Multiplexed imaging across spatial scales is crucial for understanding complex biological systems. Current methods, such as electron microscopy, MALDI imaging, and fluorescence microscopy, each have limitations in speed, depth, resolution, multiplexing capacity, and/or destructiveness. Electron microscopy offers high resolution but low speed and multiplexing; MALDI imaging provides high-throughput analysis but suffers from low resolution and sensitivity; and fluorescent microscopy is limited by spectral overlap, autofluorescence, and the diffraction limit, often requiring destructive iterative staining or complex barcode strategies. While mass cytometry (IMC) and multiplexed ion beam imaging (MIBI) offer improved multiplexing, they are mainly limited to two-dimensional imaging of surfaces and are destructive. X-ray microscopy, using tissue-penetrant X-rays, offers a potential solution, enabling nondestructive imaging at various resolutions. This study proposes MEZ-XRF, leveraging the sensitivity of XRF to detect different elements as molecular tags to achieve highly multiplexed, multiscale, nondestructive bioimaging.

Existing bioimaging techniques face trade-offs between speed, resolution, depth, and multiplexing. Iterative staining and nucleotide-based barcoding approaches have improved multiplexing in fluorescence microscopy but are slow and destructive, respectively. Mass cytometry (IMC) and multiplexed ion beam imaging (MIBI) utilize mass tags for high multiplexing, but they are limited to two-dimensional imaging and are destructive. Spatially encoded nucleotide tags, while achieving high-plex levels, also present similar limitations. Previous work demonstrated XRF detection of single molecular markers labeled with gold nanoparticles or cadmium-containing quantum dots but lacked the multiplexing capability achieved here.

MEZ-XRF employs chelating polymers conjugated to antibodies as XRF-detectable Z-tags. Formalin-fixed, paraffin-embedded (FFPE) tissue sections are stained with a panel of Z-tag-labeled antibodies. Imaging is performed using a focused X-ray beam, with the emitted X-rays characteristic of each element detected to create multichannel images. Different detectors (silicon drift detector (SDD) and germanium (GeCMOS) detectors) were used and their performance, including sensitivity and detection limits, is reported. The method's sensitivity was compared directly to IMC using multielement gelatin standards. The study utilizes cell lines (MCF10a, ZR-75-1, SKBR3, A431) with distinct marker expression profiles, manipulating some cell lines to achieve differential marker expression, to validate the multi-element Z-tag and imaging approach. For tissue-based validation, biopsies of HER2+, luminal A, and luminal B HER2+ breast tumors were imaged. To improve detection and speed, the researchers incorporated SABER-amplified Z-tags, which amplify the number of Z-tags for each target. Image analysis involved cell segmentation using Mesmer, a deep learning-based model, to quantify marker intensities and perform subcellular localization analysis. Leiden clustering was used for phenotypic classification and comparisons to IMC.

MEZ-XRF successfully imaged 20 Z-tags in parallel with subcellular resolution in cell lines and human tissues (breast tumor, tonsil, appendix). Lower Z-number elements showed greater signal-to-noise ratios. MEZ-XRF accurately recapitulated the expected marker distributions and subcellular localization, showing comparable results to IMC. Quantitative analysis using single-cell segmentation showed highly correlated intensities between different Z-tags targeting the same marker. Leiden clustering effectively separated cell lines and identified distinct cell subpopulations. The nondestructive nature of MEZ-XRF enabled rapid overview scans followed by high-resolution imaging of ROIs, allowing for multiscale analysis. SABER amplification significantly improved sensitivity, allowing detection of low-abundance markers like immune checkpoint proteins (PD1, CTLA4). With SABERx2 amplification and the GeCMOS2 detector, MEZ-XRF achieved imaging speeds up to 1.5 kHz, surpassing the speeds of existing technologies such as IMC and MIBI. Multiscale correlative imaging using MEZ-XRF combined with H&E staining demonstrated the feasibility of using low-resolution MEZ-XRF overviews to guide ROI selection for high-resolution analysis. In direct comparison, the signal-to-noise achieved using 10Hz MEZ-XRF with the GeCMOS2 detector was equivalent to 200Hz IMC. This speed improvement was primarily attributed to the elimination of the sample ablation step and the superior performance of the germanium detector.

MEZ-XRF addresses the limitations of existing multiplexed imaging techniques by offering high multiplexing, high speed, multiscale resolution, and nondestructive imaging across tissue-to-subcellular levels. The ability to perform rapid overview scans to guide subsequent high-resolution imaging is particularly advantageous for efficient use of available tissue samples. The nondestructive nature of MEZ-XRF facilitates correlative imaging with other techniques such as H&E staining. The high speed achieved with SABER amplification allows for high throughput analysis and the detection of low-abundance markers, expanding the scope of possible studies. The use of Z-tags offers advantages over fluorophores, such as improved multiplexing capabilities and the absence of photobleaching or degradation, enabling repeated measurements on the same sample. The high sensitivity, particularly when combined with the GeCMOS2 detector and SABER, offers a substantial improvement over similar technologies such as IMC.

MEZ-XRF represents a significant advancement in multiplexed bioimaging, offering a powerful and versatile platform for multiscale analysis of biological samples. The high speed, high multiplexing, nondestructive nature, and multiscale imaging capability make it suitable for a wide range of applications in biological research, including immuno-oncology. Future research directions include further improvements in speed and sensitivity through advancements in X-ray sources and detectors, exploring the use of miniaturized Z-tags for achieving molecular resolution, and developing 3D imaging capabilities.

While MEZ-XRF demonstrates superior performance in several aspects compared to existing techniques, some limitations remain. The technology currently requires access to specialized high-energy beamlines at synchrotron facilities; however, alternative laboratory-based X-ray sources are being developed. The detection limits vary between elements and detectors, impacting the choice of Z-tags for specific applications. Background levels in some MEZ-XRF images are higher than in IMC images due to Compton scatter, however this is accounted for in the data processing. Finally, achieving comparable speed to IMC will require software updates to reduce software latency in reading full spectra from the detectors.