F1000Research F1000Research 2046-1402 F1000 Research Limited London, UK 10.12688/f1000research.145171.3 Research Article Articles Tumor-driven stromal reprogramming in the pre-metastatic lymph node

[version 3; peer review: 2 approved]

 Piquet Michelle Conceptualization Data Curation Formal Analysis Investigation Methodology Software Validation Visualization Writing – Original Draft Preparation Writing – Review & Editing 1 Ruddy David A Conceptualization Resources Supervision Writing – Review & Editing 1 Cremasco Viviana Conceptualization Resources Supervision Writing – Review & Editing 2 Chang Jonathan Conceptualization Investigation Methodology Project Administration Supervision Visualization Writing – Original Draft Preparation Writing – Review & Editing https://orcid.org/0000-0001-6894-7127 a 2 Oncology Innovative Targets and Technologies, Novartis, Cambridge, MA, 02139, USA Oncology Translational Research, Novartis, Cambridge, MA, 02139, USA jonathan.chang@novartis.com

Competing interests: All authors are employees of Novartis.

 7 10 2025 2024 13 223 2 10 2025 Copyright: © 2025 Piquet M et al. 2025 This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background

Metastatic dissemination is critically reliant on the formation of a receptive niche, a process which is thought to rely on signals derived from the primary tumor. Lymph nodes are continuously exposed to such signals through the flow of afferent lymph, allowing the potential reprograming of lymphoid tissue stroma in support of metastases or immunosuppression. The objective of this study was therefore to better characterize tumor-driven transcriptomic changes occurring to specific stromal populations within the tumor-draining lymph node.

 Methods

We utilize single cell RNA sequencing of dissociated LN tissue extracted from tumor-bearing and naïve mice to profile the reprograming of tissue stroma within the pre-metastatic lymph node.

 Results

Resulting data provides transcriptomic evidence of tumor-induced imprinting on marginal reticular cells (MRCs) and floor lymphatic endothelial cells (fLECs) populating the subcapsular sinus. These alterations appear to be unique to the tumor-draining LN and are not observed during inflammatory antigenic challenge. Notably, MRCs exhibit characteristics reminiscent of early desmoplastic CAF differentiation, fLECs engage distinct chemoattractant pathways thought to facilitate recruitment of circulating cancer cells, and both stromal populations exhibit signs of metabolic reprograming and immune-modulating potential.

 Conclusions

Cumulatively, these findings build upon existing literature describing pre-metastatic niche formation and offer several promising avenues for future exploration.

 Tumor Draining Lymph Node scRNAseq metastasis Stromal Cell Fibroblast Endothelial Cell pre-metastatic Niche The author(s) declared that no grants were involved in supporting this work. Amendments from Version 2

We have amended the section discussing caveats and study limitations to highlight additional considerations brought forward by the reviewers.  In this section, we acknowledged potential caveats with Madcam1 as a marker distinguishing MRC and BRC subsets.  We also acknowledge limitations in the identification of metastatic tumor cells through the limited use of PDPN, CD31, and PDL1 surface markers.  Additionally, we made minor alterations to a claim describing Ccl19/Ccl21 chemokine gradient patterning in the SCS.  Finally, we have corrected minor wording or spelling errors in the figure legends.

 Introduction

Accumulation of genetic changes underpinning malignant transformation is the trigger point of cancer initiation. 1 , 2 However, these cell intrinsic alterations alone are insufficient for successful growth and progression of solid tumors. Rather, survival of the nascent tumor requires a co-evolved and permissive microenvironmental niche, wherein tissue stroma must be coopted to provide the structural, metabolic, and immune evasive support network necessary for tumor growth and eventual metastatic dissemination. 1 – 3 The therapeutic potential of exploiting this interdependence has spurred significant effort to unravel the complexities of the tumor microenvironment, though the processes by which normal tissues are rendered tumor-permissive remain incompletely understood.

Specific microenvironmental requirements likewise exist for the formation of secondary tumor growth, as the engraftment and survival of circulating cancer cells is equally reliant on the existence of a receptive niche. Reflective of these exigencies, fewer than 0.01% of circulating tumor cells are thought to successfully establish tumor metastases. 4 Moreover, the site of eventual metastatic engraftment is non-random, with specific organotropic patterns varying by cancer indication – an observation which serves as the basis of Stephen Paget’s longstanding “seed and soil” hypothesis. 5 – 7 Mounting evidence now suggests that successful metastasis additionally follows a reprograming of stromal elements at these distant sites, mediated by secretion of factors and extracellular vesicles originating from the primary tumor. 8 – 13 The resulting modifications form a more favorable microenvironment, termed the premetastatic niche (PMN). 9 Decoding these microenvironmental alterations and identifying the affected cell types may open a window to the requisite conditions for early tumor cell engraftment, growth, and immune evasion, and potentially reveal new points of therapeutic intervention. However, in-depth study of these processes may be experimentally challenging given variability in exact timing and location of metastases.

More predictable is the spread of cancer cells to the regional lymphatic lymph node (LN), which are frequently the first site of tumor metastasis in most forms of carcinoma. 14 This prevalence of metastatic spread to the draining LN is almost certainly a factor of direct exposure to tumor-derived signals, as such signals pool within the afferent lymphatics before invariably draining to the sentinel node. 15 – 19 Moreover, lymphangiogenesis and increased lymph flow is believed to be an essential step that precedes LN metastasis. 15 , 20 , 21 While exposure to afferent lymphatic flow facilitates the requisite acquisition of antigens and tissue signals for functional adaptive immunity, drainage of tumor-derived factors has the potential to modulate these responses, thereby conditioning the pre-metastatic LN in favor of tumor support and immune suppression. 22 Indeed, recent reports have demonstrated tumor-induced alterations to LN fibroblastic reticular cells (FRCs) – a stromal population which comprises the essential infrastructure supporting lymphocyte homeostasis and coordinating immune interactions. 23 – 25 Changes to FRCs of the tumor draining LN (tdLN) included increased activation and proliferation, disrupted homeostatic chemokine production driving aberrant immune cell localization, and altered matrix production. However, while the observed reprograming of LN stroma has a putative role in metastasis support and immune suppression, many of the described alterations overlap with the stromal response program associated with general inflammation and infection. 26 In this regard, the degree to which these previously reported findings are in fact unique or specific to tumor-derived factors requires further examination.

Here we expand on previous exploration of tumor-induced reprogramming through single cell transcriptomic analysis of stroma-enriched samples, enabling a more robust exploration of subset-specific transcriptional shifts. We demonstrated exquisite specificity in stromal reprograming associated with the premetastatic tdLN, wherein select subsets of FRCs and lymphatic endothelial cells (LECs) populating the LN subcapsular sinus (SCS) are differentially affected by tumor-derived signals. Amongst these findings, we confirm functional alterations reflective of matrix remodeling, identify engagement of distinct axes of chemotaxis, and identify transcriptional signs of altered metabolism and immunomodulatory potential. Importantly, these alterations were additionally compared to LN stromal responses following antigenic challenge in the absence of live tumor cells, allowing confirmation of the tumor-specificity of these response programs. Ultimately, we believe these findings help to broaden our understanding of pre-metastatic communication and tissue imprinting, and may direct future experimental exploration of metastatic niche dependencies with potential for therapeutic intervention.

 Methods Ethics statement

All animal work was approved and performed in accordance with the guidelines from the Institutional Care and Use Committee (IACUC) at Novartis Institutes for BioMedical Research (Protocol 20 IMO 035) and in compliance with the Guide for the Care and Use of Laboratory Animals. All efforts were made to ameliorate any potential suffering of animals.

 Mice

Experiments were performed in eight-week-old, sex-matched C57Bl/6 mice purchased from Charles River Laboratories. Mice were maintained under specific pathogen-free conditions in accordance with institutional and National Institute of Health guidelines.

 Cell lines

MC38 cells were received from NCI under MTA# 38699-15, expanded, frozen and collected in the NIBR cell line repository. Cell lines were maintained in Dulbecco’s Modified Eagle Medium (Gibco, 11965-092) containing 10% heat-inactivated FBS (VWR, 1500-500), 1% Penicillin-Streptomycin (Gibco, 15140-122), and 1% L-glutamine (Gibco, 25030-81) for one week prior to implant. Prior to inoculation into recipient animals, cell lines were tested and found to be free of mycoplasma and viral contamination in the IMPACT VIII PCR assay panel (IDEXX BioResearch, Missouri).

 Tumor models

The present study used tissues collected from mice bearing live MC38 tumors (n=3), or fixed MC38 tumor cells (n=3). Mice were randomly assigned to treatment groups. In preparation for inoculation, MC38 cells were detached using 0.25% trypsin, washed, and resuspended in sterile PBS prior to implantation. For the “fixed cell” control treatment condition, MC38 cells were additionally incubated in 4% paraformaldehyde in sterile PBS for 20 minutes at a concentration of 10 6 cells/mL before washing and resuspension in sterile PBS. For both conditions, 5 × 10 5 cells were implanted subcutaneously on the animal’s lower-right flank. Animals are anesthetized by isoflurane inhalation prior to treatment. Mice were monitored for tumor growth, changes in body weight and condition 3 times per week. Mice were euthanized by CO 2 inhalation on day 12 after treatment, and both the tumor-draining and contralateral non-draining inguinal lymph nodes were collected. No animals were excluded from analysis.

 Enzymatic digestion of lymph nodes

Single-cell suspensions of LNs were prepared for analysis by flow cytometry or single cell RNAseq. LNs were dissected and incubated at 37 °C in RPMI containing 0.1 mg ml −1 DNase I (Invitrogen), 0.2 mg ml −1 collagenase P (Roche) and 0.8 mg ml −1 dispase (Roche) for 50–60 min, as previously described. 27 Liberated cells in suspension were collected into cold medium containing 2% fetal bovine serum and 5 mM EDTA every 15–20 min and replaced with fresh digestion medium. Following complete digestion of the LN, cells were passed through a 70-μm cell strainer, washed and resuspended in cold PBS.

 Flow cytometric analysis

Single-cell suspensions were resuspended in flow cytometry buffer (PBS, 2% FBS, and 2 mmol/L EDTA). Cells were blocked with Fc block and then stained with different fluorochrome-conjugated antibodies for 20 min on ice. The following antibodies were used (clone, fluorophore, concentration): anti-CD45 (30-F11, BUV396, 1:500), anti-PD-L1 (10F.9G2, PE, 1:200), anti-CD31 (390, Pacific Blue, 1:300), and anti-PDPN (8.1.1, APC, 1:300). CountBright Absolute Counting Beads (Invitrogen, C36950) were included for quantification of cell counts. Samples were analyzed on a BD flow cytometry Fortessa instrument and analyzed with FlowJo. Raw data available. 28

Stromal enrichment

Single cells obtained from LN digestion were resuspended in mouse FC block (Miltenyi #130-092-575). Biotinylated PDPN (8.1.1, 1:100) antibody was used for positive selection, using the EasySep Selection Kit (Stemcell Technologies #18559). Cells were then washed in cold PBS, counted, and resuspended at 10 6 cells/mL for sequencing.

 Single-cell RNA sequencing and data pre-processing

The 10x Genomics Chromium Single Cell 3’ Reagents v3 kit (10× Genomics, Pleasanton, CA) was used under standard conditions and volumes to process cell suspensions for 3’ transcriptional profiling. The volumes of cell suspensions were calculated to achieve a target cell recovery of 6000 cells for all dissociated samples, following the manufacturer’s guidelines. The resultant purified cDNAs were quantified on an Agilent Tapestation (Agilent, Santa Clara, CA) using High Sensitivity D5000 ScreenTapes and Reagents. The final single cell 3’ libraries were quantified using an Agilent Tapestation using High Sensitivity D1000 ScreenTapes and Reagents. Following this, the libraries were diluted to a concentration of 10 nM in Qiagen Elution Buffer, subjected to denaturation, and loaded onto an Illumina MiSeq at 6 μM, utilizing the MiSeq Reagent Kit v3 (Illumina, San Diego, CA) to assess sample quality and achieve loading normalization for the HiSeq4000. The normalized libraries were loaded onto an Illumina cBOT at a concentration of 160 picomolar and sequenced on a HiSeq4000. The sequencing process included a 28-base-pair first read, followed by two 8-base-pair index reads, and concluded with a 91-base-pair second read, all by using 2 HiSeq4000 SBS kits at 50 cycles each. All sequence intensity files were generated using the Illumina Real Time Analysis software. The resulting intensity files were then subjected to demultiplexing and aligned to the mouse genome, version mm10, using the 10x Genomics CellRanger v3.0.1 software package.

 Single-cell RNA-Seq data quality control and processing

All computational analyses and visualizations were conducted within the R programming environment, using version 3.6.3. The Seurat package (v3.2.3) was employed for quality control and downstream analysis of single-cell RNA-seq data. 29 The count matrices were then subjected to comprehensive analysis using the Seurat workflow. Cells that exhibited low quality characteristics, defined as those with fewer than 200 expressed genes, more than 6000 expressed genes, or exceeding 30% mitochondrial content, were excluded from the dataset. Following this, the count matrices were merged and normalized using SCTransform. Principal component analysis was executed, focusing on 2000 highly variable genes. To select the appropriate number of principal components (PCs), the Elbow plot heuristic was used. A shared nearest neighbor (SNN) graph was constructed using the first 20 PCs through Seurat’s FindNeighbors function, and cell clustering was performed using a Louvain algorithm with FindClusters. A resolution parameter of 0.8 was chosen to generate an extensive collection of cell clusters, effectively capturing diverse cell types. Cluster markers were identified using Seurat’s FindAllMarkers function. Dimensionality reduction was visualized using the uniform manifold approximation and projection (UMAP) method. We assessed the quality control metrics of the dataset, including the number of RNA molecules (UMIs) and the number of genes detected per cell. All values fell within the expected range for single cells, as defined in Seurat’s documentation, and we did not observe any anomalies indicative of doublets.

 Stromal cluster identification and analysis

Cluster identification for our dataset was based on findings from previously published studies that have deeply characterized stromal subsets. 30 – 34 We aligned expression patterns of the markers validated in these studies with differential expression patterns of our identified clusters to assign subtypes. Notable markers used for cluster identification are noted below, and bubble plots for each fibroblastic or endothelial cluster are included in Supp Figure 1. Key identifiers for each subset include the following: cLEC: ackr4+, cd36+, edn1+, anxa2+. fLEC: lyve1+, glycam1+, coch+, madcam1+ Medullary LEC: lyve1+, marco+, il33+, itg2b+ Cortical LEC: mrc1+, ptx3+, reln+. Key identifiers for each subset include the following: TRC: ccl19+, ccl21+, grem1+ BRC: cxcl13+, cr2-, madcam1-, tnfsf11-, ch25h+ FDC: cxcl13+, mfge8+, madcam1+, coch+, cr2+, tmem119+, pthlh+ MRC: madcam1+, tnfsf11+, enpp2+ MedRC: cxcl12(hi), inmt+, stc1+, bst1- PRC: cd34+, fndc1+, pi16+, dpp4+ Activated SC: nr4a1+.

 Re-clustering and analysis of fLEC, MRC, BRC

The dataset was methodically divided into subsets, specifically isolating floor lymphatic endothelial cells (fLEC), marginal reticular cells (MRC), and b-zone reticular cells (BRC). These subsets underwent a re-clustering process using 15 principal components (PCs) at a resolution parameter of 0.3, with the optimal number of PCs determined using the Elbow heuristic.

 Differential gene expression and pathway enrichment analyses

Differential gene expression analyses were conducted on each subset of cells with an average log-fold change above 0.25, as part of the Seurat workflow. This analysis compared Live Draining vs. Live Non-Draining cells. Pathway enrichment analyses were executed using the ClusterProfiler package (v3.14) and its enrichGO function, using Biological Process, Cellular Component, and Molecular Function ontologies. 35 Pathway enrichment was computed for all cluster marker genes within fLEC, MRC, and BRC subsets, specifically focusing on those with an avg_logFC > 0.25 and an adjusted p-value < 0.001. Significantly enriched pathways (adjusted p-value < 0.05) were selected and visually represented using ggplot2. Finally, to effectively illustrate the significance of the top hits derived from the list of differentially expressed genes, violin plots were employed as a visual tool.

 Results and discussion Tumor-induced shifts in LN stromal composition and transcriptome

By design, lymph nodes are strategically positioned to intercept tissue-derived signals and antigen. Established tumors, however, appear to coopt the lymphatic infrastructure to facilitate metastatic dissemination. In this context, lymphatic drainage offers an efficient means of pre-conditioning tissue stroma for metastatic niche formation, and functions as a direct channel for the distribution of signals which may in fact subvert anti-tumor immunity. Indeed, efficiency of lymphatic drainage is positively correlated with metastatic potential, and lymphangiogenesis is a hallmark of metastatic niche formation. 36 In the pre-metastatic lymph node, expansion of both endothelium and FRCs has been documented. 25

However, stromal expansion in LNs is not a tumor-specific response program. Indeed, subcutaneous delivery of fixed tumor cells is sufficient to elicit LN swelling and expansion of FRCs, LECs, and BECs at a level comparable to that observed with live tumor growth ( Figure 1A). To explore more deeply the functional alterations occurring within the stromal compartment that specifically correlated with live tumor growth, we performed single cell RNA sequencing of cells isolated from draining and non-draining LNs in mice inoculated with either live or fixed MC38 colorectal carcinoma cells. We additionally enriched for PDPN-expressing LECs and FRCs to enable robust subset-level analyses. Within the resultant dataset, we identified 7 distinct clusters of fibroblastic cells and 4 clusters of lymphatic endothelial cells, each of which was annotated based on well-established identity markers ( Figure 1B, E, Supplementary Figure 1, Supplementary Tables 1 & 2). 30 – 33

Single cell RNA sequencing of the tumor-draining lymph node: <italic toggle="yes">Mice received either live or fixed MC38 cancer cells by subcutaneous injection.</italic>

Tumor draining and contralateral non-draining inguinal lymph nodes were excised and enzymatically digested for analysis. A) Quantification of cell numbers assessed by flow cytometry (n=5 mice per condition). Data are presented as cell counts relative to the average of corresponding non-draining LN. Cell populations were identified by the following markers: Fibroblastic Reticular cells (CD45-CD31-PDPN+), Lymphatic Endothelial cells (CD45-, CD31+, PDPN+), and Blood Endothelial cells (CD45-, CD31+, PDPN-). B) UMAP visualization of total lymph node fibroblasts, colored by cluster identify. C, D) Quantification of the change in population proportion (amongst fibroblast subsets), along with total number of differentially expressed genes. C. dLN vs ndLN of mice growing live tumors. D. dLN vs ndLN of mice receiving fixed tumor cells. E) UMAP visualization of total lymph node lymphatic endothelial cells, colored by cluster identity. F, G) Quantification of the change in population proportion (amongst endothelial subsets), along with total number of differentially expressed genes. F. dLN vs ndLN of mice growing live tumors. G. dLN vs ndLN of mice receiving fixed tumor cells. H) Schematic representation of lymph node architecture and anticipated location of metastases engraftment. Inset depicts a micrometastasis surrounded by the following stromal subsets of interest: fLECs, MRCs, and BRCs.

Fibroblastic populations include T-zone Reticular Cells (TRC), B zone Reticular cells (BRC), Marginal Zone Reticular Cells (MRC), Follicular Dendritic Cells (FDC), Medullary Reticular cells (MedRC), Perivascular Reticular cells (PRC), and activated stromal cells (Act SC). Endothelial cell subsets include ceiling LECs (cLEC), floor LECs (fLEC), Medullary LECs (MedLEC), and Cortical LECs (CorLEC).

The seeding of micrometastases within tumor-draining lymph nodes predictably occurs within the subcapsular sinus (SCS), where tissues are most readily exposed to lymph borne materials and migrating cancer cells ( Figure 1H). 37 , 38 Interestingly, we demonstrate here that unique cellular constituents of this specific microanatomical compartment were selectively impacted by the upstream tumor ( Figure 1C, F, Supplementary Figure 3). Marginal reticular cells (MRCs) and floor lymphatic endothelial cells (fLECs) exhibited the greatest degree of transcriptional alteration in the tumor-draining LN relative to those of contralateral non-draining LNs. Perivascular reticular cells (PRCs) likewise exhibited significant transcriptomic changes (Supplementary Table 3). Given the physical proximity or association of PRCs with the LN vasculature, transcriptional alteration of this cluster may reflect support of the expanding vascular network which accompanies LN swelling. 39 However, a subset of these cd34+ PRCs, also described as adventitial cells, has likewise been reported to populate the SCS. 33 Notably, the functional role of PRCs is not yet well characterized, but various reports 39 describe cd34+ PRCs as a stromal precursor population that gives rise to various reticular cell subsets.

Unexpectedly, we also found that the overall cellular representation of these populations, along with B-zone reticular cells (BRCs), were disproportionately increased in the tumor-draining LN ( Figure 1C, F). This appears to differ from the pattern of stromal expansion that occurs in response to normal antigenic challenge. While exposure to fixed tumor cells elicited stromal activation and expansion – as previously demonstrated – this proliferative response seemed to occur proportionately amongst fibroblast and endothelial subsets and involved comparatively fewer transcriptomic alterations ( Figure 1D, G). Ultimately, we predict that selective induction of transcriptional changes and disproportionate expansion of stromal subsets comprising the SCS in the tdLN reflects the establishment of a niche supportive of metastatic engraftment and survival.

 MRC and BRCs exhibit elevated ECM production reminiscent of the desmoplastic nature of CAFs

While the exact microenvironmental preferences of primary and secondary lesions may differ, and the means by which these alterations are driven may be distinct, there are almost certainly commonalities in tumor-induced stromal reprogramming. For instance, the aberrant production of extracellular matrix, characteristic of the desmoplastic response, is broadly observed across cancer indications. These responses are largely driven by conversion of resident fibroblast populations to tumor-supportive cancer associated fibroblasts (CAFs). Within the LN SCS, resident fibroblast populations include primarily MRCs that underly the sinus endothelium, and a proportion of BRCs that populate the interfollicular space. As both populations exhibited shifts specific to the tdLN, we re-clustered each population and performed differential transcriptomic analyses to identify tumor-specific changes ( Figure 2A-D). Amongst these changes, MRCs most notably exhibited increases in a broad spectrum of ECM components and matrix remodeling enzymes reminiscent of desmoplastic CAFs of the primary tumor, including upregulation of fibronectin, a critical ECM protein involved in development, wound healing, and cancer metastasis ( Figure 2A,B). 40 Interestingly, PDAC-derived exosomes were recently reported to induce elevated fibronectin expression in the pre-metastatic liver - a process which was found to be essential for niche formation. 41

Transcriptomic features of MRCs, BRCs, and fLECs in the tumor-draining lymph node.

A) ( top) UMAP visualization of MRCs, colored by treatment condition. ( bottom) Pathways significantly enriched in MRCs of live-tumor-draining LNs relative to non-draining LNs. B) ( top) Volcano plots of differentially expressed genes between MRCs of draining vs non-draining LNs. Select genes of interest are highlighted. ( bottom) Violin plots of select genes differentially expressed in MRCs of the live-tumor draining LN. Expression is shown across all four experimental conditions; live draining, live non-draining, fixed draining, and fixed non-draining. C) UMAP visualization of BRCs, colored by treatment condition. D) Volcano plots of differentially expressed genes between BRCs of draining vs non-draining LNs. Select genes of interest are highlighted. E) ( top) UMAP visualization of fLECs, colored by treatment condition. ( bottom) Pathways significantly enriched in fLECs of live-tumor-draining LNs relative to non-draining LNs. F) ( top) Volcano plots of differentially expressed genes between fLECs of draining vs non-draining LNs. Select genes of interest are highlighted. ( bottom) Violin plots of select genes differentially expressed in fLECs of the live-tumor draining LN. Expression is shown across all four experimental conditions; live draining, live non-draining, fixed draining, and fixed non-draining. B, D, F) The cutoffs for highlighted genes include a log2 fold change of 0.5 or above (horizontal dashed lines) and adjusted p-value above 0.001 (indicated by the vertical dashed lines). Negative log2 fold change values indicate genes that are more highly expressed in draining lymph nodes (dLNs) relative to non-draining lymph nodes (ndLNs). G) Dotplot displaying expression of transcripts upregulated across both MRCs and fLECs in the live tdLN. The size of the dot represents percentage of cells expressing transcript, while the color indicates average level of expression within the population.

In addition to ECM proteins, matrix metalloproteases (MMPs) likewise play a pivotal role in cancer progression and metastatic dissemination through matrix remodeling. 42 , 43 Along these lines, we observed increases in both mmp2 and mmp9 specifically in MRCs of the tdLN ( Figure 2B). While it is unclear from transcriptomic signature alone how the structure of the ECM is specifically remodeled in the SCS, or the extent to which this may impact the nascent metastasis, expression of both has previously been linked to niche formation and vascular remodeling in pre-metastatic tissues and are thus of particular interest. 9 , 44 , 45

Interestingly, previous reports have noted significant tumor-induced enlargement of the collagen-rich LN conduits – a branched network of fibrous structures which normally functions as a transport system for small soluble materials through the LN cortex and paracortex. 25 , 46 , 47 In this context, increased conduit thickness correlated with greater transport of larger, normally size-excluded molecules, and the authors speculated that these alterations may result in abnormal distribution of tumor-derived materials to deeper areas of the LN parenchyma. This is notably counter to the effects observed during acute inflammation-associated LN expansion. Under such conditions, LN conduits are also structurally disrupted, but size exclusivity of molecular transport through the network is nevertheless maintained. 48 , 49 This distinction is particularly intriguing, as size exclusivity of LN conduits is believed to be a function of plvap-lined gating channels that traverse the sinus-lining floor LECs (fLECs), rather than an intrinsic feature of conduit size itself. 50 Lymph-borne molecules small enough to pass through these trans-endothelial channels subsequently gain access to the conduit network. We thus speculate that matrisomal dysregulation surrounding fLECs and the underlying MRCs, such as those we find describe within this dataset, may in fact account for, or contribute to, this distinct loss of regulated conduit access in tdLNs.

BRCs likewise exhibited upregulation of some ECM-related transcripts in the tdLN, however the overall extent of differentially expressed genes in this subset was markedly less expansive ( Figure 2C, D). Nevertheless, we note that BRC representation within the dataset was significantly greater in the tdLN, possibly reflecting a greater proliferative response without significant functional divergence from baseline properties of BRCs in a resting LN. As BRCs are essential to the homeostatic maintenance of naïve B cells within the LN, proliferative expansion of the BRC subset may functionally mirror the known preferential accumulation of B cells in tumor draining nodes. 51 , 52 In turn, accumulation of B cells within the tdLN has been shown to drive pro-metastatic lymphangiogenesis, and thus expansion of the B cell support network may be an indirect, but requisite factor in support of this process. 21

fLECs engage specific chemotactic cues within the premetastatic LN niche

Cell recruitment, migration and positioning into and within the LN is a highly regulated process, guided by an array of precisely coordinated chemotactic gradients. Principal among these chemotactic cues are the homeostatic chemokines Ccl19, Ccl21, and Cxcl13, each of which is primarily produced by subsets of LN FRCs. 53 – 56 Previous studies characterizing changes in the tdLN have highlighted a transcriptional downregulation of these cues within 3 days of tumor implantation, and subsequent loss of normal immune cell compartmentalization. Such perturbations would seemingly be detrimental to the initiation of adaptive immune responses, which depends on precise cellular positioning. Indeed, genetic mouse models of homeostatic chemokine disruption exhibit impaired immunity. 57 , 58 However, it is notable that all forms of inflammation and antigenic challenge, not just exposure to tumor-derived factors, can trigger transient downregulation of these factors. 59 – 65 As such, the immunological consequences of chemokine downregulation following tumor inoculation remain unclear. Within our dataset, however, we did not observe significant transcriptional alterations to ccl19, ccl21, or cxcl13 amongst any stromal subset (data not shown). This may be reflective of the later timepoint at which we collected our data, and the transient nature of these fluctuations.

While we do not observe dysregulation of these FRC-associated chemokines, re-clustering and differential analysis of the fLEC population did in fact reveal engagement of pathways related to leukocyte migration ( Figure 2E, F). In particular, we observed dramatic increases in the inflammation-associated chemokines cxcl9, cxcl10 and ccl20 ( Figure 2F). These changes were exclusively depicted within the fLEC population, suggesting specific chemoattraction of cell types expressing the corresponding cognate receptors to the SCS. Interestingly, the Ccl20/Ccr6 axis has been well explored in the context of metastatic colorectal cancer, wherein CCr6-expressing cancer cells are recruited to the liver by Ccl20-expressing periportal stromal cells. 66 Expression of Ccl20 by floor LECs in the tdLN may be similarly utilized by migrating cancer cells in lymphatics. Critically, Ccl20 is absent on neighboring ceiling LECs, allowing the formation of a differential gradient of chemokine across the LN SCS. An analogous chemotactic gradient of Ccl21 across the SCS has been experimentally demonstrated as critical for the recruitment of migrating dendritic cells into the LN. 67

By contrast, Cxcl9 and Cxcl10 are interferon induced chemokines that are most well described in the recruitment and activation of antigen experienced, Cxcr3-expressing T and NK cells. 68 , 69 Expression of these chemokines is therefore consistent with the broader interferon response signature exhibited by fLECs of the tdLN ( Figure 2E). In the context of cancer, these chemokines are notable for their role in recruitment of tumor-infiltrating lymphocytes and are thought to be important for driving immunologically “hot” tumor phenotypes. 70 Expression of Cxcl9 and Cxcl10 within the LN is additionally known to support positioning of memory T cells to the interfollicular region and is critically important for antiviral immune responses. 71 , 72 Cumulatively, these observations may implicate expression of Cxcl9/Cxcl10 by fLECs as a beneficial means to direct antigen responsive immune cells to site of future metastatic engraftment. On the other hand, Cxcr3 may also be expressed by migrating cancer cells, and the Cxcl9/Cxcl10/Cxcr3 signaling axis has in fact been implicated in the growth and metastasis of tumors to various organs, including the LN. 73 – 76 Thus, it remains unclear whether increased expression of these chemokines in the premetastatic LN favor metastasis engraftment or anti-tumor immunity.

 Stromal expression of <italic toggle="yes">il-33</italic> is increased in the pre-metastatic niche

While MRCs and fLECs exhibited unique subset-specific responses in the tdLN, we also noted a handful of upregulated transcriptional elements common across both populations ( Figure 2G). Importantly, upregulation of each was observed only in the draining LN of mice bearing live tumors but did not occur in the LNs of mice inoculated with fixed tumor cells ( Figure 2G). As such, we believe these genes to be uniquely induced in response to tumor-derived signals.

Exposure to factors produced by an upstream tumor may elicit responses that skew immune education, and thus in this context, the striking increase in il-33 transcript – a cytokine with numerous immunological effects that are both context and cell-type specific - was particularly intriguing. 77 Notably, IL-33 is a known driver of ST2+ T-reg expansion and may have an important role in the induction of Treg-mediated tolerance. 78 – 81 In the context of cancer metastasis, ST2-expressing Tregs have been shown to support tumor development, and in vivo inhibition of IL-33 signaling has been shown to disrupt formation of metastases in mouse models. 78 Additionally, elevated IL-33 expression has been described in fibroblasts of lung metastases, wherein signaling is thought to contribute to a supportive niche through skewing immunity towards a Th2 response. 82 Tumor-induced IL33 expression across stromal subsets in the LN SCS may therefore have important implications for suppression of anti-tumor responses. Along these lines, recent reports have indeed suggested that metastatic spread to the LN can elicit immunological tolerance through the preferential induction of antigen-specific Tregs. 83 The functional consequence of this redirected immune response is thus believed to be favorable for subsequent metastatic outgrowth in distant organs.

By contrast, it is also notable that, under infectious conditions, IL33 secretion by LN fibroblasts has been shown to be a stress response program critical for CD8 T cell immunity. 84 , 85 It is thus possible that IL33 upregulation in tdLNs may alternatively engage both anti-tumor immunity or immune suppressive elements, the balance of which may be a critical determinant of tumor progression and metastatic spread.

Interestingly, a recent report has also implicated IL-33 signaling as a driver of fibroblast-to-CAF differentiation in oral squamous cell carcinoma. 86 In this context, IL-33 was found to be upregulated and stabilized by signaling from a novel lncRNA (termed Lnc-CAF), which resulted in acquisition of an activated CAF phenotype and subsequently supported tumor growth. Critically, Lnc-CAF was found to be distributed via tumor cell-derived exosomes, offering a plausible means by which this pathway might mediate preconditioning of LN stroma. Cumulatively, these studies suggest that IL-33 may function not only as a modulator of immunity in the metastatic microenvironment, but as a driving factor in the stromal acquisition of CAF-like features and niche formation.

 Metabolic reprogramming in LN stroma

The upregulation of lactate dehydrogenase ( ldha) and pyruvate kinase M1/2 ( pkm) transcript was likewise common across both MRCs and fLECs ( Figure 2G). This is particularly noteworthy within the context of niche formation and metastasis, as both these genes encode metabolic enzymes associated with the switch to aerobic glycolysis and have long been associated with metabolic adaptation in cancer cells. 87 , 88 Preferential utilization of aerobic glycolysis, even in the presence of sufficient oxygen, is termed the Warburg Effect, and is characteristic of cancer cell metabolism. 89 Despite several decades of research, a full functional account of this metabolic preference remains incompletely understood. However, disruption of this process by targeting of LDHA drives beneficial anti-tumor effects. 90 – 93

Notably, Riedel et al. recently demonstrated that inhibition of LDHA could reverse tumor-directed FRC activation phenotypes, suggesting a putative metabolic switch in stroma that may be similar to that of cancer cells. 24 The authors proposed a model wherein elevated levels of lactate produced in the primary tumor drains to the sentinel lymph node, modulating FRC function and conditioning the pre-metastatic niche. Supportive of this hypothesis, they observed higher expression of LDHA near lymphatic vessels of the primary tumor, and elevated levels of lactate in the draining LN. By contrast, our data suggest that ldha is in fact intrinsically upregulated in both MRCs and fLECs as well, potentially further contributing to the regional build-up of lactate.

Whether this response is secondary to the drainage of lactate from the primary tumor, or rather a response to other yet unidentified signals, remains unknown. However, a recent demonstration that CAFs from the primary tumor can in fact impart Warburg-like metabolism via exosomal transfer of metabolites offers an intriguing alternative whereby CAF-derived exosomes passing through lymphatic drainage may likewise contribute to induction of metabolic preconditioning at the site of future LN metastasis. 94 Nevertheless, both the means by which ldha and pkm are upregulated and the metabolic and functional consequences of these alterations bear further exploration.

 Caveats and study limitations

This study aims at characterizing transcriptomic changes to the lymph node stroma resulting from tumor-derived signals. In many instances, the nature of stromal alterations highlighted in this paper aligns well with previously proposed microenvironmental dependencies for tumor growth and metastasis, and for this reason we contextualize these changes as “pre-metastatic”. However, functional validation of these changes, particularly as they relate to metastatic potential of the tumor, necessitates further exploration. Importantly, such efforts are likely to require the use of models for which lymph node metastasis reliably takes place with relatively predictable kinetics.

Notably, MC38 tumors grown subcutaneously do not readily metastasize (at least within the timeframe at which the primary tumor remains within ethical growth limits), and overt metastatic growth was not observed in any of the LNs we analyzed at time of harvest. 95 Interestingly, putative cancer cells, identified on the basis of CD45-,PDPN-,CD31-,PD-L1+ staining pattern, were found by flow cytometric analysis within the live TdLN at the time of collection (Supplementary Figure 3). However, these cells were very few and may either represent migrating cancer cells shed from the primary tumor or the initial establishment of micrometastases. Additionally, this population may not be exclusively comprised of tumor cells, as we cannot rule out the possibility of PDPN-CD31-stromal cells upregulating PD-L1 in response to activation. Nevertheless, we feel these data support the conclusion that there is little to no metastatic growth within the LN at the time of analysis.

Additionally, we provide comparative analysis against LN stroma following antigenic challenge with fixed cancer cells, with the intent of allowing us to discern potential changes that relate to factors actively produced by tumors from generalized inflammatory response programs. An important limitation to this approach, however, is that antigenic challenge with fixed tumor cells may not fully recapitulate quantitative or qualitative changes related to the overall antigen burden present in the context of a live tumor due to time-dependent growth or maturation. As such, we must also consider the possibility that some of the tumor-induced changes identified in this analysis may relate to the amount of antigen exposure.

Finally, we note that stromal cell subset identification within our dataset was informed by previously described markers across lymph node stromal cell populations. However, within the context of a tumor-draining lymph node, wherein stromal activation or remodeling may occur, we cannot rule out the possibility that key markers used for subset identification may be modulated.

Overall, while we believe the transcriptional profiles characterized herein serve as a valuable resource for identifying potential pathways involved in tumor metastasis to the LN, we caution that the aforementioned limitations be carefully considered.

 Concluding remarks

Metastatic progression is the greatest cause of mortality across most cancer indications, and thus developing new means of clinical intervention in this process is clearly of paramount importance. We believe the findings discussed herein provide an important resource for understanding the early impact of cancer-associated signals on metastatic niche formation. Additionally, we provide comparative analysis against LN stroma following antigenic challenge with fixed cancer cells, allowing us to discern changes that relate to factors actively produced by tumors from generalized inflammatory response programs. Transcriptomic changes discussed herein thus represent tumor-specific responses that may prove to be attractive targets for preventative or therapeutic intervention. However, further experimental exploration will be required to establish causal relationships between these transcriptional changes and metastatic success rate. Additionally, the specific identity of secreted factors driving LN preconditioning are not explored in this study - though paired analyses of primary tumors and tdLN might prove informative in this regard. Ultimately, we suggest that continued exploration along these lines may critically enable the identification of novel pressure points against which transformative therapeutic or preventative intervention may be directed.

 Accession numbers

Gene Expression Omnibus at NCBI: Tumor-driven stromal reprogramming in the pre-metastatic lymph node [Mus musculus (house mouse)]. Accession number GSE248905; http://identifiers.org/geo:GSE248905

Data availability Underlying data

Open Science Framework: Underlying data for ‘Tumor-driven stromal reprogramming in the pre-metastatic lymph node’, https://www.doi.org/10.17605/OSF.IO/WDV3H. 28

This project contains the following underlying data:

Data File: Tumor dLN_A1_A01.fcs

Data File: Tumor dLN_A2_A02.fcs

Data File: Tumor dLN_A3_A03.fcs

Data File: Tumor dLN_A4_A04.fcs

Data File: Tumor dLN_A5_A05.fcs

Data File: Tumor NdLN_B1_B01.fcs

Data File: Tumor NdLN_B2_B02.fcs

Data File: Tumor NdLN_B3_B03.fcs

Data File: Tumor NdLN_B4_B04.fcs

Data File: Tumor NdLN_B5_B05.fcs

Data File: Killed dLN_C1_C01.fcs

Data File: Killed dLN_C2_C02.fcs

Data File: Killed dLN_C3_C03.fcs

Data File: Killed dLN_C4_C04.fcs

Data File: Killed dLN_C5_C05.fcs

Data File: Killed NdLN_D1_D01.fcs

Data File: Killed NdLN_D2_D02.fcs

Data File: Killed NdLN_D3_D03.fcs

Data File: Killed NdLN_D4_D04.fcs

Data File: Killed NdLN_D5_D05.fcs

Mouse FCS data key.xlsArrive Author Checklist.pdf

Extended data

Open Science Framework: Extended data for ‘Tumor-driven stromal reprogramming in the pre-metastatic lymph node’, https://doi.org/10.17605/OSF.IO/ZYCQP. 96

This project contains the following extended data:

Supplementary figure 1. (Dotplot displaying relative expression of marker genes across fibroblast and endothelial subclusters.)

Supplementary figure 2. (UMAP projections for fibroblast and endothelial subclusters, divided by treatment condition.)

Supplementary figure 3. (Flow cytometry quantification of putative cancer cells.)

Supplementary table 1. (Gene expression for 7 fibroblastic stromal cell clusters depicted in Figure 1B.)

Supplementary table 2. (Gene expression for 4 endothelial stromal cell clusters depicted in Figure 1E.)

Supplementary table 3. (Full DEG lists for MRC, BRC, PRC and fLEC subsets in the live tumor dLN vs ndLN, depicted in figure 2b,d,f.)

Supplementary figure legends. (figure legends for each supplementary figure and table.)

Reporting guidelines

Open Science Framework: ARRIVE checklist for ‘Tumor-driven stromal reprogramming in the pre-metastatic lymph node’, https://www.doi.org/10.17605/OSF.IO/WDV3H. 28

Data are available under the terms of the Creative Commons Zero “No rights reserved” data waiver (CC0 1.0 Public domain dedication).

 References Hanahan D Weinberg RA : Hallmarks of cancer: the next generation. Cell. 2011;144(5):646–674. 10.1016/j.cell.2011.02.013 Maley CC : Classifying the evolutionary and ecological features of neoplasms. Nat. Rev. Cancer. 2017;17(10):605–619. 28912577 10.1038/nrc.2017.69 PMC5811185 Valkenburg KC Groot AE de Pienta KJ : Targeting the tumour stroma to improve cancer therapy. Nat. Rev. Clin. Oncol. 2018;15(6):366–381. 29651130 10.1038/s41571-018-0007-1 PMC5960434 Fidler IJ : Metastasis: quantitative analysis of distribution and fate of tumor emboli labeled with 125 I-5-iodo-2′-deoxyuridine. J. Natl. Cancer Inst. 1970;45(4):773–782. 5513503 Fidler IJ Nicolson GL : Organ selectivity for implantation survival and growth of B16 melanoma variant tumor lines. J. Natl. Cancer Inst. 1976;57(5):1199–1202. 1003551 10.1093/jnci/57.5.1199 Paget S : The distribution of secondary growths in cancer of the breast. 1889. Cancer Metastasis Rev. 1989;8(2):98–101. 2673568 Langley RR Fidler IJ : The seed and soil hypothesis revisited--the role of tumor-stroma interactions in metastasis to different organs. Int. J. Cancer. 2011;128(11):2527–2535. 21365651 10.1002/ijc.26031 PMC3075088 Hiratsuka S : MMP9 induction by vascular endothelial growth factor receptor-1 is involved in lung-specific metastasis. Cancer Cell. 2002;2(4):289–300. 12398893 10.1016/S1535-6108(02)00153-8 Kaplan RN : VEGFR1-positive haematopoietic bone marrow progenitors initiate the pre-metastatic niche. Nature. 2005;438(7069):820–827. 16341007 10.1038/nature04186 PMC2945882 Hoshino A : Tumour exosome integrins determine organotropic metastasis. Nature. 2015;527(7578):329–335. 26524530 10.1038/nature15756 PMC4788391 Liu Y : Tumor Exosomal RNAs Promote Lung Pre-metastatic Niche Formation by Activating Alveolar Epithelial TLR3 to Recruit Neutrophils. Cancer Cell. 2016;30(2):243–256. 27505671 10.1016/j.ccell.2016.06.021 Xu WW : Cancer cell-secreted IGF2 instigates fibroblasts and bone marrow-derived vascular progenitor cells to promote cancer progression. Nat. Commun. 2017;8:14399. 28186102 10.1038/ncomms14399 PMC5309924 Sun H : Hypoxia-Inducible Exosomes Facilitate Liver-Tropic Premetastatic Niche in Colorectal Cancer. Hepatology. 2021;74(5):2633–2651. 34110633 10.1002/hep.32009 Zhou H Lei PJ Padera TP : Progression of Metastasis through Lymphatic System. Cells. 2021;10(3). 33808959 10.3390/cells10030627 PMC7999434 Hirakawa S : VEGF-A induces tumor and sentinel lymph node lymphangiogenesis and promotes lymphatic metastasis. J. Exp. Med. 2005;201(7):1089–1099. 15809353 10.1084/jem.20041896 PMC2213132 Sleeman J Schmid A Thiele W : Tumor lymphatics. Semin. Cancer Biol. 2009;19(5):285–297. 10.1016/j.semcancer.2009.05.005 Hood JL San RS Wickline SA : Exosomes released by melanoma cells prepare sentinel lymph nodes for tumor metastasis. Cancer Res. 2011;71(11):3792–3801. 10.1158/0008-5472.CAN-10-4455 Lee E : Inhibition of lymphangiogenesis and angiogenesis in breast tumor xenografts and lymph nodes by a peptide derived from transmembrane protein 45A. Neoplasia. 2013;15(2):112–IN6. 23441126 10.1593/neo.121638 PMC3579314 Maus RLG : Identification of novel, immune-mediating extracellular vesicles in human lymphatic effluent draining primary cutaneous melanoma. Onco. Targets Ther. 2019;8(12):e1667742. 10.1080/2162402X.2019.1667742 Hirakawa S : VEGF-C-induced lymphangiogenesis in sentinel lymph nodes promotes tumor metastasis to distant sites. Blood. 2007;109(3):1010–1017. 17032920 10.1182/blood-2006-05-021758 PMC1785149 Harrell MI Iritani BM Ruddell A : Tumor-induced sentinel lymph node lymphangiogenesis and increased lymph flow precede melanoma metastasis. Am. J. Pathol. 2007;170(2):774–786. 17255343 10.2353/ajpath.2007.060761 PMC1851877 Bois H du Heim TA Lund AW : Tumor-draining lymph nodes: At the crossroads of metastasis and immunity. Sci. Immunol. 2021;6(63):eabg3551. 34516744 10.1126/sciimmunol.abg3551 PMC8628268 Davidson S : Single-Cell RNA Sequencing Reveals a Dynamic Stromal Niche That Supports Tumor Growth. Cell Rep. 2020;31(7):107628. 32433953 10.1016/j.celrep.2020.107628 PMC7242909 Riedel A : Tumor-Derived Lactic Acid Modulates Activation and Metabolic Status of Draining Lymph Node Stroma. Cancer Immunol. Res. 2022;10(4):482–497. 35362044 10.1158/2326-6066.CIR-21-0778 PMC7612582 Riedel A : Tumor-induced stromal reprogramming drives lymph node transformation. Nat. Immunol. 2016;17(9):1118–1127. 27400148 10.1038/ni.3492 PMC4994871 Chang JE Turley SJ : Stromal infrastructure of the lymph node and coordination of immunity. Trends Immunol. 2015;36(1):30–39. 25499856 10.1016/j.it.2014.11.003 Fletcher AL : Reproducible isolation of lymph node stromal cells reveals site-dependent differences in fibroblastic reticular cells. Front. Immunol. 2011;2:35. 22566825 10.3389/fimmu.2011.00035 PMC3342056 Piquet M Ruddy DA Cremasco V : Tumor-driven stromal reprogramming in the pre-metastatic lymph node.[Dataset]. Open Science Framework. 2024. 10.17605/OSF.IO/WDV3H Stuart T : Comprehensive Integration of Single-Cell Data. Cell. 2019;177(7):1888–1902 e21. Buechler MB : Cross-tissue organization of the fibroblast lineage. Nature. 2021;593(7860):575–579. 33981032 10.1038/s41586-021-03549-5 Braanker H den : Single-Cell RNA Sequencing Reveals Heterogeneity and Functional Diversity of Lymphatic Endothelial Cells. Int. J. Mol. Sci. 2021;22(21). 34769408 10.3390/ijms222111976 PMC8584409 Fujimoto N : Single-cell mapping reveals new markers and functions of lymphatic endothelial cells in lymph nodes. PLoS Biol. 2020;18(4):e3000704. 32251437 10.1371/journal.pbio.3000704 PMC7162550 Rodda LB : Single-Cell RNA Sequencing of Lymph Node Stromal Cells Reveals Niche-Associated Heterogeneity. Immunity. 2018;48(5):1014–1028.e6. 29752062 10.1016/j.immuni.2018.04.006 PMC5971117 Pikor NB : Remodeling of light and dark zone follicular dendritic cells governs germinal center responses. Nat. Immunol. 2020;21:649–659. 32424359 10.1038/s41590-020-0672-y PMC7610477 Yu G : clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284–287. 22455463 10.1089/omi.2011.0118 PMC3339379 Holder AM Ziemys A : Lymphatic Transport Efficiency Determines Metastatic Potential of Cutaneous Melanoma. Front. Oncol. 2020;10:1607. 33042804 10.3389/fonc.2020.01607 PMC7518046 Brown M : Lymph node blood vessels provide exit routes for metastatic tumor cell dissemination in mice. Science. 2018;359(6382):1408–1411. 29567714 10.1126/science.aal3662 Pereira ER : Lymph node metastases can invade local blood vessels, exit the node, and colonize distant organs in mice. Science. 2018;359(6382):1403–1407. 29567713 10.1126/science.aal3622 PMC6002772 Sitnik KM : Context-Dependent Development of Lymphoid Stroma from Adult CD34(+) Adventitial Progenitors. Cell Rep. 2016;14:2375–2388. 26947077 10.1016/j.celrep.2016.02.033 Lin TC : Fibronectin in Cancer: Friend or Foe. Cells. 2019;9(1). 31861892 10.3390/cells9010027 PMC7016990 Costa-Silva B : Pancreatic cancer exosomes initiate pre-metastatic niche formation in the liver. Nat. Cell Biol. 2015;17(6):816–826. 25985394 10.1038/ncb3169 PMC5769922 Cui N Hu M Khalil RA : Biochemical and Biological Attributes of Matrix Metalloproteinases. Prog. Mol. Biol. Transl. Sci. 2017;147:1–73. 28413025 10.1016/bs.pmbts.2017.02.005 PMC5430303 Gobin E : A pan-cancer perspective of matrix metalloproteases (MMP) gene expression profile and their diagnostic/prognostic potential. BMC Cancer. 2019;19(1):581. 31200666 10.1186/s12885-019-5768-0 PMC6567474 Owyong M : MMP9 modulates the metastatic cascade and immune landscape for breast cancer anti-metastatic therapy. Life Sci. Alliance. 2019;2(6):e201800226. 31727800 10.26508/lsa.201800226 PMC6856766 Zeng ZS Cohen AM Guillem JG : Loss of basement membrane type IV collagen is associated with increased expression of metalloproteinases 2 and 9 (MMP-2 and MMP-9) during human colorectal tumorigenesis. Carcinogenesis. 1999;20(5):749–755. 10334190 10.1093/carcin/20.5.749 Gretz JE Anderson AO Shaw S : Cords, channels, corridors and conduits: critical architectural elements facilitating cell interactions in the lymph node cortex. Immunol. Rev. 1997;156:11–24. 9176696 10.1111/j.1600-065x.1997.tb00955.x Anderson AO Anderson ND : Studies on the structure and permeability of the microvasculature in normal rat lymph nodes. Am. J. Pathol. 1975;80:387–418. 1163637 Acton SE : Communication, construction, and fluid control: lymphoid organ fibroblastic reticular cell and conduit networks. Trends Immunol. 2021;42(9):782–794. 34362676 10.1016/j.it.2021.07.003 Martinez VG : Fibroblastic Reticular Cells Control Conduit Matrix Deposition during Lymph Node Expansion. Cell Rep. 2019;29(9):2810–2822.e5. 31775047 10.1016/j.celrep.2019.10.103 PMC6899512 Rantakari P : The endothelial protein PLVAP in lymphatics controls the entry of lymphocytes and antigens into lymph nodes. Nat. Immunol. 2015;16(4):386–396. 10.1038/ni.3101 Cremasco V : B cell homeostasis and follicle confines are governed by fibroblastic reticular cells. Nat. Immunol. 2014;15(10):973–981. 25151489 10.1038/ni.2965 PMC4205585 Ruddell A : B lymphocytes promote lymphogenous metastasis of lymphoma and melanoma. Neoplasia. 2011;13(8):748–757. 21847366 10.1593/neo.11756 PMC3156665 Bleul CC : A highly efficacious lymphocyte chemoattractant, stromal cell-derived factor 1 (SDF-1). J. Exp. Med. 1996;184(3):1101–1109. 9064327 10.1084/jem.184.3.1101 PMC2192798 Gunn MD : A B-cell-homing chemokine made in lymphoid follicles activates Burkitt’s lymphoma receptor-1. Nature. 1998;391(6669):799–803. 9486651 10.1038/35876 Gunn MD : A chemokine expressed in lymphoid high endothelial venules promotes the adhesion and chemotaxis of naive T lymphocytes. Proc. Natl. Acad. Sci. USA. 1998;95(1):258–263. 9419363 10.1073/pnas.95.1.258 PMC18193 Ngo VN Tang HL Cyster JG : Epstein-Barr virus-induced molecule 1 ligand chemokine is expressed by dendritic cells in lymphoid tissues and strongly attracts naive T cells and activated B cells. J. Exp. Med. 1998;188(1):181–191. 9653094 10.1084/jem.188.1.181 PMC2525549 Mori S : Mice lacking expression of the chemokines CCL21-ser and CCL19 (plt mice) demonstrate delayed but enhanced T cell immune responses. J. Exp. Med. 2001;193:207–218. 11148224 10.1084/jem.193.2.207 PMC2193340 Ansel KM : A chemokine-driven positive feedback loop organizes lymphoid follicles. Nature. 2000;406:309–314. 10917533 10.1038/35018581 Benedict CA : Specific remodeling of splenic architecture by cytomegalovirus. PLoS Pathog. 2006;2(3):e16. 16518465 10.1371/journal.ppat.0020016 PMC1386719 Cadman ET : Alterations of splenic architecture in malaria are induced independently of Toll-like receptors 2, 4, and 9 or MyD88 and may affect antibody affinity. Infect. Immun. 2008;76(9):3924–3931. 18559428 10.1128/IAI.00372-08 PMC2519400 Glatman Zaretsky A : Infection with Toxoplasma gondii alters lymphotoxin expression associated with changes in splenic architecture. Infect. Immun. 2012;80(10):3602–3610. 22851754 10.1128/IAI.00333-12 PMC3457551 John B : Dynamic Imaging of CD8(+) T cells and dendritic cells during infection with Toxoplasma gondii. PLoS Pathog. 2009;5(7):e1000505. 19578440 10.1371/journal.ppat.1000505 PMC2700268 Mueller SN : Regulation of homeostatic chemokine expression and cell trafficking during immune responses. Science. 2007;317(5838):670–674. 10.1126/science.1144830 Mueller SN : Viral targeting of fibroblastic reticular cells contributes to immunosuppression and persistence during chronic infection. Proc. Natl. Acad. Sci. USA. 2007;104(39):15430–15435. 17878315 10.1073/pnas.0702579104 PMC2000533 St John AL Abraham SN : Salmonella disrupts lymph node architecture by TLR4-mediated suppression of homeostatic chemokines. Nat. Med. 2009;15(11):1259–1265. 19855398 10.1038/nm.2036 PMC3616497 Ghadjar P : The chemokine CCL20 and its receptor CCR6 in human malignancy with focus on colorectal cancer. Int. J. Cancer. 2009;125(4):741–745. 19480006 10.1002/ijc.24468 Ulvmar MH : The atypical chemokine receptor CCRL1 shapes functional CCL21 gradients in lymph nodes. Nat. Immunol. 2014;15(7):623–630. 24813163 10.1038/ni.2889 Hiroi M Ohmori Y : The transcriptional coactivator CREB-binding protein cooperates with STAT1 and NF-kappa B for synergistic transcriptional activation of the CXC ligand 9/monokine induced by interferon-gamma gene. J. Biol. Chem. 2003;278(1):651–660. 12403783 10.1074/jbc.M204544200 Tokunaga R : CXCL9, CXCL10, CXCL11/CXCR3 axis for immune activation - A target for novel cancer therapy. Cancer Treat. Rev. 2018;63:40–47. 29207310 10.1016/j.ctrv.2017.11.007 PMC5801162 Hoch T : Multiplexed imaging mass cytometry of the chemokine milieus in melanoma characterizes features of the response to immunotherapy. Sci. Immunol. 2022;7(70):eabk1692. 35363540 10.1126/sciimmunol.abk1692 Kastenmuller W : Peripheral prepositioning and local CXCL9 chemokine-mediated guidance orchestrate rapid memory CD8+ T cell responses in the lymph node. Immunity. 2013;38(3):502–513. 23352234 10.1016/j.immuni.2012.11.012 PMC3793246 Sung JH : Chemokine guidance of central memory T cells is critical for antiviral recall responses in lymph nodes. Cell. 2012;150(6):1249–1263. 22980984 10.1016/j.cell.2012.08.015 PMC3445043 Cambien B : Organ-specific inhibition of metastatic colon carcinoma by CXCR3 antagonism. Br. J. Cancer. 2009;100(11):1755–1764. 19436305 10.1038/sj.bjc.6605078 PMC2695685 Kawada K : Chemokine receptor CXCR3 promotes colon cancer metastasis to lymph nodes. Oncogene. 2007;26(32):4679–4688. 17297455 10.1038/sj.onc.1210267 Walser TC : Antagonism of CXCR3 inhibits lung metastasis in a murine model of metastatic breast cancer. Cancer Res. 2006;66(15):7701–7707. 16885372 10.1158/0008-5472.CAN-06-0709 Zhu G : CXCR3 as a molecular target in breast cancer metastasis: inhibition of tumor cell migration and promotion of host anti-tumor immunity. Oncotarget. 2015;6(41):43408–43419. 26485767 10.18632/oncotarget.6125 PMC4791240 Afferni C : The Pleiotropic Immunomodulatory Functions of IL-33 and Its Implications in Tumor Immunity. Front. Immunol. 2018;9:2601. 30483263 10.3389/fimmu.2018.02601 PMC6242976 Li A : IL-33 Signaling Alters Regulatory T Cell Diversity in Support of Tumor Development. Cell Rep. 2019;29(10):2998–3008.e8. 31801068 10.1016/j.celrep.2019.10.120 PMC6990979 Schiering C : The alarmin IL-33 promotes regulatory T-cell function in the intestine. Nature. 2014;513(7519):564–568. 25043027 10.1038/nature13577 PMC4339042 Turnquist HR : IL-33 expands suppressive CD11b+ Gr-1(int) and regulatory T cells, including ST2L+ Foxp3+ cells, and mediates regulatory T cell-dependent promotion of cardiac allograft survival. J. Immunol. 2011;187(9):4598–4610. 21949025 10.4049/jimmunol.1100519 PMC3197898 Vasanthakumar A : Erratum: The transcriptional regulators IRF4, BATF and IL-33 orchestrate development and maintenance of adipose tissue-resident regulatory T cells. Nat. Immunol. 2015;16(5):544. 25898200 10.1038/ni0515-544d Shani O : Fibroblast-Derived IL33 Facilitates Breast Cancer Metastasis by Modifying the Immune Microenvironment and Driving Type 2 Immunity. Cancer Res. 2020;80(23):5317–5329. 33023944 10.1158/0008-5472.CAN-20-2116 PMC7611300 Reticker-Flynn NE : Lymph node colonization induces tumor-immune tolerance to promote distant metastasis. Cell. 2022;185(11):1924–1942.e23. 35525247 10.1016/j.cell.2022.04.019 PMC9149144 Aparicio-Domingo P : Fibroblast-derived IL-33 is dispensable for lymph node homeostasis but critical for CD8 T-cell responses to acute and chronic viral infection. Eur. J. Immunol. 2021;51(1):76–90. 32700362 10.1002/eji.201948413 Marx AF : The alarmin interleukin-33 promotes the expansion and preserves the stemness of Tcf-1(+) CD8(+) T cells in chronic viral infection. Immunity. 2023;56:813–828.e10. 36809763 10.1016/j.immuni.2023.01.029 Ding L : A novel stromal lncRNA signature reprograms fibroblasts to promote the growth of oral squamous cell carcinoma via LncRNA-CAF/interleukin-33. Carcinogenesis. 2018;39(3):397–406. 29346528 10.1093/carcin/bgy006 Goldman RD Kaplan NO Hall TC : Lactic Dehydrogenase in Human Neoplastic Tissues. Cancer Res. 1964;24:389–399. 14147812 Yang W Lu Z : Nuclear PKM2 regulates the Warburg effect. Cell Cycle. 2013;12(19):3154–3158. 24013426 10.4161/cc.26182 Warburg O : On the origin of cancer cells. Science. 1956;123(3191):309–314. 10.1126/science.123.3191.309 Fantin VR St-Pierre J Leder P : Attenuation of LDH-A expression uncovers a link between glycolysis, mitochondrial physiology, and tumor maintenance. Cancer Cell. 2006;9(6):425–434. 16766262 10.1016/j.ccr.2006.04.023 Le A : Inhibition of lactate dehydrogenase A induces oxidative stress and inhibits tumor progression. Proc. Natl. Acad. Sci. USA. 2010;107(5):2037–2042. 20133848 10.1073/pnas.0914433107 PMC2836706 Lu QY : Metabolic Consequences of LDHA inhibition by Epigallocatechin Gallate and Oxamate in MIA PaCa-2 Pancreatic Cancer Cells. Metabolomics. 2015;11(1):71–80. 26246802 10.1007/s11306-014-0672-8 PMC4523095 Xie H : LDH-A inhibition, a therapeutic strategy for treatment of hereditary leiomyomatosis and renal cell cancer. Mol. Cancer Ther. 2009;8(3):626–635. 19276158 10.1158/1535-7163.MCT-08-1049 PMC2671637 Zhao H : Tumor microenvironment derived exosomes pleiotropically modulate cancer cell metabolism. elife. 2016;5:e10250. 26920219 10.7554/eLife.10250 PMC4841778 Greenlee JD King MR : A syngeneic MC38 orthotopic mouse model of colorectal cancer metastasis. Biol. Methods Protoc. 2022;7:bpac024. 36225595 10.1093/biomethods/bpac024 PMC9550231 Chang J Piquet M Ruddy DA : Tumor-driven stromal reprogramming in the pre-metastatic lymph node. 2025, April 2. 10.17605/OSF.IO/QMD39 10.5256/f1000research.188508.r421293 Reviewer response for version 3 Luther Sanjiv 1 Referee https://orcid.org/0000-0002-8758-957X Department of Immunobiology, University of Lausanne, Lausanne, Switzerland

Competing interests: No competing interests were disclosed.

 22 10 2025 Copyright: © 2025 Luther S 2025 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve

I am satisfied with the changes made. I have no further comments.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Yes

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Yes

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Yes

Reviewer Expertise:

lymph nodes, stromal cells, immunity, scRNAseq

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

 10.5256/f1000research.178762.r377193 Reviewer response for version 2 Luther Sanjiv 1 Referee https://orcid.org/0000-0002-8758-957X Department of Immunobiology, University of Lausanne, Lausanne, Switzerland

Competing interests: No competing interests were disclosed.

 10 5 2025 Copyright: © 2025 Luther S 2025 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

The authors have made major changes in response to my comments by adding supplementary figures and tables that help a lot in reading and understanding the manuscript. The new caveat section is also helpful given that this study is limited to the scRNAseq datasets described.

Minor points:

Typo in figure 1 E legend: colored by cluster identity ? or a function ‘cluster identify’ and put in italic?

Supplementary figure 2 legend: say ‘pooled’ fibroblast or ‘pooled’ endothelial UMAP to be clearer?

Supplementary table 3 legend: correct ‘figure 2B,D,F’

Page 9: …chemotactic gradients of CCL19: the authors of cited ref 67 showed nothing with CCL19 but probably another paper should have been cited instead showing an ACKR4 role in DC trafficking from the SCS into the T zone due to CCL21 gradient shaping (Ulvmar, Immunity 2014). That paper did not show CCL19 gradient shaping by ACKR4 in vivo but provided only in vitro evidence; please adapt the claim accordingly.

Reply to major comment 1: Madcam1 transcript expression is being used as marker to identify MRC / FDC versus Madcam1-negative BRC, with reference to previous studies looking at LNs in homeostasis or viral infection (refs 33 and 34). While ref 33 shows that Madcam1 transcript levels are a poorly reliable subset marker, ref 34 indicates Madcam1 expression being also present in non-MRC/FDC, although at lower levels than in the latter. In the current study, the authors look at LN stroma in a different setting (tumor drainage and antitumor immunity) not previously described by scRNAseq, with Madcam1 being potentially upregulated upon stromal cell activation (by signals like LTab, similar to LN development). Thus, BRC other than MRC/FDC may express or even upregulate Madcam1 in this setting and thus be misclassified in the current study. The use of few and variable subset markers like Madcam1 is a caveat of the present study given that the authors do not provide data to confirm their claims at the protein and histological level. I recommend that the authors include this point in the caveat section of this manuscript.

Page 11 (and reply to major comment 4): Tumor cells were indirectly identified among Pdpn-CD31- cells as being PDL1+. This seems a very vague identification that may be considerably contaminated by various Pdpn-CD31- fibroblast subsets that are likely to upregulate PDL1 upon IFN sensing. Thus, this supplementary figure 3 along with the text should only be shown if the authors have solid evidence for these cells being strongly enriched in metastatic tumor cells.

Please verify validity of accession number for the identifiers.org database.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Yes

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Yes

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Yes

Reviewer Expertise:

lymph nodes, stromal cells, immunity, scRNAseq

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Chang Jonathan Oncology, Novartis, Cambridge, MA, USA

Competing interests: No competing interests were disclosed.

 23 9 2025

Response to Reviewer's comments Below:

Reviewer Comment 1:  Typo in figure 1 E legend: colored by cluster identity  ? or a function ‘cluster identify’ and put in italic?  Response:  Typo corrected

Reviewer Comment 2: Supplementary figure 2 legend: say ‘pooled’ fibroblast or ‘pooled’ endothelial UMAP to be clearer? Response: Changed according to reviewer’s suggestion.

Reviewer Comment 3: Supplementary table 3 legend: correct ‘figure  2B,D,F’.   Response : legend has been corrected

Reviewer Comment 4: Page 9: …chemotactic gradients of CCL19: the authors of cited ref 67 showed nothing with CCL19 but probably another paper should have been cited instead showing an ACKR4 role in DC trafficking from the SCS into the T zone due to CCL21 gradient shaping (Ulvmar, Immunity 2014). That paper did not show CCL19 gradient shaping by ACKR4 in vivo but provided only in vitro evidence; please adapt the claim accordingly.  Response :   Thank you for your suggestion – we believe the currently cited paper from Ulvmar et al (Nat Imm 2014) is still appropriate for the intended claim.  In the cited paper, the authors demonstrate 1) Ccl21 gradient across the SCS is shaped by cLEC expression of ACKR4 (aka CCLR1), 2) loss of ACKR4 impacts DC emigration to the LN from the lymphatics in vivo, and 3) ACKR4 could create functional gradients of Ccl19 in vitro.  You are correct that they did not specifically demonstrate Ccl19 gradients across the SCS in vivo, and this is largely because Ccl19 is soluble and cannot be readily imaged in tissues.  However, because ACKR4 scavenges both Ccl19/Ccl21, and both chemokines are similarly expressed in the SCS, the authors speculate that ACKR4 influences the patterning of both chemokines.  They state:  “Thus, we can only speculate that CCL19, produced at the same sites as CCL21, as well as potentially in the SCS itself, might be 'patterned' by CCRL1 on cLECs to form gradients similar to those of CCL21.”  Given the lack of explicit in vivo evidence regarding Ccl19 patterning, however, we will have slightly amended our original statement to reference only Ccl21 patterning.  The statement in question now reads “Analogous chemotactic gradients of Ccl21 across the SCS has been experimentally demonstrated as critical for the recruitment of migrating dendritic cells into the LN.”

Reviewer Comment 5:  (Reply to major comment 1): Madcam1 transcript expression is being used as marker to identify MRC / FDC versus Madcam1-negative BRC, with reference to previous studies looking at LNs in homeostasis or viral infection (refs 33 and 34). While ref 33 shows that Madcam1 transcript levels are a poorly reliable subset marker, ref 34 indicates Madcam1 expression being also present in non-MRC/FDC, although at lower levels than in the latter. In the current study, the authors look at LN stroma in a different setting (tumor drainage and antitumor immunity) not previously described by scRNAseq, with Madcam1 being potentially upregulated upon stromal cell activation (by signals like LTab, similar to LN development). Thus, BRC other than MRC/FDC may express or even upregulate Madcam1 in this setting and thus be misclassified in the current study. The use of few and variable subset markers like Madcam1 is a caveat of the present study given that the authors do not provide data to confirm their claims at the protein and histological level. I recommend that the authors include this point in the caveat section of this manuscript.  Response:  We appreciate the careful consideration of Madcam1 expression, and we agree that we cannot rule out the possibility of tumor-derived signals eliciting upregulation or downregulation of cluster-identification markers (whether Madcam1 or any other marker).  As you note, there is some discrepancy between references 33 and 34 (Pikor et al and Rodda et al) with regard to Madcam1 expression.  Notably, Pikor et al (Nat Imm 2020) acknowledged the discrepancy in Madcam1 expression with that seen by Rodda et al (2018), and mentioned that their dataset resolved two tnfsf11+ subsets – one of which is Madcam1-positive, and the other which is Madcam1-negative.  It is therefore also possible that the difference between these two referenced papers relates to the resolution of clustering.  However, we would also like to emphasize that beyond Madcam1 expression, specific expression of tnfsf11 and elevated enpp2 were also used as markers to help distinguish the MRC cluster.  Nevertheless, we do acknowledge that this study includes only transcriptomic data and relies on previously published studies to interpret subset identification.  And as noted by the reviewer, these previously published studies do not include an investigation of cluster identification within the context of a tumor-draining lymph node – as such we have amended the manuscript to include mention of this concern within the “Caveats and study limitations” section.

Reviewer Comment 6:  Page 11 (and reply to major comment 4): Tumor cells were indirectly identified among Pdpn-CD31- cells as being PDL1+. This seems a very vague identification that may be considerably contaminated by various Pdpn-CD31- fibroblast subsets that are likely to upregulate PDL1 upon IFN sensing. Thus, this supplementary figure 3 along with the text should only be shown if the authors have solid evidence for these cells being strongly enriched in metastatic tumor cells.  Response :  The reviewer is correct that we cannot rule out the possibility of this population (PDPN-CD31-PDL1+) containing non-tumor cells.  In the amended manuscript, we have included mention of the possibility of PDPN-CD31- stromal cells upregulating PD-L1.  This supplementary figure was included in response to the reviewer’s previous question about the possibility of active metastatic growth in the tdLN.  As noted in our previous response, this population amounted to an average of just 114 total cells in the tdLN.  Whether this PDPN-CD31-PDL1+ population purely consists of mestatic tumor cells, or additionally contains activated DN stromal cells that have upregulated PD-L1, we feel that the conclusion should still be that there are very few (if any) tumor cells within the LN at time of collection. 

Reviewer Comment 7:  Please verify validity of accession number for the identifiers.org database.  Response:  The accession number is correct.  However there appears to have been a space incorrectly placed within the URL.  The correct URL is http://identifiers.org/geo:GSE248905 and this has been updated in the revised manuscript.

10.5256/f1000research.178762.r377192 Reviewer response for version 2 McGinnis Christopher 1 Referee https://orcid.org/0000-0001-6923-9341 Department of Pathology, Stanford University, Gladstone-UCSF Institute of Genomic Immunology, Parker Institute for Cancer Immunotherapy, San Francisco, USA

Competing interests: No competing interests were disclosed.

 26 4 2025 Copyright: © 2025 McGinnis C 2025 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve

The authors satisfactorily addressed all of my major and minor comments (with the exclusion of the handling of cell-cell doublets, but this does not detract from the overall claims of the manuscript).

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Cancer immunology, metastasis biology, single-cell genomics

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

 10.5256/f1000research.159080.r296538 Reviewer response for version 1 McGinnis Christopher 1 Referee https://orcid.org/0000-0001-6923-9341 Department of Pathology, Stanford University, Gladstone-UCSF Institute of Genomic Immunology, Parker Institute for Cancer Immunotherapy, San Francisco, USA

Competing interests: No competing interests were disclosed.

 14 9 2024 Copyright: © 2024 McGinnis C 2024 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

In this study by Piquet et al, the Authors use single-cell RNA-sequencing (scRNA-seq) to compare lymph node (LN) stromal cell transcriptional profiles in mice bearing live and fixed tumors to disentangle changes in LN stromal cells related to pro-metastatic remodeling mechanisms from changes linked to antigenic challenge. The authors make the following claims:

 

Development of the LN pre-metastatic niche involves changes in the proportions and gene expression profiles of specific subtypes of fibroblasts and endothelial cells comprising the subcapsular sinus, in addition to other LN cell populations (e.g., b-zone reticular cells and perivascular cells).

MRCs in the live tdLN are induced into a desmoplastic CAF-like state associated with upregulation of ECM components and remodeling genes such as fibronectin and Mmp2/9. Authors speculate that tumor-reprogrammed MRCs contribute to altered regulation of LN conduit access.

fLECs increase expression of pro-inflammatory cytokines (e.g., Ccl20 and Cxcl9/10) in tdLNs, raising the question of whether TILs/NK cells or Cxcr3+ metastatic tumor cells (or both) are preferentially recruited to the subcapsular sinus.

MRCs and fLECs enact a coordinate gene expression program in the live tdLN associated with increased Il33 and aerobic glycolysis genes, providing new insights into stromal cell signaling and metabolism mechanisms in the LN metastatic niche.

 Major Comments

 

Do the authors have evidence showing that mice harboring MC38 tumors actually form LN metastases? Or at least what proportion of the mice form mets? The MC38 tumor model is not thought to robustly metastasize (e.g., only ~50% of orthotopically-transplanted tumors form LN metastases; Greenlee & King, 2022), especially after subcutaneous injection. Thus, it is unclear whether the signatures described in the paper truly represent the pre-metastatic niche LN. Without adequately addressing this concern, the authors need to recontextualize their observations to focus on the effects of primary tumor-mediated reprogramming on the tdLN rather than the pre-metastatic niche.

 “Within the resultant dataset, we identified 7 distinct clusters of fibroblastic cells and 4 clusters of lymphatic endothelial cells, each of which was annotated based on well-established identity markers” – Authors should include either reference to which markers were used or, better yet, a heatmap or dotplot showing expression levels for marker genes in each annotated cell type.

“Perivascular cells (PRCs), though not localized to the SCS, likewise exhibited significant transcriptomic changes, which may reflect support of vascular expansion” – While entirely possible, it is unclear how the observed increase in the numbers of DEGs suggests support of vascular expansion. Authors should analyze more deeply the genes that are differentially expressed (e.g., GSEA, targeted analysis of known proangiogenic factors expressed by PRCs, etc.) before making this claim.  

The Authors make many claims throughout the manuscript about proliferative expansion of certain fibroblast and endothelial cell subsets in tumor-draining lymph nodes, but do not show any direct evidence of this increased proliferation. Since annotating proliferative cells via Mki67 and Hells expression in scRNA-seq data is possible, I recommend the Authors compare the proportions of proliferative subsets in all experimental groups to test/strengthen these claims.

The Authors do not describe how they handled the removal of doublets during scRNA-seq analysis. The presence of doublets could significantly confound data interpretation – particularly in instances where the underlying cell type distributions between samples are known to differ, as in this case – and, thus, needs to be addressed to ensure the observations are not artefactual in nature.

In the Methods section, the Authors state that “The present study used tissues collected from naïve mice (n=4), mice bearing live MC38 tumors (n=3), or fixed MC38 tumor cells (n=3)” and then “No animals were excluded from analysis.” However, the described scRNA-seq analyses do not incorporate insights gained from the naïve mouse samples. Comparing naïve and fixed tumor samples could provide key insight for distinguishing the effects of live tumors on the tdLN. Moreover, including comparisons between naïve and fixed tumor samples would be critical for pinpointing the observed effects that are specifically due to antigenic challenge. Alternatively, if naïve samples were not used, the Methods section should be edited to clarify this point.

In Fig. 2G, the authors show genes that are specifically enriched in tdLN MRCs and fLECs. How were these genes identified? Is Il33 expression between tdLNs isolated from mice harboring live and fixed tumors statistically-significantly different? The up-regulation between tdLN and ndLN is clear in both live and fixed tumor settings, but tdLNs in fixed tumor samples also increase expression of Il33 compared to ndLNs, suggesting that this observation may not be live tumor-specific.  

Minor Comments

 

Authors state “In the pre-metastatic lymph node, expansion of both endothelium and FRCs has been documented” – true, but authors should cite appropriate literature references.

How did authors verify that MC38 tumor cells were successfully fixed after 4% PFA exposure?

Fig. 1A: Authors should clarify in the legend the reference point they used to calculate relative cell count. Imagining it was the average of the non-draining LN, but should be explicit to avoid confusion. Authors should also clarify how each cell type was identified using their flow cytometry panel.

Fig. 1B and E: Authors do not ever define some of the acronyms used here (e.g., TRC, FDC)

Page 7: “In this context, Increased conduit thickness…” – ‘Increased’ should not be capitalized.

Fig. 2B/E/F – authors should be explicit about what the volcano plot colors scheme to ensure clarity of interpretation. Also, are the p-values presented raw or adjusted?

As far as I can tell, the Authors use increased expression of fibronectin by MRCs in the tdLN to support the claim that they are induced into a desmoplastic CAF-like state. This may indeed be the case, but more thorough analyses/discussions are needed to sufficiently support this claim. I would suggest leveraging publicly-available scRNA-seq data of desmoplastic CAFs (if such data exists) to assess similarly in transcriptional signatures. Alternatively, the language can be edited to lessen the claimed connection.

“Indeed, genetic mouse models of homeostatic chemokine disruption exhibit impaired immunity.” Need reference.

Should correctly note gene/protein identifiers (e.g., italics for gene names, proteins capitalized, etc.).

The 30% mitochondrial gene expression threshold is quite high. Authors should interrogate whether any high pMito clusters were retained in analyses that could confound interpretation.

 On balance, I believe that the presented work represents a useful advance for the field, and I recommend its indexing, assuming that the concerns addressed above are satisfactorily addressed. Moreover, I would also like to specifically note that the manuscript is extremely well-written.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Cancer immunology, metastasis biology, single-cell genomics

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Chang Jonathan Oncology, Novartis, Cambridge, MA, USA

Competing interests: No competing interests were disclosed.

 28 3 2025

(Major) Reviewer Comment 1:  Do the authors have evidence showing that mice harboring MC38 tumors actually form LN metastases? Or at least what proportion of the mice form mets? The MC38 tumor model is not thought to robustly metastasize (e.g., only ~50% of orthotopically-transplanted tumors form LN metastases; Greenlee & King, 2022), especially after subcutaneous injection. Thus, it is unclear whether the signatures described in the paper truly represent the pre-metastatic niche LN. Without adequately addressing this concern, the authors need to recontextualize their observations to focus on the effects of primary tumor-mediated reprogramming on the tdLN rather than the pre-metastatic niche.  Response:  The reviewer raises a salient point in that subcutaneously implanted tumor models rarely in fact metastasize (at least within the timeframe at which the primary tumor remains within ethical growth limits).  Orthotopically implanted tumors more readily metastasize, perhaps due to the native regional tissue environment of implantation.  However, such models are often technically challenging and highly variable in terms of tumor growth kinetics.  We opted to use subcutaneous MC38 tumors to minimize technical and experimental variance which might otherwise obscure the identification of transcriptional signatures influenced by tumor-derived factors.  We feel the nature of multiple stromal alterations highlighted in this paper align well with previously proposed microenvironmental dependencies for tumor growth and metastasis highlighted within the broader literature, and it is for this reason that we contextualize these changes as “pre-metastatic”.  However, we fully recognize that data herein do not constitute experimental proof that any or all of these transcriptional changes are in fact necessary for metastasis but rather offer these results as a resource for more definitive studies to follow.  We agree that such experimental proof would necessitate the use of models for which lymph node metastasis can be explicitly documented following experimental perturbation.  Such an effort is currently beyond the scope of this limited study.  However, we do acknowledge that this is a critically important limitation, and discussion of this point has been included in the revised manuscript under the subheading “Caveats and study limitations”.

(Major) Reviewer Comment 2:    “Within the resultant dataset, we identified 7 distinct clusters of fibroblastic cells and 4 clusters of lymphatic endothelial cells, each of which was annotated based on well-established identity markers” – Authors should include either reference to which markers were used or, better yet, a heatmap or dotplot showing expression levels for marker genes in each annotated cell type.  Response: Cluster identification for our dataset was based on findings from previously published studies that have deeply characterized of stromal subsets (utilizing combined scRNAseq and IHC approaches of dissociated and intact lymph nodes respectively).  We aligned expression patterns of the markers validated in these studies with differential expression patterns of our identified clusters to assign subtypes.  Notable markers used for cluster identification are noted below, and bubble plots for each fibroblastic or endothelial cluster are now included in Supplementary Fig 1 of the revised manuscript as well as the methods section (these Key identifiers also listed below).  Additionally, DEG lists were included in Supplementary table 1 (for fibroblast subclusters) and Supplementary table 2 (for endothelial subclusters).   Lymphatic Endothelial Subsets of the LN were annotated based on characterization work from den Braanker, et al. 2021 (PMID: 34769408) and Fujimoto, et al. 2020 (PMID: 32251437).  Key identifiers for each subset include the following:  cLEC :  ackr4+, cd36+, edn1+, anxa2+.  fLEC : lyve1+, glycam1+, coch+, madcam1+  Medullary LEC :  lyve1+, marco+, il33+, itg2b+  Cortical LEC :  mrc1+, ptx3+, reln+.  Fibroblastic Subsets of the LN were annotated based on characterization collective work from multiple studies including Rodda, et al. 2018 (PMID: 29752062), Pikor et al. 2021 (PMID: 32424359) and Beuchler et al. 2021 (PMID: 33981032).  Key identifiers for each subset include the following:  TRC :  ccl19+, ccl21+, grem1+  BRC :  cxcl13+, cr2-, madcam1-, tnfsf11-, ch25h+  FDC :  cxcl13+, madcam1+, coch+, cr2+, tmem119+, pthlh+  MRC :  madcam1+, tnfsf11+, enpp2+  MedRC :  cxcl12(hi), inmt+, stc1+, bst1-  PRC :  cd34+, fndc1+, pi16+, dpp4+  Activated SC :  nr4a1+.

(Major) Reviewer Comment 3:  “Perivascular cells (PRCs), though not localized to the SCS, likewise exhibited significant transcriptomic changes, which may reflect support of vascular expansion” – While entirely possible, it is unclear how the observed increase in the numbers of DEGs suggests support of vascular expansion. Authors should analyze more deeply the genes that are differentially expressed (e.g., GSEA, targeted analysis of known proangiogenic factors expressed by PRCs, etc.) before making this claim.   Response: This was a speculative comment due to the association of PRCs with LN vasculature, and the observed increase in BEC proliferation (fig1A). We have tempered this statement in the revised manuscript and included references to studies outlining potential function of PRCs.  Importantly, we have also revised our discussion of PRC, in which we describe them as “not localized to the SCS”.  Further review of the findings by Rodda et al. 2018 in which they describe localization of CD34+ stromal cells not only in the perivascular compartment, but also within the LN capsule.  Transcriptional features identifying our PRC subset align with these CD34 SCs described by Rodda et al. and Sitnik et al. 2016.   We have additionally included the full DEG list of PRCs in tdLN vs ndLN in the newly added Supplementary table 3. 

(Major) Reviewer Comment 4:  The Authors make many claims throughout the manuscript about proliferative expansion of certain fibroblast and endothelial cell subsets in tumor-draining lymph nodes, but do not show any direct evidence of this increased proliferation. Since annotating proliferative cells via Mki67 and Hells expression in scRNA-seq data is possible, I recommend the Authors compare the proportions of proliferative subsets in all experimental groups to test/strengthen these claims.  Response: Increased cell counts of each major stromal cell population (fibroblast, LEC, and BEC) within the dLN vs ndLN (seen in Fig1A) provides us some direct evidence of proliferative expansion of each of these broader populations.  Within each stromal cell type, we then identify changes in the representation of subsets within in the scRNAseq dataset.  For instance, “Fibroblasts” are presumed to have proliferated due to the >2 fold increase in cell numbers relative to matched non-draining LN (fig1A).  Then, within all fibroblasts, there is then a higher proportion of MRC, PRC, and BRC in the dLN relative to the ndLN (Fig 1C), a differential which is not present in the control conditions of mice inoculated with fixed tumor cells (Fig 1D).  This differential in subset representation is also visually evident in the newly added Supplementary figure 2.  Unfortunately we do not have the capacity to measure absolute cell counts of individual subsets, and thus proliferative expansion is inferred from the combination of these above observations.  Additionally, ki67+ cells are indeed present in our dataset, but constitute a small proportion of every cell population present.  In our experience, ki67 has rarely been a reliable or quantitative measure of population expansion in tissues.

(Major) Reviewer Comment 5:  The Authors do not describe how they handled the removal of doublets during scRNA-seq analysis. The presence of doublets could significantly confound data interpretation – particularly in instances where the underlying cell type distributions between samples are known to differ, as in this case – and, thus, needs to be addressed to ensure the observations are not artefactual in nature.  Response: We carefully assessed the quality control metrics of the dataset, including the number of RNA molecules (UMIs) and the number of genes detected per cell. All values fell within the expected range for single cells, as defined in Seurat’s documentation, and we did not observe any anomalies indicative of doublets. Based on these observations, we do not suspect the presence of doublets in our dataset at a level that would confound the interpretation of our results. This information has been added to the methods description in the revised manuscript.

(Major) Reviewer Comment 6:  In the Methods section, the Authors state that “The present study used tissues collected from naïve mice (n=4), mice bearing live MC38 tumors (n=3), or fixed MC38 tumor cells (n=3)” and then “No animals were excluded from analysis.” However, the described scRNA-seq analyses do not incorporate insights gained from the naïve mouse samples. Comparing naïve and fixed tumor samples could provide key insight for distinguishing the effects of live tumors on the tdLN. Moreover, including comparisons between naïve and fixed tumor samples would be critical for pinpointing the observed effects that are specifically due to antigenic challenge. Alternatively, if naïve samples were not used, the Methods section should be edited to clarify this point.  Response: We appreciate the careful review and thoughtful suggestion.  However naïve samples were not included in this assessment.  The methods section was corrected to reflect the proper cohort used in this analysis.

(Major) Reviewer Comment 7:  In Fig. 2G, the authors show genes that are specifically enriched in tdLN MRCs and fLECs. How were these genes identified? Is Il33 expression between tdLNs isolated from mice harboring live and fixed tumors statistically-significantly different? The up-regulation between tdLN and ndLN is clear in both live and fixed tumor settings, but tdLNs in fixed tumor samples also increase expression of Il33 compared to ndLNs, suggesting that this observation may not be live tumor-specific. Response: The genes in Fig. 2G were identified by performing DEG analysis across dLN and ndLN of the live and fixed tumor condition.  Genes which were commonly upregulated in both the MRC and fLEC of specifically the live tdLN subset were graphed in the bubble plot.  The mean expression of IL33 in the dLN of mice receiving fixed tumor cells is overall slightly elevated over the corresponding ndLN, but did not reach statistical significance.  A statistically significant change in expression was only seen in the dLN vs ndLN of live tumor-bearing mice. 

(Minor) Reviewer Comment 1:  Authors state “In the pre-metastatic lymph node, expansion of both endothelium and FRCs has been documented” – true, but authors should cite appropriate literature references.  Response: Citation has been added.

(Minor) Reviewer Comment 2:  How did authors verify that MC38 tumor cells were successfully fixed after 4% PFA exposure?   Response: No tumor growth was observed in mice receiving fixed tumor cells at the experimental endpoint.

(Minor) Reviewer Comment 3:  Fig. 1A: Authors should clarify in the legend the reference point they used to calculate relative cell count. Imagining it was the average of the non-draining LN, but should be explicit to avoid confusion. Authors should also clarify how each cell type was identified using their flow cytometry panel.  Response: Cell counts are presented as relative to the average of corresponding non-draining lymph nodes.  Cell populations were identified by flow cytometry with the following markers:  Fibroblastic Reticular (CD45-CD31-PDPN+), Lymphatic Endothelial (CD45-, CD31+, PDPN+), and Blood Endothelial (CD45-, CD31+, PDPN-).  This information has been added to the figure legend in the revised manuscript.

(Minor) Reviewer Comment 4:  Fig. 1B and E: Authors do not ever define some of the acronyms used here (e.g., TRC, FDC)  Response: The following definitions have been added to the text:  Fibroblastic populations include T-zone Reticular Cells (TRC), B zone Reticular cells (BRC), Marginal Zone Reticular Cells (MRC), Follicular Dendritic Cells (FDC), Medulary Reticular cells (MedRC), Perivascular Reticular cells (PRC), and activated stromal cells (Act SC).  Endothelial cell subsets include ceiling LECs (cLEC), floor LECs (fLEC), Medulary LECs (MedLEC), and Cortical LECs (CorLEC). 

(Minor) Reviewer Comment 5:  Page 7: “In this context, Increased conduit thickness…” – ‘Increased’ should not be capitalized.  Response: Typo has been corrected

(Minor) Reviewer Comment 6:  Fig. 2B/E/F – authors should be explicit about what the volcano plot colors scheme to ensure clarity of interpretation. Also, are the p-values presented raw or adjusted?  Response: The cutoffs for highlighted genes include a log2 fold change of 0.5 or above (horizontal dashed lines) and adjusted p-value above 0.001 (indicated by the vertical dashed lines).  This description is included in the figure legend of the revised manuscript.

(Minor) Reviewer Comment 7:  As far as I can tell, the Authors use increased expression of fibronectin by MRCs in the tdLN to support the claim that they are induced into a desmoplastic CAF-like state. This may indeed be the case, but more thorough analyses/discussions are needed to sufficiently support this claim. I would suggest leveraging publicly-available scRNA-seq data of desmoplastic CAFs (if such data exists) to assess similarly in transcriptional signatures. Alternatively, the language can be edited to lessen the claimed connection.  Response: While we do single-out Fn1 in the text, we note that this is one of several genes differentially expressed in MRCs of the tumor dLN that relate to matrix deposition or tissue remodeling.  Many others are additionally labeled in figure 2b. Additionally, both ECM organization and wound-healing are prominent in the pathway analysis as shown in Fig2a.  For additional clarity, we have also included the full DEG list of MRC, BRC, fLEC, and PRC populations in the tdLN vs ndLN in Supplementary table 3 of the revised manuscript. 

We feel comfortable drawing the comparison to desmoplastic CAFs given that these indicators of altered matrix/remodeling arise within the context of the tumor-draining LN.  However, to soften the claim, we have changed the wording of the section title from “reminiscent of desmoplastic CAFs” to instead read “reminiscent of the desmoplastic nature of CAFs”, as we acknowledge that ECM production/tissue remodeling are only one facet of CAF function.  We do not want to make the claim that the transcriptional changes we see indicate that MRCs have become CAFs, but rather that they have taken on a key function typically associated with CAFs.

(Minor) Reviewer Comment 8:  “Indeed, genetic mouse models of homeostatic chemokine disruption exhibit impaired immunity.” Need reference.  Response: Relevant references have been added.

(Minor) Reviewer Comment 9:  Should correctly note gene/protein identifiers (e.g., italics for gene names, proteins capitalized, etc.).  Response: We have reformatted gene/protein names to match convention.

(Minor) Reviewer Comment 10:  The 30% mitochondrial gene expression threshold is quite high. Authors should interrogate whether any high pMito clusters were retained in analyses that could confound interpretation.  Response: We carefully examined the distribution of mitochondrial gene expression across cells and found no evidence of high mitochondrial clusters that would indicate stressed or damaged cells. Additionally, downstream analyses, such as clustering and differential expression, did not reveal any confounding effects that could be attributed to mitochondrial gene expression. Based on these observations, we are confident that the retained cells are biologically relevant and do not introduce bias into our results.

 10.5256/f1000research.159080.r309554 Reviewer response for version 1 Luther Sanjiv 1 Referee https://orcid.org/0000-0002-8758-957X Department of Immunobiology, University of Lausanne, Lausanne, Switzerland

Competing interests: No competing interests were disclosed.

 24 8 2024 Copyright: © 2024 Luther S 2024 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

Piquet and coauthors address in this study the interesting question of how the lymph node (LN) stroma changes or is reprorammed upon draining a tumor site with alterations presumably contributing to tumor metastasis to this site along with immune suppression. They present insightful scRNAseq datasets describing how different murine LN stromal cell types react to either a live or a fixed inoculation with the MC38 colorectal cancer cell line, when compared to a nondraining LN. While both types of inoculation lead to a comparable LN swelling, including a strong amplification of the LN fibroblasts, lymphatic endothelium (LEC) and blood endothelium (BEC), consistent with a strong inflammatory response in both settings, the impact of live tumor cell inoculation shifted much more the proportions and transcriptional profile of distinct subsets of stromal cells compared to injection with fixed tumor cells, pointing to tumor-specific factors driving these changes. The transcriptional shifts were most pronounced among two fibroblast subsets, termed MRC and BRC, as well as among two LEC subsets that are also those thought to be most exposed to the incoming lymph and its factors. The resource provided by the authors should be of great interest to all scientists interested in tumor metastasis or anti-tumor immunity. The manuscript is well written, the data are of high quality and appropriately displayed; and the findings are discussed in an insightful way.

Major points:

1) The clustering of the different subsets of fibroblasts, or lymphatic or blood endothelial cells is presented without any tables or heatmaps showing the top differentially expressed genes for each subset, so that one can judge the cells clustered for example as MRC or BRC. Currently one would be forced to redo all the bioinformatic work again, based on the individual data files provided. Such additional data would help considerably the expert reader to assess and interpret the data and the labeling of the clusters, especially as the frequencies of the FDC and MRC populations look higher compared to other scRNAseq data sets (eg. Rodda/Cyster Immunity 2018) or relative to the analysis by flow cytometry (eg. Huang/Luther, PNAS 2018), also relative to TRC and MedRC. Any histological stainings (with antibodies or by ISH) validating the key scRNAseq data, including the subset identification/clustering would greatly enhance the value of the data; eg. are the MRC-like cells really still restricted to the SCS-lining area, as are the fLEC-phenotype cells

 2) Similarly, no tables are provided as supplement to list the DE genes in the various settings as only part of them are mentioned in the figure 2, and they are limited to the ones showing an increase with none labeled showing a decrease in a live tumor cell draining LN (vs the nondraining LN). This reviewer can imagine that labeling also part of the downregulated genes may render the data and its discussion more complex but then at least tables containing these data should be provided to allow an interested reader to look into them without need to reanalyze all the data. They may contain data relevant for the main question of the paper; eg. how does the LN suppress anti-tumor immunity or get prepared for tumor metastasis. Please improve also the description in the legend so that one understands why the log2 fold changes is negative for transcripts enriched in drLN.

3) The draining LN response is investigated on day 12 after tumor cell inoculation, comparing live tumor cell vs fixed tumor cell injection which is an elegant approach. As the same number of cells is injected in the setting of live cells as of fixed cells, but the final number of live tumor cells is not stated for d12 (but is presumably much higher, by a factor of several fold), the authors should probably acknowledge in the discussion the large difference in final tumor cell material in the two settings that is likely to contribute to the transcriptional differences observed, besides larger differences in the tissue cells of the primary injection site during the time period upto d12. Thus, the difference in factors infusing the LN is not only due to the live tumor cells but also due to the difference in final tumor cell numbers and their differential effects on the peripheral tissue in the two settings.

4) It is unclear whether the authors verified if any of the live tumor injected mice showed tumor cell metastasis to the draining LNs which got analyzed. That could have been analyzed by flow cytometry (CD45- cells with specific FSC/SSC characteristics; or a non-stated epithelial marker) or by other means. This information would be valuable to know given the likely difference between premetastatic vs metastatic LNs stroma.

Minor points:

The results start with a rather lengthy intro and part of it could be incorporated into the introduction where some of these points are already raised.

BRC are claimed to populate the interfollicular space; please explain the rational or the publication stating this.

Enzymatic digestion (‘as previously described’): please add reference with a more detailed protocol as this aspect is key for appropriate reproduction

Mention at beginning that MC38 are a colorectal carcinoma cell line

In the conduit section: conduits are not simply fibers (although they appear that way histologically and contain fibers) but have also basement membranes thereby forming true tubes/channels; thus the term fiber is not ideal. Similar, the term ‘transendothelial channels’ is not appropriate even when talking of the SCS where occasionally such channels traversing the SCS can be observed but the majority of conduits do not seem to traverse the endothelium.

‘Exhibited of’ (please drop ‘of’)

IL33 discussion: may add another reference showing role of IL33 for memory T cell response (Marx et al.,2024 [Ref 1])

Please spell out LDHA and PKM at first mentioning and possibly the reaction they catalyze to inform the non-specialist reader.

 Fig.1B and D: I assume these Umaps show data compiled from all 4 groups of mice (please state clearly in legend); it would be of interest to display the Umaps of each group separately (in a supplementary figure) for readers to understand the stromal cell clustering in a 2D space for the 4 groups.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Yes

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Yes

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Yes

Reviewer Expertise:

lymph nodes, stromal cells, immunity, scRNAseq

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

References : The alarmin interleukin-33 promotes the expansion and preserves the stemness of Tcf-1+ CD8+ T cells in chronic viral infection. Immunity .2023;56(4) : 10.1016/j.immuni.2023.01.029 813-828.e10 36809763 10.1016/j.immuni.2023.01.029 Chang Jonathan Oncology, Novartis, Cambridge, MA, USA

Competing interests: No competing interests were disclosed.

 28 3 2025

(Major) Reviewer Comment 1: The clustering of the different subsets of fibroblasts, or lymphatic or blood endothelial cells is presented without any tables or heatmaps showing the top differentially expressed genes for each subset, so that one can judge the cells clustered for example as MRC or BRC. Currently one would be forced to redo all the bioinformatic work again, based on the individual data files provided. Such additional data would help considerably the expert reader to assess and interpret the data and the labeling of the clusters, especially as the frequencies of the FDC and MRC populations look higher compared to other scRNAseq data sets (eg. Rodda/Cyster Immunity 2018) or relative to the analysis by flow cytometry (eg. Huang/Luther, PNAS 2018), also relative to TRC and MedRC. Any histological stainings (with antibodies or by ISH) validating the key scRNAseq data, including the subset identification/clustering would greatly enhance the value of the data; eg. are the MRC-like cells really still restricted to the SCS-lining area, as are the fLEC-phenotype cells. 

Response: Cluster identification for our dataset was based on findings from previously published studies that have deeply characterized of stromal subsets (utilizing combined scRNAseq and IHC approaches of dissociated and intact lymph nodes respectively).  We aligned expression patterns of the markers validated in these studies with differential expression patterns of our identified clusters to assign subtypes.  Notable markers used for cluster identification are noted below, and bubble plots for each fibroblastic or endothelial cluster are now included in Supp Fig 1 of the revised manuscript as well as the methods section (these Key identifiers also listed below).  Additionally, DEG lists were included in Supplementary table 1 (for fibroblast subclusters) and Supplementary table 2 (for endothelial subclusters).   Lymphatic Endothelial Subsets of the LN were annotated based on characterization work from den Braanker, et al. 2021 (PMID: 34769408) and Fujimoto, et al. 2020 (PMID: 32251437).  Key identifiers for each subset include the following:  cLEC :  ackr4+, cd36+, edn1+, anxa2+.  fLEC : lyve1+, glycam1+, coch+, madcam1+  Medullary LEC :  lyve1+, marco+, il33+, itg2b+  Cortical LEC :  mrc1+, ptx3+, reln+.  Fibroblastic Subsets of the LN were annotated based on characterization collective work from multiple studies including Rodda, et al. 2018 (PMID: 29752062), Pikor et al. 2021 (PMID: 32424359) and Beuchler et al. 2021 (PMID: 33981032).  Key identifiers for each subset include the following:  TRC :  ccl19+, ccl21+, grem1+  BRC :  cxcl13+, cr2-, madcam1-, tnfsf11-, ch25h+  FDC :  cxcl13+, madcam1+, coch+, cr2+, tmem119+, pthlh+  MRC :  madcam1+, tnfsf11+, enpp2+  MedRC :  cxcl12(hi), inmt+, stc1+, bst1-  PRC :  cd34+, fndc1+, pi16+, dpp4+  Activated SC :  nr4a1+.     We appreciate the reviewers’ careful consideration of population frequencies within our dataset relative to the above referenced studies.  While we are unable to conclusively account for these discrepancies, we hypothesize that differences in frequencies may relate to 1) distinct methodology for tissue dissociation, 2) differences in cell enrichment prior to scRNAseq, and 3) differences in which specific peripheral LNs are included in the dataset.  Importantly, we do not make any conclusions or assumptions that are specifically about the baseline abundance of stromal subsets relative to other subsets, but rather all of the comments we make regarding population frequency is centered on changes of subset representation between tumor-draining and corresponding non-draining LNs, and thus these potential variables resulting from sample preparation methodology are internally controlled.  We are unfortunately unable to provide comprehensive IHC to corroborate the spatial representation of these markers in our dataset, but IHC of LN tissues exploring many of the above-mentioned markers has been performed in the referenced publications.  Regarding the specific example of MRC vs BRC, the primary distinction between these clusters was a lack of madcam1 and tnfsf11 on the subset we identify as “BRCs” relative to MRCs.  Both clusters expressed cxcl13 and exhibited low expression of markers typically associated with TRC (including ccl19/ ccl21).

(Major) Reviewer Comment 2:   Similarly, no tables are provided as supplement to list the DE genes in the various settings as only part of them are mentioned in the figure 2, and they are limited to the ones showing an increase with none labeled showing a decrease in a live tumor cell draining LN (vs the nondraining LN). This reviewer can imagine that labeling also part of the downregulated genes may render the data and its discussion more complex but then at least tables containing these data should be provided to allow an interested reader to look into them without need to reanalyze all the data. They may contain data relevant for the main question of the paper; eg. how does the LN suppress anti-tumor immunity or get prepared for tumor metastasis. Please improve also the description in the legend so that one understands why the log2 fold changes is negative for transcripts enriched in drLN.

Response: Tables including the complete DEG lists of the indicated subpopulations in tdLN vs ndLN have been included in Supplementary table 3 of the revised manuscript.  Regarding log₂ fold change interpretation: In volcano plots (B, D, F), negative log₂ fold change values indicate genes that are more highly expressed in draining lymph nodes (dLNs) relative to non-draining lymph nodes (ndLNs). This is because differential expression was calculated using ndLNs as the reference group. As a result, transcripts upregulated in dLNs are represented on the left side of the plot (negative fold change), while genes enriched in ndLNs appear on the right side (positive fold change).

(Major) Reviewer Comment 3:  The draining LN response is investigated on day 12 after tumor cell inoculation, comparing live tumor cell vs fixed tumor cell injection which is an elegant approach. As the same number of cells is injected in the setting of live cells as of fixed cells, but the final number of live tumor cells is not stated for d12 (but is presumably much higher, by a factor of several fold), the authors should probably acknowledge in the discussion the large difference in final tumor cell material in the two settings that is likely to contribute to the transcriptional differences observed, besides larger differences in the tissue cells of the primary injection site during the time period upto d12. Thus, the difference in factors infusing the LN is not only due to the live tumor cells but also due to the difference in final tumor cell numbers and their differential effects on the peripheral tissue in the two settings. 

Response: The reviewer is correct, in that mice inoculated with “live” tumor cell will invariably harbor more tumor cells (and thus more overall tumor antigen) at the time of harvesting the associated LN due to continued tumor growth.  This is unfortunately difficult to control for experimentally.  We have noted this concern as a potential caveat in the revised manuscript under the subheading “Caveats and study limitations”.

(Major) Reviewer Comment 4:  It is unclear whether the authors verified if any of the live tumor injected mice showed tumor cell metastasis to the draining LNs which got analyzed. That could have been analyzed by flow cytometry (CD45- cells with specific FSC/SSC characteristics; or a non-stated epithelial marker) or by other means. This information would be valuable to know given the likely difference between premetastatic vs metastatic LNs stroma. 

Response: Overt metastatic tumor growth was not observed in any live tumor-bearing mice, nor were any clusters identified as cancer cells within the total 50,381 cells sequenced in this dataset.  However, because these samples were enriched for stromal populations, it is likely that cancer cells (if present) would be lost following sample enrichment, particularly if originally present in low numbers.  As the reviewer suggested, we looked further into our (unenriched) flow cytometry data of dissociated TdLN and ndLNs to identify presence of tumor cells.  While no epithelial marker is included in this panel, MC38 cancer cells can be identified as CD45-, PDPN-, CD31-, PD-L1+ (PMID: 28507803).  Indeed we do find small numbers of these cells specifically in the tumor-draining lymph node.  Average cell counts were 114 total cells in TdLN compared to 4 in NdLN (p value = 0.0085).  This analysis has been added to the revised manuscript as supplementary figure 3.  We believe that presence of cancer cells at such low numbers, may reflect either shed/migrating tumor cells or the presence of early micrometastases.  Discussion of this topic is likewise added to the revised manuscript, and the above-mentioned flow cytometry analysis has been included as Supplementary Fig. 3.

(Minor) Reviewer Comment 1:  The results start with a rather lengthy intro and part of it could be incorporated into the introduction where some of these points are already raised.

Response: Per the reviewer’s suggestion, we have removed the first paragraph of the results.  This paragraph reiterated the role of tissue stroma in tumor progression and was mostly redundant. 

(Minor) Reviewer Comment 2:  BRC are claimed to populate the interfollicular space; please explain the rational or the publication stating this.

Response: While the exact subsetting criteria and nomenclature varies across different publications to date, the population we refer to as “BRCs” most closely aligns with interfollicular reticular cell populations described as Ccl19lo, Ch25h+, Cxcl13+ in Pikor et al. (PMID: 32424359) and Rodda et al. (PMID: 29752062).  This population (or a proportion of this population) is considered to reside in the interfollicular space due to IHC and RNAscope data included in those respective publications.  Regarding the nomenclature, this population of reticular cells is considered a subset of “B cell interacting” reticular cells in Pikor et al.  Whereas it is referred to as a subset of TRC in Rodda et al.  We use the BRC nomenclature due to low Ccl19 and high Cxcl13 expression, suggestive of a B-cell association. 

(Minor) Reviewer Comment 3:  Enzymatic digestion (‘as previously described’): please add reference with a more detailed protocol as this aspect is key for appropriate reproduction. 

Response: Reference added to revised manuscript

(Minor) Reviewer Comment 4:   Mention at beginning that MC38 are a colorectal carcinoma cell line.

Response: mention of MC38 as colorectal carcinoma cells has been added to the revised manuscript.

(Minor) Reviewer Comment 5:   In the conduit section: conduits are not simply fibers (although they appear that way histologically and contain fibers) but have also basement membranes thereby forming true tubes/channels; thus the term fiber is not ideal. Similar, the term ‘transendothelial channels’ is not appropriate even when talking of the SCS where occasionally such channels traversing the SCS can be observed but the majority of conduits do not seem to traverse the endothelium. 

Response: We have changed the description of LN conduits as “a branched network of fibers” to state instead “a branched network of fibrous structures”.  While our discussion of conduit networks does not fully address the intricacies of conduit anatomy, we have included relevant references to more in-depth exploration of this topic.  Regarding the statement of transendothelial channels, we discuss the nature of these structures as described in the referenced citation (Ranktakari et al 2015).  We do not make any statement speculating whether any conduits themselves traverse the endothelium.  Rather, we discuss the findings of the referenced study which explore the idea of size-restriction as a function of PLVAP fibrils that exist within transendothelial channels.  To clarify, in this publication, “transendothelial channels” and conduits are separate structures.  The former spans the length of sinus-lining endothelial cells, while the latter underlies the endothelium (as such this finding is not reliant on whether any or all conduits actually traverse the endothelium).  This publication suggests that molecules pass from LN sinus into the LN parenchyma through transendothelial channels, and upon traversing the endothelial lining, can subsequently access the underlying conduit network freely.  However, the structure of the transendothelial channel (dictated by PLVAP) is the gating feature for larger molecules.  We reference this finding because any structural disruption to the sinus-lining endothelium could theoretically perturb the above described regulation of molecule sizes that are allowed to pass into the conduit network.  This would be one potential explanation to the loss of size-exclusivity noted by Riedel et al within the tdLN.

(Minor) Reviewer Comment 6:  ‘Exhibited of’ (please drop ‘of’). 

Response: Typo has been corrected.

(Minor) Reviewer Comment 7:   IL33 discussion: may add another reference showing role of IL33 for memory T cell response (Marx et al.,2024 [Ref 1]). 

Response: This reference has be added.  Thank you for the suggestion

(Minor) Reviewer Comment 8:  Please spell out LDHA and PKM at first mentioning and possibly the reaction they catalyze to inform the non-specialist reader. 

Response: Definitions have been added.

(Minor) Reviewer Comment 9:  Fig.1B and D: I assume these Umaps show data compiled from all 4 groups of mice (please state clearly in legend); it would be of interest to display the Umaps of each group separately (in a supplementary figure) for readers to understand the stromal cell clustering in a 2D space for the 4 groups. 

Response: Per the reviewer’s suggestion, we have generated a new supplementary (Supplementary figure 2), which exhibits the distribution of cells from each treatment group within the corresponding endothelial (Supplementary Fig 2A) and fibroblast (Supplementary Fig 2B) Umaps.  Visualization of cell clustering in this 2D space further reflects the increased representation of specific stromal subclusters (including fLEC, MRC, and PRC) from the tumor-draining LN.