Intratumor injected gold molecular clusters for NIR-II imaging and cancer therapy

Significance

NIR-II/SWIR (1,000 to 3,000 nm) fluorescence imaging-guided surgical navigations in pre-clinical murine cancers have largely improved the tumor margin delineation and resection precision. Herein, we present a molecular Au₂₅ cluster with stealth phosphorylcholine coating (AuPC) as an intratumorally administered NIR-II/SWIR fluorescent probe, affording homogeneous tumor labeling and imaging-guided resection. A substantial fraction of i.t. injected stealth AuPC clusters entered the tumor interstitial fluid space, resulting in uniform distribution through the tumor. Probe or drug homogeneity inside tumors through local, intratumor injections has been highly challenging, and developing probes that impart negligible concentration gradient in tumor microenvironment is a breakthrough. Uniform tumor distributions of AuPC probes are also exploited for photothermal therapy for complete tumor eradication.

Abstract

Surgical resections of solid tumors guided by visual inspection of tumor margins have been performed for over a century to treat cancer. Near-infrared (NIR) fluorescence labeling/imaging of tumor in the NIR-I (800 to 900 nm) range with systemically administrated fluorophore/tumor-targeting antibody conjugates have been introduced to improve tumor margin delineation, tumor removal accuracy, and patient survival. Here, we show Au₂₅ molecular clusters functionalized with phosphorylcholine ligands (AuPC, ~2 nm in size) as a preclinical intratumorally injectable agent for NIR-II/SWIR (1,000 to 3,000 nm) fluorescence imaging-guided tumor resection. The AuPC clusters were found to be uniformly distributed in the 4T1 murine breast cancer tumor upon intratumor (i.t.) injection. The phosphocholine coating afforded highly stealth clusters, allowing a high percentage of AuPC to fill the tumor interstitial fluid space homogeneously. Intra-operative surgical navigation guided by imaging of the NIR-II fluorescence of AuPC allowed for complete and non-excessive tumor resection. The AuPC in tumors were also employed as a photothermal therapy (PTT) agent to uniformly heat up and eradicate tumors. Further, we performed in vivo NIR-IIb (1,500 to 1,700 nm) molecular imaging of the treated tumor using a quantum dot-Annexin V (QD-P³-Anx V) conjugate, revealing cancer cell apoptosis following PTT. The therapeutic functionalities of AuPC clusters combined with rapid renal excretion, high biocompatibility, and safety make them promising for clinical translation.

Surgical resection of solid malignant tumors guided by visualization of tumor-to-normal tissue margins has been an important cancer treatment approach of clinical oncology (1). Although invasive, complete surgical excisions are usually associated with better overall survival and lower recurrence rates. Currently, in practice, pre- and intraoperative assessment of tumors largely rely on palpation and visual inspection of tumor margin by surgeons. Although pre-operative imaging modalities are impeccable in surgical practice, intraoperative imaging modalities are still scarce in the clinic (2) for more precise delineation of tumor margin and detection of small residual and occult tumors.

In vivo fluorescence imaging (3) has emerged as an important imaging modality applicable to both preclinical (3) and human patients (4, 5) pre-, intra-, and post-operatively for tumor staging, resection, and monitoring of therapeutic responses. In preclinical (6) research and clinical (6, 7) trials, tumor targeting NIR-I (800 to 900 nm) fluorescent probes, i.e., dye-antibody (8), dye-peptide (9), and dye-small biomolecule (10) conjugates are intravenously (i.v.) administered for fluorescence imaging-guided surgical navigation and intervention (11, 12). Recently, much improved tumor-to-normal tissue and tumor-to-background signal ratios were achieved by deploying NIR-II (1,000 to 3,000 nm) fluorescent probes for imaging-guided surgery in preclinical settings (13–15). Compared to bioimaging in NIR-I, NIR-II fluorescence imaging benefits from reduced light scattering (16, 17) and autofluorescence, affording deeper imaging depths, higher spatial resolutions, and higher contrast (18) and improved delineation of tumor margin. Thus far, quantum dots (QDs) (14), down-conversion rear earth nanoparticles (13, 15, 19), and organic dyes (14, 20) conjugated to antibodies, affibodies, or other targeting ligands have been used for specific tumor targeting and NIR-II imaging-guided tumor resection at preclinical levels.

Photodynamic therapy (PDT) and photothermal therapy (PTT) are also promising cancer treatment approaches that rely on the generation of cytotoxic singlet oxygen/reactive oxygen species (ROS) and heat at the target site upon excitation of the probes with light. Many inorganic nanomaterials including carbon nanotubes (21), nano-graphene (22, 23), gold nanoparticles (24, 25) and clusters (26–30), and organic dyes such as ICG (31) have been used as PTT/PDT agents in combating cancers at various stages. Currently, the systemic delivery of PTT agents with or without targeting moieties relies on the extravasation of the agents from tumor blood vessels via enhanced permeability and retention (EPR) effect (32). However, abnormal and irregular vascular patterns in solid tumors cause erratic probe distributions and localized clustering near the vessel leakage (33, 34). This could lead to uneven heat distribution in PTT and hypoxic conditions within tumors in PDT, consequently compromising the treatment efficacy and survival rates.

Systemically administered cancer therapeutic drugs for chemotherapy (35) or immunotherapy (36) often are limited in tumor penetration and accumulation, hampering treatment efficacy. In addition, “off-target” toxicity and non-selective drug distributions in healthy organs can cause undesirable side effects (35, 36). Local (37), i.e., intratumoral (i.t.) (38, 39) delivery of low-dose anticancer therapeutics (40–42) and PDT/PTT agents (43, 44) has been sought to circumvent the systemic exposure (45) and overcome the inherent tumor microenvironment (TME) immunosuppression (46) by increasing the bioavailability of administered drugs inside the tumor tissue. However, vascular permeability, heterogeneity, high interstitial fluid pressure, and cellular density (33) can individually and/or collectively obstruct the uniformity and homogeneity in the spatial distribution of the drugs within solid tumors, resulting in uneven distribution and accumulation at/near the injection site (47). Thus, drug concentration gradient (48, 49) across the tumor tissue hampers the treatment effectiveness and results in low tumor ablation efficacy in the case of PTT/PDT.

In recent years, NIR-II emissive gold nanoclusters (50–55) have attracted much attention due to their ease of synthesis, size/property tuning in a controlled manner, and biocompatibility. Among all gold nanoclusters, water-soluble Au₂₅ is especially promising and has been applied for bone (56) and brain (57) imaging in vivo. Recently, we introduced “super-stealth” Au₂₅ clusters functionalization with biocompatible phosphorylcholine (PC) ligands (AuPC) for sentinel lymph node (SLN) imaging in the NIR-II window (58). The AuPC clusters showed little non-specific binding with serum proteins, cells, and tissue, fast renal excretion, and little retention in the injection sites, which differed from many other types of inorganic nanoparticles.

Here, we investigated AuPC clusters as intratumorally injected agents for NIR-II imaging-guided surgery of 4T1 murine breast cancer and as a PTT agent. Unlike ICG, we found that i.t. injected AuPC showed uniform distributions in tumors with a substantial fraction entering the tumor interstitial fluid, allowing clear delineation of tumor margins by NIR-II imaging for tumor resection and for effective PTT treatment and fully eliminating tumors. Further, we performed in vivo NIR-IIb molecular imaging of tumor cell apoptosis in vivo post PTT treatment.

Results

Synthesis and Characterization of Clusters.

We first synthesized glutathione (GSH)-coated Au₂₅ clusters as reported previously (30, 58) (denoted as AuGSH, Fig. 1 A, Left). The clusters were then conjugated to phosphorylcholine ligand (PC ligand) to form AuPC (Fig. 1 A, Right) (58) through EDC/NHS coupling in a MES pH7.0 buffer. The UV–Vis spectrum showed characteristic absorption features of AuGSH and AuPC clusters at ~670 and ~450 nm. The peak ~670 nm corresponds to the HOMO→LUMO (intraband sp→sp) transition inside the Au₁₃ core, whereas the shoulder ~450 nm correlates with mixed intraband (sp→sp)-interband (d→sp) transitions within gold core and staple motifs (59) (Fig. 1B). Additionally, an increase in absorption at shorter wavelengths is a typical trend for atomically precise gold nanoclusters related to the ligand shell contributions (60). A small peak appeared in the UV region ~250 nm (Fig. 1 B, Inset) for AuPC while absent in AuGSH (SI Appendix, Fig. S1), due to conjugation of the PC ligand. In addition, we previously showed that PC conjugation led to the appearance of stretching vibrational modes of PO₂⁻ group (ν_as 1,240 cm⁻¹ and ν_s 1,090 cm⁻¹) and choline headgroup (970 to 895 cm⁻¹) in the ATR-FTIR spectrum of AuPC (58).

Fig. 1.

(A) Crystallographic representation of Au₂₅ cluster structure using UCSF Chimera program (version 1.16) based on published crystal structure data (61). Color codes of the elements: Au: yellow, S: green. Schematic illustration of AuGSH cluster surface modification with PC ligand to result in AuPC conjugate. For simplicity, only the conjugation of γ-glutamate carboxylic functional group to PC ligand is shown while the conjugation to glycine carboxylic group in GSH is omitted. The structures of GSH and PC ligands are shown inside the frame. (B) UV−Vis absorption spectra of AuGSH and AuPC clusters in PBS buffer pH7.4. The spectra of cluster solutions (500 µg/mL) were measured using a 2-mm quartz cuvette. The inset shows the spectra in UV region and highlights the absorption of PC ligand. (C) PLE spectra of AuGSH cluster at various excitation wavelengths. (D) Visible-NIR-I and NIR-II fluorescence spectra of AuGSH cluster in PBS buffer pH7.4 (120 µg/mL) excited at 550 nm.

We observed fluorescence of Au clusters under a wide range of excitations (Fig. 1). Photoluminescence excitation/emission (PLE) spectra (Fig. 1C) in the NIR range recorded with an InGaAs detector (see details in Methods) showed strong emission in the 900 to 1,450 nm range in the excitation range of 550 to 840 nm (Fig. 1C). We also measured visible to NIR-I fluorescence of AuGSH clusters using a fluorimeter equipped with a silicon-based PMT detector (nominal detection range: 250 to 1,050 nm, see details in Methods). The clusters showed fluorescence under various excitations in the 600 to 900 nm range with an apparent peak at ~750 nm (Fig. 1D and SI Appendix, Fig. S2, excitation 550 nm) related to the silicon PMT insensitivity to >900 nm light. As the InGaAs detector was insensitive to visible light, the combined results from both detectors suggested that the Au clusters fluoresce broadly in the 600 to 1,450 nm range when excited at a shorter wavelength (e.g., at 550 nm). NIR-II emission of the Au clusters was observed under any excitation in the NIR-I range (e.g., 808 nm), useful for NIR-II imaging.

UV–Vis, Visible-NIR-I, and NIR-II PL spectra of AuPC clusters (120 µg/mL, PBS buffer pH7.4) were compared to that of AuGSH (SI Appendix, Fig. S3). First, the UV–Vis spectra of both clusters were identical indicative of no concentration change after PC conjugation (SI Appendix, Fig. S3A). In fluorescence spectra, the spectral shape and peak positions remain unchanged while the PC conjugation resulted in ~ four- and ~ one-fold intensity enhancement in Visible-NIR-I (ex@450 nm, SI Appendix, Fig. S3B) and NIR-II (ex@808 nm, SI Appendix, Fig. S3C) regions, respectively, presumably due to rigidification of the ligand shell (see details in Discussion). Also, notably, visible light excitation of the Au clusters (~550 to 650 nm, Fig. 1C) and emission > 600 nm overlapped with excitation/emission of the commonly used Cy5 dye (ex: ~650 nm, em: ~670 nm), allowing fluorescence imaging of Au clusters in the Cy5 channel using widely available microscopy instruments (SI Appendix, Fig. S4).

NIR-II Imaging Guided Tumor Resection with Intratumor Injected Gold Clusters.

Murine 4T1 breast cancer tumors (21 to 130 mm³) on Balb/c female mice (3 to 6 wk old, ~10 to 20 g) were intratumorally injected with AuPC clusters (1×, 300 µg per injection in 20 to 30 µL PBS buffer, see details in Methods) and imaged in the NIR-II window at 3 min and 10 min post injections (p.i.) using an InGaAs camera in wide field configuration (Fig. 2A, excitation 808 nm, emission > 1,100 nm, exposure time 3 ms). The resulting NIR-II images showed uniform distribution of the AuPC probe inside the tumor bed with clearly visualized and distinguishable tumor margins (n = 8, Fig. 2B, pre-surgery panel). After exposing the tumor surface by removing the covering skin, the tumors were surgically removed under NIR-II imaging-guided tumor resection (Fig. 2B, surgery panel). High-magnification images of the resected tumors showed uniform distribution of AuPC clusters across the tumor (from superficial surface to tumor base) with no remaining residue or healthy tissue staining in the resection site (Fig. 2B, post-surgery panel). High tumor-to-surrounding background tissue (TBR) ~50 was calculated for AuPC clusters (Fig. 2 C, Left) with sharp cross-sectional profiles (Fig. 2 C, Right, MFI: mean fluorescence intensity).

Fig. 2.

(A) Schematic representation of in vivo NIR-II imaging-guided tumor resection of intratumorally administered AuPC clusters in PBS buffer pH7.4 (1×, 300 µg) into a mouse bearing 4T1 tumor on a right hindlimb (3 to 6 wk old female Balb/c, n = 8). (B) Pre-surgical, intra-operative and post-surgical NIR-II imaging were perform using wide-field NIR-II fluorescence imaging system (excited by an 808-nm laser at a power density of 70 mW/cm², exposure times 3 and 7 ms for 1× and 5× magnification, respectively, using an 1,100-nm long-pass filter). The tumor area marked by a rectangle was imaged using 5× magnification (exposure time 7 ms). For surgical resection, the skin covering the tumor was carefully removed allowing NIR-II imaging of the exposed tumor. The tumor was then excised in one step. Scale bars are 1 cm and 2 mm for 1× and 5× objectives, respectively. (C) TBR ratio (Left graph) was calculated using the fluorescence intensity of tumor and nearby background after resection. The Right graph is the cross-sectional profile marked in white dotted lines represented in mean fluorescent intensity (MFI) values. Data are presented as box plots (median line, 25th and 75th percentiles; whiskers, outlier). (D) H&E staining and NIR-I fluorescence imaging of resected tumor tissue in Cy5 channel (AuPC emission). 20× objective, scale bar is 0.5 mm. The tissue sections were ~20 µm.

Ex vivo, tumor slices after H&E staining were used for concurrent fluorescence imaging of AuPC in the Cy5 channel (excitation ~650 nm, emission: 670 to 730 nm) and visible-range optical imaging of H&E-stained tumor cross-section (Fig. 2D) using a commercial instrument (see details in Methods) for overlaying AuPC signals with tumorous H&E stains. We observed a high degree of overlap of the i.t. injected AuPC fluorescence and cancerous regions stained by H&E (n = 4) (Fig. 3D), confirming uniform distribution of i.t. injected AuPC clusters in the tumor and not in the surrounding normal tissue.

Fig. 3.

(A) Schematic representation of TIF collection by centrifugation from resected tumors i.t. injected with AuGSH (1×, 300 µg in PBS buffer pH 7.4), AuPC (1×, 300 µg in PBS buffer pH 7.4), and ICG (5 mM in DI-water). (B) Ex vivo NIR-II imaging of resected tumors before and after centrifugation (n = 3). The imaging was done under an 808-nm laser excitation at a power density of 70 mW/cm² using an 1,100-nm long-pass filter, 5× objective, scale bar is 2 mm. The exposure times were 5 and 3 ms for AuPC/AuGSH and ICG, respectively. (C) Color photographs (Top) of collected TIF extracts inside Eppendorf vials. NIR-II images of collected extracts (Bottom, 10× diluted solutions, 3 ms-exposure time). (D) Tumor MFI difference before and after the centrifugation. (E) The calculated percentages of NIR-II brightness in collected extracts (10× diluted solutions, 3 ms-exposure time). Error bars represent SD of three repeated experiments.

For comparison, we performed i.t. injection of the parent AuGSH clusters and also observed uniform distribution within the tumor bed in vivo and ex vivo (SI Appendix, Fig. S5 A and B, n = 7) with high TBR ratio of ~50 (SI Appendix, Fig. S5C), but the AuGSH cluster fluorescence signals appeared slightly less homogeneous than AuPC over the H&E-stained tumor tissue regions (SI Appendix, Fig. S5D, n = 2). In another control experiment, we performed i.t. injection of an ICG dye solution (20 to 30 µL, 5 mM) and observed drastically different tumor distributions (n = 6, SI Appendix, Fig. S6 A and B) from those of AuPC and AuGSH. During injection, ICG did not spread uniformly and stayed localized near the injection site (SI Appendix, Fig. S6B) (62, 63). Due to strong non-specific binding of ICG (64), ICG stained the healthy tissue next to the tumor when it reached the tumor boundary. Upon tumor resection, the resection site showed significant staining of normal tissues under the tumor bed, resulting in much lower TBR (~4) and disturbed cross-sectional profiles (SI Appendix, Fig. S6C). Ex vivo, inhomogeneous localized ICG fluorescence was detected in H&E-stained slices (SI Appendix, Fig. S6D), consistent with in vivo NIR-II imaging and ex vivo whole-tumor imaging data.

Probing Intratumorally Injected Au Clusters and ICG in the Interstitial Fluid of Tumor.

Tumor interstitial fluid (TIF) is a major component of tumor interstitium that regulates the nutrient transport within cells and drains the metabolic waste through the lymphatic system into the bloodstream (65–67). Poor drainage of buildup fluid inside the tumor interstitium results in high-pressure gradients and consequent flow into the healthy tissue stroma and lymphatic spread (67). Thus, TIF transport was associated with tumor cell invasion and metastasis. Besides, the variable interstitial fluid pressure (IFP) as an important part of the tumor microenvironment complexity (65), imposes significant pathophysiological challenges for transvascular drug extravasation and interstitium accumulation (66). In part, non-uniform dissemination of therapeutic drugs within tumor tissue was associated with high IFP (68). Consequently, intentional, and targeted IFP decline could benefit therapeutic efficacy by increasing the local drug concentrations inside the tumor (67).

We investigated how the distributions of i.t. injected Au nanoclusters and ICG within the tumor microenvironment (TME) were associated with the tumor interstitium. To probe the AuPC, AuGSH or ICG in the TIF, we performed i.t. injections of these agents, respectively (n = 3 for each agent), followed by in vivo tumor imaging, tumor resection (SI Appendix, Fig. S7) 5 min post injection, ex vivo tumor imaging and then high-speed centrifugation of the tumors to extract and collect the interstitial fluids (see details in SI Appendix) (69) (Fig. 3A). In vivo NIR-II imaging showed that while AuPC (Movie S1) and AuGSH (Movie S2) rapidly spread out inside the tumor within seconds after injection, ICG exhibited obviously slower diffusion (Movie S3). Ex-vivo NIR-II imaging of tumors before and after centrifugation (Fig. 3B) showed that the NIR-II fluorescence of AuPC and AuGSH in the tumors reduced by ~56% and ~19%, respectively, after centrifugation (Fig. 3D), suggesting extraction of Au nanocluster from the tumor together with TIF. In strong contrast, the ICG-injected tumors exhibited even higher fluorescence brightness after centrifugation, attributed to increased association of ICG with the cancer cells or proteins in the tumor upon tumor centrifugation (Fig. 3 B and D). This was based on the well-known effect of ICG fluorescence brightening upon binding to serum proteins (70). We characterized the TIF extracted from tumors by centrifugation (Fig. 3C) and observed that relative to the fluorescence brightness of injected solution, ~33%, ~30%, and ~6% of AuPC, AuGSH, and ICG probes, respectively, were found inside the tumor interstitial fluid (Fig. 3E). These results suggested that a substantial fraction of i.t. injected AuPC clusters entered the tumor TIF and uniformly distributed through the tumor, consistent with ex vivo tissue slice NIR-II imaging and H&E staining result (Fig. 2D). For ICG, the majority of the dye was trapped inside the tumor locally near the injection site without filling the tumor TIF uniformly.

NIR-II Photothermal Therapy with AuPC Clusters.

We exploited the uniform tumor distributions of AuPC upon intratumor injections for photothermal therapy (PTT) that could benefit from homogeneous photothermal heating effects across the tumor. We first assessed PTT effect by irradiating a 25 mg/mL aqueous solution of AuPC (Fig. 4) continuously with an 808 nm laser (300 mW/cm²) for 30 min. Thermal imaging revealed photothermal heating of AuPC (Fig. 4 A, Inset), reaching ~70 °C from room temperature within 10 min post-irradiation (Fig. 4A), generating sufficient heat for PTT. Colorimetric detection of singlet oxygen generation from cluster solutions (0.5 mg/mL in water) using 1,3-diphenylisobenzofuran (DPBF) for PDT treatment (see details in SI Appendix) led to negligible reactive oxygen species (ROS) formation under the same irradiation conditions (SI Appendix, Fig. S8). Note that while the laser irradiation induced heating of the samples to higher temperatures, no significant changes were observed in the NIR-II photoluminescence intensity or spectrum features of the AuPC solution (Fig. 4B).

Fig. 4.

(A) The temperature change of AuPC cluster solutions (60 µL, 25 mg/mL in PBS buffer pH7.4) during continuous 30 min irradiation with an 808-nm laser at a power density of 300 mW/cm² (n = 3). The inset shows thermal pictures of AuPC solution inside Eppendorf vial before (0 min) and 10 min after laser irradiation recorded with a thermal camera. (B) NIR-II PL spectra of AuPC cluster solution (50× dilution) before and after irradiation using an 808-nm laser at a power density of 35 mW/cm². The Inset shows bar graphs of NIR-II peak intensity change post-irradiation and are presented as mean values ± SD. The data were analyzed by Tukey’s test.

In vivo, we performed i.t. injection of 20 to 40 μL of a AuPC cluster solution (25 mg/mL) into 4T1 tumors on mice (n = 18) for PTT (Fig. 5A) after performing NIR-II imaging (Fig. 5B) at ~3 min post-intratumor injection. PTT was done by irradiating the tumor with an 808-nm laser (300 mW/cm²) for 30 min. The tumor temperature increase was continuously recorded using a thermal camera (Fig. 5C). Within 3 min irradiation, the average temperature on the tumor reached ~53 °C and continuously increased up to 60 °C during PTT (Fig. 5E). The tumor slowly became swollen after PTT and turned dark brown/black after 24 h post treatment (Fig. 5D). One day post irradiation, the irradiated tumor visibly flattened and significantly increased in apparent size, forming a black scab on the tumor surface (Fig. 5D). Starting on day 3 (D3), post-PTT the treated tumor started to shrink and eventually disappeared by D25 (Fig. 5F) for 17 out of 18 mice. The NIR-II fluorescence signal of the clusters inside the tumor was imaged before and after PTT treatment as well as on later time points (Fig. 5B), revealing that the clusters remained inside the tumor for up to 2 to 3 wk and then faded as the tumor size shrank, suggesting gradual excretion of the i.t. injected AuPC clusters from destructed tumors after PTT. Post-PTT treatment, tumor regrowth was recorded only in one mouse in the AuPC-mediated PTT-treated group (n = 18), giving a 94.4% 60-d relative survival rate (Fig. 5H).

Fig. 5.

(A) Schematic representation of PTT treatment of intratumorally administered AuPC cluster into a mouse bearing 4T1 tumor on a right hindlimb (6 wk old female Balb/c, n = 18). Saline-injected mice irradiated with laser were used as a control group (n = 16). PTT treatment was done by an 808-nm laser irradiation at a power density of 300 mW/cm² for 30 min. (B) Wide-field NIR-II imaging before and after PTT and on later timepoints was done under an 808-nm laser excitation at a power density of 70 mW/cm² using an 1,100-nm long-pass filter. The exposure time was 40 ms. The Insets show low exposure (1 ms) NIR-II images. Arrows point to the tumor and inguinal lymph node (iLN). 1× objective, scale bar is 1 cm. (C) Thermal camera pictures of mice before and during PTT treatment, scale bar is 1 cm. (D) Color pictures of mice before and after PTT. Morphological changes on the tumor surface become evident 1-d post-treatment with obvious scab formation. Temperature increase (E) and tumor volume changes in AuPC (F) and control (G) groups during and post-PTT treatment, respectively. (H) Kaplan–Meier survival analyses of AuPC and control groups.

In the control mouse group (n = 16) injected intratumorally with only saline, the average temperature on the tumor reached 42 °C under the same laser irradiation condition (Fig. 5 C and E) without any obvious morphological changes on the tumor site post-30 min laser irradiation. The tumor size continuously increased and reached the maximum allowed size and endpoint within ~ 2 wk (Fig. 5 D and G).

Complete blood count (CBC) and hematological analyses of blood samples collected from the surviving mice (n = 5) undergone AuPC-mediated PTT showed no apparent toxic effects (SI Appendix, Figs. S9 and S10). The blood chemistry results were comparable to those of the control group. Pathological examination of histological sections of hematoxylin and eosin (H&E)-stained organs also showed no damage at the tissue level (SI Appendix, Fig. S11).

In Vivo NIR-II Molecular Imaging of Apoptosis in Tumor Undergone Photothermal Therapy.

To investigate cell death mechanism in the tumor post PTT, 30 min post AuPC-mediated PTT treatment, we performed intravenous injection of water solubilized > 1,500 nm emitting PbS/CdS core–shell quantum dots (QD) coated by three layers of cross-linked polymer (denoted as QD-P³, see details in SI Appendix) (71) conjugated to apoptosis marker Annexin V (denoted as QD-P³-Anx V) (n = 3). As control, QD-P³ without Annexin V was injected for another AuPC-mediated PTT treatment group (n = 3). Two more control groups were saline-treated mice with (+laser) and without (−laser) laser irradiation, followed by i.v. injection of QD-P³-Anx V (n = 3).

At 1.5 h post-i.v. administration (2 h post-PTT), NIR-IIb imaging revealed strong QD-P³-Anx V signal around the edge of the tumor in the AuPC-mediated PTT group (Fig. 6A, labeled “AuPC+QD-P³-Anx V”), while QD-P³ administrated AuPC-mediated PTT group showed very weak signal (Fig. 6A, “AuPC+QD-P³”). Saline-treated groups (+laser and −laser) showed signal due to QD-P³-Anx V probes still in circulation (Fig. 6A). At 24 h post-i.v. injection for the AuPC-mediated PTT group (24.5 h post-PTT), a strong > 1,500 nm NIR-IIb fluorescence was observed at the outer rim of the tumor while the middle of the tumor appeared dark, suggesting the detection of apoptotic cells in the tumor outer region and destruction of blood vessels by PTT inside the tumor restricting the particles from extravasation into the tumor core region. In contrast, the QD-P³-injected AuPC-mediated PTT group did not show any NIR-IIb fluorescence in the tumor vicinity, suggesting no targeting of apoptotic cells around the tumor by QD-P³ without Annexin V conjugation. For tumors treated by saline injection then PTT (+laser, n = 3) and without PTT (−laser, n = 3), respectively, the i.v. injected QD-P³-Anx V also showed no targeting or accumulation in the tumors, suggesting little apoptosis, consistent with rapid, continued tumor growth in these two groups. The measured tumor-to-normal tissue (T/NT) ratio of QD-P³-Anx V signals (Fig. 6B) was higher (~20) in the AuPC-mediated PTT group than those in all of the control groups with T/NT ~ 5 (Fig. 6B).

Fig. 6.

(A) In vivo apoptosis imaging post PTT treatment of intratumorally administered AuPC cluster into a mouse bearing 4T1 tumor on a right hindlimb (6 wk old female Balb/c, n = 3). QD-P³-Anx V and QD-P³ were intravenously administered to the AuPC-mediated PTT group (n = 3) 1.5 h after the treatment. Systemic administration of QD-P³-Anx V to saline-injected mice with (+laser) or without (−laser) laser irradiation was used as control groups (n = 3 each). PTT treatment was done by an 808-nm laser irradiation at a power density of 300 mW/cm² for 30 min. Wide-field NIR-II imaging 1.5 and 24 h post-i.v. administration of probes was done under an 808-nm laser excitation at a power density of 70 mW/cm² using a 1,500-nm long-pass filter. The exposure times were 40 ms and 100 ms for 1.5 h and 24 h post-i.v., respectively. 1× objective, scale bar is 1 cm. (B) The calculated T/NT values for probes at 24 h post-i.v. injection. Data are presented as box plots (median line, 25th and 75th percentiles, whiskers, outlier). Absolute P values were derived using One-Way ANOVA. (C) Flow cytometry analyses of Annexin V-FITC/PI double-stained tumor cells 24 h post-PTT. (D) Apoptosis cell death in AuPC and control groups. (E) Ex vivo confocal imaging of apoptosis 24 h post-PTT treatment of intratumorally administered AuPC cluster and saline. The cryosections of tumor tissues (~30-µm thickness) were stained with TUNEL assay kit and DAPI and imaged in DAPI (ex: 405 nm, em: 430 to 470 nm), FITC (TUNEL, ex: 488 nm, em: 500 to 540 nm), and Cy5 (AuPC cluster, ex: 640 nm, em: 650 to 750 nm) channels. 20× objective, scale bar is 50 μm.

Ex vivo flow cytometry analyses were performed to investigate cell death mechanism in the saline (n = 2) and AuPC-mediated PTT (n = 3) groups using Annexin V-FITC/PI double fluorescence staining of cells extracted from tumors 24 h post-PTT (Fig. 6C and SI Appendix, Fig. S12). Strong Annexin V-FITC and PI signals were detected in the AuPC group corresponding to cells undergoing apoptotic (both early and late stages, Annexin V-FITC staining, Fig. 6 C, Bottom and Top-Right panels) and necrotic deaths (PI staining, Fig. 6 C, Top-Left panel) (72). Since necrosis is common in cancerous tumors even without any treatment, only early and late apoptosis populations in saline and AuPC-mediated groups were compared. Then, ~1.5% and ~6.9% of cells accounted for early (Q3 values, Fig. 6 C, Bottom-Right panel) and late (Q2 values, Fig. 6 C, Top-Right panel) apoptosis in AuPC-mediated PTT group, respectively. The corresponding cell populations in the saline-mediated PTT control group were ~eight-fold (~0.19% vs. ~1.5%) and 16-fold (~0.43% vs. ~6.9%) lower. On average, the overall apoptotic cell death in saline- and AuPC-mediated PTT groups was ~0.6% and ~8%, respectively (Fig. 6D).

Finally, we performed ex vivo apoptosis imaging of tissue sections from saline- and AuPC-mediated PTT tumors using TUNEL assay kit (Fig. 6E, see details in SI Appendix). The nuclei were stained with DAPI while the TUNEL channel showed FITC-labeled damaged DNA fragments resulting from PTT treatment. The confocal fluorescence imaging of tumor tissue sections in DAPI (ex: 405 nm, em: 430 to 470 nm), FITC (TUNEL, ex: 488 nm, em: 500 to 540 nm) and Cy5 (AuPC cluster, ex: 640 nm, em: 650 to 750 nm) channels reveal striking contrast in tumor tissues from saline and AuPC-mediated PTT treated tumors (Fig. 6E). Much higher degree TUNEL staining and DNA damage due to apoptosis was observed in AuPC+PTT-treated tumor sections, while the control group showed very low apoptotic signals in the tissue. These results corroborated with in vivo NIR-II molecular imaging.

Discussion

We previously showed the applicability of stealth AuPC as sentinel lymph node (SLN) tracers for preclinical cancers (58). The AuPC clusters show negligible affinity toward serum proteins and are rapidly cleared from the injection site and excreted from the body through urine. Here, we investigated the optical properties of the AuPC nanoclusters and developed NIR-II gold nanocluster-based fluorescent probes for imaging-guided surgical resections of 4T1 murine breast cancer. The cluster exhibited fluorescence in the 600 to 1,400 nm range under visible to NIR-I excitation, useful for both visible fluorescence imaging in the Cy5 channel under ~650 nm excitation and NIR-II imaging under 808 nm excitation.

The fluorescence or photoluminescence (PL) mechanisms in molecular gold clusters have been investigated over the years but remain under debate in explaining the visible-to-NIR PL in molecular gold clusters (73). The NIR emission of Au₂₅ clusters was suggested to originate from its Au₁₃ core (74), and in a “core-regulated PL” strategy, the heteroatom doping could largely improve the luminescent properties of clusters (75, 76). In a “surface-regulated PL” strategy, the PL quantum yield could also be drastically improved by cluster core passivation, i.e., with electron-donating ligands (77) or rigidification (55), i.e., ligand shell engineering (78) or staple-motif modification (79).

In clinical practice, most chemotherapeutic agents with or without targeting moieties are delivered to the tumor site through systemic administration that relies on the drug extravasation from the blood vessel and transport within the tumor compartment via enhanced permeability and retention (EPR) effect (32). However, inefficient delivery of the drugs to the tumors and therefore low abundance within the tumor tissue results in severely reduced treatment efficacies and survival rates. Alternatively, local or intratumor administration of probes is sought to minimize the off-target exposure and attain higher dosage of probes. Although promising, inhomogeneous probe distributions within tumor tissue were typical obstacles for locally administered inorganic nanomaterials (47, 80, 81), ICG-loaded particles (81, 82), and clinical probes (48, 49, 83).

We found that a unique feature of the AuPC stealth probes was its uniform distribution inside a tumor within minutes of i.t. injection, which could be exploited for imaging-guided tumor resection or efficient photothermal therapy. Phosphorylcholine and its derivatives are indigenous to living systems and do not cause inflammatory responses (58). Moreover, the PC ligand is zwitterionic and non-fouling due to being strongly hydrated by water molecules via electro-static interactions and high resistance to non-specific interactions with proteins and other biological species (84). Thus, the stealth nature of clusters imparts uniform distributions in the tumor. The uniformity was associated with the observation that a significant fraction ~33% of AuPC entered the tumor interstitium, filling up a space crucial to tumor existence and survival. The i.t. injection route is advantageous over systemic administration in much lower dose needed, shorter wait time, higher accumulation in tumor, and no systemic off-target side effects. To our knowledge, AuPC is the only fluorescent i.t. administrated agent that enters a solid tumor homogeneously and exhibits little non-specific binding to healthy tissue adjacent to the tumor. This is useful to delineate tumor margin for high-precision imaging–guided tumor resection. In contrast, ICG is more localized near the tumor injections site and binds to nearby healthy tissue once reaching the tumor boundary.

The i.t. injected AuPC nanoclusters can also be exploited for photothermal therapy, benefiting from its uniform distribution in the tumor, efficient, homogeneous, and selective heating of tumor over healthy tissue. The AuPC PTT agent differs from previous gold-based nanomaterials such as gold nanoshells (25), nanorods (24), nanoparticles (85), and nanoclusters (28, 30). It is the only i.t. injected PTT agent, advantageous in dose, cost, PTT efficacy, and safety.

Lastly, to investigate the cell death mechanism in vivo induced by AuPC-mediated PTT, we deployed a quantum dot-based QD-P³ nanoparticles conjugated to Annexin V, QD-P³-Anx V for molecular imaging of tumor. The whole-body in vivo 1,500 to 1,700 nm NIR-II imaging revealed a dead tumor core without vascular access in ~1 d following PTT, and cells in the outer region of the tumor undergoing apoptosis, eventually leading to tumor eradication and mouse survival. Cellular apoptosis was confirmed by ex vivo flow cytometry and TUNEL assay, suggesting in vivo NIR-II molecular imaging of Annexin V targets for non-invasive apoptosis assessment that could be performed longitudinally.

Conclusions

We showed that gold molecular cluster with stealth phosphorylcholine coating (AuPC cluster) is a unique class of intra-tumor injection agent useful for NIR-II imaging-guided tumor resection in preclinical 4T1 cancer model. The AuPC probe uniformly distributed within tumor bed < 3 min post-intratumoral administration. In addition, high TBR of ~50 together with homogeneous NIR-II signal across the tumor afforded clear visualization of margins for complete surgical resection of tumor. A substantial fraction of i.t. injected AuPC clusters entered the tumor TIF and uniformly distributed through the tumor, consistent with ex vivo tissue slice NIR-II imaging and H&E staining result. Furthermore, we exploited the probe for photothermal therapy. Twenty-four hours post AuPC-mediated PTT, the tumors visibly turned black and gradually decreased in size until complete eradication. In vivo apoptosis imaging with QD-P³-Anx V probe revealed the accumulation of the targeted probe near the tumor edge 1.5 h post-i.v. injection and strong > 1,500 m NIR-IIb signal appearing in the outer rim of the tumor post-24 h i.v. The uniform distributions of AuPC probe inside the tumor and in its interstitium as well as its implementation in NIR-II PTT make the clusters promising for clinical translation.

Methods

Materials.

Hydrogen tetrachloroaurate(III) trihydrate (Sigma-Aldrich, ≥99.9% trace metals basis), L-glutathione reduced (GSH, Sigma-Aldrich, ≥98.0%), sodium borohydride (Sigma-Aldrich, ≥96%), 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS, Thermo Scientific), and 2-amino-2-(hydroxymethyl)-1,3-propanediol (tris-base) were used as received. DI water and indocyanine green (ICG) were purchased from Fisher Scientific. 4-Aminophenylphosphorylcholine (PC) was purchased from Santa Cruz Biotechnology Inc. Annexin V (His Tag) protein was purchased from AcroBiosystems. TUNEL assay kit-FITC was purchased from Abcam. DAPI was obtained from Abcam.

Synthesis of AuGSH Clusters and Conjugation to PC Ligand.

AuGSH clusters were synthesized according to the previously published procedures (30, 58). Briefly, 5 mg of HAuCl₄·3H₂O (0.013 mmol, 1.3 mM) dissolved in DI water was mixed with aqueous solution of 16 mg of L-glutathione reduced (0.052 mmol, 5.2 mM). The slightly milky solution was vigorously stirred for a few minutes and then reduced with 5 mg freshly prepared sodium borohydride solution (0.13 mmol, 13 mM) in water. The color of the solution immediately turned dark brown, indicating the formation of a mixture of different clusters. The reaction mixture was allowed to etch for 24 h at room temperature resulting in the formation of the final product, i.e., Au₂₅(GS)₁₈. Unreacted compounds and by-products were removed, and the final product was washed by centrifugation (4,400 rpm) using 15 mL Amicon 3 K filters. The purified and concentrated solution was stored in 4 °C for further use (denoted as AuGSH).

PC ligands were covalently conjugated to the AuGSH clusters using EDC/NHS chemistry according to the previously published method (58). The conjugation of Au-GSH cluster (1×, 300 µg) to PC ligand (~300 µg, in PBS pH7.4) was performed in MES pH7.0 buffer in the presence of 100 mM EDC and NHS. After orbital shaking of the reaction solution at room temperature for 3 h, the reaction was quenched for 1 h with 100 mM TRIS. The final Au-GSH-PC conjugate (denoted as AuPC) was washed with PBS pH7.4 buffer using Amicon 3KDa centrifuge filters and stored in 4 °C fridge for further use.

Characterization.

UV–Vis spectra were recorded on a Varian Cary 6000i UV/Vis/NIR spectrophotometer, using a quartz cuvette of 2-mm path length in PBS buffer.

Photoluminescence versus excitation spectra (PLE) were measured on a home-built NIR spectroscopy setup. Then, 150 W ozone-free Xenon lamp installed into Oriel 66907 Arc Lamp Source was used as an excitation source and excitation lines with a bandwidth of 15 nm were generated by a monochromator (Oriel 7400 Cornerstone 130 monochromator) to supply the excitation light in 550 to 840 nm range in 10-nm steps. Emission was collected in a 90-degree reflection geometry in 900 to 1,450 nm range. The emitted light was directed into a spectrometer (Acton SP2300i) equipped with a liquid-nitrogen-cooled InGaAs linear array detector (Princeton OMA-V). Spectra were corrected for the detector sensitivity, filter extinction features, and excitation power using the MATLAB software.

The NIR-II emission spectra were measured by an Acton SP2300i spectrometer equipped with a liquid-nitrogen-cooled InGaAs linear array detector (Princeton OMA-V, detection range 800 to 2,200 nm).

Vis-NIR-I fluorescence spectra were measured with a Horiba FluoroLog-3 Fluorimeter using thermoelectrically cooled R2658P Near IR PMT detector (detection range 250 to 1,050 nm).

Thermal images and temperature change were captured/recorded using Hti-Xintai HT-H8 high-resolution thermal camera imager.

H&E-stained tissue slices were imaged using Keyence BZ-X810 All-in-One fluorescence microscope. Visible fluorescence of clusters inside the tumor tissue was imaged in Cy5 channel (excitation ~650 nm, emission: 670 to 730 nm) and overlapped with H&E image.

Mouse Handling and Tumor Inoculation.

All animal experiments in the study were approved by Stanford Institutional Animal Care and Use Committee (IACUC). Then, 3 to 6-wk-old BALB/c female mice (weight: 10 to 20 g) were purchased from Charles River. The animals were housed in Stanford University’s Veterinary Service Center and necessary bedding, nesting material, food, and water were provided by the Stanford VSC facility. The animals were shaved using hair-removing lotion (Nair, Softening Baby Oil). For in vivo imaging and PTT, the mice were anesthetized by 2.5% isoflurane under 2 L/min oxygen flow. 4T1 murine breast cancer cells were inoculated on the right hindlimb of the mice and syngeneic 4T1 tumors were grown in a week. Animal experiments were performed when the tumor reached > 20 mm³. The maximum allowable tumor size for a mouse bearing a single tumor was 2.46 cm³ according to the guidelines of The Administrative Panel on Laboratory Animal Care (APLAC) of Stanford University. The tumor volumes were 21 to 130 and 87 to 600 mm³ for surgical resections and PTT, respectively. Then, 20 to 30 µL (1×, 300 µg) and 20 to 45 µL (25 mg/mL) AuPC solutions in PBS buffer pH7.4 were intratumorally administered for surgery and PTT, respectively. Intratumoral injections were performed very slowly using 31-gauge insulin syringe with permanently attached needle (Fisher, cat# B328449). Prior to surgical resections, the animals were humanely euthanized using isoflurane induction (3 to 5% isoflurane and oxygen as a carrying gas at a flow rate of 2 L/min) followed by cervical dislocation. During PTT treatment, a heating pad was also provided to keep the animals warm.

In Vivo Wide-Field Fluorescence Imaging.

In vivo wide-field fluorescence imaging in the NIR-II window was conducted in a two-dimensional, water-cooled 640 × 512 InGaAs array (Ninox 640, Raptor Photonics). The probes were excited by an 808-nm continuous-wave diode laser at a power density of 70 mW/cm². Then, an 1,100-nm long-pass filter was used in all the imaging experiments, while a 1,500-nm long-pass filter was used for QD-P³/Anx V imaging. AuPC, AuGSH, and ICG probes were administered intratumorally (i.t.) while QD-P³/Anx V were administered intravenously (i.v.). NIR-II fluorescence images were recorded at 3, 10, and 30 min after PTT and in the following days.

Data Processing.

LabView2009 software package was used for imaging the animals and adjusting laser exposure. The raw images were processed and analyzed using ImageJ 2.1. The structures of clusters were prepared using UCSF Chimera program (version 1.16) based on crystal structure data published in ref. 61.

Statistics and Reproducibility.

The graphs were prepared using Origin 2021 software package. Statistical analyses were performed using Paired comparison app available in Origin (mean comparison method: Tukey, one-sided) and One-way Anova. P values of <0.05 were considered statistically significant. Error bars represented SD of 3 to 18 repeated experiments. Data are presented as mean values ± SD. The mice were randomly selected from the cages and then divided into study groups.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

In situ intra-tumoral (i.t.) administration of AuPC probe (1x, 300 μg in PBS buffer pH 7.4) into a mouse bearing a 4T1 tumor on the hindlimb. The probe was excited by an 808 nm laser at a power density of 70 mW/cm², exposure time 40 ms and 1100 nm long pass filter. Total video time: ~ 41 s, frame rate: 20 fps.

Movie S2.

In situ intra-tumoral (i.t.) administration of AuGSH probe (1x, 300 μg in PBS buffer pH 7.4) into a mouse bearing a 4T1 tumor on the hindlimb. The probe was excited by an 808 nm laser at a power density of 70 mW/cm², exposure time 40 ms and 1100 nm long pass filter. Total video time: ~ 28 s, frame rate: 20 fps.

Movie S3.

In situ intra-tumoral (i.t.) administration of ICG probe (5 mM in DI-water) into a mouse bearing a 4T1 tumor on the hindlimb. The probe was excited by an 808 nm laser at a power density of 70 mW/cm², exposure time 4 ms and 1100 nm long pass filter. Total video time: ~ 45 s, frame rate: 20 fps

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.