In Vivo Molecular Imaging of Somatostatin Receptors in Pancreatic Islet Cells and Neuroendocrine Tumors by Miniaturized Confocal Laser-Scanning Fluorescence Microscopy

The aim of the study was to evaluate real time in vivo molecular imaging of somatostatin receptors (sstrs) using a handheld miniaturized confocal laser scan microscope (CLM) in conjunction with fluorescein-labeled octreotate (OcF) in healthy mice and murine models of neuroendocrine tumors. For CLM a small rigid probe (diameter 7 mm) with an integrated single line laser (488 nm) was used (optical slice thickness 7 μm; lateral resolution 0.7 μm). OcF was synthesized via Fmoc solid-phase peptide synthesis and purified by HPLC showing high-affinity binding to the sstr2 (IC₅₀ 6.2 nmol). For in vitro evaluation, rat and human pancreatic cancer cells were used and characterized with respect to its sstr subtype expression and functional properties. For in vivo confocal imaging, healthy mouse pancreatic islet and renal tubular cells as well as immunoincompetent nude mice harboring sstr-expressing tumors were evaluated. Incubation of sstr-positive cells with OcF showed a specific time- and dose-dependent staining of sstr-positive cells. CLM showed rapid internalization and homogenous cytoplasmatic distribution. After systemic application to mice (n = 8), specific time-dependent internalization and cytoplasmatic distribution into pancreatic islet cells and tubular cells of the renal cortex was recorded. After injection in tumor-harboring nude mice (n = 8), sstr-positive cells selectively displayed a cell surface and cytoplasmatic staining. CLM-targeted biopsies detected sstr-positive tumor cells with a sensitivity of 87.5% and a specificity of 100% as correlated with ex vivo immunohistochemistry. CLM with OcF permits real-time molecular, functional, and morphological imaging of sstr-expressing cell structures, allowing the specific visualization of pancreatic islet cells and neuroendocrine tumors in vivo.

Confocal laser microscopy combines focal laser illumination with a pinhole to geometrically eliminate out-of-focus light. In conjunction with fluorescent dyes and labeling techniques, it permits observation of fine subcellular morphological details and complex biological dynamic processes, even in cells located many micrometers below tissue surface (1). A major drawback of this technique, however, is its usually oversized format with bulky bench top devices not adequate for in vivo imaging in patients. Thus, much effort has been put into the miniaturization and enhanced flexibility of these microscope settings. Recently laser confocal microscopy has been introduced in clinical routine diagnostics by integrating the miniaturized components of a confocal laser scanner into the tip of a CE (Conformité Européenne)-certified, Food and Drug Administration-approved flexible endoscope. This novel technology allows subsurface histological diagnosis at the cellular and subcellular level in vivo and provides instantaneous histopathology during ongoing upper and lower endoscopy. Endomicroscopy enables immediate diagnosis of neoplastic and inflammatory lesions of the intestinal mucosa. Our own previous studies demonstrated the power of confocal endomicroscopy in screening and surveillance colonoscopy, ulcerative colitis, Barrett’s esophagus, and gastric cancer and morphodynamic analysis of liver disease in vivo (2–10). The integration of a similar miniaturized confocal laser scanner into a handheld device resulted in a unique miniaturized confocal laser scan microscope. This novel device permits straight forward fluorescence confocal microscopy in vivo, allowing exact characterization of cell morphology and dynamic imaging of cellular events, identifying various disease states like inflammation, neoplastic transformation, and imaging of microvasculature and perfusion, as has been previously published by our group (10–14).

Expression of somatostatin receptors (sstrs) in neuroendocrine tumors and endocrine organs represent the molecular basis for various clinical diagnostic and therapeutic applications. Therefore, these receptors could serve as an appropriate target for real-time molecular imaging using this novel miniaturized laser confocal microscopy system in vivo. Increased sstr expression is frequently found in neuroendocrine tumors (15) and endocrine organs like pancreatic islet cells (16, 17). These receptors are G protein-coupled transmembrane receptors of which five different subtypes have been cloned, named sstr1-5 (18). After high-affinity binding of its physiological ligand somatostatin or other pharmacologically synthesized analogs like octreotide or octreotate, the receptor is rapidly internalized and the polypeptides exert their mainly antiproliferative and antisecretory effects (19). The specific receptor-mediated uptake of somatostatin and its analogs into the cells and its accumulation represent the molecular basis for various clinical diagnostic and therapeutic applications (20, 21). The principle of dye-coupled hormone receptor imaging is not new (22, 23). However, up to now, the imaging of molecular subcellular targets like peptide receptors in tumors mainly results in an intrinsic light image for example of a specifically targeted tumor cell and do not allow a clear simultaneous in vivo imaging of the tissue anatomy together with a dynamic visualization of molecular targets like the sstrs. Therefore, the aim of this study was to develop and evaluate a miniaturized confocal laser microscopy technique, which allows the specific molecular imaging of sstrs in vivo and in real time, showing equal morphological resolution compared with standard histological tissue investigation. For selective visualization of sstrs, a novel contrast agent for laser confocal microscopy has been specifically developed by conjugating 5-carboxyfluorescein to octreotate.

Materials and Methods

Cell lines

AR42J and Panc1 cell lines were purchased from American Type Culture Collection (Manassas, VA). Cells were cultured in DMEM (PAA, Pasching, Austria) with l-glutamine, sodium pyruvate, high glucose concentrations (4.5 g/liter, Panc1 cells only), 10% fetal calf serum (Seromed Biochrom KG, Berlin, Germany), 1% penicillin /streptomycin (Invitrogen, Karlsruhe, Germany), and 1% amphotericin B (0.25 mg/ml; PAA).

Animals

Eight- to 10-wk old CD1 ν/ν (n = 8) and BALB/c mice (n = 8) (Charles River, Wilmington, MA) were used for the in vivo studies. Animals were bred under specific pathogen-free conditions in the Central Animal Facility of the University of Mainz. All animal experiments were performed according to German guidelines for animal care and in accord with accepted standards of humane animal care (regulation number by German authorities: 1.5 177-07-04/051-59).

Flow cytometry

Surface phenotyping of the cells was performed by staining 5 × 10⁵ cells with specific monoclonal antibodies for 30 min at 4 C. Antibodies used were rabbit polyclonal sstr1, rabbit polyclonal sstr2, rabbit polyclonal sstr3, rabbit polyclonal sstr4, and rabbit polyclonal sstr5. Cells were washed with PBS containing 1% fetal calf serum; the cells were incubated for 30 min with the secondary antibody phytoerythrin conjugated-donkey antirabbit IgG (all antibodies from Abcam, Cambridge, MA). After another washing step, the cells were analyzed in a FACSCalibur flow cytometry unit (BD Biosciences, Heidelberg, Germany) equipped with CELLQuest software (BD Biosciences).

Cell proliferation assay

For proliferation assays, Quick Cell proliferation assay kit (BioCat GmbH, Heidelberg, Germany) was used. The 3 × 10⁵ cells were incubated in cell culture medium in 96-well flat-bottom tissue culture plates in 100 μl total volume per well. Experiments were performed in triplicate with or without the addition of unlabeled octreotide (Sandostatin; Novartis, Basel, Switzerland) or conjugated octreotate labeled with 5-carboxy-fluorescein (OcF; Merck KGA, Darmstadt, Germany). After 48 h, 10 μl of water-soluble tetrazolium salt were added. After an incubation time of 3 h, the reaction was stopped by adding 1% SDS (Sigma-Aldrich GmbH, Hamburg, Germany). The extinction was measured in a SmartSpec3000 (Bio-Rad, München, Germany) spectrophotometer using a reference wavelength of 650 nm and a test wavelength of 405 nm.

RNA isolation and cDNA synthesis

Total RNA was isolated from 1 × 10⁶ cells of each cell line immediately after harvesting and counting using the high pure RNA isolation kit from Roche (Mannheim, Germany). RNA purification was performed following the manufacturer’s instructions. The isolated RNA was quantified with a SpecSmart3000 (Bio-Rad) spectrophotometer. One microgram of RNA was reverse transcribed using the Transcriptor first strand cDNA synthesis kit (Roche). For priming we used the included anchored-oligo(deoxythymidine)₁₈ primers. cDNA synthesis was performed according to the specified instructions.

RT-PCR

PCR was done using Red Taq DNA polymerase (Invitrogen). Reactions were performed using 2 μl cDNA per reaction and primers specific for human and mouse sstr1-5. PCR was performed according to the package inserts. Thermocycler (iCycler; Bio-Rad) conditions were as follows: 95 C for 6 min, 12 cycles of 95 C for 45 sec, 56 C for 30 sec, 72 C for 40 sec; decrease temperature after cycle 1 by 0.5 C every cycle; and 18 cycles of 95 C for 45 sec, 52 C for 30 sec, 72 C for 40 sec with a 2-min extension at 72 C. PCR products were run on a 1.5% agarose gel in 1× Tris base acetic acid buffer along with 1 kb DNA ladder (PegLab, Erlangen, Germany) and visualized using ethidium bromide and a Bio-Rad Gel Doc 2000. Primer specifications used were the following: mouse (5′-3′): sstr1b sense, ctactttgccgcctggtgctc, 62.0 C; sstr1 antisense, tggcaatgatgagcacgtaac, 61.0 C; sstr2 sense, ttgacggtcatgagcatc, 58.5 C; sstr2 antisense, acagacacggacgagacattg, 63.0 C; sstr3 sense, ggccgctgttacctatccttc, 63.0 C; sstr3 antisense, ggcactcctgagaacacaacc, 63.5 C; sstr4 sense, cggaggcgctcagagaagaag, 64.0 C; sstr4 antisense, tggtcttggtgaaagggactc, 61.5 C; sstr5 sense, catgagtgttgaccgctacc, 62.5 C; sstr5 antisense, ggcacagctattggcataag, 60.0 C. Human (5′-3′): sstr1 sense, tgcgcttactcctctgac, 58.5 C; sstr1 antisense, gcttaaacaaacagaaataacgac, 57.5 C; sstr2 sense, catgagtccaattcagagaac, 56.5 C; sstr2 antisense, gtttctaaagatcatatccaggca, 59.0 C; sstr sense, accagttcaccagcatat, 58.0 C; sstr3 antisense, gggcactcccgagaaga, 58.5 C; sstr sense, catgttcaccagcgtctt, 59.0 C; sstr4 antisense, gagagtgaccaacaggga, 59.0 C; sstr5 sense, ctggtcatctgcctgtg, 58.5 C; sstr5 antisense, gttgacgatgttgacggt, 59.0 C.

Synthesis of 5-carboxy-fluorescein-labeled octreotate

OcF was synthesized via Fmoc solid-phase peptide synthesis. The labeling with 5-carboxyfluorescein was performed on solid phase with N,N′-diisopropyl-carbodiimide and 1-hydroxy-benzotriazole (Sigma-Aldrich). The product was cleaved from the resin with 95% trifluoroacetic acid and purified by HPLC. HPLC and Maldi analysis showed the lyophilized OcF to be pure (>99%). The affinity of OcF to the sstr was tested by using a membrane preparation from rat cerebral cortex, which shows a strong predominant expression of sstr2, as described previously (24). The affinity of OcF was tested against ¹²⁵I-Tyr3-labeled octreotide. The IC₅₀ of OcF was 6.2 nmol, showing comparable binding characteristics like [¹¹¹In-DTPA]octreotide or [⁹⁰Y-DOTA-Tyr]octreotide (Ic₅₀ of 22 and 11 nm, respectively), which is used for routine clinical imaging purposes.

Fluorescence microscopy

Cells were disseminated in chamber slides (Nunc, Langenselbold, Germany) with a final volume of 400 μl. After 24 h, cell medium was replaced by medium including the fluorescent agents in different concentrations (0.1–100 μm) and incubated for 30 min. After being washed twice, detection of fluorescence and visualization of binding to tumor cells was investigated using conventional fluorescence microscopy (Olympus-X10, Hamburg, Germany). The binding and internalization of OcF was tested by using a benchtop confocal laser-scanning microscope (LSM 510UV; Zeiss, Oberkochen, Germany).

In vivo confocal imaging

In the miniaturized confocal laser-scanning microscope, a prototype (FIVE1) provided by Optiscan Pty. Ltd. (Notting Hill, Victoria, Australia), the x-y scan mechanism and the z-axis actuator have been integrated into the distal tip of the microscopy probe (see Fig. 1). The probe is flexibly connected to the laser source and the detection unit. A solid-state laser delivers an excitation wavelength of 488 nm, and light emission is detected at 505 to 585 nm. Actuation of the imaging plane depth along the range of the z-axis (from surface to 250 μm) is controlled using two remote control buttons. Laser power output can be adjusted during imaging from 0 to 1000 μW to achieve appropriate tissue contrast. Serial images are collected at a scan rate of 0.8 frames/sec at 1024 × 1024 pixels or 1.6 frames/sec at 1024 × 512 pixels, approximating a 1000-fold magnification on a 19-in. screen. Details could be further magnified by changing into a review mode during ongoing examination. For in vivo imaging, the probe was directly placed onto the tissue by using it handheld or mounting the probe onto a stereotactic frame. A midlaser power of 300–600 μW was found adequate. A total of 16 animals were examined. Healthy mice were examined to establish the staining protocols as well as investigate the distribution pattern of the fluorescence agents. For tumor induction 10 × 10⁶ cells were injected sc into the back of CD1 ν/ν mice. After 2–4 wk, when the tumor size had reached approximately 5–10 mm, the animals were evaluated. The animals were deeply anesthetized by ip injection of ketamine/xylazine (3 μl/g body weight; Pfizer Pharma GmbH, Berlin, Germany). After in vivo imaging, the animals were killed by ketamine/xylazine overdose. Confocal images were collected after intracardial application of fluorescence agents. Fluorescein sodium (1 g per 10 ml; Alcon Pharma, Freiburg, Germany) or OcF (1 mg per 250 μl) were used as fluorescent agents using 10 μl/g body weight. For investigation of nonsuperficial organs like the mouse pancreas or the kidney, the organs were surgically exposed after a median laparotomy.

Fig. 1.

The miniaturized laser confocal microscopy system. The miniaturized laser confocal microscopy system consisting of a control and acquisition unit, the display, and control interface (A) and the laser illumination and detection unit (B). Microscopic images are generated in vivo using the handheld confocal microscopy probe by direct placement onto the tissue to be evaluated (C). Miniaturized laser confocal imaging provides instantaneous high-resolution histopathological evaluation for instances of tumor tissue implanted in immunoincompetent nude mice (D).

Ex vivo histology

Tissue specimens were fixed in 4% buffered formalin and embedded in paraffin. Both vertical and transverse serial sections (4 μm) were obtained and stained with hematoxylin and eosin. For immunohistochemical analysis of sstr2 expression, a rabbit polyclonal antibody directed against the amino acid sequence 355-369 (ETQRTLLNGDLQTSI) of sstr2 was used (final dilution 1:500; Gramsch Laboratories, Schwabhausen, Germany) known to detect mouse sstr2. Serotonin-producing tumors of the terminal ileum served as positive controls. Deparaffinized tissue sections were rehydrated and subjected to heat-induced epitope retrieval procedures (25). After blocking with nonimmune serum, the sections were incubated with the primary antibody (45 min), followed by species-specific biotinylated secondary antibodies (45 min; Dianova, Hamburg, Germany). After being washed, the slides were incubated with the ABC reagents for 30 min (Vectastain Elite ABC kit, Boehringer, Ingelheim, Germany). The immunoreaction was visualized with 3′3-diaminobenzidine (Sigma, Deisenhofen, Germany) and counterstained with hematoxylin. Sections were analyzed and photographed with an Axioskop 50 microscope (Zeiss, Oberkochen, Germany).

Statistical analysis

Arithmetic means and sds were calculated using SPSS Software (SPSS Inc., Chicago, IL). Data were analyzed using Scheffé’s F test for testing the variance of the probability distribution and Student’s t test for unpaired samples. Statistical significance was defined as P < 0.05 unless otherwise mentioned.

Results

sstr expression of AR42J and Panc1 cancer cells

For in vitro and in vivo characterization of the sstr-specific contrast agent, different cancer cell lines have been investigated with respect to its sstr expression at the mRNA and the protein level to serve as positive and negative controls, respectively. Two different cancer cell lines have been identified being suitable for further analysis. Evaluation of sstr mRNA of all five subtypes by RT-PCR revealed expression of all sstr subtypes in the rat pancreatic cancer cell line AR42J, with the sstr subtypes −2 to −5 being most abundant (Fig. 2A). Similar results were obtained by characterizing the receptor expression of the cells at the protein level by fluorescence-activated cell sorter (FACS) analysis (Fig. 2B). As in many malignant NET cells, the predominant sstr subtype expressed was sstr2, which was detectable in more than 70% of all AR42J cells and shows high-affinity binding to various somatostatin analogs like octreotide or octreotate. In contrast, sstr1-5 mRNA expression was barely detectable in the human pancreatic adenocarcinoma cell line Panc1 (Fig. 2A), and no relevant sstr expression at the protein level was detectable on Panc1 cells, as demonstrated by FACS analysis.

Fig. 2.

sstr expression of AR42J and Panc1 cells at the mRNA and protein level. Analysis of sstr subtype expression of AR42J and Panc1 cells at the mRNA and protein-level by RT-PCR (A) and FACS analysis (B). Data are represented as mean values ± sem (n = 3). *, P < 0.05 vs. sstr1; **, P < 0.001 vs. sstr3-5.

Bioactivity of OcF in vivo

To demonstrate uncompromised biological activity of the OcF, we compared the antiproliferative effect of unlabeled octreotide compared with OcF in vitro (Fig. 3). Administration of unlabeled octreotide resulted in a significant reduction in proliferation compared with control cells with a more than 50% suppression of proliferative activity at 5 μm. In contrast, OcF showed significant antiproliferative activity only at concentrations higher than 100 μm but had a similar maximum antiproliferative effect at 250 μm. This demonstrates a comparable maximal antiproliferative effect of OcF at higher but still physiological concentrations of OcF. In sstr-negative Panc1 cells, none of the different somatostatin analogs exerted any antiproliferative effect.

Fig. 3.

Biological effect of OcF in vitro. Incubation of sstr-positive AR42J cells with octreotide resulted in a significant antiproliferative effect already at low concentrations of 5 μm, reaching a maximum effect at high physiological concentrations of 250 μm. OcF exerted a significant antiproliferative effect only at concentrations greater 100 μm but, however, reaching a similar maximum antiproliferative effect like unlabeled octreotide at 250 μm (A). No effect of either compound was found on the proliferation of sstr-negative Panc1 cells (B). Data are represented as box plots of all performed experiments (n = 9) representing the median and range of the values 25% above and below the median as boxes as well as the maximum and minimum values as error bars.

In vitro sstr imaging

For in vitro testing of the developed contrast agent, incubation studies with the aforementioned cell lines were performed, using conventional fluorescence microscopy for visualization of binding to the tumor cells. Incubation with OcF showed a time- and dose-dependent staining of the cells with a maximum intensity at 100 μm OcF and a still detectable contrast at 0.1 μm OcF (Fig. 4A). The staining was already detectable after 10 min of incubation and showed a maximum intensity after 2 h. Best contrast to background ratio and a homogenous cytoplasmatic staining of sstr-positive cells was detected at a final concentration of 10 μm OcF after a 30-min incubation time. Similar results were obtained when the miniaturized laser confocal microscope was used for visualization of OcF-stained AR42J-cells in vitro (Fig. 4B). The binding and subsequent rapid internalization of OcF could be further demonstrated by conventional confocal laser-scanning microscopy, in which OcF was rapidly bound and completely internalized to the cytoplasm over a time period of 60 min (Fig. 4C). Specific binding of OcF to the sstr was demonstrated by adding an excess amount of unlabeled octreotide to AR42J cells Beevor staining the cells with OcF. After being washed, only a weak residual binding of OcF was still detectable after preincubation with 10 mm unlabeled octreotide. A complete displacement of OcF binding from its receptor has been observed after increasing the amount of unlabeled octreotide to 100 mm (ratio OcF to unlabeled octreotide = 1:10,000) before staining the cells with OcF, demonstrating specific high-affinity receptor binding of the synthesized fluorescein-labeled somatostatin analog to its receptor. When sstr-negative exocrine pancreatic cancer cells (Panc1) were used, no fluorescence was observed after incubation with OcF as expected.

Fig. 4.

Imaging of sstrs in vitro. Incubation of AR42J cells with OcF showed a time- and dose-dependent staining of sstr-positive cells with a maximum intensity at 100 μm and a still detectable concentration (c) at 0.1 μm OcF as demonstrated by fluorescence microscopy in vitro (A). Similar staining results were obtained by using the miniaturized handheld laser confocal microscope (B). Dynamic imaging of sstr binding and internalization of OcF with a benchtop confocal laser scan microscope demonstrated binding of OcF to the sstr after 2 min and a subsequent rapid internalization (10 min) and homogenous cytoplasmatic distribution (20–60 min) (C).

In vivo tissue imaging with miniaturized laser confocal microscopy

For real-time in vivo imaging using the miniaturized confocal laser scan microscope, normal tissue of healthy adult mice was used to study tissue architecture and anatomy. In normal healthy mice, iv injection of 10 μl/g body weight fluorescein sodium (1 mg per 250 μl) resulted in a quick distribution of the fluorescein compound throughout the tissue, allowing the visualization of tissue architecture and blood vessels. For in vivo imaging, the probe of the confocal laser scan microscope was directly placed onto the tissue by using it handheld device. Intravenous application of fluorescein sodium immediately allowed exact in vivo subsurface imaging of normal tissue architecture, showing excellent correlation with ex vivo histological examination (Fig. 5) but in addition allowing an ongoing dynamic real-time tissue monitoring.

Fig. 5.

In vivo imaging of normal cell morphology using the miniaturized confocal laser scan microscope. Microscopic images using the miniaturized confocal microscopy system in healthy normal and nude mice are generated by placing the handheld probe directly on to the tissue to be examined after iv application of fluorescein sodium. In mouse pancreatic tissue, acinar cells (arrow) could be easily visualized as well as fibrous tissue (arrowhead) and interspersed adipocytes (solid arrow) (A). Confocal laser scan microscopy immediately allowed in vivo subsurface imaging of neuroendocrine tumors (B) as well as tubular cells of the renal cortex (blank arrow) (C), showing good correlation with ex vivo histological examination. HE, Hematoxylin and eosin.

In vivo molecular imaging of sstrs

For in vivo-specific molecular imaging of sstr-positive cells, either sstr-expressing pancreatic islet and renal tubular cells or sstr-positive-implanted AR42J tumors in immunoincompetent nude mice were used. After systemic application of OcF in the mouse pancreas, a subpopulation of cells was contrasted at the cell surface 15 min after the injection. Internalization and homogenous cytoplasmatic distribution was recorded after an additional 15–30 min. The bright cells clearly contrasted with surrounding tissue representing sstr-positive islet cells of the pancreas as confirmed by ex vivo immunohistochemistry (Fig. 6).

Fig. 6.

In vivo molecular imaging of sstrs in pancreatic mouse islets. After iv application of OcF in the mouse pancreas, a subpopulation of cells was contrasted with a homogenous cytoplasmatic distribution leaving the nucleus unstained (A). The bright cells clearly contrasted to surrounding tissue. In contrast to in vivo imaging, ex vivo histological (hematoxylin and eosin staining, B) and immunohistochemical staining (anti-sstr2-diaminobenzidine, C) of a corresponding section through a mouse pancreas specimen did show a predominant superficial membrane-bound immunostaining of the sstr2 without significant cytoplasmatic staining.

Molecular imaging of sstr-positive cancer cells in vivo was tested after tumor induction by sc injection of AR42J cells or Panc1 cells in nude mice. After the application of OcF, sstr-expressing tumor cells displayed a bright cell surface and cytoplasmatic staining (Fig. 7), whereas sstr-negative Panc1 tumor cells were detectable only by background fluorescence using twice as much contrast agent. The morphological visualization of the tumor cells showed an excellent correlation compared with the classical immunohistochemical tissue examination ex vivo using a sstr2 antibody for tissue staining. After targeted core needle biopsy of the tumor tissue specimens examined with confocal laser scan microscopy after OcF application, correlation with immunohistochemical sstr2 staining was performed and sstr-positive cells could be detected with a sensitivity of 87.5% and a specificity of 100% by miniaturized confocal laser scan microscopy. Comparable with ex vivo immunohistochemistry, real-time in vivo examination of the tumor cells with confocal laser scan microscopy after OcF application showed irregular sstr expression throughout the tumor tissue with areas of strong receptor expression contrasting with regions of sstr negativity as is frequently observed in less differentiated neuroendocrine cancer cells.

Fig. 7.

In vivo molecular imaging of sstrs in pancreatic cancer cells. After tumor induction by sstr-positive rat pancreatic tumor cells (AR42-J) in nude mice, sstr-positive tumor cells were selectively visualized after OcF application using the miniaturized confocal laser scan microscopy system (A and B). After 30–60 min, a subset of tumor cells displayed a homogenous cytoplasmatic staining pattern. Sstr distribution was irregular among tumor cells and showed excellent correlation to ex vivo immunohistochemical staining of AR42-J tumor cells (anti-sstr2-diaminobenzidine) (C).

When mouse kidneys were investigated after systemic application of OcF, tubular cells of the renal cortex displayed a faint superficial staining already a few minutes after application. Fifteen to 30 min after OcF application, a bright surface staining of most renal tubular cells was recorded as well as a beginning cytoplasmatic distribution of the contrast agent, reaching its maximum intensity after 30–45 min and gradually attenuating after 60 min with a bright-contrast enhancement of the renal tubular system after tubular secretion (Fig. 8). Not all tubular cells of the renal cortex contrasted after OcF application, which is in good accordance with ex vivo immunohistochemistry, showing an irregular expression of the sstr2 in renal tubular cells.

Fig. 8.

Dynamic in vivo molecular imaging of sstr in tubular cells of the renal cortex. The sstr-positive tubular cells of the renal cortex (arrowhead) contrasted with sstr-negative cells (arrow) as correlated with ex vivo immunohistochemistry, showing an irregular expression of the sstr2 in renal tubular cells (A). After systemic application of OcF, tubular cells of the renal cortex displayed a faint superficial staining already a few minutes after application (arrow). After 30 min, a brighter surface staining of most renal tubular cells was recorded as well as a beginning cytoplasmatic distribution of the contrast agent, reaching its maximum intensity after 30–45 min and gradually attenuating after 60 min, contrasting the tubular luminal system (B).

Discussion

In the present study, we demonstrate for the first time in vivo molecular, functional, and morphological imaging of sstr-expressing cell structures in real time using a novel OcF specifically synthesized as a contrast agent in conjunction with a miniaturized laser confocal microscopy system. This unique fluorescein-labeled somatostatin analog has been demonstrated to specifically bind to sstr in a time- and dose-dependent manner and exerts a functional antiproliferative effect on sstr-positive tumor cells. After systemic application, OcF allows specific dynamic in vivo imaging of sstr-positive pancreatic islet and cortical renal tubular cells as well as sstr-positive neuroendocrine tumor cells as correlated with ex vivo immunohistochemistry. This novel technique not only provides instantaneous high resolution histopathological tissue evaluation but also permits identification and characterization of subcellular molecular structures like sstrs and opens new horizons for dynamic real-time morphological, functional, and molecular imaging in vivo.

Although confocal microscopy allows detailed cellular analyses in vitro, its in vivo use has been limited by stationary and bulky optics and relatively few studies have successfully used this technique for imaging in disease models or humans in vivo. Previous attempts to build confocal microscopes with miniaturized scanning heads have incurred significant performance compromises, resulting in poor resolution, sensitivity, or both (26, 27), and targeted fluorescence technology has been thoroughly exploited but not linked to in vivo microscopic imaging (28, 29). For the present study, we evaluated a rigid confocal minimicroscope that allows full-resolution, high-sensitivity point scanning real-time confocal fluorescence imaging, suitable for flexible, handheld imaging in living animals but also appropriate for use in humans. In vivo confocal minimicroscopy is easy to apply and confocal images can be captured in a single examination without major readjustments within the procedure. In addition, the combination with particular fluorescent agents for the first time opens the opportunity for in vivo molecular imaging.

In this study, we used the sstr as molecular target to proof the feasibility of a true molecular real-time in vivo imaging technique. sstrs are overexpressed in the vast majority of GEP-NET cells and have been widely used for diagnostic purposes. Labeling of somatostatin with radioactive molecules results in a specific trapping and accumulation of the contrast agent in the target cell, allowing visualization of sstr-overexpressing cells (21). Evaluation of the binding characteristics of the fluorescein-labeled octreotate used in this study revealed an IC₅₀ of 6.2 nm, showing comparable binding affinities like [¹¹¹In-DTPA] octreotide or [⁹⁰Y-DOTA-Tyr]octreotide, used for routine clinical diagnostic and therapeutic purposes in NET (30). Visualization of sstr-positive cells by high-affinity-labeled octreotate demonstrates that in vivo molecular imaging is feasible with the confocal minimicroscopy probe by selectively and specifically targeting single molecules, such as peptide receptors. This approach identified sstr-positive cancer cells as well as pancreatic islet cells in vivo after systemic application of fluorescein-labeled octreotate. A limitation of confocal in vivo imaging of pancreatic islet cells, however, includes the limited infiltration depth of a maximum of 250 μm. Thus, only few superficial pancreatic islets could be visualized, being a drawback of this technique. In the future, however, imaging depth will be increased by integrating longer wavelength laser lines, e.g. near infrared, or using nonlinear imaging techniques such as two-photon microscopy (31, 32). In analogy to pancreatic islet cells, confocal laser scan microscopy allowed visualization of sstr in tubular cells of the renal cortex. In addition, with confocal laser scan microscopy, a dynamic visualization of sstr binding and uptake could be demonstrated in vivo.

Dynamic observation of such specific molecular events in the intact organ system is unique to an in vivo microscopic imaging approach and adds significant insight to our understanding of signaling pathways of cells in their normal environment but also can influence in vivo molecular diagnosis of human diseases. With routine endoscopic diagnostics, intestinal NET cells can usually not be distinguished from other polypoid structures by macroscopy alone. Using fluorescein-labeled octreotate, these lesions could be detected with confocal endoscopy. Additionally, molecular imaging of sstr-positive cells with confocal laser scan microscopy could allow intraoperative tumor visualization for guided surgical resection, e.g. during endoscopic transphenoidal surgery for pituitary tumors, which frequently overexpress sstrs, thereby defining tumor-free resection margins. In laparoscopy, confocal imaging can guide biopsy to focal lesions, e.g. of the liver (10). In general, the confocal minimicroscopy probe allows the immediate microscopic morphological, functional, and molecular evaluation of all easily accessible organs, thus facilitating screening for and early diagnosis of sstr-positive neoplasias. However, the development of specific molecular tracers still needs thorough evaluation as to specificity, sensitivity, and safety before in vivo application in humans is conceivable. But having established the feasibility of in vivo molecular imaging in this study and given the fact that both fluorescein and the somatostatin analogs are already approved for the use in men, it is tempting to speculate that rapid technological improvement will allow similar technical features in confocal minimicroscopy as in bench-top confocal microscopy, making confocal laser scan microscopy a promising tool for morphological and molecular imaging in vivo.

Disclosure Summary: P.D. is director of technology at Optiscan Pty., Ltd. All other authors have nothing to disclose.

First Published Online March 16, 2010

¹

C.F. and E.M. contributed equally to this paper.

Abbreviations

 
• FACS,

  Fluorescence-activated cell sorter;

 
• OcF,

  5-carboxy-fluorescein;

 
• sstr,

  somatostatin receptor.