In Vivo Assessment of VCAM-1 Expression by SPECT/CT Imaging in Mice Models of Human Triple Negative Breast Cancer

Recent progress in breast cancer research has led to the identification of Vascular Cell Adhesion Molecule-1 (VCAM-1) as a key actor of metastatic colonization. VCAM-1 promotes lung-metastases and is associated with clinical early recurrence and poor outcome in triple negative breast cancer (TNBC). Our objective was to perform the in vivo imaging of VCAM-1 in mice models of TNBC. The Cancer Genomic Atlas (TCGA) database was analyzed to evaluate the prognostic role of VCAM-1 in TNBC. MDA-MB-231 (VCAM-1+) and control HCC70 (VCAM-1-) TNBC cells were subcutaneously xenografted in mice and VCAM-1 expression was assessed in vivo by single-photon emission computed tomography (SPECT) imaging using 99mTc-cAbVCAM1-5. Then, MDA-MB-231 cells were intravenously injected in mice and VCAM-1 expression in lung metastasis was assessed by SPECT imaging after 8 weeks. TCGA analysis showed that VCAM-1 is associated with a poor prognosis in TNBC patients. In subcutaneous tumor models, 99mTc-cAbVCAM1-5 uptake was 2-fold higher in MDA-MB-231 than in HCC70 (p < 0.01), and 4-fold higher than that of the irrelevant control (p < 0.01). Moreover, 99mTc-cAbVCAM1-5 uptake in MDA-MB-231 lung metastases was also higher than that of 99mTc-Ctl (p < 0.05). 99mTc-cAbVCAM1-5 is therefore a suitable tool to evaluate the role of VCAM-1 as a marker of tumor aggressiveness of TNBC.


Introduction
Breast cancer (BC) is the most common female malignancy, accounting for more than 30% of all malignant tumors in women [1]. Breast cancer is a heterogeneous disease consisting of various subtypes with distinct molecular and pathological profiles [2,3]. Triple Negative Breast Cancer (TNBC) subtypes represent 10% to 20% of BC and are characterized by the lack of progesterone receptor, estrogen receptor and human epidermal growth factor receptor 2 expression [4]. TNBC are mostly associated with poor clinical outcome and high rate of metastasis and relapse following treatment [5,6]. Lung, bones, brain and liver are the most common sites of distant metastasis. Despite intense clinical research efforts, only limited advances have been obtained in the management of BC metastases.
Recent studies have led to the identification of new genes and mechanisms that mediate metastatic colonization [7]. Among them, Vascular Cell Adhesion Molecule-1 (VCAM-1) plays a key role in BC progression and metastatic processes [8,9]. VCAM-1 belongs to the immunoglobulin super family group of adhesion molecules. It is a 110 kDa glycoprotein mainly expressed at the endothelial cells surface during inflammation, but also by macrophages and dendritic cells [10]. The ability of VCAM-1 expressed by endothelial cells to bind tumor cells suggest that it could contribute to metastatic spread. Indeed, VCAM-1 expression on endothelial cells plays a key role in angiogenesis, in tumor cell transmigration, and therefore promotes tumor development and dissemination of tumor cells [11].
In the last few years, a growing interest into VCAM-1 expression on tumor cells has emerged. In BC, the overexpression of VCAM-1 on tumor cells correlates with early relapse and poor patient outcome [12,13]. The direct interaction of VCAM-1 with its ligand Very-Late Antigen -4 (VLA-4) expressed on leukocytes allows tumor cell survival in lungs and consequently lung metastasis [14]. Moreover, VCAM-1 is involved in the transition from dormant micro-metastases to overt macro-metastases in bones, a turning point in BC progression [13,15]. Inhibition of the VCAM-1 and VLA-4 interaction impairs bone metastasis progression and lung colonization [13,14,16]. VCAM-1 is also involved into chemoresistance and tumor immune escape [17][18][19]. Therefore, VCAM-1 is a key player with multiple functionalities in directing the metastatic spread. VCAM-1 expression could be induced by pro-inflammatory cytokines, such as Tumor Necrosis Factor-α (TNF-α), IL-1 or IL-6, major mediators of tumor progression [19].
To assess the role of VCAM-1 in the metastatic phenotype of BC, nuclear imaging may represent a powerful tool. We recently developed a single domain antibody (sdAb)-based radiotracer targeting VCAM-1 called 99m Technetium-cAbVCAM1-5 ( 99m Tc-cAbVCAM1-5), which is ongoing clinical transfer for detection of inflamed atherosclerosis lesions [20,21]. The aim of the present study was to assess VCAM-1 expression using 99m Tc-cAbVCAM1-5 in subcutaneous and lung metastasis mice models of TNBC.

Ex Vivo Assessment of VCAM-1 Expression in Subcutaneous Tumors
Since VCAM-1 is also involved in inflammatory and angiogenic processes, we next determined which part of its expression was attributable to tumor cells (human VCAM-1), or to the inflammatory and endothelial cells (mouse VCAM-1) using RT-qPCR ( Figure 4). Consistently with in vitro results, hVCAM-1 mRNA was expressed by MDA-MD-231 but was not detectable in HCC70 tumors. Moreover, hVCAM-1 expression was nearly 4-fold higher than that of mVCAM-1 in MDA-MB-231 xenograft tumors (p < 0.001), suggesting that the hVCAM-1 expression by tumor cells was the predominant form present in MDA-MB-231 subcutaneous xenografts. Interestingly, no significant difference was observed in mVCAM-1 mRNA expression between HCC70 and MDA-MB-231 tumors (1.6 ± 1.0 vs. 1.0 ± 0.1). This results strongly suggests that the signal obtained on SPECT/CT images reflects VCAM-1-expressing tumor cells.

SPECT/CT Imaging of VCAM-1 in an Experimental Metastasis Model
Because VCAM-1 is involved in lung colonization we next investigated the ability of 99m Tc-cAbVCAM1-5 to perform its imaging in an experimental metastasis model. Representative SPECT/CT images are presented in Figure 5A and in Figure S6.  99m Tc-cAbVCAM1-5 lung uptake was readily observable, whereas a weak signal was obtained with 99m Tc-Ctl ( Figure 5A). Quantification of SPECT/CT images confirmed that the lung uptake was 2.5-fold higher with 99m Tc-cAbVCAM1-5 than 99m Tc-Ctl (1.7 ± 0.4 vs. 0.7 ± 0.1% ID/cm 3 , p < 0.05, Figure 5B). When performing image quantification of pulmonary activity, the volume of interest contains a mixture of tissue and air, thereby leading to an underestimation of the uptake. 99m Tc-cAbVCAM1-5 lung uptake was therefore further evaluated using ex vivo gamma-well counting ( Figure 5C) and autoradiography ( Figure 6A,B). Using these two technics, 99m Tc-cAbVCAM1-5 uptake was found to represent~3% ID/g and 4-5 fold-higher value than that obtained with 99m Tc-Ctl (p < 0.05 for both technics). 99m Tc-cAbVCAM1-5 uptake was also evaluated in lung-metastasis free mice and found to be 3-fold lower than lung-metastasis bearing mice (p < 0.05, Figure S7).

Discussion
Despite major advances in fundamental knowledge and therapeutic opportunities, BC remains the leading form of malignancy among women [1]. If the 5-year relative survival of BC is almost 100% when the cancer is restricted to the breast, the prognosis of patients with metastatic BC is unfavorable with a 5-year survival rate of 25% [22]. Moreover, among women initially diagnosed without metastasis, 20 to 25% will develop a metastatic disease in the next 5 years [23]. Despite intense clinical research efforts, there is still a strong need for novel molecular target and therapies to improve management of BC metastases [24]. The identification of genes and mechanisms involved in metastatic processes and the development of effective treatments against metastatic BC are outstanding challenges in current experimental and clinical research. In the past few years, growing interest into tumorigenicity and metastatic processes has led to the identification of VCAM-1 as a key actor for tumor growth, metastasis and angiogenesis [25].
In BC, VCAM-1 expression on tumor cells is an important actor for metastatic colonization of lungs and bones [9]. In the lungs VCAM-1 binds to α4β1 integrin expressed on macrophages triggering the activation of the PI3K/Akt survival pathway in cancer cells [14]. In bone, VCAM-1 expressing tumor cells binds to α4β1 integrin-expressing osteoclast progenitors to mediate osteolytic metastasis [13]. Moreover, VCAM-1 expression seems to be correlated to poor outcome in BC [12]. The results provided by the TCGA analysis showed that VCAM-1 is overexpressed in TNBC in comparison to Non-TNBC. Moreover, VCAM-1 expression is associated with a poor prognosis only in TNBC patients. In addition to BC, VCAM-1 increased expression has also been described in other cancer types such as glioblastoma, gastric and ovarian cancer. In these cancers, VCAM-1 expression correlated with the tumor grade [26][27][28]. Considering the role of VCAM-1 in directing the metastatic spread, VCAM-1 imaging agents could be used to (1) understand the biological role of VCAM-1 in metastatic processes and (2) to evaluate the prognostic value of VCAM-1 expression in clinical practice. Therefore, radiotracers have been developed for VCAM-1 imaging in tumors [29,30].
Our group recently developed 99m Tc-cAbVCAM1-5, a radiotracer targeting VCAM-1 which is ongoing clinical transfer for the detection of inflamed atherosclerotic lesions [20,21,31]. The purpose of the present study was to evaluate this new tool for the pre-clinical nuclear imaging of VCAM-1 in TNBC. Two TNBC cell lines were employed, MDA-MB-231 and HCC70. The MDA-MB-231 cell line is highly metastatic whereas few results were available on HCC70. In vitro experiments demonstrated that MDA-MB-231 expressed hVCAM-1 mRNA and protein, and that expression was increased after TNF-α stimulation which is consistent with previous studies [19]. TNF-α is a key activator of VCAM-1 through NF-κB signaling [32]. A bundle of evidences links TNF-α and NF-κB pathway to tumor survival, growth and invasion [33,34]. The results provided by the TCGA showed higher TNF-α expression in TNBC in comparison to Non-TNBC tumors. These results suggest that the TNF-α/VCAM-1 axis is a relevant target in TNBC. As demonstrated by in vitro competition studies, the 99m Tc-cAbVCAM1-5 tracer specifically binds human VCAM-1. Using this imaging agent, subcutaneous MDA-MB-231 tumors were successfully visualized by SPECT/CT imaging, whereas significantly lower signals were found in HCC70 tumors (which do not express hVCAM-1 mRNA and protein) or using the negative irrelevant control sdAb, thereby indicating that the signal was specific and predominantly attributable to tumor cell expressing VCAM-1 rather than murine endothelial or inflammatory cells. These results were in agreement with the level of murine and human VCAM-1 mRNA determined in tumor biopsies by RT-qPCR. Additional studies are however warranted in order to investigate if, similarly to that has been observed in vitro, in vivo hVCAM-1 expression in MDA-MB-231 tumor cells can be increased by TNF-α or other cytokines present in the microenvironment, leading to increased 99m Tc-cAbVCAM1-5 uptake.
Therefore, 99m Tc-cAbVCAM1-5 is a validated tool to investigate the role of VCAM-1 in metastatic processes. However our results support that using VCAM-1 as a tumor inflammatory marker should be carefully considered due to the potential expression of VCAM-1 by tumor cells themselves.
Because of VCAM-1 is aberrantly expressed in BC cells and mediates lung metastasis, 99m Tc-cAbVCAM1-5 imaging was further studied in an experimental lung metastasis model with MDA-MB-231 cells. According to SPECT/CT quantifications, 99m Tc-cAbVCAM1-5 lung uptake was 3-fold higher than that of 99m Tc-Ctl whereas it was found to be 5-fold higher by ex vivo gamma-well counting. This difference can be attributed to the fact that when performing in vivo lung imaging quantification, the volume of interest contains a mixture of tissue and air, thereby leading to underestimation of the uptake.
Autoradiography further confirmed that 99m Tc-cAbVCAM1-5 uptake was localized in metastatic nodules, and RT-qPCR showed that, in the whole lung, hVCAM-1 expression was found to be 3-fold higher than mVCAM-1 indicating that 99m Tc-cAbVCAM1-5 uptake highlights presence of tumor cells rather than inflammatory processes. This result is consistent with a previous study demonstrating VCAM-1 tumor expression rather than endothelial cells one in lung metastasis of BC patients [14]. 99m Tc-cAbVCAM-1 is therefore a validate tool to study the prognostic value of VCAM-1 in the metastatic disease.
Other VCAM-1-targeting radiotracers have been developed. 99m Tc-cAbVCAM1-5 tumor uptake was comparable to that obtained by Scalici et al. using 111 In-tVCAM-4, an indium-111 labeled peptide targeting VCAM-1 [30]. Indeed, in an experimental mouse model of metastatic ovarian cancer, 111 In-tVCAM-4 uptake in tumor was of~2% ID/g at 4 h post-injection. Using a 68 Ga-labeled Single Chain Variable Fragment, Zhang et al. obtained a~5% ID/g tumor uptake in a mouse model of melanoma. Nevertheless, due to the fast blood clearance of sdAbs-based imaging agents, 99m Tc-cAbVCAM1-5 tumor-to-blood ratio (5 at 2 h) favorably compared to that of this imaging agent (2 at 3 h) [29].

Patients-Online Data
Normalized RNA sequencing (RNA-Seq) data of The Cancer Genome Atlas (TCGA) were downloaded from cBioportal (Breast Invasive Carcinoma-TCGA Provisional; RNA-Seq V2) [35]. Data were available for 980 BC tumor samples (866 Non-TNBC and 114 TNBC). The results published here are based upon data generated by the TCGA Research Network [36,37].

Cell Lines and Culture Conditions
MDA-MB-231 cells were cultured with Dulbecco's Modified Eagle's Medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. HCC70 cells were cultured using Roswell Park Memorial Institute-1640 medium, supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. nodules, 20-50 mg of tissue was ground (PureLink TM RNA Mini Kit; ThermoFisher Scientific), and the same protocol as for the in vitro assay was applied. Quantitative Polymerase Chain Reaction (qPCR) was performed on the resulting cDNA with the Fast SYBR Green Master Mix (ThermoFisher Scientific, Illkirch, France), using a Real Time PCR system (Applied Biosystem StepOne Plus; ThermoFisher Scientific).

Western Blot Assay
HCC70 and MDA-MB-231 cells were treated or not with TNF-α (50 ng/mL) for 18 h. Total proteins were extracted with RIPA buffer. Proteins were separated by 7.5% denaturing SDS-PAGE and transferred onto a nitrocellulose membrane. The membrane was incubated with the anti-VCAM-1 antibody (1/2000; Rabbit anti-VCAM-1, ab134047 Abcam®, Cambridge, UK). As a loading control, the membrane was probed with an anti-β-actin antibody (1/10,000; Beckton Dickinson). Revelation was assessed using the chemiluminescence ECL kit (BioRad). Bands were quantified by densitometry using ImageJ software.

Saturation Binding Experiments and In Vitro Competition Studies
Saturation binding experiments was determined on 96-well plates coated with mouse or human recombinant VCAM-1 protein (100 ng/well; R&D Systems, Minneapolis, MN, USA). Serial diluations of 99m Tc-cAbVCAM1-5 were incubated for 1 h at room temperature before being washed 5 times with PBS-polysorbate 0.1%. The radioactivity in each well was then determined using a γ-counter (Wizard 2 ; Perkin Elmer, Courtaboeuf, France) and corrected for unspecific binding. Binding curves were fitted using a non linear regression equation (specific binding: y = B max × x/(K D + x), with x being the radioligand concentration, K D being the dissociation constant, and B max being the maximum number of binding sites, or receptor density) (GraphPad Prism, version 6, software, San Diego, CA, USA), to determine K D values.
For in vitro competition studies, 12,000 HCC70 and MDA-MB-231 cells were coated in 96-well plates (Stripwell TM Plate, Corning ® , Corning, NY, USA) in their respective culture medium. After 24 h, cells were treated with 50 ng/mL of TNF-α for 18 h. Then, cells were rinsed with PBS and fixed with paraformaldehyde (PFA) 4% for 10 min. Following 1 h of saturation with PBS-BSA 1%, cells were incubated with 30 nM of 99m Tc-cAbVCAM1-5 in the absence or presence of a 100-fold excess of unlabeled cAbVCAM1-5 at room temperature for 1 h before being washed 5 times with PBS-Tween 0.05%. Bound activity was determined using a gamma-counter (Wizard 2 , Perkin Elmer) and results were corrected from nonspecific binding determined on HCC70 cells.

SPECT/CT Imaging
For the subcutaneous xenograft study, mice were divided in 2 groups: tumor bearing mice injected with (i) the human/mouse cross-reactive sdAb 99m Tc-cAbVCAM1-5 (n = 6) or (ii) with the irrelevant control sdAb, 99m Tc-Ctl (n = 6). The previously described BcII10a control sdAb was used in this study (20). SPECT/CT acquisitions were performed 1 hour after intravenous injection of 57.6 ± 9.8 MBq of 99m Tc-cAbVCAM1-5 or 99m Tc-Ctl. The cAbVCAM1-5 and control sdAbs were radiolabeled as previously described using the tricarbonyl method [20]. Whole body SPECT/CT acquisitions were performed using a dedicated system (nanoSPECT-CT; Mediso, Budapest, Hungary). For the experimental metastasis study, mice were injected with 61.9 ± 16.2 MBq of 99m Tc-cAbVCAM1-5 (n = 5) or the irrelevant 99m Tc-Ctl (n = 4). Biodistributions were also performed on healthy mice (n = 4). SPECT/CT acquisition was centered on the thoracic region. CT and SPECT acquisitions were reconstructed and fused using Nucline software (Mediso), and SPECT quantification based on CT was performed using VivoQuant TM (InviCRO, Boston, MA, USA). For the xenograft tumor model, a 50 mm 3 sphere was drawn at the center of the tumor on the basis of the CT image. For the metastasis assay, total quantification of pulmonary tracer uptake based on the CT image was performed. 99m Tc-sdAbs activity was expressed as a percentage of the injected dose per cm 3 (% ID/cm 3 ).

Post-Mortem Analysis
Two hours after injection and immediately following SPECT/CT image acquisitions, anesthetized mice were euthanized using CO 2 and tumors (either subcutaneous or the whole lungs with metastasis) were harvested along with other organs. Samples were weighed and their radioactivity determined with a γ-counter (Wizard 2 , PerkinElmer, Courtaboeuf, France). Results were corrected for decay, injected dose and organ weight and expressed as % ID/g. Subcutaneous tumors were then immediately frozen in −40 • C isopentane, whereas metastatic lungs were inflated with a 1:1 mixture of PBS/Optimal Cutting Temperature prior being frozen. In order to investigate whether the VCAM-1 expression status determined in vitro on HCC70 and MDA-MB-231 cells was preserved in vivo, RT-qPCR was performed on HCC70 and MDA-MB-231 subcutaneous tumor slices and tissue samples. Moreover, due to the fact that ex vivo counting on lung with metastases reflects a mixture of healthy tissue and tumor uptake, lung autoradiography was performed to precisely evaluate the tumor uptake. To that purpose, 20 µm-thick lung slices, together with reference organs of known activities, were exposed overnight on an autoradiographic film which was then scanned using a phosphoimager (Fujifilm BAS-5000, FUJIFILM, Montigny, France). Slices were then stained with hematoxylin-eosin and regions of interest were delineated on tumor nodules, thereby allowing the quantification of 99m Tc-cAbVCAM1-5 or 99m Tc-Ctl uptake as a percentage of injected dose per gram (% ID/g).

Statistics
For the in vitro and mice experiments: Data were expressed as mean ± standard deviation and analyzed using an unpaired Mann Whitney test for inter group analysis. Significance of linear correlations was assessed with a Pearson's test. p values < 0.05 were considered significant. Data were analyzed with Prism 7.0 (GraphPad Software).
For patients: The Student's t-test was used to compare continuous variables. Overall survival (OS) was defined as the time between surgery and the date of death from any cause. The Kaplan-Meier method was used to produce survival curves and analyses of censored data were performed using Cox models. Data were analyzed with Prism 7.0 (GraphPad Software).

Conclusions
In the two mouse models of TNBC, SPECT imaging of VCAM-1 was successfully performed and the signal originated from the tumor was found to reflect hVCAM-1 expression by cancer cells rather than mVCAM-1 expression by the tumor stroma. 99m Tc-cAbVCAM1-5 can therefore be used as a preclinical tool to evaluate the role of VCAM-1 expression by tumor cells in tumor development and metastasis. In clinical practice, VCAM-1 expression has been reported to be correlated with poorer outcome in TNBC but also in other cancer type such as glioblastoma and ovarian cancer, and VCAM-1 imaging has been proposed as a tool for the assessment of tumor aggressiveness. Further studies will however be necessary to evaluate the prognostic value of 99m Tc-cAbVCAM1-5 tumor imaging in clinical practice.