Optical imaging (OI) provides real-time clinical imaging capability and simultaneous molecular, morphological, and functional information of disease processes. In this study, we present a new interventional OI technique, which enables in vivo visualization of three distinct pathologic zones of ablated tumor periphery for immediate detection of residual tumors during a radiofrequency ablation (RFA) session. Rabbits with orthotopic hepatic tumors were divided into two groups (n = 8/group): incomplete RFA and complete RFA. Indocyanine green-based interventional OI was used to differentiate three pathological zones: ablated tumor, transition margin, and residual tumor or surrounding normal liver—with quantitative comparison of signal-to-background ratios among the three zones and between incompletely and completely ablated tumors. Subsequent ex vivo OI and pathologic correlation were performed to confirm the findings of interventional OI. Interventional OI could differentiate incompletely or completely ablated tumor peripheries, thus permitting identification of residual tumor. This technique may open new avenues for immediate assessment of tumor eradication during a single interventional ablation session.
Interventional optical imaging can instantly visualize pathologic zones of ablated tumor peripheries to detect residual tumors, which could revolutionize current image-guided interventional oncologic ablation techniques.
Advances in percutaneous ablation technologies, such as radiofrequency ablation (RFA), microwave ablation, cryoablation, and irreversible electroporation, have greatly improved the management of patients with solid organ malignancy. These interventional oncologic therapies have significant advantages: they are minimally invasive and repeatable, offer low risk of complications, and require only a short hospital stay. However, with current technology, these interventional ablation techniques are most effective for eradicating small (<3 cm) lesions (1, 2). For ablation of larger or irregular lesions, these methods often suffer from incomplete tumor eradication at the periphery of the lesion (3, 4). Ultimately, residual viable tumor cells then serve as the source for tumor persistence and recurrence, leading to treatment failure (5, 6).
Incomplete tumor ablation is attributed to the challenge of creating an effective and safe ablation periphery with a surgical margin or “clear safety margin” beyond the tumor confinement (7, 8). To date, CT, MRI, ultrasound imaging, and PET/CT, have been used for assessment of residual tumors after completion of initial ablation treatment (9–11). These tools usually detect residual or persistent tumor days or weeks to months after terminating ablation treatments (12–14). Thus, it is imperative to work towards immediate or instant detection of residual tumor during the actual ablation session.
For immediate detection of residual tumors, it is essential to have a tool that enable visualization, in vivo, of tissue at the ablated tumor periphery. Pathologically, the periphery of an ablated tumor possesses three distinct zones (from inner to outer): (1) the ablation zone, described as the necrotic area of ablated tumor; (2) the transition zone (or hyperemic rim), described as the boundary between the ablation zone and surrounding normal tissue; and (3) the surrounding normal tissue (15, 16). To date, assessment of the three pathologic zones has only been accomplished after the completion of ablation treatments via either follow-up imaging, or confirmed by postoperative ex vivo pathology, if available.
Optical imaging (OI), the newest member of the modern clinical medical imaging family, provides real-time imaging capability, simultaneous molecular/morphologic and functional information of disease process. It is cost-effective, portable, nonionizing, and generally well tolerated by patients (17, 18). Application of OI dyes (optical contrast agents), such as indocyanine green (ICG) or targeted optical probes, has further enhanced the usefulness of OI techniques, providing better detection of relatively deep-seated lesions and better guidance of cancer treatments (19–21). After intravenous administration of ICG, it has a short half-time of 150 to 180 seconds (22). ICG uptake into normal hepatocytes is mediated by two principal sinusoidal surface transporters, Na+/taurocholate cotransporting polypeptide and organic anion transporting polypeptide 8, and it is cleared exclusively by the liver (23). It has been used previously in large animals, highly successfully particularly in pigs and interventional procedures (24–26). In open liver surgery for example, the operators use an ICG-based external OI imager to obtain surface fluorescence images for planning of liver tumor resection (23, 27). However, the surface OI for detection of deep-seated lesions is challenging, because the tissue penetration depth of current ICG-based fluorescence imaging is only approximately 1 cm (22, 23). This problem, the limited penetration depth of current OI, can be solved by image-guided interventional approach, which is capable of precisely guiding micro-OI detectors to the targets, avoiding tissue scattering and reflection along the pathway of OI light.
The purpose of this study was to fully apply the advantages of OI by extending its use from open surgeries to minimally invasive interventional oncology for thermal ablation of larger tumors. The innovative design of the ICG-based interventional real-time OI technique should enable visualization, in vivo, of the three distinct pathologic zones and thereby provide immediate detection of small residual tumor at the ablated tumor periphery during a single interventional ablation session.
Materials and Methods
This study included three phases: (i) optimizing in vitro experimental protocol of ICG parameters, such as dose and detection time-window to achieve high-contrast and high-quality ICG-based OI for in vitro experiments using VX2 tumor cells, and observing ICG uptake by primary hepatocytes; (ii) in vitro evaluation of the functionality and sensitivity of the interventional OI system in differentiation 1.5 mm3 living tumor cell clusters from 1.5 mm3 dead tumor cell clusters; and (iii) validating the technical feasibility of using the new ICG-based interventional OI system to instantly visualize the three pathologic zones of ablated tumor periphery and thereby identify residual tumors during an ablation session, via serial in vivo experiments using living animal models with orthotopic hepatic VX2 tumors.
In vitro evaluation and confirmation
The cell lines VX2 tumor cells were obtained in 2019 from Cell Resource Center for Biomedical Research, Tohoku University, Japan. The hepatocyte cell line (catalog no. HUCPG) was purchased in 2021 from Lonza Biosciences Inc. These cells were tested by short tandem repeat analysis, validated to be free of Mycoplasma, which were used within 3 months. In addition, the cells were cultured within 10 passages for all experiments, and all the cell lines were authenticated by Laboratory of the Government Chemist standards. VX2 tumor cells, which were derived from a papilloma virus-induced rabbit skin squamous cells, and hepatocytes (Lonza Biosciences Inc., catalog no. HUCPG) were cultured with (i) 1 × 105 cells seeded in 35 × 10-mm petri dishes (Falcon; Becton Dickinson and Company) for in vitro optical/x-ray imaging; and (ii) 5 × 104 cells per chamber seeded in four-well chamber slides (Lab-Tek; Thermo Fisher Scientific) for fluorescent microscopy confirmation. All VX2 tumor cells were cultured in RPMI1640 Medium (Gibco) supplemented with 10% FBS (Gibco), and incubated at 37° with a 5% carbon dioxide atmosphere. When 90% confluence via two passes was reached, VX2 tumor cells were treated with ICG (Patheon Italia S.P.A.) at different concentrations for various incubation times. The hepatocytes were cultured with thawing medium (Lonza Biosciences Inc., catalog no. MCHT50), plating medium (Lonza Biosciences Inc., catalog no. MP100), and maintenance medium (Lonza Biosciences Inc., catalog no. CC-3198), and incubated at 37° with a 5% carbon dioxide atmosphere. The experiment for each of the different cell groups with various treatments was repeated six times for statistical analysis.
Optimizing in vitro experimental protocol for sufficient ICG uptake by VX2 cells and hepatocytes
VX2 tumor cells were treated with ICG at (i) concentrations of 0, 25, 50, 75, 100, 125, 150, 175, and 200 μg/mL for 24 hours; (ii) a concentration of 100 μg/mL with various incubation times of 1, 2, 4, 8, 12, 24, and 48 hours, and (iii) nontreated cell groups served as control. Hepatocytes were treated with ICG at (i) a concentration of 100 μg/mL with various incubation times of 1, 2, 4, 8, 12, 24, and 48 hours, and (ii) nontreated cell groups served as control. For fluorescence microscopy, these ICG treated cells were washed twice with PBS to remove the free ICG, fixed with 4% paraformaldehyde, and then dried at room temperature. The cells were then counterstained with 4′,6-diamidino-2-phenylindole (DAPI; SouthernBiotech) and imaged with a fluorescence microscope (IX73; Olympus), which was connected with a 130-W light source (U-HGLGPS; Olympus) and equipped with band pass filters (Excitation: FF01-769/41-25, Emission: FF01-832/37-25; Semrock).
For in vitro optical/x-ray imaging using the optimized ICG parameters above, the treated cells were rinsed twice with PBS, trypsinized, centrifuged, and then 3 × 105 VX2 cells or hepatocytes were resuspended in 0.6-mL microcentrifuge tubes (Fisherbrand; Thermo Fisher Scientific Inc.) with 0.2-mL cell culture medium. Subsequently, the cell-containing tubes were imaged with the optical/x-ray imaging system (In vivo Xtreme; Bruker) at the excitation wavelength of 760 nm, emission wavelength of 830 nm, field of view of 120 × 120 mm, and exposure time of 1 minute, whereas the fluorescence SIs of different cell groups were measured using the Bruker's software and compared quantitatively.
Building the interventional OI system
The interventional OI system was constructed using a 17-gauge 30-cm micro-OI needle. The OI needle was connected with (i) a 250-W halogen light source (KL 2500 LCD; Schott), which was equipped with a filter wheel containing an excitation filter (ET775/50x; Chroma Technology Corp.); and (ii) a metal-oxide-semiconductor (sCMOS) camera (Pco.edge 4.2 bi; PCO AG) through a band pass emission filter (ET845/55m; Chroma Technology Corp.). The sCMOS camera was connected to a personal computer, which displayed the OI images with an image receiving software (Camware64. Ink). The interventional OI system acquired the near-infrared fluorescence signals of ICG-cells with a resolution of 2,048 × 2,048 ppi, field of view of 1.1 × 1.1 mm, and pixel size of 6.5 × 6.5 μm2 over an exposure of 2 seconds.
In vitro interventional OI of ICG-treated VX2 cells (ICG-cells) with and without ablation-heated treatments
VX2 tumor cells were treated using the optimized protocol (i.e., using the ICG concentration of 100 μg/mL with the detection time-window at 24 hours after ICG administration, as established above for in vitro experiments). A total of 1 × 107 ICG-cells were transferred into a 15-mL tube (SuperClear; VWR International, Inc.). Twelve tubes of cells were randomly divided into two groups with six tubes per group including: (i) a dead (nonsurviving) ICG-cells group that was treated with an ablation heat at 80°C for 10 minutes; and (ii) a living (surviving) ICG-cells group without ablation heat treatment to serve as the control. The dead ICG-cells and living ICG-cells were confirmed by fluorescent microscopy.
In vivo technical validation
Creation of animal models with orthotopic hepatic tumors
The animal protocol was approved by the Institutional Animal Care and Use Committee. Under general anesthesia, the left liver lobe of each of 16 adult female New Zealand white rabbits, weighing 2 to 3 kg, was exposed through a subxiphoid abdominal incision. Fresh VX2 tumor tissues (1 mm³) were inoculated into the left liver lobe. The abdominal incision was closed with layered sutures.
Two weeks after the tumor implantation, the orthotopic hepatic tumors received RFA treatment. Under real-time ultrasound imaging guidance, a multipolar RFA electrode, connected to a generator (WE7568-II; Welfare Electronics Co.) was positioned in the tumor. The animals were randomly divided into two tumor model groups (n = 8/group): (i) incomplete ablation by partially deploying RFA electrode prongs; and (ii) complete ablation by fully deploying electrode prongs. The partial or full prong deployment at the target tumor margins were precisely guided by real-time ultrasound imaging and further confirmed by x-ray imaging. All tumors were ablated at a lethal heat of 80°C for 4 minutes.
In vivo interventional OI
Twenty-four hours before RFA procedures, all animals received intravenous administration of ICG via an auricular vein at a dose of 0.5 mg/kg body weight (28). Before and immediately after RFA, the interventional micro-OI needle was positioned in the normal liver parenchyma to achieve background OI. Then, under real-time ultrasound imaging-guidance, the micro-OI needle was advanced to the tumor periphery, to generate intratumoral “inside-output” OI at six different directions/points around the tumor periphery (i.e., at 3, 6, 9, 12 o'clock along the equator, as well as the north and south poles).
Ex vivo OI confirmation
After achieving in vivo OI, the rabbits were euthanized, and the left hepatic lobes were harvested. The liver specimens were then imaged using the Bruker In-vivo Xtreme OI system at the excitation wavelength of 760 nm, emission wavelength of 830 nm, field of view of 120 × 120 mm, 1.40 F-Stop, 2 × 2 binning, and an exposure time of 1.0 minute. The ex vivo OI for the gross liver specimens of ablated tumors functioned as the “gold standard” to corroborate the findings achieved by the new interventional OI system.
Pathologic and laboratory confirmation
The samples were cut off along the needle tract. One portion of the tumor-containing specimen was flash-frozen and cryosectioned at 8-μmol/L thickness. The sections were counterstained with DAPI for fluorescent microscopy and were analyzed with NADH staining. The remaining tumor-containing specimen was fixed with 4% paraformaldehyde, embedded in paraffin, sectioned at 4-μmol/L thickness, and analyzed with hematoxylin and eosin (H&E) staining.
The optical/x-ray images were analyzed using the Bruker imaging analysis software, and the interventional optical images were analyzed using the NIH ImageJ software. All regions of interest on the optical images were set at different locations, including the ablated tumors, residual tumors, transition zone, and adjacent liver parenchyma of in vivo experiments, to obtain the average fluorescence SIs. The SBRs were then calculated by dividing the mean fluorescence SI of the residual, transition zone, or ablated tumors by the mean fluorescence SI of the adjacent liver parenchyma.
All statistical analyses were performed using GraphPad Prism (version 8). As indicated in the figure legends, results were analyzed by a paired t test. Data were expressed as mean ± SD. A probability level of less <0.05 was statistically significant.
In vitro optimization of ICG dose and time window for OI of ICG-cells
Fluorescent microscopy showed that ICG was taken up by VX2 tumor cells through emission of red fluorescence, which was not seen in VX2 cells without ICG treatment. The red fluorescence signal intensity (SI) of ICG-cells became more intense with increased ICG concentrations from 0 to 100 μg/mL. Due to the quenching effect (29), the fluorescence SI decreased as the ICG concentrations increased from 100 to 200 μg/mL (Fig. 1A). In vitro quantitative optical/x-ray images of ICG-cells further confirmed similar results of fluorescent microscopy, demonstrating the fluorescence SIs of ICG-cells increased as ICG concentrations increased from 0 to the peak of 100 μg/mL, and then decreased as ICG concentrations increased (Fig. 1B and C). Thus, we determined the optimum ICG concentration at 100 μg/mL to further investigate the optimum time-window for the best detection of ICG-cells in the in vitro experiments.
For optimization of the detection time-window, fluorescent microscopy demonstrated that ICG-emission become more intense as the incubation times increased from 0 to 24 to 48 hours (Fig. 2A and B) and in vitro quantitative optical/x-ray images of ICG-cells and ICG-treated hepatocytes further confirmed the results (Fig. 2C, D, and E). Thus, we decided that 12 hours post-ICG treatment for ICG-treated hepatocytes and 24 hours post-ICG treatment for ICG-cells were the best time window for in vitro interventional OI.
In vitro evaluation on functionality and sensitivity of the interventional OI system
We first successfully built the interventional OI system (Fig. 3A and B). To evaluate the efficacy of interventional OI in differentiating living (viable) ICG-cells from dead (nonviable) ICG-cells, a cell phantom was created, consisting of a clear centrifuge tube that contained two separate chambers in the inferior portion (Fig. 4A and B). The living cells were centrifuged in one chamber with PBS, whereas the dead tumor cells (treated by 80°C of ablation heat) were present in the opposite chamber. Thus, the living and dead cell clusters, divided by the septum, mimicked the ablated tumor periphery. Volumes of both living and dead tumor cell clusters in the two chambers of the centrifuge tube was approximately 1.5 mm3 (Fig. 4B).
The interventional micro-OI needle was vertically positioned above the two cell-containing chambers, to generate in vitro interventional OI (Fig. 4A–C). The interventional OI could clearly visualize the three distinct zones: living cells, septum, and dead cells. Meanwhile, it detected the high fluorescence signals emitted from the 1.5-mm3 living cell cluster in sharp contrast to the low fluorescence signals from the 1.5-mm3 nonviable cell cluster (Fig. 4C), which was subsequently confirmed by fluorescent microscopy, displaying as the decreased ICG signals of cells with smudged nuclei after the ablation heat treatment (Fig. 4D and E). The quantitative analysis of fluorescence SIs further confirmed a significantly higher average fluorescence SI of the living tumor cells, in comparison with the ablation-treated nonviable tumor cells (187.0 ± 9.4 au vs. 65.1 ± 4.0 au, P < 0.001; Fig. 4F). Thus, these in vitro experimental results demonstrated the capability of interventional OI in differentiation of 1.5-mm3 living ICG-cell clusters from 1.5-mm3 dead ICG-cell clusters.
Creation of hepatic VX2 tumor models
All animal models with orthotopic hepatic VX2 tumors were successfully created in the left lobes of rabbit livers (Fig. 5A), detected as hyperechoic lesions on ultrasound imaging (Fig. 5B). The axial (x) and longitudinal (y) diameters of tumors and tumor depths (z) were measured on the ultrasound images at the largest dimension. The volume of each tumor was then calculated according to the following equation: volume = xyzπ/6. The mean tumor volume in the two groups were 3.95 ± 0.44 cm3 and 3.91 ± 0.50 cm3, respectively, between which there was no significant difference (P = 0.872). The tumor perfusion was assessed by contrast-enhanced ultrasonography before ablation, and all the lesions were well perfused (Fig. 5C).
Under real-time ultrasound imaging guidance, the interventional micro-OI needle was precisely placed along the ablated tumor margin for intratumoral “inside-output” imaging (Fig. 5D). The interventional OI system could clearly visualize the three pathologic zones at the ablated tumor periphery: (i) the completely ablated tumor as the area with the lowest optical signal; (ii) the transition zone as a high optical signal zone between the ablated tumor and the adjacent normal liver parenchyma; and (iii) the adjacent normal liver parenchyma (Fig. 5E). These findings of interventional OI were further confirmed by both fluorescent microscopy and pathology with H&E staining (Fig. 5F and G). These results revealed that the interventional OI can sensitively identify the three pathologic zones of ablated tumor periphery.
Validation of technical feasibility of interventional OI to immediately detect residual tumor
To specifically create incomplete and complete tumor ablation models, the prongs of RFA electrode were partially or fully deployed (Figs. 6A–C and F–H). For the incomplete tumor ablation model, the interventional OI system detected clearly the three ablation zones of residual tumor, the transition zone, and fully eradicated tumor. The optical signal of residual tumor was higher than those of the transition zone and fully ablated tumors (Fig. 6D). Quantitative analysis further confirmed the significantly higher mean signal-to-background ratio (SBR) of residual tumor, compared with those of the transition zone and fully ablated tumor (residual tumors vs. transition zone vs. ablated tumors = 2.77 ± 0.32 vs. 1.41 ± 0.04 vs. 0.57 ± 0.05, P < 0.001; Fig. 6E). For the complete tumor ablation model, the optical signal of the ablated tumor was lower than that of transition zone (Fig. 6I), which was further confirmed by the interventional OI quantification, as a significantly lower mean SBR for the completely ablated tumors compared with the transition zone (0.57 ± 0.05 vs. 1.40 ± 0.07, P < 0.001; Fig. 6I). These results suggested that the interventional OI can instantly detect residual tumor tissue during an ablation session.
Confirmation with ex vivo OI
The complementary ex vivo OI of the grossly sectioned specimens functioned as a “gold standard” to ensure the precise information or accurate findings as obtained by the new interventional OI system, which was ultimately confirmed by pathology. The ex vivo OI confirmed the findings by in vivo interventional OI, detecting residual tumor as high optical signal compared with nearly absent signal in the completely ablated tumors. The transition zone appeared as a rim with moderate optical signal surrounding the completely ablated tumor (Fig. 7A and G).
The ex vivo OI results were further confirmed by corresponding histopathologic examinations (Fig. 7B, C–E, and H–K). The high optical signals of residual tumors on ex vivo OI were pathologically detected as (i) persistent (nonablated) tumors with H&E staining (Fig. 7C); (ii) viable or living tumor cells (as darkest blue) with NADH staining (Fig. 7D; ref. 30); and (iii) high fluorescent signals from ICG-outlined residual tumor cells with fluorescent microscopy (Fig. 7E). The transition zone reflected the transient physiologic inflammatory response to ablation heat (15, 16), also so-called benign peri-ablational enhancement (BPE) as recommended by the International Working Group on Image-Guided Tumor Ablation (31). The results of pathology examinations in our study confirmed these statements above, demonstrated as the transition zone with no living tumor cells by H&E staining (Fig. 7C and I) and proven to possess no viable tumor cells (light blue) by NADH staining (Fig. 7D and J; ref. 16).
Furthermore, the quantitative SI analysis demonstrated that (i) the mean SBR of the residual viable tumors was significantly higher than those of transition zones and ablated tumor zones in incomplete tumor models (residual tumors vs. transition zone vs. ablated tumors = 2.78 ± 0.27 vs. 1.41 ± 0.05 vs. 0.58 ± 0.05, P < 0.001; Fig. 7F); and (ii) the mean SBR of the ablated tumors was significantly lower than that of the transition zones in complete ablation tumors (0.57 ± 0.05 vs. 1.41 ± 0.06, P < 0.001; Fig. 7I). These results confirmed that the interventional OI could visualize of three distinct pathologic zones of ablated tumor periphery and thereby differentiate incomplete and completed tumor ablation.
Viable tumor cells, usually located at the periphery of solid malignancies after incomplete ablation, are largely responsible for tumor recurrence. One primary reason for incomplete tumor ablation is inability to detect residual tumor that would allow additional target ablation during the treatment session. Thus, it is essential to develop new intraprocedural real-time imaging tools that would allow immediate visualization of the various pathologic zones at the ablated tumor periphery, and thereby accurately differentiate residual tumors from ablated tumor periphery before terminating the ablation treatment.
OI has several important advantages over other imaging modalities such as CT or PET-CT (with no real-time imaging capability) and MRI (with high cost and environment requirements for MR safety). There is growing interest in development of targeted OI probes to strengthen the OI detection of cancers (32). However, most of these OI compounds are still in preclinical development phase and a long way from regulatory approval (33). To shorten the time to actual clinical practice, we have specifically selected the FDA-approved fluorescent agent, ICG as the OI contrast agent for development of the new interventional OI technique.
ICG is a relatively nontoxic protein-bound compound that is taken up by some cells (19, 34). The potential mechanisms for intracellular localization of ICG have not been fully understood yet. A few studies suggest that hepatic malignant tumors can take up ICG via the transporters of Na+/taurocholate coransporting polypeptide and organic anion transporting polypeptide 8 (23). Intravenously administered ICG is excreted from normal hepatocytes exclusively into the biliary system. However, due to impaired biliary excretion, ICG accumulates in liver tumor cells, leading to ICG enhancement throughout the tumor.
In this study, we attempted to fully exploit the advantages of ICG-based OI to guide minimally invasive interventional oncology treatment for thermal ablation of hepatic tumors as the first target. The innovative design of the new interventional real-time OI technique enabled us to visualize, in vivo, the three distinct pathologic zones of the ablated tumor periphery, and thereby instantly detect residual tumor tissues during an ablation session.
For in vitro confirmation with VX2 tumor cells, we first needed to optimize the in vitro experimental protocol for ICG dose, and then determine the optimum detection time-window for achieving the best ICG-based interventional OI 24 hours after ICG treatment of VX2 tumor cells. The results of our in vitro experiments showed well intracellular uptake of ICG by VX2 cells, with the significantly higher fluorescence SI of living (non-ablation-heated) VX2 tumor cells than dead (ablation-heated) tumor cells. In addition, in vitro evaluation further confirmed the sensitivity of the new interventional OI system in differentiating 1.5-mm3 living tumor cell clusters from 1.5-mm3 dead tumor cell clusters.
Subsequently, we validated the technical feasibility of the new interventional OI system through serial in vivo experiments using rabbit models with orthotopic liver VX2 tumors. The interventional OI system could differentiate residual tumor sites from the transition zone and fully ablated tumor zone during an ablation session. As a “gold standard,” ex vivo OI with pathologic correlations further confirmed these findings. Thus, these comprehensive in vitro, ex vivo, and in vivo experiments in this study have validated a new interventional OI technique.
In conclusion, we present a new interventional OI technique, which allows instant visualization of the three post-ablation pathological zones and thus accurate identification of residual viable tumor at the periphery of an ablation-treated lesion. This capability is expected to revolutionize current interventional oncologic techniques by changing the landscape of percutaneous ablation and decision-making during the intervention procedures, rather than after completion of the initial ablation treatment, which thus could significantly improve survival and quality of life in patients with large or irregular unresectable tumors in solid organs.
No disclosures were reported.
X. Kan: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing, performed the experiments. G. Zhou: Validation, methodology, performed the experiments. F. Zhang: Data curation, validation, methodology, writing–original draft, performed the experiments. H. Ji: Data curation. H. Zheng: Performed the experiments. J.F.B. Chick: Writing–review and editing. K. Valji: Writing–review and editing. C. Zheng: Writing–review and editing. X. Yang: Conceptualization, data curation, supervision, funding acquisition, validation, project administration, writing–review and editing.
This study was supported by the NIH R01EB028095 grant.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.