Molecular imaging of aberrant crypt foci in the human colon targeting glutathione S-transferase P1-1

Aberrant crypt foci (ACF), the earliest precursor lesion of colorectal cancers (CRCs), are a good surrogate marker for CRC risk stratification and chemoprevention. However, the conventional ACF detection method with dye-spraying by magnifying colonoscopy is labor- and skill-intensive. We sought to identify rat and human ACF using a fluorescent imaging technique that targets a molecule specific for ACF. We found that glutathione S-transferase (GST) P1-1 was overexpressed in ACF tissues in a screening experiment. We then synthesized the fluorogenic probe, DNAT-Me, which is fluorescently quenched but is activated by GSTP1-1. A CRC cell line incubated with DNAT-Me showed strong fluorescence in the cytosol. Fluorescence intensities correlated significantly with GST activities in cancer cell lines. When we sprayed DNAT-Me onto colorectal mucosa excised from azoxymethane-treated rats and surgically resected from CRC patients, ACF with strong fluorescent signals were clearly observed. The ACF number determined by postoperative DNAT-Me imaging was almost identical to that determined by preoperative methylene blue staining. The signal-to-noise ratio for ACF in DNAT-Me images was significantly higher than that in methylene blue staining. Thus, we sensitively visualized ACF on rat and human colorectal mucosa by using a GST-activated fluorogenic probe without dye-spraying and magnifying colonoscopy.


Protein expression of candidate genes in human ACF tissues
Protein expression of the seven candidate genes (Glut-1, GSTP1-1, c-MET, β-catenin, Cadherin-1, SLC7A7, EGFR) in five human ACF tissue specimens were examined by immunohistochemistry. Representative results are shown in Supplementary Figure 2. Glut-1 stained strongly in the membrane of ACF cells. GSTP1-1 stained strongly, mainly in the cytosol of ACF cells. Moreover, Cadherin-1 and β-catenin were appreciably stained in ACF cells but were also stained in normal epithelia, indicating that protein expression by these genes in ACF was relatively weak. On the other hand, the staining signals for c-MET, SLC7A7 and EGFR were weak in ACF cells. Similar results were obtained in the remaining four ACF tissues.

Protein expression of GSTP1-1 and Glut-1 in rat ACF tissues
Expression of GSTP1-1 and Glut-1 proteins in five rat ACF tissues was examined by immunohistochemistry. Representative results are shown in Supplementary Figure 3. Glut-1 stained strongly in the membrane of ACF cells. GSTP1-1 also stained strongly, mainly in the cytosol of ACF cells. Similar results were obtained in the remaining four ACF specimens.

In vivo molecular detection of ACF in AOM-treated rats
We observed ACFs in vivo on the rectosigmoidal mucosa of five AOM-treated rats using a thin veterinary endoscope after enema administration of DNAT-Me. Representative images are shown in Supplementary Figure 4. ACF were clearly visualized at 10 min on the sigmoidal mucosa, then the signal became slightly attenuated at 20 min and almost disappeared at 30 min.
Similar results were obtained in the remaining four rats. These results clearly suggest that the topical application of DNAT-Me is able to visualize rat ACF not only in vitro but also in vivo.

Ex vivo molecular detection of human adenoma and cancer
We performed ex vivo molecular imaging of each adenoma in five patients and each adenocarcinoma in seven patients by DNAT-Me. Representative lesions of adenoma and adenocarcinoma are shown in Supplementary Figure 5. A strong fluorescence signal was observed in the entire adenoma lesion although a partly dappled appearance was apparent.
Likewise, a strong fluorescence signal was observed in the entire cancer lesion. Similar results were obtained in the remaining four adenomas and six cancers. Thus, human adenomas and cancers were clearly visualized ex vivo by DNAT-Me. These data support the theory of an ACF-Adenoma-Carcinoma sequence in colorectal carcinogenesis.

Human ACF tissues for immunohistochemistry
ACF tissues were obtained by biopsy under magnifying colonoscopy from five patients with colorectal adenoma or cancer, as described previously 13,14,39 . The specimens were fixed in 10% formalin, embedded in paraffin, and sliced into 5-µm sections for immunohistochemistry, as described previously 30 .

Rat ACF tissues for immunohistochemistry
Azoxymethane (AOM) was administered to five male F344 rats subcutaneously at a dose of 15 mg/kg once a week for two weeks. The animals were sacrificed at eight weeks, and the colorectum was removed and fixed in 10% formalin. After ACF observation under a stereomicroscope, the colorectal mucosa including ACF was excised, embedded in paraffin, and sliced into 4-µm sections for immunohistochemistry as well as for haematoxylin and eosin stain, as described previously 41 .

Immunohistochemistry
Immunohistochemical staining of human ACF tissues was performed using a streptavidin-biotin peroxidase method with labeled streptavidin-biotin (Dako, Tokyo, Japan), as Immunohistochemistry for GSTP1-1 and Glut-1 in rat ACF tissue was also performed using the streptavidin-biotin peroxidase method with labeled streptavidin-biotin (Dako). Rabbit anti-human GSTP1-1 polyclonal antibody (MSA-102, Assay Designs Inc.) and rabbit anti-human glut-1 polyclonal antibody (Abcam, Cambridge Science Park), which crossreacts with rat counterparts, were used as primary antibodies.

In vivo molecular imaging of ACF in AOM-treated rats
AOM was administered subcutaneously to five male F344 rats, as described above. At eight weeks, the rats were anesthetized by inhalation of 3% isoflurane. The colorectum was cleaned by saline enema and infused with 3 mL DNAT-Me (200 µM). A rigid miniature probe (AE-R16150, AVS Co.,Ltd., Tokyo, Japan) was inserted into the colorectum, and the colorectal mucosa was carefully observed. The fluorescence image was monitored chronologically using a prototype fluorescence imaging system that was developed by modifying a veterinary endoscopic system (Olympus VR Type 7142A, AVS Co.,Ltd.) with a blue excitation filter (WRATTEN Gelatin Filter No.47, blue 410-500nm, Kodak, Rochester, NY) and a yellow emission filter (WRATTEN Gelatin Filter No.12, yellow 510+, Kodak).

Ex vivo molecular imaging of human adenoma and cancer
Five patients with adenomas (three patients with sigmoid colon adenomas and two patients with rectal adenomas) were enrolled. The average age and male-to-female ratio were 58.7 ± 14.1 years and 3:2, respectively. Colorectal adenomas were removed by endoscopic mucosal resection. The specimens were fixed on rubber, washed once with phosphate-buffered saline  with PBS and sprayed with 20 mL of DNAT-Me (200 µM), as described in the text. They were observed using a fluorescent imaging system ( Supplementary Fig. 7).

Ethics
This study was approved by the institutional review board of Tokushima University Hospital (Tokushima, Japan), as described in text. All human samples were obtained from patients who

Supplementary Figure 3
Immunohistochemical analysis for Glut-1 and GSTP1-1 in rat ACF. Immunohistochemical staining was performed on 4-µm sections of ACF tissue using the LSAB method. Original magnification 100 x.

Supplementary Figure 4
In