68Ga-labelled exendin-3, a new agent for the detection of insulinomas with PET
- First Online:
- Cite this article as:
- Brom, M., Oyen, W.J.G., Joosten, L. et al. Eur J Nucl Med Mol Imaging (2010) 37: 1345. doi:10.1007/s00259-009-1363-y
- 834 Downloads
Insulinomas are neuroendocrine tumours derived from pancreatic β-cells. The glucagon-like peptide 1 receptor (GLP-1R) is expressed with a high incidence (>90%) and high density in insulinomas. Glucagon-like peptide 1 (GLP-1), the natural ligand of GLP-1R, is rapidly degraded in vivo. A more stable agonist of GLP-1R is exendin-3. We investigated imaging of insulinomas with DOTA-conjugated exendin-3 labelled with 68Ga.
Targeting of insulinomas with [Lys40(DOTA)]exendin-3 labelled with either 111In or 68Ga was investigated in vitro using insulinoma tumour cells (INS-1). [Lys40(111In-DTPA)]Exendin-3 was used as a reference in this study. In vivo targeting was investigated in BALB/c nude mice with subcutaneous INS-1 tumours. PET imaging was performed using a preclinical PET/CT scanner.
In vitro exendin-3 specifically bound and was internalized by GLP-1R-positive cells. In BALB/c nude mice with subcutaneous INS-1 tumours a high uptake of [Lys40(111In-DTPA)]exendin-3 in the tumour was observed (33.5 ± 11.6%ID/g at 4 h after injection). Uptake was specific, as determined by coinjection of an excess of unlabelled [Lys40]exendin-3 (1.8 ± 0.1%ID/g). The pancreas also exhibited high and specific uptake (11.3 ± 1.0%ID/g). High uptake was also found in the kidneys (144 ± 24%ID/g) and this uptake was not receptor-mediated. In this murine tumour model optimal targeting of the GLP-1R expressing tumour was obtained at exendin doses ≤0.1 µg. Remarkably, tumour uptake of 68Ga-labelled [Lys40(DOTA)]exendin-3 was lower (8.9 ± 3.1%ID/g) than tumour uptake of 111In-labelled [Lys40(DTPA)]exendin-3 (25.4 ± 7.2%ID/g). The subcutaneous tumours were clearly visualized by small-animal PET imaging after injection of 3 MBq of [Lys40(68Ga-DOTA)]exendin-3.
[Lys40(68Ga-DOTA)]Exendin-3 specifically accumulates in insulinomas, although the uptake is lower than that of [Lys40(111In-DTPA)]exendin-3. Therefore, [Lys40(68Ga-DOTA)]exendin-3 is a promising tracer to visualize insulinomas with PET.
Insulinomas are neuroendocrine tumours (NET) derived from pancreatic beta-cells. The diagnosis is based on clinical symptoms (such as neuroglycopenic symptoms) and biochemical parameters (endogenous hyperinsulinism confirmed by hypoglycaemia). The standard treatment for insulinomas is surgical resection, requiring exact localization of the tumour lesions. Therefore, patients usually undergo computed tomography (CT) and magnetic resonance imaging (MRI) to precisely localize the primary tumour and potential metastases in those with malignant insulinoma. Somatostatin receptor scintigraphy (SRS) is considered the most sensitive method for detection of NET . In contrast to other NET, the sensitivity of SRS for detecting insulinomas is only 40–60% due to the relatively low incidence of somatostatin subtype receptor (sstr) 2 and sstr5 expression in insulinomas [1–3]. Therefore, SRS is not suitable for localizing insulinomas. Recent studies have shown a high detection rate of insulinomas with endoscopic ultrasonography . However, the high detection rate is limited to tumours that are located in the head and the body of the pancreas. Moreover, this technique requires specialized expertise and is observer-dependent. Angiography shows reasonable localization of insulinomas in approximately 70% of patients. However, the localization varies from 43% to 94% due to observer-dependency . Arterial stimulation and venous sampling is a sensitive method that is able to localize over 90% of insulinomas . The major drawback is that the procedure is invasive. Thus, there is a need for new noninvasive strategies for localization of insulinomas.
Amino acid structures of exendin-3 and exendin-4. Homologous amino acids are underlined
Amino acid sequence
Because even small insulinomas may induce clinical symptoms, imaging methods for the detection of these tumours needs to be highly sensitive. Positron emission tomography (PET) may therefore offer an advantage over SPECT because of its excellent sensitivity in combination with its improved spatial resolution. As the radionuclide for PET imaging of GLP-1Rs with exendin, 68Ga seems to be the optimal candidate since DOTA-conjugated probes can be rapidly and efficiently labelled with 68Ga, and because 68Ga is a generator-produced positron emitter, it is widely available and relatively cheap. The conjugation of exendin-3 with DOTA could change the pharmacokinetics of the peptide. Therefore we compared the biodistribution of [Lys40(DTPA)]exendin-3 and [Lys40(DOTA)]exendin-3 labelled with 111In.
We determined the potential of 68Ga-labelled [Lys40(DOTA)]exendin-3 for insulinoma imaging. [Lys40(111In-DTPA)]exendin-4 was successfully used for this application [12, 14, 15]. Here, biodistributions of [Lys40(111In-DTPA)]exendin-3 and [Lys40(111In-DTPA)]exendin-4 were compared. The effect of the chelator (DOTA or DTPA) was examined by comparing the biodistributions of [Lys40(111In-DTPA)]exendin-3 and [Lys40(111In-DOTA)]exendin-3. The biodistributions of 68Ga- and 111In-labelled exendin-3 were compared. Finally, microPET images were acquired of insulinoma-bearing BALB/c nude mice after injection of [Lys40(68Ga-DOTA)]exendin-3.
Materials and methods
Peptides and radionuclides
[Lys40(DTPA)]Exendin-3 was dissolved in 0.1 M MES (2-(N-morpholino)ethanesulphonic acid) buffer, pH 5.5, and 150 MBq 111In was added to 1 µg peptide followed by five volumes of 0.1 M MES buffer, pH 5.5, resulting in a total volume of reaction mixture ranging from 800 µl to 2,345 µl. After incubation at room temperature for 30 min, quality control was performed using reversed-phase high-performance liquid chromatography (RP-HPLC) on a C18 reversed-phase column (Zorbax Rx-C18; 4.6 mm × 25 cm; Agilent Technologies) and instant thin-layer chromatography (ITLC). The column was eluted with 10 mM ammonium acetate, pH 5.5, with a linear gradient from 3% to 100% acetonitrile in 10 min (flow rate 1 ml/min). ITLC was performed on silica gel ITLC (Pall Corporation Life Sciences, New York, NY). Two mobile phases were used: 0.1 M EDTA in 0.1 M NH4Ac, pH 5.5 (Rf111In-labelled exendin = 0, Rf unbound 111In = 1) and 0.25 M NH4Ac (pH 5.5)/methanol (1:1) (Rf colloidal residues = 0: , Rf111In-labelled exendin and unbound 111In = 1).
[Lys40(DOTA)]Exendin-3 was labelled with 111In by adding five volumes of 0.25 M ammonium acetate buffer, pH 5.5, containing 5 µg peptide and 55 MBq 111In, resulting in a total volume of reaction mixture ranging from 359 µl to 1,043 µl. After 20 min incubation at 95°C, quality control was performed using RP-HPLC and ITLC as described above. [Lys40(DOTA)]exendin-3 was labelled with 68Ga by adding 185 MBq of 68Ga in 350 µl 0.1 N Ultrapure HCl to 42 µl 2.5 M HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulphonic acid) buffer, pH 7, containing 8 µg [Lys40(DOTA)]exendin-3. The final pH of the reaction mixture was 3.0. The mixture was incubated at 95°C for 15 min, then EDTA was added to a final concentration of 5 mM, and the mixture was incubated at room temperature for 5 min. The reaction mixture was purified by HPLC using a C8 reversed-phase column (Zorbax eclipse XDB C8 4.6 × 150 mm, 5 µm; Agilent Technologies). The column was eluted with 10 mM ammonium acetate, pH 5.5 (0–5 min) and 40% ethanol (5–10 min) followed by a linear gradient from 40% to 90% ethanol over 5 min (flow rate 1 ml/min). The fractions containing [Lys40(68Ga-DOTA)]exendin-3 were collected and diluted with PBS containing 0.5% BSA to a final ethanol concentration of less than 10% before injection into mice (injection volume 0.2 ml). The contents of unincorporated 68Ga and 68Ga colloid in the purified 68Ga-labelled [Lys40(DOTA)]exendin-3 preparations were determined using ITLC as described above.
The rat insulinoma cell line INS-1  was maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum, 2 mM glutamine, 10 mM HEPES, 50 µM β-mercaptoethanol, 1 mM sodium pyruvate, 100 U/ml penicillin and 100 µg/ml streptomycin, in a humidified atmosphere containing 5% CO2 at 37°C. The cells were harvested by trypsinization with trypsin/EDTA.
The 50% inhibitory concentrations (IC50) of [Lys40(natIn-DTPA)]exendin-3, [Lys40(natIn-DOTA)]exendin-3 and [Lys40(natGa-DOTA)]exendin-3 were determined using suspensions of INS-1 cells. [Lys40(DTPA)]Exendin-3 was labelled with stable natIn by adding 10.6 nmol natInCl3 (Merck, Darmstadt, Germany) to 2.12 nmol [Lys40(DTPA)]exendin-3 (0.21 nmol/µl) in 30 µl 0.1 M MES buffer, pH 5.5, and incubated at room temperature for 30 min. [Lys40(DOTA)]Exendin-3 was labelled with natIn in 0.25 M NH4Ac, pH 5.5, containing 2.12 nmol [Lys40(DOTA)]exendin-3. After adding 10.6 nmol natIn, the reaction mixture was incubated at 95°C for 30 min. [Lys40(DOTA)]exendin-3 was labelled with stable natGa by adding 10.6 nmol natGa(NO)3 (Sigma Aldrich, St Louis, MO) to 2.12 nmol [Lys40(DOTA)]exendin-3 in 30 µl 0.25 M NH4Ac, pH 5.5, followed by incubation at room temperature for 3 h. INS-1 cells were resuspended in binding buffer (HEPES-buffered RPMI containing 1% BSA) to a concentration of 3 million cells/0.5 ml. After 10 min incubation at 0°C, natIn- or natGa-labelled peptide was added to a final concentration in the range 0.10–350 nM along with radiolabelled [Lys40(111In-DTPA)]exendin-3 (60,000 dpm). To prevent internalization, cells were incubated on ice for 4 h. The cell suspension was centrifuged at 3,000 g for 5 min and the supernatant was discarded. The cells were washed twice with ice-cold PBS, centrifuged, and the radioactivity in the cell pellet was determined in a gamma-counter (Wizard, Wallac-Oy, Uppsala, Sweden), and the bound fraction was determined.
In vitro internalization
The kinetics of internalization of [Lys40(111In-DTPA)]exendin-3 were determined as described previously . INS-1 cells were seeded at a density of 500,000 cells/well in six-well plates and incubated overnight at 37°C. The cells were washed twice with binding buffer and subsequently incubated at 37°C for 10 min in 1.0 ml binding buffer. Approximately 60,000 dpm of 111In-labelled peptide (55 fmol) was added followed by incubation for 0.5, 1, 2 or 4 h at 37 C. To determine nonspecific binding, 1 µg unlabelled [Lys40]exendin-3 was added together with the 111In-labelled peptide to three wells per time point. After incubation, the cells were washed twice with binding buffer. To determine the surface-bound radiolabelled peptide, ice-cold acid buffer (0.1 M acetic acid, 154 mM NaCl, pH 2.5) was added and cells were incubated for 10 min at 4°C. After washing the cells twice with binding buffer, they were harvested with a cotton stick and the cell-associated activity was measured in a gamma-counter. The internalized fraction (cell-associated activity after acid wash) and the receptor-bound fraction (activity removed by acid wash) were determined. The values of the internalized and receptor bound fractions were corrected for nonspecific binding as determined by coincubation with an excess of unlabelled [Lys40]exendin-3 (1 µg/well).
Biodistribution studies of [Lys40(111In-DTPA)]exendin-3 and [Lys40(111In-DTPA)]exendin-4
The biodistributions of [Lys40(111In-DTPA)]exendin-3 and [Lys40(111In-DTPA)]exendin-4 were compared. Female BALB/c nude mice at 6–8 weeks of age were injected subcutaneously in the right flank with 0.2 ml of a suspension of INS-1 cells containing 5 × 107 cells/ml. When the tumours were 2–5 mm in diameter, the mice were randomly divided into groups of five mice each. The mice were injected intravenously with 370 kBq (peptide dose 0.1 µg) of 111In-labelled [Lys40(DTPA)]exendin-3 (n = 5) or 111In-labelled [Lys40(DTPA)]exendin-4 (n = 5) via the tail vein. The mice were killed 4 h after injection by CO2/O2 suffocation, and blood, tumour and relevant tissues (muscle, heart, lung, spleen, pancreas, kidney, liver, stomach and duodenum) were removed, weighed and counted in a gamma-counter. The percentage injected dose per gram (%ID/g) was determined for each tissue. To determine whether the exendin uptake is GLP-1R-mediated, a 100-fold excess (10 μg) of unlabelled [Lys40]exendin was coinjected in a separate group of two mice.
Biodistribution studies of [Lys40(111In-DTPA)]exendin-3 and [Lys40(111In-DOTA)]exendin-3
In order to determine the effect of the chelator, the biodistributions of [Lys40(111In-DTPA)]exendin-3 and [Lys40(111In-DOTA)]exendin-3 were determined in BALB/c nude mice with a subcutaneous tumour. The mice were injected intravenously with 370 kBq 111In-labelled [Lys40(DTPA)]exendin-3 (n = 5) or [Lys40(DOTA)]exendin-3 (n = 5). To determine the GLP-1R-mediated uptake, a separate group of two mice were coinjected with an excess of unlabelled [Lys40]exendin. The mice were killed 4 h after injection by CO2 suffocation, and blood, tumour and major organs were removed, weighed and counted.
Biodistribution studies of [Lys40(111In-DTPA)]exendin-3 and [Lys40(68Ga-DOTA)]exendin-3
The biodistributions of [Lys40(111In-DTPA)]exendin-3 and [Lys40(68Ga-DOTA)]exendin-3 were determined in BALB/c nude mice with subcutaneous INS tumours. Mice were injected intravenously with 370 kBq 111In (n = 5) or 1 MBq 68Ga-labelled exendin-3 (n = 5) at a peptide dose of 0.1 µg. To determine the GLP-1R-mediated uptake, a separate group of two mice were co-injected with an excess of unlabelled [Lys40]exendin. Another group of two mice were injected with [Lys40(68Ga-DOTA)]exendin-3 that was purified on a disposable PD10 column instead of by RP-HPLC. The mice were killed 1 h after injection by CO2 suffocation, and blood, tumour and major organs were removed, weighed and counted.
Peptide dose-escalation study
To determine the optimal peptide dose of exendin-3 in this model for tumour imaging, BALB/c nude mice with a subcutaneous INS-1 tumour were injected intravenously with 370 kBq [Lys40(111In-DTPA)]exendin-3 (peptide dose 0.003 µg) and were coinjected with various amounts of unlabelled [Lys40]exendin-3 to final peptide doses in the range 0.003–10.0 µg (five mice per peptide dose). The mice were killed 4 h after injection by CO2/O2 suffocation, and the radioactivity in the blood, tumour and relevant tissues was measured in a gamma counter.
PET images were acquired of BALB/c nude mice (n = 3) with a subcutaneous INS-1 tumour in the hind limb using a small-animal PET/CT scanner (Inveon; Siemens Medical Solutions, Knoxville, TN). The scanner had an axial field of view of 12.7 cm and a 1.4-mm FWHM spatial resolution in images reconstructed using Fourier rebinning and filtered back projection. When the tumours had a diameter of 2–5 mm, the mice were injected intravenously via the tail vein with 3 MBq HPLC-purified [Lys40(68Ga-DOTA)]exendin-3 (specific activity 110 GBq at completion of the labelling, 50 GBq/µmol at time of injection) corresponding with an exendin dose of 0.3 µg. PET images were acquired 1 h after injection under isoflurane/O2/N2O inhalation anaesthesia. The PET images were acquired over 45 min for the emission scan, and the images were reconstructed by OSEM3D/MAP reconstruction.
Specific activities up to 700 GBq/μmol for [Lys40(111In-DTPA)]exendin-3 were obtained with a radiochemical purity exceeding 95%. [Lys40(DOTA)]exendin-3 was labelled efficiently with 111In and 68Ga. Specific activities of 20 GBq/µmol with a radiochemical purity >95% as determined by HPLC were obtained for [Lys40(111In-DOTA)]exendin-3. [Lys40(DOTA)]exendin-3 was labelled with 68Ga with a specific activity of 110 GBq/µmol with a radiochemical purity >99% after HPLC purification. Without HPLC purification, a specific activity of 17 GBq/µmol was obtained. Unbound 111In and 68Ga eluted from the column after 3 min, whereas 111In- and 68Ga-labelled exendin had a retention time of about 13 min.
Biodistribution studies of [Lys40(111In-DTPA)]exendin-3, [Lys40(111In-DTPA)]exendin-4, and of [Lys40(111In-DOTA)]exendin-3
Biodistribution of [Lys40(68Ga-DOTA)]exendin-3
In contrast to the RP-HPLC-purified preparation, [Lys40(68Ga-DOTA)]exendin-3 purified by gel filtration on a Sephadex G-25 column showed enhanced uptake in the spleen and liver (16.8 ± 0.9%ID/g and 20.9 ± 1.6%ID/g, respectively). This enhanced hepatic and splenic uptake was not receptor-mediated as it could not be blocked by coinjection of an excess of unlabelled [Lys40]exendin (data not shown).Very low uptake in the liver and spleen was observed when RP-HPLC-purified [Lys40(68Ga-DOTA)]exendin-3 was injected (0.44 ± 0.05%ID/g and 0.25 ± 0.02%ID/g, respectively). Again, no specific uptake was observed in the liver and spleen as determined by coadministration of an excess of unlabelled exendin (0.31 ± 0.02%ID/g and 0.20 ± 0.01%ID/g, respectively).
Peptide dose escalation study
SRS, the standard method for imaging of NET, detects only around 50% of insulinomas [1, 2, 12]. Clinical studies with 18F-DOPA PET for the detection of insulinomas have shown promising results . However, physiological uptake of 18F-DOPA in the pancreas and duodenum may cause false-positive results or mask small lesions. An exendin-based imaging method targeting the GLP-1R specifically expressed on insulinomas might therefore improve the visualization of insulinomas.
Preclinical studies with [Lys40(111In-DTPA)]exendin-4 in different animal models have shown high accumulation of 111In-labelled exendin in insulinomas [12, 13]. Wild et al. demonstrated that in vivo imaging of insulinomas by SPECT with radiolabelled exendin is feasible in patients . In a transgenic mouse model, they detected insulinomas of less than 1 mm in diameter using [Lys40(111In-DTPA)]exendin-4 as tracer. Although SPECT has a higher resolution in small-animal imaging, for clinical imaging PET offers a higher spatial resolution, a higher sensitivity and the possibility to accurately quantify uptake, which makes it preferable to SPECT. Recent studies have shown that PET imaging of NET with 68Ga-labelled octreotide is superior to SPECT imaging with 111In-labelled octreotide [19, 20].
In this study we therefore investigated the potential of [Lys40(68Ga-DOTA)]exendin-3 for the detection of insulinomas by PET in a mouse model. Conjugation of [Lys40]exendin-3 with DOTA instead of DTPA allows radiolabelling of the tracer with a broad range of radionuclides including the positron emitter 68Ga. We selected 68Ga as positron emitter because it is a generator-derived radionuclide (and therefore readily available), the labelling strategy is fast and relatively simple, and the half-life matches the pharmacokinetics of the tracer (exendin) . A potential disadvantage of 68Ga for animal imaging may be the high energy of the emitted β+ particles limiting the spatial resolution of the images to 3 mm while the intrinsic resolution of the scanner is 1.5 mm . This is only a relevant issue in preclinical scanners, as in clinical scanners the intrinsic resolution of the scanner rather than the β+ energy will limit the resolution of the images.
Biodistribution studies in BALB/c nude mice with subcutaneous INS-1 tumours showed high and specific uptake of [Lys40(68Ga-DOTA)]exendin-3 in the tumour and pancreas. Subcutaneous INS-1 tumours in the hind limb of BALB/c nude mice were clearly visualized by microPET imaging. Therefore, our data show that imaging of insulinomas with 68Ga-labelled exendin is feasible.
[Lys40(111In-DTPA)]exendin-3 was used as a reference in all experiments. Biodistribution studies with [Lys40(111In-DTPA)]exendin-3 showed specific uptake in GLP-1R-expressing tissues as described previously . The peptide dose escalation study revealed that in this murine model a peptide dose of≤0.1 µg results in maximum accumulation of radiolabelled exendin in the tumour and in GLP-1R positive organs. Apparently, at higher doses in vivo receptors saturation occurs. The biodistribution of [Lys40(111In-DTPA)]exendin-3 was identical to the biodistribution of [Lys40(111In-DTPA)]exendin-4. Also there was no measurable effect of the chelator on receptor targeting as the biodistributions of [Lys40(111In-DTPA)]exendin-3 and [Lys40(DOTA)]exendin-3 were identical.
The uptake of [Lys40(68Ga-DOTA)]exendin-3 in the tumour and the pancreas was significantly lower than that of [Lys40(111In-DTPA)]exendin-3. It has been demonstrated that the affinity of a peptide can be influenced by the chelator as well as by the radionuclide. Reubi et al. showed that 68Ga-DOTA-TATE has a higher affinity for the somatostatin receptor subtype 2 than 111In-DOTA-TATE . However, there was no significant difference between the IC50 values of Lys40(natIn-DTPA)]exendin-3, [Lys40(natIn-DOTA)]exendin-3 and [Lys40(natGa-DOTA)]exendin-3. The differences in tumour and pancreatic uptake between [Lys40(68Ga-DOTA)]exendin-3 and [Lys40(111In-DTPA)]exendin-3 could not be attributed to the substitution of DTPA by DOTA, because in our mouse model, the uptakes of [Lys40(111In-DTPA)]exendin-3 and [Lys40(DOTA)]exendin-3 in the tumour and other receptor-positive organs were identical.
Enhanced uptake of [Lys40(68Ga-DOTA)]exendin-3 was seen in the liver and spleen when the preparation was only purified by gel filtration, probably because this preparation still contained some colloidal 68Ga. Colloidal 68Ga will accumulate in the liver and spleen. RP-HPLC purification removed the colloidal contamination, since this preparation had the same very low uptake in the liver and spleen as the 111In-labelled preparation. Previous studies with 68Ga-labelled DOTA-conjugated compounds not purified by RP-HPLC have also shown enhanced liver and spleen uptake [23, 24]. It is still a matter of debate whether the enhanced uptake in liver and spleen is caused by colloidal 68Ga or by instability of the 68Ga-DOTA complex. In the present study, uptakes of HPLC-purified [Lys40(68Ga-DOTA)]exendin-3 in the liver and spleen were identical to that of 111In-labelled exendin-3. Our data suggest that the enhanced uptake in the liver and spleen of 68Ga-labelled compounds found in previous studies is not caused by instability of the 68Ga-DOTA complex but by minor amounts of colloid.
To obtain high-quality PET images a 68Ga activity of 3 MBq was required. This necessitated the administration of an exendin dose of 0.3 µg, and although this resulted in suboptimal uptake (as demonstrated in the peptide dose escalation study), the tumours could exactly be visualized.
GLP-1R is also expressed on native pancreatic beta-cells. Radiolabelled exendin can potentially be used to target beta-cells. This is an attractive approach for the determination of beta-cell mass, a key issue in diabetes mellitus. [Lys40(111In-DTPA)]exendin and [Lys40(68Ga-DOTA)]exendin potentially can be used to noninvasively determine the beta-cell mass in the pancreas by SPECT or PET. A previous study has shown a correlation between the uptake of 111In-labelled exendin and the beta-cell mass , which makes this approach promising for the in vivo determination of the beta-cell mass.
In conclusion, 68Ga-labelled [Lys40(DOTA)]exendin-3 is a suitable tracer for PET imaging of GLP-1R-expressing insulinomas. 68Ga-labelled [Lys40(DOTA)]exendin-3 showed high uptake in subcutaneous INS tumours in BALB/c mice. Subcutaneous INS-1 tumours were clearly visualized with a microPET/CT scanner after injection of 68Ga-labelled [Lys40(DOTA)]exendin-3. Therefore, clinical studies should be conducted to elucidate the potential of [Lys40(68Ga-DOTA)]exendin-3 for insulinoma imaging in humans.
This work was supported by NIH grant 1R01 AG 030328-01.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.