Colorectal cancer (CRC) is the third most common cancer, affecting males and females usually at ages above 50 years.1, 2 Worldwide, as many as 1,000,000 new cases are diagnosed annually. Early stages of the disease are asymptomatic, thus many patients are diagnosed at progressed stages, when the average 5-year survival rate is only ∼60%. However, if diagnosed and treated in its early stages, CRC is curable. Obviously, early detection may significantly reduce mortality from CRC.3, 4
Despite considerable technological progress in the development of new CRC screening methods (e.g., laboratory-based diagnostics, CT-colonography and MR-colonography), limitations in sensitivity, specificity and patient acceptance constitute barriers to wider clinical application.5 Currently, routine white-light optical colonoscopy is considered by many to be the gold standard for CRC diagnosis.6 However, identification of small polyps and neoplasms with nonpolypoid morphology (flat and depressed), representing a significant proportion of cases,7, 8 is a serious problem. To improve visualization of surface features of the intestine and the detection abilities of this technique, addition of colorizing agents to the gastrointestinal tract can be applied.9, 10 However, this method is also not specific enough; in particular, inflammation can lead to false-positive results.11 Contrast agents with specific binding to malignant cells are desirable to significantly enhance diagnostic accuracy. Somatostatin (SST) based targeting seems to be a promising route toward improved CRC detection.
SST plays an important role in controlling biological functionality of the gastrointestinal tract via membrane coupled somatostatin receptors (SSTRs).12 Messenger ribonucleic acids (mRNAs) of 5 SSTR subtypes are found in the gut of rodents13 and humans.14, 15 In colon, SST regulates cellular proliferation, ion transport, secretory activity and motility.16–19 Scintigraphy, radiographic imaging with radiolabeled SST analogs, was shown to be efficient for detection of human neuroendocrine gastro-entero-pancreatic tumors,20 including colorectal carcinoids.21
However, dissimilar results are published about the expression of SSTRs in CRC. For example, Laws et al. detected SSTR2 mRNA in nearly 90% of 32 examined colonic tumor samples.22 Similarly, Vuaroqueaux et al. showed that 97% from 53 samples of primary colorectal tumors expressed at least one SSTR mRNA subtype.23 Radulovic et al. found high affinity of radiolabeled SST-14 to all 5 SSTRs, in 8 of 15 (53%) specimens of CRC,24 while in the study of Iftikhar et al., only 12 from 50 samples (24%) of primary CRCs were SSTR positive.25 Raggi et al. using real-time RT-PCR showed measurable amounts of SSTR2 mRNA in 100% of 100 colorectal carcinoma tumors, whereas preoperative imaging of CRC patients with 111In-pentetreotide depicted an increased tumor-to-normal tissue (T/NT) ratio of the radiolabeled tracer only in 7 of 17 cases.26
Previously, we demonstrated highly selective targeting of SSTR-positive human lung cancer in a mouse model, by optical methods, using a novel fluorescent conjugate of a synthetic SST analog.27 In our study, we investigated the potential of this diagnostic agent, conjugate 3207-86, in the detection of human colon cancer.
Female BALB/c nude mice, ∼8 weeks of age, were purchased from Harlan Ltd. (Jerusalem, Israel) and maintained under pathogen-limited conditions. Food and water were supplied ad libitum. Tumors were induced by subcutaneous injection of 1.5 × 107 HT-29 human colon carcinoma cells in saline suspension into the groin area of the mice. Experiments were started at days 10–14 postinjection, as soon as tumors were palpable. Experimental procedures were performed on animals anesthetized by intraperitoneal injection of a combination of 80/20 mg/kg ketamine/xylazine in saline. All experiments were conducted in compliance with regulations of the Animal Welfare Committee at the Sheba Medical Center.
Synthesis and characterization of conjugate 3207-86
The fluorescent SSTR-specific conjugate 3207-86 consists of a backbone cyclic core peptide, Phe-Cys-Phe-Trp-(D)Trp-Lys-Thr-Phe-Gly-NH2, with a building unit type GlyS2 (denoting glycine with a 2-carbon linker bearing a thiol group), covalently conjugated to a fluorescein moiety via γ-aminobutyric acid as a linker.
The compound was synthesized by DeveloGen Israel Ltd and purified by Novetide Ltd (Israel). Design, synthesis, purification and structure-activity relations have been described in detail.27
Briefly, assorted combinations of peptide-linker-fluorophores were synthesized using an automated plate synthesizer. Each peptide was conjugated to a fluorescein derivative by a linker via the amino terminal group. The peptides were purified using reversed phase (C-18 column) high performance liquid chromatography, which was later scaled-up for purification of the lead compound. Characterization was done using chromatography, electrospray ionization mass spectrometry and spectrofluorometry. The purified substance was stored at −20°C until used.
The appropriate amount was weighed and dissolved in dimethyl sulfoxide (DMSO) (Sigma, St. Louis, MO) to form a stock solution. Before injection, a further 1:20 dilution was made in saline (final concentration of DMSO in the injected solution is 5%).
Conjugate 3207-86 was administered as a bolus injection (6 mg/kg body weight, ∼200 μl/mouse) in a lateral tail vein of an HT-29 tumor-bearing mouse. At various time periods after administration, a fascial-cutaneous flap was elevated in the tumor area location. After imaging, animals were sacrificed.
Low-magnification fluorescence microscopy was performed using an Olympus (Japan) model SZX-12 fluorescence microscope, equipped with a digital color camera (Olympus DP50) and a standard SZX-MGFP filter-cube unit (Olympus).
Spectrally-resolved fluorescence imaging was carried out using a spectral imaging (SI) system (SD300, Applied Spectral Imaging, Migdal Ha'Emek, Israel).
The system comprises a Sagnac (triangular) common path interferometer placed in front of a Peltier-cooled black-and-white digital camera. Dedicated software converts interferograms into spectra by Fourier transformation and creates a SpectraCube™ file. The file is displayed as a false-color image in which the colors are adapted to the spectral wavelength scale. Each pixel has spatial and spectral characteristics; the spatial resolution is down to ∼10μm/pixel (depending on the magnification of the objective) and the spectral resolution is ∼3 nm.
Fluorescence was excited at 455 nm ± 20 nm from a xenon-lamp light source (MultiSpec, SeNet, Israel) in combination with a 450FS40-25 band-pass filter (Andover, Salem, NH). Macroscopic (total-body and close-up) spectrally-resolved images were acquired by the system using Nikkor (Nikon, Japan) objectives via a long-pass cut-on filter (type GG495, Rolyn Optics, Covina, CA). Images were analyzed qualitatively and quantitatively by SpectraView™ software (Applied Spectral Imaging).
High-magnification fluorescence imaging of tumor samples was performed by confocal laser scanning microscopy (CLSM), using a Zeiss (Oberkochen, Germany) model CLSM 410 confocal laser scanning microscope and excitation at 488 nm. For this purpose, whole mount tumor sections of a few mm thickness were prepared. Additional technical details of this methodology and analysis have been previously described.28
HT-29 tumor-bearing mice (5 mice per group) were sacrificed at 24, 48 and 72 hr after administration of the agent (6 mg/kg body weight, ∼200 μl/mouse).
Excised tissue samples were studied ex vivo using fiber-optic spectroscopy. For this purpose, an argon ion laser (43 series, Melles Griot, Carlsbad, CA) was coupled to one leg of a bifurcated fiber bundle (model 77533, Oriel, Stratford, CT), while the second leg was mounted in front of the entrance slit of a spectrofluorophotometer (model RF-5301PC, Shimadzu, Japan). The common-end tip of the bundle was fixed at a distance of 1 mm from the tissue sample. Fluorescence spectra were recorded in the range of 500–700 nm at 488 nm excitation. Three recordings at different locations were averaged for each data point. Tumor-to-normal tissue (T/NT) accumulation ratios of the agent were determined as the corresponding ratios of fluorescence intensities at 525 nm between tumors and normal tissue samples.
Conjugate 3207-86 dissolved in saline with 5% DMSO was administered intraperitoneally into BALB/c mice (Harlan, Jerusalem, Israel) at escalating doses of 100, 200, 300, 500 and 1000 mg/kg body weight (5 mice per group). Three control mice were injected with vehicle (10% DMSO in saline solution). Animal behavior and survival were checked daily for 20 days.
Visualization of the conjugate 3207-86 biodistribution was investigated in HT-29 colon carcinoma bearing mice using various imaging techniques (regular fluorescence microscopy, SI and CLSM) at low- and high-magnification levels.
Low-magnification microscopy at the time points 1, 3 and 6 hr postadministration did not reveal differences in fluorescence intensity between tumors and normal tissues (e.g., skin plexus) at the site of tumor location (data not shown). After prolonged time periods (≥24 hr), enhanced uptake of conjugate 3207-86 by tumors was observed with considerable contrast between tumor and normal tissues.
Figure 1 shows standard color and corresponding fluorescence images of a subcutaneously growing tumor nodule in vivo (skin flap was elevated 24 hr after administration of the agent). Bright fluorescence of the conjugate is observed in the tumor area, while the muscles and blood vessels do not fluoresce.
In vivo fluorescence images of agent-injected mice were acquired and analyzed using the SI technique. Tumor nodules were easily detected in total-body images of the mice. A high-intensity fluorescence signal at 520 nm, specific for conjugate 3207-86, was identified in the tumors. Figure 2a is a typical fluorescence image obtained with the SI system. Selective accumulation of the conjugate is seen in the tumor area. Red fluorescence, peaked at 670 nm (spectrum not shown), which is due to porphyrinic food additives, is registered in the intestine. Slight green fluorescence is observed in skin. Other tissues do not fluoresce significantly. Figure 2b depicts the same image after spectrally resolved image processing using a spectral unmixing algorithm.
Figure 3a shows spectral images of excised sections from normal colon and tumor nodule, while Figure 3b demonstrates corresponding spectra in the region of interest. In these samples, a 5-fold higher fluorescence intensity at 525 nm is detected in tumors compared to colon tissue. The presence of porphyrins in colon is also noted.
CLSM of excised tumors revealed that the fluorescence is localized in the tumor cell cytoplasm as well as in intratumoral microvessels (Fig. 4).
To assess quantitatively biodistribution patterns, the fluorescence intensity of the agent at 525 nm was measured, using fiber-optic spectroscopy, in excised tissue samples at 24, 48 and 72 hr after administration of conjugate 3207-86. At these time points, fluorescence in the tumor samples was significantly higher than in all normal tissues resulting in positive T/NT ratios (Table I). From all normal tissues, skin and colon fluorescence were maximal. Twenty-four hour after injection, T/NT ratios varied from 4–5-fold for skin and colon to 50–70-fold for liver, kidney and spleen, which were the lesser fluorescent tissues. At longer time periods, T/NT values have a tendency to increase.
Table I. In Vivo Biodistribution of Conjugate 3207-86 in Nude Mice after Intravenous Administration of 6 mg/kg Body Weight
Time after administration (hr)
Tumor/normal tissue ratio (mean ± SEM)
Fluorescence intensity ratios at 525 nm, excitation at 488 nm. (Fiber-optic spectroscopy measurements).
5.4 ± 0.6
4.7 ± 0.7
16.2 ± 3.5
17.5 ± 3.9
47.1 ± 8.2
59.1 ± 9.5
71.0 ± 11.3
5.9 ± 0.7
4.6 ± 0.9
21.2 ± 3.2
17.2 ± 4.2
55.0 ± 12.2
63.6 ± 14.9
88.3 ± 13.5
8.4 ± 1.1
4.3 ± 0.8
28.0 ± 4.7
23.9 ± 5.6
62.9 ± 18.5
84.2 ± 16.6
93.3 ± 18.3
No evidence of change in animal behavior and mortality was observed after administration of conjugate 3207-86 at all tested doses, up to 1000 mg/kg body weight. This dose, which is 160-times higher than our diagnostic dose, was well tolerated by all mice.
Significant variations and even controversial data regarding the expression profile of SSTRs in excised colorectal tumors are reported in the literature.22–25, 29, 30 These variations may be associated with technical peculiarities of the methods used for SSTR identification, or with real differences in SSTR expression between types of colorectal tumors, as well as differences among tumors of the same type. Cellular heterogeneity of CRC may be an important additional factor. Indeed, CRC is classically specified as a tumor type of nonneuroendocrine origin,31 yet endocrine cells and their hormone products were found in a fraction of colorectal adenomas and adenocarcinomas.32, 33
Using various optical techniques, we clearly demonstrated selective accumulation in vivo of the SSTR targeted conjugate 3207-86 in HT-29 tumor xenografts. This selectivity was time-dependent. Preliminary studies showed that during the first few hours after injection of the agent, background fluorescence from normal tissues at the site of tumor location was relatively high, which may be attributed to a high concentration of the compound in blood. At prolonged time periods (24–72 hr), T/NT ratios increased, because of retention of the conjugate in the tumor and clearance from normal tissues.
Obviously in vivo pharmacokinetics is a specific characteristic of any drug. The biodistribution patterns of the 3207-86 conjugate differ from other fluorescent34 or radiolabeled35 SST analogs, which demonstrate faster tumor uptake. Nevertheless, a 24-hr time point is generally useful for diagnostic SSTR imaging.20, 21, 35
When in vivo low-magnification fluorescence microscopy was used 24 hr after 3207-86 administration, it was noted that relatively large blood vessels surrounding a tumor nodule did not fluoresce against the background of normal and tumor tissues (Fig. 1). In contrast, ex vivo CLSM demonstrated fluorescence of the agent in both tumor cells and in tumor microcapillaries (Fig. 4). This discrepancy may be explained by significant absorption of excitation light and the fluorescence signal by blood in thick vessels throughout in vivo imaging procedures. Moreover, real differences may exist between SSTR expression in normal and intratumoral vasculature.36 The presence of SSTR1, SSTR2 and SSTR5 subtypes has been identified in growing endothelial cells, and in peritumoral and intratumoral capillaries.14, 37–40 This phenomenon is explained by participation of SST in the angiogenic switch from resting (quiescent) to proliferating status of endothelium.36
The high binding specificity of the 3207-86 conjugate to SSTR2 was established previously by a standard radioligand binding inhibition assay using membranes of CHO-K1 cells, stably expressing SSTR2.27 Considering the high affinity of conjugate 3207-86 to SSTR2, it may be hypothesized that vascular binding of the agent is realized also via this receptor subtype.
The HT-29 cell line has been shown to have specific binding sites for SST and to be sensitive to SST analogs, which caused characteristic enzymatic (e.g., stimulation of tyrosine phosphatase) and antiproliferative effects.24, 41–43 Buscail et al. found the presence of human SSTR1 mRNA in these cells.44 Further studies identified additionally a high-level expression of human SSTR5 subtype.14, 41
On the basis of literature data, we may expect the involvement of SSTR1 and/or SSTR5 in the mechanism of agent uptake by HT-29 tumor cells by receptor mediated endocytosis. In fact, conjugate 3207-86 may have a multi-receptor specificity, similar to other SST analogs.19, 45, 46
Obviously, the T/NT ratio of a diagnostic agent is one of the main parameters characterizing its detection ability. The biodistribution data showed that conjugate 3207-86 is accumulated in tumor tissue with significant selectivity. Specifically, at 24 hr, the fluorescence signal from tumor was 5-fold higher than from normal mouse colon, while at 72 hr this difference increased to 8-fold, mainly as a result of continuing decrease of fluorescence in normal colon tissue. Interestingly, the 24-hr T/NT ratio of the conjugate in our mouse xenograft model, is comparable to the 2–4.5-fold difference between excised human colon tumors and normal colon tissue in expression of SSTR5 mRNA, found in a clinical study by Vuaroqueaux et al.14 It should be noted that despite the only 81% homology in coding sequence between human SSTR5 and SSTR5 of BALB/c mice, the affinities of some SST analogs to human SSTRs and mouse SSTRs may still be similar.47
We suggest that selective uptake of conjugate 3207-86 by HT-29 tumors may probably be attributed to increased binding of the agent to the tumor cells (related to SSTR overexpression), as well as to increased angiogenesis, which is characteristic of CRC.48 Intracellular internalization of the agent after membrane binding is also one of the important factors providing selectivity of tumor accumulation. In view of recent data demonstrating that expression of SSTRs may serve as a marker of cell differentiation and prognosis for CRC,29, 49 the application of fluorescent SSTR specific agents may be important, not only for early detection but also for diagnosis/prognosis for colon cancer patients.
In conclusion, high selectivity in addition to low toxicity of the conjugate 3207-86 makes this agent a promising candidate for CRC screening. Clearly, more comprehensive studies including development of a water-soluble drug formulation and investigation of receptor specificity using other tumor models are needed to prove that conjugate 3207-86 is suitable for general SSTR-based fluorescence diagnosis of colon cancer.
In summary, human colon adenocarcinoma HT-29 was selectively targeted in mouse xenografts by a novel fluorescent SST-analog. Preferential tumor uptake of the agent was demonstrated by optical imaging and spectroscopic techniques allowing quantitative analysis. Total-body fluorescence imaging clearly showed selective accumulation of the conjugate in subcutaneously located tumors. CLSM of tumor nodules revealed fluorescence staining of cellular components and of intratumoral capillaries.
The authors thank Ms. Sharona Salomon (Sheba Medical Center) for animal care and for professional technical assistance.