Host deficiency in ephrin-A1 inhibits breast cancer metastasis

The conventional dogma of treating cancer by focusing on Background: the elimination of tumor cells has been recently refined to include consideration of the tumor microenvironment, which includes host stromal cells. Ephrin-A1, a cell surface protein involved in adhesion and migration, has been shown to be tumor suppressive in the context of the cancer cell. However, its role in the host has not been fully investigated. Here, we examine how ephrin-A1 host deficiency affects cancer growth and metastasis in a murine model of breast cancer. 4T1 cells were orthotopically implanted into the mammary fat Methods: pads or injected into the tail veins of ephrin-A1 wild-type ( ), Efna1 heterozygous ( ), or knockout ( ) mice. Tumor growth, lung Efna1 Efna1 metastasis, and tumor recurrence after surgical resection were measured. Flow cytometry and immunohistochemistry (IHC) were used to analyze various cell populations in primary tumors and tumor-bearing lungs. While primary tumor growth did not differ between  , Results: Efna1 Efna1 , and   mice, lung metastasis and primary tumor recurrence were Efna1 significantly decreased in knockout mice.   mice had reduced lung Efna1 colonization of 4T1 cells compared to   littermate controls as early Efna1 as 24 hours after tail vein injection. Furthermore, established lung lesions in mice had reduced proliferation compared to those in Efna1 Efna1 controls. Our studies demonstrate that host deficiency of ephrin-A1 Conclusions: does not impact primary tumor growth but does affect metastasis by providing a less favorable metastatic niche for cancer cell colonization and growth. Elucidating the mechanisms by which host ephrin-A1 impacts cancer relapse and metastasis may shed new light on novel therapeutic strategies. 1,2 1 3,4 3,4


Introduction
Over the past several decades, the conventional dogma of treating cancer by focusing on the elimination of rapidly dividing tumor cells has been gradually refined to include consideration of the environment in which the tumor thrives -the tumor microenvironment. The tumor microenvironment consists of both cancer cells and host stromal cells, such as endothelial cells, immune populations, and fibroblasts. Prominent discoveries regarding tumor-associated endothelium and immune cells have notably led to breakthrough therapeutic strategies with anti-angiogenic agents and immunotherapies, respectively [1][2][3][4] . Thus, understanding the host-tumor interactions involved in tumor growth and metastasis is critical for the development and application of new anti-cancer therapies.
As a result of new advancements in targeted and immunotherapies, the majority of patients with early stage disease have a very favorable prognosis. However, patients who later develop distant metastasis or who are diagnosed with disseminated disease at the onset are typically very difficult to treat effectively 5,6 . This is largely because our knowledge of how cancer cells spread is still limited. Cancer metastasis is a dynamic and complex process that requires tumor cells to undergo many steps, including adopting invasive properties, intravasating into proximal vasculature, surviving in circulation, evading immunosurveillance, extravasating from distant vasculature, and finally adapting to selective pressures of a new environment 7,8 . Each of these steps involves multiple interactions between cancer cells and different types of host stromal cells. As an example, breast cancer most commonly metastasizes to the lung, bone, liver, and brain, but how and why these cells travel and colonize these particular organs is still unknown 5,6 . A better understanding of how breast cancer metastasizes to these distant sites is greatly needed in order to develop more effective therapies and prevent spread of malignant disease.
Ephrin-A1 is a cell surface protein that regulates cell adhesion and migration 9-28 , and its role in cancer has recently been investigated in several different solid tumors [29][30][31][32][33][34][35] . It belongs to the group of ephrin ligands that interact with the largest family of receptor tyrosine kinases (RTKs), the Eph receptors, and regulates various developmental processes, such as embryonic cardiovascular development and angiogenic remodeling [36][37][38] . It is expressed in various cell types, including epithelial, endothelial, and immune cells and is the primary ligand for EphA2 RTK, which has been implicated in cancer growth and metastasis in various solid tumors [36][37][38][39][40][41] . While ephrin-A1 expression in cancer cells has been shown to be tumor suppressive 25,26,42,43 , its role in the host, has not been fully investigated. Here, we use ephrin-A1 knockout mice to examine how ephrin-A1 host deficiency affects cancer growth and metastasis in a murine model of breast cancer.
To test the impact of ephrin-A1 host deficiency on cancer progression, we utilized an orthotopic 4T1 mammary tumor model, as well as two different models of metastasis. While primary tumor growth did not significantly differ between ephrin-A1 wild-type (Efna1 +/+ ), heterozygous (Efna1 +/-), and knockout (Efna1 -/-) mice, metastasis and primary tumor recurrence were significantly decreased in Efna1 -/mice. Results of analysis on tumor-infiltrating immune cell populations and vascularity in the primary tumor did not evidently explain the differences in metastasis between Efna1 +/+ and Efna1 -/mice. However, tumor cell lung colonization was reduced in Efna1 -/mice, and lung metastases in Efna1 -/mice were less proliferative than in their wild-type counterparts, suggesting that the metastatic niche in Efna1 -/mice is less hospitable for invading tumor cells. Together, our studies suggest that host deficiency of ephrin-A1 does not impact initial tumor growth but does affect metastasis through inhibiting cancer cell extravasation and proliferation at the metastatic niche.

Animal models
Animals were housed in a non-barrier animal facility under pathogen-free conditions, 12-hour light/dark cycle, and access to standard rodent diet and water ad libitum. Experiments were performed in accordance with AAALAC guidelines and with Vanderbilt University Medical Center Institutional Animal Care and Use Committee approval. All mice used in this study were immunocompetent BALB/c mice. Ephrin-A1 knockout (Efna1 -/-) mice were previously characterized by our lab 44 . To generate littermate controls, wild-type BALB/c mice were purchased from Jackson Laboratory and mated with Efna1 -/mice to generate heterozygote mating pairs. Efna1 +/+ , Efna1 +/and Efna1 -/animals were identified by PCR analysis of genomic DNA using the following primers: Forward primer (5'-TGGTTATATCCCCCCACCTCACAC-3') and two allele-specific reverse primers (WT 5'-AAGGACTCCCATATCT-CAGCGACG-3') and (KO 5'-AGACTGCCTTGGGAAAAGCG-3'). Mice were co-housed with one to four littermates for at least two weeks prior to and during all experiments and compared with littermate controls whenever possible. All mice used for tumor experiments were six to ten weeks old at the onset on the experiment. Experimental cohorts were limited to litters that were born within two consecutive weeks and that also had at least one Efna1 -/and Efna1 +/+ gender-matched littermate pair and, when applicable, at least one Efna1 +/gender-matched littermate. Sample sizes are as shown in the figures and range from three to twelve mice per group. At experimental endpoints, mice were euthanized by cervical dislocation.

Tumor models
To reflect human breast cancer, only female mice were used for tumor experiments. For orthotopic mammary tumor implantations, 1×10 5 4T1 cells suspended in a 1:1 mixture of PBS and Growth Factor-Reduced Matrigel (Corning #354230) were injected through the nipple into the fourth mammary fat pads of six to eight-week-old female mice. Tumor dimensions were measured by digital caliper at given time points every other day, and volume was calculated using the following formula: volume = length × width 2 × 0.52. To observe spontaneous lung metastases and primary tumor recurrence, mammary tumors were resected at day 14 post-implantation, along with draining inguinal lymph nodes and surrounding fat pads, and mice were ultimately sacrificed at day 32. At the time of surgical resection of primary tumors on day 14, tumors were weighed and cut in half to provide tissue for both flow cytometry analysis and cryosection staining. At the experimental endpoint on day 32, recurrent tumors were weighed, and lung metastases were counted in a blinded manner. For lung colonization experiments, 4T1-GFP-luciferase cells suspended in PBS were injected via tail vein, and mice underwent in vivo bioluminescence imaging with a PerkinElmer IVIS Spectrum several hours post-injection to verify successful and equal delivery of 4T1 cells. To observe gradual formation of GFP+ metastases, 1×10 5 4T1-GFP-luciferase cells were injected via tail vein, and mice were sacrificed at day 17. GFP+ lung metastases were counted in a blinded manner. The left lung lobe of each mouse was fixed in 10% formalin for subsequent formalin-fixed paraffin-embedded (FFPE) processing, sectioning, and H&E staining, while the other lung lobes were processed for flow cytometry analysis. To observe early colonization and proliferation of 4T1 cells, 5×10 5 4T1-GFP-luciferase cells labeled with CellTrace Violet dye (Invitrogen #C34571) were injected via tail vein. At 24 hours, mice were sacrificed, and lungs were perfused with PBS and processed for flow cytometry analysis.
Immunohistochemistry (IHC) and Immunofluorescence (IF) FFPE lung sections were prepared and stained for PCNA (1:100, BD Biosciences #555567 raised in mouse, RRID: AB_395947) as described previously 45 . Slides were blinded, and the number of metastatic foci per section of lobe was quantified. Nuclear PCNA staining was analyzed using ImageJ v1.52o with the IHC Profiler plugin 46 and percentage of PCNA+ tumor cell nuclei were quantified. Each data point is an average of two sections of the left lung from an individual mouse. To prepare cryosections, mammary tumors were frozen in OCT Compound (Thermo Fisher Scientific #23-730-571) on dry ice and stored at -80°C. Sections (8 µm) were cut on a Leica Cryostat CM1950, fixed in 4% PFA, washed with PBS, permeabilized with 0.3% Triton X-100 (Sigma-Aldrich #X100), and blocked using M.O.M. Mouse Ig Blocking Reagent and Protein Concentrate (Vector Laboratories #PK-2200) per manufacturer recommendations and with 2.5% goat serum (Sigma-Aldrich #G9023) in PBS. Slides were then incubated over two nights at 4°C with primary antibodies against CD31 (1:150, Biolegend #102501 raised in rat, RRID: AB_312908) and αSMA (1:150, Dako #M085129-2 raised in mouse, RRID: AB_2811108) in blocking buffer. After washing with PBS, slides were incubated for one hour at room temperature in secondary antibodies goat anti-rat Ax594 (1:500, Invitrogen #A11007, RRID: AB_10561522) and anti-mouse Ax488 (1:500, Invitrogen #A11001, RRID: AB_2534069), washed with PBS, and mounted with ProLong Gold Antifade Mountant with DAPI (Invitrogen #P36931). Slides were blinded, and images were taken by an Olympus DP72 camera through a BX60 inverted fluorescence microscope and processed using CellSens Dimension software. A total of 12-40 20x fields of view were analyzed from each section using ImageJ. For αSMA analysis, images were evaluated for colocalization with CD31 staining, and data was displayed as a percentage of αSMA+ out of CD31+ area or integrated intensity. Each data point is an average of all fields of view of two to three tumor sections from an individual mouse.

Flow cytometry
Tumors and lungs were minced and dissociated in RPMI-1640 media (Corning #MT10040CV) containing 2.5% FBS, 1 mg/ml collagenase IA (Sigma-Aldrich #C9891), and 0.25 mg/ml DNase I (Sigma-Aldrich #DN25) for 45 minutes at 37°C. Digested tissue was then filtered through a 70-µm strainer, and red blood cells were lysed using ACK Lysis Buffer (KD Medical #RGF-3015). Samples were washed with PBS and stained with Ghost Dye Violet V510 (Tonbo Biosciences #13-0870) to exclude dead cells. After washing with buffer (0.5% BSA, 2mM EDTA in PBS), samples were blocked in αCD16/32 mouse Fc block (Tonbo Biosciences #70-0161) and stained for extracellular proteins using an antibody master mix made in buffer. After washing with buffer, cells were fixed with 2% PFA. For FoxP3 intracellular staining, cells were permeabilized using the FoxP3 Transcription Factor Staining Kit (Tonbo Biosciences #TNB-0607-KIT) per manufacturer protocol. Flow cytometry data was obtained on a BD 4-laser Fortessa using BD FACS Diva software v8.0.1 and analyzed using FlowJo software v10.6.1. Fluorescence minus one (FMO) samples were used as gating controls when needed. Antibodies used in flow panels are detailed in Table 1, and gating strategies used in analysis are detailed in Table 2. Each data point is generated after analyzing at least 5×10 5 viable cells from a specimen from an individual mouse.

Statistical analysis
All graphs and statistical analysis were completed using GraphPad Prism software v6.07. For comparisons between two groups, an unpaired Mann-Whitney U-test was performed. For comparisons between three groups, a Kruskal-Wallis H-test was performed, followed by post-hoc Mann-Whitney U-tests evaluating differences between the knockout and either the wild-type or heterozygote animals. A P-value less than 0.05 was considered statistically significant.

Ephrin-A1-deficient hosts have reduced metastasis in vivo
We initially investigated the impact of ephrin-A1 host deficiency on primary tumor growth by implanting 4T1 cells orthotopically into the mammary fat pads of syngeneic BALB/c female Efna1 +/+ and Efna1 -/mice. No difference in primary tumor growth or weight at 21 days post-implantation was observed ( Figure 1A). To test the impact of ephrin-A1 host deficiency on spontaneous metastasis, 4T1 cells were implanted orthotopically as described above and surgically resected on day 14 post-implantation to allow for gradual development of endogenous metastases by day 32 ( Figure 1B). As expected, primary tumors resected from Efna1 +/+ and Efna1 -/mice were not different in size ( Figure 1C), and this was additionally verified with Efna1 +/+ , Efna1 +/-, and Efna1 -/littermates ( Figure 1D). However, at the experimental endpoint, the number of visible lung metastases and lung weights were significantly decreased in knockout mice ( Figure 1E). Many of these mice not only harbored lung metastases but also tumors that had regrown at the original site of the resected primary tumor. Similar to our findings in lung metastases, the size of recurrent primary tumors was significantly reduced in knockout mice ( Figure 1F). Together, these results demonstrate that while host deficiency in ephrin-A1 may not affect initial tumor growth, it can impact metastatic spread and recurrence. Underlying data are available 47,48 .
To complement our findings in our model of spontaneous metastasis, we evaluated the impact of ephrin-A1 host deficiency on experimental metastasis. 4T1 cells engineered to express GFP and luciferase (4T1-GFP-luciferase) were injected into the tail veins of Efna1 +/+ , Efna1 +/-, and Efna1 -/littermates. In vivo bioluminescence imaging several hours after injection illustrated comparable signal across all mice (Figure 2A, B), indicating ephrin-A1 host deficiency did not impact tumor cell trafficking and lodging within the lung, at least in this short time frame. After harvesting the lungs 17 days later, we observed decreased GFP+ metastases in Efna1 -/mice, compared to both Efna1 +/+ and Efna1 +/littermates, which was additionally confirmed by flow cytometry (Figure 2C, D). Similarly, histological analysis revealed fewer metastatic foci in lungs from Efna1 -/mice ( Figure 2E). These data align with our previous observations on endogenous metastasis and suggest that host deficiency in ephrin-A1 inhibits circulating cancer cells from colonizing the lung. Underlying data are available 49,50 .
Tumor-infiltrating immune populations are not significantly different in ephrin-A1-deficient hosts Ephrin-A1 is expressed in several types of host cells, including immune cells and endothelial cells 39-41 . Thus, we sought to determine how the ephrin-A1-deficient host immune system and endothelium may mitigate metastasis. Among immune cells, Ephrin-A1 can be expressed in B and T cells, monocytes, and macrophages 39 . The role of ephrin-A1 in B cells is largely unknown 51 . However, in T cells, monocytes, and macrophages, ephrin-A1 has been shown to regulate cell adhesion and migration 11,12,15,27,52-55 . These immune cell populations play a critical role in overall anti-tumor immunity and immunosurveillance. Dendritic cells and T cells, particularly CD8 cytotoxic T cells, are the primary drivers of the adaptive anti-tumor response in solid tumors and increased infiltration of these cell types is correlated with better prognosis and enhanced response to immunotherapies 56,57 . Conversely, T regulatory cells (Tregs) suppress effector functions of T cells and typically inhibit the anti-tumor response 58,59 . Between the two ends of this spectrum, myeloid populations, such as monocytes, macrophages, and granulocytes, can either promote or suppress an anti-tumor response, depending on their polarization and functionality 60,61 .
Because of ephrin-A1's known role in adhesion and chemotaxis of immune cells, we performed flow cytometry analysis on 4T1 primary tumors harvested from Efna1 +/+ , Efna1 +/-, and Efna -/littermates. To our surprise, we found no significant differences in total infiltrating immune cells, CD4 or CD8 T cells, dendritic cells, Tregs, or myeloid populations in Efna1 +/+ , Efna1 +/-, and Efna -/littermates ( Figure 3A, B). While there were no apparent differences in the immune microenvironment of the mammary tumors, the immune microenvironment of the lung is distinct from that of the mammary gland and may impact the metastatic niche. To investigate this, we performed flow cytometry analysis on 4T1 tumor-bearing lungs generated from our model of experimental metastasis. Similar to the results we obtained from the 4T1 primary tumors, we did not see significant differences in immune populations in tumor-bearing lungs harvested from Efna1 +/+ , Efna1 +/-, and Efna1 -/littermates, except for a modest increase in macrophages in knockout mice ( Figure 3C, D). Underlying data are available 62,63 .
Although the percentage of tumor infiltrating T cells in Efna1 +/+ and Efna1 -/mice is comparable, their activation status and effector function may still be different. Tumor-infiltrating T cells with upregulated expression of activation markers, such as CD44, CD69, and CD25, and downregulated expression of antigen-naïve markers like CD62L and exhaustion markers like PD-1 and CTLA-4 indicate a higher T cell functional status that mediates a stronger and more enduring anti-tumor response 58,59 . We assessed these markers on T cells in 4T1 primary tumors and tumor-bearing lungs from Efna1 +/+ and Efna1 -/littermates using flow cytometry. However, we did not observe consistent increases in activation or decreases in naïve or exhaustion markers in knockout-derived T cells (data not shown, included in Underlying data) 62 . In summary, host deficiency in ephrin-A1 does not significantly affect tumorinfiltrating immune cells in both primary tumors and tumorbearing lungs. Thus, the reduction of lung metastases in Efna1 -/hosts in vivo is unlikely due to host immunity.

Tumor vascularity and pericyte coverage are not significantly different in ephrin-A1-deficient hosts
In addition to the anti-tumor immune response, another host factor that can impact metastasis is the tumor vasculature. Angiogenesis is the formation of new blood vessels from a pre-existing network and is required for solid tumor growth and progression. Blood vessels can supply nutrients that support tumor growth and provide an entry for hematological dissemination and invasion 1,64 . These new blood vessels are typically hastily constructed in response to the high release of growth factors, such as vascular endothelial growth factor (VEGF), from tumor cells 65,66 . Thus, tumor vessels tend to be disorganized, leaky, and poorly covered by pericytes, which normally support the integrity of the endothelium. Ephrin-A1 is expressed in the vascular endothelium and has been shown to promote angiogenesis in vitro and in several in vivo models 67-72 . Therefore, we hypothesized that tumors in Efna1 -/mice may have reduced tumor vasculature and increased endothelial pericyte coverage compared to Efna1 +/+ controls.
To evaluate tumor vascularity and vessel function, we co-stained cryosections of 4T1 primary tumors from Efna1 +/+ and Efna1 -/littermates with CD31 and αSMA, markers for endothelial cells and pericytes, respectively. Colocalization of αSMA with CD31 acts as an indicator for functional endothelium within tumors. Surprisingly, we did not observe a change in CD31+ area or intensity in 4T1 tumors from Efna1 +/+ and Efna1 -/littermates ( Figure 4A, B). Furthermore, pericyte coverage on tumor vessels remained the same in tumors from Efna1 +/+ and Efna1 -/littermates ( Figure 4A, C). Together, these data suggest that loss of ephrin-A1 in the host does not affect tumor vessel formation and function in the primary tumor. Underlying data are available 73 .
Ephrin-A1-deficient lung microenvironment provides a less favorable metastatic niche Our results from analysis of the immune infiltrate and vasculature of primary tumors, coupled with the significant difference in experimental lung metastasis between Efna1 +/+ and Efna1 -/mice, suggest that host factors critical to this metastatic phenotype are more likely to lie downstream of the primary tumor site. These steps include tumor cell trafficking to the lung vasculature, extravasation, and adaptation to new selective pressures of the lung microenvironment. 4T1 cell trafficking to the lung was not significantly different between Efna1 +/+ and Efna1 -/littermates after tail vein injections (Figure 2A, B). Thus, we aimed to evaluate extravasation and adaptation to the lung metastatic niche in Efna1 +/+ and Efna1 -/hosts.
To test extravasation of 4T1 cells in vivo, we injected 4T1-GFP-luciferase cells labeled with CellTrace Violet dye into the tail veins of Efna1 +/+ and Efna1 -/littermates. This dye is retained only in the labeled tumor cells and diminished after subsequent cell divisions, enabling quantification of short-term cell proliferation. At 24 hours after injection, we perfused the lungs with PBS to flush out remaining cells in the pulmonary vasculature and processed the lungs for flow cytometry. Decreased GFP+ 4T1 cells were found in ephrin-A1-deficient lungs compared to wild-type controls ( Figure 5A), suggesting that fewer cancer cells had extravasated into the lung parenchyma at this timepoint. This result was not due to decreased vascularity of ephrin-A1-deficient lungs at baseline, as CD31+ endothelial cells were not significantly different in knockout and wild-type lungs ( Figure 5B). Moreover, this was not due to decreased proliferation of the 4T1 cells within the 24-hour timeframe, as the amount of retained CellTrace Violet dye was not increased in 4T1 cells that had extravasated in knockout lungs compared to wild-type lungs ( Figure 5C). Together, these data suggest that extravasation of 4T1 cells is inhibited in knockout mice, compared to wild-type controls, and ephrin-A1 deficiency in the host lung may play a role in this process.
While decreased extravasation of tumor cells may explain in part the decreased lung metastases in Efna1 -/mice, another possibility is that once tumor cells have extravasated and established in the lung, they have reduced fitness of survival in ephrin-A1-deficient lungs, compared to wild-type lungs.
There are many stressors in the lung metastatic niche that could impact the adaptability of the tumor cell. We used tumor proliferation index as a marker to evaluate how well tumor cells have adapted to a metastatic niche. Since no differences were observed in proliferation of 4T1 cells that had newly extravasated into the lung parenchyma of Efna1 +/+ and Efna1 -/littermates within the short 24-hour timeframe ( Figure 5C), we next assessed proliferation of tumor cells in lung micrometastases that had established over 17 days after tail vein injection (Figure 2). There was a significant decrease in cell proliferation in metastatic foci established in Efna1 -/mice, compared to Efna1 +/+ controls, as indicated by PCNA staining ( Figure 5D). These findings suggest that reduced tumor cell lung colonization in Efna1 -/hosts is due to both decreased extravasation of cancer cells and decreased proliferation in the metastatic niche.

Discussion
In conclusion, host deficiency in ephrin-A1 inhibits metastasis by providing a less hospitable metastatic niche for cancer cell extravasation and colonization of the lung. Our data from 4T1 primary tumor specimens demonstrated no differences in primary tumor growth, infiltrating immune cell populations, and vascularity. This led us to investigate the metastatic process downstream from the primary tumors. We then found that lung colonization in knockout mice was decreased compared to wild-type mice as early as 24 hours post-tail vein injection of 4T1 cells, in part due to decreased extravasation. Moreover, the metastases that established in Efna1 -/lungs were not only reduced in number but also less proliferative compared to those in wild-type lungs. These studies offer insight on how host expression of ephrin-A1 may impact tumor growth and dissemination, but they also lead to additional questions.
Our ephrin-A1 knockout model is not tissue-specific nor inducible, which creates challenges in identifying specific mechanisms that contribute to our observed phenotype. For example, ephrin-A1 is highly expressed in embryonic stages of development and plays a known role in neuronal and mammary development 67,76-80 . The transcriptional and epigenetic changes that occur in utero and during early physiological development as a result of ephrin-A1 deficiency in various tissues may all contribute to the observed phenotype; however, dissecting which changes are directly downstream of ephrin-A1 and critical to metastasis may be quite difficult. This challenge is further augmented when we consider the many cell types that can express ephrin-A1, especially immune, endothelial, and epithelial cells.
Many studies have demonstrated ephrin-A1's role in immune cell adhesion and migration. Although we did not observe significant differences in tumor immune infiltrate, this does not preclude a role for ephrin-A1 in these cell populations. Immune cells engage in a complex network of crosstalk, and it is possible that loss of ephrin-A1 in one cell type may mask the effects it has in another. One intriguing difference we observed was an increase in macrophages in Efna1 -/tumor-bearing lungs. However, we have not determined if this difference occurs in the specific context of a stressor, such as tumor metastasis, or if knockout mice have increased macrophages at baseline. Furthermore, it remains to be seen if these macrophages are polarized towards an anti-tumor or a pro-tumor response. Nevertheless, this offers evidence of a novel role of ephrin-A1 in macrophage differentiation or survival, which requires further investigation.
In addition, ephrin-A1 has been shown to regulate expression of adhesion molecules on endothelial cells and promote angiogenesis. Modulation of surface expression of adhesion proteins, such as ICAM-1 and VCAM-1, on endothelial cells impact binding to immune cells and cancer cells 8,40 . Thus, it is possible that ephrin-A1 on endothelial cells may mediate cancer cell transendothelial migration through modulation of these adhesion proteins. While this result may be consistent with published literature, in contrast to ephrin-A1's known role in angiogenesis, we did not observe differences in angiogenesis between tumors from Efna1 +/+ and Efna1 -/hosts. This discrepancy may be due to a couple reasons. First, most studies reporting on ephrin-A1's impact on angiogenesis has shown its effect through EphA receptor signaling on the endothelial cell, not necessarily through ephrin-A1 directly in the endothelium 67-70 . Loss of ephrin-A1 in the endothelium and other host tissues is unlikely to completely abrogate EphA receptor signaling in the endothelium, as other ephrin ligands are able to promiscuously bind to the same EphA receptors and may even compensate for the loss of ephrin-A1 37 . Second, some of these studies use soluble ephrin-A1, instead of membrane-bound or cell-surface ephrin-A1. In our Efna1 -/model, both cell-surface, membrane-bound ephrin-A1 and soluble, secreted ephrin-A1 are lost in vivo, and these two forms of ephrin-A1 have been shown to have competing effects 13 .
The different forms of ephrin-A1, as well as the range of interactions with various Eph receptors, show how potentially complex the molecular mechanisms can be when considering host deficiency of ephrin-A1. A clue into this complicated investigation can be found in our data obtained with ephrin-A1 heterozygote controls. When comparing tumor metastasis and immune infiltrate, results from Efna1 +/mice were much more comparable to wild-type than knockout littermate controls. This suggests that ephrin-A1 has a genetically dominant effect -one wild-type allele may be sufficient to induce the wild-type phenotype.
Although we focused our inquiries on primary mammary tumors and lung metastases, there is much more to be explored. 4T1 cells, like human breast cancer, metastasize to other organ sites, such as the bone, liver, and brain. The lungs in Efna1 -/hosts may or may not be the only organ that provides a less favorable environment for colonizing tumor cells than those in Efna1 +/+ hosts. We observed differences in recurrent primary tumor, in addition to lung metastases, which may indicate that tumor cell apoptosis or senescence is altered in knockout hosts. Further elucidating the mechanisms by which ephrin-A1 host deficiency limits cancer relapse and metastasis may enhance our understanding of the metastatic process and ultimately shed new light on novel therapeutic strategies.  Figure 1A (Flow cytometry files (fcs), gating and analysis (wsp), and panels (xlsx) containing immune profiling of 4T1 primary mammary tumors from Efna1 +/+ , Efna1 +/-, and Efna1 -/littermate mice (related to Figure 3A, B)) Harvard Dataverse: 4T1-GFP-luciferase tumor-bearing lung flow cytometry. https://doi.org/10.7910/DVN/S06NQ1 63 . (Flow cytometry files (fcs), gating and analysis (wsp), and panels (xlsx) containing immune profiling of 4T1-GFP-luciferase tumor-bearing lungs from Efna1 +/+ , Efna1 +/-, and Efna1 -/littermate mice (related to Figure 2D 1.

Nikki Cheng
Department of Pathology and Laboratory Medicine, University of Kansas Medical Center (KUMC), Kansas City, KS, USA EphrinA1 is a cell surface bound protein that has been shown to be tumor suppressive when expressed in cancer cells. In this study, the authors characterize the effects of the microenvironment on mammary tumor progression using variations of the 4T1 mammary tumor model-spontaneous, and experimental metastasis. By injecting 4T1 cells in ephrinA1 knockout mice, they show that ephrinA1 knockout does not affect primary tumor growth and inhibits metastasis and recurrence. The study is generally well written, and the phenotypes are interesting. Comments are described below: The discussion could be better expanded.
How do expression of ephrinA1 and corresponding receptors in the stroma of primary tumor vs. lung tissue compare in normal and tumor bearing mice? How do results fit into the bigger picture of what is known about this?
Recommend expanding on the discussion of the mechanism regarding ephrinA1 mediated lung metastasis. In particular: a) How might ephrinA1 signaling might affect macrophage function in the lung. b) Please expand on what is known about macrophage polarization, effects on T cells in the lung. How would ephrinA1 signaling affect macrophage function in the formation of pre-metastatic vs. metastatic niche? c) How would ephrinA1 ko affect Eph receptorA1 vs. A2 signaling in the lung stroma? d) Further discussion the clinical implications of the study would be helpful, especially since the tumor suppressive/oncogenic effects of ephrinA1 appear to be cell type dependent.
More minor comments: Figure 1A, C, D, Figure 3, Figure 4, Figure 5B-C: Statistics are not consistently shown. Even if data show no statistical significance, it would be helpful to include p-value or indicate n.s to indicate the analysis was done. 3. Figure 4: CO-IF staining is a bit difficult to see. It would be helpful to show magnified insets for clearer details. Figure 5: recommend revising figure legend. From the way the figure legend is stated " ," I Ephrin-A1-deficient lung microenvironment provides a less favorable metastatic niche would expect data correlating decreased metastasis with changes in the lung microenvironment of ephinA1 KO mice. However, this is not the case -no changes in endothelial cells were observed. Alternatively, for Figures 3 and 5 -recommend reorganizing some of the data so that the messages are clearer. Figure 3 figure legend indicates that "Tumor-infiltrating immune populations are However, Figure 3D show increased in not significantly different in ephrin-A1-deficient hosts" macrophage recruitment to lung tissues in ephrinA1 KO mice. This data would be more appropriate in Figure 5, which shows decreased lung metastasis, along with characterization of endothelial cells in the lung.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? No source data required

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed. Competing Interests: 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.
Author Response 07 May 2020 , Vanderbilt University, Nashville, USA

Eileen Shiuan
Thank you for reviewing this manuscript! We have addressed your suggestions and comments with the following revisions: 1) Expanded discussion on potential impact of ephrin-A1 deficiency on other ephrin-A1 ligands and EphA receptors in stromal cells and their signaling.
2) Expanded discussion on what is known about the role of macrophages in the lung metastatic niche and how ephrin-A1 deficiency may impact recruitment and function of macrophages.
3) Expanded discussion of clinical implications of the study. confounding effects of Matrigel on primary tumor volume in the Discussion.
2) For data with small sample sizes showing a nonsignificant trend, revised interpretation of data to include possibility of significant differences if more mice were used.
3) Indicated in all figure legends that each data point corresponds to an individual mouse.

4) Minor text changes.
No competing interests were disclosed. Competing Interests: