Critical Role of Tissue Mast Cells in Controlling Long Term Glucose Sensor Function in Vivo

Abstract

Little is known about the specific cells, mediators and mechanisms involved in the loss of glucose sensor function (GSF) in vivo. Since mast cells (MC) are known to be key effector cells in inflammation and wound healing, we hypothesized that MC and their products are major contributors to the skin inflammation and wound healing that controls GSF at sites of sensor implantation. To test this hypothesis we utilized a murine model of continuous glucose monitoring (CGM) in vivo in both normal C57BL/6 mice (mast cell sufficient), as well as mast cell deficient B6.Cg-Kit^W-sh/HNihrJaeBsmJ (Sash) mice over a 28 day CGM period. As expected, both strains of mice displayed excellent CGM for the first 7 days post sensor implantation (PSI). CGM in the mast cell sufficient C57BL/6 mice was erratic over the remaining 21 days PSI. CGM in the mast cell deficient Sash mice displayed excellent sensor function for the entire 28 day of CGM. Histopathologic evaluation of implantation sites demonstrated that tissue reactions in Sash mice were dramatically less compared to the reactions in normal C57BL/6 mice. Additionally, mast cells were also seen to be consistently associated with the margins of sensor tissue reactions in normal C57BL/6 mice. Finally, direct injection of bone marrow derived mast cells at sites of sensor implantation induced an acute and dramatic loss of sensor function in both C57BL/6 and Sash mice. These results demonstrate the key role of mast cells in controlling glucose sensor function in vivo.

Introduction

Diabetes is a chronic disease that afflicts over 20 million people in the United States alone, with an annual cost to the U.S. of over $150 billion in direct and indirect expenditures (1). The public health significance of diabetes is manifested in the various long-term complications resulting in premature death, disability and a compromised quality of life (1). Central to minimizing these devastating complications, and their associate health and economic costs, is maintaining normal blood glucose levels. Currently repeated, “finger sticks” to obtain capillary blood samples is the major approach to monitoring blood glucose levels. Because of the pain and inconvenience of this procedure, patient compliance is often poor. Clearly, there is a critical need for a method that would allow long-term continuous glucose monitoring (CGM) in vivo. Implantable glucose sensors hold significant promise to fill this critical clinical need for people with diabetes. Unfortunately, although the concept of implantable sensors has existed for over 30 years; there has only been limited success in developing a viable implantable glucose sensor that lasts continuously or consistently for more than a few days (2–5). Although, there are short duration transdermal glucose sensors on the market, they are only approved for up to 7 days (3, 6). Generally, both health insurers and governmental agencies have not been willing to pay for these short duration implantable glucose sensors, so their benefits for patient with diabetes is limited. Therefore, developing sensors with extended life span would be more appealing to the health care community. Additionally, glucose sensors with extended lifespan hold the promise of “closed-loop” systems that would utilize these sensors with insulin pumps (artificial pancreas) to control blood glucose levels in patients with diabetes. However, one of the major hurtles to CGM in general, and to closed loop systems specifically is the loss of sensor function due to foreign body tissue reactions to the implanted sensors. Developing a better understanding of the role of the cells, factors and tissue reactions that occur at sites of sensor implantation and their relationship to sensor function will likely provide better rationales and approaches to extending glucose sensor function in vivo.

One of the first, and most explosive tissue responses to the implantation of any object into the skin, including sensors is mast cell activation. This mast cell activation is productive of a myriad of preformed pro-inflammatory mediators and cytokines that immediately initiates tissue inflammation (Figure 1). If the insult is sustained (e.g. chronic inflammation) mast cells can synthesize additional pro-inflammatory mediators and cytokines to recruit and activate a wide variety of leukocytes including macrophages and lymphocytes, which can then contribute to further tissue destruction and inflammation. These mast cells can also produce potent angiogenic factors, such as vascular endothelial cell growth factor (VEGF), which are critical to neovascularization in injured tissue. Additionally these activated mast cells can produce a wide variety of cytokines and growth factors that recruit and activate fibroblasts productive of tissue fibrosis. Surprisingly, although there are over 20,000 publications on mast cells listed in Medline, there is not a single publication on the in vivo role of mast cells or their products on: 1) tissue reactions induced by sensors, or 2) on the impact of mast cells on sensor function in vitro or in vivo. In fact, we were able to find only one study on murine mast cells and biomaterials in vivo in the entire literature (7). This study in fact demonstrated the importance of mast cells in biomaterial (polyethylene terephthalate (PET) disks) induced tissue inflammation. Mast cells play a critical role in cutaneous inflammation and fibrosis and therefore we believe that cutaneous mast cells likely play a central role in controlling tissue reactions and sensor function in vivo. Specifically, we hypothesize that mast cells and their products control the tissue reaction triad of inflammation, neovascularization and fibrosis at sites of sensor implantation by releasing preformed and newly synthesized mediators, cytokines and growth factors (Figure 1). We further hypothesize that the deficiency of mast cells at sites of sensor implantation will extend sensor lifespan by decreasing inflammation and fibrosis. To test this hypothesis we utilized our previously described murine model of continuous glucose sensing in vivo in both normal C57BL/6 mice (mast cell sufficient), as well as mast cell deficient B6.Cg-Kit^W-sh/HNihrJaeBsmJ mice (Sash mice).

Figure 1. Hypothetical model of the role of mast cells and their products in sensor inducted tissue reactions.

This hypothetical model outlines the various possible pathways involved in glucose sensor activation of skin mast cells, as well as the potential preformed and newly synthesized mediator produced by these activated mast cells. Additionally the model presents the relationship between the production of these mediators and the tissue triad of inflammation, neovascularization and fibrosis that are induced directly or indirectly by these mast cell derived mediators. Abbreviations include, arachidonic acid (AA) and Platelet activating factor (PAF).

Methods and Materials

Mast Cell Sufficient and Deficient Mouse models

For these in vivo studies female mast cell sufficient (C57BL/6) and mast cell deficient Sash mice (B6.Cg-Kit^W-sh/HNihrJaeBsmJ) (8) were utilized. Both strains of mice were obtained from Jackson Laboratory (Bar Harbor, Maine).

Glucose Sensors, Implantation and Murine Continuous Glucose Sensor System

All modified Navigator glucose sensors used in these in vivo studies were obtained from Abbott Diabetes Care. Glucose sensors were implanted into mice and continuous glucose monitoring (CGM) was undertaken as described recently with modifications (9, 10). Implanted sensors were secured to the mouse skin with a mesh, and CGM was initiated as previously described (10). For the present studies, all sensor data was presented as raw current signals (nA) in order to evaluate the true non-calibrated signal dynamics, i.e. no sensor calibration or recalibration. Current data were obtained at 60-second intervals. Blood glucose reference measurements were obtained at least daily using blood obtained from the tail vein of the mouse and a FreeStyle® Blood Glucose Monitor. The Institutional Animal Care and Use Committee of the University of Connecticut Health Center (Farmington, CT) approved all the studies involving mice.

Bone Marrow Derived Mast Cells

Bone marrow derived mast cells were obtained from long bones of C57BL/6 mouse as previously described by Razin et al (11). Briefly, bone marrow cells were obtained from the femurs of 6–8 weeks of C57BL/6 mice (Jackson Laboratory, Bar Harbor, ME). Cells were then cultured at a starting density of 1×10⁶ cells/ml in RPMI 1640 media at 37C and at a pH 7.2 in a humidified atmosphere containing 5% CO₂. RPMI 1640 media was supplemented with 10% FBS, 2mM L-glutamine, 0.1mM nonessential amino acids, 100U/ml penicillin/streptomycin, 50 uM 2-mercaptoethanol (Invitrogen, Carlsbad, California, USA) and murine interleukin 3 (IL-3) (10ng/ml, Peprotech, Rocky Hill, N.J., USA). Non-adherent cells were passaged every 4 days for 2 weeks and maintained at a density of 1×10⁶ cells/ml. At the beginning of week 3, murine stem cell factor (mSCF, Peprotech, Rocky Hill, N.J., USA) was added to the cultures at a concentration of 50 ng/ml media every week for the next 3 weeks. Mast cells were identified in the cultures using standard toludine blue (12) and mast cell esterase staining (13–15). Using this protocol bone marrow derived mast cells represented >95 percent of the cells in the final cultures (see Figure 2D).

Figure 2. Procedure for in situ injections of bone marrow derived mast cells (BMDMC) in mouse model of CGM.

Sensor implantation into mouse skin (A), with mesh to secure sensor to mouse skin (B) and 24 hours post implantation and prior to fluid/cell injection (C). Morphology of injected BMDMC is represented in Figure 2D.

Bone Marrow Derived Mast Cell (BMDMC) Injections at Sites of Sensor Implantation

To determine if mast cells interfere with sensor functionality in vivo, we utilized direct injection of cells or buffer at sites of glucose sensor implantation in our mouse model of continuous glucose sensing. Sensors were implanted as described above (Figure 2A and 2B) and after a sensor run-in time of about 24 hours, mice were placed under anesthesia (isoflurane) and the mesh around the implanted sensor was removed (Figure 2C) and either saline or 10⁴ to 10⁵ mast cells in saline were injected in a volume of 30 ul at site of sensor tip location. The injected fluid was allowed to be completely absorbed prior to removing the mouse from the isoflurane anesthesia and returning the mouse into their cage.

Histopathologic Analysis of Tissue Reactions at Glucose Sensor Implantation Sites

In order to evaluate the tissue responses to the glucose sensor implantation at various time points, individual mice were euthanized and the tissue containing the implanted sensors was removed, fixed in 10% buffered formalin for 24 hours, followed by standard processing, embedded in paraffin and sectioned. The resulting 4–6um sections were then stained using standard protocols for H&E (16) and Masson Trichrome (fibrosis) (17), mast cell esterase (13–15), Giemsa May-Grünwald staining (Sigma-Aldrich, Inc. St. Louis, MO) and toludine blue (12). Histopathologic evaluation of tissue reactions at sites of sensor implantation was performed on mouse specimens obtained at 7 days post implantation (DPI) of the glucose sensor, 14 DPI, 21 DPI and 28 DPI. The tissue samples were examined for signs of mast cells, inflammation, including necrosis, fibrosis, and vessel regression. Resulting tissue sections were evaluated directly and documented by digitized imaging using an Olympus Digital Microscope.

Results

Glucose Sensor Function in Mast Cell Sufficient Mice (C57BL/6)

To begin our studies, we first determined the CGM profile of normal C57BL/6 mice (mast cell sufficient mice) over a 28-day post sensor implantation time period (Figure 3A–D). As expected, the ADC modified Navigator sensors displayed excellent CGM during the first 7 days post implantation with glucose sensing closely following blood glucose levels (Figure 3A–D). During the first 7 days the Navigator sensor tracked both hyperglycemic and hypoglycemic events in the normal mice (Figure 3B and 3C). Interestingly sensor function beyond 7–10 days post sensor implantation displayed very heterogeneous patterns of CGM (Figure 3A–D). Representative sensor responses in mast cell sufficient mice C57BL/6 mice included sensors that: 1) lost function at days 14–21 but had good responses at both early and late time frames (Figure 3A); 2) had good response for the first 3 weeks but lost responsiveness in week 4 post implantation (Figure 3B); 3) had a poor response throughout week 2–4 post sensor implantation (Figure 3C) as well as 4) showed episodes of no or poor sensor response past 7 days to 14 days and 16 days to 18 days post implantation (Figure 3D). It is important to note that although the sensor response after 7–10 days post implantation was erratic, there were periods where the sensor response did correspond to sensor output even at 28 days post sensor implantation (Figure 3A and 3C). This data demonstrates that during long-term glucose sensing in normal mice sensor function can be lost but regained, which is comparable to cycles of varying tissue reactions (i.e. cycles of inflammation, neovascularization, fibrosis). In conclusion, these data clearly demonstrate that although the modified Navigator sensor has a very good response profile within the first week post implantation, but the sensor response in normal mice becomes highly variable from weeks 2 to 4 post sensor implantation.

Figure 3. Continuous glucose monitoring (CGM) in mast cell sufficient and deficient mice for up to 28 days post sensor implantation.

Figures 3A – 3D are representative of CGM in mast cell sufficient C57BL/6 mice for up to 28 days post sensor implantation. Figures 3E – 3H are representative of CGM in mast cell deficient Sash mice for up to 28 days post sensor implantation. Sensor output is expressed as CGS output (nA) and is represented by the blue lines. Blood glucose levels are represented by red diamonds. For this study we evaluated a total of 3 to 4 mice for 21 DPI and 8 to 9 mice for 28 DPI each for C57BL/6 and Sash mice.

Glucose Sensor Function in Mast Cell Deficient Mice (Sash: B6.Cg-Kit^W-sh/HNihrJaeBsmJ)

With the completion of the long-term CGM studies in normal (mast cell sufficient) mice, we next determined the impact of mast cell deficiency on CGM using Sash (mast deficient) mice. As was the case with mast cell sufficient mice (C57BL/6), sensor function in the first 7 days post sensor implantation in the mast cell deficient mice (Sash mice) displayed consistent and high sensor output (see Figure 3E – 3H). Analysis of sensor function beyond 7 days post sensor implantation in the MC deficient Sash mice clearly demonstrated that glucose sensor function continued to have a consistent stable sensor output that correlated well with the reference tail vein blood glucose levels (Figure 3E– 3H). In fact, excellent sensor function was consistently seen in Sash mice up to 28 days (Figure 3E – 3H). These studies not only demonstrated that mast cells play a critical role in controlling glucose sensor function in vivo, but also indicated that the high variability in sensor responsiveness post 7 day sensor implantation seen in the mast cell sufficient mice was not due to sensor variability or malfunction, but likely due to variability in tissue responses between mast cell sufficient and deficient mice. We believe this is the case because if sensor malfunction caused the variation in sensor response in normal mice, the mast cell deficient mice would have shown the same wide fluctuations in sensor responses. Therefore, we believe the differences seen in sensor response in these 2 mouse strains are the result of differences in tissue responses. The question remains, are tissue reactions at sites of sensor implantation different in mast cell sufficient mice when compared to mast cell deficient mice?

Inflammation and Fibrosis at the Sites of Glucose Sensor Implantation

The sensor function in normal and mast cell deficient mouse studies described above clearly demonstrate the key role of mast cells in controlling sensor function in vivo. The next obvious question is how do mast cells influence sensor function in vivo? We hypothesized that mast cells and their products drive inflammation and fibrosis at sites of sensor implantation. Therefore we would predict removing mast cells (i.e. mast cell deficiency) would decrease inflammation and fibrosis at sites of sensor implantation. To investigate this possibility we evaluated sensor tissue sites using H&E as well as trichrome staining technology at 7, 14, 21 and 28 days post sensor implantation. As can be seen in Figure 4 mast cell deficiency dramatically decreased tissue reactions of inflammation and fibrosis (Figure 4I – 4Q) when compared to normal/mast cell sufficient mice (Figure 4A – 4H). Specifically, inflammation was consistently more intense in the mast cell sufficient mice both at early stages (day 7) as well as later stages post sensor implantation (days 14–28). Of particular interest was the fact that there was dramatically less macrophage accumulation at the interface of the sensor with tissue in the mast cell deficient mice (Sash) when compared to mast cell sufficient mice (C57BL/6) (Figure 4). This decrease in macrophages at the interface of the sensor is likely significant since it is known that macrophages are key cells in controlling inflammation and fibrosis at sites of tissue injury including foreign body reactions. Using trichrome staining techniques we evaluated the impact of mast cell deficiency on fibrosis at the site of sensor implantation. As expected in normal mice (Figure 4E – 4H) significant fibrosis was seen surrounding the implanted sensors by 7 days post sensor implantation (Figure 4E). This fibrosis increased in density and intensity over the remaining 3 weeks of the study (Figure 4F – 4H). Alternatively, in the mast cell deficient mice there was very little fibrosis throughout the entire 28 day study and even the limited fibrosis that was seen was very sparse and lacked organization (Figure 4M – 4P). The dramatic impact of mast cell deficiency on fibrosis at the tissue sensor interface would likely contribute positively to extending sensor function in vivo by allowing rapid diffusion of glucose to the sensor, as opposed to the inhibitory effects that would occur in the dense fibrosis seen in the normal mast cell sufficient mice (Figure 4E – 4H). Since mast cells are known to have a major role in controlling fibroblast function in vivo, it is possible that the lack of mast cells at the sensor tissue interface in the mast cell deficient mice may have contributed to the decrease if fibrosis. Alternatively, since it is known that mast cell control fibroblast function it is possible that the lack of mast cells and mast cell products may directly decrease both the recruitment and activation of fibroblasts at site of sensor implantation. Of course it is likely that both of the factors equally contributed to the decrease in fibrosis seen in mast cell deficient mice.

Figure 4. Tissue reactions induced at sites of glucose sensor implantation in mast cell sufficient (C57BL/6) and deficient (Sash) mice over a 28-day period.

Histopathologic analysis of tissue from sensor implantation sites in C57BL/6 (Figure 4A–H) and Sash (Figure 4 I–P) mice was done using both standard H&E as well as Masson Trichrome staining techniques. Location of the sensor in the tissue is designated by the asterisk symbol (*). In H&E sections the residual sensor coating appears as a black layer associated with the asterisk symbol. In the Masson Trichrome sections the residual sensor coating appears as a orange layer associated with the asterisk symbol.

Mast cell Distribution at Sensor Implantation Site in Normal and MC deficient mice

Clearly, the data presented above demonstrates that mast cells play a critical role in long-term glucose sensor function in vivo. To begin to unravel the potential mediators and mechanisms responsible for the effects on sensor function in vivo, we evaluated the distribution of mast cells in normal and inflamed tissue at sites of sensor implantation in normal and mast cell deficient mice. Mast cells are distributed throughout the skin of normal C57BL/6 mice. As expected no detectable mast cells were seen in the skin of mast cell deficient Sash mice. In the case of mast cell sufficient mice we also characterized the distribution of mast cells at sensor implantation sites and found that mast cells were never seen in direct contact with the sensor or the sensor-tissue interface where the most intense inflammation and fibrosis occurred, but rather at the margins of the tissue reactions (Figure 5). This pattern of MC distribution at the sensor implantation sites in the normal mice was seen throughout the entire 4 week time period (Figure 5). We also noted that throughout the entire 4 week period of CGM in the MC sufficient mice the mast cells in the margins of the tissue reactions consistently organized themselves in small lines or chains of cells (see Figure 5), giving the appearance of encircling the tissue reactions. As expected related to the mast cell deficient Sash mice there were no detectible mast cells in the non-sensor implanted mouse skin or at the sensor implantation site anytime during the 4 weeks of CGM (Figure 5). This observation supports that the Sash mice are truly mast cell deficient and are not able to mobilize mast cell precursors from the bone morrow.

Figure 5. Mast cell distribution at sites of glucose sensor implantation in the mast cell sufficient (C57BL/6) and Deficient (Sash) mice over a 28-day period.

Mast cell distribution in the tissue obtained from senor implantation sites in mast cell sufficient (C57BL/6) and deficient (Sash) mice was detected using mast cell esterase staining. Using this esterase stain mast cell appear as redish orange granulated cells within the tissue. These mast cells are highlighted with red triangles. Location of the sensor in the tissue is designated by the asterisk symbol (*). Esterase positive mast cell were seen surrounding the implanted sensor in mast cell sufficient C57BL/6 mice (Figure 5A–D). As expected no mast cells were seen at the sensor implantation sites in mast cell deficient mice.

Impact of mast cell injections on CGM in normal and mast cell deficient mice

To eliminate the possibility that non-mast cell related factors/cells could be responsible for the extended life span seen in mast cell deficient Sash mice we considered mast cell reconstitution studies. Intravascular injections of total bone marrow cells or bone marrow derived mast cells (BMDMC) have been used to reconstitute mast cell deficient mice. Unfortunately previous studies have clearly demonstrated that reconstitution of skin mast cells in mast cell deficient mice is not successful when mast cells are delivered via vascular routes (18, 19). Thus, usually direct injection of the mast cells into the skin sites is utilized to achieve reconstitution (18, 19). For our MC reconstitution studies we first implanted the glucose sensors into the mouse and after a 24 hr run in period we injected 10⁴ or 10⁵ bone marrow derived mast cells at the site of sensor implantation. It should be noted that control saline injection at the sensor implantation site of either the mast cell sufficient or deficient mice showed only a brief drop in sensor response (data not shown). This sensor output drop was likely the result of dilution of interstitial glucose levels by the saline volume. Thus, the sensor is actually reading the lower interstitial glucose levels at the site of saline injection. As the saline was rapidly absorbed into the tissue and blood glucose-interstitial glucose levels at site of saline injection returned to normal, sensor output also rapidly returned to normal, usually within 15 minutes after the saline injection (data not shown). Evaluation of the impact of BMDMC injection of sensor function in mast cell sufficient C57BL/6 mice demonstrated that injection of either 10⁴ (Figure 6A) or 10⁵ (Figure 6B & 6C) bone marrow derived mast cells resulted in a dramatic fall in senor output usually 1–2 days following a BMDMC injection (injection time point designated by yellow triangle in Figure 6). Glucose sensor function and blood glucose levels were monitored in the mast cell injected mice for an additional 6 days post cell injection, in which glucose sensor function failed to return in the mast cell injected mouse. As part of evaluating the effect of local injection of mast cells on sensor function, periodically mice were administered an intra-peritoneal (ip) glucose bolus injection (Figure 6: green arrows designate the glucose injection time point). The ip bolus glucose injections did not increase the sensor output in the mast cell sufficient mice despite the fact that blood glucose level increased (Figures 6A – 6C). Similar results were seen in the mast cell deficient Sash mice injected with 10⁴ (Figure 6D) or 10⁵ (Figure 6E and 6F) bone marrow derived mast cells at the site of sensor implantation, i.e. fall of sensor output within 1–2 days post BMDMC and failure to detect significant rise in sensor output after ip glucose bolus injections (Figure 6D). Although a rise in sensor output post ip glucose bolus injection can be detected in Figure 6E about 24 hours post mast cell injection, this phenomena was not reproduced in the following days and was never repeated in other representative studies. Clearly, these studies demonstrate that mast cells including their products have a profound effect on sensor function in vivo. It is possible that these effects include: 1) direct sensor toxic effects on the sensors; 2) mast cells induce tissue responses such as inflammation which limit sensor function; 3) mast cell derived products, such as vasoregulator (e.g. vasoconstriction), vasoactive (histamine and serotonin) as well as vasocoagulation (vascular thrombus (fibrin clot) formation) may be restricting blood flow into the implantation sites or 4) all of the above. Of course future studies to dissect these possibilities will have to be conducted in order to unravel the mechanisms involved. These future studies are important in order to indicate that mast cells and their products are likely important to sensor function directly (sensor toxic) and or indirectly by controlling tissue reactions.

Figure 6. Impact of insitu injections of bone marrow derived mast cells (BMDMC) on continuous glucose monitoring in mast cell sufficient (C57BL/6) and Deficient (Sash) mice.

Navigator glucose sensors were implanted in mast cell sufficient (Figure 6 A–C) and deficient mice (Figure 6 D–F), and 24 hrs later BMDMC (10⁴ or 10⁵ cells) were injected at the site of sensor implantation (green asterisk symbol (*). Figures 6A and 6D are representatives for 10⁴ BMDMC injected cells and Figure 6B, 6C, 6E and 6F are representatives for 10⁵ BMDMC injected cells. Periodically mice received an ip bolus glucose injection (green arrows) and blood glucose was determined by external monitoring (red diamonds).

Discussion

Implantable materials and devices, such as drug delivery systems, pacemakers, artificial joints, and organs play an important role in health care today. In addition to these devices, implantable monitoring devices such as glucose sensors have great potential for improving both the quality of care and quality of life of patients and animals. Unfortunately the development of these implantable sensors has been hampered by the inability of currently designed implantable sensors to overcome their relatively short lifespan in vivo, which is typically less than a week. Unfortunately there is relatively little known about the specific cells, mediators and mechanisms that control this TRT at sensor implantation sites. Clearly, understanding of the cells and factors that drive TRT at sites of senor implantation would allow the development of a rationale approach to extending the lifespan of glucose sensors in vivo. One of the first, and most explosive tissue responses to injury in the skin, e.g. sensor implantation, is mast cell activation. This mast cell activation triggers immediate release of preformed vasoactive and inflammatory mediators in the skin. Interestingly although mast cells are known to play key roles in controlling inflammation, fibrosis and neovascularization in tissues, their role(s) in controlling TRT and sensor function in vivo have not been investigated. In fact, a recent search of the Medline literature database indicated that there is not a single paper on mast cells and glucose sensors! To begin to fill this gap in our knowledge we hypothesized that mast cells in the skin play a central role in controlling local tissue reaction to implanted sensors, and thereby control the function and lifespan of glucose sensors in vivo. We further hypothesized that tissue reactions would be decreased and sensor function and lifespan would be increased in mice that are genetically deficient in mast cells. The results of our present studies not only confirm these hypothesis but provide additional insights into the nature of long-term glucose sensing in our mouse model of continuous glucose monitoring. The basis for these and other related insights and conclusions derived from our present studies are discussed in detail below.

Mast Cells and Tissue Reactions

Mast cells are granulated leukocytes that represent resident leukocytes in a wide variety of tissues including both connective and mucosal surfaces. Mast cells play key roles in both innate and acquired immunity by controlling inflammation and wound healing. In innate immunity mast cells are sentential cells in connective and mucosal tissues, usually located in close proximity to the vasculature. When triggered (activated) by a variety of receptor and non-receptor mediated stimuli, mast cells release a wide variety of preformed and newly synthesized pro inflammatory mediators that control blood flow, vasopermeability, leukocyte recruitment and activation, to name a few examples (also see Figure 1 and Figure 7). Mast cells also control neovascularization and fibrosis by the expression of a wide variety of cytokines and growth factors (20). Additionally, in acquired immunity mast cells are central to tissue reaction triggered by acute hyper-allergic immune responses (e.g. IgE mediated vasoamine release in response to allergens) as well as non-allergic immune responses (20). Clearly the number, distribution and activation state of mast cells at sites of injury and inflammation have a major impact on the nature, extent, duration and reoccurrence of tissue responses, occurring at sites of sensor implantation. Although mast cells are key players in controlling tissue reactions in tissue such as the skin, a growing body of literature suggests that mast cells-macrophage interactions may actually result in synergistic regulation of tissue reactions. Additionally these interactions likely involve subpopulation of mast cells and macrophages. Using immunohistochemical studies we have recently demonstrated the presence of macrophages at the interface between the tissue and the sensor (unpublished data). The presence of these macrophages in close proximity with mast cells supports the potential synergy in controlling tissue reactions (Figure 8A). This observation also supports the concept that blocking the 2-way communication between mast cells and macrophages may be critical in extending sensor life span in vivo (Figure 8B). It is likely that there is specific interplay between activated mast cells and polarized macrophage subpopulations such as M1 (drives inflammation) and M2 (drives wound healing) macrophages, which further regulates tissues reactions (Figure 9). It should also be noted that these same mast cell interactions likely occur between mast cells and fibroblasts, as well as the triad of mast cells, macrophages and fibroblasts. These cellular interactions would likely have profound impacts on sensor function in vivo (Figure 9). Alternatively selectively inhibiting these interactions would likely have a positive impact on both tissue reactions and sensor function (Figure 9). Ultimately we will need to not only define these interactions but also the mediators that drive these interactions if we wish to develop long-term implantable glucose sensors.

Figure 7. Examples of major receptors and mediators associated with mast cells.

Figure 8. Hypothetical Model of Mast Cell-Macrophage-fibroblast interactions at sites of glucose sensor implantation in normal tissue.

Our hypothetical model of mast cell (orange cells) interactions with sensor interface cells (macrophage and fibroblasts) controlling tissue responses (Figure 10A). Tissue reactions include: inflammation (inflam.), neovascularization (Neovas.), fibrosis, regeneration (regen.), vasopermeability (VP), and blood flow (BF). Figure 10B presents potential targets for inhibition that would decrease mast cell-macrophage-fibroblasts driven tissue reactions that control glucose sensor function in vivo.

Figure 9. Model of mast cell macrophage interactions that control tissue reactions and its impact on glucose sensor function in vivo.

Decreased sensor function is represented by red arrows (down), and increased / extended glucose sensor function is indicated by green arrows (up). Abbreviations: proinflammatory M1 macrophages (M1 MQ); wound healing promoting M2 macrophages (M2 MQ); mast cells (MC); activated mast cells (AMC).

Continuous Glucose Monitoring in Normal Mice

Currently all commercial continuous glucose monitors are approved for 3 – 7 days in humans (3, 6). Unfortunately there is limited data on continuous glucose sensor function beyond 7 days in the literature. Thus, we began our studies by characterizing long-term sensor function in our mouse model of CGM (9, 10). As expected, CGM during the first 7 days was excellent, and sensor output closely paralleled blood glucose levels monitored externally (Figure 3). Our modified Abbott Navigator glucose sensor consistently detected both hyperglycemic and hypoglycemic events during the initial 7 days of CGM (Figure 3B – 3C). Interestingly after the initial 7–10 days there was a very heterogeneous response in CGM (Figure 3), with segments in which sensor function matched blood glucose levels and other times when it did not. We believe that these fluctuations in sensor function in normal mice are the results of cycles of inflammation, neovascularization and fibrosis, triggering changes/cycles in the tissue reactions at the site of sensor implantation. This phenomenon is the result from: 1) mechanical tissue injury from the movement of the sensor, 2) subclincial infections at the sensor site, as well as 3) differential recruitment, proliferation and polarization of various leukocyte populations (e.g. macrophages and mast cells). It will be critical to dissect the cells, factors and events that cause these cycles of tissue reactions and their impact on sensor function in vivo.

Continuous Glucose Monitoring in mast cell deficient mice

The analysis of CGM in normal (mast cell sufficient) mice clearly demonstrated that CGM was quite effective during the first 7–10 days post sensor implantation; it was highly variable after this initial period. Parallel studies of CGM in mast cell deficient mice demonstrated that sensor function effectively reflected blood glucose levels during the first 7–10 days post sensor implantation, comparable to CGM in normal mice (Figure 3). Interestingly, further analyses of the CGM results in mast cell deficient mice demonstrated that sensor function continued to accurately and consistently reflected blood glucose levels in the mouse of up to 28 days (Figure 3). We believe that this dramatic difference in CGM between mast cell deficient mice and mast cell sufficient mice is the result of various preformed and newly synthesized mast cell derived mediators (see Figure 7) that drive inflammation and fibrosis. However, these mast cell derived mediators are diminished in the mast cell deficient mice. Therefore, this decrease in inflammation and fibrosis in the mast cell deficient mouse likely results in less tissue destruction, more neovascularization, as well as rapid diffusion of glucose through the interstitium, due to the decrease in mechanical (fibrosis) and metabolic (inflammatory cells) barriers. In fact, our histologic evaluation of the sensor implantation sites supports this hypothesis, since there was dramatically lower levels of inflammation and fibrosis surrounding the sensor in the mast cell deficient mice when compared to the mast cell sufficient mice (Figure 4). Observations in the mast cell sufficient mice demonstrated that the mast cells appeared to remain at the periphery of the tissue reactions, not intermingled with the inflammatory cells (e.g. macrophages) at the sensor tissue interface. This is very reminiscence of the cell distribution pattern seen in chronic inflammation in which lymphocytes “ring” the outer edges of macrophage-laden granulomas. This peripheral regulation of macrophages by mast cells or lymphocytes likely provides better control of the nature and extent of inflammation and repair at sites of foreign body reactions. This positioning of the mast cells likely provides optimal location for mast cells communication (e.g. cytokines and growth factors) with both inflammatory cells (M1 and M2 macrophages) and fibroblasts that are at the sensor tissue interface. Additionally, this ringing of the sensor with mast cells allows the mast cells to act as sentinel cells, which trigger inflammation. For example, if there was significant movement of the sensor this mechanical irritation of the tissues would likely trigger mast cells and initiate a cycle of inflammation, neovascularization and fibrosis (Figure 10). Alternatively since these sensors are transdermal sensors and if there were an infection arising from microbial entry thru the sensor channel of the skin, the mast cells would trigger inflammatory reactions, including the recruitment of leukocytes, which would limit the spread of the infections. It is important to note that mast cells normally function as barrier cells, whether it is in the skin or on mucosal surfaces.

Figure 10. Model of mast cell – macrophage driven cycles of tissue reaction and sensor function at sites of glucose sensor implantation.

Hypothetical models of mast cell driven cycles of tissue reactions that control both tissue reaction and sensor function utilizing M1 and M2 macrophages to control cycles of inflammation and wound healing in vivo. Abbreviations: proinflammatory M1 macrophages (M1 MQ); wound healing promoting M2 macrophages (M2 MQ); mast cells (MC); activated mast cells (AMC); regeneration (Regen.); mast cell recruitment (recuit.).

Mast Cell Injections and CGM

To support our observations that mast cells are critical players in controlling tissue reactions and senor function in vivo, we initially planned to undertake mast cell reconstitution of mast cell deficient Sash mice. Typically for these types of reconstitution studies the entire bone marrow cell population from compatible donors are used. The disadvantage of using whole bone marrow reconstitution is that this reconstitution results in not only mast cells but also all other stem/progenitor cells that are present in the bone marrow. An alternative approach is to induce mast cells subpopulations in vitro from whole bone marrow cell populations. For this approach mouse interleukin-3 (IL-3) plus stem cell factor (SCF) was utilized to selectively induce mast cell populations from whole bone marrow (11). Using this protocol, we were able to induce bone marrow cells into mast cells that are > 95% mast cells. We hypothesized that by directly injecting bone marrow derived mast cells at the site of sensor implantation that these mast cells would reconstitute the skin. We further hypothesized that mast cells that were reconstituted in the skin of the mast cell deficient mice would be activated by the tissue reactions at the site of sensor implantation. This mast cells activation would in turn trigger more inflammation with resulting loss of sensor function. The data from these studies not only confirmed our hypothesis but in fact indicated that even small numbers of mast cells could produce profound loss of sensor function in both mast cell sufficient and efficient mice (Figure 6). However, what was particularly surprising to us was that once sensor function was lost after mast cell injection not even high blood glucose levels resulting from repeated intraperitoneal glucose injections were detected by the sensors (Figure 6). These studies not only underscore the ability of mast cells to control sensor function in vivo, but this also raises questions on the possible mechanisms involved. It is possible that mast cells, when injected, are rapidly activated and trigger major inflammatory reactions in the tissue. Although that could be the case, our prior experience related to acute loss of sensor function in vivo associated with inflammation is that the leukocytes surrounding the sensor create a “metabolic barrier” that consumes the glucose and thereby prevents the glucose from diffusing to the glucose sensor. In the case of the cell based metabolic barrier (e.g. macrophages), sensor response is only seen during acute hyperglycemia, e.g. bolus glucose injection. However, we did not see this effect in the mast cell injected mice. Alternatively it is possible that factors released from the mast cells themselves, i.e. granule contents, could be directly biofouling the glucose sensor. It is also possible that since mast cells produce very potent vasoactive factors, that these factors may limit the blood flow (vasocontrictors) into the sensor tissue site which then results in the failure of the exogenously injected glucose from reaching the sensor. However, it is also possible that all of the above occur at the site of mast cell injection. Regardless of the explanation, it appears that mast cells are extremely important cells in not only controlling tissue reactions at sites of tissue trauma such as sensor implantation, but that these cells, and their related factors likely directly and indirectly control sensor function in vivo.

Conclusion

These studies clearly demonstrate the association and importance of mast cells in GSF in vivo. Future studies to dissect the mediators and mechanism involved in this mast cells effect on sensor function will lay the foundation for not only our understanding of the cells, factors and mechanism that control sensor function in vivo, but also develop a rational approach to extend glucose sensor function in vivo. Extending glucose senor function in vivo is critical to the development of long term closed loop systems that will be of great benefit in the treatment of patients with diabetes.