An intravital microscopy model to study early pancreatic inflammation in type 1 diabetes in NOD mice

ABSTRACT

Intravital microscopy (IVM) of the pancreas has been proven to be an invaluable tool in pancreatitis, transplantation and ischemia/reperfusion research. Also in type 1 diabetes (T1D) pancreatic IVM offers unique advantages for the elucidation of the disease process. Female non-obese diabetic (NOD) mice develop T1D spontaneously by 40 weeks of age. Our goal was to establish an IVM-based method to study early pancreatic inflammation in NOD mice, which can be used to screen novel medications to prevent or delay T1D in future studies. This included evaluation of leukocyte-endothelial interactions as well as disturbances of capillary perfusion in the pancreatic microcirculation.

KEYWORDS:

• functional capillary density
• inflammation
• intravital microscopy
• leukocyte adherence
• Type 1 diabetes

Introduction

Intravital microscopy (IVM) represents a powerful tool to study biological processes in living organisms. IVM offers the unique option to explore highly dynamic cellular processes that cannot be reconstituted in vitro or ex vivo, or when a link between cellular events and tissue pathophysiology is being pursued.¹ This is the case for studies of inflammatory processes within the microcirculation. Inflammatory changes in the microcirculation, such as margination and rolling as well as transition to adhesion and migration of immune cells can only be visualized by intravital imaging.²

Experimental IVM of the pancreas has been proven to be a useful method in areas of biomedical research such as pancreatitis, transplantation, ischemia/reperfusion injury and diabetes.^3-5 In particular, in type 1 diabetes (T1D) as acute inflammatory process pancreatic IVM provides advantages for the elucidation of disease pathogenesis. For studies in T1D several models and mouse strains are in use, most frequently non-obese diabetic (NOD) mice. As in humans T1D is a dynamic disease in NOD mice and by 10–12 weeks of age, immune cells (mostly CD4⁺ and CD8⁺ T cells) are infiltrating pancreatic islets (insulitis), β-cells are destroyed and overt hyperglycemia occurs when 70–75 % destruction of the functional β-cell mass has been achieved.⁶ Up to 95% of female NOD mice develop T1D spontaneously by 40 weeks of age. To our knowledge only 2 studies thus far used pancreatic IVM in NOD mice. Reiner and coworkers described a novel near-infrared fluorescent probe for imaging of pancreatic β-cells.⁷ Others focused on pathognomonic late T-cell mediated islet infiltration.⁸ Our goal was to establish an IVM-based method to study early immune response in NOD mice, during the onset of T1D, which can be used to screen novel medications to prevent or delay T1D in future studies. This included evaluation of leukocyte-endothelial interactions as well as disturbances of capillary perfusion in the pancreatic microcirculation.

Results

Figures 3 and 4 show IVM micrographs of pancreatic tissues of NOD mice under conditions of low (Fig. 3A, 4A) and high inflammation (Fig. 3B, 4B). NOD mice with low inflammation did not show signs of significant leukocyte activation (Fig. 3A). Pancreatic inflammation is shown in Figure 3B: significant increased numbers of firmly adherent leukocytes. Capillary perfusion was reduced after T1D onset compared to animals with low inflammation (Fig. 4A and 4B). It should be noted that individuals with lower levels of pancreatic inflammation displayed more organized, less chaotic and more defined branching blood vessels (Fig. 4A), when compared to pancreatic tissues under higher degrees of inflammation (Fig. 4B).

Figure 1. Setup for pancreatic intravital microscopy. The stand (ST) was constructed from immunohistochemistry tissue block molds stacked on top of one another (to a desired height) with a histology slide glued to the top of the stacked molds. During each IVM procedure, the pancreas (P) of the animal was lifted out of the body and placed onto the microscopy slide. Sterile normal saline was then applied to the tissue, in order to prevent desiccation of the pancreas. Saline was able to accumulate around the tissue during each procedure because of a barrier of ethylene-vinyl acetate polymer that was applied to the boarders of the slide and the immunohistochemistry tissue block mold. A single coverslip (CS) was placed on top of the suspended pancreas in order to facilitate IVM.

Figure 2. Overview of female NOD mouse pancreatic microvasculature (5 fold magnification). Fifteen minutes prior to IVM, Rhodamine-6G and FITC-albumin was administered via tail vein injection.

Figure 3. Intravital micrographs of female NOD mouse pancreatic microvasculature (20 fold magnification), depicting low (A) and elevated (B) degrees of leukocyte adhesion (L and arrows). Fifteen minutes prior to IVM, NOD mice received Rodamine-6G (via tail vein injection) which allowed observation of leukocyte rolling and adhesion via IVM.

Figure 4. Intravital micrographs depicting functional capillary density (FCD) in NOD mice with low (A) and high (B) pancreatic inflammation. Fifteen minutes prior to IVM, NOD mice were administered bovine FITC-albumin via tail vein injection. It can be noted that animals with low degrees of inflammation possessed pancreatic microvasculature with very clearly defined and organized branching vessels (A); whereas, animals with high degrees of leukocyte activation had highly opaque pancreatic tissue, containing fewer perfused vessels and displayed irregular branching patterns in smaller vessels that branched from larger vessels (B).

Quantification of leukocyte activation and functional capillary density is shown in Figures 5A–C. Rolling behavior as well as firm adhesion of leukocyte to the endothelium were significantly increased in early T1D. The reduction of FCD in the pancreas of NOD mice with high grade inflammation (leukocyte rolling >1 cell/second) did not reach statistical significance; however, pancreatic capillary perfusion in individuals with low inflammation (leukocyte rolling <1 cell/second) was found to be 13% higher, on average, when compared to that of individuals with high inflammation.

Figure 5. Leukocyte rolling (A), adhesion (B) and FCD (C). Leukocyte rolling behavior was significantly reduced within pancreatic vessels of individuals with lower degrees of inflammation, when compared to that of individuals with higher pancreatic inflammation (A). We also observed a significant increase in the number of leukocytes firmly adhering to the endothelium in the higher pancreatic inflammation (B). Functional capillary density was not significantly different between female NOD mice with low and high degrees of pancreatic inflammation (C); however individuals with lower degrees of inflammation possessed 13% higher FCD, on average, than that of individuals with a higher degree of inflammation.

Blood glucose levels correlated with the degree of inflammatory changes in the pancreatic microvasculature: animals with significant leukocyte activation (n = 7) showed higher blood glucose levels (Fig. 6).

Figure 6. Blood glucose values (mean and SEM) of NOD mice with low and high degrees of inflammation. Blood glucose levels were assayed via weekly tail vein puncture between 17 and 21 weeks of age.

Discussion

We established an intravital microscopy method to identify and quantify early inflammatory changes of the pancreas in NOD mice in order to study onset and potential therapeutic delay of T1D.

T1D represents an acute onset disease with dramatic consequences for the individual; however, it is also a gradual process in the beginning, offering a therapeutic treatment window to avoid full-blown islet cell destruction.⁹ In contrast to histological or immunohistochemical methods for studying pancreatic pathology, early stages of inflammation (such as leukocyte rolling and adhesion) can be observed and functional consequences, for instance, impact on microvascular blood flow can be studied by using IVM.¹⁰ IVM using an epifluorescence microscope is of advantage over confocal or 2-photon microscopy for the observation of fast moving cells. Furthermore, the equipment is much less expensive, which makes the method more attractive for screening purposes.

IVM of the pancreatic microcirculation is well established.¹¹ Relevant information is obtained immediately by in vivo observation followed by offline quantification using manual or semi-automated methods. Standard intravital fluorescence staining of leukocytes using Rhodamine 6G allows for screening of early inflammatory processes, which are mainly mediated by neutrophils.¹² The typical picture in acute inflammation is an increase in leukocyte rolling and significant firm leukocyte-endothelial adhesion, mostly in postcapillary venules.^13,14 We were able to confirm those changes in the inflammatory response associated with early T1D in the pancreatic microvasculature of NOD mice.

Disturbance of capillary blood flow within the microcirculation represents one of the consequences of increased leukocyte adhesion to the endothelium. However, other factors additionally contribute to the reduced capillary blood flow in inflammation such as impaired erythrocyte deformability, capillary leakage and the resulting tissue edema compressing microvascular perfusion or activation of intravascular coagulation.¹⁵ We observed a 13% decrease in FCD in pancreatic tissue of diabetic NOD mice, most likely related to the described inflammatory changes including leukocyte adhesion.

Clinical signs of T1D onset in the NOD mice accompanied the IVM changes. We observed blood glucose levels above 13.3 mmol/L, indicating T1D onset in the experimental animals with IVM markers of inflammation. This confirms the validity of the model and the relevance of the observed changes within the microcirculation of the pancreas by using IVM.

Pancreatic IVM does have its limitations – it is an invasive method, only permitting single time point observations, even if the observation can be done over a period of time under general anesthesia of the experimental animals. Potential solutions are the use of endoscopic microscopy techniques or implantation of pancreatic tissue into the anterior chamber of the eye.^16,17 Also, with Rhodamine 6G as fluorescence dye, the method does not permit differentiation of immune cells subsets, which can be done, e.g. by using specific antibodies. Intravenous administration of Rhodamine 6G also stains endothelial cells, which are clearly different from immune cells, or platelets, which are much smaller in size and less frequent in the blood stream.

In conclusion, the described method offers the opportunity to study early changes within the pancreatic microcirculation in a mouse model of T1D. This method can be used to screen for novel drugs to control or delay the onset of T1D.

Material and methods

Animals

Thirteen female NOD/ShiLtJ mice (6 weeks of age) were purchased from Jackson Laboratories (Bar Harbor, ME, USA), housed in chip-bedded cages and prior to the experiments acclimatized for one week in the air-filtered institutional Carlton Animal Care Facility of the Faculty of Medicine at Dalhousie University, Halifax, Canada. Animals were kept on a 12 hours light/ dark cycle, with the room temperature kept at 22°C and humidity at 55–60%. A standard diet of rodent chow and sterile drinking water were available ad libitum. All experimental procedures were performed in accordance with the standards and procedures set forth by the Canadian Council on Animal Care.

Blood glucose measurements

Starting at week 17 weeks of age, blood glucose levels were assayed using a One Touch Ultra 2 blood glucose monitor (LifeScan, product code: 021098) and One Touch Ultra Blue blood glucose test strips (LifeScan, product code: 022895) in all individuals weekly until week 21 (end of experiments). The blood glucose monitor was calibrated against a control glucose solution (LifeScan, product code: 010458) at 3 random time points throughout the entire length of the study, in order to ensure the accuracy of the blood glucose results. Clinical diabetes in mice was defined as follows: blood glucose level of at least 13.3 mmol/L following 2 fasting (4 hours) blood glucose tests, on 2 separate days. While fasting, all animals did not receive access to food for 4 hours; however, they did receive water ad libitum. Blood samples were collected from the subject's tail vein, via tail vein puncture with a 25G × 5/8 mm needle (BD Canada, reference code: 305122), following bathing the subject's tail in warm (37°C) sterile normal saline (Hospira, DIN: 00037842) for 30 seconds to one minute. Once the tail had been bathed in warm saline, it was dried with sterilized gauze pads (AMD-RITMED Inc., product code: A2101-CH). Each mouse was then properly restrained, while a second researcher collected blood samples. Once blood samples were collected, a sterile gauze pad was applied to the injection site on the tail of each mouse until the wound had clotted. When all samples were collected, all mice were returned to their home cages, placed back onto ventilation racks and received food ad libitum.

Anesthesia

Animals were weighed with a commercial scale. Using a 1 ml syringe (BD Canada, Mississauga, ON, Canada) equipped with a G25 × 5/8 inch needle (BD Canada, Mississauga, ON, Canada), 54.6 mg/kg of pentobarbital (Ceva Sante Animale, index number: FR/V2770465 3/1992) were administered intraperitoneally. Sufficient depth of anesthesia during the procedure was assessed by the animal's response to ear or tail pinch and, when needed, 5 mg/kg pentobarbital were supplemented. Body temperature of the animals was monitored throughout the experiments using a rectal probe and maintained at 37-38 °C using a heating pad.

Surgery

Prior to surgery, each animal's abdomen was cleaned and disinfected with 70% ethyl alcohol. Laparotomy was performed using a scalpel for the skin incision, followed by lifting of the muscular layer by forceps and cutting with curved tip scissors along the linea alba to open up the muscular layer. We prepared the duodenal loop with the pancreas for IVM by placement of a 22×22 mm-sized cover slip (Fisher Scientific Co., Ottawa, ON, Canada), on the tissue to facilitate IVM. Surrounding portions of the intestine were covered with gauze sponges soaked in warm normal saline (Hospira, DIN: 00037842). During the microscopic procedure, the intestine will be perfused with thermostat-controlled (37°C) saline solution to avoid drying. Animals were transferred on the heating pad to the microscope stage (Fig. 1).

Intravital microscopy

IVM was performed using the following technical devices: an epifluorescent microscope (Leica DMLM, Wetzlar, Germany), light source (LEG EBQ 100, Jena, Germany), BC-71 black/white CCD camera (11 × 11 µm pixel size; Horn Imaging, Aalen, Germany), black/white monitor (Speco Technologies, Texas, US). Video sequences of 30 seconds were recorded using WinDV 1.2.3 (Petr Mourek, Czech Republic; free software for non-commercial use) for off-line evaluation on a PC (Asus Essentid Series).

Fifteen minutes before the start of the IVM leukocytes were stained in vivo by intravenous (tail vein) injection of 0.05% Rhodamine-6G (1.5 ml/kg; Sigma-Aldrich, CAS number: 989-38-8) and the plasma was stained with 5% bovine FITC albumin (1 ml/kg; Sigma-Aldrich, product number: A 9771). Tail vein administration of Rodamine-6G and FITC were both conducted with the use of a 29G × ½ inch 0.5ml U-100 insulin syringe (BD Canada, reference code: 324703), under conditions of minimal light exposure.

The microscope was set to focus the surface of the pancreas (Fig. 2; 5X long distance objective; Leica, Wetzlar, Germany; NA 0.4). Leukocytes were visible in the venules under 20X long distance objective (Leica, Wetzlar, Germany; NA 0.4). Three to 5 visual fields containing non-branching venules over a length of at least 300 μm were recorded for 30 seconds (Movie 1; Fig. 3A & B). Filter set was changed for examinations with FITC-albumin. Video sequences (30 seconds) of 3 to 5 randomly selected visual fields with capillaries were recorded (Fig. 4A & B).

Evaluation of the video sequences was carried out off-line using ImageJ software (version 1.50g; National Institutes of Health, USA; free software for non-commercial use). For measurements of leukocyte activation, the lengths and diameters of the venules under the study were recorded. Rolling leukocytes were defined as the number of cells that during an observation period of 30 seconds passed in a rolling motion through a selected vascular diameter (cells/minute). Adhering leukocytes were quantified by the number of cells that stayed immobile on the vessel wall over the 30-second observation period (cells/μm²). For measurements of capillary perfusion, the functional capillary density (FCD) was calculated by measuring the length of capillaries with red blood cell flow in relation to a predetermined rectangular field (micrometer/μm²).

Statistics

Data are expressed in terms of means and standard errors of the mean (SEM). All statistical results were acquired and graphically represented through computation of experimental data via GraphPad Prism6® software (GraphPad Software Inc., La Jolla, CA, USA). Normal distribution was tested using the Kolmogorov-Smirnov test with Dallal-Wilkinson-Lilliefor correction. Students t-test for independent samples was used for comparisons between groups. Significance was defined as p < .05.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgment

The authors would like to thank Lei Jiang for providing Figure 1.