Dynamic Immune Cell Composition, Phenotypes, and Signaling in an Engineered Metastatic Niche.

1
Introduction
Metastasis is the stage of disease at which the prognosis substantially worsens for many types of cancer. In breast cancer, metastasis is associated with a 5‐year survival rate of approximately 30%, while early‐stage, localized breast cancer has a 5‐year survival rate of 99% (Siegel et al. 2025). The formation of metastatic lesions occurs through a multi‐step process that includes the mobilization of immune cells from the bone marrow to distant tissues, thus establishing a permissive microenvironment, called the metastatic niche (MN) (Peinado et al. 2017). The seeding of immune cells and their associated signaling alters native tissues, such as the lungs in breast cancer, which can lead to increased vascular leakiness, alterations in resident fibroblasts, and additional recruitment of immune cells (Peinado et al. 2017). Over time, circulating tumor cells are recruited to the MN via altered cellular signaling and, in later disease stages, form proliferative metastases (Hapach et al. 2019). The presence of these tumor cells can disrupt the normal function of the organ and contribute to decreased survival.
The detection of metastatic tumors often occurs late in disease progression, with patients self‐reporting symptoms associated with organ dysfunction or pain (Pesapane et al. 2020). Radiographic imaging followed by tissue biopsy is the diagnostic standard for metastatic disease, yet there are several limitations in this approach. First, current imaging modalities can only detect macroscopic metastatic tumors, which are often indicative of late‐stage, untreatable disease. Second, metastatic lesions are stochastic in nature, making visualization and biopsy difficult. Furthermore, biopsy of vital organs to confirm the presence of metastatic lesions has associated risks (Wiener et al. 2013). More recently, liquid biopsy has emerged as a diagnostic method that measures circulating tumor DNA (ctDNA) to monitor for disease progression or recurrence (Vlataki et al. 2023). ctDNA has been correlated with increased tumor size, meaning that patients with elevated ctDNA levels already have radiologically detectable, advanced‐stage tumors. In fact, the recent ZEST trial (NCT04915755) was permanently discontinued due to an insufficient number of patients with elevated ctDNA levels but no radiological evidence of disease. Altogether, current diagnostic tools are unable to detect metastatic tumors prior to substantial systemic disease burden.
Towards addressing this clinical gap, we are pursuing an alternative disease identification strategy that is based on the immune responses associated with metastasis. We have developed a microporous scaffold, formed from either polycaprolactone or poly(lactide‐co‐glycolide) (Azarin et al. 2015; Rao et al. 2016), that serves as a synthetic metastatic niche. In a murine model of triple‐negative breast cancer (TNBC), scaffolds are implanted subcutaneously and are infiltrated by immune cells, recruit tumor cells, and are easily biopsied (Oakes et al. 2020; Rao et al. 2016). Further, we have established a correlation between gene expression in the lung and that in the scaffold (Oakes et al. 2020; Orbach et al. 2022; Rao et al. 2016), suggesting that the scaffold has the potential for disease surveillance and monitoring disease progression. The use of the scaffold has been applied to non‐metastatic (e.g., 4T07) and metastatic (e.g., 4T1) TNBC cell lines, and a gene signature can delineate disease aggressiveness (Orbach et al. 2024). For the metastatic 4T1 line, we observed that tumor cells are recruited to the scaffold prior to the lung, yet the tumor cell numbers are relatively stable in the scaffold but increase in the lung. The steady level of tumor cell numbers has been referred to as dormancy, which could imply a lack of both recruitment and proliferation or may indicate that the rate at which cells are recruited is balanced with tumor cell death, the latter having been termed metastatic dormancy (J. Wang et al. 2023).
Herein, we investigated the dynamic cellular phenotypes and signaling within the synthetic metastatic niche, aiming to capture the phenotypes that underlie metastatic dormancy. We have previously reported on the dynamic cell populations and gene expression within the lung, which demonstrated that the lung MN has an anti‐cancer phenotype that transitions to pro‐cancer as disease progresses, a process primarily attributed to myeloid cell accumulation and phenotype shifts (Orbach et al. 2022). Herein, we analyze the synthetic niche and assess its dynamic cell populations and gene expression. The scaffold microenvironment contains numerous innate immune cell types (e.g., neutrophils and monocytes) that transition from anti‐tumor to pro‐tumor phenotypes as disease progresses, though this transition is more gradual than the lung. A combination of sequencing analysis and cellular assays was performed to analyze the dynamic processes within the scaffold. This synthetic niche holds promise as a surrogate tissue that is accessible and can provide cells for analysis that represent the distant organ pathophysiology.

2
Results
2.1
Scaffolds Support Dynamic Immune Cell Subtypes Throughout Disease Progression
We initially investigated the cellular dynamics within the scaffold across multiple stages of disease progression (Figure 1a). The 4T1 model is a highly metastatic TNBC cell line that was derived from BALB/c mice, with the lung as the initial metastatic organ (Aslakson and Miller 1992). Scaffolds were collected at time points corresponding to distinct stages of disease progression: Day 7, which represents pre‐metastasis; Day 14, which represents micro‐metastasis; and Day 21, which represents macro‐metastasis (Oakes et al. 2020). Scaffolds were also collected from time‐matched healthy controls (Day 0). Cell types identified at the scaffolds include monocytes, macrophages, neutrophils, dendritic cells (DCs), natural killer (NK) cells, and T cells (Figure 1b,c, Supporting Information S1: Figure S1). These cell populations are dynamic through disease progression at both the scaffold and lung (Figure 1d,e). Both tissues exhibit a decrease in macrophages over time (lungs: 24% at Day 0 to 7% at Day 21; scaffold: 27% at Day 0 to 16% at Day 21) and a large influx of neutrophils from Days 0 to 21 (lung: 8%–82%; scaffold: 9%–44%). However, the scaffold has an expanded DC population and increased non‐immune cell populations compared to the lung, the latter comprising about 20% of the scaffold's total cell composition at all time points. The spleen, which recruits and distributes cells into circulation, has a cell composition that differs substantially from the scaffold and lung (Supporting Information S1: Figure S2).

2.2
N2 Neutrophils and MDSCs Drive the Scaffold's Pro‐Tumor Phenotype at Late‐Stage Disease
The expansion of neutrophils throughout metastatic progression led to an investigation of neutrophil subsets. We identified four neutrophil subsets in the scaffold: N1 neutrophils, chemotactic N1 neutrophils, N2 neutrophils, and myeloid‐derived suppressor cells (MDSCs) (Figure 2a). N1 neutrophils are pro‐inflammatory cells that were identified by increased expression of Isg15 and prostaglandin 2 (Ptgs2), and chemotactic N1 neutrophils were defined by high expression of Cxcl2, a migratory marker (Supporting Information S1: Figure S3) (Capucetti et al. 2020). N2 neutrophils secrete immunosuppressive factors, such as reactive oxygen species and matrix metalloproteinases (X. Wang et al. 2018). MDSCs are immature cells that have a similar immunosuppressive phenotype as N2 neutrophils. In our dataset, MDSCs were distinguished from N2 neutrophils by high expression of lactoferrin (Ltf) (Y. Liu et al. 2019). We found that N1 neutrophils comprise the vast majority of all neutrophil subtypes in the scaffold on Days 0, 7, and 14 (Figure 2b). On Day 14, N2 neutrophils and MDSCs account for about 10% of all neutrophils, yet by Day 21, the N2 and MDSC populations account for over 90% of all neutrophils in the scaffold, indicating the microenvironment of the scaffold is progressing from anti‐tumor characteristics towards pro‐tumor characteristics. Chemotactic N1 neutrophils are primarily responsible for inflammatory signaling and TNFɑ production (Figure 2c–e), and these changes in signaling are likely due to the decrease in chemotactic N1 neutrophils from Days 14 to 21. Furthermore, N1 neutrophils express Il1b (Figure 2f), another pro‐inflammatory gene with anti‐tumor signaling that decreases from Days 14 to 21 as MDSCs and N2 neutrophils expand. The expanded MDSC and N2 neutrophil populations at Day 21 were validated by analysis of Arg1+ pro‐tumor neutrophils in the scaffold (Figure 2g).
We next examined the changes in signaling from neutrophils across disease progression. Using single‐cell pathway analysis (SCPA) (Bibby et al. 2022), we identified 15 pathways with regulation that was significantly altered for at least two of the three time point comparisons (Day 0 vs. 7, Day 7 vs. 14, and Day 14 vs. 21) (Figure 2h). The majority of inflammatory pathways, such as TNF signaling, are significantly upregulated from Days 7 to 14 but then downregulated from Days 14 to 21. TNF family members, particularly TNFɑ, are pro‐inflammatory molecules that can activate T cells and aid in tumor suppression (X. Wang et al. 2018).

2.3
Macrophages Contribute to the Pro‐Tumor Environment in Late Disease Stages
Scaffolds exhibited expanded macrophage and DC populations compared to the lungs, which motivated a deeper analysis of these cell populations. Macrophages can be polarized into different functional states depending on the activation signals they receive and their localization within tissues (Figure 3a) (W. Zhang et al. 2024). Pro‐inflammatory macrophages (termed M1) can kill tumor cells via phagocytosis and antibody‐dependent cell‐mediated cytotoxicity (W. Zhang et al. 2024). Conversely, anti‐inflammatory macrophages (termed M2) can promote the epithelial‐mesenchymal transition (EMT) and immunosuppression (W. Zhang et al. 2024). In our sequencing data, tumor‐associated macrophages (TAMs) were distinguished by significantly upregulated expression of S100a9 and Fosb (Supporting Information S1: Figure S4). We have previously reported that the expression of S100 family members by immune cells is associated with increased metastasis and tumor cell recruitment (J. Wang et al. 2023). Fosb is one of several downstream genes previously found to be increased in TAMs (Radharani et al. 2022). Interstitial macrophages were termed tissue‐resident macrophages (Supporting Information S1: Figure S4) (Chakarov et al. 2019). We identified that the relative number of M1 macrophages peaked at Day 7 and then steadily decreased in later time points. Additionally, TAMs were identified at the scaffold on Day 21, but not at any earlier time points (Figure 3b). The percentage of Arg1+ macrophages (representing M2 macrophages and a portion of TAMs) in the scaffold on Day 21 was enriched, similar to the lungs (Orbach et al. 2022), indicating that the cell types at the scaffold become progressively more pro‐tumor (Figure 3c). For the initial time points, few signaling pathways related to inflammation were significantly altered in macrophages (Figure 3d). By Day 21, however, significant downregulation was observed with many pathways, reflecting a loss of inflammatory signaling as M2 macrophages and TAMs expand.
The analysis of DCs, which can be categorized as classical (or conventional), inflammatory, or plasmacytoid, among other classifications (Supporting Information S1: Figure S5) (Eisenbarth 2019), indicated more modest fluctuations relative to macrophages as disease progressed. Classical DCs produce high amounts of type III interferons, which lead to durable immune responses, and are highly effective at presenting antigens to CD4+ T cells (Macri et al. 2018). Antigen‐presenting DCs display upregulated histocompatibility proteins, and they are likely an activated subset of classical DCs. Inflammatory DCs (also called monocyte‐derived DCs) are derived from monocytes that differentiate in the presence of inflammation (Del Prete et al. 2023). In cancer patients, they have been associated with improved responses to therapeutics (Del Prete et al. 2023). Plasmacytoid DCs are responsible for peripheral immunity, as they release type I interferons to rapidly activate other cell types in inflamed tissues (Pinto et al. 2012). Each of these DC subtypes was identified in our sequencing data (Figure 3e). We observed that inflammatory and classical DCs comprised the majority of all DCs in healthy scaffolds (Figure 3f). At Day 7, inflammatory DCs expanded to account for about 65% of all DCs, with a concomitant decrease in classical DCs. By Day 21, though, these proportions once again resemble the proportions found in healthy scaffolds. An analysis of signaling revealed that the majority of pathways were downregulated from Days 0 to 7 (Figure 3g), similar to macrophages. One exception was the antigen processing and presentation pathway, which was upregulated. By Day 21, only a few pathways associated with DC activation become upregulated. These observations suggest that, unlike neutrophils and macrophages, DCs maintain relatively consistent pro‐inflammatory signaling throughout disease progression.

2.4
Macrophage and DC Signaling Is Replaced by Neutrophil Signaling in the Scaffold as Disease Progresses
We next characterized the dynamic signaling at the scaffold throughout disease progression. We identified 14 pathways related to cancer progression that were significantly upregulated or downregulated between consecutive time points from all enriched pathways identified through SCPA (Supporting Information S1: Figure S6). Many of these pathways have been reported for their association with TNBC progression in particular (Figure 4a) (Lin et al. 2021; Zhao et al. 2022), indicating that the scaffold microenvironment captures systemic changes corresponding with disease as early as 7 days post‐inoculation. From Days 14 to 21, the scaffold mimics changes observed in the lung at late disease stages (Orbach et al. 2022). All but four pathways (MAPK signaling, TNF signaling via NFκB, hypoxia, and apoptosis), are significantly upregulated by Day 21. Similar to the scaffold, MAPK signaling in the lungs was decreased between Days 14 and 21 (Orbach et al. 2022), which may reflect the changing cell types and inflammation during this time frame (Kyriakis and Avruch 2012). NFκB is partly responsible for M1 macrophage polarization and functionality, and TNFɑ is a key inflammatory cytokine produced by M1 macrophages (T. Liu et al. 2017); thus, a decrease in TNFɑ signaling may be associated with decreased anti‐tumor macrophage activity. Additionally, TNF receptors are associated with NFκB‐mediated activation of inflammatory T cells (T. Liu et al. 2017). This observation suggests that decreased TNFɑ signaling could also lead to reduced anti‐tumor T cell responses. The downregulation of the hypoxia and apoptosis pathways in the scaffold was unexpected, as we had previously reported that these pathways were upregulated in the lungs at Day 21 (Orbach et al. 2022). Collectively, the 14 pathways we identified reflect changes associated with disease progression. Upregulation of interferon responses, antigen processing and presentation, and the complement system on Day 21 suggests that some anti‐tumor characteristics persist in the scaffold at late disease stages.
We next applied CellChat to investigate the intercellular communication networks occurring within the scaffold throughout disease progression (Figure 4b–e). Between Days 0 and 14, macrophages and DCs have the greatest probability of connections to other cell types within the system. However, by Day 21, the number of signaling interactions for macrophages and DCs substantially declined. Neutrophils had few signaling connections at Day 0, and they progressively increase from Days 7 to 21 (Figure 4e). This dominance in intercellular signaling is consistent with the influx of neutrophils at Day 21, which actively drives pro‐tumor dysregulation at both the scaffold and lung (Orbach et al. 2022).

2.5
Microenvironment Signaling Constrains Tumor Cell Numbers in the Scaffold Relative to the Lung
Pro‐tumor neutrophils are identified in the scaffold, yet the scaffold was previously reported to not support proliferative metastases as the lung does (J. Wang et al. 2023). We thus examined signaling from the scaffold and lung microenvironment to tumor cells using a series of in vitro assays as a means to identify differential signaling that may underlie the distinct responses. 4T1 cells were either co‐cultured with bulk microenvironmental cells from the scaffold and lung or with conditioned media (CM) generated from scaffolds and lungs isolated at each time point.
CM was applied to cancer cells for measurement of transcription factor (TF) activity using a TRanscriptional Activity Cell aRray (TRACER) assay (Orbach et al. 2022). TF reporters that are involved in immune signaling (STAT3, NFκB, and STAT1) were selected for analysis. STAT3 is involved in multiple oncogenic processes, including increased PD‐L1 expression for immune escape, proliferation, angiogenesis, and invasion (W. Wang et al. 2023). NFκB upregulates anti‐apoptotic genes in cancer cells and confers resistance to immune‐related death signals (Luo et al. 2004; Verzella et al. 2020). STAT1 has traditionally been viewed as a tumor suppressor, but more recent evidence suggests that it may be oncogenic in breast cancer by promoting MDSC infiltration and PD‐L1 expression (Meissl et al. 2017). For these three TFs, activity was similar for lung and scaffold CM at Days 0, 7, and 14 (Figure 5a–c). At Day 21, the TF activity was significantly increased with the lung conditioned media. We also examined two TFs involved in the epithelial‐mesenchymal transition (EMT), HIF1 (Z. Liu et al. 2015) and ZEB1 (P. Zhang et al. 2015). For these TFs associated with EMT, activity was similar between scaffold and lung CM at Days 0 to 14. (Figure 5d,e). By Day 21, however, the TF activity for these EMT pathways was significantly increased in the lung CM compared to the scaffold CM, which is consistent with the lung having greater disease progression than the scaffold.
We next examined the phenotypic responses by the cancer cells as a consequence of the signaling from the microenvironment. Using a migration assay, tumor cells co‐cultured with Day 0 bulk cells in a Transwell had migration that was similar for the scaffold and lung, and migration increased with bulk cells from later time points for both tissues (Figure 5f). Transwell invasion assays were also performed and provided results comparable to the migration assay, with similar invasion for Day 0 co‐cultures and increased invasion using cells from later time points for both tissues (Figure 5g).
With signaling from the lung and scaffold being permissive to tumor cell migration and invasion, we next examined tumor cell proliferation with direct co‐culture of tumor cells and cells derived from lung or scaffolds. For co‐cultures involving cells isolated between Days 0 and 14, tumor cell proliferation was similar, with increases for both tissues on Day 21 (Figure 5h). Apoptosis assays based on annexin V were performed and indicated that 4T1 cells cultured in scaffold‐derived co‐cultures had similar levels of apoptosis compared to lung‐derived co‐cultures (Figure 5i). The apoptosis assays were also performed in the presence of CD8+ T cells with CM. In co‐cultures with T cells, 4T1 cells in scaffold CM underwent more apoptosis at Days 14 and 21 than 4T1 cells in lung CM, supporting greater anti‐tumor responses at later time points within the scaffold environment (Figure 5j).
We subsequently assessed tumor cell metabolism in response to the signaling from the scaffold microenvironment. Tumor cells maintain high proliferation rates partly through deregulated cellular metabolism (Hanahan 2022; Hanahan and Weinberg 2011). One measure of metabolic health is the mitochondrial membrane potential, which can be measured through a JC‐1 assay. As expected, tumor cells incubated in Day 21 lung CM exhibited poor mitochondrial health, as indicated by a low red/green ratio in the assay (Figure 5k). However, tumor cells incubated in scaffold CM generated throughout disease progression maintained their mitochondrial health, indicating that the scaffold environment does not deregulate cellular metabolism as was observed with the lung. We also examined the ratio of reduced glutathione (GSH) to oxidized glutathione (GSSG) in tumor cells. Compared to non‐pathological cells, tumor cells often have an increased GSH/GSSG ratio to compensate for their excessive production of reactive oxygen species (Kennedy et al. 2020; Traverso et al. 2013). Tumor cells incubated with scaffold CM exhibited progressively higher GSH/GSSG ratios from Days 0 to 14, similar to the lung CM (Figure 5l). However, on Day 21, the GSH/GSSG ratio of the tumor cells in the lung CM was significantly greater than that of the tumor cells in scaffold CM, consistent with the tumor cell expansion in the lung at that time point (J. Wang et al. 2023).

3
Discussion
The significant mortality associated with metastatic cancer has the potential to be improved through early detection that would allow treatment with a lower disease burden and less heterogeneity, which motivates the development of technologies for disease surveillance. Metastatic spread occurs through mobilized immune cells that colonize distant organs and subsequently recruit tumor cells (Peinado et al. 2017). We have developed a microporous polymer scaffold that integrates with the host tissue, becomes vascularized, and similarly recruits immune cells through the foreign body response. Prior research has demonstrated that the scaffold functions as an engineered metastatic niche, which is a non‐vital tissue from which cells can be sampled without substantial risk (Oakes et al. 2020; Rao et al. 2016). The synthetic niche has been fabricated from either PCL or PLG, with both demonstrating the ability to recruit a similar cohort of immune and tumor cells, though alternative materials may have disparate effects (Rao et al. 2016). This surrogate niche has recapitulated multiple aspects of the microenvironment observed within native metastatic organs (Orbach et al. 2022; J. Wang et al. 2023). Interestingly, tumor cell numbers at the scaffold remained steady throughout disease, suggesting metastatic dormancy that results from a balance of tumor cell recruitment and elimination (J. Wang et al. 2023). Herein, we applied single‐cell RNA sequencing to provide a more comprehensive analysis of the immune cell dynamics throughout disease progression. We identified multiple immune and cellular processes within the scaffold that correspond with the biology of metastatic disease. At early and intermediate stages of metastatic disease, we observed recruitment of immune cell subtypes associated with cancer, tumor cell migration capacity, activity of pro‐metastatic transcription factors, and tumor‐microenvironment‐associated inflammation. At an advanced stage of disease, the engineered metastatic niche continues the trend towards disease progression yet has fewer pro‐tumor characteristics than the native metastatic niche. Collectively, the engineered metastatic niche provides a tool to analyze the dynamics and mechanisms of metastatic progression.
The early stages of disease progression were marked by increasing percentages of neutrophils and declining levels of macrophages, with signaling following a similar trend. Innate immune cells, in particular, have a central, dynamic role in disease progression. Within the metastatic niche, the arrival of innate immune cells prepares the site for colonization by tumor cells (Peinado et al. 2017). The microenvironment within solid organs generally has anti‐tumor functions, and the recruited immune cells initially contribute to an inflammatory environment that activates T cells (Wculek et al. 2020). Gr1 cell depletion at early stages leads to a decline in survival, consistent with an anti‐tumor function (Orbach et al. 2022). However, the environment gradually evolves towards a pro‐tumor environment that recruits tumor cells and suppresses T cell function (Wculek et al. 2020). In the scaffold, innate immune cells, including neutrophils, DCs, macrophages, and monocytes, were present across these early stages of the triple‐negative breast cancer model 4T1 (Figure 1). 4T1 is associated with an increase in neutrophil populations, and as the disease progresses, both the lungs and the scaffold become enriched in neutrophils that secrete factors such as S100A8/A9 and MMP8 (Supporting Information S1: Figure S3) (J. Wang et al. 2023). The increased presence of neutrophils led to a decline in the percentages of other cell populations, such as macrophages, though the phenotypic subpopulations of these macrophages indicate a decline in M1 macrophages and an increase in TAMs (Figure 3b). This finding is consistent with the progression from anti‐tumor responses towards pro‐tumor responses in the scaffold. Interestingly, DCs had modest changes in relative distribution across disease progression (Figure 3f). Macrophage signaling is more predominant in the scaffold than the lung, with a decline in the number of active pathways between Days 0 and 14 (Figure 4b–d). Neutrophil signaling is low initially in both the scaffold and lung, yet the number of interactions gradually increases (Figure 4b–e). Analysis of Tnf, Il1b, and S100a6/8/9 expression suggests that both the scaffold and lung become more pro‐tumor over time (Figure 2e,f) (J. Wang et al. 2023). Collectively, canonical pro‐tumor pathways became more upregulated over time in the scaffold, similar to the lungs (Figure 4a) (Orbach et al. 2022).
The last stage of disease in the model, corresponding to Day 21, is associated with significant changes in cell composition and signaling that promote immune suppression and tumor cell expansion (Oakes et al. 2020). Immune suppression within the 4T1 model is reflected in the extensive accumulation of N2 neutrophils and MDSCs (Figure 2b), which are reported to inhibit immune responses through production of reactive oxygen species and matrix metalloproteinases (X. Wang et al. 2018). Signaling from neutrophils increased through Day 14 and remained elevated at Day 21 (Figure 4b–e). The increase in neutrophil populations at these stages caused a concomitant decline in the macrophage population in both the lung and the scaffold. Consistent with this finding, the signals emanating from macrophages declined by Day 21 (Figure 4b–e). At that time point, an increase in the TAM phenotype was observed within the macrophage population. TAMs have been reported to signal to tumor cells and support processes such as tumor cell proliferation, which was increased in our studies of tumor cell culture with CM (Figure 5h). This CM also led to increases in apoptosis, with slightly greater apoptosis in the scaffolds relative to lungs, consistent with the scaffold being more anti‐tumor than the lungs (Figure 5i,j).
Signaling to tumor cells from the scaffold environment drives tumor cell recruitment and supports proliferation, yet the environment constrains tumor cell expansion through immune‐mediated cytotoxicity that is consistent with a state of metastatic dormancy. This constraining of tumor cell expansion has been observed for both the 4T1 and 4T07 models (Orbach et al. 2024). Tumor cell recruitment and invasion into metastatic sites are driven by cytokines and chemokines that are secreted within the microenvironment (J. Wang et al. 2023), and we observed that co‐culture enhanced migration and invasion (Figure 5f,g). We did not observe tumor cells in the sequencing data collected from either the scaffolds or the lungs because too few cells were sampled, but we have identified tumor cells through flow cytometry (Orbach et al. 2022; J. Wang et al. 2023). An analysis of tumor cell numbers at metastatic sites indicated that the tumor cells arrive to the scaffold prior to the native metastatic site, yet the numbers of tumor cells remain relatively steady over time (Rao et al. 2016; J. Wang et al. 2023). This stability of tumor cell numbers at the scaffold contrasts with the lung, which had modest increases in cell numbers followed by exponential increase (J. Wang et al. 2023). Changes in immune cell communication increased tumor cell migration and invasion (Figure 5f, g) and altered the TF activity within tumor cells in ways that were similar to the lungs until late disease stages (Figure 5a–e). Our analysis of signaling suggests a state of metastatic dormancy, as the environment has the capacity to recruit tumor cells over time yet also supports tumor cell apoptosis, consistent with the scaffold maintaining some anti‐tumor characteristics. Metastatic dormancy may occur transiently within the lung; however, at later time points, the lung microenvironment provides greater immune suppression and support for metabolic function and cell proliferation relative to the scaffold. This state of metastatic dormancy in the scaffold is likely transient, as the cell phenotypes are trending towards more pro‐tumor phenotypes as time progresses.
In conclusion, the engineered metastatic niche recruits immune cells that subsequently lead to the arrival and colonization of tumor cells in a manner that is similar to the native metastatic niche. The early stages of disease progression are characterized by increasing numbers of neutrophils and a corresponding decline in macrophages. Increases are observed in N2 neutrophils and MDSCs along with the ratio of M2–M1 macrophages, consistent with the progression of cells from anti‐tumor to pro‐tumor phenotypes. Signaling at these stages is influenced by the dynamic macrophage and neutrophil populations, similar to those in the lungs from early to intermediate stages of disease. The latter stages of disease are marked by significant increases in TAMs, N2 neutrophils, and MDSCs. The scaffold and the lung microenvironments had similar trends in in vitro functional assays with tumor cells until late‐stage disease, when signals from the native metastatic niche reduced tumor cell death and increased proliferation of tumor cells. A limitation of the current study was number of cells sequenced, which prevented distinguishing tumor cells or T cells due to their low percentages in the populations. Collectively, the data highlight the dynamic progression of the synthetic metastatic niche from an environment with anti‐tumor characteristics towards one that is more supportive of tumor growth, yet the scaffold microenvironment never obtains the extent of pro‐tumor characteristics observed within the native metastatic niche. These findings support the use of the scaffold as a synthetic metastatic niche that can provide comprehensive insights on disease progression.

4
Methods and Materials
4.1
Scaffold Fabrication
Scaffolds were fabricated as previously described (Oakes et al. 2020; Schrack et al. 2024). Briefly, scaffolds were created by pressing a PCL‐sodium chloride melt dispersion in a steel die for 30 s at 1000 psi. Melt dispersion was generated by mixing PCL (Lactel) and sodium chloride (250–425 µm in diameter) at a 1:30 ratio in a dual shaft mixer (Ross). The resulting disks were heated at approximately 60°C for 5 min per side to melt the polymer into a continuous structure. The scaffolds were then immersed in MilliQ water for 1.5 h to leach out salt crystals and to create a microporous structure. Prior to implantation, scaffolds were sterilized by immersion in 70% ethanol and rinsed twice with sterile water, then dried on sterile gauze and frozen at −80°C until implantation.

4.2
Scaffold Implantation
All animal studies were performed following institutional guidelines and protocols approved by the University of Michigan Institutional Animal Care and Use Committee. 7‐ to 8‐week‐old female BALB/c mice (Jackson Laboratory) were anesthetized via inhalation of 2% isoflurane and administered 5 mg/kg carprofen subcutaneously as an analgesic. The back was shaved and sterilized with betadine and ethanol wipes, and a fenestrated sterile field was draped over the surgical area. As previously described, one 1‐cm incision was made on the upper back and was used to create subcutaneous pockets in which frozen scaffolds were inserted (Oakes et al. 2020; Rao et al. 2016). Four scaffolds were implanted per mouse on the upper back away from the fourth right mammary fat pad. Incisions were closed using sterile wound clips (Reflex 7 mm, Roboz Surgical Instrument Co.). Mice were administered additional carprofen the next day and monitored for 2 weeks after surgery.

4.3
4T1 Tumor Model
Mice received orthotopic injections of 4T1‐tdTomato‐luc2 cells (Perkin Elmer) (2 million cells/50 μL phosphate buffered saline [PBS]) in the fourth right mammary fat pad 14 days following scaffold implantation. The 4T1 murine cell line is a model of triple‐negative breast cancer with spontaneous metastases to the lung and bone marrow (Pulaski and Ostrand‐Rosenberg 2000). Tumors were allowed to progress for 7, 14, or 21 days following inoculation. The body condition of the mice was monitored daily throughout the study, and mice were euthanized if the tumor diameter exceeded 2 cm, ulceration encompassed greater than 50% of the tumor surface area, tumor presence caused partial paralysis, or the mouse was found moribund (Oakes et al. 2020).

4.4
Tissue Processing
At 7, 14, and 21 days post‐inoculation, lungs and scaffolds were extracted from mice and from time‐matched healthy controls (Day 0). Samples from three diseased mice or two healthy mice were pooled together at each time point to account for biological variability. Tissues were minced with a scalpel and incubated in RPMI containing 0.2 U/mL Liberase TL (Roche) and 150 U/mL DNase I (Sigma‐Aldrich) at 37°C for 20 min. Samples were then passed through a 70 μm mesh filter to create a single cell suspension. Cells were washed with PBS containing 30% (w/v) bovine serum albumin (BSA) and 0.5 M ethylenediaminetetraacetic acid (EDTA). Erythrocyte lysis was conducted by sequential exposure to hypotonic (0.2% w/v) and hypertonic (1.6% w/v) solutions of sodium hydroxide. Cells were washed twice before they were counted with the CountessTM Automated Cell Counter. For single‐cell RNA‐sequencing, cells were resuspended to a final concentration of 100,000 cells/mL in 0.01% (w/v) BSA in PBS. For co‐culture assays and CM, cells were resuspended in serum‐free RPMI.

4.5
Library Preparation and Sequencing
Single‐cell library preparation and RNA‐sequencing was completed as described previously (Orbach et al. 2022). Briefly, RNA was extracted from single cells and labeled using the Drop‐Seq platform (FlowJEM) as described previously (Macosko et al. 2015; Picelli et al. 2014). Approximately 1000 cell‐bead pairs were processed for sequencing from each sample. Samples were analyzed with Illumina HiSeq sequencers and approximately 30,000 reads/cell were generated with asymmetric reads.

4.6
Single‐Cell Sequencing Analysis
We aligned raw Illumina reads to the mouse genome (mm10) and generated the digital gene expression file using the Drop‐Seq pipeline (v1.2) (Macosko et al. 2015). These files were processed using Seurat v3, developed by the Satija research group, for cell type identification and analysis (Butler et al. 2018; Stuart et al. 2019). A comprehensive Seurat object was created, encompassing all the healthy and diseased samples. Cells were filtered to include those with 200–5000 RNA fragments and less than 25% mitochondrial genes. The cell types were annotated using the gene markers shown in Supporting Information S1: Figures S1, S3, S4, and S5 and visualized using Uniform Manifold Approximation and Project (UMAP). The differentially expressed pathways between the diseased and healthy conditions over time were identified using the Single Cell Pathway Analysis (SCPA) package in R (Bibby et al. 2022). The gene sets used for pathway analysis were curated from the Molecular Signature database (MSigDB) (Liberzon et al. 2015). Differentially expressed genes that were statistically significant (p < 0.05) in at least one of the time points were reported. Cell‐cell interactions based on the expression of known ligand‐receptor pairs in different cell types were inferred using CellChat v2 (Jin et al. 2021, 2023).

4.7
Conditioned Media Generation
Isolated cells were suspended at 5 million cells/mL and cultured for 24 h in serum‐free phenol red‐free RPMI. The medium and cells were collected and centrifuged at 12,000 × g for 5 min. The supernatant was sterile filtered and stored at −80°C until use. Protein concentrations of the CM were determined using a BCA assay and samples were diluted to 300 μg/mL with phenol red‐free RPMI.

4.8
Transwell Assays
The invasion, migration, and standard apoptosis assays were completed using the Transwell system in 24‐well plates. The lung and scaffold bulk cell suspensions were added to the basal compartment at 400,000 cells/mL. 4T1 cells labeled with Deep Red CellTracker were seeded on the apical side of the membrane. For the invasion assays, 100 uL Matrigel (300 ug/mL in DMEM) was added to the Transwell inserts prior to the addition of the cells. Cultures were incubated at 37°C for 24 h before adhered cells from the apical side of the membrane were removed using a cotton swab soaked in PBS (to isolate the migrated cells) and the membranes were excised. Apoptosis staining was completed using Annexin V‐FITC for the designated samples before all membranes were fixed with 2% paraformaldehyde. The membranes were stored at 4°C until they were imaged using a Zeiss Axio Observer.Z1 inverted microscope. The number of migrated tumor cells and apoptotic cells in response to co‐culture were normalized to models using only RPMI in the basal compartment.

4.9
Reporter Arrays
A transcriptional activity cell array (TRACER) was used to identify central transcription factors in tumor cell responses. Unlabeled 4T1 cells were plated into a black 384 well plate with an existing library of transcription factor activity reporters for HIF1, NFκB, STAT1, STAT3, and ZEB1. After 48 h, growth media was exchanged for media containing 630 μM d‐luciferin. Transcription factor activity was monitored after 48 h using an IVIS Spectrum (Perkin Elmer). TRACER data were processed as previously described (Aguado et al. 2018). Briefly, activity measurements were background subtracted, normalized to the empty control reporter and log2 transformed prior to analysis, as previously described (Aguado et al. 2018).

4.10
EdU Staining
Unlabeled 4T1 cells were seeded in a 96‐well plate at a concentration of 10,000 cells/well. After the cells were allowed to adhere for 4 h, bulk microenvironmental cells were added in direct co‐culture at a ratio of 5:1 with the 4T1 cells. The EdU FITC‐labeling solution from the Click‐iT EdU Cell Proliferation Kit (Thermo Fisher) was added to the culture and allowed to incubate for 18 h at 37°C. Cultures were washed to removed non‐adherent microenvironmental cells. EdU was detected according to the manufacturer's protocol and the cultures were imaged using the Zeiss Axio Observer.Z1 inverted microscope.

4.11
JC‐1 Assay
Unlabeled 4T1 cells were seeded in a 96‐well plate at a concentration of 10,000 cells/well and allowed to adhere. After 24 h, the media was replaced with 50% CM and 50% RPMI with 10% FBS and incubated for an additional 24 h. The JC‐1 assay was completed according to the protocol for adherent cells from the Mitochondrial Membrane Potential Assay Kit (Abcam) provided by the manufacturer. Plates were read on a Biotek Synergy H1 Microplate Reader (Winooski) at Ex/Em of 475/590 for the red mitochondrial aggregates and 475/530 for the green monomers.

4.12
GSH/GSSG Assay
Unlabeled 4T1 cells were seeded in a 96‐well plate at a concentration of 10,000 cells/well and allowed to adhere. After 24 h, the media was replaced with 50% CM and 50% RPMI with 10% FBS and incubated for an additional 24 h. The GSH/GSSG‐Glo Assay (Promega) was then performed according to manufacturer instructions. Half of the samples were processed for total glutathione (GSH) and half of the samples were processed for oxidized glutathione (GSSG). The ratio of GSH/GSSG was measured for matched samples. Luminescence was quantified using the Biotek Synergy H1 Microplate Reader.

4.13
CD8+ T Cell‐Induced Apoptosis
Unlabeled 4T1 cells were labeled with Deep Red CellTracker and seeded in a 96‐well plate at a concentration of 10,000 cells/well and allowed to adhere. CD8+ T cells were extracted from the spleens of healthy mice using the tissue processing techniques described above and isolated using a CD8+ T Cell Isolation Kit (Miltenyi Biotec) and magnetic activated cell sorting. T cells were added to the 4T1 cultures at a ratio of 10:1 concurrent with CM. Cultures were incubated for 24 h, washed to remove non‐adherent T cells, stained for Annexin V‐FITC, and imaged as described above.

4.14
Statistical Analysis
Significance between samples for the in vitro assays was completed using ANOVA with the Tukey multiple comparison test. Experiments were conducted with n = 4–6 biological replicates and n = 3–4 technical replicates. All figures show mean ± SEM with α = 0.05, unless otherwise stated. English letters are used to identify statistically significant groups across temporal scaffold data. Similarly, Greek letters are used to identify statistically significant groups across temporal lung data.

Author Contributions
Conceptualization: Sophia M. Orbach, Rebecca S. Pereles, Jyotirmoy Roy, Jacqueline S. Jeruss, Lonnie D. Shea. Methodology: Sophia M. Orbach, Rebecca S. Pereles, Jyotirmoy Roy, Michael D. Brooks, Lonnie D. Shea. Investigation: Sophia M. Orbach, Rebecca S. Pereles, Jyotirmoy Roy. Visualization: Rebecca S. Pereles, Jyotirmoy Roy, Sophia M. Orbach. Funding acquisition: Sophia M. Orbach, Max S. Wicha, Jacqueline S. Jeruss, Lonnie D. Shea. Supervision: Max S. Wicha, Jacqueline S. Jeruss, Lonnie D. Shea. Writing – original draft: Rebecca S. Pereles. Writing – review and editing: Rebecca S. Pereles, Jyotirmoy Roy, Sophia M. Orbach, Lonnie D. Shea.

Conflicts of Interest
L.D.S. has patents and patent applications on the scaffold technology. All other authors declare they have no competing interests.

Supporting information

Figure S1. Scaffold cell subtype annotation. Cells isolated from scaffolds were identified by differential gene expression of canonical genes. Figure S2. Cell populations in the diseased spleen (non‐metastatic site). The scaffold, as a synthetic metastatic site, differs from the spleen in immune composition. The spleen has a higher proportion of adaptive immune cells and proliferating cells than the scaffold at all time points. Figure S3. Neutrophil subtype annotation. Neutrophil subtypes isolated from scaffolds were identified by differential gene expression of canonical neutrophil‐related genes. Figure S4. Macrophage subtype annotation. Macrophage subtypes isolated from scaffolds were identified by differential gene expression of canonical macrophage‐related genes. Figure S5. DC subtype annotation. DC subtypes isolated from scaffolds were identified by differential gene expression of canonical DC‐related genes. Figure S6. Temporal dynamics of differentially expressed pathways over time in diseased mice compared to healthy controls. Single‐cell pathway analysis identified 42 pathways that are significantly enriched at one or more time points during disease progression. Among these pathways are well‐known cancer‐related pathways such as KRAS and P53, as well as novel pathways, including metabolism‐related pathways like glycolysis and oxidative phosphorylation (OXPHOS).