Factors Influencing Quantification of in Vivo Bioluminescence Imaging: Application to Assessment of Pancreatic Islet Transplants

Abstract

The aim of this study is to determine and characterize factors influencing in vivo bioluminescence imaging (BLI) and apply them to the specific application of imaging transplanted pancreatic islets. Noninvasive quantitative assessment of transplanted pancreatic islets poses a formidable challenge. Murine pancreatic islets expressing firefly luciferase were transplanted under the renal capsule or into the portal vein of nonobese diabetic–severe combined immunodeficiency mice and the bioluminescence was quantified with a cooled charge coupled device camera and digital photon image analysis. The important, but often neglected, effects of wound healing, mouse positioning, and transplantation site on bioluminescence measurements were investigated by imaging a constant emission, isotropic light-emitting bead (λ = 600) implanted at the renal or hepatic site. The renal beads emitted nearly four times more light than hepatic beads with a smaller spot size, indicating that light absorption and scatter are greatly influenced by the transplant site and must be accounted for in BLI measurements. Detected luminescence decreased with increasing angle between the mouse surface normal and optical axis. By defining imaging parameters such as postsurgical effects, animal positioning, and light attenuation as a function of transplant site, this study develops BLI as a useful imaging modality for quantitative assessment of islets post-transplantation.

Introduction

Pancreatic islet transplantation has great potential for the treatment of type 1 diabetes mellitus [1–3]. However, the number of transplanted islets needed to overcome diabetes presents a major obstacle precluding islet transplantation from being adopted as a routine treatment. In addition, diabetes reversal in most patients requires islets isolated from at least two donor pancreata [2]; the supply of donor pancreata for islet isolation falls well short of the number needed to treat type 1 diabetics in the United States [2,3]. Thus, significant efforts are focused on preserving or increasing islet mass post-transplantation [4–6]. The survival rate of transplanted islets is incompletely defined; in addition to immune factors, studies indicate that hypoxia, nutrient deprivation, and inflammation hamper islet engraftment and survival and result in significant islet loss in the early post-transplantation period [7,8]. Efforts to overcome these obstacles are limited because no suitable method of noninvasively measuring the number of surviving islets or islet mass currently exists. Islet mass is commonly estimated from insulin secretion following glucose tolerance testing, but this method assesses islet function, which does not necessarily correlate with islet mass. Morphometric analysis of histological sections of islet grafts can be used to measure islet mass, but requires removal of the organ containing the islets, preventing any sequential studies [9]. Additionally, this morphometric analysis is difficult to perform when islets are scattered, as they are when embolized throughout the liver, the most common site of transplant [2,3].

This study sought to further develop in vivo bioluminescence imaging (BLI) as a method to quantify the number of islets surviving post transplantation. BLI refers to the generation of photons by a biologic source such as cells or bacteria as a result of an ATP- and oxygen-dependent enzyme reaction (usually luciferase) with the enzyme substrate [10,11]. BLI has been utilized to assess a number of biologic processes such as inflammation, wound healing, and tumor cell growth in vivo [12–15]. As recently described by Lu et al. [16] and by our group [17], BLI has recently shown promise for monitoring transplanted pancreatic islets. However, a number of imaging parameters remain undefined and these must be addressed before BLI can be used to accurately quantify transplanted islet mass. For example, islets are commonly transplanted to different anatomical locations in animal models. Correlation of light emission and transplanted islet mass must take into account the factors that influence light transmission from the bioluminescent source to the charge coupled device (CCD) camera aperture. Light transmission is determined by the optical properties of the tissue through which the light must pass [18]; different islet graft locations are subject to different degrees of light attenuation. In this report, constant emission, isotropic light-emitting beads with spectral emission similar to the luciferase reaction were implanted beneath the renal capsule or into the liver of mice to serve as a model of transplanted islet bioluminescence. The luminescent beads provide a constant, known light intensity that is reliable and reproducible, allowing for validation and calibration of the imaging method. In contrast, bioluminescent islets are subject to biological variability as islet light emission depends on the health and size of the islets and survival of islets post-transplantation. The effects of wound healing, mouse positioning, and light attenuation by tissues overlying the islet grafts were determined by imaging these bead-bearing mice to identify factors that must be taken into account when correlating light emission with islet mass. Although applied to the BLI of transplanted islets, the findings in this study are relevant for quantitative BLI in general where subtleties of confounding factors that may alter the amount of light detected are often ignored.

Materials and Methods

Animal Model

Nonobese diabetic–severe combined immunodeficiency (NOD-SCID) mice from Jackson Laboratories (Bar Harbor, ME) were used for transplant studies as previously described [19]. The NOD-SCID strain is homozygous for the severe combined immune deficiency (SCID) spontaneous mutation, characterized by an absence of functional T cells and B cells. The NOD-SCID strain accepts allografts without immune rejection.

Mouse Islet Isolation and Luciferase Expression

Murine pancreatic islets were isolated from adult B6D2 mice as previously described [19]. Briefly, mouse pancreata were digested with collagenase P (Roche Molecular Biochemicals, Indianapolis, IN) in Hanks buffered saline (0.6 mg/mL) using a wrist action shaker. Some islets were then handpicked under microscopic guidance. Others were purified by histopaque gradient centrifugation and washed three times with 10 mM PBS containing 1% mouse serum. Islets were suspended in 30 μL of 10 mM PBS with 1% mouse serum for transplantation. Murine islets were cultured in RPMI 1640 (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and 11 mM glucose. Islets were infected with a recombinant adenovirus that bicistronically encodes the dual-reporter genes luciferase and green fluorescence protein (Adv-luciferase) under control of the CMV promoter at a multiplicity of infection (MOI) of 1000 for 16 hr using techniques previously described [20]. Luciferase activity of extracts from islets transduced with Adv-luciferase was measured in a Pharmingen Monolight 3010 Luminometer.

Islet Transplantation

Murine islets were transplanted beneath the renal capsule or infused into the portal vein of NOD-SCID mice as previously described [19]. Briefly, the islet suspension was injected in a 30-μL volume just beneath the renal capsule with a 23-gauge butterfly needle. The needle was withdrawn and the insertion point was cauterized. For the hepatic transplants, islets were infused into the portal vein via PE10 tubing attached to a 30-gauge needle. Slight pressure was applied to the insertion point to stop blood loss. Incisions were closed with black subcutaneous sutures (Prolene, Ethicon, Somerville, NJ) and aluminum skin staples (Autoclips, 9 mm size, Clay Adams, Parsippany, NJ). Clips were removed 6 days after surgery.

Luminescent Beads

As a surrogate for luciferase-expressing islets, we used luminescent beads (Mb-Microtec, Bern, Switzerland) that consisted of glass capillaries (0.9 mm diameter and 2 mm long). These beads are filled with tritium (a β-emitter with a half life of over 10 years) that excites a phosphor and isotropically emits constant intensity light (Figure 2A and B). The spectral emission of these beads was measured using a fiber-optic probe attached to a spectrometer (Ocean Optics, Dunedin, FL) equipped with a 360-nm cutoff filter. Spectral emission of a single bead within a centrifuge tube was measured with 1-sec integration time. An attached laptop computer was used to record the spectral data. To implant a luminescent bead in an NOD-SCID mouse, an incision was made above either the kidney or liver, as described previously for the islet transplantations [19]. The luminescent bead was glued onto the kidney or liver at a site that was deemed anatomically similar to the site of islet engraftment using Vetbond tissue adhesive (3M, St. Paul, Minnesota). The incision was closed with subcutaneous sutures and skin staples as previously described [19].

Bioluminescence Imaging

For luminescence imaging of islets and beads, we used a liquid nitrogen cooled, back thinned, back illuminated CCD camera with a 1300 × 1340 pixel chip (EEV 1300 series, Roper Scientific, Trenton, NJ). Prior to luminescent imaging, a black and white image of the field of view was taken to allow correlation of the bioluminescent signal to anatomical sites on the animal. A 1-msec background image (shutter closed) was taken prior to each bioluminescence image. Background subtraction was performed on all images. Metamorph software (Version 4.6r6, Universal Imaging, Downingtown, PA) was used to analyze the bioluminescence image with peak intensity (regions of equal area were drawn around the region of interest [ROI]). On chip binning of 5 was used for imaging to increase signal to noise ratio. Pixel intensities within the ROI were summed to yield integrated intensity of luminescence.

Islet Bioluminescence Imaging

For in vitro imaging, the luciferase substrate d-Luciferin (Biosynth International, Naperville, IL) was added in excess (10 μL of 0.15 mg/mL concentration) to murine islets in six-well plates with 100 μL of phosphate-buffered saline (PBS). After placing the plate in the imaging chamber, bioluminescence was imaged with a 4-min exposure taken with the CCD camera. Light emission was integrated from 2 to 6 min after luciferin addition to capture peak bioluminescence activity. To image mice bearing transplants of luciferase-expressing pancreatic islets, the hair overlying the islet graft was shaved in the anesthetized mice (intraperitoneal injection of 50 mg/kg body weight sodium pentobarbital). The substrate d-luciferin, dissolved in sterile deionized water, was injected intraperitoneally (150 mg/kg body weight). Mice were gently secured to a black felt pad to minimize any motion artifacts in lateral decubitus orientation (graft facing up) for renal grafts and supine orientation for hepatic grafts. This pad was then placed in the light-tight imaging chamber. Bioluminescence images were taken with 4-min integration time. Luminescence emission was found to peak/plateau approximately 8 min postsubstrate administration, hence, images used for quantification was taken from approximately 6 min postluciferin administration to 10 min postinjection.

Imaging of Luminescent Beads

Prior to implantation, a black felt pad holding the luminescent beads was imaged with the CCD camera. Beads were first imaged with a 1-sec exposure time and then reimaged four times, reorienting the bead between each image, in order to quantify variability in bead light emission. Bead luminescence was quantified using circles of equal area drawn around the ROI. Luminescence was quantified by summing pixel intensities within the ROI to yield integrated intensity.

Imaging of mice with an implanted bead was performed as described above for mice bearing islet transplants except no luciferin was injected. A 1-sec exposure was taken for all images. Background subtraction was performed and implanted bead luminescence was quantified by photon counting of the ROI. Spot size was determined using Metamorph. The maximum intensity pixel from each bead was found and measured. A threshold was then applied at half the maximum intensity for all pixels above that value. The number of pixels exceeding threshold was determined. This area is the spot size; full width at half maximum (FWHM) was calculated as the diameter of this circular spot size.

$\text{FWHM}=\sqrt{\frac{4{.\text{Area}}_{\text{threshold}}}{\pi }}$

(1)

Rotational Variability Study

The effect of mouse rotation (relative to the imaging axis) on the measured photon flux from an implanted bead was measured using a rotational stage (Figure 4A and B). This stage consisted of a hinged black felt platform that allowed 50° rotation in either direction. An anesthetized mouse with a luminescent bead was placed on the stage with the mouse gently secured in position with a black strap; the stage was then placed in the light-tight imaging box of the CCD camera. Mice with the bead on the renal capsule were placed in a lateral decubitus position (bead facing up); mice with a hepatic bead were placed in the supine position. The stage was rotated in 10° increments from −50° to 50° (with 0° indicating parallel to the floor). Positive rotation was defined as clockwise rotation when viewed from the head of the mouse. Thus, for renal beads, positive rotation was defined as rotation toward the prone orientation, whereas negative rotation indicated rotation towards the supine orientation. One-second camera exposures were taken at each angle. Luminescence was quantified using Metamorph's photon counting ROI analysis, as previously described. Luminescence at each angle was normalized to the measured luminescence at 0°.

Monte Carlo Simulation

Monte Carlo simulation was used to model the propagation of photons from the luminescent bead to the aperture of the CCD camera. Monte Carlo simulation provides a well-accepted numerical simulation of light transport in multilayer tissues close to tissue boundaries. Monte Carlo simulation has been used to model light propagation in a variety of applications [21–23]. The Monte Carlo code used was based on the MCML code developed by Steven Jacques and others, modified for isotropic sources [24]. Transmission of light through the tissue was determined as a function of radial position; photon transmission reaching the camera aperture was calculated as the photon weight transmitted within the radius of the camera aperture. Monte Carlo simulation was run for three conditions: the bead alone, the bead implanted in the renal capsule, and the bead implanted beneath the liver (Figure 3). The bead simulation consisted of a single layer of nonabsorbing, nonscattering media corresponding to the air between the camera stage and the aperture. The renal bead simulation added a layer of skin to the air layer. The hepatic bead simulation added a layer of liver tissue to the skin and air layers for a three-layer model. Thickness of the tissue layers was determined by sacrificing the animal and measuring tissue thickness overlying the bead using calipers. Tissue optical properties used in the simulations were obtained from work by Cheong et al. [25]. The following optical properties were used in simulation: Air layer: n = 1, g = 1, μ_a = 10⁻⁹ cm⁻¹, μ_s = 0 cm⁻¹; Skin layer: n = 1.37, g = 0.9, μ_a = 10 cm⁻¹, μ_s = 20 cm⁻¹; Liver layer: n = 1.37, g = 0.9, μ_a = 9.6 cm⁻¹, μ_s = 89 cm⁻¹.

Results

Bioluminescence of Luciferase-Expressing Islets

Pancreatic islets transduced with an increasing MOI of Adv-luciferase expressed increasing luciferase activity (Figure 1A). For subsequent studies, an MOI of 1000 was used. Adenovirus infection of murine islets did not alter glucose-stimulated insulin secretion of islets in a cell perfusion system (data not shown). Luminescence of luciferase-expressing murine islets in culture was easily detected (Figure 1B) after addition of luciferin to the culture media. By imaging different numbers of islets transduced with a MOI of 1000, bioluminescence as assessed by the CCD camera correlated with luciferase activity in islet extracts as assessed with a luminometer [17]. Bioluminescence intensity increased linearly with the number of islets/well. Two different anatomical sites are typically used for murine islet transplantation. Islets transplanted beneath the renal capsule form an islet graft localized in a small area; this site is widely used in murine models of islet transplantation. However, liver transplantation is more applicable to clinical studies, as the liver is the site of human islet transplantations. Following an intraperitoneal injection of luciferin, bioluminescence emission was detected in mice bearing luciferase-expressing islets transplanted beneath the renal capsule or into the liver (Figure 1C).

Figure 1.

Luminescence of pancreatic islets expressing luciferase. (A) Luciferase activity of murine islets transduced with an increasing multiplicity of infection of Adv-luciferase (50 islets/point; expressed as number of viral particles/cell with the assumption of 500 cells/islet) was measured in a luminometer. (B) Luciferase-expressing murine islets in a six-well plate. Islets in quantities of 50, 100, and 200 islets from left to right; upper and lower rows duplicates. (C) Luciferase-expressing murine islets (100) transplanted beneath the renal capsule (left) or transplanted into the liver (right).

Luminescent Beads In Vitro

Light emission from the luminescent beads was quantified using a CCD camera and integrated photon measurements (Figure 2A and B). Multiple images of the same bead resulted in similar photon measurements with less than 3% variation from the mean (n = 10). Additionally, photon measurements of the same bead did not change over a period of several months. Luminescent beads provide a constant light source with a half-life of 10 years, hence, they can be considered isotropic and constant intensity emitters for the duration of the experiment. The spectral emission of these beads measured using a fiber-optic probe attached to a spectrometer was similar to that of light generated by the firefly luciferase reaction (Figure 2C).

Luminescent Beads In Vivo

Luminescent beads were implanted at the sites used for islet transplantation: the renal capsule and the liver (Figure 2D and E). The renal beads show a brighter, more concise region of luminescence. Light emission from the hepatic beads is less bright and spread over a larger area. Note the difference in scale between the two images.

Mice (n = 4) with luminescent beads implanted at the renal capsule were imaged weekly for 6 weeks postimplantation to determine the temporal variation in luminescence intensity (Figure 2F). Bead intensity is quantified as the ratio of bead light luminescence in vivo (implanted bead) to the constant light emission of the bead in vitro (preimplantation). At 1-week postimplantation, the mice with renal beads showed significantly lower luminescence than in later weeks (t test, α = 0.05). The detected luminescence increased twofold over the Week 1 measurement. Six weeks postimplantation, the ratio of implanted bead luminescence to preimplantation bead luminescence was 0.2394 ± 0.0261. The renal beads also showed the greatest mouse-to-mouse difference in detected light the week immediately postimplantation. By the second week postimplantation, the spread in detected light among different mice had decreased.

Figure 2.

(A) Luminescent beads shown next to a ruler (each mark corresponds to 1 mm). (B) A CCD camera image taken of four luminescent beads with a 1-sec exposure. (C) The emission spectrum of the luminescent bead (shown in blue) plotted with the emission spectrum of the luciferase reaction (shown in magenta). (D) Luminescent bead implanted beneath the renal capsule. (E) Luminescent bead implanted in the liver. (F) Quantification of luminescence from mice with renal or hepatic bead weekly postimplantation. Data are shown as the ratio of implanted bead luminescence to bead luminescence preimplantation. Each point represents an average of three to four mice. Error bars indicate the standard error of the mean.

Luminescent beads implanted beneath the liver showed lower light emission (Figure 2F). In the first week postimplantation, the detected luminescence was lower than in later weeks. At 6 weeks, the ratio of implanted bead intensity to preimplantation bead intensity was at a steady state value of 0.0645 ± 0.0140.

The luminescence from the renal and hepatic implanted beads was also analyzed for light scattering. Spot size and FWHM were determined each week postimplantation. There was no statistically significant change in spot size over the 6 weeks post-transplantation. The FWHM of the liver implanted beads was 17% larger than the FWHM of the renal beads. As in the intensity measurements, the greatest variability was seen a week following surgery.

Monte Carlo Simulation

Light propagation from the luminescent bead to the camera aperture was modeled using Monte Carlo simulation. Simulation was run for the bead alone, for the bead implanted on the renal capsule, and for the bead implanted beneath the renal capsule. Geometry and optical properties used in simulation are listed in the caption of Figure 3. The results of this simulation are in agreement with experimental values found using the constant light-emitting bead (Table 1).

Figure 3.

Geometry used for Monte Carlo simulation of light propagation from either the renal or hepatic bead to the camera aperture. Modeling of the bead alone included only the nonabsorbing, nonscattering air layer. The following optical properties were used in simulation: Air layer: n = 1, g = 1, μ_a = 10⁻⁹ cm⁻¹, μ_s = 0 cm⁻¹; Skin layer: n = 1.37, g = 0.9, μ_a = 10 cm⁻¹, μ_s = 20 cm⁻¹; Liver layer: n = 1.37, g = 0.9, μ_a = 9.6 cm⁻¹, μ_s = 89 cm⁻¹.

Table 1.

The Ratio of In Vivo to In Vitro Bioluminescence Intensity for Renal Transplantation Site (Left Column) and Hepatic Transplantation Site (Right Column)

	Renal In Vivo Intensity/In Vitro Intensity	Hepatic In Vivo Intensity/In Vitro Intensity
Monte Carlo Simulation	0.2744	0.0495
Beads	0.2394 ± 0.0261	0.0645 ± 0.0140
100 Islets	0.0476	0.0116
200 Islets	0.0284	0.0112

Rotational Variation

Figure 4 shows the luminescence intensity from the implanted beads as a function of rotational angle. As the mouse was rotated from a normal (i.e., surface of animal normal to the optical axis of the imaging system) position (lateral decubitus for renal beads, supine for hepatic beads), the luminescence intensity decreased. At small deviation from the optical axis, intensity decreased only slightly. But at 50° rotation from flat, luminescence intensity decreased to approximately one quarter (0.27) of the normal angle intensity for the renal bead (Figure 4D) and half (0.48) of the normal angle intensity for the hepatic bead (Figure 4C). The angular dependence was stronger for beads at the renal capsule site than for hepatic beads.

Bioluminescence of Islets in Vivo

Bioluminescence intensity correlates linearly with the number of islets transplanted [17]. As expected from the studies with the luminescent bead, light emission from hepatical transplanted islets is less than that from renally transplanted islets (Table 1). The spot size of hepatic islet grafts is greater than the spot size of renal islet grafts.

To investigate the survival of transplanted islets, we calculated the in vivo/in vitro luminescence ratio of the luminescent beads and the luciferase-expressing islets for both the hepatic and renal sites. Because the effects of the surrounding tissue on light transmission should be similar for beads and islets, the in vivo/in vitro luminescence ratio provides an estimate of the amount of luciferase expression that has survived post-transplantation. As mentioned above, the in vivo/in vitro luminescence ratio for hepatic bead implants was nearly four times lower than the ratio for renal bead implants (0.0645 ± 0.0140 vs. 0.239 ± 0.0261). Transplanted islets show a similar relationship, with higher light emission from renal islet grafts than hepatic grafts. However, the in vivo/in vitro luminescence ratio for transplanted islets was markedly less than the ratio for implanted beads at both the renal and hepatic sites (Table 1). This suggests that only a minority of the luciferase-expressing islets survive after transplantation.

Figure 4.

(A) Image of a mouse with an implanted hepatic bead. The mouse was secured to the rotational stage. (B) Schematic of the rotational stage, showing the relationship between the rotational angle of the stage and the angle between the camera optical axis and surface normal of the mouse. (C) Change in hepatic bead luminescence intensity with angle of rotation. Bioluminescence intensity at each angle is normalized to normal position (θ = 0°) (supine for hepatic implantation). Shown is the mean of four mice; error bars indicate the standard error of the mean. (D) Change in renal bead luminescence intensity with angle of rotation. Bioluminescence intensity at each angle is normalized to normal position (θ = 0°) (lateral decubitus for renal implantation). Shown is the mean of four mice; error bars indicate the standard error of the mean.

Discussion

In this report, we characterized imaging parameters that must be understood and accounted for to accurately quantify BLI and applied these findings to BLI of transplanted islets. Bioluminescence from islet grafts was simulated by implanting a constant light-emitting bead, allowing investigation of important imaging parameters, including the effect of geometric features and tissue optical properties on bioluminescence measurements. Using this constant light-emitting source, we have shown that postsurgical effects (i.e., wound healing), animal positioning, and the anatomic location of the light source influence bioluminescence measurements.

A major parameter of interest when tracking transplanted islets is the number of islets surviving post-transplantation. The relationship between number of bioluminescent islets and emitted light intensity could be useful for monitoring the survival of transplanted islets. However, a number of biologic variables and imaging parameters influence bioluminescence. For example, the measured light intensity depends on the transduction efficiency of the adenovirus, the size of the islets, and various other geometric and biological factors such as thickness of overlying tissue (and changes in tissue thickness due to for example wound healing and/or scar formation), optical properties of tissue, perfusion/hemoglobin content, and substrate availability. The implanted beads mimic bioluminescence by emitting a constant and known intensity light from the site used for islet transplantation. The emission spectrum of the beads closely mimics that of firefly luciferase induced bioluminescence (Figure 2C). The peak bead luminescence (600 nm) is slightly higher than the peak emission of the luciferase reaction (563 nm). However, studies indicate that the firefly luciferase reaction shows temperature dependence, with peak emission at 590 nm at physiological temperatures (personal communication with Dr. Brad Rice, Xenogen, Alameda, CA). The strategy of measuring light emission from the beads in vivo allows us to distinguish changes in measured bioluminescence owing to actual changes in islet light emission from changes caused by other factors that may influence light transport in the animal and toward the imaging detector.

Light transmitted to the CCD camera is dependent on transplantation site. Use of the constant light-emitting bead implanted at a transplantation site allowed quantification of light attenuation at each implantation site. Comparison of renally implanted beads to hepatical implanted beads showed a marked difference between light transmission from each site. Bioluminescence signals collected from renal implants are brighter and spatially more concise than signals from implants in the liver. Monte Carlo modeling of light propagation supports the finding that light is attenuated four times more strongly from the hepatic site. The renal capsule implantation site is more superficial; the mean optical path length for light emitted from this site is shorter than that for light emitted from the liver. Light emanating from the hepatic transplants must pass through liver tissue, a highly perfused tissue that contributes significant light scattering and absorption, resulting in lower light emission and increased spatial spread of light at the skin surface.

The spot size of luminescence differed between hepatic islets and renal islets. Islets infused into the portal vein are thought to embolize throughout the liver, leading to a more diffuse light source. Renally transplanted islets are typically confined to a contiguous and highly localized graft. However, spot size is also a function of the tissue optics through which propagating photons pass. Bioluminescence from the liver passes through hepatic tissue, leading to increased scattering and thus a larger spot size than renal grafts. The extent to which increased hepatic light scattering leads to larger spot size was analyzed using the mice with implanted beads. The FWHM for the hepatic beads was 17% higher than the renal bead implants. The islet transplants showed a much greater difference in spot size between hepatic and renal sites than the bead implantations. This indicates that hepatic islet dispersion is likely the primary reason for larger hepatic islet spot size. Intrahepatic islets are more spatially dispersed throughout the organ than those beneath the renal capsule where the graft is clustered in a contiguous area.

The luminescence from implanted beads was tracked temporally to investigate postsurgical effects on light transmission. Our results with the light-emitting beads demonstrate that the effect of the inflammatory response on bioluminescence measurements is dynamic as the wound healing progresses. The metal clips interfered with imaging less than 1-week postsurgery by blocking underlying light emission, preventing accurate measurement of bioluminescence. After surgical clip removal, tissue changes at the surgical site also effected light emission for 1–2 weeks postimplantation before stabilizing to consistent measurements in later weeks. Both the renal and hepatic bead implants showed less luminescence the first week post-transplantation, with the renal implants showing the greatest variation in light transmission 1 week post-op. This finding emphasizes that the surgical site may induce imaging artifacts that must be considered when BLI is used to quantitatively assess biological processes. Lu et al. [16] found that islet bioluminescence decreased considerably the week immediately following surgery. It was postulated that insufficient vascularization of the islets 1-week post-op could hamper delivery of the substrate luciferin to the islet graft. Thus, dynamic changes in light transport by surgery-related edema, angiogenesis, or scar tissue overlying the bioluminescent source or even the presence of tissue ridges and/or sutures or clips at the suture site can significantly decrease bioluminescence measurements by increasing optical path length, resulting in light quantification that does not accurately reflect light emission from the bioluminescent source. For accurate quantitative use of bioluminescence as a surrogate marker for biological processes, whether by islet mass, tumor volume, gene transcription, or one of many other uses of this technology, these effects must be taken into consideration.

Bioluminescence measurements can also be affected by rotation of the sample. Detected photon emission depends on the angle the surface normal makes with the optical axis of the camera system. The practical implication of this finding is that the exact positioning of the mouse relative to the camera axis is an important parameter that must be carefully controlled. Failure to do so can induce variations in the photon counting measurements that are not representative of the biological processes for which bioluminescence is the surrogate marker. The application of tomographic analysis to BLI, thus far still in a developmental state, must take into account these rotational effects. The effect of rotation on luminescence measurements was determined for both the renal and hepatic implanted beads. Both implantation sites showed a parabolic relationship between angle and luminescence, with luminescence decreasing with an increasing angle from normal. At rotations of 50° from normal, measured luminescence decreased to approximately a quarter compared to the normal measurement for renal beads. For hepatic beads, the decrease at 50° from normal was less, resulting in measured intensity less than half of the normal measurement. The drop in renal intensity was sharper and more pronounced than that seen from the hepatic bead. Supine orientation served as normal placement for hepatic bead imaging, whereas lateral orientation was used for renal imaging to minimize the optical path length of light through tissue. The animal surface overlying the liver bead thus has more surface area normal to the camera axis than the animal surface overlying the renal bead. This, in turn, results in a slower decrease in luminescence with increasing rotation for the hepatic implantation.

Light emitted from islet grafts is significantly lower than light emission from islets in vitro. However, it was unclear how much of this decrease in luminescence was due to islet death post-transplantation and how much could be attributed to light attenuation by tissues overlying the implants. The light-emitting beads transplanted at the site of islet grafts were used to quantify light transmission through the tissues overlying the luminescent source. For renal implantation, the ratio between luminescence after implantation (in vivo) to luminescence of the bead alone (prior to implantation) stabilized at a constant value 2 weeks postimplantation. Experimental results for both renal and hepatic implanted beads are in agreement with Monte Carlo modeling of photon propagation. Light attenuation by tissue overlying hepatic transplants results in a nearly fourfold less light transmission from hepatic islets than renal islets. This is in agreement with the ratio found using the constant light-emitting bead, suggesting that the survival rate of islets at the renal and hepatic site is similar. Interestingly, the ratio between in vivo and in vitro luminescence of the beads was much greater than the ratio for islets, suggesting a large (six- to eightfold) drop in viable islets post transplantation (Table 1). This conclusion assumes that the BLI of islets reflects islet survival of luciferase-expressing islet cells. However, bioluminescence measurements could be affected by loss of the luciferase transgene and differences in substrate availability. Additional work is needed to conclusively prove that transplanted islet mass, assessed by morphologic and histologic approaches, correlates with BLI.

Bioluminescence measurements are subject to some inherent limitations. In vivo BLI can be applied only to small animal models; the tissue thickness of larger animal models or humans currently prevents BLI studies. Additionally, islets transduced with our adenovirus constitutively express the reporter gene luciferase in all islet cell types, not just β-cells. This limitation can be circumvented by the using a β-cell-specific promoter or by creating transgenic islets that express luciferase solely in the β-cells. The adenovirus system used for islet transduction also has a limitation in that dividing cells the transgene is passed to only one of the daughter cells and thus β-cell replication post-transplantation would not be reflected in bioluminescence measurements. As reported by Lu et al. [16], the use of lentivirus-mediated gene transfer may allow for integration of the luciferase DNA into the genome before cell division but typically has much lower transfection efficiency compared to the adenovirus.

The correlation of light emission to islet number in vivo enables the use of BLI as a noninvasive means of islet assessment in mouse models of diabetes and islet transplants. In turn, these models may be used as models for assessing and screening of novel therapeutic approaches to improve islet survival. Information regarding relative islet mass or number as function of time or in response to pharmacological intervention should be readily obtainable from the bioluminescence measurements (i.e., increase or decrease of signal). Absolute information regarding the number of islets surviving requires additional investigation correlating islet histology with BLI using the imaging parameters optimized in this study. In conclusion, BLI is shown to be a valuable method to assess transplanted islet mass in vivo. However, we have shown using our luminescent bead studies that postsurgical dynamics, animal positioning, and light source location can significantly alter the measured bioluminescence. Clearly, these findings are not unique to the application of transplanted islet imaging, but hold important clues for the field of BLI in general. Our findings suggest that for accurate use of BLI as a quantitative surrogate marker for biological processes, detailed and careful system characterization and calibration is required.