Using mass spectrometry-based proteomics to power spatial biology for deep proteomic profiling

Significance of the topic

Spatial proteomics integrates high-plex imaging with mass spectrometry to map protein expression across tissue sections while retaining cellular and microenvironment context. This capability is critical for disciplines such as oncology, neurobiology and developmental biology where local cell states, rare phenotypes and cell–cell interactions determine function and disease progression. Methods that combine unbiased, deep proteome profiling with precise spatial annotation enable discovery of unexpected cell states, biomarkers and therapeutic targets that are inaccessible to imaging-only or bulk proteomic approaches.

Objectives and study overview

The study presents mxDVP (multiplexed Deep Visual Proteomics), a modular pipeline developed to (1) perform highly multiplexed immunofluorescence imaging to identify and phenotype cells in situ, (2) transfer precise contours to laser microdissection for physical isolation, and (3) generate ultra-deep LC–MS/MS proteomes from ultra-low input samples. The workflow was applied to human pancreatic islets to characterize cellular heterogeneity, discover rare polyhormonal and progenitor-like endocrine cells, and link spatial phenotypes to unbiased proteomic signatures.

Methodology and workflow

The mxDVP pipeline comprises seven coordinated steps:

• Design and validation of a customized, high-plex antibody panel to resolve cell types and states.
• Acquisition of multiplexed immunofluorescence images on the same FFPE tissue sections used for proteomics.
• Interactive and automated image analysis using in-house software (PIPΣX and BIAS) and visualization tools (TissUUmaps) to segment, cluster and annotate cells.
• Polychromatic staining (Toluidine Blue) to aid high-precision contour transfer for laser microdissection.
• Laser microdissection (LMD) to isolate pooled contours corresponding to selected cell populations.
• Ultra-sensitive LC–MS/MS proteomic profiling (Orbitrap Astral MS) of pooled low-input samples.
• Integrated bioinformatic analysis linking imaging-derived phenotypes to proteomic data for spatially resolved discovery.

Key practical optimizations included testing pooling strategies (optimal pooling identified as ~250 contours ≈ 100 cell equivalents), stringent post-imaging washes to avoid antibody or buffer interference in MS, and iterative contour refinement to maximize purity of isolated populations.

Used instrumentation

• High-plex immunofluorescence imaging platforms (multiplex staining cycles, CODEX-style multiplexing implied).
• Laser microdissection microscope with precise contour transfer capability.
• Thermo Scientific Orbitrap Astral Mass Spectrometer for ultra-high sensitivity LC–MS/MS acquisition.
• PPS membrane slides for tissue handling (preferred over PEN), with Toluidine Blue polychromatic staining to assist contour visualization.
• Image analysis and integration tools: PIPΣX pipeline, BIAS integration, and TissUUmaps Spot Inspector plugin.

Main results and discussion

The mxDVP workflow achieved the following notable outcomes:

• Deep proteome coverage: The Orbitrap Astral MS enabled identification of >6,000 proteins from approximately 100 cell equivalents, and robust profiling across ~3,600 cell subpopulations isolated from human pancreatic tissue.
• Revealed cellular heterogeneity: Multiplexed imaging combined with MS uncovered rare polyhormonal endocrine cells co-expressing markers (e.g., CHGA, GCG, NPDC1, SST) and proteomic signatures suggestive of metabolic, mitochondrial and stress-adaptation programs—features not captured by antibody panels alone.
• High reproducibility: Quantitative metrics showed coefficients of variation typically below 10%, protein ID overlap >85% between replicates, and Pearson correlation coefficients >0.95, demonstrating both identification and quantitation robustness.
• Effective contamination control: Optimized washing and protocol adjustments minimized residual antibody or imaging-buffer signal in downstream MS, with control experiments indicating negligible impact on proteome quality.
• Spatially informed discoveries: Integration of imaging and MS data revealed spatial relationships such as correlations between islet size, polyhormonal states and local vascularization—observations dependent on maintained tissue context.

These results illustrate that combining targeted spatial phenotyping with unbiased proteomics expands discovery space: imaging guides selection of biologically relevant cells while MS provides the depth to detect low-abundance transcription factors, signaling and metabolic proteins critical for understanding cell plasticity.

Benefits and practical applications

mxDVP provides several advantages for analytical and biomedical applications:

• Discovery of rare or transitional cell states in situ, enabling studies of tissue regeneration, disease initiation and heterogeneity-driven therapy resistance.
• Compatibility with archived FFPE material and multiplexed imaging workflows, facilitating retrospective clinical cohort analyses and biomarker discovery.
• Flexibility to tune the workflow for depth (maximize protein IDs from low-input pools) or spatial granularity (focus on neighborhood or single-cell mapping) depending on the research question.
• High quantitative reproducibility suitable for comparative studies, method validation and translational pipelines.

Future trends and applications

Anticipated evolutions and opportunities for spatial proteomics and mxDVP include:

• Increased throughput through faster, higher-sensitivity MS acquisition modes and further automation of LMD and contour export steps, enabling larger clinical cohorts and tissue banks to be interrogated.
• Integration with multi-omics modalities (spatial transcriptomics, metabolomics) for richer mechanistic models of tissue ecosystems and disease processes.
• Refined computational tools for automated contour selection, batch correction and spatial statistics to scale discovery and reduce human-in-the-loop bottlenecks.
• Clinical translation: targeted panels derived from deep spatial proteomes could inform diagnostics, prognostics and spatially guided therapeutic strategies.

Conclusion

mxDVP demonstrates that coupling high-plex spatial phenotyping with ultra-sensitive LC–MS/MS yields deep, spatially resolved proteomes from ultra-low inputs. The approach successfully identifies rare cell types and uncovers molecular programs linked to cell plasticity and microenvironmental context, with strong reproducibility metrics. As instrumentation and automation advance, workflows like mxDVP are poised to become scalable tools for translational research and biomarker discovery across complex tissues.

References

1. Mardamshina M., et al. Image Analysis with PIPΣX: An Integrated Automated Pipeline for Image Processing and EXploration for Diverse Tissue Types. bioRxiv 2025.05.04.652145. doi:10.1101/2025.05.04.652145
2. Mardamshina M., et al. Streamlining Multiplexed Tissue — (citation details incomplete in source; referenced in original material).