Exogenous marker-engineered mesenchymal stem cells detect cancer and metastases in a simple blood assay

Mesenchymal stem cells (MSCs) are adult multipotent stem cells that possess regenerative and immunomodulatory properties. They have been widely investigated as therapeutic agents for a variety of disease conditions, including tissue repair, inflammation, autoimmunity, and organ transplantation. Importantly, systemically infused MSCs selectively home to primary and metastatic tumors, though the molecular mechanisms of tumor tropism of MSCs remain incompletely understood. We have exploited the active and selective MSCs homing to cancer microenvironments to develop a rapid and selective blood test for the presence of cancer. We tested the concept of using transplanted MSCs as the basis for a simple cancer blood test. MSCs were engineered to express humanized Gaussia luciferase (hGluc). In a minimally invasive fashion, hGluc secreted by MSCs into circulation as a reporter for cancer presence, was assayed to probe whether MSCs co-localize with and persist in cancerous tissue. In vitro, hGluc secreted by engineered MSCs was detected stably over a period of days in the presence of serum. In vivo imaging showed that MSCs homed to breast cancer lung metastases and persisted longer in tumor-bearing mice than in tumor-free mice (P < 0.05). hGluc activity in blood of tumor-bearing mice was significantly higher than in their tumor-free counterparts (P < 0.05). Both in vitro and in vivo data show that MSCs expressing hGluc can identify and report small tumors or metastases in a simple blood test format. Our novel and simple stem cell-based blood test can potentially be used to screen, detect, and monitor cancer and metastasis at early stages and during treatment.


Introduction
Cancer is a leading cause of human morbidity and mortality, and its origins, biomarkers, and detection remain difficult to pinpoint [1]. Although early detection has proven to be a useful and often necessary first step to effectively manage and treat cancer [2], it remains a challenge at early stages to identify cancer, especially small tumors and metastases which account for over 90 % of cancer mortality [3,4]. Methods of cancer detection based on imaging are non-invasive, but common drawbacks include high cost, low specificity or resolution, and the use of potentially irritating contrast agents [2]. For instance, positron emission tomography (PET), computed tomography (CT), and their combinations (PET-CT) are widely used for identifying and staging tumors but require high doses of ionizing radiation and have limited specificity and resolution [5]. Other imaging modalities, such as magnetic resonance imaging (MRI) and ultrasound, do not use radiation but are still unable to achieve spatial resolution smaller than several millimeters [6,7]. On the other hand, tissue biopsies are invasive and suffer from false negatives for heterogeneous tumors, and obtaining biopsies from multiple small disseminated tumors (e.g., metastases) is impractical. Cancer screening also uses tests for biomarkers, including circulating tumor cells, exosomes, proteins, and nucleic acids. Recently, scientists have developed nanoparticle-based synthetic biomarkers composed of mass-encoded peptides that can be released upon tumor protease cleavage and then detected in urine [8,9]. Such approaches, however, still rely on passive delivery of nanoparticles to tumors via the enhanced permeability and retention (EPR) effect and on limited types of endogenous proteins, both of which are cancer typespecific. More recently, scientists have also reported a probiotic microbe-based system to deliver synthetic biomarker for cancer detection in urine [10]. Nevertheless, cancer biomarker discovery has led to only a few biomarkers used in clinical diagnosis since cancer biomarkers frequently suffer from low sensitivity and specificity [11].
In particular, cancer heterogeneity and evolution make it challenging to rely on molecular biomarkers for cancer detection [1]. For example, the commonly used cancer biomarkers prostate-specific antigen for prostate cancer and BRCA1/2 gene mutations for breast cancer can identify only about 25 % and 10 % to 25 % of the patients in each cancer type, respectively [12]. Indeed, it has been widely accepted that a single biomarker typically lacks the sensitivity and specificity that are necessary for useful diagnosis. Intriguingly, recent research indicates that most cancers are caused by stochastic events rather than predictable mutations [13]. Thus, finding biomarkers that recognize multiple types of cancers with no common genetic basis is likely less promising than previously thought. In summary, there is clearly an unmet clinical need for sensitive early-stage cancer and metastasis tests that can "universally" identify many types of cancers independently of specific biomarkers from healthy controls and other conditions that share similar symptoms (e.g., inflammation) as well as to discriminate different (sub)types of cancers at different stages.
Cells, including immune and stem cells, act as autonomous and adaptive agents and these properties have recently been used for cancer treatment and drug delivery [14][15][16][17]. In particular, mesenchymal stem (or stromal) cells (MSCs) have been tested as therapeutic agents because of their intrinsic regenerative and immunomodulatory features [18][19][20][21][22][23]. MSCs are under investigation for treating a wide array of diseases, including diabetes, myocardial infarction, stroke, and autoimmune diseases [24][25][26]. MSCs are also the world's first manufactured stem cell product to receive clinical approval (i.e., Prochymal manufactured by Osiris (Columbia, MD, USA) was approved in Canada to treat graft-versus-host disease) [26], suggesting that they may be a safe source for diagnostic and therapeutic uses in humans. Importantly, systemically infused MSCs preferentially home to and integrate with tumors, including both primary tumors and metastases in different anatomical locations [24]. As we have recently reviewed [22], mounting evidence now suggests that MSCs possess leukocyte-like, active homing mechanisms for tumor tropism involving a variety of adhesion molecules (e.g., P-selectin and vascular cell adhesion molecule-1) and tumor-derived cytokines, chemokines, and growth factors (e.g., CXCL12 and platelet-derived growth factor). This selective and active homing ability makes MSCs appealing vectors for localized delivery of therapeutics to treat cancers, including gliomas, melanomas, breast cancer, and lung metastases, in ongoing clinical trials [15,24]. In addition, MSCs engineered with probes (such as luciferase) have been used to detect and image tumors in situ [19,27]. However, imaging methods such as PET/single-photon emission computed tomography

Box 1. About Weian Zhao
and MRI, which are currently used for cell tracking after infusion are limited by the same aforementioned disadvantages of cancer detection [2].
In this article, we present the concept of using exogenous MSCs as the basis for a simple cancer blood test (Scheme 1). Here, we hypothesize that, owing to their tumor tropism property, MSCs engineered with a secreted reporter can actively and specifically home to tumor sites regardless of the type and location of the tumors and persist there longer compared with MSCs in healthy microenvironments. MSCs engineered to express humanized Gaussia luciferase (hGluc) [28][29][30][31] were systemically administered to mice harboring breast cancer cells, exhibited tumor tropism and persistence, and secreted hGluc into the bloodstream of tumor-bearing mice. Thus, MSCs engineered with secreted reporters can potentially be developed into a blood test for broad cancer screening and monitoring.

Cell lines and cell culture
Human bone marrow MSCs were obtained from the Texas A&M Health Science Center and were expanded to within passages 3-6. The cells were routinely maintained in minimum essential medium alpha (MEMα) (Life Technologies, Carlsbad, CA, USA) supplemented with 15 % fetal bovine serum (FBS) (Atlanta Biologicals, Norcross, GA, USA) and 1 % penicillin-streptomycin Scheme 1 Using engineered mesenchymal stem cells (MSCs) to detect cancer. Engineered MSCs (gray) secreting humanized Gaussia luciferase (hGluc) (green) are systemically administered into patients with cancer (breast cancer lung metastasis in this case). Engineered MSCs home to tumor (cyan) niche and persist, secreting hGluc into blood. Then patient blood can be collected and hGluc activity measured

Statistical analysis
Data were analyzed by Student's t test when comparing two groups and by analysis of variance when comparing more than two groups. Data were expressed as mean ± standard deviation or as mean ± standard error of the mean, and differences were considered significant at P values of less than 0.05.

Results
Humanized Gaussia luciferase is secreted from engineered MSCs in vitro and is stable and detectable in blood Human bone marrow MSCs were stably transduced with lentivirus to express secreted humanized Gaussia luciferase (hGluc) as described above. To determine whether hGluc is secreted in an active form by MSC, cell-free CM was harvested from hGluc-MSCs 24 hours after MSC seeding at different concentrations (100, 1000, 2500, or 5000 cells per cm 2 ). The substrate CTZ was added and hGluc activity was measured for both cells and CM (Fig. 1a). hGluc activity increased with increasing cell number (Fig. 1a). In addition, hGluc activity in CM was 3-to 6-fold higher than inside cells (Fig. 1a), indicating that hGluc expressed by engineered MSCs is secreted in active form, as expected. hGluc-MSC CM was serially diluted with PBS and hGluc activity was measured in vitro and found to exhibit a linear function of concentration, in agreement with earlier reports [33,36,37] (Fig. 1b). To demonstrate whether luciferase activity from hGluc-MSCs is detectable and sufficiently stable in blood, human serum either directly (100 %) or serially diluted in PBS was mixed with hGluc-MSCs CM. hGluc activity remained detectable (P < 0.0001) after 24 hours co-incubation and was not decreased significantly over time (Fig. 1c), indicating that hGluc-MSCs can be a stable marker in blood assays in vitro. Finally, since both firefly luciferase (Fluc-tdT) and hGluc would be used in vivo (below), any potential cross-reactivity between Fluc-tdT and hGluc-MSCs was measured (Additional file 1: Figure S1). These two luciferases were substrate-specific and no cross-reaction was observed, as reported. Overall, these data show that hGluc expressed by engineered MSCs is secreted in vitro, is stable in human serum for up to 24 hours, and exhibits substrate-specific enzyme activity.
Engineered MSCs home to tumor sites and persist longer in the lungs of the tumor-bearing mice As MSCs are reported to naturally home to tumor sites [18,19], we tested this phenomenon in our experiment as a preliminary step to using MSCs that secrete hGluc Human breast cancer-derived MDA-MB 231 cells were labeled with eGFP or Fluc-tdT and implanted intravenously (i.v.) into immunodeficient NSG mice (Fig. 2) to establish a simple in vivo mouse model of breast cancer that has metastasized in the lungs [38,39]. Tumor mass was observed in lungs both in vivo (Fig. 2a) and ex vivo (Fig. 2b, d), whereas no tumor-related signal was seen in healthy lungs (Fig. 2a, c). Owing to the fact that hGluc is secreted by MSCs and to its diluted and limited signal under whole animal imaging conditions with IVIS Lumina [40] (data not shown), we used MSCs engineered with intracellular Fluc-tdT [41] for real-time imaging and localization of MSCs in tumors in situ. Fluc-tdT-MSCs were simultaneously labeled with red fluorescent protein (RFP) to assess Fluc transduction efficiency and to image any co-localized MSCs and tumor cells in subsequent ex vivo immunohistochemistry. Both Fluc activity and RFP signal from Fluc-tdT-MSCs were observed in vitro (Additional file 2: Figure S2), demonstrating that engineered MSCs express Fluc (Additional file 2: Figure S2A) with high transduction efficiency (>90 % RFP + ; Additional file 2: Figure S2B-D).
To investigate any differences in MSCs homing between cancer-bearing and healthy mice, 10 6 Fluc-tdT-MSCs were systemically infused into mice with or without breast cancer. Mice were anesthetized and in vivo Fluc activity was measured after i.p. administration of D-luciferin Fig. 3 Mesenchymal stem cells home to tumor site and persist longer than in healthy mice. a Five weeks after eGFP-231 were seeded intravenously into NSG mice, 10 6 Fluc-tdT-MSCs were administered systemically into both tumor-free (top) and tumor-bearing (bottom) mice. Then mice were injected intraperitoneally with D-Luciferin (150 mg/kg in Dulbecco's phosphate-buffered saline), and in vivo Fluc activity was measured at different time points (2, 6, 24, and 48 hours and 7 and 10 days after MSC infusion) by using an IVIS Lumina to begin data acquisition 10 minutes after substrate administration (exposure time = 60 s; n=4 in each group). MSCs were cleared out faster in tumor-free mice. Color scale: minimum = 6.50×10 4 , maximum = 7.50×10 5 . Frozen sections of lungs of b tumor-free mice and c eGFP-231 tumor-bearing mice sacrificed 10 days after Fluc-tdT-MSC infusion were stained with anti-eGFP (green) and anti-Fluc (red) antibodies. MSCs were observed to home to tumor niche. Scale bar: 50 μm. d Fluc activity measured at different time points was quantified and normalized to the time point of 2 hours. Error bar: mean ± standard error of the mean. *P <0.05. n=4 in each group. eGFP enhanced green fluorescent protein, Fluc firefly luciferase, MSC mesenchymal stem cell, NSG nonobese diabetic/severe combined immunodeficiency gamma, tdT tdTomato red fluorescent protein substrate into mice at the indicated time points. In vivo imaging demonstrated that MSCs were detectable in tumor-bearing mice for as long as 10 days after systemic administration (Fig. 3a). Ex vivo immunohistochemistry data confirmed that engineered MSCs homed to the tumor niche in vivo (Figs. 3c and 4a). As we hypothesized, engineered MSCs persisted significantly longer in tumor-bearing lungs, especially at later time points (Fig. 3a). We then quantified the Fluc signal and found that significant differences between tumor-bearing and tumor-free mice emerged 24 hours after MSC infusion and lasted until 10 days after infusion (Fig. 3d, n=4, P < 0.05). To test whether our technology can be applied to other types of cancer, we investigated fused Fluc-tdT-MSCs into Fig. 4 Gaussia luciferase (hGluc) is active in murine blood and the signal is elevated in tumor-bearing mice. a Frozen sections of lungs of tumor-bearing mice sacrificed 10 days after Dil-labeled hGluc-MSC administration were stained with DAPI and then imaged by fluorescence microscopy. MSCs (red) were observed to home to tumor niche (dense blue). Scale bar: 100 μm. b Five weeks after Fluc-tdT-231 were seeded intravenously into NSG mice, 10 6 hGluc-MSCs were administered systemically into both tumor-free and tumor-bearing mice. Then murine blood was harvested and hGluc activity was measured at different time points (6, 24, and 48 hours and 7 and 10 days after MSC infusion) with IVIS Lumina immediately after substrate was added. hGluc activity measured at different time points was quantified and normalized to the time point of 6 hours. The inset graph shows that the hGluc activity in blood between tumor-bearing and tumor-free mice is significantly different from 48 hours after MSC infusion. Error bar: mean ± standard error of the mean. *P <0.05. Exposure time = 30 s. n=4 in each group. DAPI 4',6-Diamidino-2-phenylindole, MSC mesenchymal stem cell mice with lung metastasis of colon cancer. Similar results were observed (Additional file 3: Figure S3) which demonstrate engineered MSCs could home to and stay in tumor-bearing lungs for a significantly longer time compared with tumor-free lungs. Our data, along with mounting evidence of MSC tumor tropism in the literature [18,22,42,43], suggest that the in vivo persistence of engineered MSCs in tumor-bearing compared with healthy animals provides a viable "marker" for broad cancer detection.
hGluc secreted by engineered MSCs can be assayed in the blood of tumor-bearing mice We next investigated whether MSCs that were engineered to express hGluc can be used to detect metastasis of breast cancer to the lungs. hGluc was chosen as the reporter in this study because of its high sensitivity, lack of nonspecific cross-reactivity to other substrates (e.g., Additional file 1: Figure S1), and linear signal over a wide concentration range (Fig. 1b). In addition, hGluc has a short half-life in vivo (20 minutes), allowing repeated real-time testing without undesirable excessive signal accumulation, but a long half-life in vitro (6 days), allowing convenient sample storage [33]. As hGluc is secreted, it cannot be used as a marker to co-localize MSCs and tumor as seen in Fig. 3c for intracellular Fluc. Therefore, in this set of experiments, we stained hGluc-MSCs with the Dil lipophilic dye before they were infused i.v. into mice. Like Fluc-tdT-MSCs, Dil-MSCs were detectable in the tumor niche up to 10 days postinfusion (Fig. 4a). Mouse blood was collected at the indicated time points, and hGluc activity was measured. Although the detected signal decayed rapidly over time as expected, the difference of hGluc activity in blood between tumor-bearing and tumor-free mice was significant starting from 48 hours after MSC administration and lasting until 10 days post-infusion (Fig. 4b), suggesting that systemically infused hGluc-MSC can be used for the potential development of a simple blood assay for cancer detection in this murine model. In summary, this set of data supports the feasibility of using engineered MSCs with secreted hGluc as a blood test for the presence of cancer.

Discussion
Early detection of cancer, especially metastasis, is a necessary and often critical first step to effectively treat and eradiate cancer. Traditional imaging tools and molecular biomarker-based assays are typically complex, expensive, and/or invasive for routine screening for most cancers; most importantly, they frequently do not possess the sensitivity and specificity to identify heterogeneous cancers at early stages. In our study, we developed a stem cell-based detection system that can detect cancer, including metastases, by collecting small amounts of blood with a minimally invasive procedure. Our engineered MSCs could home to tumor sites and persist there for significantly longer durations compared with healthy mice. The signal derived from engineered stem cells lasted longer compared with current imaging tracers [5], and no repeat administration was needed. With one single administration, the presence of tumor could be monitored continuously through a prolonged period of time, making MSCs a convenient tool for real-time cancer detection. Compared with acellular systems (e.g., antibodies and nanoparticles), the natural interactions between MSCs and tumor involve complex adaptive sensing and responding systems that enable more efficient and specific reporting of cancer and metastases. This intrinsic biological property of tumor homing therefore potentially allows our stem cell approach to "universally" identify many cancers regardless of their origins, types, and anatomical sites. In addition, stem cell-based probe delivery circumvents many hurdles associated with passive delivery (i.e., by direct administration or polymeric nanoparticles via the EPR effect), including penetrating the endothelium and the increased pressure associated with tumors. In addition, the use of distinct, exogenous markers (hGluc in this article) as surrogate markers to detect and monitor cancer is more advantageous than endogenous markers because of the lack of unique cancer biomarkers. In our assay, a positive detection of hGluc (even with a small signal) would indicate the presence of cancer, which therefore helps to eliminate the need for sophisticated signal normalization over background as required in conventional cancer detection assays. Therefore, our simple, noninvasive stem cell-based blood test might be useful for routine cancer screening, detecting small tumors and metastases, and monitoring cancer progression and recurrence during the course of treatment.
Since MSCs possess not only tumor tropism but also tropism for bone marrow and sites of inflammation and injury [20,23], it remains important to distinguish those conditions from cancer when using MSC-based methods to detect cancer. In addition, given high cancer heterogeneity, our next-generation systems aim to engineer MSCs with activatable, cancer type-specific probes to further increase the assay specificity. The long-term goal is to establish a panel of tests that can effectively discriminate between cancer (sub)types and stages and distinguish between cancer and other disorders that share similar symptoms, including inflammation and injury.
MSCs were chosen in our current (first-generation) system because they can be easily obtained from multiple adult tissues [44], including bone marrow and fat, therefore avoiding ethical concerns. MSCs are also relatively easy to expand in culture and can be readily engineered to express functional therapeutics or reporters [14,23]. Importantly, the clinically approved Prochymal and hundreds of other ongoing clinical trials have demonstrated that allogeneic MSCs are generally safe for use in the human without harsh immunosuppressive regimens. Nonetheless, as MSCs themselves may participate in cancer progression or regression [22], further considerations are required. The interactions between MSCs and cancer remain incompletely understood [14,22], with different reports indicating conflicting findings from endogenous and exogenous MSCs on cancer progression [22,45,46]. Thus, safety tests and optimizations will likely be required to better control the fate of our engineered MSCs after cancer detection, though no obvious MSC-mediated cancer growth was observed within our detection window (Additional file 4: Figure S4). To mitigate this potential issue, for example, a suicide gene [47] can be engineered into our MSC-based system so that after completion of the cancer detection test, the remaining engineered MSCs can be eliminated by using exogenously administered drugs. For example, inducible human caspase-9 (iC9), which can be activated by a bio-inert small-molecule drug, has been used as a safety switch in clinical trials of cell therapy with limited immunogenicity [48]. Another limitation of our study is that we used a relatively large tumor burden as our model to demonstrate our proof-of-concept because of its technical simplicity. In the future, we will evaluate our engineered stem cell approach to detect early-stage cancer and metastases when they are small by using cancer models with smaller tumor burden by either reducing the cell number administered or at the early stages of the beast cancer progression. These future experiments will allow us to determine the smallest tumor size we can detect with our technology. Furthermore, our system may be used as companion diagnostics combined with other treatments, for example, identifying certain patients and monitoring side effects. Finally, our cell-based blood assay may represent a new platform for monitoring the fate and functions of transplanted cells as well as for assessing the in vivo microenvironment where they reside.

Conclusions
We demonstrate for the first time, to the best of our knowledge, a simple blood test for cancer detection. This test is based on the premise of exploiting the natural tumor-homing ability of MSCs to further engineer them to express a secreted luciferase with optimal biocompatibility and kinetic parameters. Similar to our current murine studies, these "reporter MSCs" could be developed to identify the presence of small tumors or metastases in humans that would otherwise be undetectable by existing imaging modalities. We hope this simple "off the shelf" allogeneic stem cell-based diagnostic test can be used to screen, detect, and monitor cancer on a routine basis.

Additional files
Additional file 1: Figure S1. Additional file 3: Figure S3. Systemically infused MSCs persist in the lungs of the LoVo cancer cell-bearing mice. Five weeks after LoVo colon cancer cells were seeded intravenously into NSG mice, 10 6 Fluc-tdT-MSCs were administered systemically into both tumor-free (blue) and tumor-bearing (red) mice. Then mice were injected intraperitoneally with D-Luciferin (150 mg/kg in Dulbecco's phosphate-buffered saline), and in vivo Fluc activity was measured at different time points (6,24, and 48 hours and 7 and 10 days after MSC infusion) by using an IVIS Lumina to begin data acquisition 10 minutes after substrate administration (exposure time = 60 s). Fluc activity measured at different time points was quantified. Similar to the results with MDA-MB-231 breast cancer cells, MSCs were cleared out faster in tumor-free mice, showing that the tumor tropism of MSCs is applicable to multiple types of cancers. Error bar: mean ± standard error of the mean. *P <0.05. n=4 for tumor-bearing mice and n=3 for tumor-free mice. Fluc firefly luciferase, MSC mesenchymal stem cell, NSG nonobese diabetic/severe combined immunodeficiency gamma, tdT tdTomato red fluorescent protein. (PNG 128 kb) Additional file 4: Figure S4. Engineered mesenchymal stem cell (hGluc-MSC) infusion has no influence on the growth of cancer metastasis size in vivo. Five weeks after Fluc-tdT-231 were seeded intravenously into NSG mice, 10 6 hGluc-MSCs or PBS was administered systemically into tumor-bearing mice (day 0). In vivo Fluc activity was measured with IVIS Lumina 10 minutes after substrate administration before (day 0) and 10 days after MSC infusion (day 10). Exposure time = 5 s. Relative metastasis index (RMI) = Luciferase activity on day 10 (after) / Luciferase activity on day 0 (before). N=4 for each group. hGluc humanized Gaussia luciferase, MSC mesenchymal stem cell, n.s. not significant, NSG nonobese diabetic/severe combined immunodeficiency gamma, PBS phosphate-buffered saline, tdT