Intra-patient spatial comparison of non-metastatic and metastatic lymph nodes reveals a reduction in CD169⁺ macrophages within metastatic breast cancers

Summary

Background

Breast cancer cells suppress the host immune system to efficiently invade the lymph nodes; however, the underlying mechanism remains incompletely understood. Here, we aimed to comprehensively characterise the effects of breast cancers on immune cells in the lymph nodes.

Methods

We collected non-metastatic and metastatic lymph node samples from 6 patients with breast cancer with lymph node metastasis. We performed bulk transcriptomics, spatial transcriptomics, and imaging mass cytometry to analyse the obtained lymph nodes. Furthermore, we conducted histological analyses against a larger patient cohort (474 slices from 58 patients).

Findings

The comparison between paired lymph nodes with and without metastasis from the same patients demonstrated that the number of CD169⁺ lymph node sinus macrophages, an initiator of anti-cancer immunity, was reduced in metastatic lymph nodes (36.7 ± 21.1 vs 7.3 ± 7.0 cells/mm², p = 0.0087), whereas the numbers of other major immune cell types were unaltered. We also detected that the infiltration of CD169⁺ macrophages into metastasised cancer tissues differed by section location within tumours, suggesting that CD169⁺ macrophages were gradually decreased after anti-cancer reactions. Furthermore, CD169⁺ macrophage elimination was prevalent in major breast cancer subtypes and correlated with breast cancer staging (p = 0.022).

Interpretation

We concluded that lymph nodes with breast cancer metastases have fewer CD169⁺ macrophages, which may be detrimental to the activity of anti-cancer immunity.

Funding

JSPS KAKENHI (16H06279, 20H03451, 20H04842, 22H04925, 19K16770, and 21K15530, 24K02236), JSPS Fellows (JP22KJ1822), AMED (JP21ck0106698), JST FOREST (JPMJFR2062), Caravel, Co., Ltd, Japan Foundation for Applied Enzymology, and Sumitomo Pharma Co., Ltd. under SKIPS.

Research in context.

Evidence before this study

How cancer cells evade immune surveillance is a fundamental question. Lymph node metastasis is particularly intriguing, given that lymph nodes harbour many immune cells. Regulatory T cells have been a major focus in this area of research, but our knowledge regarding other immune cell types that cancer cells target to achieve metastasis remains limited.

Added value of this study

We thoroughly characterised lymph nodes from patients with breast cancer using multi-scale transcriptome analyses. The essence of our approach is to directly compare non-metastatic and metastatic lymph nodes within the same patients. This comparison allows us to examine the effects of metastasised breast cancer cells on immune cells in lymph nodes, subtracting the effects of primary breast cancer cells. Our relatively unbiased approach led to a finding that CD169⁺ macrophages, a critical part of anti-cancer immunity, are selectively reduced in metastatic lymph nodes. In contrast, other major immune cell types, such as regulatory T cells, were not significantly affected in our patient cohort, establishing this particular macrophage subtype as important in understanding how breast cancer cells evade immune surveillance.

Implications of all the available evidence

We expect that CD169⁺ macrophages will be measured to predict the efficacy of anti-cancer immunotherapy. Although anti-cancer immunotherapy is very effective, it also has unwanted side effects, as is the case for other anti-cancer therapies. In this sense, predicting the efficacy of anti-cancer immunotherapy is a critical medical need. Moreover, leveraging CD169⁺ macrophages in patients may further enhance the strength of anti-cancer immunotherapy.

Introduction

Cancers affect host cells in various ways to ensure their survival and evade the host immune system.1, 2, 3, 4 Cancer cells directly contact host immune cells via cell surface molecules, inhibiting immune cell activity.^5,6 Cancer cells also secrete soluble factors, including cytokines, that influence nearby cells.^7,8 Cancer-derived factors can have distant effects when they enter the blood and lymphatic systems.9, 10, 11 Such direct, proximal, and distal interactions between cancer and host cells eventually disrupt host homeostasis.^4,9,12, 13, 14, 15 These cancer–host interactions are even more complex in the presence of more than two cancer tissues (e.g., primary and metastasised cancer tissues) within the same body.¹⁶

The lymph nodes represent the main site of action of such complex cancer–host interactions.¹⁷ The interaction between cancer cells and lymph nodes is of particular interest because metastasis to the lymph nodes requires the efficient suppression of host immune cells.¹⁷ Previous studies demonstrate immune cell suppression in the lymph nodes of cancer patients, as exemplified by the increase in the regulatory T cell (T_reg) population.^18,19 Given the immunosuppressive function of T_reg cells, this is a reasonable mode of immune suppression by cancer cells.^18,19 However, it remains incompletely understood whether other immune cell types are affected in the lymph nodes of cancer-bearing organisms. Moreover, distinguishing the direct, proximal, and long-range effects of primary and secondary cancer tissues on host immune cells in the lymph nodes is challenging. Addressing these issues requires a comprehensive dissection of cancer–host interactions within the lymph nodes, which is critical for precisely understanding the mechanisms by which cancer cells colonise this immune cell-rich organs of the body.

CD169⁺ macrophages are a unique type of resident macrophages in the lymphoid organs that contribute to anti-cancer immunity.20, 21, 22, 23 CD169⁺ macrophages phagocytose dead cancer cells and present cancer-derived antigens to CD8⁺ T cells.²⁰ Animal studies have revealed the significant contribution of CD169⁺ macrophages to the suppression of cancer growth and metastasis in vivo.²¹ Clinical studies have also supported the importance of CD169⁺ macrophages; the number of CD169⁺ macrophages in the lymph nodes is positively correlated with the prognosis of patients with various cancer types.24, 25, 26, 27, 28, 29

In the current study, we aimed to comprehensively characterise the effects of breast cancers on immune cells in the lymph nodes. In patients with breast cancer with lymph node metastasis, we analysed both metastatic and non-metastatic lymph node samples using multi-scale transcriptome analyses. The obtained datasets were used to capture the direct and proximal effects of metastasised cancer cells on host immune cells in a relatively unbiased manner, providing insight into CD169⁺ macrophages as targets of metastasised breast cancer cells in the lymph nodes. This study thus deepens our understanding of immunosuppression by metastasised breast cancer cells in the lymph nodes, establishing CD169⁺ macrophages as crucial therapeutic targets.

Methods

Clinical samples and tissue processing

Lymph nodes were collected from six female patients with breast cancer who underwent axillary dissection at Kyoto University Hospital (Kyoto, Japan). Sex was verified against Japanese health insurance cards. The clinical and pathological characteristics of the patients are summarised in Table 1. Lymph nodes were collected in the surgical room and cut in half. Enlarged lymph nodes were cut into one-quarter or one-eighth sections. The number of lymph nodes collected from each patient are summarised in Table 1. The collected tissues were embedded in OCT compound (Sakura Finetek Japan), frozen in liquid nitrogen, and immediately stored at −80 °C. The remaining lymph nodes were formalin-fixed and paraffin-embedded (FFPE). Tissue blocks were sectioned at 15 μm thickness using a Leica CM1950 cryostat microtome. Each section was mounted on polyethylene naphthalate membrane 4.0 μm slides (Leica). The sections were stained with haematoxylin and eosin (H&E). Among the metastatic lymph nodes, only those with clearly isolated metastatic masses were selected for transcriptome analyses. Lymphocyte regions were isolated using an LMD7000 laser micro-dissection system (Leica Microsystems) following the manufacturer’s protocol.

Table 1.

Patient characteristics and information of the collected lymph node samples.

	Patient 1	Patient 2	Patient 3	Patient 4	Patient 5	Patient 6
Age	54	41	45	76	69	68
Pathological T category	4b	2	2	2	2	2
Pathological N category	2a	3a	2a	1a	1a	3a
N status	4/14	12/20	4/12	2/13	2/15	11/16
Tumour subtype	Luminal	Luminal	Luminal	TNBC	Luminal	TNBC
Ki67 (%)	33.2	22	98.6	65	62	80.2
No. of collected lymph nodes
Metastatic	1	2	1	1	1	2
Non-metastatic	1	1	1	2	2	2
No. of laser micro-dissected sections
Metastatic	1	2	1	4	1	2
Non-metastatic	1	1	1	2	2	2

TNBC, triple-negative breast cancer.

Bulk RNA-sequencing

RNA was extracted using the RNeasy Micro Kit (QIAGEN) following the manufacturer’s protocol. Library preparation and sequencing were performed by Macrogen (Japan). Libraries were prepared using the SMART-Seq v4 Ultra Low Input RNA Kit and TrueSeq RNA Sample Prep Kit v2 following the manufacturer protocols (Library protocol SMART-Seq v4 Ultra Low Input RNA). Paired-end sequencing was performed using the Illumina NovaSeq 6000 platform with a 100-bp read length.

Transcriptome analysis

RNA-sequencing reads were mapped to the human reference genome h38/GRCh38 using Hisat2^30,31 and counted using featureCounts³² in Galaxy (https://usegalaxy.org/). Read counts were normalised using the transcripts per kilobase million (TPM) method, and the low-expression genes (TPM < 1) and non-protein-coding genes were removed. Gene expression matrices generated using the TPM scores are listed in Supplemental Table S1.

Spatial transcriptomics

FFPE samples were used for spatial transcriptomic analysis. Tissue slides were prepared using Visium CysAssist Spatial Gene Expression for FFPE, following the manufacturer’s protocol (CG000518, 10× Genomics). Deparaffinisation, H&E staining, and de-crosslinking were performed following the manufacturer’s protocol (CG000520, 10× Genomics). Libraries for Visium were prepared following the Visium Spatial Gene Expression User Guide (CG000495, 10× Genomics). Paired-end sequencing was performed using a NovaSeq 6000 system (Illumina), following the manufacturer’s protocol.

Spatial transcriptomics data analysis

Data processing: Spatial transcriptomics data, including UMI (Unique Molecular Identifier) counts and spot coordinates, were analysed using the R Seurat package (version 4.3.0).³³ The four images and their data were processed into Seurat objects using the Load10X_Spatial function and normalised using the NormalizeData function. Visual inspection revealed that each sample contained spots with low numbers of reads and detected genes or spots detached from the main part of the tissue slice. Spots were manually filtered from each slice. Ultimately, 12,625 spots remained. The data of the four samples were merged, and batch effects among the four samples were removed using Harmony (version 0.1.1, function RunHarmony).³⁴ The first 10 Harmony dimensions were used to conduct further dimensionality reduction (with the RunUMAP function) and spot clustering (with the FindNeighbors and FindClusters functions; resolution = 0.5) into 11 clusters. To interpret the cell-type composition of each spot, we inspected the expression patterns of known marker genes (Supplemental Table S2).

Analysis of biological pathways: We defined sets of genes associated with Gene Ontology (GO) biological process functional annotations using the R package msigdbr (version 7.5.1). We filtered out GO terms associated with <20 or >250 genes present in the data. For each of the remaining 3152 GO terms and their associated genes (Supplemental Table S2), we calculated module scores using Seurat’s AddModuleScore function in the spots of the spatial transcriptomic data.³³ In brief, module scores reflected the average expression of the set of genes within each spot or cell subtracted from the average expression of several control genes with similar average expression levels as the input set. AddModuleScore was used with default parameters. In addition, within each of the 11 spot clusters, we predicted differentially expressed genes between lymph nodes with and without metastasis (function FindMarkers). We performed this analysis for samples from each patient separately to avoid patient-specific biases.

Antibody panel design

Information on the antibodies, metals, and their concentrations is provided in Supplemental Table S3. Unlabelled antibodies were conjugated with metals using the Maxpar X8 Multimetal Labeling Kit-40 Rxn (Standard BioTools) following the manufacturer’s protocol. Antibodies were validated by the manufacturer. For those from Standard BioTools, clones were selected based on manufacturer data and existing literature and were tested by imaging mass cytometry to assess method comparability and compatibility with the staining protocol; positive and negative control tissues were stained with an imaging mass cytometry panel containing antibodies directed against proteins that co-localise and counter-localise with the antibody of interest; a verification report was generated for each antibody and the staining was evaluated by an independent pathologist. The CD169 (NB600-534; clone HSn7D2; Novus) antibody has been validated by the manufacturer for use in Western blot, flow cytometry, enzyme-linked immunosorbent assays (ELISA), functional assays, immunocytochemistry, and immunohistochemistry. For the pan-cytokeratin antibody (914,204; clone AE-1/AE-3; BioLegend), the manufacturer tested on one or more tissue type and compared against one or more benchmark antibodies of the same target, and ensured that (paraffin-embedded) immunohistochemistry antibodies met their signal-to-noise standards under multiple dilutions and other staining conditions.

Imaging mass cytometry

FFPE tissue samples were prepared at Kyoto University Hospital in Japan. FFPE section staining follows by modified manufacturer’s protocol (Fluidigm, PN 400322 A3). The slides were baked for >2 h until all visible wax was removed. The slides were then deparaffinised in fresh xylene and ethanol. The slides were then inserted into preheated antigen retrieval ethylenediaminetetraacetic acid (EDTA) buffer (pH 9.0) and incubated for 30 min. Following incubation, the slides were cooled to 70 °C in the antigen retrieval EDTA buffer. The slides were washed twice with distilled water for 5 min each, followed by washing in phosphate-buffered saline (PBS) for 10 min each. The slides were blocked with 0.5% bovine serum albumin/PBS for 45 min at room temperature and incubated overnight with the antibody cocktail at 4 °C in a hydration chamber. The slides were washed twice with 0.2% Triton X-100 in PBS for 5 min and twice with PBS for 5 min each. The tissue was stained with the intercalator Ir in PBS (1:400) for 30 min at room temperature, followed by washing with distilled water for 5 min. The slides were air-dried for 20 min at room temperature. Imaging mass cytometry was performed using a Hyperion Imaging System (Standard BioTools). Regions of interest (ROIs) were selected from at least three points, including the nearest part of the invading tumour. The ROI was selected at 1000 μm width and 1000 μm height. Imaging was performed according to the manufacturer’s protocol at a laser frequency of 200 Hz and power of 3 dB.

Images were exported as TIFF files (OME-TIFF 32-bit) in an MCD viewer (Standard BioTools) and loaded into HALO version 3.5.3577 (Indica Labs). Imaging analysis was performed using the HighPlex FL module (v4.2.5) following the manufacturer’s protocol. Positive thresholds for individual markers were determined according to published nuclear or cytoplasmic staining patterns.

Immunostaining and cell count

FFPE sections from the lymph nodes of patients with breast cancer were obtained from Kyoto University Hospital (Kyoto, Japan). Immunohistochemistry for CD169 (sc-53442; clone HSn7D2; Santa Cruz Biotechnology, Santa Cruz, CA, RRID: AB_2254730, diluted 1:25) and CD68 (Roche/Ventana, 790–2931, RRID: AB_2335972) were performed using the Ventana Discovery Ultra platform (Roche, Basel, Switzerland). Both CD169 and CD68 antibodies were validated by the manufacturers according to their internal standards. Antigen retrieval was performed using the CC1 reagent (Ventana) for 60 min. The primary antibody was incubated for 32 min, and detection was performed using UltraView Universal DAB Detection Kit (Ventana). CD169 positive staining was counted in whole sliced lymph nodes on the slide, excluding lymph nodes completely occupied by cancer, by an experienced pathologist blinded to patient information. Consistent with previous reports, most CD169⁺ macrophages existed in the lymph node sinuses.

Statistics

The sample size was determined based on feasibility. Statistical significance was determined using Wilcoxon signed rank test for paired samples and the Mann–Whitney U test to compare cell count data between groups. Statistical analyses were performed using GraphPad Prism Software (Prism 9). The Jonckheere–Terpstra test was used to evaluate the CD169 expression trend in cells according to pN and pT (SAS for Windows release 9.4; SAS Institute Inc., Cary, NC, USA). Statistical significance was determined at α = 0.05.

Ethics

Lymph nodes were collected from six patients with breast cancer who underwent axillary dissection at Kyoto University Hospital (Kyoto, Japan) under the institutional ethical guidelines and regulations/ethical principles of the Declaration of Helsinki. The study protocol was approved by the Institutional Ethical Committee (G0424-18) (Kyoto University Graduate School and Faculty of Medicine) and informed consent was obtained. FFPE sections from the lymph nodes of patients with breast cancer were obtained from Kyoto University Hospital (Kyoto, Japan) according to the protocol approved by the Institutional Ethical Committee (R2889-1) (Kyoto University Graduate School and Faculty of Medicine).

Role of funders

The funding organisations were not involved in the design, data collection, analysis, interpretation, or reporting of the study.

Results

Searching for lymph node genes affected by breast cancer metastasis

To determine the direct and proximal effects of metastatic breast cancer cells on host cells in the lymph nodes, we collected 20 laser micro-dissected sections from 17 lymph nodes of six patients with breast cancer who had lymph node metastasis without clinical evidence of distant organ metastasis (i.e., stages II–III; Fig. 1a and Table 1). Four patients had breast cancers expressing hormone receptor proteins (luminal-type breast cancer), and breast cancers of two patients did not express hormone receptor proteins and did not exhibit HER2 amplification (triple-negative breast cancer subtype). Each patient had at least one non-metastatic and one metastatic lymph node, allowing direct comparison between paired non-metastatic and metastatic lymph nodes in the same patient (Table 1). Given the long-range effects of primary breast cancers via the lymph,35, 36, 37, 38, 39 this comparison was considered to be useful for identifying the direct and proximal effects of cancer cells on host immune cells in the lymph nodes. In four cases, the patient had more than two non-metastatic and/or metastatic lymph nodes (Table 1). In these cases, we averaged the obtained gene expression data to normalise variations among lymph nodes from the same patient. As a result, we generated 12 gene expression matrices (six non-metastatic and six metastatic lymph nodes) (Supplemental Table S1).

Fig. 1.

Reduced expression of macrophage-related genes in lymph nodes with breast cancer metastasis.a. Experimental workflow. b. Heatmap of genes differentially expressed in the metastatic lymph nodes. Gene expression in metastatic lymph nodes was normalised to that in non-metastatic lymph nodes for each patient (log₂ fold change). The top 20 most downregulated genes are highlighted; n = 6. c. Expression of SIGLEC1 (RNA-seq). The average fold-change data normalised to non-metastatic lymph nodes are presented. The p-value was calculated using the Wilcoxon signed rank test; n = 6. d. Expression of macrophage-related genes in the lymph nodes (RNA-seq). Gene expression in the metastatic lymph nodes was normalised to that in non-metastatic lymph nodes for each patient (log₂ fold change); n = 6.

A total of 489 differentially expressed genes (p < 0.05, paired t-test) were identified between the paired non-metastatic and metastatic lymph nodes. Notably, most of the affected genes were downregulated in the metastatic samples (Fig. 1b). These differentially expressed genes included gene expression changes that were reported previously.³⁷ For example, the expression levels of Interleukin-7 (IL-7) was reduced in the metastatic lymph nodes (Supplemental Fig. S1a). This gene is a representative of cancer-dependent lymph node reprogramming,³⁷ which is considered to occur via long-range mechanisms. Thus, our data suggested that cancer cells further remodel the structure of the lymph nodes upon arrival.

A closer examination of the heatmap of differential gene expression revealed that genes expressed in macrophages were preferentially affected by the presence of lymph node metastasis. Notably, the expression level of sialic acid binding IG-like lectin1 (SIGLEC1; also known as CD169), the canonical marker of CD169⁺ macrophages,^23,40 was severely reduced in lymph nodes harbouring metastasis (Fig. 1c and d). In addition, reductions in macrophage receptor with collagenous structures (MARCO) and apoptosis inhibitor of macrophage (AIM; also known as CD5L) in metastatic lymph nodes were also notable (Fig. 1d and Supplemental Fig. S1b). Collectively, our transcriptome analyses indicated that the direct and proximal effects of metastatic breast cancer suppress a set of genes expressed in macrophages.

Spatial transcriptomics reveals the reduction of lymph node macrophages

We next investigated whether the identified changes in host genes reflected the quantity and quality of changes in immune cell populations using spatial transcriptome analysis. We selected two sets of non-metastatic and metastatic lymph node pairs from two patients (patients #4 and #6) to obtain four slices. After a quality check and batch-effect correction, 12,625 spots were obtained for further analysis. The datasets were subjected to dimensionality reduction using principal component analysis and Uniform Manifold Approximation and Projection (UMAP), revealing 10 clusters in the lymph nodes (Fig. 2a). These clusters were annotated according to known marker genes (Supplemental Table S2). We also calculated “module scores” (averaged expression levels of genes involved in the same biological processes; see Methods) using Seurat for sets of genes associated with 3152 Gene Ontology (GO) terms for annotation. These analyses identified B cells, T cells, macrophages, plasmacytoid dendritic cells (pDCs), endothelial cells, and cancer cells in our lymph node datasets (Fig. 2a and b and Supplemental Fig. S2a–f).

Fig. 2.

Spatial transcriptomics landscape of lymph nodes of breast cancer patients.a. UMAP plot of VISIUM spots of non-metastatic and metastatic lymph nodes. Two pairs of non-metastatic and metastatic lymph nodes from patients #4 and #6 are presented. b. Compartmentalization of lymph node cells. Each cell type was identified using a specific set of marker genes listed in Supplemental Table S2 (also see Supplemental Fig. S2). c. Gene Ontology (GO)-based annotation of lymph node cell types. VISIUM spots enriched for B cells, T cells, plasmacytoid dendritic cells (pDCs), and macrophage markers are presented. VISIUM slices for B cells, T cells, and macrophage markers in non-metastatic lymph nodes are also presented in Fig. 3d, and Supplemental Fig. 3a, and b respectively. d. Expression of EPCAM in UMAP plots. EPCAM-high spots are highlighted as metastatic breast cancer cells. e. Expression of EPCAM in the VISIUM slices from metastatic lymph nodes. EPCAM-high spots correspond to histologically annotated tumour areas. The upper slice of EPCAM expression is also presented in Supplemental Fig. S2g. f. Module scores for “ATP metabolic process” in the VISIUM slices from metastatic lymph nodes.

Consistent with the previously established structure of the lymph nodes,⁴¹ these different immune cell types were observed to be compartmentalised rather than uniformly distributed in the lymph nodes (Fig. 2b and c). Using two slices from non-metastatic lymph nodes as examples, B cell- and T cell-enriched spots demonstrated distinct patterns of localisation in an almost mutually exclusive manner (Fig. 2c). Although our spatial transcriptomic data were not obtained at the single-cell level, based on these results, each spot was expected to be enriched in specific cell types. Macrophage and pDC spots also exhibited a unique distribution and rarely overlapped with B cell and T cell-enriched spots (Fig. 2c).

Our analyses successfully captured metastasised breast cancer cells (Fig. 2d; cluster 10). The cluster corresponding to the metastasised breast cancer cells highly expressed EPCAM, an epithelial marker gene (Fig. 2e). These cancer cells were enriched for a GO term “ATP metabolic process,” suggestive of active energy production and usage (Fig. 2f). Spots surrounding cancer cells, which expressed NNMT at a high level, were expected to be metastasis-associated stroma as mentioned previously in ovarian cancers (Supplemental Fig. S2g).⁴² Furthermore, the metastasised lymph nodes had more than two foci of breast cancer cells within the sections (Fig. 2e). Thus, these lymph nodes appeared in the process of being occupied by metastasised breast cancer cells. Although similar spots were also detected in the non-metastatic lymph nodes, unlike those observed in the metastatic lymph nodes, these spots were sparsely distributed (Supplemental Fig. S2h), and they were not histologically considered to be metastasised cancer cells. In addition to EPICAM, KRT5 and KRT7, markers of triple-negative breast cancer cells,⁴³ were enriched in metastasised cancer cells (Supplemental Fig. S2i).

Comparing the non-metastatic and metastatic lymph nodes demonstrated that macrophages (cluster 1) represented the most strongly reduced population in breast cancer metastasis (Fig. 3a). The number of spots corresponding to cluster 1 was reduced by more than two-fold in metastatic lymph nodes compared with that in non-metastatic lymph nodes. This observation aligned with our bulk transcriptome data, demonstrating a decrease in a series of macrophage markers (Fig. 1b–d). Notably, SIGLEC1 was abundantly expressed in this macrophage cluster (Fig. 3b). Furthermore, spots with high expression of SIGLEC1 were localised in the lymph node sinuses (Fig. 3c), suggesting that these spots contained CD169⁺ lymph node sinus macrophages. Visualisation of whole macrophage-enriched spots and SIGLEC1 (CD169)-positive spots in the four slices demonstrated a severe reduction in SIGLEC1-positive spots in metastatic lymph nodes (Fig. 3d and e). We detected almost no SIGLEC1-positive spots near cancer tissues. In contrast, other cell types (e.g., B and T cells) were not strongly affected by breast cancer metastasis in these sections (Supplemental Fig. S3). These results suggested that macrophages, including CD169⁺ macrophages, in the lymph nodes are selectively affected by metastatic breast cancer cells.

Fig. 3.

Spatial transcriptomics reveals macrophage reduction in metastatic lymph nodes.a. UMAP plots of non-metastatic and metastatic lymph nodes. The fraction of macrophage-enriched spots is presented as the percentage. b. Expression of SIGLEC1 in the UMAP plot. Macrophage-rich spots (Cluster 1) are highlighted. c. Localization of SIGLEC1-high spots in the lymph nodes. Upper panel: haematoxylin-eosin (H&E) staining; lower panel: expression of SIGLEC1 in VISIUM slices. Arrowheads indicate lymph node sinuses. d. Macrophage-rich spots in non-metastatic and metastatic lymph nodes. Metastatic breast cancer cells are outlined. VISIUM slices for macrophage markers in non-metastatic lymph nodes are also presented in Fig. 2c. e. Expression of SIGLEC1 in non-metastatic and metastatic lymph nodes. Metastatic breast cancer cells are outlined.

CD169⁺ macrophages are select targets of metastasised breast cancers

We further tested if CD169⁺ macrophages are selectively reduced in metastatic lymph nodes using the Hyperion imaging system,⁴⁴ with which we can visualise cells of interest at the single-cell resolution. We used anti-CD20 antibody for B cells, anti-CD4 antibody for CD4⁺ T cells, anti-CD8 antibody for CD8⁺ T cells, and anti-FOXP3 antibody for T_reg cells, anti-CD68 antibody for total macrophages, anti-CD169 (SIGLEC1) antibody for CD169⁺ macrophages, anti-CD11c antibody for CD11c⁺ cells (e.g., dendritic cells), anti-α-smooth muscle actin (SMA) for myoepithelial cells, and anti-pan-cytokeratin for cancer cells (Fig. 4a and Supplemental Table S3: the list of antibodies used) against the same set of samples we used for spatial transcriptomics, enabling direct comparison. We selected 3 regions of interest (ROI) and counted the number of these cell types in each slice (Supplemental Fig. S4).

Fig. 4.

CD169⁺ macrophages are select targets of metastasised breast cancers.a. Representative imaging mass cytometry images of non-metastatic and metastatic lymph nodes from patient #6. Displayed channels are as follows: CD20 (B cells; cyan), CD4 (CD4⁺ T cells; red), CD8 (CD8⁺ T cells; yellow), FOXP3 (T_reg cells; pale purple), CD68 (macrophages; purple), CD169 (CD169⁺ macrophages; pink), CD11c (CD11c⁺ cells; green), pan-cytokeratin (cancer cells; white), DNA intercalator (nucleus; blue), and anti-α-smooth muscle actin (SMA) (myoepithelial cells; blue). DNA intercalators are displayed in each panel. The SMA staining is depicted in the merged images. Cell phenotypes were determined using HALO software. Images of non-metastatic and metastatic lymph nodes from patient #6 are also presented as ROI3 in Supplemental Fig. S4. b. The density of CD68⁺ macrophages in non-metastatic and metastatic lymph nodes. c. The density of CD169⁺ macrophages in non-metastatic and metastatic lymph nodes. d. The density of CD20⁺ B cells in non-metastatic and metastatic lymph nodes. e. The density of CD4⁺ T cells in non-metastatic and metastatic lymph nodes. f. The density of CD8⁺ T cells in non-metastatic and metastatic lymph nodes. g. The density of FOXP3⁺ T_reg cells in non-metastatic and metastatic lymph nodes. h. The density of CD11c⁺ cells in non-metastatic and metastatic lymph nodes. (b–h) Data were calculated from three regions of interest (ROIs) per section using HALO software (six ROIs per group). Data are presented as the mean ± SEM. The p values were calculated with the two-tailed Mann–Whitney U test.

As displayed in Fig. 4a, distinct immune cell types were successfully visualised. The compartmentalised localisation of these cell types was consistent with the spatial transcriptomic analysis (Fig. 2, Fig. 3). Importantly, our data demonstrated that the numbers of total macrophages (Fig. 4b) and CD169⁺ macrophages (Fig. 4c) were smaller in the metastatic lymph nodes than in the non-metastatic lymph nodes, implying that cancer cells might uproot CD169⁺ macrophages from the lymph nodes.

In contrast to macrophages, the numbers of B, T, T_reg, and CD11c⁺ cells were comparable in non-metastatic and metastatic lymph nodes (Fig. 4d–h). Furthermore, T_reg cells appeared enriched around metastasised cancer tissues, supporting previous findings of the T_reg-dependent suppression of anti-cancer immunity (Fig. 4a).^18,19 Given the unaltered number of T_reg cells in metastasised lymph nodes (Fig. 4g), these results indicate the migration of T_reg cells toward cancer tissues (Fig. 4a). We also observed that CD11c⁺ cells were enriched around the cancer tissues (Fig. 4a).

In addition, Hyperion analysis was performed on lymph node slices from three additional patients (patients #1–3 in Table 1). In this experiment, slices were different from those we performed bulk transcriptome analysis. Yet, we still observed the reduction of CD169⁺ macrophages in metastatic lymph nodes compared to non-metastatic lymph nodes (Supplemental Fig. S5). Among cell types examined, only CD169⁺ macrophages showed a consistent decrease in metastatic lymph nodes.

Collectively, we concluded that CD169⁺ macrophages were reduced in lymph nodes having breast cancer metastasis, which may lead to the suppression of the earlier step of anti-cancer immunity of the host. The examined patients had breast cancers with lymph node metastasis but without clinical evidence of distant organ metastasis (i.e., stage II to III). They had normal levels (i.e., the range of healthy subjects) of albumin and C-reactive protein (CRP), which are used to establish the patient’s systemic status (Supplemental Table S4).^45,46 Together with the relatively earlier breast cancer stages of the examined patients, their systemic statuses were not considered terminal. Based on these results, we speculate that cancer-dependent suppression of CD169⁺ macrophages precedes other cancer-dependent abnormalities of the host, such as systemic inflammation.

The elimination of CD169⁺ macrophages is a generalisable event in breast cancer

As described earlier, CD169⁺ macrophages play a critical role in anti-cancer immunity.20, 21, 22, 23 In this sense, it is unlikely that CD169⁺ macrophages are suppressed from the beginning without participating in anti-cancer immunity. It is more likely that CD169⁺ macrophages are overwhelmed by cancer cells after anti-cancer reactions, and consequently become less present in the lymph nodes. In other words, there must be a stepwise mechanism of cancer-dependent suppression of CD169⁺ macrophages. The mechanisms underlying our assumption are to be investigated in the future. Related to these assumptions, it is of note that our analyses investigated only snapshots of cancer-dependent elimination of CD169⁺ macrophages. In addition, our spatial analysis was limited to a small number of patients. These situations were not ideal for examining such a phased mechanism of elimination of CD169⁺ macrophages. Moreover, the limited number of samples prevented us from generalising our findings from spatial analyses. To address these issues, we analysed an additional 315 non-metastatic lymph nodes and 159 metastatic lymph nodes from 58 patients with breast cancer (Supplemental Table S5).

These experiments revealed several crucial findings. First, we confirmed that the number of CD169⁺ macrophages was reduced in the metastatic lymph nodes compared with that in the non-metastatic lymph nodes (Fig. 5a). Thirty-seven of the 159 metastatic lymph nodes revealed no detectable CD169⁺ macrophages, suggesting complete elimination of CD169⁺ macrophages in these samples. Most CD169⁺ macrophages existed in the lymph node sinuses, validating the important nature of this particular macrophage subtype (Fig. 5b). We also noted the consistent reduction of CD169⁺ macrophages in three major breast cancer subtypes (luminal, HER2-positive, and triple-negative breast cancers) (Fig. 5c). These results indicated that the elimination of CD169⁺ macrophages is a general phenomenon in patients with breast cancer.

Fig. 5.

The elimination of CD169⁺ macrophages is a generalisable event in patients with breast cancer.a. Density of CD169⁺ macrophages in non-metastatic and metastatic lymph nodes. Data are presented as the mean ± SEM. The p values were calculated using the Wilcoxon signed rank test. A total of 315 non-metastatic and 159 metastatic lymph nodes were obtained from 58 patients, averaged the number of CD169⁺ macrophages in metastatic and non-metastatic lymph nodes in the same patient and compared the number of CD169⁺ macrophages in the same patient. b. Representative immunohistochemical images of CD169 staining in non-metastatic lymph nodes. Dotted lines indicate the lymph node sinus. c. The density of CD169⁺ macrophages in non-metastatic and metastatic lymph nodes from luminal, HER2-positive, and triple-negative breast cancers. Data are presented as the mean ± SEM. The p values were calculated with the Wilcoxon signed rank test. A total of 180 non-metastatic and 100 metastatic lymph nodes were obtained from 36 patients with luminal breast cancer; 59 non-metastatic and 32 metastatic lymph nodes were obtained from 10 patients with HER2-positive breast cancer; and 76 non-metastatic and 27 metastatic lymph nodes were obtained from 12 patients with triple-negative breast cancer (TNBC), averaged the number of CD169⁺ macrophages in metastatic and non-metastatic lymph nodes in the same patient and compared the number of CD169⁺ macrophages in the same patient. See also Supplemental Table S5 for patient characteristics. d. The density of CD169⁺ macrophages in metastatic lymph nodes and pathological N classification. pN1 refers to 1–3 metastatic lymph nodes, pN2 refers to 4–9 metastatic lymph nodes, and pN3 refers to ≥10 metastatic lymph nodes. Data are presented as the mean ± SEM. The p values were calculated with the Jonckheere–Terpstra test. A total of 51, 62, and 46 metastatic lymph nodes were identified from 29, 18, and 11 patients with pN1, pN2, and pN3 disease, respectively. e. Graphical summary of the proposed mechanism. LN, lymph node.

We next investigated the correlation between the number of CD169⁺ macrophages and pN classification for breast cancer staging (pN1 = 1–3 metastatic lymph nodes; pN2 = 4–9 metastatic lymph nodes; and pN3 ≥ 10 metastatic lymph nodes).⁴⁷ Our data demonstrated a gradual decrease in CD169⁺ macrophages correlating with the pN classification (Fig. 5d). These results suggested that when a patient has more metastatic lymph nodes, more CD169⁺ macrophages are eliminated from each metastatic lymph node. However, pathological tumour size classification (pT) and volume of the metastasised cancer did not correlate with the number of CD169⁺ macrophages in the lymph nodes (Supplemental Fig. S6a and b). Furthermore, our slides included metastatic lymph nodes in which CD169⁺ macrophages accumulated in metastasised cancer tissues, despite being rare (15 of 159 investigated lymph nodes and 7.67 ± 14.1 cells/tumour in the 15 lymph nodes; Supplemental Fig. S6c). This was in contrast to our spatial transcriptome analyses, in which CD169⁺ macrophages were observed to be completely depleted around cancer tissues (Fig. 3e). These differences may reflect which stage of cancer-dependent suppression of CD169⁺ macrophages was captured in each lymph node section. In summary, our data strongly indicated that CD169⁺ macrophages were consistently decreased from the metastatic lymph nodes of breast cancer patients, and that our experiments detected various different phases of CD169⁺ macrophage reduction (Fig. 5e).

We could not exclude the possibility that metastatic breast cancers reduced the expression of CD169 in otherwise CD169⁺ macrophages. Yet, based on the reduction of CD68⁺ positive cells (Supplemental Fig. S6d), our datasets strongly indicate the reduction in the number of CD169⁺ macrophages in metastatic lymph nodes.

Discussion

The effective suppression of host anti-cancer immunity is vital for cancer cells to survive in the host environment. Cancer suppresses various immune cell types in the lymph node via remote, proximal, and direct mechanisms.^17,35,38,39 The list of immune cell types suppressed by cancers in the lymph nodes remains incomplete. Moreover, in many cases, the precise mechanisms through which cancer suppresses immune cells and the invasion stage in which cancer cells achieve this suppression are unclear.

The current study establishes CD169⁺ macrophages as a target for proximal and/or direct suppression by cancers in the lymph nodes. Other lymph node cell types suppressed in cancer patients and mouse cancer models include CD4⁺ T cells, CD8⁺ T cells, and innate immune cells.^3,17, 18, 19^,35,38,39,48, 49, 50, 51, 52, 53, 54 T_reg cells are an important cell type involved in the suppression of CD4⁺ and CD8⁺ T cells in the lymph nodes.^18,19,38,39 The elimination of CD169⁺ macrophages we identified in this study can be an additional mechanism leading to inability of T cells to combat cancer cells. In addition, cancer cells can remotely condition lymphatic vessels (i.e., non-immune cell type) to modulate the immune system via, for example, C–C motif chemokine 5 (CCL5).35, 36, 37^,55 Such conditioning may contribute to the elimination of CD169⁺ macrophages from the lymph nodes. Testing this possibility requires additional studies that compare lymph nodes from, for example, non-invasive breast cancer patients and non-metastatic lymph nodes from patients having lymph node metastasis.

We identified CD169⁺ macrophages as “early” targets of cancer cells in two aspects. First, CD169⁺ macrophages phagocytose cancer-derived antigens and present them to CD8⁺ T cells.^20,23 This process represents an early step in anti-cancer immunity, the failure of which corrupts broader anti-cancer immunity processes.20, 21, 22 Metastasised breast cancers inhibit this early process of anti-cancer immunity. Second, our data indicate that the elimination of CD169⁺ macrophages precedes other reported immune cell abnormalities, such as abnormalities in the altered number of immune cells^18,19,35,48 and systemic inflammation defined by CRP (Supplemental Table S4).^45,46 Thus, there is a possibility that the elimination of CD169⁺ macrophages might be an early requirement for metastasised cancer cells to achieve sufficient immunosuppression, given the critical role of CD169⁺ macrophages in anti-cancer immunity.20, 21, 22, 23 This leads to the next important question, that of the underlying mechanism exploited by cancer cells to achieve this immune-evading effect.

The mechanism underlying the elimination of CD169⁺ macrophages from the lymph nodes is currently unclear, which is a major limitation of this study. However, our datasets successfully captured different stages of CD169⁺ macrophage suppression in patients (Fig. 5 and Supplemental Fig. S6c). According to these results, we propose a phased mechanism (i.e., sequence) of CD169⁺ macrophage suppression in the lymph nodes of breast cancer patients (Fig. 5e). Antigens from primary cancer tissues transported via the lymph may activate CD169⁺ macrophages, triggering anti-cancer immunity.²⁰ Cancer cells may subsequently break this defence, invading lymph nodes. Upon cancer cell arrival, CD169⁺ macrophages show some anti-cancer activity against metastasised cancer cells (Supplemental Fig. S6c). However, metastasised cancer cells can directly and/or indirectly expel CD169⁺ macrophages from the lymph nodes, achieving down-regulation of anti-cancer immunity. Such immunosuppression may allow expansion of metastasised breast cancer cells in the lymph nodes (Fig. 5d) and may cause other adverse effects, including lowered efficiency of anti-cancer immunotherapy. Addressing the exact mechanisms behind CD169⁺ macrophage suppression in the lymph nodes is our next challenge.

We also suggest that the conflict between breast cancer cells and CD169⁺ macrophages in the lymph nodes is of significant pathophysiological importance. This is strongly supported by the correlation between breast cancer staging and the number of CD169⁺ macrophages in the metastatic lymph nodes (Fig. 5d). Given these implications, we envision that retaining CD169⁺ macrophages in the lymph nodes in patients may be a way to increase the efficiency of anti-cancer immunotherapy in the future. Our samples are currently obtained from a relatively small number of patients in one country (Japan), which is a limitation of our study.

In summary, the current study identified reduction of CD169⁺ macrophages as the most prominent pathological phenotype in lymph nodes with breast cancer metastasis, establishing this reduction as a critical therapeutic target in the future.

Contributors

K.K. conceived and supervised the study, provided patient samples and clinical information, designed experiments, analysed data, constructed figures, and revised and edited the manuscript. S.K. supervised the study, designed experiments, analysed data, constructed figures, and wrote the manuscript. Y.M. collected patients’ samples, performed experiments, analysed data, constructed figures, and wrote the manuscript. T.R.K. provided patient samples, contributed to the experimental design, and performed the immunohistochemistry evaluation. A.V. performed spatial transcriptome analyses and constructed figures. M.H. prepared tissue samples for spatial transcriptomic experiments. Y.S. performed spatial transcriptomic experiments. Y.T. and H.H. provided patient samples and contributed to the experimental design. Y.F. collected patient samples and performed experiments. M.K. and M.Ta. provided patient samples. Y.I. and S.M. conducted the statistical analysis. M.To. conceived and supervised the study. All authors provided intellectual input and reviewed the paper. Y.M. and K.K. have directly accessed and verified the underlying data reported in the manuscript. K.K. and S.K. were responsible for the decision to submit the manuscript.

Data sharing statement

The bulk RNA transcriptome datasets used in this study are available in DNA Databank of Japan (DDBJ) under the accession numbers of DRR483580-DRR483599. The Visium datasets used in this study are available under accession number in the JGAS000616. Other data will be provided upon request of the corresponding author.

Declaration of interests

KK: grants from TERUMO, Astellas, Eli Lilly, Kyoto Breast Cancer Research Network; consulting fee from Becton Dickinson Japan; honoraria from Eisai, Chugai, and Takeda; MK: grants from Japan Science and Technology Agency (JST), FOREST, Kyowa Kirin, Pfizer, Japan Society for the Promotion of Science (JSPS) KAKENHI, and Japan Society for Fluorescence Guided Surgery; honoraria from Pfizer, Chugai, Daiichi-Sankyo, Eisai, and Guardant Health AMEA. MTa: grants from AstraZeneca, Daiichi-Sankyo, Kyoto Breast Cancer Research Network, JBCRG, ABCSG, Yakult, Medbis, and IQVIA Japan; honoraria from AstraZeneca, Daiichi-Sankyo, Chugai, Taiho, Lilly, Pfizer, and MSD. MTo: grants from Chugai, Takeda, Pfizer, Taiho, JBCRG Assoc., KBCRN Assoc., Eisai, Eli Lilly, Daiichi-Sankyo, AstraZeneca, Astellas, Shimadzu, Yakult, Nippon Kayaku, AFI Technology, Luxonus, Shionogi, GL Science, and Sanwa Shurui; honoraria from Chugai, Takeda, Pfizer, Kyowa-Kirin, Taiho, Eisai, Daiichi-Sankyo, AstraZeneca, Eli Lilly, MSD, Exact Science, Novartis, Shimadzu, Yakult, Nippon Kayaku, Devicore Medical Japan, Sysmex; Advisory Board of Daiichi-Sankyo, Eli Lilly, BMS, Athenex Oncology, Bertis, Terumo, Kansai Medical Net; Board of Directors of JBCRG Assoc., KBCRN, NPO org. OOTR, and JBCS Assoc; Associate Editor of the British Journal of Cancer, Scientific Reports, Breast Cancer Research and Treatment, Cancer Science, Frontiers in Women’s Cancer, Asian Journal of Surgery, Asian Journal of Breast Surgery.

All remaining authors declare no conflicts of interest.