CODEX Multiplex Imaging for Bacterial Spatial Mapping in Breast Cancer Tumors

Fresh human breast cancer tissues were aseptically collected, transferred to DMEM containing 10 μmol DARMR bacterial probe, and incubated at 37°C for 2 h. Tissues were fixed with 4% PFA for 24 h, dehydrated, paraffin-embedded, and sectioned into 5-μm slices on Poly-L-lysine slides (stored at 4°C after air-drying).The 18-plex antibody panel included CD3 (Akoya #4550119), CD8 (Akoya #4250012), PanCK (Akoya #4450020), and immune markers (full list in Table S1), conjugated to Alexa Fluor 647/750/Atto 550 and stored in Akoya Storage Buffer.Slides were baked at 55°C for 1 h, dewaxed with Histochoice (VWR #H103-4L), and antigen-retrieved using Tris-EDTA (pH 9, Solarbio #C1038) in a pressure cooker. Antibody cocktails (1:200 dilution) were applied overnight at 4°C (first round) and 3 h at RT (second round), followed by fixation with 1.6% PFA and methanol.Slides were imaged on a PhenoCycler-Fusion (Akoya) with DAPI (1 ms), AF647 (150 ms), and AF750 (150 ms) channels. Bacterial localization was analyzed by U-Net model-based single-cell segmentation of DAPI-stained nuclei, followed by quantification of DARMR probe fluorescence (excitation/emission 488/660 nm) in QuPath with positive signals defined as mean intensity > 3× background. Bacterial cells were co-registered with immune cells (CD3⁺/CD8⁺) and tumor cells (PanCK⁺) using Weighted Unifrac distance for spatial correlation analysis, while spatial clustering of bacteria and immune cells was tested by Ripley’s K-function (p < 0.05). Data visualization combined UMAP for dimensionality reduction and heatmaps for marker co-expression.