Original Article - diabetes.diabetesjournals.org · In Vivo Imaging of Immune Rejection in Transplanted Pancreatic Islets Natalia V. Evgenov, 1Zdravka Medarova, John Pratt,1 Pamela

Original Article

In Vivo Imaging of Immune Rejection in TransplantedPancreatic IsletsNatalia V. Evgenov,

1Zdravka Medarova,

1John Pratt,

1Pamela Pantazopoulos,

1Simone Leyting,

1

Susan Bonner-Weir,2

and Anna Moore1

As islet transplantation becomes an acceptable clinicalmodality for restoring normoglycemia in type 1 diabeticpatients, there is a crucial need for noninvasive assessmentof the fate of the grafts. In spite of the success of theEdmonton Protocol, a significant graft loss occurs due toimmunological and nonimmunological events immediatelyafter transplantation. Allogeneic rejection in graft recipi-ents is one of the major reasons for islet death and graftfailure. Therefore, monitoring the islet rejection usingreliable noninvasive methods would significantly aid inclinical assessment of graft success. We have previouslydeveloped a method to detect transplanted islets noninva-sively using magnetic resonance imaging (MRI). For thisprocedure, human pancreatic islets are labeled with an MRIcontrast agent that enables their visualization on magneticresonance images. In our present study, we not only de-tected labeled human islets in a preclinical intrahepaticmodel of human islet transplantation in mice but alsoshowed that islet rejection can be monitored noninvasivelyand repeatedly in real time by MRI. In addition, in thisstudy, we have adapted, for islet cell labeling, a Food andDrug Administration–approved commercially availablecontrast agent, Feridex, that is used clinically for liverimaging. We believe that this agent, in combination withour preclinical model of islet transplantation, will facili-tate the transition of imaging immune rejection to clinicaltrials. Diabetes 55:2419–2428, 2006

Pancreatic islet transplantation has recentlyemerged as an effective therapy and potentialcure for patients with type 1 diabetes (1–3).Studies in experimental models and recent clini-

cal trials have shown that islet transplantation may repre-

sent a superior alternative to insulin injections sincetransplantation, unlike insulin therapy, results in normal-ization of metabolic control (4). For clinical transplanta-tion, islets are infused in the liver through the portal veinwhile the subject is under local anesthesia. For an average-size person (70 kg), a typical transplant requires about 1million islets, extracted from two donor pancreases (5).Allogeneic human islet rejection and/or recurrence ofautoimmunity are the two major challenges hamperingsuccessful transplantation. Significant improvement in thisarea was achieved by researchers from the University ofAlberta, who introduced an improved antirejection regi-men, which became known as the Edmonton Protocol (1).In a 5-year follow-up report from this group, it is empha-sized that the results, though promising, still point to theneed for improving islet engraftment and preserving isletfunction (6).

It is clear at this point that immunological and nonim-munological events, which take place immediately aftertransplantation, lead to significant graft loss. Conse-quently, the fate of islets after transplantation is mostlyunknown. In spite of the sophisticated immunosuppres-sion regimen, a significant number of pancreatic islets dieduring the first 10–14 days after the procedure. In fact,even in syngeneic models of islet transplantation, a 60%loss of transplanted islets was reported in this earlypostprocedure period (3). The causes for islet death,besides immune rejection, are numerous and includemechanical injury, ischemia, nonalloantigen-specific in-flammatory events in the liver after transplantation, andrecurrent autoimmunity. Mediators of islet dysfunctioninclude the inflammatory cytokines interleukin-1�, inter-feron-�, and tumor necrosis factor-� (7–9). Islets exposedto allogeneic blood undergo instant blood-mediated in-flammatory reaction elicited by tissue factor, which isproduced by the endocrine cells and involves coagulationand complement activation (10–12). In addition, there isevidence of “metabolic” rejection of intraportal islet trans-plants involving lipotoxic destruction of islets by hepaticlipids (13,14). Furthermore, elevated levels of apoptosishave been shown in pancreatic islets exposed to chronichyperglycemia immediately after transplantation (15). Atpresent, however, immune rejection of pancreatic isletspresents the biggest challenge in transplantation. A con-siderable fraction of islet transplants undergo chronic failuredue to immune rejection, despite the novel use of rapam-ycin-based immunosuppressive agents and other means ofprevention of allogeneic human islet rejection (4). Regard-less of the specific immunosuppressive strategy used aftertransplantation, there is a critical need for monitoring isletrejection using reliable noninvasive methods.

From the 1Department of Radiology, Molecular Imaging Laboratory, Massa-chusetts General Hospital/Massachusetts Institute of Technology/HarvardMedical School Athinoula A. Martinos Center for Biomedical Imaging, Massa-chusetts General Hospital and Harvard Medical School, Boston, Massachu-setts; and the 2Joslin Diabetes Center, Harvard Medical School, Boston,Massachusetts.

Address correspondence and reprint requests to Anna Moore, PhD, Molec-ular Imaging Laboratory, MGH/MIT/HMS Athinoula A. Martinos Center forBiomedical Imaging, Department of Radiology, Massachusetts General Hos-pital/Harvard Medical School, Rm. 2301, Bldg. 149, 13th St., Charlestown, MA02129. E-mail: [email protected].

Received for publication 11 April 2006 and accepted in revised form 16 June2006.

Z.M. and J.P. contributed equally to this work.Additional information for this article can be found in an online appendix at

http://diabetes.diabetesjournals.org.FDA, Food and Drug Administration; MRI, magnetic resonance imaging;

TUNEL, transferase-mediated dNTP nick-end labeling.DOI: 10.2337/db06-0484© 2006 by the American Diabetes Association.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked “advertisement” in accordance

with 18 U.S.C. Section 1734 solely to indicate this fact.

DIABETES, VOL. 55, SEPTEMBER 2006 2419

In vivo magnetic resonance imaging (MRI) is a powerfulmodality that allows for acquiring images at near-micro-scopic resolution with simultaneous detection of anatom-ical and physiological parameters (16). It can provideimportant information regarding graft location and sur-vival after transplantation. Recently, we have reported onthe in vivo imaging of human pancreatic islets labeled witha magnetic resonance contrast agent and transplantedunder the kidney capsule in a mouse model of type 1diabetes (17). We showed that, using this model, labeledpancreatic islets could be reliably detected in vivo by MRI.Furthermore, we demonstrated that labeled islets werecapable of restoring normoglycemia in diabetic animalswith the same efficiency as their unlabeled counterparts.The basis for our present report builds upon our previousfindings with the ultimate goal of translating this methodinto clinical practice. Therefore, the goal of this study wastwofold. First, for clinical application, we needed to adoptthe clinically relevant model of intrahepatic transplanta-tion and demonstrate the possibility of in vivo detection oftransplanted islets in this organ. Second, using this model,we intended to detect an input of immune rejection aftertransplantation in immunocompetent animals by in vivoMRI. As mentioned above, there are multiple reasons forislet death after transplantation. Here, we specificallyfocused on immune rejection since it presents the biggestchallenge in clinical islet transplantation (4). In our previ-ous work, we used superparamagnetic iron oxide nano-particles synthesized in our laboratory as a magneticresonance contrast agent for islet labeling (17). Here, weutilized the Food and Drug Administration (FDA)-ap-proved commercially available contrast agent Feridex,which is routinely used in clinics for liver imaging. Weadapted it for islet labeling, which we believe wouldfacilitate the transition of this imaging method into clinicalpractice.

As a result of this study, we demonstrated that intrahe-patic islet transplantation could be visualized using in vivoMRI in immunocompetent and immunocompromisedmice. Furthermore, islet death was detected by MRI inboth models immediately after transplantation. However,there was a clear acceleration of this process in immuno-competent animals pointing to severe immune rejection.We anticipate that these studies could make a significantimpact on allotransplantation and the development of newand improved immunosuppressive regiments.

RESEARCH DESIGN AND METHODS

Labeling human pancreatic islets with an FDA-approved commercially

available contrast agent. Human pancreatic islets were obtained throughThe Islet Cell Resource Center at the National Institutes of Health and the IsletDistribution Program at the Juvenile Diabetes Research Foundation. Isletswere shipped overnight to Massachusetts General Hospital, handpicked andcultured in Miami Medium #1 culture media (Cellgro; Mediatech, Herndon,VA), and supplemented with 20 mg/l ciprofloxacin hydrochloride (FisherScientific, Pittsburgh, PA) and 10 mg/l L-glutathione (Sigma, St. Louis, MO).

For labeling experiments, we used superparamagnetic iron oxide nanopar-ticles Feridex (ferumoxide; Advanced Magnetics, Cambridge, MA), which areFDA approved for human use as a liver imaging contrast agent. Here, weutilized it for labeling human pancreatic islets. One thousand human pancre-atic islets were incubated overnight with Feridex in culture medium (200 �giron/ml). After incubation, islets were washed with culture medium threetimes and used for in vitro studies or islet transplantation.

To calculate the amount of probe associated with the islets, we utilized aniron binding assay. One hundred pancreatic islets were incubated overnightwith increasing concentrations of Feridex (10–300 �g iron/ml). After incuba-tion, islets were washed three times with PBS, resuspended in 6N HCl, andincubated at 70°C for 30 min. The iron content of labeled islets was

determined using a total iron reagent set (Pointe Scientific, Canton, MI).Incubation with each iron concentration was repeated five times.Insulin secretion and viability of Feridex-labeled islets. Insulin secretionwas evaluated using static incubation of labeled and nonlabeled islets at lowand high glucose concentrations. Briefly, islets were washed in CMRL 1066media without glucose (Cellgro), supplemented with 0.5% BSA, radioimmu-noassay grade (Sigma), 25 mmol/l HEPES, and 1.7 mmol/l glucose andpreincubated for 60 min in the same medium. The medium was then discardedand replaced by fresh medium containing 2.5 mmol/l glucose for 1 h for basalsecretion, followed by an additional 1-h incubation at 16.7 mmol/l glucose.Supernatants were collected and frozen for insulin assays. Thereafter, isletswere washed with PBS and extracted with 0.18 N HC1 in 70% ethanol for 24 hat 4°C. The acid-ethanol extracts were collected for determination of insulincontent and to normalize the values measured for insulin secretion. Insulinconcentration was measured using a human insulin enzyme-linked immu-nosorbent assay kit (Mercodia, Uppsala, Sweden). A stimulation index wascalculated as the ratio of stimulated to basal insulin secretion normalized bythe insulin content. Nonlabeled islets from the same batch served as controlsin all experiments. Results are representative of three separate experiments.

Islet viability after labeling was assessed by a standard 3-[4,5-dimethylthia-zol-2-yl]-2,5-diphenyltetrazolium bromide assay (Promega, Madison, WI) andcompared with that of unlabeled islets.

To measure apoptosis in Feridex-labeled and unlabeled islets, we used afluorometric caspase-3 assay kit (Sigma). Human pancreatic islets wereincubated with 200 �g iron/ml Feridex for 0, 1, 2, 4, 6, and 24 h, lysed, andassayed according to the manufacturer’s protocol. Incubation with 1 �g/mlstaurosporine served as a positive control.In vitro imaging of Feridex-labeled islet phantoms. To show that labelingwith Feridex produces a change in signal intensity on T2-weighted images, weperformed MRI of islet phantoms. Labeled and nonlabeled pancreatic isletswere placed in Eppendorf tubes (200 islets/tube) and allowed to settle on thebottom by gravity sedimentation. MRI was performed using a 9.4T Brukerhorizontal bore scanner (Billerica, MA) equipped with ParaVision 3.0.1software. Acquisition of T2-weighted spin echo pulse sequences was based onthe following protocol: TR/TE � 2,000/15, 30, 45, 60, 75, 90, 105, 120 ms; FoV �4 � 4 cm2; matrix size � 256 � 256; resolution � 156 � 156 �m; slicethickness � 1 mm; and imaging time of 8 min 32 s.Mice and islet transplantation procedure. All animal experiments wereperformed in compliance with institutional guidelines and approved by thesubcommittee on research animal care at the Massachusetts General Hospital.

Pancreatic islet transplantation was performed using immunocompro-mised (NOD.scid, n � 12) and immunocompetent (Balb/C, n � 8) mice. Onethousand Feridex-labeled human islets were infused through the portal vein inanesthetized animals. Control animals were transplanted with nonlabeledislets.In vivo MRI of transplanted islets. In vivo MRI was performed using a 4.7TBruker horizontal bore scanner (Billerica, MA) equipped with ParaVision 3.0.1software. Imaging was performed on days 1, 2, 3, 4–5, 6–7, 10–11, and 14–15after transplantation The imaging protocol consisted of T2*-weighted gradientecho pulse sequences with the following parameters: TR/TE � 200/8 ms,number of averages � 32, FOV � 3.2 � 3.2 cm2, matrix size � 256 � 256,resolution � 0.125 � 0.125 mm2, slice thickness � 0.5 mm, and a total scantime of 27 min 18 s.

Labeled pancreatic islets (or islet clusters) appeared as distinct dark signalvoids in the liver. To compare islet death in immunocompetent and immuno-compromised animals, these signal voids were manually scored in the livers ofall imaged animals (13 slices each) for each day of imaging. On every day ofimaging, after each MRI session, we killed one of the imaged mice to use forex vivo histology. This study was performed by two independent investigators(N.V.E. and S.L.). The number of islets on day 1 was considered 100%. As aresult, our data are based on time-course longitudinal measurements of thesame set of animals, where this set was reduced by one on each subsequentday of imaging.Immunohistochemistry. To analyze our imaging findings at the microscopiclevel, we excised livers from the animals at each day of imaging. Mouse liversamples were embedded in Tissue-Tek OCT Compound (Sakura Finetek,Tokyo, Japan), snap frozen in liquid nitrogen, and axial sections prepared ata thickness of 7 �m.

The in vivo immune rejection of transplanted islets was evaluated semi-quantitatively by immunohistochemical staining of liver sections with anti-bodies targeting various types of immune cells. Cryosections were fixed in 4%paraformaldehyde and incubated with primary antibody for 1 h followed bybiotinylated rabbit anti-rat IgG (H�L) secondary antibody (Vector) andVectastain Elite ABC kit (Vector). Bound peroxidase was developed with a3,3�-diaminobenzidine SIGMAFAST kit (Sigma). Some sections were pro-cessed for Prussian Blue staining to identify the presence of iron in trans-planted islets. Briefly, sections were immersed in Prussian Blue solution

IN VIVO IMAGING OF REJECTED PANCREATIC ISLETS

2420 DIABETES, VOL. 55, SEPTEMBER 2006

containing 5% potassium ferrocyanide (ACROS Organics, Fairlawn, NJ) and5% hydrochloric acid (Aldrich, Milwaukee, WI) for 30 min and counterstainedwith nuclear fast red (Sigma-Aldrich). Other sections were counterstainedwith hematoxylin. The primary antibodies used for this study are summarizedin Table 1.

To determine which types of cells were labeled with Feridex in the islet, weperformed colocalization studies by staining for islets hormones, residentmacrophages, and iron nanoparticles. For these studies, we used mouseanti-human CD68 (Y1/82A) antibody (BD Pharmingen, San Diego, CA), rabbitanti-human insulin polyclonal antibody (Santa Cruz Biotechnology, SantaCruz, CA), rabbit anti-porcine glucagon antibody, and rabbit anti-humansomatostatin polyclonal antibody (MP Pharmaceuticals, Irvine, CA), followedby corresponding secondary biotinylated goat anti-mouse or anti-rabbit IgG(H�L) (Vector). Prussian Blue staining was performed as described above.In situ apoptosis detection. To evaluate levels of apoptosis in transplantedpancreatic islets, we performed a terminal deoxynucleotidyl transferase-mediated dNTP nick-end labeling (TUNEL) assay on frozen liver sections(Apoptag Fluorescein In Situ Apoptosis Detection kit; Chemicon Interna-tional, Temecula, CA) according to the manufacturer’s protocol. The nucleiwere counterstained with VECTASHIELD Mounting Medium with DAPI(Vector), and slides were examined using a Nikon Eclipse 50i fluorescencemicroscope equipped with an appropriate filter set (Chroma Technology,Rockingham, VT). Images were acquired using a charged-coupled devicecamera with near-infrared sensitivity (SPOT 7.4 Slider RTKE; DiagnosticInstruments, Sterling Heights, MI) and analyzed using SPOT 4.0 advancedversion software (Diagnostic Instruments). The percentage of cells undergo-ing apoptosis within each islet was calculated from the number of islet nucleipositive for DNA fragmentation versus the total number of cells present asobserved by two independent investigators (N.V.E. and S.L.).Statistical analysis. The data from the iron internalization assay wereexpressed as means SD. The results from the time course of the ironinternalization assay were analyzed by ANOVA. For the MRI data, the numberof dark signal voids in the livers of immunocompromised and immunocom-petent transplanted animals was compared for each day of imaging by a pairedStudent’s t test (SigmaStat 3.0; Systat Software, Richmond, CA). A value ofP 0.05 was considered statistically significant.

RESULTS

Human pancreatic islet labeling with the FDA-ap-proved contrast agent Feridex. To test whether humanpancreatic islets can be labeled with Feridex, we incu-bated them with the contrast agent and subjected them toMRI. We compared labeled and unlabeled islets and founda significant decrease in signal intensity on T2-weightedimages in labeled islets (Fig. 1A). By contrast, we observedno differences in signal intensity when we imaged the layerof medium directly above the islets. These results indicatethat the signal indeed originated from Feridex-labeledhuman pancreatic islets and that the labeling produced achange in signal intensity on T2-weighted images. Ironuptake by islet cells resulted in 1.21 0.36 to 18.26 1.36pg iron/islet cell, considering that an average pancreaticislet consists of 2,000 islet cells (Fig. 1B). As we showlater, iron distribution within islet cells was not uniform,and these numbers reflect only average iron accumulation.These numbers were similar to those obtained with the“in-house” made nanoparticles used in our previous study(17) and to the reported levels of nonspecific iron uptakeby other specialized cells (19–22). Time course studies

showed an overall increase in the islet iron content withincreasing incubation time, with a maximum at 12 h (P 0.001) (Fig. 1C). The results of the retention assay indi-cated that there was no iron loss in labeled islets in culture(Fig. 1D). We did not attempt to investigate iron retentionfor longer periods of time since we generally receive isletsbetween 3 and 4 days of culture and have, on average,another 3 days to perform experiments while they are stillviable. However, it is our experience that labeled isletsretained their label in vivo for up to 188 days aftertransplantation under the kidney capsule (17).

Preservation of islet viability and function after labelingwith Feridex is crucial for future clinical transplantation.Therefore, we assessed islet viability after overnight expo-sure to the contrast agent and found that it remainedunchanged compared with nonlabeled cultured islets (P �0.05). In addition, there was no elevated level of apoptosisin pancreatic islets after 24 h exposure to Feridex com-pared with nonlabeled cultured islets (online appendixFig. 1 [available at http://diabetes.diabetesjournals.org]).Glucose-stimulated insulin secretion, as seen by a stimu-lation index, was unchanged in labeled versus nonlabeledislets (5.9 2.0 vs. 6.2 2.0, respectively; P � 0.05). Basedon these experiments, we concluded that labeling ofhuman islets with Feridex does not compromise theirviability and function.

Similar to our previous findings with related iron oxides(17), we observed that within the pancreatic islet, Feridexlocalized to all islet cells including �-, �-, and �-cells, aswell as to islet macrophages (Fig. 2A). Most of the labelwas associated with insulin-producing �-cells since thesecells represent the largest cell population in the islet.Feridex does not have any specificity toward islet cells,and cellular uptake is most likely associated with conven-tional fluid phase endocytosis. These findings are in agree-ment with previous results regarding iron uptake byvarious cell types shown by us (17) and other investigators(20,23–25). Labeling of pancreatic islets with Feridex hada heterogeneous pattern, and some of the islets labeledmore efficiently than the others with overall efficiencyranging from 10 to 70% (Fig. 2B). These numbers representonly a histological cross section of a particular islet andmight not reflect an overall volumetric accumulation of theiron. In addition, we did not find any correlation betweenislet size and labeling efficiency. It seems that this processdepends more on the ”freshness“ of the particular isletisolation with islets being labeled more efficiently imme-diately to 2–3 days after isolation. Detailed investigationon this matter, however, is outside the scope of thismanuscript and is currently underway in our laboratory. Itis noteworthy that the average labeling efficiency achievedin this and similar studies was enough to detect a singleislet by MRI in islet phantoms (17).

TABLE 1Antibodies used for the immunohistochemical staining

Antigen Clone Isotype Specificity Dilution

CD4 YTS191.1* IgG2b Helper-inducer T-cells 1:50CD8 YTS169.4* IgG2b Cytotoxic/suppressor T-cells 1:50CD19 6D5* IgG2a B-cells 1:50CD68 FA-11* IgG2a Monocytes, macrophages 1:50Undefined membrane structure NIMP-R14† IgG2b Neutrophils 1:100

Source: *Serotec, Kidlington, U.K.; †Cell Sciences, Canton, Massachusetts.

N.V. EVGENOV AND ASSOCIATES


In vivo imaging of immune rejection after transplan-tation. To test the feasibility of in vivo MRI of intrahepatichuman islet grafts, we transplanted Feridex-labeled isletsin immunocompetent (Balb/c) and immunocompromised(NOD.scid) mice. We found that islets transplanted intothe liver appeared as dark hypointense foci representingsingle islets and/or, possibly, islet clusters (Fig. 3A),whereas nonlabeled islets did not cause any change insignal intensity on T2*-weighted images (Fig. 3B). Thesesignal voids were detectable in both models for theduration of the study, although the number of these voidschanged with time. To evaluate the number of islets/isletclusters in the liver, we manually scored the number ofsignal voids found on each slice. The validity of thismethod was first confirmed by creating a calibration curvewhere the known number of transplanted islets directlycorrelated to the number of dark voids representinglabeled islets/islet clusters found on T2*-weighted images(R2 � 0.997; Fig. 3D). Since islet death has been reportedin islet grafts during the first 10–14 days after transplan-tation (9,15,26,27), we performed MRI starting on the 1stday after transplantation. Visual coregistration of match-

ing slices on magnetic resonance images showed that thedisappearance of signal voids in Balb/c mice was morepronounced than in NOD.scid mice over time (Fig. 3C).Quantitative analysis of relative islet loss was performedfor immunocompromised and immunocompetent animalsbased on 13 slices covering the entire liver. This analysisshowed that the number of islets in both models started todecline immediately after transplantation. The rate of isletloss gradually decreased during the course of the studywith plateau between days 10 and 14 (Fig. 3E). Thisimmediate decrease is consistent with islet death aftertransplantation due to mechanical injury, ischemia, andnonantigen-specific inflammatory events (27). However, itis evident from Fig. 3E that immunocompetent miceexhibited a significantly higher rate of islet disappearanceon magnetic resonance images compared with immuno-compromised animals, especially pronounced on day 10,and resulting in a 20% difference in relative islet number by14 days after transplantation, presumably due to the inputfrom severe immune rejection. To correlate our MRI dataof islet death with apoptotic rates, we performed a TUNELassay on excised livers. In NOD.scid and Balb/c mice,

FIG. 1. A: Human islets labeled with Feridex produced a decrease in signal intensity on T2-weighted images. Unlabeled islets and the mediumabove islets did not show any darkening using the same magnetic resonance settings. B: Iron accumulation by islets labeled with Feridex. Isletswere incubated with increasing concentrations of the contrast agent. Iron uptake varied from 1.21 � 0.36 to 18.26 � 1.36 pg iron/islet cell forFeridex (assuming 2,000 cells/islet). C: Time course studies. Iron uptake by the islets increased as a function of time (P < 0.001). D: Retentionassay. After labeling, the iron content of human islets remained constant after incubation in culture for up to 24 h.



apoptosis was most pronounced on day 1 (12.1 and 18.6%,respectively). The number of apoptotic cells in immuno-deficient mice decreased to 1.8% by day 4 and stayed atthis level for the duration of the study. In contrast, thenumber of apoptotic cells in immunocompetent micegradually decreased to 3.3% on days 4–5 followed bysevere antigen-specific rejection, which destroyed up to13.2% of the islets on days 10–11 (Fig. 3F). The apoptosislevel gradually decreased by day 14 since by that time,almost all cells in the islets were destroyed. This rejectionwas evident from our correlative immunohistochemicalstudies that showed that the higher rate of islet deathdetected by MRI in immunocompetent mice was due tosignificant immune cell infiltration. As evident from Table2, this infiltration in Balb/c mice started in the first daysafter transplantation. The transplanted pancreatic islets in

these mice contained CD8� and CD4� T-cells, B-cells,macrophages (Fig. 4A), and neutrophils (Table 2). Thelevel of infiltration by these cells increased with time andpeaked on days 10–11 in agreement with the TUNEL assayresults. In contrast, immunocompromised animals showedonly marginal presence of immune cells in islets aftertransplantation except for CD68� invading macrophages,which are unaffected in animals with the scid mutation(28) (Fig. 4B). Evidently, in NOD.scid mice, islet deathoccurred mostly from the input from infiltrating macro-phages and from cell damage from mechanical injury,ischemia, and nonantigen-specific inflammatory events(27). Similar results were obtained by other investigatorsstudying immune rejection in allogeneic transplants indifferent species (29–31).

Overall, the results of this study suggested that the

FIG. 2. Feridex accumulation in human pancreatic islets. A: Light microscopy of islet sections stained for hormones and resident macrophagesshowing Feridex distribution in islet cells. Separate sections were incubated with anti-insulin, anti-glucagon, anti-somatostatin, and anti-CD68�antibodies followed by Prussian Blue stain. Feridex colocalized with insulin-producing �-cells and to a lesser degree, with � and � cells. Residentmacrophages showed an accumulation of Feridex (arrows). B: The accumulation of Feridex in human pancreatic islets was heterogeneous withrepresentative efficiency ranging from 10 to 70% (magnification bar � 10 �m).

TABLE 2Histochemical analysis of the islet infiltration by immune cells in immunocompromised and immunocompetent models

Days after treatment1 2 3 4 5 6 / 7 10 / 11 14 / 15

CD4�Balb/c /� � �� Scid

CD8�Balb/c � �� Scid /�

CD19�Balb/c � �� Scid /� /�

CD68�Balb/c � �� Scid � � � � ��

NeutrophilsBalb/c � � �� Scid /� /� � /� /� /�

Scoring: /�, at least one cell in the islet was positive; � to ��, increasing degree of infiltration.



disappearance of pancreatic islets on magnetic resonanceimages was due to islet death, which was confirmed byhistological studies. This event was more pronounced inimmunocompetent animals compared with immunocom-promised animals due to severe immune rejection in theformer.

In our next set of experiments, we attempted to tracethe history of the iron label after transplantation and as afunction of islet death. As seen in Fig. 5A, immediatelyafter transplantation in the liver, islets retained Feridexand, therefore, were visible on magnetic resonance im-ages. However, as immune rejection and other factors

FIG. 3. In vivo imaging of intrahepatically transplanted human islets. A: Representative images of NOD.scid mice with transplanted islets. On invivo images, Feridex-labeled islets appeared as signal voids scattered throughout the liver. B: Nonlabeled islets were not detectable using thesame imaging parameters. C: In vivo time course imaging of immune rejection in immunocompetent (upper row) and immunocompromised (lower

row) animals. Representative images are shown from days 2, 10, and 14 after transplantation. Note that the signal voids (arrows) representinglabeled islets/islet clusters tend to disappear faster in Balb/c mice. D: Calibration curve establishing a direct correlation between the number of signalvoids on T2* magnetic resonance images and the number of transplanted islets (R2 � 0.997). E: The number of intrahepatically transplanted isletsdecreased immediately after transplantation in immunocompetent (Balb/c, black columns) and immunocompromised (NOD.scid, white columns) mice.Note that immunocompetent mice exhibited on average a 20% higher rate of islet disappearance on magnetic resonance images due to severe immunerejection, especially pronounced on day 10. F: Apoptotic rates in islet grafts of Balb/c (solid line) and NOD.scid (dashed line) mice. Both animal modelsshowed increased apoptosis immediately after transplantation. Apoptotic rates in immunocompetent mice were higher on days 10–11 due to severeimmune rejection and were consistent with islet death observed by MRI.



were affecting their viability, transplanted islet cells under-went apoptosis releasing Feridex and decreasing its con-tent significantly. Figure 5B shows a pancreatic islet froman immunocompetent mouse 10 days after transplanta-tion, which was infiltrated with immune cells but was stillmaintaining some of the label allowing for its detection byMRI. Interestingly, the remaining label in the islet was notinternalized by invading cells but stayed associated withislet cells. At the same time, as islets were dying, theyreleased their content into the liver parenchyma where itwas internalized and processed by Kuppfer cells (Fig. 5C).

By 14 days after transplantation, however, there was noevidence of iron associated with Kuppfer cells or liverparenchyma, suggesting rapid clearance of the label con-sistent with the known pharmacokinetics of the com-pound (32) (Fig. 5D). From an imaging perspective, ironreleased by dying islet cells in small amounts diffusesthroughout the liver and sparsely distributes over a largerarea and, therefore, is not expected to create “false-positives” on magnetic resonance images. By contrast,Feridex concentrated upon internalization within viablepancreatic islet cells creates a signal void due to compart-

FIG. 4. Histochemical analysis of immune rejection. Consecutive liver sections from Balb/c and NOD.scid mice were stained for the presence ofimmune cells. A: Immunocompetent mice demonstrated the presence of CD4�, CD8�, CD19�, and CD68� cells in transplanted islets.Representative microphotographs from day 10 after transplantation showed severe infiltration by immune cells. B: NOD.scid mice did not showany noticeable infiltration of immune cells except for occasional CD68� macrophages (arrows) (magnification bar � 20 �m).

FIG. 5. Fate of Feridex in transplanted islets. A: Liver section with a transplanted pancreatic islet stained for iron (Prussian Blue stain, blue)indicating the presence of Feridex immediately after transplantation (counterstained with Nuclear Fast Red). B: Representative microphoto-graph showing accumulation of invading immune cells (CD4�, CD8�, and CD68�) in Balb/c mice 10 days after transplantation. Residual Feridexwas present in pancreatic islets and was still associated with islet cells but not with invading immune cells. C: The content of dying islet cells wasreleased into the liver parenchyma where it was internalized and processed by Kuppfer cells (arrows). E: By 14 days after transplantation, therewas no evidence of Feridex associated with liver parenchyma suggesting its clearance.



mentalization of the label in a membrane-enclosed envi-ronment (33,34). The persistence of iron within Kuppfercells is short lived due to its rapid processing by these cellsand its release into the physiologic iron pool (32). There-fore, the only entity within the liver, which would retainiron at a high concentration and with a steady time-coursewould be viable islet grafts. Overall, based on theseresults, we have demonstrated that the labeling of humanpancreatic islets with an FDA-approved contrast agent andtheir imaging by MRI is feasible and represents a promis-ing strategy for the in vivo tracking of transplanted isletfate over time as well as for the detection of islet deathimmediately after transplantation.

DISCUSSION

Type 1 diabetes results from autoimmune destruction ofthe insulin-producing �-cells (35). In diabetic patients,daily treatment with exogenous insulin is required in orderto achieve normoglycemia. However, because of difficul-ties in achieving physiological control of blood glucoseconcentrations, chronic and degenerative complicationsstill occur in a marked number of patients. Pancreatic islettransplantation has recently emerged as an attractivealternative to insulin injection and as a promising clinicalmodality, which can render patients with type 1 diabetesinsulin independent (1,2). In general, islet transplants arehighly susceptible to inflammatory damage and to isch-emia and anoxia, as a consequence of the transplantprocedures. During harvesting, islets are stripped fromtheir blood vessels. Consequently, the perfusion of thegraft is severely compromised until revascularization isestablished (4). However, with the establishment of bloodflow, the islets become a target for ischemia-reperfusioninjury, in addition to inherent ischemia and anoxia, duringthe period of compromised blood flow, coagulation, andthrombosis (36). The major challenge, however, lies inallograft rejection, which occurs after transplantation.Clearly, for islet transplantation to become widely appli-cable, development of successful and safe strategies forinducing tolerance to islet transplants and reducing im-mune rejection is a necessity. Successful immunosuppres-sive regimens provide hope for continous improvements inachieving insulin independence (2). To evaluate the suc-cess of islet transplants, repeated noninvasive monitoringof surviving grafts is needed. MRI, with its high resolutionand availability of contrast agents, seems to be an idealmodality to perform such monitoring.

We have previously shown that in vivo MRI of trans-planted islets is possible, providing that islets are labeledwith a contrast agent (17). Here, we built on our previousfindings and performed in vivo imaging of transplantedislets using Feridex, which is currently used in clinics as acontrast agent for the detection of liver lesions. Similar tonanoparticles used in our proof-of-principle studies (17),Feridex consists of a superparamagnetic iron oxide corecovered with a dextran coat. When used for hepatic MRI,the iron oxide nanoparticles are phagocytosed and accu-mulate in the endosomes of Kuppfer cells and reticuloen-dothelial cells (18). Iron oxide particles are nontoxic,biodegradable, and have been used in the clinic as intra-venous contrast agents. In addition, we utilized the clini-cally relevant intrahepatic model of islet transplantation.Currently, this is the only clinically established protocol(2) that, together with the use of an FDA-approved con-

trast reagent, serves as a solid base for transferring ourstudies into clinical trials.

As a first step, we established the suitability of Feridexfor labeling human pancreatic islets. Specifically, we con-firmed that islets could be labeled with Feridex withoutaltering their function and causing toxicity. Furthermore,we showed that labeled islets were readily detectable onT2-weighted magnetic resonance images due to the pres-ence of the label, a finding that was confirmed microscop-ically. Based on our in vitro results, we performed in vivoMRI of transplanted pancreatic islets in immunocompro-mised and immunocompetent mice. Severe immune rejec-tion in immunocompetent animals was evident from MRIas the gradual disappearance of the islet-associated label.The rate of islet death in these animals was significantlyhigher than in NOD.scid mice and correlated with the lossof signal voids on magnetic resonance images. Interest-ingly, using our in vivo imaging technique, we were able toconfirm previous findings regarding early islet death aftertransplantation, which were exclusively based on histolog-ical assessment. Namely, in vivo MRI showed that thisprocess started immediately after the transplantation pro-cedure in both animal models and continued for theduration of the study. In this study, we deliberatelyavoided using diabetic animals to exclude the effect ofglucose toxicity on islet death (15); instead, we focused onthe input from immune rejection, which was noted as themajor challenge in islet transplantation (4). However, webelieve that the input of glucose toxicity is an extremelyimportant factor and deserves special consideration,which we are currently pursuing in a separate study.

Our findings are in drastic contrast with the recentpublication by Kriz et al. (37), where no islet death wasreported in syngeneic animals up to 6 weeks aftertransplantation. Furthermore, these authors claimedthat islet death in allogeneic transplants began as late as1 week after transplantation. We believe that this cannotbe the case since islet death in both allogeneic andsyngeneic models has been detected immediately aftertransplantation in previous studies and is well docu-mented (9,15,26,27). The authors may not have takeninto consideration islet distribution in the entire liversince they counted dark voids in only three slices onmagnetic resonance images. This provides very limitedcoverage of the liver, especially taking into account theimaging parameters used in this study (slice thickness 2mm with slice separation of 3 mm). In addition, wecould not support the authors’ claim that iron was onlylocalized to islet macrophages, which actually contra-dicts their own previous findings (38). If all of the ironlabel was localized to macrophages, then calculating theiron concentration per cell based on their reported ironaccumulation (38) and the known number of residentmacrophages in an islet (39), the average macrophagewould contain an unrealistic 10 –20 ng of iron. Thepossibility of iron being trapped between islet cells, assuggested by Jirak et al. (38), should be confirmedexperimentally.

Our previously reported observations (17), and thedata presented here, suggest that labeling with ironoxides involves all islet cell types and that massiveaccumulation of the label limited to a small volume ofthe islet causes the dark contrast on magnetic reso-nance images that we observed. Overall, the applicationof contrast agents helps to overcome a natural lowsensitivity of MRI, which seems to be the most suitable



modality for islet imaging. MRI is advantageous incomparison to other imaging modalities because it doesnot utilize damaging ionizing radiation, has tomographiccapabilities, can deliver the highest resolution images invivo, and has unlimited depth penetration (16). Webelieve that in the current study, we have established aconvincing case for utilizing the FDA-approved contrastagent Feridex for in vivo MRI of immune rejection oflabeled grafts, pending approval by appropriate regula-tory agencies. We believe that these studies will betranslated into clinical trials in the near future. Pancre-atic islet transplantation holds hope for patients withtype 1 diabetes. Therefore, the ability to follow the fateof the graft is crucial for the success of this procedure.

ACKNOWLEDGMENTS

This work was supported in part by the National Institutesof Health Grants DK071225 and DK072137 (to A.M.).

We thank the following Islet Cell Resource Centers forproviding human pancreatic islets: Diabetes Research In-stitute, Miami, Florida; City of Hope National MedicalCenter, Duarte, California; Puget Sound Blood Center,Seattle, Washington; Human Islet Laboratory, Universityof Pennsylvania, Philadelphia, Pennsylvania; College ofMedicine, University of Tennessee, Memphis, Tennessee;College of Physicians and Surgeons, Columbia University,New York, New York; Joslin Diabetes Center, Boston,Massachusetts; University of Minnesota, Minneapolis, Min-nesota; Children’s Hospital of Pittsburgh, Pittsburgh, Penn-sylvania; and University of Alberta, Edmonton, Canada.

We acknowledge John Moore, BS, for his exceptionalsupport in animal surgery and Guangping Dai, PhD, for hisassistance with in vivo MRI.

REFERENCES

1. Shapiro A, Lakey J, Ryan E, Korbutt G, Toth E, Warnock G, Kneteman N,Rajotte R: Islet transplantation in seven patients with type 1 diabetesmellitus using a glucocorticoid-free immunosuppressive regimen. N Engl

J Med 343:230–238, 20002. Shapiro A, Ricordi C, Hering B: Edmonton’s islet success has indeed been

replicated elsewhere (Letter). Lancet 362:1242, 20033. Barshes N, Lee T, Goodpasture S, Brunicardi F, Alejandro R, Ricordi C,

Soltes G, Barth M, Hamilton D, Goss J: Achievement of insulin indepen-dence via pancreatic islet transplantation using a remote isolation center:a first-year review. Transplant Proc 36:1127–1129, 2004

4. Ricordi C, Strom T: Clinical islet transplantation: advances and immuno-logical challenges. Nat Rev Immunol 4:259–268, 2004

5. Ryan E, Lakey J, Paty B, Imes S, Korbutt G, Kneteman N, Bigam D, RajotteR, Shapiro A: Successful islet transplantation: continued insulin reserveprovides long-term glycemic control. Diabetes 51:2148–2157, 2002

6. Ryan E, Paty B, Senior P, Bigam D, Alfadhli E, Kneteman N, Lakey J,Shapiro A: Five-year follow-up after clinical islet transplantation. Diabetes

54:2060–2069, 20057. Gysemans C, Waer M, Valckx D, Laureys J, Mihkalsky D, Bouillon R,

Mathieu C: Early graft failure of xenogeneic islets in NOD mice isaccompanied by high levels of interleukin-1 and low levels of transforminggrowth factor-� mRNA in the grafts. Diabetes 49:1992–1997, 2000

8. Tan M, Di Carlo A, Liu S, Tector A, Tchervenkov J, Metrakos P: Hepaticsinusoidal endothelium upregulates IL-1alpha, IFN-gamma, and iNOS inresponse to discordant xenogeneic islets in an in vitro model of xenoislettransplantation. J Surg Res 102:229–236, 2002

9. Bottino R, Fernandez L, Ricordi C, Lehmann R, Tsan M, Oliver R, InverardiL: Transplantation of allogeneic islets of Langerhans in the rat liver: effectsof macrophage depletion on graft survival and microenvironment activa-tion. Diabetes 47:316–323, 1998

10. Kaufman D, Gores P, Field M, Farney A, Gruber S, Stephanian E,Sutherland D: Effect of 15-deoxyspergualin on immediate function andlong-term survival of transplanted islets in murine recipients of a marginalislet mass. Diabetes 43:778–783, 1994

11. Bennet W, Sundberg B, Groth C, Brendel M, Brandhorst D, Brandhorst H,

Bretzel R, Elgue G, Larsson R, Nilsson B, Korsgren O: Incompatibilitybetween human blood and isolated islets of Langerhans: a finding withimplications for clinical intraportal islet transplantation? Diabetes 48:1907–1914, 1999

12. Johansson H, Lukinius A, Moberg L, Lundgren T, Berne C, Foss A, FelldinM, Kallen R, Salmela K, Tibell A, Tufveson G, Ekdahl K, Elgue G, KorsgrenO, Nilsson B: Tissue factor produced by the endocrine cells of the islets ofLangerhans is associated with a negative outcome of clinical islet trans-plantation. Diabetes 54:1755–1762, 2005

13. Shimabukuro M, Zhou Y, Levi M, Unger R: Fatty acid-induced beta cellapoptosis: a link between obesity and diabetes. Proc Natl Acad Sci U S A

95:2498–2502, 199814. Kakuma T, Lee Y, Higa M, Wang Z, Pan W, Shimomura I, Unger R: Leptin,

troglitazone, and the expression of sterol regulatory element bindingproteins in liver and pancreatic islets. Proc Natl Acad Sci U S A 97:8536–8541, 2000

15. Biarnes M, Montolio M, Nacher V, Raurell M, Soler J, Montanya E: �-Celldeath and mass in syngeneically transplanted islets exposed to short- andlong-term hyperglycemia. Diabetes 51:66–72, 2002

16. Massoud T, Gambhir S: Molecular imaging in living subjects: seeingfundamental biological processes in a new light (Review). Genes Dev

17:545–580, 200317. Evgenov N, Medarova Z, Dai G, Bonner-Weir S, Moore A: In vivo imaging

of islet transplantation. Nat Med 12:144–148, 200618. Ferrucci J, Stark D: Iron oxide-enhanced MR imaging of the liver and

spleen: review of the first 5 years (Review). AJR Am J Roentgenol

155:943–950, 199019. Moore A, Weissleder R, Bodganov A: Uptake of dextran-coated monocrys-

talline iron oxides in tumor cells and macrophages. J Magn Reson Imaging

7:1140–1145, 199720. Bulte JW, Douglas T, Witwer B, Zhang SC, Strable E, Lewis BK, Zywicke H,

Miller B, van Gelderen P, Moskowitz BM, Duncan ID, Frank JA: Magneto-dendrimers allow endosomal magnetic labeling and in vivo tracking ofstem cells. Nat Biotechnol 19:1141–1147, 2001

21. Bulte J, Duncan I, Frank J: In vivo magnetic resonance tracking ofmagnetically labeled cells after transplantation (Review). J Cereb Blood

Flow Metab 22:899–907, 200222. Arbab A, Yocum G, Kalish H, Jordan E, Anderson S, Khakoo A, Read E,

Frank J: Efficient magnetic cell labeling with protamine sulfate complexedto ferumoxides for cellular MRI. Blood 104:1217–1223, 2004

23. Bulte J, Zhang S, van Gelderen P, Herynek V, Jordan E, Duncan I, Frank J:Neurotransplantation of magnetically labeled oligodendrocyte progeni-tors: magnetic resonance tracking of cell migration and myelination. Proc

Natl Acad Sci U S A 96:15256–15261, 199924. Josephson L, Tung C, Moore A, Weissleder R: High-efficiency intracellular

magnetic labeling with novel superparamagnetic-Tat peptide conjugates.Bioconjug Chem 10:186–191, 1999

25. Lewin M, Carlesso N, Tung C, Tang X, Cory D, Scadden D, Weissleder R:Tat peptide-derivatized magnetic nanoparticles allow in vivo tracking andrecovery of progenitor cells. Nat Biotechnol 18:410–414, 2000

26. Davalli A, Scaglia L, Zangen D, Hollister J, Bonner-Weir S, Weir G:Vulnerability of islets in the immediate posttransplantation period:dynamic changes in structure and function. Diabetes 45:1161–1167, 1996

27. Barshes N, Wyllie S, Goss J: Inflammation-mediated dysfunction andapoptosis in pancreatic islet transplantation: implications for intrahepaticgrafts (Review). J Leukoc Biol 77:587–597, 2005

28. Prochazka M, Gaskins H, Shultz L, Leiter E: The nonobese diabetic scid

mouse: model for spontaneous thymomagenesis associated with immuno-deficiency. Proc Natl Acad Sci U S A 89:3290–3294, 1992

29. Kirchhof N, Shibata S, Wijkstrom M, Kulick DM, Salerno CT, ClemmingsYH, Galili U, Sutherland DER, Dalmasso AP, Hering BJ: Reversal ofdiabetes in non-immunosuppressed rhesus macaques by intraportal por-cine islet xenografts precedes acute cellular rejection. Xenotransplanta-

tion 11:396–407, 200430. Yasunami Y, Maki T, Nagai Y, Ikehara S, Kodama S, Ikeda S: Cellular

immune response in the liver of mice rejecting intrahepatic rat and guineapig islet xenografts: expansion of intermediate TCR cells. Transplant Proc

30:581, 199831. Li H, Ricordi C, Inverardi L: Effect of graft-versus-host reaction on

intrahepatic islet transplants. Diabetes 48:2292–2299, 199932. Weissleder R, Stark D, Engelstad B, Bacon B, Compton C, White D, Jacobs

P, Lewis J: Superparamagnetic iron oxide: pharmacokinetics and toxicity.AJR Am J Roentgenol 152:167–173, 1989

33. Bodganov A, Martin C, Weissleder R, Brady T: Trapping of dextran-coatedcolloids in liposomes by transient binding to aminophospholipid: prepara-tion of ferrosomes. Biochim Biophys Acta 1193:212–218, 1994



34. Moore A, Marecos E, Bogdanov A, Weissleder R: Tumoral distribution oflong-circulating dextran-coated iron oxide nanoparticles in a rodentmodel. Radiology 214:568–574, 2000

35. Atkinson M, Maclaren N: What causes diabetes? (Review). Sci Am

263:62–67, 199036. Strom T: Saving islets from allograft rejection (Comment). Proc Natl Acad

Sci U S A 102:12651–12652, 200537. Kriz J, Jirak D, Girman P, Berkova Z, Zacharovova K, Honsova E,

Lodererova A, Hajek M, Saudek F: Magnetic resonance imaging of pancre-atic islets in tolerance and rejection. Transplantation 80:1596–1603, 2005

38. Jirak D, Kriz J, Herynek V, Andersson B, Girman P, Burian M, Saudek F,Hajek M: MRI of transplanted pancreatic islets. Magn Reson Med 52:1228–1233, 2004

39. Arnush M, Heltmeler M, Scarlm A, Marino M, Manning P, Corbett J: IL-1produced and released endogenously within human islets inhibits b-cellfunction. J Clin Invest 102:516–526, 1998



Documents

Original Article - diabetes.diabetesjournals.org · In Vivo Imaging of Immune Rejection in Transplanted Pancreatic Islets Natalia V. Evgenov, 1Zdravka Medarova, John Pratt,1 Pamela