Real-Time Detection of Circulating Tumor Cells in Living Animals

The ability to count and track single cells flowing in the bloodstream of a live animal is a central goal in the growing field of circulating tumor ce...
1 downloads 0 Views 1MB Size
Download PDF
Subscriber access provided by EDINBURGH UNIVERSITY LIBRARY | @ http://www.lib.ed.ac.uk

Communication

Real-time detection of circulating tumor cells in living animals using functionalized large gold nanorods Rebecca Dutta, Orly Liba, Elliott Daniel SoRelle, Yonatan Winetraub, Vishnu C Ramani, Stefanie S. Jeffrey, George W. Sledge, and Adam de la Zerda Nano Lett., Just Accepted Manuscript • DOI: 10.1021/acs.nanolett.8b05005 • Publication Date (Web): 21 Mar 2019 Downloaded from http://pubs.acs.org on March 21, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

Real-time detection of circulating tumor cells in living animals using functionalized large gold nanorods Rebecca Dutta1,2, Orly Liba1,2,3, Elliott D. SoRelle1,2,4, Yonatan Winetraub1,2,4, Vishnu C. Ramani5, Stefanie S. Jeffrey5, George W. Sledge6, & Adam de la Zerda1,2,3,4,7* 1Department 2Molecular 3Electrical

of Structural Biology, Stanford University, Stanford, CA 94305.

Imaging Program at Stanford and the Bio-X Program, Stanford, CA 94305.

Engineering, Stanford University, Stanford, CA 94305.

4Biophysics

Program at Stanford, Stanford, CA 94305.

5Department

of Surgery, Stanford University School of Medicine, Stanford, CA, 94305.

6Department

of Medicine, Stanford University Medicine, Stanford, California

7The

Chan Zuckerberg Biohub, San Francisco, California 94158, USA

*Correspondence to: [email protected], +1(650)575-1597 BioRxiv doi: https://doi.org/10.1101/498188

ACS Paragon Plus Environment

1

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 19

Abstract Optical coherence tomography (OCT) can be utilized with significant speckle reduction techniques and highly-scattering contrast agents for noninvasive, contrast-enhanced imaging of living tissues at the cellular scale. The advantages of reduced speckle noise and improved targeted contrast can be harnessed to track objects as small as 2 μm in vivo, which enables applications for cell tracking and quantification in living subjects. Here we demonstrate the use of Large Gold Nanorods (LGNRs) as contrast agents for detecting individual micron-sized polystyrene beads and single myeloma cells in blood circulation using speckle modulating-OCT (SM-OCT). This report marks the first time that OCT has been used to detect individual cells within blood in vivo. This technical capability unlocks exciting opportunities for dynamic detection and quantification of tumor cells circulating in living subjects. Keywords: OCT, Circulating tumor cells, Gold nanorods The ability to count and track single cells flowing in the bloodstream of a live animal is a central goal in the growing field of circulating tumor cell (CTC) research.1,2 CTCs are cells that have shed from a primary tumor and entered into the vasculature, and they may act as seeds for future metastases. Thus, the enumeration of CTCs in cancer patients has important prognostic and therapeutic value. Most current in vitro attempts at enumerating and isolating CTCs such as CellSearch have focused on blood biopsies.3–5 However, factors including the rarity of CTCs (~1-10 CTCs/mL of blood), limited blood sample volume, and typical sampling frequency virtually preclude the acquisition of dynamic information about the number of tumor cells in the bloodstream using these methods.6,7 In principle, in vivo imaging could enable dynamic tracking of tumor cells in diagnostic contexts as well as for therapeutic applications such as screening drug effects on CTC numbers in the bloodstream. In recent years, in vivo imaging techniques such as MRI, photoacoustic tomography, and positron emission tomography (PET) have benefited from the use of tailored contrast agents, which can enhance detection sensitivity and specificity for targets of interest. However, due to the limited spatial resolution of these techniques, each is incapable of examining tissues at the single-cell scale.8,9 Single-impulse panoramic photoacoustic computed tomography (SIP-PACT) has been used to image CTC clusters in the microvasculature of living mice; however, SIP-PACT cannot resolve individual cells and further relies on the intrinsic scattering of specific cell types, such as melanoma cells.10 In vivo flow cytometry has also been used to detect moving cells, but this method requires the use of fluorescent proteins.11 Optical coherence tomography (OCT) is a noninvasive, label-free, optical imaging technique that uses low-coherence interferometry to visualize tissue at millimeters depth of optical penetration and subcellular resolution.12 OCT is used widely for ophthalmological applications13; however, as with all imaging technologies that employ coherence-based detection, speckle noise hinders the diagnostic potential of OCT at the cellular level by effectively and significantly reducing spatial resolution.14 The challenges arising from speckle noise are especially pronounced when attempting to detect small, fast-moving structures (e.g., CTCs) since rapid blood flow precludes the use of B-scan (frame) averaging, a standard technique for speckle reduction.15 Instead we sought to develop an approach based on A-scan (or line) averaging, which is substantially faster than frame averaging and thus minimizes local image blur. While Ascan averaging reduces time-varying artifacts such as photon shot-noise, it typically does not

ACS Paragon Plus Environment

2

Page 3 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

remove speckle because the imaged volume is static in this small amount of time. In order to reduce speckle while leveraging the benefits of the A-scan averaging approach, we performed image acquisition with our recently-reported OCT variant called Speckle-Modulating OCT (SMOCT). SM-OCT dynamically manipulates the phase of the incident (and reflected) light interacting with the sample through the use of a diffuser to virtually eliminate speckle noise, enabling OCT to have the unprecedented ability to image individual cells within turbid tissue.16 The resulting A-scan-averaged SM-OCT images exhibit negligible speckle noise without degradation of image resolution; this enables the detection of structures that were otherwise obscured. Nanoparticles with tunable size and spectral characteriztics can be used to meet specific imaging needs for OCT. Previous work in our lab has developed and employed biostable large gold nanorod contrast agents (LGNRs, ~100 x 30 nm), which backscatter significantly more light than commonly-used gold nanorods. For a typical experiment, LGNRs are synthesized to achieve a well-defined longitudinal plasmonic resonance band matched to the OCT light source. The LGNRs are then purified and coated with poly(sodium 4-styrenesulfonate) (PSS) for stability and biocompatibility. Biostable LGNRs can then be incubated with living cells for passive labeling through nonspecific uptake of the contrast agent. LGNRs can also be subsequently functionalized through the bioconjugation of cell-specific antibodies like antiepithelial cell adhesion molecule (anti-EpCAM).17,18 By combining the use of optimized speckle reduction techniques and contrast agents, the spatial resolution inherent to OCT can be harnessed to visualize individual labeled microscale species moving in the bloodstream. This new technical capability is first demonstrated by imaging LGNR-labeled 2 µm polystyrene beads circulating within the pinna vasculature of a living mouse after intravenous injection. Then, the number of labeled beads detected at a cross section of a blood vessel in the mouse ear is counted temporally to determine their circulation time. Next a similar capability is demonstrated for imaging intravenously injected RPMI-8226 myeloma cells, pre-incubated with LGNRs, as they flow through blood vessels. Figure 1 shows the schematic of the experimental setup for the imaging of injected LGNR-labeled cells in the blood vessel of a living mouse. Objects in the bloodstream that take up less volume than a single voxel and exhibit poor scattering need to be adequately labeled with LGNRs (Fig. 1a) to achieve sufficient scattering enhancement for subsequent visualization with SM-OCT. To test the feasibility of this approach, we initially produced mimics of labeled cell-sized objects for characterization. To this end, 2 µm diameter NeutrAvidin-coated polystyrene beads, which are smaller than a single imaging voxel, were incubated with biotin-functionalized LGNRs. SEM images indicated that each bead had an average of 159 LGNRs attached in random orientations and distribution across the bead surface after excess LGNRs were washed off (Fig. 2a,b). Next, suspensions of unlabeled beads and LGNR-beads were prepared into water or freshly-collected whole blood and imaged in capillary tubes using SM-OCT. While individual unlabeled and labeled beads were visible in water, only the LGNR-labeled beads could be visually identified in blood. This result demonstrated that, in principle, cell-sized objects previously invisible to SMOCT could be readily detected with adequate LGNR labeling (Fig. 2c). The SM-OCT signal (on linear scale) from LGNR-labeled beads was two-fold higher than that from unlabeled beads when prepared in water (Fig. 2d). When prepared in blood, unlabeled beads were no longer detectable; however, the LGNR-labeled beads were readily discernible from surrounding blood background signal using SM-OCT. (Fig. 2e). The number of beads detected in a single frame (only considering the top half of each tube, which is less influenced by blood absorption) was representative of the initial concentration of beads that were mixed into the whole blood. As the ear of a mouse is approximately 100 µm thick and the top half of the capillary tube is 200 µm,

ACS Paragon Plus Environment

3

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 19

this mode of analysis did not affect subsequent in vivo imaging. The SM-OCT signal of the labeled beads was compared to regions of blood in the blood-only capillary tube that were at the same depth, to account for attenuation of light from red blood cells. When the SM-OCT signal was normalized for depth-dependent attenuation of signal, the signal-to-background ratio (SBR) of labeled beads to blood is 6.3 ± 1.0 (Mean ± SD) fold higher. Given the ability to detect sub-cellular-sized labeled polystyrene beads using SM-OCT, we proceeded with labeling cells derived from tissue culture with LGNRs to determine their detection in blood vessels of living animals. Due the difficulty of obtaining and propagating primary cultures of CTCs from human patients19–21, we used cells from the RPMI-8226 myeloma cell line, which is grown as a suspension culture and is a reasonable proxy for CTCs with regard to circulation mechanics, as suggested by the line’s similarity to circulating white blood cells.22 Additionally, unlike cancer cells of epithelial origin, myeloma cells express low levels of cell surface integrins and therefore exhibit minimal cell-cell or cell-ECM interactions23 that would encourage cell clustering or adhesion to blood vessel walls, respectively. RPMI-8226 cells were incubated for one hour with an excess of LGNRs and washed thoroughly to remove any unbound LGNRs. To verify LGNR labeling of RPMI-8226 cells, labeled cells and unlabeled cells was imaged using a hyperspectral dark-field microscope over a spectral range of 400-1000 nm. A custom machine-learning algorithm for adaptive detection of LGNRs based on their unique optical scattering signature24 was used to verify LGNR labeling of the RPMI-8226 cells. The majority of cells incubated with LGNRs show a significant amount of LGNR+ pixels that have strong scattering in the near-infrared region, suggesting dense cellular uptake of LGNRs. Pixels that were classified as LGNR+ are depicted in orange on the hyperspectral image (Fig. 3a). As expected, the control sample of unlabeled cells show no presence of this LGNR scattering spectrum. Suspensions of labeled and unlabeled cells in complete cell culture media were imaged in adjacent capillary tubes (Fig. 3b). While single unlabeled cells generally produced no detectable SM-OCT signals due to poor scattering relative to the media background, cells labeled with LGNRs could be clearly distinguished in the capillary tube owing to the enhanced scattering from the LGNR contrast agents. LGNR-labeled RPMI-8226 cells had three-fold higher SM-OCT signal compared to unlabeled cells (Fig.3c). Unlabeled and labeled cells were also suspended in whole mouse blood to determine whether the signal intensity of the labeled cells was higher than the baseline intensity from the whole blood. Unlike unlabeled cells, LGNR-labeled RPMI8226 cells in whole blood could be observed in the top half of the capillary tube (the lower half of each tube was not analyzed because of light attenuation from blood) (Fig.3b). The signal from labeled cells was compared to the signal from blood at similar depths, and it was found that the LGNR-labeled RPMI-8226 cells had two-fold greater signal than blood (Fig. 3d). As observed for the labeled beads, the signal from labeled cells was highly punctate, allowing for their clear demarcation even against high backgrounds in physiological conditions. These results indicated that RPMI-8226 cells can be labeled sufficiently with LGNRs such that their increased backscattering should enable their detection in blood vessels using SM-OCT. To mimic blood flow, a programmable syringe pump was used to pass whole blood with labeled beads through a capillary tube at biologically relevant speeds. We were able to detect labeled beads in whole blood within the capillary tubes at flow rates of 10 cm/s and 20 cm/s, proving that labeled beads can be detected with SM-OCT at different velocities including the physiological upper bounds of arterial flow in humans 25(Fig. 4a,b). To confirm detection of labeled beads in a background of flowing blood, time-lapse SM-OCT images were acquired at the location of the punctate signal that was suspected to be a flowing bead. We found that flowing LGNR-beads had a distinct SM-OCT signal pattern over consecutive A-scans compared

ACS Paragon Plus Environment

4

Page 5 of 19 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Nano Letters

to the background signal from blood. The blood background exhibited low, spatially-diffuse signal and the punctate signal exhibited a more predictable localized pattern over the time interval in which the bead was present within the imaged volume (Fig. 4c-e). In an analogous setup to assess detection under flow conditions, LGNR-labeled RPMI-8226 cells in whole blood were passed through a capillary tube using the automatic syringe pump. The labeled cells were also detectable at flow rates of 10 cm/s and 20 cm/s, indicating that labeled cells can be detected with SM-OCT in a range of biologically-relevant flow speeds (Fig. 4f,g). To confirm detection of labeled cells in a background of flowing blood, a time-lapse analysis was performed of the SM-OCT signal at the location of the punctate signals that were suspected to be flowing cells (Fig. 4h-j). As was observed for labeled beads, flowing LGNR-cells produced a distinctly different SM-OCT signal pattern over consecutive A-scans relative to blood background signal. Notably, the number of LGNR-cells counted in any given frame corresponded to the concentration of cells that were added into whole blood. We predicted a count of 37.8 beads in every frame, and found that we could count 17.4 ± 5.4 (Mean ± SD), or approximately half of the expected bead count. This count is consistent with the predicted concentration, because only the top half of the capillary tube can be analyzed due to the attenuation of light by the red blood cells and indicates that the scan speed of SM-OCT is fast enough to capture every cell flowing through the cross-section of the capillary tube at flow speeds up to 20 cm/s. Next, we tested our detection scheme for labeled beads and RPMI-8226 cells in the bloodstream of live mice (n=3 for each injection type). Beads and cells were labeled via the same procedures described for the capillary tube experiments. In separate experiments, each labeled sample was injected intravenously into the tail vein of a mouse. Image acquisition at the blood vessels of the mouse pinna was performed every two minutes post-injection for 30 minutes, after which point no beads or cells were detected in the blood vessel. Labeled beads showed up in the blood vessel as high-intensity, localized puncta that had at least two-fold higher signal intensity in comparison to the immediate background (Fig. 5a). The puncta pattern changed with every frame since the blood flow within the blood vessel is faster than the image acquisition rate. These punctate signals were similar in size and signal intensity to those seen in the capillary tube imaging experiments. To estimate the circulation time of the beads, the punctate signals inside the vessels were manually counted in each frame. The number of puncta was graphed over time (Fig. 5b). Sixteen minutes post-injection, the number of puncta counted over 100 consecutive frames was found to be approximately similar to the number of puncta counted pre-injection, indicating that the labeled beads had largely been cleared from the circulation (two-tailed t-test, p < 0.05). For the labeled cell experiments, approximately five million cells were tail-vein injected incrementally to allow for sufficient clearance of cells through the capillary vessels of the lungs. Images were acquired after each injection increment at the longitudinal cross-section of a blood vessel in the pinna of the mouse. Lengthwise sections were imaged at the middle of the blood vessel (parallel to blood flow) since the probability of observing a cell within an individual transverse cross-section is rare, whereas a lengthwise optical slice allows for the sampling of a much larger region of the blood vessel at any given time. Bright punctate signal determined to be LGNR-labeled RPMI-8226 cells were manually counted every two minutes over a 30 minute period. As with the beads, there was an increase in localized punctate regions with at least twofold higher signal intensity compared to the background after the injection, indicating the presence of LGNR-labeled RPMI-8226 cells flowing in vivo (Fig. 5a). The ability to detect these cells in the blood vessel of a living mouse once they have been sufficiently labeled with the LGNR contrast agents demonstrates that SM-OCT has a sufficient resolution and sensitivity to

ACS Paragon Plus Environment

5

Nano Letters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 19

identify circulating cells. The number of observed puncta were plotted versus time to better understand the circulation dynamics of the injected cells (Fig. 5c). Unlike the rapid circulation and clearance of labeled beads in the bloodstream, labeled cells exhibited a less-predictable circulation pattern defined by a delayed gradual increase in the number of punctate signals counted over time as well as a more gradual return to baseline. There are several possible explanations for these dynamics. While beads do not have a tendency to adhere to vasculature, cells needed to be injected incrementally to prevent cellular aggregates in pulmonary microcirculation, which could potentially lead to the death of the mouse. Prior research has noted the limited clearance of intravenously injected tumor cells that arrest in the pulmonary vasculature26. Coupled with the distal location of the ear pinna blood vessel, the circulation pattern of cells becomes more stochastic and depends on a variety of factors such as the tendency of the cells to adhere to vessels and their deformability. Bright-field microscopy of blood collected via cardiac puncture from a mouse injected with labeled cells showed that less than 1% of the 1095 cells counted are not single cells. Of the 10 multi-cellular clumps that were counted in the sample, 9 were clumps consisting of 2 RPMI-8226 cells, and 1 was a clump of 3 RPMI-8226 cells. If all five million of the cells injected made it into circulation, punctate signals should be seen around once every 8 frames in a blood vessel that is approximately 50 µm in size; however, bright punctate signals were spotted in the blood vessel approximately once every 14 frames, suggesting that less than half of the cells injected make it into systemic circulation. If all 1% of cells that are multicellular clumps make it into systemic circulation, 97% of cells detected would still be single cells. This strongly suggests that the cells in circulation that are detected by SM-OCT are, indeed, single cells and not clusters of cells. We encountered two types of rare, endogenous features with noticeable signal intensity in the blood vessels of the six mice prior to injection with labeled beads or cells. The first type involved punctate signals that showed up consistently in the same region, indicating that they may be inherent static scatterers, such as proteins or collagenous fibers, found in the blood vessel architecture. If consistent scatterers were located in a specific region of the blood vessel prior to the injection, punctate signals were not counted in that location. This eliminated false positive detection caused by inherent scatterers in the vessel. The second and rarer type of endogenous signals noticed were seen in the lumen of the blood vessel and did not appear in the same spot over multiple frames. In the in vivo experiments in which labeled beads were injected, we looked at 100 pre-injection frames in blood vessels with approximate volumes of 2.50*10^3 µm2 and found an average of 1.33± 0.942 (mean ±SD) punctate signals. In the labeled cells in vivo experiments, we looked at 100 pre-injection frames in longitudinally imaged blood vessels with approximate volumes of 3.75*10^3 µm2 and found an average of 9.00± 1.41(mean ±SD) punctate signals. Using a student’s t-test, the number of punctate signals during the preinjection phase was found to be significantly lower than the number of punctate signals found at minute 2 post-injection in the labeled beads experiment (t(4)=4.62, p