|Home | About | Journals | Submit | Contact Us | Français|
In vivo multicolor photoacoustic (PA) flow cytometry for ultra-sensitive molecular detection of the CD44+ circulating tumor cells (CTCs) is demonstrated on a mouse model of human breast cancer. Targeting of CTCs with stem-like phenotype, which are naturally shed from parent tumors, was performed with functionalized gold and magnetic nanoparticles. Results in vivo were verified in vitro with a multifunctional microscope, which integrates PA, photothermal (PT), fluorescent and transmission modules. Magnet-induced clustering of magnetic nanoparticles in individual cells significantly amplified PT and PA signals. The novel noninvasive platform, which integrates multispectral PA detection and PT therapy with a potential for multiplex targeting of many cancer biomarkers using multicolor nanoparticles, may prospectively solve grand challenges in cancer research for diagnosis and purging of undetectable yet tumor-initiating cells in circulation before they form metastasis.
Advanced flow cytometry in vitro is a powerful tool for rapid multiparameter quantitative analyses of many individual cells at subcellular and molecular levels [1-2]. However, this technique is unable to detect rare cells (≤ 1 cell/mL) in blood circulation because the sensitivity threshold of conventional assays is primarily limited by a small sample volume (5-20 mL) [3-7]. Moreover, numerous studies have demonstrated that current sample preparation procedures fail to reproduce in vitro the native environment in vivo, and thus do not provide information on native cell-cell interactions, control of cell apoptosis and proliferation by the host environment, or cell migration [5-7]. Removal of cells from the host may alter their properties, including cell morphology and biomarker expression. This can confound interpretation, in translating in vitro data to living organisms.
To overcome these problems, our and other research groups introduced in vivo flow cytometry, in which cells of interests, in particular circulating tumor cells (CTCs), are detected directly in bloodstream of superficial vessels [4-12]. This in vivo approach allows potentially assessing rare CTCs in much larger blood volume compared to sample volume ex vivo. The most frequently used fluorescent labeling of cells in bloodstream demonstrated promising results on animal models although translation of this approach to human may be problematic due to (1) potential toxicity of the available fluorescent labels; (2) immune response to fluorescent tags; and (3) influence of light scattering and autofluorescence of the background, which limit its application to the assessment of superficial microvessels with a slow flow rate that may significantly lengthen the examination of the large blood volume ([13,14] and references there).
As an alternative, we developed in vivo blood and lymph flow cytometry with photothermal (PT), photoacoustic (PA), Raman and scattering detection technique [4,8-12,15,16], which allows either label-free detection of cells with appropriate intrinsic properties (e.g., strong pigmentation or scattering) or using low toxic metal nanoparticles (NPs) as PT, PA and scattering molecular contrast agents. Here we show that the recent advances in in vivo PA flow cytometry (PAFC) combined with PT technique and novel bioconjugated NPs can provide ultra-sensitive detection and counting of extremely rare subpopulation cancer cells with stem-like phenotype in bulk CTCs.
The identification of tumor-initiating cancer stem cells (CSCs) is a keen interest in cancer research because it was hypothesized that CSCs could be exclusively responsible for the growth and re-growth of primary and metastatic tumors [17-25]. The high metastatic potential of the CSCs and sometimes their drug- and radio-resistance might explain tumor progression and recovery despite intensive therapy [19,26,27]. A current theory, based on advances in genomic and molecular pathology, suggests that CSCs represent a small percentage (0.1-2%) of the unfractionated bulk tumor cells [18,25]. It can be logically suggested that, to develop metastatic disease, CSCs should be disseminated from parent tumors to metastatic sites by blood or lymph systems. Thus, the circulating CSCs (termed stem CTCs) should exist in small amount among bulk CTCs. Our assumption is supported by the fact that only rare (~0.01% and even less) tumor cells (likely stem CTCs) in circulation may form metastases . However, the stem CTCs remain unexplored area of cancer research. The major obstacle is technical limitations (e.g., low sensitivity) of current assays to identify rare stem CTCs from the small population of CTCs disseminated with a large background of blood cells. Currently, the sensitivity threshold of existing CTC assays, such as reverse transcriptase polymerase chain reaction (RT-PCR), CellSearch® system, flow cytometry, microchip fluid technology and others, is 1-10 CTCs per 1 mL of whole blood or 5,000-50,000 CTCs in the entire blood volume (~5 L in adult) [3,29]. The aforementioned data suggest that stem CTCs are expected to be present in circulation at a proportion of 0.01-2 % of bulk CTCs. Therefore, at the threshold of sensitivity of 1 cell/mL for bulk CTCs, the detection limit for stem CTCs must be around 1 cell per 50-1000 mL of whole human blood. Although these estimations are very preliminary and require careful verification, they clearly support that the insuperable obstacle in studying rare stem CTCs in vitro is the restricted volume of blood sample, typically 5-20 mL. Therefore, novel in vivo diagnostic strategies are required to detect such extremely low concentrations of stem CTCs.
We demonstrate here the proof-of-concept that the previously developed integrated nanotechnology-based multicolor PA and PT flow cytometry platform [8-12], after further upgrading and using novel gold and magnetic NPs as molecular PA and PT contrast agents [32-33], may hold promise not only for real-time detection of stem CTCs in blood circulation in vivo but also for their simultaneous purging when integrated with PT technique. To provide the first-step verification of this approach, folic acid and antibodies (Abs) to CD44 (hualuronic acid receptor) were selected for targeting CTCs and stem CTCs because the available data [17,18,22,30,31] support that folate receptors are highly expressed (70-90%) in breast bulk CTCs while CD44 expression happens preferentially in CSCs within some breast cancer cells.
PAFC measurements were performed by irradiation with a focused pulsed laser beam of the individual cells of interest (e.g., Folate+ and CD44+ CTCs) targeted by bioconjugated NPs directly in the blood flow. Laser-induced, non-invasive heating (≤1°C) of NPs was accompanied by their fast thermal expansion leading to generation of PA waves, which were detected with an ultrasound transducer attached to the skin (Fig. 1a). The integrated PA setup was built as described previously [32-36] on the technical platform of an Olympus BX51 microscope (Olympus America, Inc.) and a tunable optical parametric oscillator (OPO, Lotis Ltd., Minsk, Belarus) with a pulse of 8 ns in width, a repetition rate of 10 and 50 Hz, a wavelength in the range of 420-2,300 nm and a fluence range, 1-104 mJ/cm2.
PA signals from the ultrasound transducer (XMS-310; 3 mm in diameter, 10-MHz frequency band; Panametrics) and amplifier (model 5670, 10 MHz; 40-60 dB; Panametrics) were recorded with a Boxcar system (Stanford Research Systems Inc.) and a Tektronix TDS 3032B oscilloscope, and analyzed with standard and customized software. The transducer was gently attached to microscopic slides (in vitro study), or to the skin near the examined vessels (study in vivo). The warmed water or conventional ultrasound gel was topically applied for better acoustic matching between the transducer and the samples.
To extend the diagnostic and therapeutic capability, the PA flow cytometer was integrated with well-established PT technique. In PT thermolens (diagnostic) mode, the laser-induced temperature-dependent variations of the refractive index in single cells around heated NPs caused defocusing of a collinear continuous wave (CW) He-Ne laser probe (pilot) beam (wavelength, 633 nm; power, 1.4 mW) and, hence, a reduction in beam intensity at its center as detected by a photodiode (C5658; Hamamatsu Corp.) with a pinhole. PT nanophotothermolysis of breast cancer cells was performed by irradiation of individual labeled cells at relatively high OPO energy fluences ranged from 50 mJ/cm2 to 1 mJ/cm2.
Navigation of the laser beams was controlled with high-resolution (~300 nm) transmission digital microscopy (TDM) with a Cascade 650 CCD camera (Photometrics). To verify PA/PT data, fluorescence imaging was added using the following CCD cameras: a color Nikon DXM1200 and a high-sensitive PentaMAX (Princeton Instruments Inc.).
The integrated PA/PT microscope was used for time-resolved study of linear and nonlinear PA/PT signals from single cells as well as assessment of the labeling efficiency by monitoring PA/PT signals from individual treated and control cells (total 300 cells for each experiment) at 639 nm and 850 nm. To count the percentage of signals with specific amplitudes, the slide with cells was scanned across with a fixed laser beam using an automatic stage (Conix Research) and visual basic software .
The golden carbon nanotubes (GNTs) were used as super contrast PA and PT agents. The GNTs consist of hollow single-walled carbon nanotubes (CNTs) with gold layers around CNTs with average total dimensions of 12.8 nm (diameter) × 91.7 nm (length). The GNTs were synthesized as described elsewhere with detailed reports on their physicochemical characteristics . The GNTs exhibited high water solubility, biocompatibility, low cytotoxicity (due to the protective layer of gold around the CNTs) and high plasmon resonance in the near-infrared (NIR) range (similar to gold nanorods) at 850-900 nm. In particular, cell viability and proliferation assays revealed no apparent adverse toxicity effects on live cells after their exposure to various concentrations of GNTs (0.05–0.5 mg/mL) for 10 days .
The 30-nm spherical magnetic nanoparticles (MNPs) (Ocean NanoTech, Springdale, AR). were used as a triple (magnetic, PT and PA) contrast agent. MNPs exhibited strongest intrinsic absorption at short visible wavelengths, monotonically decreasing through the visible to NIR range . In selected experiments the permanent magnetic field was provided for the manipulation of MNPs by a cylindrical Neodymium-Iron-Boron (NdFeB) magnet with Ni-Cu-Ni coating, 3.2 mm in diameter and 9.5 mm long and surface field strength of 0.39 Tesla (MAGCRAFT, Vienna, VA). The magnet tip was gently attached to the top cover of the slides with MNPs alone (1011/mL, 8uL PBS) or with cell suspension labeled by MNPs.
To target breast cancer cells' overexpressed folic receptors, GNTs were conjugated with folate (GNTs-Folate) in the presence of 1% polyethylene glycol (PEG) through electrostatic interactions (Fig. 1b). To target breast cancer cells overexpressed CD44 receptors, GNTs and MNPs were conjugated with Abs specific to human CD44 receptor, yielding the complexes of GNTs-CD44 and MNPs-CD44, respectively (Fig. 1c,d). The Abs were additionally stained with fluorescent labels (fluorescein isothiocyanate–dextran [FITC]) according the manufacturer's specification (BD Pharmateuticals). After 10-min incubation with NPs at 23°C, the residual Abs were removed by three-time washing with 10-mM phosphate buffer saline (PBS, pH 8.2) containing 1% bovine serum albumin after centrifugation at 10,000×g for 5 min. The conjugation was verified by presence of fluorescence in individual NPs (Fig. 2a).
MDA-MB-231 human breast cancer cells (American Type Culture Collection, Manassas, VA) were cultured according to the vendor's specifications. The cells were cultured to confluency in vitro, detached with 0.25% trypsin-0.53 mM EDTA, washed and resuspended in PBS, and then used for in vitro and in vivo studies. Viability of cells was > 98.5% according to trypan blue exclusion test.
To perform in vitro study, the cells were incubated with GNT-Folate, GNT-CD44 or MNP-CD44 for 1 hr at 37°C. The necessary concentration of NPs was calculated based on our previous data: (0.5-1)×104NPs per one cell . Subsequently, double-washed cells were resuspended in PBS and placed in 8.6-μl wells (Molecular Probes). In selected experiments, the tumor cells were double labeled with functionalized NPs and FITC using standard procedure (15 min at 37°C).
The nude nu/nu mice weighing 20–25 g, was purchased from Harlan Sprague-Dawley (Harlan Sprague-Dawley, Indianapolis, IN). The animals were used in accordance with protocols approved by the University of Arkansas for Medical Sciences Institutional Animal Care and Use Committee. Mice were anesthetized with ketamine/xylazine, 50/10 mg/kg, by intraperitoneal injection. After anesthesia, an animal was placed on the customized heated microscope stage. The detection of CTCs was performed in 200-300 μm abdominal-skin blood vessels at depth of 0.3-0.5 mm.
To establish primary tumors, 106 MDA-MB-231 breast cancer cells in 50 μl of PBS were inoculated subcutaneously with a 30-gauge needle. The tumor growth was monitored on a weekly basis by measuring tumor size (2 dimensions with calipers).
To verify in vivo results, the blood was collected from donor mice using a plastic sterile syringe with anticoagulant to prepare samples of stabilized whole blood with controlled number of cancer cells.
Results were expressed as means plus/minus the standard error of at least three independent experiments (P <0.05). Statistica 5.11 software (StatSoft, Inc.) and MATLAB 7.0.1 (MathWorks) was used for the statistical calculations.
To verify target specificity of integrated PAFC and PT technique in vitro, the MDA-MB-231 cells in PBS were treated with conjugated MNP-CD44 and GNT-CD44 complexes, as well as unconjugated MNPs and GNTs. The presence of NPs in individual cancer cells was estimated by PA/PT scanning cytometry through increase of signals from cells with MNPs at 639 nm and cells with GNTs at 850 nm, and by fluorescent imaging of targeted cells (Fig. 2b). Counting of labeled cells was performed by automatic scanning of a microscopic slide on the microscopic stage at a fixed position of focused laser beam with diameter of ~10 μm and a laser fluence level of 15-100 mJ/cm2. Several PA signals from the same cells were acquired and transformed to unipolar signals with a width determined by the transit time of cells through the laser beam . At relatively high laser fluences of 300-500 mJ/cm2, readable PA/PT signals were obtained from 40-60% of cancer cells as well as the space between them (i.e., PBS with unbound NPs). However, most PA/PT signals from targeted cells were significantly higher (10-50 fold) than those from unbound NPs randomly distributed in suspension between cells. At a reduced laser fluence of 15 mJ/cm2, detectable signals with signal-to-noise ratio [SNR] ≥ 2 were obtained solely from 2.05 ± 0.10% of total cancer cell population in the absence of background signals from unbound NPs above electronic noise (Fig. 2c). The fluorescent imaging confirmed the presence of conjugated NPs in these cells (Fig. 2b) and, thus, verified that the PA signals were associated with CD44+ cancer cells. The cancer cells incubated with the unconjugated NPs at the same concentration as conjugated NPs yielded the level of detectable signals ≤0.2%, demonstrating negligible influence of non-specific binding at the adjusted low pulse energy of 15 mJ/cm2. These findings suggest that 1) the PA/PT signals from GNT-CD44 or MNP-CD44 conjugates were associated with binding of Abs to CD44+ breast cancer cells, 2) the cells overexpressed CD44 might be selected by adjusted laser energies, and 3) the influence of background signals from unbound or non-specifically bound NPs could be alleviated by also properly decrease laser fluencies, which are low enough to provide signals only from cells of interest with highest NP concentration.
To explore the capability of PAFC to detect CD44+ breast cancer cells in the real blood background, the whole blood from donor mice were spiked with breast cancer cells and incubated with MNP-CD44 complex. To ensure that rare PA/PT signals were from cancer cells, we priory labeled cancer cells with FITC (Fig. 2d) and controlled the position of NIR laser beam on samples using additional co-axial pilot He-Ne laser beam. PA/PT and fluorescent scanning cytometry revealed that 1.98 ± 0.12% of fluorescent labeled cancer cells in whole blood provided significant PA signals that are 10-20 fold higher than background signals from surrounding blood. These data are in good agreement with the above result using cancer cell suspension in PBS (i.e., 2.05±0.10% cells overexpressed CD44+). The signal amplitudes from the background blood cells were also significantly less than those from CD44+ cancer cells. This high signal contrast allowed us to reduce the background signals from normal blood cells to the level below electronic noise by decreasing laser energy to 10-20 mJ/cm2. Even under this adjusted low pulse energy, NP-labeled CD44+ cancer cells still provided clearly distinguishable signals with SNR in the range of 5-10. It is also important to note that the percentage of CD44+ cells obtained with MNPs was correlated with that of CD44+ cells obtained with GNTs using both PT and PA cytometry technique.
The attachment of a magnet tip to the top cover of the slides (Fig. 3a) with MNP-CD44 alone led to concentration of MNPs to the area of the magnet. This was visualized even by the naked eyes as a dark spot (Fig. 3b) and verified by fluorescent microscopy (Fig. 3d), which showed increases in fluorescent intensity from magnet-induced MNP clusters compared to the intact area before magnet action (Fig. 3a,c). The PT signal from the area with high local concentration of MNPs after magnet action (Fig. 3f) was significantly higher (10-20 fold) compared to that from intact sample (Fig. 3e). Similar effects were obtained from single cancer cells labeled with MNP-CD44 before and after magnet exposure. Specifically, the magnet action led to notable enhancement of local fluorescence gradient within one cell (Fig. 3h) compared to relatively homogenous spatial fluorescent light distribution before magnet action (Fig. 3g). This suggests magnet-induced clustering of MNPs within single cell, yielding significant (6.6-fold in this particular experiment) enhancement of PT signals from these cells (Fig. 3i,k). The appearance of the local intracellular zones with dense MNPs is likely due to the accumulation of moving (under magnet action) MNPs near cellular membranes as the mechanical obstacles.
We applied the PT technique to demonstrate ablation of the CD44+ cancer cells labeled by GNT-CD44 conjugates (Fig. 4). At a low fluence (35 mJ/cm2), a classic positive PT response was observed, which is related to laser-induced cell heating (due to redistribution of heat from strongly absorbing GNTs) and cooling (due to diffusion of heat from cell as whole to surroinding meduim) (Fig. 4e) without notable cell damage (Fig. 5a,c) [34-35]. The increase of laser fluence to 0.5 J/cm2 led to the formation of a negative peak associated with laser-induced microbubble formation (and hence local optical refraction changes) around overheated strongly absorbing NP assembly or clusters (Fig. 4f). It was known that microbubbles may lead to cell damage . Indeed, the high-resolution optical (Fig. 4b) and fluorescent (Fig. 4d) imaging confirmed cell membrane damage and change in the spatial configuration of the cell.
To detect rare CTCs in vivo, we applied mouse model of human breast cancer (Fig. 5a) on the basis of our previous experience on in vivo PA detection of circulating NPs (e.g., gold nanorods, CNTs, MNPs and GNTs) and CTCs (e.g., melanoma and breast cancer cells) [4,8-12,32,33]. In particular, because PA signals from unbound GNTs or nonspecifically uptaken GNTs by blood cells (e.g., blood-derived macrophages) at low NP concentration (109 NPs/mL in 100 μL of injected PBS) were below the PA background from blood in mouse vasculature , we chose this concentration for in vivo targeting and detection of stem-like CTCs originated from a primary tumor. The use of anti-human CD44 allowed us to reduce binding of GNT-CD44 conjugates with mouse white blood cells, which may likely express CD44, and, thus, to estimate capability of PAFC for the detection of CD44+ CTCs. In particular, at week 4 of tumor inoculation (Fig. 5) when the metastatic disease was well-recognized by metastasis in the liver (Fig. 5b), GNT-Folate and GNT-CD44 were separately injected i.v. through mouse tail vein at concentrations of 109/mL in 100 μL PBS (n=3). To allow effective labeling of CTCs in bloodstream, PA monitoring of blood vessels began at 10 min after injection [6,7,33]. The flashing readable PA signals above PA background of blood with SNR in the range of 3 to 20 were detected within 2 hours of the observation. The result is in agreement with previous studies, which demonstrated the monitoring of labeled CTCs with lifetime a few hours in the circulation (NPs alone were more quickly cleared, typically within 15-30 min) [6,7,10,33,36]. The observed variations in PA signal amplitudes are likely due to the heterogeneous cell labeling and variable biomarker expression in vivo [37,38]. Furthermore, almost half of PA signals (46%) had relatively low SNR at the level of 1.5-2, while approximately 8% signals had SNR ≥10. As a result, the average mean rate of PA signals, which is equivalent to the cell rate (i.e., one PA signal corresponds to one cell or cell cluster crossing laser beam) after introduction of GNT-Folate was 42.9 ± 6.5 cells/min. After subsequent injection of GNT-CD44, the rate of flashing PA signals increased to 46.7 ± 6.8 cells/min. It means that 3.8 ± 0.6 cells/min or 8.8% of all detectable CTCs can be exclusively associated with CD44+/Folate-CTCs, likely stem CTCs.
The current study reports the first noninvasive real-time PA-molecular detection of single CD44+ CTCs (i.e., with cancer stem phenotype) among other (bulk) CTCs in bloodstream of breast cancer bearing mice in vivo. We also demonstrated the poof-of-concept that this diagnosis can be integrated with targeted eradication of very aggressive metastasis initiating tumor cells using clinically-relevant multifunctional PAFC-PT-nanotechnology-based platform. No other currently available in vivo diagnostic techniques, such as magnetic resonance imaging (MRI), positron emission tomography (PET) or optical assays, have similar dynamic capability. In vivo fluorescent flow cytometry [5-7] only showed detection of bulk CTCs targeted by fluorescent conjugates (e.g., folate)  with no apparent application for identification and counting of stem CTCs. Moreover, compared to the robust PA and PT technique that can easily integrate diagnosis and therapy with the same laser, the fluorescent flow cytometry is primarily used for diagnostics. In the fluorescent technique, it is really hard to use the same laser simultaneously for label excitation and therapy of the fast moving CTCs because of different requirements to laser power and spectral range. Recently we also introduced in vivo Raman flow cytometry , which can be integrated with PA and PT technique, allowing detection of CTCs targeted by NPs with strong Raman contrast properties (e.g., CNTs or SERS). However, all these techniques require further sensitivity improvement so lesser laser energy can be applied to avoid potential photodamage of normal tissue.
Various strongly absorbing NPs have been used for both PA diagnostics and PT cancer therapy, including gold nanospheres, nanoshells, nanorods, nanocages, CNTs and others [39-41]. To the best of our knowledge, we demonstrate here for the first time that NPs, especially advanced low toxic GNTs with one of the highest PA/PT contrast properties, can be used for ultra-sensitive molecular detection and ablation of single stem CTCs in vivo. The population of CSCs are likely heterogenous [17,18,22,30,38] and identification of stem CTCs requires the use of several biomarkers (e.g., CD44, CD24, CD133, ALDH) among bulk CTCs. In this preliminary study, as a first-step of mimicking the real situations, folic acid and Abs to CD44 and combination of GNTs with conventional MNPs as multimodal agents were selected. Although we focused on intrinsic PA/PT contrast properties of MNPs, they can be also used for magnetic capturing of stem CTCs directly in bloodstream  (with their possible ablation or isolation) or MRI detection  in static condition (due to relatively slow MRI signal acquisition algorithm) after CTC extravasations and localization in specific organs. Because the clinical potential of PA devices and the required low laser energy within safety standard (45–100 mJ/cm2 at 700–1,100 nm, respectively ) has been demonstrated in several pilot trials in humans [44,45], and some gold and magnetic NPs are also approved for pilot clinical trials (see discussion in ), the PA diagnosis and PT ablation of stem CTCs has high potential for use on humans.
The important questions are the threshold sensitivity and the targeting efficiency of stem CTC in real biological environment in vivo. We demonstrated the threshold sensitivity of PAFC around 1 CTC/mL in blood in vivo, which was primarily limited by the relatively small blood pool volume (~2 mL) present in mice and realistic monitoring times (a few hrs), rather than by PAFC parameters . We predict that the use of large animals and especially translation of this technology to humans with much larger blood volume can significantly increase the detection sensitivity of stem CTCs up to 103 fold as the ratio of assessed blood volumes in vivo to that ex vivo. This can also reduce the examination time because the blood volume of ~5 L circulates through a ~3 mm in diameter human vein within 1 hr . Such threshold of sensitivity is unachievable with other existing techniques and represents sufficient level to rapidly detect extremely rare stem CTCs among bulk CTCs (see above) disseminated with a large background of blood cells. For MNPs with less intrinsic NIR absorption, higher concentrations should be required compared to GNTs. However, as we shown in this study, PA/PT response from cells with MNPs can be significantly amplified by magnet-induced MNP clustering, or/and by developing of hybrid MNPs with gold or other strongly NIR absorbing layers as discussed in .
The requirements for the labeling of CTCs, and especially stem CTCs in the bloodstream are much less strict than those for targeting primary tumors because of 1) lower NP concentration required due to lower CTC concentration, 2) faster labeling in flow (up to 5-10 min ), and 3) easy optical access to a peripheral blood vessels. The high efficiency of CTC labeling is related with frequent NP-CTC collisions in blood flow with the local turbulence  even at similar NP and CTC velocity; on the contrary, for the cancer cell targeting in solid tumor, fast moving NPs even in capillary have to adhere to endothelial cells, penetrate through vessel wall, and migrate to nearest malignant tissue. The low background signals from unbound NPs can be achieved by selecting optimal NP concentration for injection to minimize signals in normal blood but generate detectable signals from each CTC targeted by NPs. Indeed, according to recent data , injection of 109 NPs/mL in 50-100 μL of PBS into the mouse blood circulation provided low background signal from unbound NPs (just few NPs in the irradiated blood volume) and nonspecifically bound NPs by blood macrophages compared to the signals from single CTCs with high local concentration of NPs. In addition, the identification of signals from targeted stem CTCs in the presence of background signals from unbound NPs (e.g., at high NP concentration injected) in blood is possible due to the fast clearance of unbound NPs from circulation (15–30 min, which can be adjusted by proper selection of NP properties) compared to the longer circulation of labeled CTCs . In this study, the optimal time for detecting stem CTCs is 30-40 min after NP injection to allow unbound NP to wash-out.
In our previous [33,36] and current study, PA technique demonstrated both high sensitivity and high dynamic linear range, allowing to detect single CTCs targeted by various number of NPs from minimal level of 10-50 to maximum up to 104 -105 depending on marker expression, amount of injected NPs, targeting efficiency, and sizes of NPs and biomarkers. As a result, PAFC has potential to determine the marker expression in individual cells in vivo because PA signal amplitude is proportional to total number of NPs targeting specific biomarkers in one CTC assuming that NP amount per one biomarker (likely 1-3) can be determined. Because PA signal amplitude also depends on laser energy, this parameter should be carefully optimized for identification of cancer stem cells. For example, at increased laser energy (300-500 mJ/cm2), we could detect PA signals from 30-50% of cancer cells labeled by NP-CD44 complexes even at low CD44 expression. However, only small portion of cells provided strong PA signals indicating significantly higher biomarker expression (overexpression).
The interpretation of the observed rate of CD44+cells in vivo is difficult at current stage of research because the literature data are not quite sufficient. Thus, we demonstrated a novel biophotonic noninvasive platform and its potential for detection of rare stem CTCs rather than the exact quantitative data of percentage of stem CTTs in the circulation. Indeed, the values of CSCs expressed CD44 (e.g., with the phenotype of CD44+/CD24-) varies from a few persantage to 30-80% in bulk breast cancer cells [18,25,46,47]. Moreover, it was recently suggested that the proportion of CSCs varies widely with various factors, including 2-fold change during tumor development . In addition, to estimate the rate of false-positive signals the expression of CD44 in WBCs and/or GNT-CD44 nonspecific binding to other cells should be estimated. The proportion and role of CD44+/Folate+ cells are also unclear yet and it has not yet been determined the level of possible folate receptor expression in breast stem CTCs. Further study, to address significance of various marker(s) for stem CTCs, is currently in progress in our laboratories using an advanced time-resolved multi-spectral PAFC with multicolor NPs for multiplex biomarker targeting . In the current study, we used only one biomarker, i.e., CD44, to identify CSCs because this marker is most relevant to breast CSCs (as reported by other groups [17,18,22,30]). However, as discussed above, biomarker expressions may be heterogeneous among stem and another CTCs [17,18,22,30,38]. Therefore, as the next step, we plan to test the labeling of breast CSCs with Abs specific to other known markers (e.g., CD133 and ALDH) [33,49]. To discriminate PA signals from CD44+ stem CTCs in the background of other potentially CD44+ cancer cells (i.e., bulk CTCs and leukocytes), multi-molecular targeting can be applied using CD44, CD45 and CD24 markers. We assume that CD44+/CD24-/CD45- cells could be related to cancer stem cells, and CD44+/ CD45+ could be leukocytes.
In this pilot study, to make sure that CTCs and, especially, stem CTCs (tumor-initiating cells) were present in the blood and disseminated to distant organs at the point of PA detection, we chose later (not early) stage of cancer with visible hematogenous metastases in liver. However, the high sensitivity of PAFC technique may allow assessing the value of stem CTCs for metastasis prevention that should hold a tremendous importance in cancer research . Our planned future investigations of earliest stages of breast cancer (i.e., the stages of sentinel lymph node metastasis  and even before overt metastasis) would elucidate this issue.
In general, the developed biophotonic technology may fill some gaps in in vivo cancer stem cell research related to (1) stem CTC molecular targeting without the potential influence of the novel labels on the host microenvironment, (2) counting of stem CTCs in low concentrations against a background of blood, (3) estimation of the proportion of stem CTCs within the entire CTC population, (4) estimation of CTC/CSC lifetime in circulation, (5) correlation of stem CTC number with metastasis stages, and (6) significance of stem CTC for early metastasis diagnosis and prevention.
Besides breast tumors, the presented strategy may be broadly applicable to all metastatic cancers and would contribute to a better understanding of cancer stem cell biology at the molecular and cellular levels in vivo, and it would accelerate progress towards attacking the earliest cancer metastasis. We predict that such comprehensive understanding along with this relatively simple technology platform using safe laser parameters and low toxic molecular contrast agents can be quickly translated to humans, possibly as a portable wrist device (e.g., bracelet: see the Supplementary notes in ) for non-invasive diagnosis, specific targeted therapies, individualized selection of therapeutic modalities for a specific patient and monitoring of cancer recurrence and disease-free status.
This work was supported in part by the National Institute of Health grant nos R01EB000873, R01CA131164, R01 EB009230, and R21CA139373 (V.P.Z), the National Science Foundation grant nos DBI-0852737 (V.P.Z) and CMMI-0709121 (J.-W.K.) and the Arkansas Biosciences Institute (J.-W.K and V.P.Z). We would like to thank Y.A. Wang of Ocean Nanotech, LLC for providing MNPs, H.-M. Moon for his assistance with GNT synthesis, T. Kelly for cell culturing and S. Fergusson for his assistance with laser measurements.