Optical in vivo imaging detection of preclinical models of gut tumors through the expression of integrin αVβ3

Optical imaging and Fluorescent Molecular Tomography (FMT) are becoming increasingly important for the study of different preclinical models of cancer, providing a non-invasive method for the evaluation of tumor progression in a relatively simple and fast way. Intestinal tumors, in particular colorectal cancer (CRC), represent a major cause of cancer-related death in Western countries: despite the presence of a number of preclinical models of intestinal carcinogenesis, there is a paucity of information about the possibility to detect intestinal tumors using fluorescent probes and optical in vivo imaging. Herein, we identify the detection of integrin αvβ3 by FMT and optical imaging as an effective approach to assess the occurrence and progression of intestinal carcinogenesis in genetic and chemically-induced mouse models. For this purpose, a commercially available probe (IntegriSense), recognizing integrin αvβ3, was injected in APC+/min mice bearing small intestinal adenomas or CRC: FMT analysis allowed a specific tumor detection, further confirmed by subsequent ex vivo imaging or conventional histology. In addition, IntegriSense detection by FMT allowed the longitudinal monitoring of tumor growth. Taken together, our data indicate the possibility to use integrin αvβ3 for the visualization of intestinal tumors in preclinical models.


INTRODUCTION
Colorectal cancer (CRC) is a frequent human neoplasia and still one of the major leading causes of cancer-related death, especially in Western Countries [1]. Both transplanted orthotopic and spontaneous mouse models of colorectal cancer are available, representing a fundamental tool to understand the biological mechanisms of tumor progression and to evaluate the efficacy of potential therapeutic approaches. However, to assess disease state without sacrificing the animals, only indirect parameters are commonly used, such as animal weight loss, food and water uptake, stool consistency or the presence of occult blood in the feces. On the other hand, invasive procedures such as conventional optical endoscopy can be operated, but with the risk of perforating the colon and bleeding, thus impacting on the quality of the analysis [2]. In addition, a 6-27% miss rate recognition of small polyps have been estimated when samples are analysed by white light colonoscopy [3]. Different molecular in vivo imaging methods are now available, allowing the detection and the non-invasive characterization of tumor development in living animals, reducing the number of mice required per experiments and increasing the statistical power of the analysis. In preclinical oncological studies, in addition to conventional imaging techniques Research Paper www.oncotarget.com such as positron emission tomography (PET), magnetic resonance imaging (MRI), computed tomography (CT) and ultrasounds, optical imaging procedures can be used for the visualization of biological processes in small animal models expressing specific reporters (luciferase or fluorescent proteins) or injected with specific fluorescent probes [4][5][6][7]. However, a major drawback of conventional two-dimensional optical imaging is that the acquired signal is attenuated as much as the light source is deeper embedded in tissue. In this context, Fluorescent Molecular Tomography (FMT), a relatively new optical imaging technology, significantly improves the quality and the spatial resolution of optical imaging analysis, being able to localize and, more importantly, quantify fluorescent probes three-dimensionally in deep tissues at high sensitivity [8,9].
The characterization of processes involved in neoplastic progression and the description of tumor features are of primary importance to determine the correct cancer staging and to identify the best prognostic, predictive or therapeutic approach for oncologic patients. In this scenario, the expression of integrins, has been demonstrated to influence the prognosis in various cancer types [10]. Integrins are small transmembrane heterodimeric glycoproteins involved in cell adhesion, cell-matrix interaction and cell signalling pathways, whose expression has been correlated with tumor cell migration and metastatic processes [11]. Among members of this family, integrin αvβ3 is one of the most extensively studied and has been associated with the early phase of angiogenesis in a variety of tumors, including colorectal cancer [12,13]. In addition, studies demonstrated the efficacy of targeting integrin αvβ3 for the imaging identification and therapeutic treatment of several tumors, such as breast cancer, lung cancer and glioblastomas [14,15].
In this study, we provide evidence that imaging of integrin αvβ3 can be successfully used for the visualization and detection of intestinal tumors in preclinical models. The methodology described here may represent an evaluable tool to analyse the pathogenesis of intestinal tumors and therapeutic drug development.

In vivo detection by αvβ3 integrin expression of tumor formation in the small intestine
To test the efficacy of targeting integrin αvβ3 to visualize spontaneous genetic model of small intestine tumors, we took advantage of the APC +/min (APC, Adenomatous Polyposis Coli) mouse model. APC +/min mice are characterized by a single point mutation in the murine homolog of the APC gene. These mice develop multiple adenomas in the small intestine, representing a very powerful and commonly used model to study intestinal carcinogenesis [16]. In addition, APC +/min mice were previously crossed with mice lacking the expression of the chemokine receptor CX 3 CR1, known to have a role in regulating intestinal inflammation and whose absence induces intestinal carcinogenesis [17][18][19][20], in order to generate mice either competent (APC +/min -CX 3 CR1 +/-) or not (APC +/min -CX 3 CR1 -/-) for CX 3 CR1 expression and carrying concomitantly the mutation in APC gene. The double-mutant mice allowed us to compared not only tumor-bearing mice with healthy subjects, but also mice with a different tumor-load, due to the CX 3 CR1-dependent phenotype.
The occurrence of small intestine tumors was analysed by histology in 18 weeks-old APC +/min mice and age-related WT mice ( Figure 1A). Subsequently, we investigated the expression of integrin αvβ3 in small intestine tumors. We excised tumor lesions or corresponding healthy mucosa from APC +/min and WT mice, respectively: mRNA analysis by q-PCR indicated that the levels of both αv and β3 subunits were significantly upregulated in small intestine tumors in comparison to corresponding healthy mucosa ( Figure 1B).
To evaluate whether the expression of the integrin αvβ3 could be used to detect small intestine tumor formation in vivo, 24 h before the imaging, 18 weeksold APC +/min mice and age-related WT mice were shaved and injected IV with 0.08 nmol/g of body weight of IntegriSense750 (Perkin Elmer), a commercial available probe specifically recognizing integrin αvβ3 [21] and emitting a near infrared (NIR) fluorescent signal at 750 nm wavelength. Animals have been fed with the AIN76A, alfalfa-free, rodent diet (Mucedola srl) for 2 weeks, in order to minimize the autofluorescence originating from the animals' intestinal contents. Mice were positioned in a dedicated imaging cassette and IntegriSense750 signal acquired with the fluorescent molecular tomography (FMT) imaging system FMT2000 (Perkin Elmer). A 3D ellipsoidal region of interest (ROI) was outlined to select, as much as possible, the small intestine. Subsequently, we quantified probe concentration (as pmol/mm 3 ) within the ROI: independently from the CX 3 CR1 status, IntegriSense accumulated in the small intestine and this accumulation strongly increased in small intestine of APC +/min mice compared to WT mice (Figure 2A and 2B).
To further confirm the specificity of the in vivo signal acquisition, both APC +/min and WT mice were sacrificed and small intestine collected for ex vivo imaging analysis, performed with the IVIS Lumina III optical imaging system (Perkin Elmer). Ex vivo analysis confirmed the increased accumulation of IntegriSense750 probe in the small intestine of tumor-bearing APC +/min mice compared to WT mice ( Figure 2C and 2D): of note, several spots of probe accumulation can be identified in the small intestine of APC +/min mice ( Figure 2C, white arrows), likely corresponding to the localization of the developed multiple small adenomas. To further verify the accuracy and the correct positioning of the in vivo ROIs, we correlated the in vivo quantification of the IntegriSense750 probe and ex vivo acquired signal, finding a significant correlation between in vivo and ex vivo results (r 2 0.72, p < 0.0001), thus confirming the specificity of the detection ( Figure 2E). In conclusion, APC +/min tumor-bearing mice show a higher accumulation of integriSense750 probe in comparison to WT mice, suggesting that more fluorescent signal corresponds to an increased tumor load in APC +/min mice.
Finally, given the role of CX 3 CR1 in regulating intestinal inflammation, we compared the formation of small intestine adenomas in APC +/min -CX 3 CR1 +/and APC +/ min -CX 3 CR1 -/-. When the CX 3 CR1 status was considered, we found that APC +/min -CX 3 CR1 -/mice showed a slightly but significant increased accumulation of IntegriSense750 probe compared to APC +/min -CX 3 CR1 +/mice both in in vivo FMT and ex vivo imaging analysis ( Figure 3A, 3B and Figure 3C, 3D, respectively), confirming our (and other) previous results of increased tumor occurrence in CX 3 CR1-deficient mice [20,22]. This result indicates that IntegriSense750 is able not only to distinguish tumor tissue from healthy mucosa, but also to discriminate between different tumor load in small intestine adenomasbearing mice.

In vivo detection of tumor formation in the colon by αvβ3 integrin expression
We subsequently evaluated the feasibility of targeting integrin αvβ3 in a chemically-induced mouse model of colon cancer. APC +/min mice spontaneously develop multiple adenomas in the small intestine but few in the colon. To induce the formation of colorectal tumors, APC +/min and WT mice were administered with dextran sulfate sodium (DSS) in drinking water: DSS is a widely used irritant of the colonic mucosa, able to cause intestinal inflammation (in a B-and T-cell independent manner), likely through the damaging of epithelial monolayer lining the large intestine and the consequent dissemination of pro-inflammatory luminal content into the tissue [23]. We administered 3% DSS to 4 weeks-old APC +/min and WT mice for 7 days, followed by 4 weeks of fresh water [20,24]. The occurrence of tumors was assessed by histology, comparing colon from APC +/min and WT DSStreated mice ( Figure 4A). Subsequently, we evaluated the mRNA expression of integrin αvβ3. We excised tumor lesions or corresponding healthy mucosa from the colon of DSS-treated APC +/min and DSS-treated WT mice, respectively: mRNA analysis by qPCR indicated a significant increase in the expression of the β3 subunit in tumor lesions of DSS-treated APC +/min mice ( Figure 4B).
We then evaluated in vivo the expression of integrin αvβ3. 24 h before the acquisition, mice were shaved and injected IV with 0.08 nmol/g of body weight of IntegrinSense750. Also in this case, animals have been accumulation in the small intestine of 18 weeks old APC +/min -CX 3 CR1 +/and APC +/min -CX 3 CR1 -/mice. Region of interest (ROI) was tridimensionally defined to outline the small intestine. The colorbar shows the concentration (in nM) of the IntegriSense750 probe. (B) FMT quantification of IntegriSense750 probe in the small intestine of APC +/min -CX 3 CR1 +/and APC +/min -CX 3 CR1 -/mice. The total amount of probe within the ROI (in pmoles) was calculated relative to the internal standard generated with a known concentration of the appropriate NIR dye, following manufacturer's instruction. The amount of pmoles of the probe was normalized based on the volume (mm 3 ) of the ROI, after adjustment using the statistically derived threshold value. Data are presented as min and max whiskers plot. Student's t Test, * p < 0.05 (C) Ex vivo analysis of the small intestine of APC +/min -CX 3 CR1 +/and APC +/min -CX 3 CR1 -/mice previously acquired by FMT. The colorbar shows the Radiant Efficiency. (D) Ex vivo quantification of the IntegriSense750 probe in the small intestine of APC +/min -CX 3 CR1 +/and APC +/min -CX 3 CR1 -/mice. Data are presented as min and max whiskers plot. Student's t Test, * p < 0.05. fed for 2 weeks with AIN76A, alfalfa-free, rodent diet. Mice were positioned in the dedicated imaging cassette and IntegriSense750 signal acquired with FMT2000 system. A cylindrical 3D region of interest (ROI) was outlined in order to select, as much as possible, the colon. Independently from the CX 3 CR1 status, probe quantification revealed higher accumulation of the IntegriSense750 in the colon of DSS-treated APC +/min mice compared to DSS-treated WT mice ( Figure 5A and 5B).
Subsequently, both APC +/min and WT DSS-treated mice were sacrificed and colon collected for ex vivo imaging analysis by IVIS system, clearly confirming the increased concentration of IntegriSense750 probe in the colon of APC +/min mice compared to WT mice ( Figure 5C and 5D): in particular, IntegriSense750 accumulation specifically corresponds to the visible colonic neoplastic lesions of APC +/min DSS-treated mice ( Figure 5C). Also in this case, even if the r 2 is poor, a significant linear correlation was found between the in vivo FMT concentration of the IntegriSense750 probe and ex vivo IVIS acquired signal, confirming the specificity of the detection (r 2 0.21, p < 0.02, Figure 5E).

Correlation between in vivo and ex vivo αvβ3 integrin detection and histological analysis
Histology still represents the leading technique for the evaluation of tumor occurrence and dimension. We therefore compared the results obtained from the in vivo and ex vivo imaging acquisition with the histological analysis. Small intestine tumor and colon cancer from mice previously analysed for the expression of IntegriSense750 probe were collected and stained with hematoxylin and eosin. Sections were scanned with VS120-S5 Virtual Slide System (Olympus). Subsequently, we calculated the percentage of tumor area as described in "Material and Methods". We then correlated the results obtained from the in vivo and ex vivo imaging analysis with the percentage of tumor area. Interestingly, we found a significant correlation between the small intestine tumor area and the quantification of IntegriSense750 acquired both in vivo with FMT (r 2 0.79, p 0.02, Figure 6A) and ex vivo with the IVIS system (r 2 0.68, p 0.04, Figure 6B). In particular, some spots of IntegriSense750 accumulation distinguished with the ex vivo analysis corresponded to small adenomas identified by conventional histological analysis ( Figure 6C, whitedotted line squares, and Figure 6D).
The same correlation was performed on colon tumors. In this case, a significant correlation was found only between histological analysis and the ex vivo acquisition with IVIS system (r 2 0.44, p 0.03, Figure 6F), even if a tendency was observed also for the FMT analysis (r 2 0.08, p 0.39, Figure 6E). As for the small intestine tumors, the spots of IntegriSense750 accumulation detected by ex vivo imaging analysis matched with colonic tumors recognizable by conventional histological analysis ( Figure 6G, white-dotted line squares, and Figure 6H).

Expression of IntegriSense probe within tumor tissue
Integrin αvβ3 is probably the integrin most strongly involved in the regulation of angiogenesis. Moreover, it is extensively reported that it binds to a wide range of extracellular matrix (ECM) proteins. To investigate the expression pattern of IntegriSense probe, mice bearing tumors in the small intestine or in the colon were IV injected with 0.08 nmol/g of body weight of IntegriSense680 (Perkin Elmer), a probe analogous to the IntegriSense750 previously used but emitting a fluorescent signal at 680 nm wavelength. After 24 h, mice were sacrificed and small intestine and colon collected. Ex vivo IVIS imaging confirmed that the expression pattern of IntegriSense680 was comparable to the one previously observed with IntegriSense750 (Supplementary Figure 1A and 1B). Subsequently, tissues were analysed by immunofluorescence: confocal microscopy analysis confirmed the upregulation of IntegriSense probe in neoplastic tissues (APC +/min ) compared to healthy mucosa (WT), both in small intestine and colon tumors ( Figure  7A and 7B, respectively). Of note, the localization of the probe was similar between small intestine tumor and colon tumors.
To confirm the colocalization of IntegriSense probe with ECM proteins, tissues were analysed for the expression of α-smooth muscle actin (α-SMA) and fibrin: as expected, we observed several spots of colocalization between IntegriSense and both α-SMA and fibrin in small intestine and colon tumors ( Figure 7C and 7D, respectively).
As extensively reported in the literature, integrin αvβ3 is strongly involved in the regulation of angiogenesis. We therefore stained tissues with an anti-CD31 antibody to specifically identify vessels: white arrows in Figure 7C and 7D indicate the colocalization between IntegriSense and CD31 + vessels, confirming the expression of integrin αvβ3 by tumor-associated vessels in both small intestine and colon tumor ( Figure 7C and 7D, respectively).

In vivo monitoring of small intestine and colon cancers
The main advantage of the non-invasive in vivo imaging technique is the possibility to monitor deep tumor growth without the need to sacrifice mice. We therefore investigated if the expression of integrin αvβ3 could be used as a marker to follow both small intestine tumor and colon cancer progression.
For the small intestine tumor model, independently from the CX 3 CR1 status, APC +/min and WT mice were shaved and injected, 24 h before the acquisition, with 0.08 nmol/g of body weight of IntegriSense750 probe after 14 and 18 weeks of age. As shown in Figure 8A and 8B, a significant increased accumulation of IntegriSense750 was already detectable in 14 weeks-old APC +/min mice compared to age-matched WT mice; in addition, we observed a close to significant (p = 0.28) signal increase from 14 to 18 weeks, a result presumably compatible with a progression of the disease, while no difference was found over time in WT mice ( Figure 8A and 8B).
In the colon cancer model, a close to significant higher IntegriSense750 accumulation was already detectable 2 weeks after the end of the DSS administration ( Figure 8C and 8D, p = 0.06). Interestingly, IntegriSense750 accumulation increased between the second and the fourth week after DSS administration in APC +/min tumor-bearing mice, while no differences were observed in WT mice. Of note, at this latter time point (4 weeks), IntegriSense750 accumulation is significantly higher in DSS-treated APC +/min tumor-bearing mice compared to DSS-treated WT mice ( Figure 8C and 8D). These results indicate that IntegriSense750 probe can be used to monitor tumor growth in both small and large intestine tumor progression.

DISCUSSION
The identification of new tumor biomarkers increases the opportunity of an early detection and the possibility of a more effective therapeutic intervention [25]. In this study, we demonstrated that integrin αvβ3 is overexpressed in intestinal lesions compared to corresponding healthy mucosa and its visualization by Fluorescent Molecular Tomography (FMT) can be used for the early in vivo detection of both small and large intestine tumors. We found that the commercially available www.oncotarget.com probe IntegriSense (Perkin Elmer) [21] specifically accumulates in cancer tissues compared to healthy mucosa, allowing tumor visualization and the monitoring of neoplastic progression in a genetically-induced mouse model of the small intestine tumor (APC +/min ) as well as in a chemically-induced model of colon cancer (DSStreated APC +/min mice). In addition, data obtained from in vivo FMT detection of IntegriSense correlated with ex vivo imaging acquisitions. Moreover, subsequent immunofluorescence analysis by confocal microscopy revealed a high specificity of IntegriSense signal, showing a precise colocalization with ECM proteins and markers of vascularization.
Collectively, these results strongly support the potential utility of the IntegriSense probe and, more in general, of integrin αvβ3 expression in studies aimed at evaluating intestinal tumor occurrence and progression.
The application of optical molecular imaging techniques in preclinical studies is becoming increasingly important, providing the possibility to investigate tumor progression in small animal models in relatively inexpensive, simple and fast way [4,5,7]. Colorectal cancer, a major cause of morbidity and mortality in Western countries, has been investigated in the last years using optical imaging approaches [26,27]. A variety of NIR-conjugated probes have been generated to recognize specific target involved in CRC progression and eventually evaluate the efficacy of therapeutic treatments. However, the majority of these studies have the limitation of having been performed using transplanted orthotopic or xenograft mouse model of CRC [28][29][30][31][32][33][34]. In addition, several fluorescent probes have been already used for the detection of chemically-induced mouse models of CRC, but only for ex vivo validation of tumor occurrence or metastatic dissemination [35,36]. More recently, a protease-activatable NIRF imaging probe has been used for the FMT imaging of azoxymethane (AOM)-induced colonic tumors, presenting anyway some limitations as a valid non-endoscopic method for cancer detection [37]. Of note, the advantage of FMT versus conventional 2D optical imaging has been already discussed [9] and recently applied for the detection and quantification of inflammation in a mouse model of inflammatory bowel disease [38].
Although integrin αvβ3 has been already used to visualize transplanted mouse model of CRC [28,30], to our knowledge this is the first study where its detection by optical in vivo imaging has been used for the non-invasive evaluation of genetic or chemically-induced mouse model of intestinal tumor. In addition, the use of fluorescent molecular tomography (FMT) technique, combined with the 750 nm wavelength NIRF probe, allows a more precise three-dimensional and deep localization of the signal and the quantification of the fluorescent probe within the outlined regions of interest. Anyway, the anatomical resolution of optical in vivo imaging techniques is quite limited due to light scattering and absorption by surrounding tissues: in the present study, two different ROIs were used in order to isolate as much as possible the small intestine (located in the anterior region of the abdominal cavity and identified with an ellipsoid ROI) and the colon (located in the posterior region of the abdominal cavity, in cranio-caudal direction, and identified with a cylindrical ROI). For this reason, ex vivo optical imaging was used as a confirmation of the signal detected by in vivo FMT imaging. While a clear correlation between in vivo and ex vivo imaging was found for the small intestine, a poor but still significant relationship between in vivo and ex vivo analysis was found for the colon. This difference can be explained by considering the deep position of the colon in the abdominal cavity and the consequent effect of scattering and absorption by surrounding tissues affecting both excitation and emission light, that eventually limits the specificity of the results. These aspects have to be carefully considered, suggesting cautions in the interpretation of in vivo imaging results.
We also evaluated the feasibility of IntegriSense probe to distinguish mouse models with a different tumor load. For this purpose, APC +/min mice were previously breed with CX 3 CR1 +/and CX 3 CR1 -/mice. CX 3 CR1 is considered a marker for resident intestinal macrophages, which play a central role in regulating inflammation, and its absence has been correlated with dysregulated inflammatory response, tissue damage and higher tumor occurrence [18-20, 22, 39]. Our data, in line with the literature, showed that in the APC +/min mouse model of small intestine carcinogenesis, the accumulation of IntegriSense probe was slightly but significantly higher in APC +/min -CX 3 CR1 -/compared to APC +/min -CX 3 CR1 +/-, confirming the protective effect of CX 3 CR1 in intestinal inflammation.
Furthermore, we compared optical imaging results with conventional histological analysis. Even if a correlation between histology and optical imaging analysis was found (especially in the small intestine carcinogenesis model), this result needs to be carefully discussed. In fact, while optical imaging allows the visualization and analysis of the entire tissue, histology is limited by the physical sectioning of the sample. This sectioning could result in the impossibility to have all tumors in the same slide, especially when bigger tissues and small lesions are considered (as indeed occurs in the APC +/min model of small intestine carcinogenesis). Of note, this consideration could also explain the fact that the slightly difference in small intestine tumor formation between APC +/min -CX 3 CR1 -/and APC +/min -CX 3 CR1 +/detected by both in vivo and ex vivo optical imaging was not confirmed by histological analysis (data not shown). In this context, optical imaging and histology can be considered as two complementary, rather than overlapping, techniques, able to provide different features of the same sample.
The role of integrins in cancer progression has been extensively described, being involved in tumor cell migration, invasion and control of angiogenesis, lymphangiogenesis and inflammation [10,11]. Among integrins, αvβ3 is probably the most strongly involved in the regulation of angiogenesis [11,13,40]. It has been reported that, differently from the quiescent endothelium, integrin αvβ3 is highly expressed by tumor-associated vessels, possibly mediating the interaction between endothelial cells and provisional matrix protein, such as vitronectin, fibrinogen and proteolized collagen [11,15,41,42]. These observations are in line with our confocal acquisitions, showing a colocalization between the IntegriSense probe and the matrix protein (α-SMA and fibrin) as well as the CD31 + tumor-associated vessels.
The expression of integrin αvβ3 correlated with disease progression in various type of cancer, including breast, pancreatic and prostate cancer and glioblastomas [43][44][45][46][47][48][49]. Given its role in cancer progression and neoangiogenesis processes, it is not surprising that integrin αvβ3 represents an interesting target for the visualization and treatment of a variety of tumors [15,40,[50][51][52][53][54][55][56]. Coupling integrin αvβ3 antagonists or antibodies (LM609) to paramagnetic contrast agents or radionuclides provided the detection of vascularization in experimental tumor models [15,[56][57][58]. In addition, combination of 18 F-galacto-RGD and PET allowed the quantification of integrin αvβ3 expression in murine tumor models and in cancer patients [59]. In CRC, it has been demonstrated that vascular expression of integrin αvβ3 correlated with the presence of liver metastasis, reduced relapse-free interval and patients' overall survival [12]. In addition, treatment with an integrin αvβ3 antagonist resulted in prolonged survival in an orthotopic murine model of colon cancer liver metastasis, with a reduction of vessels in liver metastasis, an increase in endothelial cell apoptosis and a significant decrease in pericyte coverage [60]. Moreover, optical in vivo imaging detection of integrin αvβ3 has been successfully used in combination with PET/CT analysis for the visualization of subcutaneous and orthotopic transplanted mouse model of CRC [28]. Accordingly, our in vivo, ex vivo and confocal data showed that integrin αvβ3 is already expressed by intestinal tissue, and its expression strongly increased in tumors. These observations confirmed the role of integrin αvβ3 in intestinal carcinogenesis, additionally showing that its expression could be used for the early tumor detection by FMT in preclinical mouse models. In fact, our results demonstrated that a higher IntegriSense accumulation was already detectable after 14 weeks of age in small intestine adenomas-bearing APC +/min mice and 2 weeks after DSS treatment in colorectal cancer-bearing APC +/min mice. Of note, the signal slightly increased in time, indicating that IntegriSense detection can be used to monitor tumor progression both in the small and large intestine. These observations raise the possibility of using integrin αvβ3 imaging for early detection of human CRC. Of course, the application of fluorescent reporter in human is limited to intra-operative imaging-guided surgery, thus the integrin αvβ3 imaging should be approached with MRI or PET techniques.
In conclusion, our results support the use of integrin αvβ3 detection probe for the in vivo FMT and ex vivo 2D-imaging detection of tumor in preclinical mouse models of intestinal carcinogenesis. In particular, FMT technique, together with the use of NIRF probe, increases the specificity of tumor detection compared to 2D conventional optical imaging methods, ameliorating the anatomical resolution. Combining integrin αvβ3 FMT application with other imaging techniques, such as microCT or MRI, could improve the specificity and sensitivity of intestinal cancer visualization, laying the foundations for the potential use of integrin αvβ3 expression for the early detection of intestinal tumors in humans. All mice were on C57/B6 background. We compared APC +/min -CX 3 CR1 +/gfp and APC +/min -CX 3 CR1 gfp/gfp mice (indicated as APC +/min , when the CX 3 CR1 status were not considered) with CX 3 CR1 +/gfp and CX 3 CR1 gfp/gfp (indicated as WT, when the CX 3 CR1 status were not considered). Since GFP expression was not considered in this study, we indicated heterozygous CX 3 CR1 +/gfp mice as CX 3 CR1 +/and knock-out CX 3 CR1 gfp/gfp mice as CX 3 CR1 -/-. Mice were maintained in a specific-pathogen free facility, and given ad libitum access to food and water. Where indicated, mice were feed with the specific AIN76A, alfalfa-free rodent diet (Mucedola srl).

RNA extraction and quantitative real-time PCR
RNA was extracted from adenomas or corresponding healthy mucosa independently. Samples were homogenized in TRIzol (Ambion) using Tissue Lyser II (Quiagen). A quantity amounting to 1 µg total RNA was reverse transcribed using the High-Capacity cDNA Archive kit (Applied Biosystems), according to manufacturer's instruction. cDNA was analyzed by quantitative real-time PCR, performed on CFX96 Touch™ Real-time PCR detection system (BioRad), using Fast SYBR Green (Applied Biosystems). Actin gene was used as internal control. The sequences of primers are as follow:

Histological analysis
Tissues were harvested, cleaned from the luminal content, longitudinally opened and rolled up transversely (Swiss roll). Tissues were then fixed in 4% paraformaldehyde (PFA) overnight (O/N) at 4° C and paraffin-embedded. To evaluate the histologic architecture of the tissue and the percentage of the tumor area over the total tissue, 2-µm thick sections were cut and stained with hematoxylin and eosin (H&E stain). Slides were scanned with VS120-S5 Virtual Slide System (Olympus) and analyzed with OlyVIA Software (Olympus): subsequently, we calculated the percentage of global tumor area, by summing the dimensions of the multiple neoplastic lesions and dividing them by the dimension of the entire tissue section.
The following formula was used: Percentage of tumor area: (Global dimension of the neoplastic lesions × 100)/Dimension of the entire tissue section

Fluorescent molecular tomography (FMT)
Fluorescent molecular tomography (FMT) was performed using FMT2000 system (Perkin Elmer). Briefly, animals were fed with AIN76A, alfalfa-free rodent diet (Mucedola srl) for at least 2 weeks before the acquisition in order to reduce autofluorescence background. 24 h before imaging session, mice were shaved to avoid the interference of the fur and IV injected with 0.08 nmol/g of body weight of IntegriSense750 probe (Perkin Elmer) [21]. Prior to the imaging, mice were anesthetized with a mixture of ketamine and xylazine and positioned in a dedicated imaging cassette. The imaging cassette was adjusted to the proper depth to gently restrain the mice. Mice were inserted in supine position into the heated docking system (~37° C) into the FMT imaging chamber and trans-illuminated with 750 nm near infrared (NIR) laser. The entire acquisition procedure took approximately 6-8 minute per mouse. The collected images were reconstructed and analysed using TrueQuant3.1 software (Perkin Elmer). Three-dimensional regions of interest (ROI) were designed to identify the small intestine and the colon. Threshold was applied equal to the 40% of the mean fluorescence (in nM) of the control mice. The total amount of probe within the ROI (in pmoles) was automatically calculated relative to the internal standard generated with a known concentration of the appropriate NIR dye, following manufacturer's instruction. The amount of pmoles of the probe was normalized based on the volume (mm 3 ) of ROI, after adjustment using the statistically derived threshold value.

IVIS Lumina III ex vivo imaging
Ex vivo imaging was performed using IVIS Lumina III system (Perkin Elmer). Briefly, after in vivo FMT analysis, mice were sacrificed, the organs collected and acquired using the following parameters: excitation filter 740 nm, emission filter 790 nm, binning 4 or 8, f/Stop 2. Collected images were analysed using Living Image 4.3.1 software (Perkin Elmer): ROIs were designed in order to appropriately select each organs and radiant efficiency calculated.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 6 (GraphPad software). Student's t test, multiple student's t test with Holm-Sidak multiple comparison test and linear regression analysis were used, as indicated in the figure legends. A p value < 0.05 was considered as statistically significant.