Indium-111-labeled CD166-targeted peptide as a potential nuclear imaging agent for detecting colorectal cancer stem-like cells in a xenograft mouse model

Background Cancer stem cells (CSCs) are involved in drug resistance, metastasis, and relapse of cancers, which can significantly affect tumor therapy. Hence, to develop specifically therapeutic target probe at CSCs for improvement of survival and quality of life of cancer patients is urgently needed. The CD166 protein has been suggested to be involved in colorectal cancer (CRC) tumorigenesis and to be considered a marker for colorectal CSCs (CRCSCs) detection. In this study, therefore, we attend to apply a nuclear imaging agent probe, Glycine18-Cystine-linked CD166-targeted peptides (CD166tp-G18C), to detect the changes of CD166 level in a CRC xenograft mouse model. Results We isolated the CD166-positive cells from the HCT15 CRC cell line (CD166+HCT15) and evaluated their morphology and ability of clone formation, migration, protein expression, and drug resistance. The CD166-positive HCT15 cells display the CSCs characteristics. We discovered and designed a CD166-targeted peptide (CD166tp-G18C) as a targeted probe of CRC stem-like cell for cell binding assay. The CD166tp-G18C confirmed the CD166 protein targeting ability in CD166+HCT15 cells. The diethylenetriaminopentaacetic acid (DTPA)-conjugated CD166tp-G18C further was labeled with indium-111 (111In-DTPA-CD166tp-G18C) as nuclear imaging agent for imaging and bio-distribution analysis in vivo. Finally, we observed that the 111In-DTPA-CD166tp-G18C was significantly enhanced in tumor tissues of CD166+HCT15 xenograft mice as compared to the non-CD166tp-G18C control. Conclusions Our results indicated that the indium-111-labeled CD166tp-G18C may be served as a powerful tool for colorectal CSCs nuclear imaging in the CRC patients.


Background
Colorectal cancer (CRC) is the third most frequent occurring cancer in men and the second most frequent occurring cancer in women, with nearly 1.65 million new diagnosed cases and about 832,000 deaths in 2015 [1][2][3]. Although diagnostic techniques have improved dramatically, the rate of early diagnosis in CRC patients is still less than 40%. In the last decade, CRC has risen rapidly in Asia, and the cure rate has not improved remarkable, which leads to mortality increase and remains at about 12-14% in 5-year survival rate of patients with metastasis [4]. One possible cause of treatment failure is that the tumor site contains a small population of tumor-initiating cells termed cancer stem cells (CSCs). The CSCs presented high metastatic potential and contributed to chemoresistance and radioresistance through specific intracellular signal pathways, such as Wnt, Notch, and Hedgehogand [5]. The colorectal cancer stem cells (CRCSCs) had been identified in human CRC tumor tissues [6,7] and were considered to be the crucial targets for the effective treatment of CRC [8]. Several studies showed that CRCSCs-targeted therapy effectively reduced the chance of tumor recurrence and metastasis [9][10][11]. The accurately diagnostic tool contributes to the efficacy in assessment of CRCSCs maintenance and metastatic sites. Colonoscopy is a most versatile diagnostic technique in symptomatic individuals including colon cancer [12]. However, the CSCs in the tumor tissue biopsy are not easily removed by colonoscopy. The small population of CSCs may affect the results of biopsy and interpretation of subsequent analysis. Therefore, development of non-invasive nuclear medicine imaging with high specific biomarkers is suitable as a tool for CRCSCs detection.
CD166, also known as activated leukocyte cell adhesion molecule (ALCAM), is a member of the immunoglobulin superfamily of cell adhesion molecules for intercellular adhesion regulation [13,14]. It is a highly conserved 110-kDa type-I transmembrane glycoprotein, which comprises five extracellular Ig domains: two amino-terminal variable (V) type Ig domains followed by three constant (C) type Ig domains (V1V2C1C2C3), a lipophilic transmembrane domain, and a short cytoplasmic tail [15]. CD166 has been reported to be involved in many biological activities, such as T cell activation and proliferation, monocyte migration, axon fasciculation, angiogenesis, and hematopoiesis [16,17], and has been suggested to be correlated with aggressive disease phenotype in many cancers including CRC [18], esophageal cancer [19], melanoma [20], breast [21], ovarian [22], and prostate [23]. It is also one of the biomarkers associated with CRCSCs, which are responsible for CRC tumor initiation in xenografts, colony formation, and further enrichment [24,25]. In addition, CD166 also contributed to the survival time and cancer regression of patients with CRC in which the frequently upregulated CD166 can act as an independent prognostic marker in the progression of CRC [26,27]. These findings indicated that CD166 is a key factor for maintaining the characteristics of CRCSCs and possesses the potential as a prognostic biomarker in the clinical management of CRC patients. Therefore, development of a highly specific CD166targeted probe for estimating CD166 status is helpful for the optimal care in patients with advanced CRC.
Antibody-drug conjugates (ADC) have been widely used as the specific biomarker ligands for tumor targeting in clinical and preclinical studies [28,29]. However, the macromolecule antibodies might lead to an immune response, low tumor penetration, high accumulation in specific tissues/organs, and resistance [30]. Therefore, in order to overcome ADC limitations, the development of new specific targeting peptide ligands as imaging probe is a more effective strategy [31]. To date, the phage-displayed peptide library is a potential tool for identifying specific peptide ligands. The method has been applied in lymphocyte epitope mapping, selection of bioactive peptides bound to receptors/proteins, and development of drug delivery systems [32]. Recent studies have reported that the peptide selection method successfully discovered novel peptides as imaging probe for tumor detection [33,34].
In this study, we aimed to design an effective CD166targeted peptide as a nuclear imaging probe for CRCSCs detection. We developed an optimal CD166-targeted peptide (CD166tp) by using phage-displayed peptide library, which linked with Glycine 18 -Cysteine (G 18 C), and conjugated with diethylenetriaminopentaacetic acid (DTPA) for isotope (Indium-111) labeling. We used this CD166-targeted radiopharmaceutical for CD166 detection in a CRC xenograft mouse model.

Cell culture
The human CRC-derived cell line (HCT15) was obtained from the Bioresource Collection and Research Center (Hsinchu, Taiwan). Cells were cultured in RPMI-1640 medium (Thermo Fisher Scientific, Waltham, MA, USA) with 10% fetal bovine serum (FBS) at 37°C in a humid incubator with 5% CO 2 .

Animals
The 10-week-old male BALB/c nude mice were obtained from BioLASCO (Taipei, Taiwan) and housed under the standard condition (a 12:12-h light:dark cycle at 22 ± 2°C). Mice were supplied rodent laboratory chow diet and drinking water ad libitum. Animal protocols were approved (INER-107175) by the institutive ethical review committee and were in accordance with regulations in Taiwan and National Institute of Health guidelines on the care and welfare of laboratory animals.
Magnetic-activated cell sorting of CD166 cells HCT15 cells were treated with type I collagenase, and then the digested cells were collected and passed through a 40-μm nylon cell strainers (1 × 10 7 cells/mL). These cells were suspended in PBS and FcR blocking buffer and then added anti-CD166 human polyclonal immune-magnetic beads (Thermo Fisher Scientific). After washing bead-bound cells, the CD166-positive HCT15 cells (CD166 + HCT15) were obtained. The purity of both CD166 + HCT15 and CD166-negative HCT15 cells (CD166 − HCT15) was detected using standard flow cytometric analysis (see below).
Screening of CD166-targeted peptides by phage display The Ph.D.-12™ Phage Display Peptide Library Kit (New England Biolabs, Ipswich, MA, USA) was used for screening human CD166-targeted peptides according to the manual protocol of manufacturer. Briefly, the 20 μg human CD166 recombinant protein (Sigma-Aldrich) was coated on 96-well plate with 100 μL NaHCO 3 solution (0.1 M, pH 8.6) overnight at 4°C. Next day, each well was blocked with 250 μL blocking buffer (0.1 M NaHCO 3 , pH 8.6, 5 mg/mL BSA) for 2 h at room temperature (RT). All wells were washed 6 times with TBST (0.1% Tween-20, pH 7.5) for biopanning. For the phage selection procedure, each well was added with phage library (10 11 plaque forming unit (pfu)/100 μL in TBST with 0.1% Tween-20) and incubated for 1 h at RT. After washing 10 times with TBST (0.1% Tween-20), the bound phages were eluted with 150 μL elution buffer (0.2 M Glycine-HCl, pH 2.2, 1 mg/mL BSA) and neutralized immediately with 15 μL neutralizing buffer (1 M Tris-HCl, pH 9.1). Five microliters of eluted phages was then used for tittering. The remaining phages were amplified for subsequent rounds of biopanning (second to fourth rounds) that the detergents at the concentrations of 0.3% (second round), 0.5% (third round), and 0.5% (fourth round) were used. For the amplification procedure, the eluted phages were infected in chemically competent E. coli cells (ER2738) for 5 h at 37°C. After bacteriolysis, the phages were recovered from the culture supernatant by using centrifugation and PEG precipitation. The titer was determined by a blue/white colony screening on LB/IPTG/Xgal plate and used as input phages for the next cycle of biopanning. The phage recovery rate of each round was calculated as follows: titer of output phage/titer of input phage × 100%. After the final round of biopanning, the phage clones with higher binding affinity to CD166 protein were collected for DNA extraction. To the amplified phages, 100 μL were added iodide buffer (10 mM Tris-HCl pH 8.0, 0.1 mM EDTA, 4 M NaCl) by vigorously tapping the tube and added 250 μL 95% ethanol for 20 min at RT. After centrifugation, the pellets were washed with 0.5 mL 75% ethanol and briefly dried under vacuum. Finally, the pellets were resuspended in ddH 2 O for DNA sequencing. The resulting sequences were analyzed to identify homologous amino acid sequences (Table 1). To further select the best targeted peptide sequence, the phage clones (clone number 1, 2, 3, 4, 5, 7, 10, 11) corresponding to sequence peptide (P1-P8) were chosen for phage ELISA assay and cell-based phage ELISA assay.

Cell-based phage ELISA
Both CD166 + HCT15 and CD166 − HCT15 cells were used to evaluate the binding of selected phage clones on cell surface. Both cell lines were cultured in 96-well plates to 80% confluence and fixed with 4% paraformaldehyde. After blocking with BSA (5 mg/mL) for 2 h at RT, 10 11 individual phages were added to each well and incubated at 37°C for 2 h. After washing the plate with PBST for 6 times, the cell-bound phages were detected with HRP-conjugated M13-monoclone antibody (1:5000; Abcam) as described above.

Flow cytometry analysis
For CD166 detection on the cellular surface, the optimized density (1 × 10 6 cell) of CD166 + HCT15 and CD166 − HCT15 cells were added with 1 mL PBS with

Clone formation experiment
Both CD166 + and CD166 − HCT15 cells were separated into single cells (2000 cells/well) and plated into culture dishes (diameter, 6 cm) to grow for 16 days. The medium (MSC Nutristem® XF medium supplemented without FBS) was replaced every 3 days. The cell colonies were fixed with 10% neutral buffered formalin solution for 30 min and stained with 0.05% (g/L) crystal violet solution for 30 min.

Migration assay
The cells with 90% confluence in the six-well plate were gently created a horizontal wound in monolayers using a 200-μL sterile pipette tip. The scratch images were acquired at × 100 magnification at 0 h (T0) and 24 h (T24).
The migration distance was determined by using ImageJ software to detect the reduction of the wound gap.

Immunofluorescence staining
The CD166tp-G 18 C binding assay was determined by immunofluorescence staining in both CD166 − HCT15 and CD166 + HCT15 cells. Briefly, cells (1 × 10 4 cells) were seeded in 4-well Millicell EZ slides (Merck Millipore, Germany) with culture medium at 37°C and 5% CO 2 for 16 h. Cells were fixed using 4% paraformaldehyde for 30 min at 37°C. The fixed-cells were then incubated in the SuperBlock Blocking Buffers (Thermo Fisher Scientific, USA) for 30 min. Cells were incubated with CD166tp-G 18 C-FITC (20 μg/mL) for 2 h. DAPI was used to stain the cell nuclei at a concentration of 0.2 μg DAPI/mL PBS for 10 min. The images were captured by a fluorescence microscope (BX53, Olympus, Japan) coupled to a CCD camera. Digital images were analyzed by an Image-Pro Plus software (Media Cybernetics, USA).

CD166 fluorescence imaging and distribution in CRC xenograft mice
Both CD166 + HCT15 cells and CD166 − HCT15 cells were collected for subcutaneously inoculating into the flank of nude mice for 2 weeks. For the CD166 imaging assay, the Cy5.5-labeled CD166 antibody was intravenously injected into tail vein of both CD166 + HCT15 and CD166 − HCT15 xenograft mice. The fluorescent images of whole animal body were detected by an in vivo fluorescent imaging system (Perkin Elmer). Mice were sacrificed and the organs (brain, heart, lung, liver, spleen, kidneys, stomach, small intestine, colon, muscle, and tumor) were collected for fluorescent distribution imaging detection. In-DTPA-CD166tp-G 18 C (148 MBq/kg/mouse) were intravenously injected into mice for 2 weeks. After 2, 4, 24, and 48 h post-injection, mice were sacrificed and collected blood and organs including the brain, heart, lung, liver, spleen, kidneys, stomach, small intestine, colon, muscle, and tumor. The organ weight was measured. The radioactivity in blood and organs was determined by a 1470 WIZARD gamma counter (PerkinElmer). The percentage of injected dose per gram of tissue (%ID/g) was calculated.

Statistical analysis
Data were presented as mean ± standard deviation. Groups more than two were determined by one-way analysis of variance (ANOVA) followed by post hoc analysis with Bonferroni's test. The significant differences (p value < 0.05) among experimental groups were analyzed by GraphPad Prism V5.01 software (San Diego, CA, USA).

CD166 + cells have characteristics of CSCs and present drug resistance
The CD166 antibody-conjugated magnetic beads were used to isolate the CD166 + HCT15 cells (Fig. 1a). After the immunomagnetic sorting with CD166 beads, the CD166-expressed cells were determined by Western blotting and cytometry. The results indicated that the CD166 + HCT15 and CD166 − HCT15 cells could be separated by magnetic beads (Fig. 1b, c). CD166 was considered a biomarker for colorectal cancer stem cells (CSCs). We next tested the features of CSCs in these CD166 + cells using the in vitro sphere formation assay, clone formation assay, and migration assay. CD166 + HCT15 cells presented spheroids and higher colony-forming efficiency, and migration ability than that of CD166 − HCT15 cells (Fig. 2).
In addition, several studies have found that the expressions of Nanog, c-Myc, OCT4, and survivin are linked to cancer stem cells [35][36][37]. In this study, we found that CD166 + HCT15 cells showed significantly higher levels of Nanog, c-Myc, OCT4, and Survivin than that of CD166 − HCT15 cells (Fig. 3a). We also found that CD166 + HCT15 cells exhibited significant resistance to doxorubicin and 5-fluouracil as compared to CD166 − HCT15 cells (Fig. 3b).

Screening and identification of CD166-targeted peptides
In order to obtain the high-affinity target CD166 peptides, the phage display system was performed. The phage yields from each round of biopanning were used to indicate the enrichment of phages. The results showed that the fourth round of biopanning was highest for CD166 binding phages enrichment compared to third round of biopanning (Fig. 4a). The 30 phage clones were randomly picked from fourth round of biopanning for DNA sequencing, and obtained 8 corresponding amino acid sequences (P1-P8) ( Table 1). To compare the binding capacity of these peptides to CD166 protein, the 8 phage clones (clone number 1, 2, 3, 4, 5, 7, 10, 11), which carried different peptide sequences, were picked and their binding affinities for CD166 were detected using phage ELISA and Cell-based phage ELISA. The results indicated that the one selection phage clone (clone number 7) showed highest binding to CD166 protein and CD166 + HCT15 compared to other phage clones (Fig. 4b, c). Therefore, the P6 peptide (amino acid sequence: DSEGNSNLCSQS) was chemically synthesized for subsequent experiments.

Effects of CD166-targeted peptides on binding ability in CD166 + CRC cells
To provide chelating agent conjugation, reducing opsonization, and prolonging circulation time in vivo, the P6 peptide was extended an additional amino acid sequence (Glycine 18 -Cysteine, G 18 C). In order to understand the CD166 binding ability of CD166tp-G 18 C, the flow cytometric analysis was performed to detect the interaction between CD166tp-G 18 C and CD166 + HCT15 cells. The fluorescence of CD166tp-G 18 C-FITC was significantly increased in CD166 + HCT15 cells, but not in CD166 − HCT15 cells (Fig. 5a). Pretreatment with non-fluorescence of CD166tp-G 18 C markedly decreased the binding efficiency of CD166tp-G 18 C-FITC in CD166 + HCT15 cells (Fig. 5a). We further tested the binding ability of CD166tp-G 18 C-FITC in CD166 + and CD166 − HCT15 cells determined by immunofluorescence. The fluorescence imaging indicated that the binding ability of CD166tp-G 18 C-FITC was increased in CD166 + HCT15 cell, but not in CD166 − HCT15 cells (Fig. 5b). We further confirmed whether CD166tp-G 18 C-FITC targeted to the CD166 on the surface of CD166 + HCT15 cells. The immunoprecipitation assay was performed with CD166 antibody. The results revealed that CD166tp-G 18 C-FITC could bind to CD166 + HCT15 cells, but not CD166 − HCT15 cells, via interacting with CD166 protein (Fig. 5c). The increased CD166tp-G 18 C-FITC binding could be inhibited by nonfluorescence CD166tp-G 18 C pretreatment.

CD166 imaging in tumors and organs of CRC xenograft mice
In order to assess whether CD166tp-G 18 C can be used as a diagnostic probe for CD166 + colorectal cancer (CRC) in vivo, the CD166 + HCT15 and CD166 − HCT15 xenograft mouse models were established. The fluorescence signaling was remarkably increased in the tumor area of CD166 + HCT15 xenograft mice (Fig. 6a). However, weak fluorescence signaling was detected in CD166 − HCT15 xenograft mice (Fig. 6a). We further detected the bio-distribution of CD166 in several organs of CD166 + HCT15 xenograft mice. The fluorescence signaling was significantly and markedly increased in tumor tissue as compared to other organs in CD166 + HCT15 xenograft mice (Fig. 6b). The expression of CD166 protein was also markedly upregulated in tumor tissue compared to other organs in CD166 + HCT15 xenograft mice (Fig. 6c), which was consistent with the CD166 biodistribution imaging. Moreover, the CD166 protein was significantly overexpressed in tumor tissue of CD166 + HCT15 xenograft mice as compared to the CD166 − HCT15 group (Fig. 6d). Fig. 4 Screening and identification of CD166-targeted peptide. The phage display (12 phage library) was used to screen CD166 binding phages with four rounds of biopanning. a The titers of recovered phages from each round of biopanning were eluted by a blue/white colony screening on LB/IPTG/Xgal plate. The phage enrichment rate was calculated as output number/input number. b Phage ELISA assay. The carrying different peptide sequences of phage clones were exposed to CD166 protein and BSA (as a control) for detecting the binding affinity. c Phage binding to CD166 expression cells. The phage clones were further screened by cell-based phage ELISA. The CD166 + HCT15 and CD166 − HCT15 cells were positive and negative CD166 protein expression cells, respectively. Data are presented as mean ± SD (n ≥ 3) 111 In-DTPA-CD166tp-G 18 C as a diagnostic radiopharmaceutical for CD166 + CRC tumor imaging detection in mice In order to understand whether CD166tp-G 18 C can be used as a potential probe for CD166 detection in vivo, the radionuclide-labeled CD166tp-G 18 C was synthesized for detecting the CD166 + CRC in a xenograft mouse model. DTPA is one of the metal chelators for linking diagnostic probe and a corresponding radionuclides. Therefore, CD166tp-G 18 C was conjugated with DTPA for indium-111 labeling ( 111 In-DTPA-CD166tp-G 18 C). The flowchart of 111 In-DTPA-CD166tp-G 18 C synthesis and treatment method was shown in Fig. 7a. DTPA-CD166tp-G 18 C was successfully synthesized by conjugating the SH group of terminal cysteine of CD166tp-G18C with maleimide-DTPA. The molecular weight was determined by mass spectrometry (Fig. 7b). We further tested the labeling efficacy of Indium-111 onto the DTPA-CD166tp-G 18 C determined by instant thin-layer chromatography. The labeling efficacy of 111 In-DTPA-CD166tp-G 18 C was more than 98% (Fig. 7c). Therefore, the molar activity at the end of labeling procedure was about 11.05 GBq/μmol. We further examined the stability of 111 In-DTPA-CD166tp-G 18 C in sera of mouse, fetal bovine, and human during 144-h period. The result indicated that 111 In-DTPA-CD166tp-G 18 C in these sera was more than 98% during 144 h period (Fig. 7d).
To confirm the above results, the bio-distribution of 111 In-DTPA-CD166tp-G 18 C was detected in tumor tissue and other organs of CD166 + HCT15 xenograft mice. The results indicated that 111 In-DTPA-CD166tp-G 18 C offered the highest level of radionuclide signal in the tumor tissue compared to radionuclide-labeled control compounds ( 111 In-DTPA, 111 In-DTPA-G 18 C, and 111 In-DTPA-CD166tp-C) (Fig. 9). Moreover, the ratios of tumor-to-muscle (T/M) and tumor-to-blood (T/B) were evaluated in 111 In-DTPA-, 111 In-DTPA-G 18 C-, 111 In-DTPA-CD166tp-C-or 111 In-DTPA-CD166tp-G 18 Ctreated CD166 + HCT-15 xenograft mice. The ratios of T/M and T/B were remarkably increased in 111 In-DTPA-CD166tp-G 18 C-treated group as compared to radionuclidelabeled control compounds (Fig. 10). These results confirmed the targeted efficiency of 111 In-DTPA-CD166tp-G 18 C in CD166 + CRC tumor tissues.

Discussion
In this study, we successfully identified CD166-positive CRC cells (CD166 + HCT15) isolated from CRC cells and Fig. 6 The CD166 imaging in tumors and organs of CRC xenograft mice. a The CD166 imaging of both CD166 + HCT15 and CD166 − HCT15 xenograft model. Both CD166 + HCT15 and CD166 − HCT15 cells were subcutaneously inoculated into the right flank of nude mice for establishing CD166 highand non-expressed xenograft model, respectively. The Cy5.5-conjugated CD166tp-G 18 C was intravenously injected into both xenograft mice for 6 h and observed by in vivo imaging system. b The distribution of CD166 imaging in various organs and tumor tissue of CD166 + HCT15 xenograft mice. The top panel presented the CD166 imaging in various organs. The button panel presented the quantitation of fluorescent signals. Data are presented as mean ± SD (n = 3). *P < 0.05, versus other organs. c The protein expressions of CD166 in tumors and various organs of CD166 + HCT15 xenograft mice. d The CD166 expressions in tumors of both CD166 + HCT15 and CD166 − HCT15 xenograft mice. In c and d, protein expressions were determined by Western blotting and quantified by densitometry and normalized by GAPDH levels. In c, data are presented as mean ± SD (n = 3). *P < 0.05, versus other organs. In d, data are presented as mean ± SD (n = 3). *P < 0.05, versus 100 mm 3 (CD166 − HCT15); #P < 0.05, versus 150 mm 3 (CD166 − HCT15) the characteristics of the cancer stem cells. We also screened the optimal CD166 targeting peptide (CD166tp) and added an additional amino acid sequence (G 18 C) to form a nuclear imaging probe (CD166tp-G 18 C). The CD166tp-G 18 C was confirmed to possess the ability of targeting the CD166-positive CRCs, which directly and specifically bound to CD166 on surface of CD166-positive CRCs. The fluorescent-labeled CD166tp-G 18 C was remarkably increased in the tumor of CD166 + HCT15 xenograft mice compared to CD166 − HCT15 xenograft mice. Moreover, we developed a nuclear imaging agent ( 111 In-DTPA-CD166tp-G 18 C) using CD166tp-G 18 C as a probe for CD166-positive CRCs detection in a xenograft mouse model. In this xenograft model, when the tumor size achieved about 150 mm 3 which possessed about 1 × 10 7 CD166-postive cells (cancer cell average diameter: 15 μm), the nanoSPECT/CT detection started to perform. The difference of counts between background and tumor area was about 8 times. Therefore, we estimated that at least 1.25 × 10 6 (or volume 18.75 mm 3 ) CD166-postive cells were required for nanoSPECT/CT detection. These results suggest that 111 In-DTPA-CD166tp-G 18 C can successfully target tumor site of CD166-positive CRCs for tumor imaging in vivo.
The present study identified a CD166 high-affinity target-binding peptide by using phage display system, which possessed the powerful targeted peptide screening ability [38]. The standard operating procedure recommended to perform 3 rounds of biopanning. Here, we performed 4 rounds of biopanning for confirming the credibility of CD166 affinity and reducing the number of peptides. The data of phage recovery rate indicated that the fourth round of biopanning was more than 3 times of CD166 binding phages enrichment compared to third round of biopanning (Fig. 4a). Therefore, if only 3 rounds were performed, 24 peptide sequences would be selected to cause more complicated screening operations and error rate. This concept was supported by previous studies [39,40]. We have obtained a peptide sequence (P6) that has significantly greater affinity than other sequences (Fig. 4b, c) from assays of phage ELISA and cell-based phage ELISA. We further confirmed that the P6 peptide could effectively bind to CD166 protein with higher affinity in vitro and in vivo (Figs. 5 and 6). Moreover, Glycine (18)-Cysteine played the linking role between chelator and CD166 targeted probe. The uncharged Glycine is an ideal amino acid for extending peptide sequences. The sufficient length of linker is required for reducing the destruction of the spatial structure of the probe by the chelator and maintains high yields during peptide synthesis. We chose DTPA as a chelator, which not only could be labeled with Indium-111 for diagnosis, but also could be labeled with Lutetium-177 or Yttrium-90 for therapy [41].
The small population of cancer stem cells in tumor tissue is difficult to isolate from tumor tissue, which limits the ability to study the development and the pathogenesis of the cancer. In order to solve this problem, cell-based models of cancer stem-like cells were developed to extend our knowledge about CSCs, together with tumor biology, microenvironment, carcinogenesis, biomarker discovery, and improvement in oncologic therapies. CSCs can be obtained from tumor tissues or cancer cell lines through reprogramming [42,43], expression of specific surface markers [44,45], detection of the side population [46], selection of cells resistant to anoikis [47,48], or culture in specific conditions [49]. Previous studies have confirmed that CD166 is a cell surface marker of CRCSCs and suitable for CRCSCs isolation [50,51]. Therefore, we chose CD166 as a target of CRCSCs surface marker for CRCSCs The circled positions in images were quantified by a 3D analysis software. Data are presented as mean ± SD (n ≥ 3). *P < 0.05, versus control group. c The competitive study of 111 In-DTPA-CD166tp-G 18 C in CD166 + HCT15 xenograft mice. After tumor xenograft mice were intravenously injected with CD166tp-G 18 C (0, 10, and 50 mg/kg) for 6 h, 111 In-DTPA-CD166tp-G 18 C (740 MBq/kg/mouse) was intravenously injected into mice for 24 and 48 h and detected by a nanoSPECT/CT. d Quantification of nuclear images in tumor areas of colorectal tumor xenograft mice. Data are presented as mean ± SD (n ≥ 3). *P < 0.05, versus 0 mg/kg CD166tp-G 18 C group, #P < 0.05, versus 0 mg/kg CD166tp-G 18 C group isolation and identification in HCT15 cells. Fluorescenceactivated cell sorting (FACS) and magnetic-activated cell sorting (MACS) are two commonly used methods. FACS is based on fluorochrome-conjugated antibodies in direct or indirect immune fluorescence staining. MACS is an alternative isolation method via super-paramagnetic and biodegradable microbeads coated with specific monoclonal antibodies that enrich a specific antigen cell type from a mixed population. However, FACS has some disadvantages such as high cost, low viability of recovered cells, and difficulty of using the complex equipment. Therefore, in this study, MACS was performed to collect high purity of CD166 + cells from HCT15 cells.
Some non-invasive molecular imaging technologies, such as optical techniques and nuclear imaging have been developed as detection tools for CSCs monitoring and tracking [52]. The macroscopic fluorescence and NIR imaging have been developed for CRCSC detection in vivo [53,54]; however, the accumulation rate of agents was low and high background signaling was exhibited. Nuclear medicine imaging improved the limitation of tissue penetration in optical techniques and showed to be highly sensitive. In this study, we selected nuclear medicine imaging technique (single-photon emission computed tomography/computed tomography, SPECT/CT) and obtained the clear imaging and high accumulation rate more than 10 %ID/g at 4 h. Even though positron emission tomography/computed tomography (PET/CT) provides a much better sensitivity and spatial resolution than SPECT, the cost effectiveness and the greater availability of SPECT/CT scanners at most clinical centers contribute to more usage frequency of SPECT imaging than PET imaging [55].
Recently, the CD133-positive colorectal cancer stemlike cell was detected in vivo by using nanoSPECT/CT [56]. Liu et al. indicated that the radiolabeled CD133 targeted tracer ( 99m Tc-SHNH-AC133) showed high tumor uptake (8.82 ± 0.73%ID/g) and high tumor-to-muscle ratios (13.18 ± 2.84) at 36 h. Jin and colleagues have also developed the Iodine-125-labeled ANC9C5 as a radionuclide  111 In-DTPA-G 18 C and 111 In-DTPA-CD166tp-C) (148 MBq/kg/per mouse) intravenous injection for 2, 4, 24, and 48 h, CD166 + HCT15 xenograft mice were sacrificed and obtained tumor, blood, and organs. The radioactivity of samples was detected by a gamma counter. Values are presented as the percentage of injected dose per gram organ (ID%/g). Data are presented as mean ± SD (n ≥ 3). *P < 0.05, versus control group imaging agent for CSCs detection in colon carcinoma xenografts. But they obtained an unfavorable biodistribution profile [57]. The CD166-targeted agent that we developed in this study showed higher tumor uptake (11.1 ± 2.2 %ID/g, 24 h) and tumor-to-muscle ratios (48.9 ± 5.03, 48 h) in CRCSCs xenograft. Compared to current CRCSCs nuclear imaging, we considered that the high specificity and good tumor targeting properties of 111 In-DTPA-CD166tp-G 18 C may provide an optional imaging for tracking CRCSCs.
To accurately carry the drug to the lesion area, many molecules with specific binding ability, such as antibodies, anti-peptides, and pharmacological inhibitors were developed as probes for targeting therapy and diagnosis. Demcizumab and vanticumab, the IgG2 monoclonal antibodies, blocked formation of active receptor signaling complex for CRCSCs therapy in clinical trials In-DTPA, 111 In-DTPA-G 18 C, 111 In-DTPA-CD166tp-C, and 111 In-DTPA-CD166tp-G 18 C treated-mice after 2, 4, 24, and 48 h injection were shown. Data are presented as mean ± SD (n ≥ 3). *P < 0.05, versus 111 In-DTPA, 111 In-DTPA-G 18 C and 111 In-DTPA-CD166tp-C [59]. A FDA-approved drug Bevacizumab has been shown that it can target the vascular endothelial growth factor and possesses the potential against CRCSCs [60]. In addition, Gaedicke and colleagues have developed the antibody-based agent ( 64 Cu-NOTA-AC133) for CD133 + CSCs detection in subcutaneous and orthotopic glioma xenografts [61]. The antibody-based molecular probes were commonly used for cancer imaging, but the size limitation (~150 kDa) and immunogenicity made the antibodies no longer widely available for developing in vivo imaging [62]. In contrast, the peptidebased small-molecule probe that we developed in this study has the advantages of less immunogenicity, faster rates for tumor localization, convenient synthesis, and easier conjugation process. In the present study, comparing the ratio of tumor/blood between 111 In-DTPA-CD166tp-C and 111 In-DTPA-CD166tp-G 18 C (Fig. 10), we found that the CD166-targeted peptide with additional amino acid sequence improved the excretion rate (tumor/blood) of imaging agent in vivo and increased the bio-distribution in the tumor.
Di Stefano et al. have shown that the expression and localization of survivin, which plays the roles in cell death counteraction and mitotic progression control in CRCSCs, can be regulated by interleukin (IL)-4 [35]. The c-Mycassociated signaling has also been found to play the roles in CRCSCs regulation, chemotherapy resistance, and CRC organoids [36]. The induction of an embryonic stem celllike transcriptional program consisting of specific reprogramming factors, such as Oct4, Nanog, Sox2, and KLF4, has been suggested to be involved in hypoxia-inducible factors (HIFs)-induced stem-like phenotype in cancer progenitors from multiple solid tumors during tumor hypoxia [37]. In the present study, we found that CD166 + HCT15 cells showed significantly higher levels of Nanog, c-Myc, OCT4, and survivin than that of CD166 − HCT15 cells, indicating that CD166 + cells have characteristics of CSCs.

Conclusions
We demonstrated that CD166-positive CRC exhibited characteristics of CSCs, suggesting that it may be a useful drug screening tool for CRC diagnosis. The CD166tp-G 18 C compound specifically targeted to CD166-positive CRCs in vitro. We successfully synthesized DTPA-CD166tp-G 18 C and radiolabeled with Indium-111 for detecting CD166 imaging by using nanoSPECT/CT in CD166-positive CRC xenograft mice. The bio-distribution of 111 In-DTPA-CD166tp-G 18 C confirmed the accumulation of CD166-positive cells in tumors. Therefore, 111 In-DTPA-CD166tp-G 18 C may be a potential nuclear imaging agent for diagnosis of CRCSCs. The CD166 bound peptide-based nuclear imaging may provide physicians to classify cancer cells before treatment and monitor patients with a history of CRC after surgery or drug treatment.