfected BHK cell line for counter screen. The 7F38 clone was isolated and purified. Binding of the antibody to the Glp-1r was validated in a flow cytometry experiment using BHK-Glp-1r cells and BHK mock cells. These data will be published elsewhere.

Probing of cells

GLP-1R-transfected Human Embryonic Kidney (HEK) 293 cells¹⁶ and non-transfected HEK 293 control cells were seeded on an 8-chamber slide with 12000 cells per chamber in DMEM with 10 % heat-inactivated fetal bovine serum (Life Technologies). Cells were probed with 10 and 100 nM concentrations of Ex4-Cy3 diluted in 1 X DPBS and incubated for 20 min at 37 C with 5 % CO₂. Negative control cells were incubated with 1 X DPBS. The cells were fixed before mounting using Vectashield mounting medium with DAPI (Vector Laboratories, Burlingame, CA, USA).

Probing of mice and preparation for histology

Mice were injected subcutaneously with 20.1 μg (4 nmol) of Ex4-probe (diluted in 200 μl 1 X DPBS), based on doses applied previously using similar probes.^7,8,20 In the db/db time study a weight-adjusted dose of 804 μg/kg (160 nmol/kg) was used. Vehicle controls were injected with 200 μl 1 X DPBS. The probe was allowed to circulate 5.5-6.5 hours before sacrifice by transcardial perfusion fixation. Pancreata were post-fixed for 2 h before dehydration in 30 % sucrose in mQ water solution overnight, followed by processing using a simplified approach to the smooth fractionator method.^21,22 10 μm cryosections were cut on a Leica CM 3050 (Leica Biosystems, Nussloch, Germany). One section per block was used for analysis.

Immunohistochemistry

Immunohistochemistry was performed using standard protocols. Primary antibodies used were guinea pig α-insulin (Dako, 1:1000), guinea pig α-pancreatic polypeptide (Linco, 1:500), rabbit α-glucagon (Dako, 1:2000), rabbit α-somatostatin (Dako, 1:8000), rabbit α-amylase (Abcam, 1:4000) and mouse α-Glp-1r (7F38A2, Novo Nordisk A/S, 1:1000). Secondary antibodies used were donkey α-rabbit/guinea pig DyLight 488, 643 and donkey α-mouse Cy2 Fab fragment (Jackson ImmunoResearch, 1:500). Nuclei were stained with DAPI (Jackson ImmunoResearch, 1:1000). Duct cells were visualized with FITC-conjugated Dolichos Biflorus Agglutinin (DBA, Vector Laboratories, 1:500). Slides were mounted using Fluorescent Mounting Medium (Dako).

Imaging and quantification

Images were acquired with an Olympus VS-120 slide scanner using the UIS-2 UPlanSApo 20x (NA=0.75) objective (Olympus, Tokyo, Japan). Confocal images were acquired with an Olympus Fluoview FV10i microscope with 60x magnification. Editing was performed in respective microscope software and Adobe Photoshop Elements v 9.0 (Adobe Systems, San Jose, CA, USA). Mean pixel intensity and islet area were quantified using Visiopharm Integrator System v 4.6.1.630 (Visiopharm A/S, Hoersholm, Denmark). Counting of cells was performed manually at 20x magnification and presence of co-localization was visually determined.

Optical Projection Tomography

Head sections of Ex4-Cy5-probed pancreata were mounted in 1.5 % low-melt agarose (Invitrogen, Carlsbad, CA, USA). Agarose plugs were dehydrated in 100 % methanol (Merck, Darmstadt, Germany) for 24 h followed by optical clearing in BABB solution²³ for 48 h. Samples were scanned in the Bioptonics 3001M scanner (Bioptonics, Edinburgh, United Kingdom). Images were reconstructed using NRecon v. 1.6.10.1 (Bruker microCT, Kontich, Belgium), and viewed in Imaris v. 7.1.1 (Bitplane AG, Zürich, Switzerland).

Statistics

The Student's t-test was used (p-value: 0.05, 95 % CI). Welch's correction was applied for unequal sample sizes. GraphPad Prism 6 (GraphPad Software, San Diego, CA, USA) was used for statistical analyses.

Results

Ex4-Cy3 probes cells in vitro and mouse islets in vivo in a Glp-1r-dependent manner

GLP-1R-transfected HEK 293 cells displayed uptake and internalization of Ex4-Cy3 at concentrations of 10 and 100 nM (Fig. 1B and C). The probe was not detected in non-transfected control cells (Fig. 1D-F). In line with previous results using liraglutide,²⁰ Ex4-Cy3 targeted islets of wt mice (Fig. 1G), colocalizing with insulin (Fig. 1H and I). No Ex4-Cy3 was detected in Glp-1r^−/− mouse islets (Fig. 1J–L).

Figure 1.

Ex4-Cy3 specifically binds the GLP-1R on cells in vitro and islets in vivo. (A-C) GLP-1R transfected HEK 293 cells and (D-F) non-transfected HEK 293 control cells probed with Ex4-Cy3. (G-L) representative images of Ex4-Cy3-probed wt and Glp-1r^−/− islets. (A-F) are confocal images, (G-L) are flatfield images. Scale bars: (F) = 40 μm for (A-F). (L) = 25 μm for (G-L).

Exendin-4 probes specifically label β cells in the mouse pancreas

To investigate Ex4-Cy3 probing specificity in the pancreas, sections from probed wt, db/+ and Glp-1r^−/− pancreata were immunostained for endocrine, acinar and duct cells. Ex4-Cy3 targeted insulin-immunoreactive β cells with 99.9 % frequency (Suppl. Table 1, Fig. 2A). In general, no signal was detected in glucagon-immunoreactive α and somatostatin-immunoreactive δ cells (Fig. 2D and E), apart from a few weakly probed cells (Suppl. Table 1). PP cells were negative for Ex4-Cy3 (Suppl. Table 1, Fig. 2F). Duct cells were negative for Ex4-Cy3 in general (Suppl. Table 1). Infrequently, we observed single cells located in or near ducts that were strongly Ex4-Cy3-positive, however, in all cases these were confirmed to be isolated β cells based on immunofluorescent detection of insulin (Fig. 2B and C). No clear evidence of probing was seen in acinar cells (Suppl. Fig. 1A-B) as no significant fluorescence intensity difference was measured in acinar cells between wt and Glp-1r^−/− pancreata (Suppl. Fig. 1C). Islet-specific labeling was further confirmed on an intact organ level when Ex4-Cy5-probed pancreata were analyzed using optical projection tomography (Fig. 2G and H).

Ex4-Cy3 probing is reduced in 3-month old db/db mice

Downregulation of Glp-1r expression in β cells has previously been reported in the db/db mouse.²⁴ We investigated if this downregulation translates into reduced Ex4-Cy3 probing in db/db islets. Prior to probing, weight and glucose levels were measured in the mice. The db/db mice were obese and hyperglycaemic compared to db/+ controls (Suppl. Table 2). A 3-fold reduction in Ex4-Cy3 probing intensity was observed in the db/db mice (Fig. 3A-B and G). However, the β cell specificity was retained when comparing the probing against insulin-immunoreactive cells (Fig. 3B and D). There was no significant difference in the insulin immunostaining intensity between db/+ and db/db mice (Fig. 3C, D and H).

Figure 3.

Ex4-Cy3 mean probing intensity is reduced in the db/db mouse. Representative images of islets from Ex4-Cy3-probed and insulin-immunostained db/+ (A, C and E) and db/db (B, D and F) mice. (G and H) displays the mean islet fluorescence intensity levels measured for the Ex4-Cy3 probe and the insulin immunostaining, respectively. The line in the graphs represents the mean fluorescence intensity measured for the surrounding exocrine tissue. Data are normalized to controls and presented as means with SEM. ***P < 0.001, n=3. Scale bar = 25 μm

Reduced Ex4-Cy3 probing intensity in db/db islets correlates with reduced Glp-1r expression

We have developed an antibody against the mouse Glp-1r (7F38A2). Here, the specificity of antibody 7F38A2 was tested on pancreatic sections. In wt islets, the antibody labeled β cells (Fig. 4A), and the signal was predominantly located to the membrane (Fig. 4A, insert). Previous studies have shown that the GLP-1R internalizes upon ligand activation.^16,20 This translocation was observed in Ex4-Cy3-probed islets, as the labeling of 7F38A2 was shifted to a cytoplasmic pattern (Fig. 4B, insert), co-localizing with Ex4-Cy3 (Fig. 4E and H). No immunoreactivity was detected in Ex4-Cy3 probed Glp-1r^−/− islets (Fig. 4C, F and I). Due to the reduced Ex4-Cy3 probing intensities observed in the db/db mice, sections from Ex4-Cy3 probed 3-month old db/+ and db/db pancreata were immunostained with antibody 7F38A2 (Fig. 4J and K). A significant reduction in the mean fluorescence intensity of 7F38A2 in db/db islets was observed (Fig. 4L). The decrease was similar to what was measured for the Ex4-Cy3 probe. This suggests that the reduction in Ex4-Cy3 probing is a direct consequence of reduced receptor expression.

Figure 4.

The novel antibody 7F38A2 recognizes the mouse Glp-1r specifically; staining intensity is reduced in db/db islets. 7F38A2-immunostained (Glp-1r) islets from vehicle control (A, D and G), and Ex4-Cy3-probed (B, E and H), wt pancreata. (C, F and I): Representative images of an Ex4-Cy3-probed Glp-1r^−/− mouse islet immunostained with 7F38A2 and insulin. (J and K): representative islets from Ex4-Cy3-probed and 7F38A2-immunostained db/+ and db/db pancreata. The mean islet fluorescence intensity for 7F38A2 in db/+ and db/db islets is shown in (L). The line in the graph represents the mean fluorescence intensity measured for the surrounding exocrine tissue. Data are normalized to controls and presented as means with SEM. Arrows in (A), (B), (J) and (K) point at the location of inserts. (A, B, D, E, G and H) are confocal images. (C, F, I, J and K) are flatfield images. **P < 0.01, n=3. Scale bars: (H) = 40 μm for (A, B, D, E, G and H). (B, insert) = 10 μm for inserts in (A) and (B). (K) = 25 μm for (C, F, I, J and K)

Ex4-Cy3 probing and 7F38A2 immunoreactivity declines progressively in db/db islets with time

Next, we investigated whether the reduced Ex4-Cy3 probing is an inherent phenotype linked to leptin deficiency or whether it is progressive, associated with the physiological conditions observed in the db/db mice. Furthermore, we examined whether Glp-1r immunoreactivity corresponds with the Ex4-Cy3 probing in db/db islets with time using the novel 7F38A2 antibody. A group of db/+ and db/db mice were followed between ages 6-12 weeks where weight and glucose levels were monitored (Fig. 5A-C). Another group of weaning-age mice (4 weeks) was used for the earliest time point (Suppl. Table 3). Prior to probing and sacrifice, plasma insulin levels were measured in all mice (Fig. 5D). In the 4-week age group (Suppl. Table 3), there was no difference in weight between the mice. At 6 weeks, the db/db mice had gained more weight compared to controls, and at 9 weeks the db/db mice presented with obesity. This had further developed at the final 12-week time point where the db/db mice were severely obese in comparison to lean db/+ controls (Fig. 5A). Glucose levels were within normal values between mice in the 4-week group (Suppl. Table 3). The glucose levels and HbA_1c values, respectively, were similar between db/db and db/+ mice at the 6-week time point. At 9 weeks the db/db mice displayed higher glucose levels although there was no significant difference in the HbA_1c levels. At 12 weeks, glucose levels had further increased in the db/db mice with HbA_1c levels now also being significantly elevated (Fig. 5B-C). Plasma insulin levels increased rapidly and were consistently higher in db/db mice throughout the study (Fig. 5D). Analysis of Ex4-Cy3 mean probing intensity in islets did not reveal a difference between db/+ and db/db mice at 4 weeks of age (Fig. 6A, A' and M). At 6 weeks the mean probing intensity was significantly declined in db/db mice (Fig. 6B, B' and M). This decline continued progressively through the 9-week time point (Fig. 6C, C' and M), to the 12-week point (Fig. 6D, D' and M). The immunoreactivity of 7F38A2 also declined progressively in db/db islets throughout the study, matching the decrease in Ex4-Cy3 probing (Fig. 6E-H' and N). No significant difference in insulin immunoreactivity was measured between mice during the study (Fig. 6I-L' and O). These results suggest that the reduction in Glp-1r expression is progressive in the db/db mice and is possibly linked to changes in β cell health.

Figure 5.

Metabolic parameters for db/+ and db/db mice. Weight and blood glucose were monitored twice a week in db/db mice and lean db/+ controls between ages 6 weeks (42 days) and 12 weeks (84 days) (A-B). HbA_1c levels were measured once weekly (C). Plasma insulin levels were measured before sacrifice (including 4-week mice) (D). The “X” in (A-C) marks the time points for probing and sacrifice of mice. Data in (A-D) are presented as means with SEM. (A and B) 1^st measurement n = 20 and 15 for db/db and db/+ mice, respectively. 2^nd measurement n = 16 and 11 for db/db and db/+ mice, respectively. 3^rd – 7^th measurement n = 9 and 11 for db/db and db/+ mice, respectively. 8^th– 14^th measurement n = 5 and 7 for db/db and db/+ mice, respectively. (D) n = 3 and 4 for db/db and db/+ mice aged 4 weeks (28 days), respectively. n = 4 for db/db and db/+ mice aged 6 and 12 weeks (42 and 84 days), respectively. n = 1 and 4 for db/+ and db/db mice aged 9 weeks (63 days), respectively. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Figure 6.

Ex4-Cy3 mean probing intensity and antibody 7F38A2 immunoreactivity declines progressively in db/db mouse islets over time. (A-D) and (A'-D'): representative images of islets from Ex4-Cy3-probed db/+, and db/db islets, respectively, between 4-12 weeks of age. (E-H) and (E'-H'): representative images of Ex4-Cy3-probed db/+ and db/db islets, respectively, immunostained with antibody 7F38A2 (Glp-1r). (I-L) and (I'-L'): representative images of Ex4-Cy3-probed db/+ and db/db islets, respectively, immunostained with insulin. (M-O) display the mean islet fluorescence intensity values measured for the Ex4-Cy3 probe, antibody 7F38A2 and insulin, respectively. The lines in the graphs represent the mean fluorescence intensity for the surrounding exocrine tissue for the respective channel. Mean fluorescence intensity graphs are presented as means with SEM. Data are normalized to the respective control of each age group. **P < 0.01, ****P < 0.0001, n = 3. Scale bar = 50 μm

Reduced Ex4-Cy3 probing was also observed in DIO mice

To determine whether the phenomenon of decreased Ex4-Cy3 probing and 7F38A2 immunoreactivity seen in db/db islets could be applied to other T2D mouse models, we performed a probing experiment using DIO mice. At both 15 and 23 weeks of age, DIO mice were heavier than their lean littermates (Table 1). Glucose levels did not differ significantly at either time point, however, plasma insulin levels were elevated in DIO mice at the 23-week time point (Table 1). Mean islet Ex4-Cy3 probing intensities and antibody 7F38A2 immunoreactivity at 16 weeks of age did not differ between the mice (Fig. 7A-B, D-E, M and N). The mean Ex4-Cy3 probing intensity was significantly lower in the DIO mice at 23 weeks (Fig. 7G, J and M). 7F38A2 immunoreactivity in the DIO mice was decreased to a similar level as the probe at 23 weeks, although the decrease was not statistically significant (Fig. 7H, K and N). The mean insulin immunoreactivity between DIO mice and lean littermates did not differ statistically during the study (Fig. 7C, F, I, L and O).

Table 1.

Metabolic parameters for DIO mice and lean littermate controls.

	Non-fasting
	Age 15 weeks		Age 23 weeks
	Lean (13)	DIO (15)	Lean (9)	DIO (6)
Weight (g)	26.1 ± 1	34 ± 0.7****	32.9 ± 2.1	40.6 ± 0.6*
Glucose (mmol/l)	8.4 ± 0.7	9.3 ± 0.6	8.9 ± 0.2	9.8 ± 0.5
Insulin (pmol/l)	180.2 ± 32.3	190.5 ± 31.9	239.5 ± 38.8	572.4 ± 116.3*

Data are presented as means with SEM. Numbers within brackets indicate the number of animals in each group. *P < 0.05, ****P < 0.0001

Figure 7.

Ex4-Cy3 probing intensity is reduced in 23 week-old Diet-Induced Obese (DIO) mice. (A-F): representative islets from 16-week old lean control and DIO mice, respectively, probed with Ex4-Cy3 and immunostained with antibody 7F38A2 (Glp-1r) and insulin. The 23-week old counterparts are shown in (G-L). (M-O): mean islet fluorescence intensity levels measured for the Ex4-Cy3 probe, antibody 7F38A2 and insulin in the 16 and 23-week old mice, respectively. Data are normalized to the respective control and presented as means with SEM. *P < 0.05, n = 3. Scale bar = 50 μm.

Discussion

In this study, we investigated the use of fluorescent exendin-4 as in vivo β cell probes in mice. A detailed evaluation ex vivo using the Ex4-Cy3 probe revealed a near-100 % β cell probing specificity with less than 1 % probing frequency observed in other examined pancreatic cell types. The probing intensity was strongest in β cells compared to very weak intensities observed in the rare α, δ and duct cells classified as Ex4-Cy3-positive. This demonstrates that exendin-4 probes are β cell specific in the rodent pancreas, which is in line with recent findings.^13,25,26 Besides β cells, the exact pancreatic expression sites of the GLP-1R have been a subject of debate due to conflicting data reported in earlier studies.^27-32 Recent studies have demonstrated that many commercially available GLP-1R antibodies are unspecific, and that there exists interspecies variation in pancreatic GLP-1R expression.^14,15 These findings present a likely source for the variable results obtained in localization studies. We recently developed a novel antibody against the primate GLP-1R, and predominant β cell labeling was observed within islets in humans and monkeys.³³ Targeting the same epitope we have developed and are currently validating an antibody against the rodent Glp-1r. Here, we tested the specificity of this antibody (7F38A2) in wt and Glp-1r^−/− mouse pancreata. We observed a Glp-1r-dependent labeling of β cells in islets, and the antibody labeling pattern shifted from membranous to cytoplasmic when Ex4-Cy3-probed tissue was immunostained with 7F38A2, indicating binding to the internalized receptor. A manuscript describing the full validation of 7F38A2 will be published elsewhere. Next, we proceeded to test the applicability of the Ex4-Cy3 probe as a β cell labeling agent in T2D-like conditions in mice, performing studies with the db/db and the DIO mouse. The islet probing intensity of Ex4-Cy3 was progressively reduced in db/db mice, rendering the islet-specific signal nearly equal to background levels in 3-month old mice. Although physical factors might hamper the delivery of Ex4-Cy3 probe to the islets, such as changes in islet vasculature due to leptin deficiency,^34,35 the simultaneous decrease in the immunoreactivity of antibody 7F38A2 suggests that the reduced probing is due to downregulated expression of the Glp-1r protein on the β cell surface. Interestingly, the β cell probing specificity was retained despite the decrease in Ex4-Cy3 probing intensity. This indicates that all β cells are still capable of probe binding and internalization. It is important to note that although the probing is reduced there is still sufficient expression of Glp-1r in the β cells to mediate an incretin response since studies have shown a positive glycaemic response following treatment with GLP-1R agonists in db/db mice.^36,37 In a previous study, the downregulation of Glp-1r expression was indeed suggested to be caused by hyperglycaemia in the db/db mouse and in the pancreatectomized hyperglycaemic rat, as the normalization of glucose levels also normalized Glp-1r expression.²⁴ However, due to the rather mild glycaemic profile observed in the db/db mice in our study, we suggest that hyperglycaemia is not the sole cause of Glp-1r downregulation as other factors including hyperinsulinaemia, dyslipidaemia and obesity might affect the expression. Although the progressive reduction in Ex4-Cy3 probing indicates that downregulation of the Glp-1r is not a direct cause of leptin deficiency, the absence of leptin signaling might aggravate the situation in the db/db mice. In addition to the observations in db/db mice, we found the probing intensity to gradually decrease also in the DIO mouse, a commonly used T2D mouse model.^38,39 However, the change in Ex4-Cy3 probing intensity was subtle in DIO mouse islets in contrast to the decrease in db/db mice. Whether the decreased probing intensity seen in DIO mice can be attributed to decreased Glp-1r expression remains to be investigated, though it can be considered likely since we observed a similar decrease in the immunoreactivity of antibody 7F38A2 in the islets of the older DIO mice. The change in 7F38A2 immunoreactivity was non-significant in contrast to the Ex4-Cy3 probe, however. Moreover, previous studies have demonstrated that expression of the Glp-1r is negatively affected in T2D-like conditions in rodent pancreata.^24,40,41 Whether this phenomenon also occurs in humans is not clear. A previous study reported T2D-related GLP-1R downregulation in human islets,⁴⁰ whereas in a recent study conducted by us there was little or no change in GLP-1R antibody immunoreactivity between normal islets and islets from a T2D patient.³³ As noted above, expression levels of the Glp-1r could be normalized following re-establishment of normoglycaemia in the pancreatectomized hyperglycaemic rat.²⁴ An interesting possibility is therefore that treatment with plasma glucose-lowering drugs over an extended period of time could restore the levels of the GLP-1R on β cells.

In summary, we have demonstrated that exendin-4 probes are β cell specific in the mouse pancreas. This study also suggests that changes in the probe signal intensity will not directly reflect changes in β cell mass; emphasizing that probing intensity will be a combined readout of β cell mass and function in line with the progression toward a T2D-like condition in mice. Our data indicate that even a mild degree of such a condition impacts Glp-1r expression negatively in mice, resulting in decreased probing levels. The true β cell mass will therefore be underestimated should exendin-4 probes be used in these circumstances, emphasizing that other probing alternatives need to be considered should the exact β cell mass be measured in T2D mouse pancreata. This also indicates that a combination of 2 or more probes together with known biomarkers for β cell function, combined with longitudinal studies in the same individual, might be a solution for collecting more information regarding β cell status in these conditions. In T2D mouse models, our data indicate that exendin-4 probes can have potential as markers to predict β cell function as the reduced probing intensity might be linked to changes in β cell health in these circumstances. Since treatment with GLP-1R agonists reduces β cell stress,^37,42 future studies should elucidate whether such treatment restores Glp-1r expression levels which would subsequently restore probing levels in these mouse models.

Disclosure of Potential Conflicts of Interest

The authors are full-time employees at Novo Nordisk A/S, which markets liraglutide for diabetes treatment, and hold minor stock portions in the company as part of an employee offering program.

Acknowledgments

The authors wish to acknowledge: F. Strauss, H. Duus Laustsen, H. Solvang Nielsen, H. Hvid, A. Bowman and S. L. Riisberg for assistance with animal experiments. P. Proidl for the assistance with cell culture experiments. Special thanks to Dr. R. S. Heller for inspiring discussions and for proofreading the manuscript.

Data included in this study was presented at the Joint Meeting of the Islet Study Group and Beta Cell Workshop 2015, and at the Pre-congress Symposium on Beta Cell Imaging 2015

Author Contributions

J. L., J. H-S. and J. A-R. outlined and designed research strategy. L.S. designed and synthesized the exendin-4 probes. M. G. R. designed and synthesized the Glp-1r antibody. J. L. performed research. J. L., J. H-S. and J. A-R. analyzed the data. J. L. wrote the manuscript. J. H-S. and J. A-R reviewed and edited the manuscript.

Funding

The research leading to these results has received funding from the People Program (Marie Curie Actions) of the European Union's Seventh Framework Program FP7/2007-2013/ under REA grant agreement n 289932.

Supplemental Material

Supplemental material for this article can be accessed on the publisher's website.