Multiphoton Luminescence Imaging: A Label-Free Tool for Visualizing the Long-Term Fate of Gold Nanoparticles in Tissue

Abstract

Gold nanoparticles (AuNPs) are increasingly used in applications across the biomedical domain, yet their long-term biodistribution and biocompatibility remain poorly understood. Conventional brightfield microscopy imaging techniques often fail to detect AuNPs due to optical diffraction limits and lack of chromogenic contrast. Understanding the biodistribution and ultimate fate of these nonbiodegradable NPs is crucial for further development of AuNP-based therapeutics and diagnostics. Here, we present a label-free multiphoton luminescence (MPL) imaging workflow that enables sensitive detection of AuNPs in liver histology sections, even 1 year after intravenous (IV) administration. MPL imaging exploits the intrinsic nonlinear optical properties of AuNPs to generate broadband emission under ultrafast pulsed laser excitation, enabling subcellular localization without exogenous labels while having the ability to rapidly image entire organ sections. The intrinsic, distinct broadband MPL emission produced by gold allows us to study these NPs in their biological context without extrinsic labels while also faithfully representing the surrounding tissue architecture via autofluorescence and second harmonic generation. We demonstrate that MPL imaging detects up to 98% more AuNP-positive regions than brightfield microscopy in challenging low-dose (1 nM) conditions and requires no modification of standard histology workflows. Correlative imaging with SEM−EDS confirms high spatial specificity (AUC = 0.955) of MPL for AuNP localization. Dose-dependent retention patterns were observed across liver tissue, and MPL analysis showed strong correlation with ICP−MS quantification. Importantly, histological and immunohistochemical analyses (Masson’s trichrome, CD3, and TUNEL) revealed no significant fibrosis, immune activation, or apoptosis in liver tissue at either low (1 nM) or high (10 nM) doses at 1 year post IV administration. These findings establish MPL imaging as a robust, label-free tool for long-term tracking of AuNPs in biological tissue and highlight its potential for improving biodistribution and safety assessments.

Graphical Abstract

INTRODUCTION

Gold nanoparticles (NPs) are widely utilized across biomedical and industrial applications due to their unique optical and electronic properties.^1–5 Although this broad range of use cases has brought about technological revolutions in multiple fields, the effects of widespread production and subsequent human exposure to these NPs have yet to be fully understood. AuNPs are under active investigation for several biomedical applications, including photothermal cancer therapy, cancer detection, nanotheranostics, antimicrobial treatment, acne treatment, and targeted drug delivery.^6–18 With the rapid adoption of these nonbiodegradable NPs come concerns, particularly with their prolonged retention in the body. The long-term fate and toxicity of gold NPs in biological systems remain poorly understood.^19–23 This presents a critical need for evaluating their biodistribution, clearance, and potential accumulation in biological tissues over extended timeframes.^24–26

In clinical and research pathology, formalin-fixed paraffin-embedded (FFPE) tissue sections or cryosections are stained with hematoxylin and eosin (H&E) and analyzed via brightfield microscopy for morphological evaluation of tissue structure, fibrosis, inflammation, and cellular abnormalities.^27–29 However, gold NPs (~50 nm) are well below the diffraction limit of conventional brightfield microscopes used in pathology clinics.³⁰ They also lack distinct intrinsic chromogenic contrast, making them invisible under standard H&E-stained brightfield imaging unless they present in large aggregates.³¹ Alternative imaging techniques require labeling NPs with either fluorophores or radiolabels that can dissociate or lose signal intensity over time, limiting the time span one can interrogate their biodistribution.^32–34 Elemental mapping with electron microscopy (EM), can offer sensitive NP detection but with limited fields of view and require additional sample preparation steps that can take days and disrupt pathology workflows.^27,35 With sensitivity in the parts per billion, inductively coupled plasma mass spectrometry (ICP−MS) offers high accuracy for detecting trace metals.³² Unfortunately, the technique requires acid digestion of the tissue samples destroying the spatial context of where the NPs reside across the cellular landscape of the tissue.

To address these limitations, we introduce multiphoton luminescence (MPL) imaging, a label-free technique for NP detection that seamlessly integrates into standard histopathology workflows without additional sample preparation. This underutilized optical phenomenon, first observed in metals by Mooradian in 1969, arises from the nonlinear interaction of femtosecond pulsed light with the conduction electrons in metallic NPs, producing a strong broadband emission.³⁶ Unlike more traditional label-free imaging techniques such as dark-field microscopy,^37,38 hyperspectral imaging,^39,40 or phase-contrast microscopy,⁴¹ MPL offers intrinsic subcellular spatial resolution, three-dimensional sectioning capabilities, and spectrally distinct multiphoton signals that are minimally affected by tissue scattering.⁴² While dark-field imaging can visualize NPs based on scattered light, it is limited by background interference from tissue structures and poor depth resolution.⁴³ Hyperspectral imaging provides spectral discrimination but lacks the subcellular spatial resolution, and morphological detail necessary for correlating NP localization with fine tissue architecture, especially in stained sections, and also requires sophisticated unmixing algorithms. Phase-contrast techniques, while well suited for imaging live cells in culture, are not compatible with thick or fixed tissue analysis.

MPL imaging, by contrast, directly exploits the intrinsic optical properties of gold NPs and provides optically sectioned images with minimal background from paraffinized tissue, even under H&E staining, while preserving tissue structure. Moreover, unlike fluorophore-based or radiolabeled techniques, MPL imaging relies on endogenous contrast from the particles themselves, enabling stable long-term detection without photo-bleaching or label dissociation. However, MPL has seldom been used in preclinical studies to visualize metallic NPs within biological tissue.⁴⁴ MPL has been applied to investigate NP uptake, intracellular trafficking, and retention in stem cells and tumor models using in vitro models.^45,46 Our group has also previously demonstrated MPL-based imaging using in vivo models for tracking gold NPs; including visualizing their interactions in free-flowing vasculature in real time.^42,47

In this work, we extend MPL imaging to long-term biodistribution analysis, presenting the first demonstration of label-free MPL tracking of gold NPs in FFPE tissues by integrating it into routine histopathology workflows and filling the substantial gap in understanding the localization and ultimate fate of gold NPs across intact tissue sections. The intrinsic broadband emission produced by gold NPs allowed us to study them in their biological context without the need for extrinsic labels while also faithfully representing the surrounding tissues via autofluorescence and second harmonic generation distinct from the NP spectra.⁴² As a result, we were able to capture the microscopic distribution of these gold NPs in the liver, demonstrating where these NPs reside 1 year post IV administration. This study represents the first demonstration of MPL imaging for NP biodistribution tracking at a 1 year postexposure time point, providing unprecedented insight into the long-term retention and potential biological interactions of gold NPs in intact tissues at multiple concentrations. It is important to note that our MPL imaging method is nondestructive, and the tissue sections can be repeatedly imaged between routine brightfield microscopy and our MPL imaging strategy. This ensures compatibility with routine histopathological analysis while providing highly sensitive label-free detection of gold NPs. By leveraging MPL imaging alongside standard histopathology, we also assessed long-term NP fate, retention, and indicators of potential toxicity. In collaboration with the NIH Nanotechnology Characterization Laboratory (NCL), we further assessed potential toxicity via histopathological and immunohistochemical (IHC) analyses, including Masson’s trichrome staining for fibrosis, CD3 IHC for immune activation, and terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) staining for apoptosis at this 1 year time point. We hope to promote this imaging strategy as a viable, label-free solution to track and understand gold NP behavior and distribution in living models from the subcellular level to the whole organ level.

RESULTS AND DISCUSSION

Localization of Gold NPs in Histology Sections.

FFPE histology section analysis, commonly paired with H&E staining, is widely used to assess tissue morphology in response to disease or exogenous agents. However, brightfield microscopy, the primary technique for imaging H&E-stained tissues, poses significant challenges for visualizing NPs. To illustrate this challenge, we intravenously (IV) injected two groups of mice with 50 nm gold NPs (Figure S1). The “high dose” group received a 10 nM injection, and the “low dose” group received a 1 nM injection of gold NPs. Liver lobes were fixed and prepared for histology using conventional FFPE and H&E staining techniques. The liver was chosen for imaging because it is the predominant organ responsible for clearing NPs from the blood.⁴⁸ Gold NPs, whether as discrete particles or small aggregates, often fall below the diffraction limit of brightfield microscopy, preventing their resolution as distinct structures. Beyond diffraction limitations, additional factors such as low contrast, optical scattering, and background interference may further obscure NP detection. H&E staining specifically can mask gold NPs due to the purple and pink hues resulting from the H&E stain (Figure 1A,D).

Figure 1.

Brightfield and MPL images of H&E-stained FFPE mouse liver sections following IV administration of gold nanoparticles (AuNPs) at high (10 nM) and low (1 nM) doses. (A,D) Brightfield images of H&E-stained liver sections. NP identification is not apparent in the low-dose condition (A), whereas in the high-dose group (D), larger NP accumulations are visible as dark intracellular pigment. (B,E) Corresponding MPL composite images of the same H&E-stained liver sections. AuNP luminescence (purple) is revealed through multiphoton excitation, while eosin autofluorescence and tissue autofluorescence/SHG delineate hepatocytes and tissue structure (yellow/green). Nuclei appear as dark voids due to the absence of autofluorescent emission from the hematoxylin stain. (C,F) NP emission overlays from 425 to 475 nm and 650–720 nm illustrate NP distributions not detectable in brightfield imaging. MPL imaging enhances NP localization while preserving spatial and structural context. (Scale bars: 20 μm).

In cases of high-dose NP exposure, some large NP aggregates become visible as dark black or purple intracellular features (Figure 1D). However, at lower doses, where NPs are more diffusely distributed, they remain mostly undistinguishable with brightfield imaging (Figure 1A). This limitation creates a gap in understanding NP distribution and biological interactions at physiologically relevant concentrations. Traditionally, pathologists have inferred NP presence based on darkly pigmented regions in H&E-stained tissues, but this lacks specificity and cannot capture the full biodistribution profile of NPs at the cellular level.

To address these limitations, we applied MPL imaging, a label-free, nondestructive technique that leverages the intrinsic optical properties of gold NPs. Details of our image acquisition and optimization process are provided in the following “Multiphoton Luminescence Imaging Workflow” section. Unlike brightfield imaging, MPL imaging selectively detects gold NPs based on their multiphoton-excited emission, overcoming spectral interference from hematoxylin or eosin. In both high- and low-dose groups, MPL imaging reveals gold NPs that are undetectable in brightfield microscopy (Figure 1). Additionally, tissue structure remains apparent due to the autofluorescence and second harmonic generation (SHG) effects from the H&E stains providing clear visualization of hepatocytes and tissue architecture (Figure 1B,E). The portal triad, consisting of the portal vein (PV), bile duct (BD), and the hepatic artery (HA), is also easily identified in the MPL composite images (Figure 1B,E) offering spatial context of NP accumulation within the tissue. As expected, the NPs appear to accumulate within the sinusoids of the liver, where Kupffer cells reside and sequester NPs. Visibility of morphological context is crucial for further inferences into biodistribution and ability to study other organs in the future.

Multiphoton Luminescence Imaging Workflow.

To determine optimal excitation conditions, infrared multiphoton (IR-MP) excitation wavelengths from 800 to 1300 nm were investigated. Gold NPs exhibited luminescence when excited between 800 and 1200 nm, with peak emission at 1040 nm (Figure S2B). Notably, this wavelength also excites eosin autofluorescence (Figure S2A), allowing simultaneous visualization of both tissue morphology and NP localization (Figure 2C), providing essential spatial context for biodistribution analysis.

Figure 2.

Microscope schematic for MPL histology imaging and tissue/gold NP emission curves. (A) Microscope configuration for MPL imaging of histological sections, utilizing four detection channels corresponding to conventional fluorescence filter sets. (B) Emission spectra of gold NPs and tissue at 1040 nm pulsed excitation. While NP luminescence overlaps with endogenous tissue autofluorescence, spectral differences enable selective NP detection. (C) Representative MPL images from an H&E-stained liver section, showing eosin and tissue autofluorescence (500–625 nm) alongside distinct NP signals detected in the “DAPI” (425−475 nm) and “Cy5” (650–720 nm) channels. (D) NP localization was determined by correlating Cy5 and DAPI emission channels, generating a composite NP detection channel independent of tissue autofluorescence. (E) Tissue localization was determined by correlating FITC and TRITC emission channels, generating a structural map with details of tissue architecture and hepatocytes. (F) Four-channel composite image displaying both tissue and NP emission, where tissue structure is delineated in yellow, nuclei appear as dark voids, and gold NPs are visualized in purple. (Scale bars: 20 μm).

Gold NPs produce a broad multiphoton-excited luminescence spectrum spanning 380–730 nm (Figure S2A), which overlaps with tissue autofluorescence and eosin emission (500–700 nm). To isolate NP signal from background fluorescence, we implemented a four-channel detection system aligned with standard fluorescence filters: DAPI (425–475 nm), FITC (500–550 nm), TRITC (570−625 nm), and Cy5 (650–720 nm). While tissue autofluorescence contributes across multiple channels, NP luminescence is most distinct in the DAPI and Cy5 ranges. By first correlating these two channels, we can generate a binary NP localization map, effectively removing background autofluorescence and ensuring high specificity in NP detection. NP images without tissue channels are presented as an overlay of DAPI and Cy5 (Figure 2D).

A key advantage of MPL imaging is its broadband spectral emission. Unlike endogenous fluorophores and standard fluorescent dyes, which typically emit within narrow spectral windows (~100 nm range), gold NPs exhibit emission from the UV to NIR (380–730 nm). This unique property allows us to select detection channels that exclude tissue autofluorescence, significantly enhancing imaging contrast. The broad emission of gold NPs also enables flexible integration with histological stains (e.g., H&E, trichrome) without additional labeling (Figure 2F). This MPL imaging workflow is also compatible with the standard slides that have already been prepared for clinical histology (Figure S3). While FFPE tissue preparation with H&E staining is a widely adopted standard for clinical histology, the methodology we have developed for MPL imaging is also adaptable to cryosections and alternative unstained histological processing techniques (Figure S3). However, staining with H&E appears to provide better cellular contrast and tissue architecture visibility as opposed to unstained tissues due to the strong intrinsic autofluorescence seen in the TRITC and FITC channels.

Validation of MPL Imaging for Gold NP Localization.

To evaluate the accuracy of MPL imaging for NP localization, we compared it with gold-standard electron microscopy techniques, including scanning electron microscopy (SEM) and energy-dispersive X-ray spectroscopy (EDS). While SEM and EDS provide high-resolution imaging of gold NPs, their application is limited by the destructive sample preparation and small restricted fields of view. In contrast, MPL imaging offers nondestructive NP detection within intact histological sections, preserving spatial context and enabling analysis across whole organ tissue sections.

Figure 3A presents an MPL image of a mouse liver that has been H&E stained, where luminescence signals (pink/white) indicate NP accumulation with varying degrees of aggregation. Due to the emission intensity and optical properties of gold NPs, MPL imaging can detect both individual NPs appearing as single pixels and larger aggregates spanning multiple pixels. Data supporting that our MPL imaging technique is capable of detecting single gold NPs has been previously published by our group.⁴² This capability is particularly important given the heterogeneous nature of NP accumulation in the liver, where distribution ranges from isolated NPs to densely packed agglomerates. To further refine NP segmentation, we employed a dual-channel correlation approach, which isolates NP signal from tissue autofluorescence, ensuring high sensitivity and specificity for NP detection (Figures S4 and S5). This method remains effective across a broad spectrum of NP distributions, including diffuse, low-density accumulations, which were challenging to segment using conventional image processing techniques. Correlation of multiple nanoparticle channels with a >300 nm spectral separation allowed for diffraction limited sensitivity and specificity while effectively suppressing noise (Figure S5). Threshold-based segmentation methods on single emission channels commonly used in cell studies struggled with irregular NP clustering, leading to an under-estimation of NPs in histological contexts.

Figure 3.

MPL, SEM, and EDS imaging colocalizing gold NPs in FFPE H&E-stained liver tissue section. (A) MPL image of mouse liver tissue exposed to gold NPs. Red blood cells highlighted in red, hepatocytes in orange and gold NPs in pink/white. (B) Backscatter SEM image of the same FFPE section highlights gold NP location in white. (C) EDS image of the section with gold NP location highlighted in orange via characteristic X-ray emission of gold. (D) Correlation of MPL and EDS localization of gold NP accumulation demonstrated the high sensitivity and linearity of both techniques for positive gold identification with no errant tissue positives. Gold can also be localized in the MPL images from diffraction-limited aggregations. (Scale bars: 25 μm).

Following MPL imaging where a “map” of the tissue was generated (Video S1), the same tissue sections were processed for SEM (Figure 3B) and EDS (Figure 3C) to provide a direct comparison of NP localization. Our MPL signal intensity showed a strong linear correlation with the EDS-derived elemental gold detection (R² = 0.9907, AUC = 0.955) confirming the high sensitivity and specificity of MPL imaging (Figures 3D and S5). While SEM and EDS remain essential for ultrastructural NP characterization, their destructive sample preparation, limited scalability, and preparation time make them impractical for large-scale analysis in biodistribution studies. In contrast, MPL imaging enables high-throughput NP detection in intact histological sections while maintaining compatibility with conventional brightfield microscopy and pathology workflows. The strong agreement between MPL and EDS reinforces the detection capabilities of MPL, which achieves a significantly larger field of view while preserving tissue architecture.

Whole-Tissue Imaging and Integration within Pathology Workflows.

A significant advantage of MPL imaging is its ability to perform whole-tissue imaging while maintaining high resolution and compatibility with existing pathology workflows. Unlike electron microscopy-based approaches, which require lengthy sample preparation, destructive processing, and are constrained to small fields of view, MPL imaging needs no additional sample preparation, is nondestructive, and enables large-scale tissue visualization while preserving spatial context. This capability facilitates a more comprehensive assessment of gold NP biodistribution across entire histological sections, within the intact biological environment of the tissue, minimizing the risk of misinterpretation.

Figure 4 illustrates the workflow integration of MPL imaging with standard histological and pathology techniques. MPL imaging can be conducted directly on FFPE sections without additional sample preparation, preserving compatibility with brightfield microscopy, histological stains, and IHC. SEM-based approaches require coverslip removal, dehydration, and carbon coating, rendering samples unsuitable for subsequent optical imaging or further pathological analysis (Figure 4B). These preparation steps can take several days to process for generating a relatively small field of view (100 μm × 100 μm) in tissue using EM-based methods. In comparison, MPL imaging (Figure 4D) does not modify histological sections, allowing for quick (~15 min) whole tissue imaging with much larger fields of view. The same tissue sample can then be imaged across multiple modalities for correlative imaging evaluation if desired.

Figure 4.

Comparison of MPL imaging and SEM sample preparation, field of view, and imaging time. (A) MPL imaging workflow showing no additional preparation of histology slide before MPL imaging. (B) Workflow for preparing histological slides for SEM imaging, including coverslip removal, dehydration, and carbon coating, a process requiring multiple days. Histology samples must also be trimmed to fit SEM sample stubs. (C) Comparison of imaging techniques in terms of preparation time and field of view. MPL imaging requires no additional sample modification, allowing for direct imaging of intact histological sections. Differences in estimated imaging time between SEM and MPL for whole-tissue section imaging are also highlighted. MPL imaging enables both subcellular-resolution NP detection and whole-slide imaging, facilitating comprehensive biodistribution assessment. (D) Example of whole-slide MPL imaging providing rapid NP localization, achieving high-resolution detection within minutes. Video S2 demonstrates the versatility of whole slide and subcellular MPL imaging in one imaging platform.

Beyond its ease of integration, MPL imaging offers scalability across different spatial resolutions (Video S2). This technique enables whole-slide scanning for comprehensive biodistribution analysis (Figure 4D) while also providing high-resolution imaging for subcellular NP localization (Figure S6). The ability to seamlessly transition between macroscopic and microscopic scales facilitates a more robust interpretation of NP accumulation within relevant tissue structures.

Long-Term Distribution of Gold NPs in the Liver: 1 Year Post NP Exposure.

Since gold NPs are not biodegradable, there remain major concerns with regard to their potential lasting toxicity effects in humans post exposure. Understanding the long-term retention and spatial distribution of gold NPs in biological tissues is essential for evaluating their fate and potential biological effects.⁴⁹ Standard biodistribution studies that label NPs with either a radiolabel or fluorophore to follow their localization over time may not be suitable for long-term evaluation in histology tissues. The label would have likely either dissociated from the NP or the signal would no longer be detectable at such long time points post exposure. Researchers have utilized ICP−MS to evaluate total gold content of various organs post administration, but this requires destructive acid digestion of the organs and thus the loss of any spatial information or cellular changes from the tissue.^32,50–54 Our label-free MPL imaging technique is well suited to localize gold NPs and assess their long-term fate and morphological effects across intact histology sections with subcellular detail due to both the intrinsic luminescent phenomenon and high specificity of the technique. While MPL imaging of intact thick tissues can be inherently limited by optical scattering and absorption (with typical imaging depths of ca. 100–300 μm depending on tissue type and excitation wavelength),⁵⁵ the thin FFPE sections used in this study (4–5 μm) eliminate such constraints. Thus, resolution and depth limitations of MPL imaging are negligible in our experimental context, enabling high-fidelity detection of gold NPs at subcellular detail across entire tissue sections. For future studies aiming to examine deeper tissue architecture in 3D, optical clearing techniques may be employed to enhance MPL imaging depth while preserving spatial context.⁵⁶ Studies have already demonstrated that conventional multiphoton microscopy in conjunction with tissue clearing enables deep tissue imaging with high resolution.^57,58 Tissue clearing of whole organs coupled with our MPL imaging workflow offer a promising tool for volumetric analysis of gold NP distribution in biological tissues.

To investigate these long-term dynamics, we employed MPL imaging to assess gold NP accumulation in mouse tissues 1 year post-IV administration. Exposure groups included high dose (10 nM, n = 4), low-dose (1 nM, n = 3), and control (no NPs, n = 2) (Figure S8). ICP−MS was utilized to first establish which organs had accumulated the most gold NPs at 1 year post exposure (Figure S8). As expected, the liver showed the most accumulation of the 50 nm gold NPs. Therefore, we proceeded with establishing our MPL imaging workflow using the liver tissue for the remainder of our study. Figure 5 presents a whole-liver lobe section analysis, with a blue binary overlay highlighting NP localization with distinct dose-dependent distribution patterns observed (Figure 5B,D). To assess the quantitative accuracy of MPL imaging, we compared NP distributions from MPL imaging with ICP−MS. To account for differences in value units and measurement scales, data from each technique was normalized within the groups. After normalization between the two techniques, the distribution ratio of NP accumulation via MPL imaging closely matches ICP−MS data (Figure 5E). Importantly, unlike ICP−MS, which requires complete tissue digestion and sample destruction, MPL imaging preserves the sample and spatial context, enabling simultaneous quantification and high-resolution mapping of NP distribution within intact histological sections.

Figure 5.

MPL imaging of gold NP distribution in mouse liver sections 1 year post intravenous exposure. (A−D) Dose-dependent accumulation of gold NPs in liver visualized by MPL imaging, where NP localization is represented in the blue binary mask, and eosin/tissue autofluorescence appears in orange. (B,D) MPL imaging reveals a dose-dependent pattern of NP aggregation, with larger, more concentrated NP pockets in the high-dose (10 nM) group and smaller, more dispersed accumulations in the low-dose (1 nM) group. (E) Comparison of ICP−MS quantification and MPL imaging shows a strong correlation in gold distribution, demonstrating the potential of MPL imaging for semiquantitative NP analysis. No significant difference between MPL and ICP−MS was found for all groups (p > 0.05). (F) Spatial analysis of NP aggregation highlights that at higher doses, NPs form fewer but denser clusters, whereas at lower doses, they are more frequently distributed but in smaller accumulations, consistent with the visual dose-dependent trends observed. Significant differences in accumulation size distribution were found for aggregates >100 μm and <15 μm (p < 0.001) indicated by *.

At 1 year, NPs were more evenly dispersed, forming smaller and more numerous aggregates in 1 nM liver tissue. In contrast, the high-dose group exhibited fewer but larger NP accumulations, suggesting a localized sequestration effect, likely within hepatic structures such as Kupffer cells (Table S1). This condensation effect indicates that higher NP doses favor aggregation, whereas lower doses result in a more diffuse distribution throughout the tissue. In high-dose samples, NP clusters frequently exceeded 150 nm, reinforcing the trend toward localized sequestration at higher doses (Figure 5F).

Comparison of Brightfield Imaging and Multiphoton Luminescence Imaging to Identify NP Localization in Liver Tissue.

We collaborated with the Nanotechnology Characterization Lab (NCL) to compare our MPL imaging technique with an established brightfield imaging method that is widely used to localize gold NPs. The NCL is a specialized resource that was first established by the National Cancer Institute (NCI) within NIH to provide preclinical characterization and safety testing of NPs. The data they compile is routinely evaluated by regulatory agencies to help determine safety and efficacy of new NP constructs.

FFPE liver tissue blocks from our 1-year study (both experimental and control groups) were evaluated in parallel with the NCL. Sections of FFPE livers from the same organ were sectioned and stained independently at their facility with H&E and trichrome to evaluate morphology and identify NPs. Their standard histopathological workflow relies on using brightfield microscopy to look for concentrated areas of dark black pigment that represent gold NPs (Figure S9). Brightfield microscopy remains the standard for histological NP evaluation, particularly in pathology workflows. However, gold NPs lack an intrinsic chromogenic signature, making their identification in brightfield microscopy dependent on visualizing dark intracellular pigment. While this approach is effective for detecting large NP aggregates, it is inherently limited in its ability to resolve smaller, more diffuse NP distributions. NCL independently evaluated these areas of concentrated black pigment by applying a machine learning (ML)-assisted segmentation approach using the QuPath digital pathology software package to enhance NP detection in brightfield images, classifying dark regions as gold NP-positive. While this method improves semiquantitative NP assessment, it remains susceptible to false negatives, particularly in low-dose conditions or in tissue regions where NP accumulation does not produce overt pigmentary changes. The same samples were returned to us and evaluated with our MPL imaging workflow.

MPL imaging provides a fundamentally different approach to NP detection, leveraging the intrinsic optical properties of gold rather than indirect chromogenic contrast. By utilizing broadband spectral emission, we demonstrate that MPL imaging circumvents the limitations of color-based histology, enabling direct visualization of both large aggregates and smaller, dispersed gold NP populations. Figure 6 demonstrates that MPL imaging detects significantly more NP-positive regions than brightfield segmentation, revealing up to a 50-fold increase in NP signal per field of view. Brightfield imaging struggled to identify gold NPs to a greater degree in low dose groups, failing to identify up to 98% of the gold NP accumulations as compared to MPL imaging. These findings indicate that NP distributions inferred solely from brightfield microscopy likely underestimate the true extent of NP localization, an inherent limitation of solely using brightfield microscopy to estimate NP distribution in histology sections.

Figure 6.

Quantification and analysis of dose dependent gold nanoparticle biodistribution in the liver 1 year post intravenous exposure. (A) Brightfield imaging of trichrome stained liver section reveal large pockets of gold nanoparticle aggregates, but smaller pockets are not visible. Multiphoton imaging of the same section allows for visualization of both the biological context and precise NP localization. Although gold aggregates are visible in higher dose fields of view, the majority of gold present in the tissue section is not able to be visualized in brightfield images. (B) Brightfield analysis of the trichrome sections using a QuPath random forest machine learning model can identify larger clusters of nanoparticles in bright field sections but fails to differentiate smaller clusters. Brightfield imaging failed to identify 98% and 94% of nanoparticles in the frames compared to multiphoton imaging for the low and high dose, respectively. Significant differences between MPL and brightfield analysis was found for experimental groups (p < 0.05) indicated by *. (Scale bars: 100 μm).

Multi-site Assessment of NPs near Fibrotic Regions.

To further evaluate the clinical implications of these differences, we examined fibrotic regions identified by our collaborating pathologist at the NCL and correlated them with NP accumulation patterns. Figure 7 highlights trichrome-stained liver sections, where fibrotic areas appear in blue, comparing NP detection across brightfield and MPL imaging. While brightfield analysis suggested only sporadic NP-fibrosis colocalization, MPL imaging consistently identified NP accumulations within fibrotic regions.

Figure 7.

Areas of fibrosis identified by pathologist stained with trichrome blue. Pathological analysis of gold NP localization in relation to fibrosis in trichrome-stained liver sections. (A) Brightfield imaging reveals isolated fibrotic regions (blue) in both low and high dose groups, with or without visible NP aggregates. (B) MPL composite images of same slide and region revealing the clear distribution of gold NPs (yellow binary) across the respective dosing groups. (C) Despite the presence of visible NP aggregates in brightfield microscopy, quantification indicates that 52.4% and 91.6% of total NP accumulation in high and low dose samples, respectively, remained undetected. In fibrotic regions where NPs were not visibly apparent in brightfield microscopy, 67.4% and 100% of NPs in high and low dose groups, respectively, were undetected. These findings highlight the inherent limitations of standard brightfield imaging in assessing NP-fibrosis colocalization, underscoring the need for more sensitive detection methods such as MPL imaging. Significant differences between MPL and brightfield analysis was found for experimental groups (p < 0.05) indicated by *. (Scale bars: 50 μm).

Quantitative analysis of these fibrotic regions revealed a significant discrepancy between brightfield and MPL-based NP detection. In fibrotic regions where nanoparticles were already visible in brightfield microscopy, traditional analysis still failed to identify 91% and 52% of NP accumulation in the low- and high-dose groups, respectively. The limitations of brightfield imaging were even more pronounced in fibrotic regions without visible NP aggregates, where 100% and 67% of nanoparticles in low-and high-dose tissues, respectively, were undetected. These findings suggest that conventional histopathological workflows may systematically underestimate NP retention in fibrotic tissues, which could affect conclusions regarding long-term gold NP biocompatibility and clearance.

By collaborating with the NCL, we demonstrated that current histopathology-based NP assessments, while valuable, may not fully capture the extent of NP distribution. MPL imaging provides a more complete, high-resolution assessment, particularly for identifying small NP aggregates that evade detection in brightfield microscopy. These findings highlight the importance of complementary imaging strategies in NP biodistribution studies, particularly when evaluating long-term retention and potential tissue effects.

Histopathological Evaluation of Liver Tissue: 1 Year Post NP Exposure.

To assess potential tissue-level effects associated with long-term gold NP retention in the liver, histopathological analysis was conducted in conjunction with our collaborators at the NCL. The same tissue blocks discussed above were stained with Masson’s trichrome to evaluate fibrosis and collagen deposition. In addition, IHC staining for T cell visualization via CD3 was used to assess inflammation and TUNEL staining was used to evaluate apoptosis.

Histological examination revealed gold NP pigment localized primarily within Kupffer cells and perivascular spindle cells, with greater pigment accumulation in the high-dose group (Figures S9 and S10). Some portal inflammation was observed, with mild-to-moderate infiltration of neutrophils, histiocytes, and lymphocytes, but this response was not clearly dose-dependent (Figure S11). Notably, fibrotic changes were limited, with localized increases in collagen deposition (Figure S10) rather than a diffuse fibrotic response. The presence of trichrome-positive regions between the portal tract and central vein was atypical but remained focal, indicating subclinical structural changes rather than widespread hepatic fibrosis.

Further evaluation of immune activation and apoptosis showed no significant increases in T-cell infiltration (CD3 staining) or apoptotic activity (TUNEL staining) across exposure groups (Figure S11). Occasional CD3-positive cells were observed near NP deposits, but quantitative image analysis confirmed no significant immune-mediated tissue damage (Table S1 and Figure S11). Similarly, apoptotic cells were rare and did not colocalize with gold NP accumulation, reinforcing the conclusion that gold NP exposure did not induce sustained apoptotic responses at 1 year (Figure S12).

Taken together, these findings indicate that, while gold NPs persist in the liver at 1 year, they do not induce overt hepatotoxicity, significant immune activation, or widespread fibrosis. The localized fibrotic response suggests a tissue adaptation process rather than a pathological consequence, supporting the long-term biocompatibility of gold NPs in the liver at the tested doses.

CONCLUSIONS

This study demonstrates the use of MPL imaging as a highly sensitive and label-free approach for detecting and characterizing gold NP biodistribution in histological sections. Utilizing the intrinsic luminescence properties of gold, MPL imaging overcomes the limitations imposed by other imaging approaches such as fluorescence-based methods or radiolabel-based imaging techniques, which are hindered by the need for exogenous labels that may not remain with the NPs over long periods. The ability to quantitatively assess NP distribution at long time points while preserving spatial context makes MPL imaging a valuable tool for studying NP fate in biological systems.

At 1 year post IV administration, MPL imaging revealed dose-dependent NP accumulation patterns in liver tissue, with larger, localized NP clusters in high-dose samples and more diffuse distributions in low-dose samples. Comparisons of MPL imaging with EDS gold quantification showed a strong correlation (R² ≈ 0.99), reinforcing MPL’s reliability for gold NP detection while maintaining tissue integrity. ICP−MS quantification and MPL imaging also showed a strong correlation, confirming that MPL imaging accurately reflects gold NP distribution trends observed with bulk elemental analysis. While ICP−MS provides precise quantification of total gold content, it requires tissue digestion and thus a loss of spatial information, whereas MPL imaging preserves histological context, enabling both quantitative assessment and spatially resolved NP localization within intact tissue sections.

Beyond biodistribution, our study also highlights the limitations of the current standard used for gold NP detection in histopathology that uses brightfield microscopy. Brightfield imaging failed to identify up to 98% of NPs in low-dose conditions. MPL imaging, on the other hand, was capable of identifying gold NPs throughout the tissue, even in fibrotic liver regions where no NPs were detected with brightfield microscopy. These results suggest that brightfield microscopy underestimates NP accumulation, particularly in tissues with structural alterations, emphasizing the need for more sensitive imaging techniques in nanomedicine research.

Histopathological and IHC analyses showed no clinically significant immune or apoptotic responses in liver tissue 1 year after NP exposure in either high (10 nM) or low (1 nM) dose groups. TUNEL staining indicated no increase in apoptotic activity, and CD3 immunostaining revealed no abnormal T-cell infiltration across exposure groups. Similarly, Masson’s trichrome staining showed localized, nonprogressive fibrotic changes, suggesting a tissue adaptation response rather than a pathological effect of NP retention. These results demonstrate the long-term biocompatibility of gold NPs in liver tissue at the doses examined.

As NPs have revolutionized several technological and industrial sectors, inadvertent human exposure rates to engineered NPs are increasing, generating the need to better understand their short- and long-term effects to ensure the public’s safety. While our study focused on long-term effects, this label-free imaging technique could also readily investigate earlier time points postexposure to potentially evaluate acute inflammation triggered by NP administration. Looking ahead, MPL imaging could be used to investigate biodistribution following alternative exposure routes, such as inhalation or ingestion to assess tissue-specific NP accumulation and clearance. Extending this technique to study other metallic NPs (e.g., silver, titanium dioxide, iron oxide) that humans encounter could improve our understanding of their ultimate fate in the body and reveal any adverse biological effects in harvested tissues with microscopic detail. Beyond gold NPs, MPL imaging holds promise for visualizing other metallic NPs that exhibit intrinsic nonlinear optical properties. Silver nanoparticles, for example, are known to display strong two-photon-induced luminescence due to their plasmonic behavior,^59,60 while titanium dioxide and iron oxide NPs have demonstrated nonlinear optical responses under specific excitation conditions.^61,62 Prior reports suggest that MPL could be extended to these materials for label-free detection. Future work exploring particle-specific excitation/emission properties will help define the broader applicability of MPL in nanotoxicology and biodistribution studies involving diverse nanomaterials. We hope this work demonstrates the many advantages of utilizing MPL imaging and offers the nanocommunity a new label-free imaging approach for visualizing and tracking gold NPs on the microscopic scale across a broad range of times post exposure.

EXPERIMENTAL SECTION

NP Fabrication and Characterization.

A modified Turkevich method was used to fabricate spherical AuNPs.⁶³ Prior to synthesis, all glassware was cleaned with aqua regia. Gold(III) chloride hydrate (HAuCl₄·xH₂O, 99.995%) and trisodium citrate dihydrate (C₆H₅Na₃O₇·2H₂O, 99.0%) were purchased from Sigma-Aldrich. Briefly, 2 mL of 30 mg mL⁻¹ HAuCl₄ was added to 300 mL of boiling water under vigorous stirring. Then, 2.4 mL of 30 mg mL⁻¹ sodium citrate solution was rapidly added. The colloidal solution was stirred for an additional 10 min. The prepared AuNPs were concentrated via centrifugation at 1100g for 30 min. Deionized water (Milli-Q grade, Millipore) with a resistivity of 18.2 MΩ cm was used throughout the experiment.

The resulting colloidal solution was characterized using UV−vis and NP tracking analysis (NTA) with a NanoDrop2000 (Thermo Fisher Scientific, US) and a NanoSight NS300 (Malvern Panalytical, UK), respectively. The maximal absorption band of the AuNPs was observed at 525 nm. The size and concentration were verified with NTA measurements, which indicated the size distribution of 54 ± 6 nm. Final stock was diluted to 10 nM with deionized water. All NP concentrations reported correspond to the particle concentration in the colloidal suspension.

Animal Experiments.

Eight-week-old female Balb/c mice (Charles River) were used for all NP biodistribution and imaging studies. All animal procedures were approved by the University’s Institutional Animal Care and Use Committee (IACUC no. 21311) and conducted in accordance with established guidelines for the humane care and use of laboratory animals.

Animal Injections.

Animals were divided into three groups for the 1-year time point to evaluate biodistribution differences based on injection concentration at this extended time point. The three groups, referred to as high (n = 4), low (n = 3), and control (n = 2), received IV tail vein injections of gold NPs at concentrations of 10 nM, 1 nM, and 0 nM (no injection), respectively. Each mouse in the high- and low-dose group received a 200 μL IV injection of 54 nm gold NPs using a 26 gauge needle. The low dose solution was directly diluted from the high dose stock to ensure accurate dosing.

To validate the dosing strategy and confirm visualization of NP biodistribution in the liver, we also included a one-month injection cohort. This validation group, consisting of one mouse per dose received 200 μL of 54 nm AuNPs at either 10 nM or 1 nM concentrations.

Organ Harvesting and Processing.

Each mouse was euthanized by cervical dislocation under deep isoflurane anesthesia. Tissues were harvested, rinsed in PBS, weighed, and subsequently imaged. Organs were subdivided into two lobes or sections to evaluate the effects of fixation and processing methods. The liver was separated into lobes, with the right medial lobe bisected; lungs were divided into left and right sides; the brain into two hemispheres; kidneys were harvested individually; and the spleen was cut along the transverse plane into two pieces.

The two one-month organ groups were further divided to evaluate differences in the intrisic tissue autofluorescence generated by the fixative itself. All organ sections were initially fixed for 72 h in 20 mL of fixative, then bisected and processed either for paraffin embedding or cryosectioning. For paraffinization, one portion of each organ was further fixed in 10% neutral-buffered formalin (NBF) for 24 h at room temperature. Samples then underwent sequential dehydration in graded ethanol (70%, 80%, 95%, and 100%), followed by xylene clearing and paraffin embedding to produce FFPE tissue blocks. The blocks were sectioned into 5-μm-thick slices using a rotary microtome (Leica RM2235, Leica Biosystems) for subsequent histological and IHC analysis.

The second portion of each organ was subjected to cryosectioning to preserve native biomolecular architecture. Samples were snap-frozen in optimal cutting temperature (OCT) compound (Tissue-Tek, Sakura Finetek) using liquid nitrogen. Frozen tissue blocks were sectioned at 10-μm thickness using a cryostat (Leica CM1950, Leica Biosystems) at −20 °C. Sections were mounted onto glass slides and stored at −20 °C until further analysis to ensure preservation of structural and molecular integrity.

Histological Staining and Mounting.

Sections from paraffin-embedded tissues were deparaffinized in xylene and rehydrated through a graded ethanol series (100%, 95%, 80%, and 70%) into distilled water. H&E staining was performed at the USC’s Translational Research Lab (TRLab) using an automated stainer (Varistain Gemini ES Automated Slide Stainer, Thermo Fisher), following the manufacturer’s protocol. Slides were stained with Gill’s hematoxylin (Sigma-Aldrich) for nuclear visualization, followed by eosin Y solution (Thermo Fisher Scientific) for cytoplasmic contrast. After staining, sections were dehydrated, cleared in xylene, and mounted with Cytoseal 60 (Thermo Fisher Scientific) for imaging and long-term preservation.

Cryosections were processed in a similar manner, with staining carried out using the same automated system. Prior to staining, sections were incubated in 100% ethanol and subsequently loaded into the stainer. Unstained cryosections were air-dried at room temperature for 10 min and directly mounted with Cytoseal 60. Both stained and unstained slides were stored at room temperature under light-protected conditions to preserve tissue integrity.

Acid Digestion and Gold Content ICP−MS Analysis.

Organ samples from the 1 year cohort were processed for gold quantification using ICP−MS. Tissue samples were digested according to the protocol utilized by the Nano and Pico Characterization Laboratory (NPC) at the California NanoSystems Institute (CNSI), University of California, Los Angeles (UCLA). Tissue sections were weighed and placed in glass vessels and stored at −80 °C until digestion.

Gold quantification was performed using a PerkinElmer NexION 2000 ICP−MS instrument housed at the NPC Core Facility at UCLA. Calibration standards were prepared in-house by serial dilutions of gold NP stock solutions (from the IV injection stock administered to the mice) into Milli-Q water. NP injections were provided to the facility to ensure consistent and accurate measurements. The quantified gold content in each sample was normalized to tissue mass and expressed as nanograms of gold per milligram of tissue. Quality control measures included analyzing blank samples and standard reference materials to verify the accuracy and reproducibility of the measurements. Each sample was run three times for gold content.

SEM and EDS Histology Preparation.

To prepare the tissue sections for SEM, coverslips were removed by floating them off in xylene through a 24 h incubation at 22 °C. Organ sections remain affixed to the glass slide. The sections were then dehydrated by three successive incubations in 100% ethanol, followed by immersion in hexamethyldisilazane (HMDS, Sigma-Aldrich) with three media changes as an alternative to critical point drying. Once the samples were fully dried, they were mounted on aluminum stubs using carbon tape and sputter-coated with carbon.

SEM imaging was performed on an FEI Nova NanoSEM 450 (Thermo Fisher Scientific, Waltham, MA). Images were acquired using a spot size of 2.0, an accelerating voltage of 5 kV, and a stage bias of 2000 V. A backscatter detector was used to visualize gold NPs in the tissue sections. Areas of interest were coregistered with multiphoton autofluorescence images to ensure precise alignment.

Following initial SEM imaging, elemental confirmation of gold within the identified regions was achieved using EDS on a Helios G4 PFIB UXe DualBeam FIB/SEM (Thermo Fisher Scientific) equipped with an Oxford UltimMax 170 silicon drift detector for EDS and Oxford AZtecHKL electron backscatter diffraction (EBSD). EDS spectra were collected at an accelerating voltage of 15 kV, and elemental maps were generated to confirm the presence of gold. These maps were overlaid with SEM images to validate NP localization.

Brightfield and MPL Imaging.

Brightfield imaging was conducted using a Leica Mica system (Leica Microsystems) to capture the overall histological features of the tissue sections. Whole tissue stitched images were acquired as necessary at 10× magnification.

Following brightfield imaging, MPL imaging was performed using a Leica SP8 inverted microscope (Leica Microsystems) equipped with an Insight DeepSee ultrafast tunable laser (Spectra-Physics) for spectral emission analysis of gold NPs and tissue autofluorescence. Additional MP imaging for gold quantitation and MP histology was performed on a Nikon A1R HD upright microscope (Nikon Instruments Inc., Melville, NY) equipped with a 25× 1.1 NA water immersion objective and a Chameleon Discovery ultrafast tunable laser (Coherent Corp.).

Excitation on the Nikon A1R HD system was set at 1040 nm with 1% laser power, corresponding to an 80 MHz repetition rate, yielding an average power (P_avg) of 100 mW, pulse energy (E_pulse) of 1.25 nJ, and peak power (P_peak) of 8.93 kW. To ensure stability, the laser power was allowed to equilibrate for 1 h prior to imaging. Emission signals were collected through a Nikon IR ND emission filter and split into four descanned DU4 detectors with bandpass emission filters at 450/50 nm, 525/50 nm, 600/50 nm, and 685/70 nm. Laser power at the sample was carefully optimized to avoid photothermal melting or deformation of the AuNPs, and no morphological changes or artifacts were observed during or after imaging of nanoparticles or tissue.

Multiphoton autofluorescence imaging was used to identify regions of interest (ROIs), particularly for visualizing gold NP distribution within tissue sections. Tiled images were acquired to cover the entire sample, and a software-based overlay was generated to enhance localization and quantification of gold NP signal.

NCL Histological Staining and Processing.

Liver tissues were collected from three experimental groups: control, 1 nM gold NPs, and 10 nM gold NPs. Two serial sections were obtained from each paraffin-embedded tissue block: one stained with H&E to assess gold NP distribution and tissue morphology, and the other stained with Masson’s trichrome to evaluate collagen deposition and fibrosis. Histological examination was visually performed to assess NP accumulation and associated tissue responses. To further characterize tissue responses, deeper sections from the same tissue blocks were analyzed using IHC for CD3+ T cells to localize areas of inflammation and TUNEL staining for localizing apoptotic cells. Sections were stained according to standard protocols, and positive/negative controls were included for validation (CD3: mouse spleen; TUNEL: mouse testes). Images were acquired and analyzed using QuPath (v0.4.4) for automated quantification of gold NP accumulation (H&E), fibrosis (Masson’s trichrome), CD3+ T-cell density (IHC), and apoptotic cells (TUNEL). Positive pixel percentage was calculated for each parameter and reported for all samples. Gold NP pigment was distinguished from IHC-positive staining using spectral differentiation in the analysis pipeline. Representative images were acquired to illustrate findings, with overlays created to visualize colocalization patterns.

Histology and NP Quantitation.

MPL Imaging Quantification.

Quantitative analysis of gold NPs in tissue sections from MPL histology imaging was performed using Nikon Elements software (version 5.30.06). Images were captured across four channels, with DAPI and Cy5 channels used to create a derived channel representing correlated emission. The following workflow was used for NP identification:

1. A correlation channel was created to identify gold NPs by analyzing the overlap between the DAPI and Cy5 channels, using a pixel neighborhood of 10 pixels to identify spatially overlapping signals.

   1. To segment regions of MPL signal corresponding to gold nanoparticles we employed a spatial−spectral correlation approach based on local Pearson’s correlation coefficients between emission channels. Raw MPL image stacks were first preprocessed in NIS-Elements AR to isolate the DAPI and Cy5 channels, which exhibit strong AuNP luminescence and minimal background autofluorescence in tissue.
   2. For each pixel, the Pearson correlation coefficient (r) was computed over a 10 × 10 pixel window between the DAPI and Cy5 channels. This step produced a floating-point correlation map (range: −1 to 1), where high values indicate spectrally and spatially co-occurring emission. Because AuNPs emit across both channels due to their broadband MPL characteristics, while most endogenous fluorophores emit only in a single channel, this dual-channel correlation suppresses tissue autofluorescence and random noise.
   3. The resulting r-map was thresholded to generate a binary segmentation mask. While the threshold (typically r > 0.6–0.8) was selected manually, the correlation image itself was highly selective and stable across samples, minimizing subjectivity. Following thresholding, a 4 pixel square dilation was applied to ensure connectivity of closely spaced pixels, and contiguous regions (“blobs”) were labeled and extracted. Metadata including local r-values and spatial properties were retained for each segmented region.

2. The correlation channel was thresholded to generate a binary mask, isolating regions corresponding to NP emission.

3. The binary mask was applied to the original image, enabling quantification of emission intensity and total NP emission per frame, expressed as the ratio of gold-positive pixels to total frame pixels.

4. The binary mask was further processed to extract metrics, including area, emission intensity, aggregate size, and other NP related parameters.

5. To facilitate visualization, the binary image was assigned an arbitrary false color and either overlaid onto the original image or displayed independently to illustrate NP distribution.

Brightfield Quantification.

Brightfield analysis of tissue sections was performed using QuPath software (version 0.4.4). Gold NPs were identified through a random trees (Rtrees) pixel classifier, trained on manually defined regions of interest (ROIs) within high-dose sections where gold aggregates were clearly visible. The classifier was applied uniformly across all experimental groups. Results from brightfield image analysis were normalized within each group and compared to normalized ICP−MS data obtained from the same animal.

SEM−EDS and MP Co-localization.

For correlative imaging, SEM coupled with EDS was used to identify gold NP and compare them to MP histology results. Binary masks derived from EDS imaging were coregistered with the binary masks generated from MP images. Corresponding intensities between the two imaging modalities were calculated from the binary masks, normalized, and compared across experimental groups.

MPL−EDS ROC Analysis.

To quantitatively assess the accuracy of MPL-based segmentation, we performed a pixel-wise receiver operating characteristic (ROC) analysis using SEM−EDS as the gold standard. A representative tissue region containing visible AuNP clusters in both modalities was selected for analysis. The raw MPL and EDS images were FIJI (ImageJ) and individually normalized and Gaussian blur was applied.

A binary ground truth mask was generated from the EDS image using a fixed intensity threshold applied after normalization. The MPL image was then thresholded across the full intensity range in 1-unit steps (0–255) after segmentation. At each threshold, a binary prediction mask was created and compared to the EDS ground truth to compute the true positive rate (TPR) and false positive rate (FPR). This ROC curve was constructed without spatial binning to preserve the pixel-level resolution of the data and reflect the system’s ability to localize nanoscale AuNP signals. The area under the curve (AUC) was calculated using the trapezoidal rule and reported as a threshold-independent summary metric of classification performance.

Data Normalization, Image Processing and Statistical Analysis.

All quantitative data were normalized within experimental groups and compared across imaging modalities including MP, brightfield, SEM−EDS, and ICP−MS. Statistical analyses and additional data processing were conducted in Microsoft Excel. Spearman’s correlation was used for comparing EDS and MPL emission localization along with coefficient of determination (R²). Where mentioned, statistical significance is given at a 95% confidence level (p < 0.05). Error bars are given as ± the standard error of the mean where present. Determination of statistically significant differences between ICP−MS and MPL imaging was performed with the Wilcoxon signed-rank test. Determination of statistically significant differences between AuNP aggregates identified in MPL and brightfield images at 1 year was done with the χ-square test.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsnano.5c05069.

Video S1 depicting zoom in of MPL map from trimmed sample of a whole liver lobe prepared for SEM−EDS. Mouse was injected with 10 nM AuNPs 1 year post IV injection. End of the video shows region of interest used for correlative imaging between MPL and SEM−EDS analysis (MP4)

Video S2 depicting zoom in of MPL map of whole liver lobe. Mouse was injected with 10 nM AuNPs 1 year post IV injection. Liver shown here is the same as depicted in Figure 4D. As we zoom in, you can visualize a pocket of sequestered AuNPs in the liver. This demonstrates how we can easily image a whole liver lobe tissue section and then zoom in to see subcellular detail in any region across the whole organ (MP4)

Nanoparticle characterization, histology imaging, SEM and EDS data, high-resolution MPL images, 1-year study design and corresponding data, histopathological analysis of the 1-year study, descriptions of supporting videos (PDF)

ABBREVIATIONS

NP

nanoparticle

MPL

multiphoton luminescence

AuNP

gold nanoparticle

FFPE

formalin-fixed paraffin-embedded

IHC

immunohistochemistry

H&E

hematoxylin and eosin

TUNEL

terminal deoxynucleotidyl transferase dUTP nick end labeling

PV

portal vein

BD

bile duct

HA

hepatic artery

FOV

field of view

EDS

energy-dispersive X-ray spectroscopy

NCI

National Cancer Institute

NCL

Nano Characterization Lab

OCT

optimal cutting temperature

ICP−MS

inductively coupled plasma mass spectrometry

EM

electron microscopy

SEM

scanning electron microscopy

SHG

second-harmonic generation

HMDS

hexamethyldisilazane

IV

intravenous

DAPI

4′,6-diamidino-2-phenylindole

Cy5

cyanine5

TRITC

tetramethylrhodamine

FITC

fluorescein isothiocyanate

Data Availability Statement

Raw data available upon reasonable request.