Reelin Deficiency Delays Mammary Tumor Growth and Metastatic Progression

Abstract

Reelin is a regulator of cell migration in the nervous system, and has other functions in the development of a number of non-neuronal tissues. In addition, alterations in reelin expression levels have been reported in breast, pancreatic, liver, gastric, and other cancers. Reelin is normally expressed in mammary gland stromal cells, but whether stromal reelin contributes to breast cancer progression is unknown. Herein, we used a syngeneic mouse mammary tumor transplantation model to examine the impact of host-derived reelin on breast cancer progression. We found that transplanted syngeneic tumors grew more slowly in reelin-deficient (rl^{Orl −/−}) mice and had delayed metastatic colonization of the lungs. Immunohistochemistry of primary tumors revealed that tumors grown in rl^{Orl −/−} animals had fewer blood vessels and increased macrophage infiltration. Gene expression studies from tumor tissues indicate that loss of host derived reelin alters the balance of M1- and M2-associated macrophage markers, suggesting that reelin may influence the polarization of these cells. Consistent with this, rl^{Orl −/−} M1-polarized bone marrow-derived macrophages have heightened levels of the M1-associated cytokines iNOS and IL-6. Based on these observations, we propose a novel function for the reelin protein in breast cancer progression.

Introduction

The reelin signaling pathway is widely recognized as an important regulator of cell migration in the developing central nervous system [1–4]. Reelin is a glycoprotein secreted into the extracellular matrix, and canonically signals by binding to low-density lipoprotein receptors Apo E receptor 2 (ApoER2) and very low-density lipoprotein receptor (VLDLR) located on the surface of reelin-responsive cells [5]. Binding of reelin to its receptors recruits the intracellular adaptor protein Disabled-1 (Dab1), which is phosphorylated by Src family kinases on tyrosine residues [6,7]. Phosphorylated Dab1 activates multiple downstream effectors, including phosphatidylinositol-3-kinase (PI3K)/Akt and C3G/Rap1 [2]. This cascade of signaling events leads to changes in cytoskeleton stabilization, allowing reelin to orchestrate cell migration in the central nervous system.

Recent studies have identified a number of non-neuronal functions for reelin [8–12]. For example, reelin signaling was found to regulate the homeostasis of the intestinal crypt-villus unit [11], formation of thrombin clots [10], and to be important for establishment of the lymphatic vasculature [12]. Our laboratory found that canonical reelin signaling is essential for proper mammary gland development [13]. Reelin is normally expressed in the myoepithelial layer and the stroma of the developing and mature mammary ducts, while Dab1 is expressed in the luminal epithelium. Reelin and Dab1 coordinate ductal extension and refine the morphology of the mammary ducts. Absence of functional reelin signaling results in ductal growth delays, abnormal branching patterns and disorganized cellular layers within the mammary ducts [13]. Alterations in reelin signaling have been reported in pancreatic, gastric, prostate, and esophageal cancers [14–17]. Interestingly, reelin expression is lost in many human breast tumors, and the loss of reelin protein correlates with poor survival [18], but the mechanisms by which reelin may affect breast cancer progression in vivo remain to be determined.

To better understand the relationship between reelin signaling and breast cancer, we monitored mammary tumor growth and metastatic progression following transplantation of 4T1 mouse mammary tumor cells into mice that lack functional reelin protein (rl^{Orl −/−}). We chose the 4T1 mammary tumor transplantation model for two reasons. First, 4T1 cells mimic stage IV human breast cancer, and rapidly form primary tumors and spontaneous lung metastases when orthotopically injected into Balb/C mice [19]. Second, the 4T1 model allows us to study tumor progression in immunocompetent mice carrying the rl^Orl mutation, and provides us with the ability to address the contribution of the host immune response to the growth of primary tumors and metastasis.

We report that the absence of reelin from the host environment delays primary tumor growth and metastatic spread of mammary carcinoma cells, possibly via alterations in the cytokine expression profile of tumor-associated macrophages (TAMs). Loss of reelin does not directly affect proliferation or migration of tumor cells, but may modulate the activation of macrophages in the tumor microenvironment, diminishing their tumor-promoting properties. Our results indicate a novel function for the reelin protein in mammary tumor progression, and suggest possible roles for reelin in macrophage activation.

Materials and Methods

Mice

Balb/C mice were purchased from Charles River Laboratories. Reeler Orleans (rl^{Orl −/−}) mice were obtained from a breeding colony maintained at UCLA. These mice carry a naturally occurring mutation, in which a transposon insertion leads to exon skipping and a 220 bp deletion in the reelin mRNA [20]. The resulting reelin protein is truncated and is not secreted [21]. Rl^Orl mice were initially on a mixed, 70–75% Balb/C and 20–25% 129/Sv background, and were backcrossed to the Balb/C strain for three generations to obtain ≥95% Balb/C offspring. The genetic background of the rl^Orl line was confirmed by single nucleotide polymorphism (SNP) scanning (The Jackson Laboratory). The use of a Balb/C background is necessary for histocompatibility, as 4T1 cells are derived from Balb/C mice [19]. Homozygous mutant and wild type control female offspring were obtained from intercrosses of heterozygous rl^{Orl +/−} animals. Rl^Orl mice were genotyped by PCR as described [20].

Cell Lines

The 4T1 cell line was purchased from American Type Culture Collection, and maintained according to ATCC guidelines. Cells were cultured in RPMI-1640 medium (Life Technologies) supplemented with 10% fetal bovine serum (FBS, Omega) and 100 u/mL penicillin/streptomycin (Life Technologies). Sub-confluent cultures were treated with 0.25% trypsin-EDTA (Life Technologies) and passaged, or counted using a hemocytometer and used for in vitro or in vivo experiments.

The reelin-secreting HEK293T cell line (stably transfected with a full-length reelin clone) [22] was kindly provided by Dr. Tom Curran, Children’s Hospital of Philadelphia, PA, USA. The control HEK293T cell line was kindly provided by Dr. Harley Kornblum, University of California Los Angeles, CA, USA. Both cell lines were cultured in DMEM (Life Technologies) supplemented with 10% FBS and 100 u/mL penicillin/streptomycin. Conditioned media was collected from confluent cells after 48 hours of culture, centrifuged at 600 g for 10 min, and the supernatant was collected and used immediately for treatment of 4T1 cells and migration assays.

Mammary Epithelial Cell (MEC) Purification

MECs were purified as previously described [23]. Briefly, pairs of #3 thoracic and #4 inguinal mammary glands were dissected from 8–10 week-old female mice, minced and incubated in DMEM/F12 (Corning) containing 5% FBS, 100 u/mL penicillin/streptomycin, 2 mg/mL collagenase IV (Sigma), 2 mg/mL trypsin (Sigma), and 5 µg/mL insulin (Life Technologies) on an orbital shaker at 100 RPM, 37°C, for 1 hr. Digested tissue was treated with 4 u/mL DNAse (Sigma), and organoids containing MECs were purified by repeated pulse centrifugation.

Transwell Migration Assay

10⁵ 4T1 cells in serum-free DMEM were seeded on top of Boyden transwells fitted with membranes containing 8 µm pores, and allowed to migrate overnight in response to conditioned media from reelin-secreting HEK293T cells or control HEK293T cells with 10% FBS. Cell nuclei were stained with DAPI, and membranes were imaged on a Zeiss Axioskop with a cooled CCD camera. Four evenly spaced 100X fields per membrane were photographed and used to count the number of cells that migrated through the membrane. Cell counts were used to determine the average number of migrating cells for each membrane.

Tumor Challenge

10⁵ 4T1 cells in HBSS were injected into the left #4 mammary fat pad of homozygous mutant and control wild type rl^Orl 8–10 week-old female mice. The tumor diameters were measured every 3–4 days beginning on day 11 after transplantation using electronic calipers. The tumor volume was calculated using the formula L × W² × 0.52, where L = longest diameter, and W = perpendicular diameter [24]. All mice were sacrificed before the primary tumor diameter reached 1.5 cm or when the animals became moribund.

Micrometastasis Assay

Lungs from tumor-bearing mutant and wild type rl^Orl mice were dissected in sterile conditions, and lung tissue was processed as described [19]. Briefly, lungs were minced and incubated in digestion buffer containing collagenase I (Sigma) for 1 hour at 37°C. Digested tissue was sieved through 70 µm cell strainers, and red blood cells were lysed in RBC lysis buffer (StemCell Technologies). The resulting cell suspension was serially diluted into 6-well plates in RPMI-1640 medium supplemented with FBS and penicillin/streptomycin. The next day, medium was replaced with complete RPMI-1640 containing 60 µM 6-thioguanine. After 10 days of incubation, resistant 4T1 colonies were stained with methylene blue and counted. The number of 4T1 colonies was equated to the number of 4T1 cells in the seeded cell suspension, normalized to the number of cells plated, and the resulting percentage was reported as the metastatic burden.

Immunohistochemistry and Histology

For paraffin-embedded samples, dissected tumor slices were fixed overnight in Bouin’s fixative, washed, dehydrated, and embedded in paraffin. For frozen samples, tumor slices were fixed in 4% paraformaldehyde for 4 hours, cryoprotected in 30% sucrose and embedded in Optimal Cutting Temperature compound (Sakura).

For Ki-67 and CD31 labeling, paraffin-embedded samples were sectioned at 10 µm. The sections were rehydrated through graded ethanols and boiled in 10 mM sodium citrate buffer to retrieve antigens. After blocking endogenous peroxidases, as well as avidin and biotin, sections were incubated with primary, and then biotinylated secondary antibodies. The antibodies used were rabbit anti-Ki-67 (1:500, Vector Labs), rat anti-CD31 (1:200, Dianova), biotinylated goat anti-rabbit (1:500, Jackson Immunoresearch), biotinylated donkey anti-rat (1:500, Jackson Immunoresearch). The antibodies were visualized using an ABC Elite Kit (Vector Labs) and diaminobenzidine (DAB) staining. Cell nuclei were counterstained with hematoxylin.

For F4/80 labeling, 10 µm frozen sections were treated with 1% Triton X-100 (Sigma), blocked with donkey serum, and incubated with primary antibody, then AlexaFluor 594-conjugated secondary antibody. Cell nuclei were counterstained with DAPI. The antibodies used were rat anti-F4/80 (1:1000, Abbiotec), Alexa Fluor 594 donkey anti-rat (1:500, Life Technologies).

For histological evaluation of tumor sections, 10 µm paraffin sections were rehydrated through graded ethanols and stained with hematoxylin and eosin.

For Ki-67 labeling analysis, Ki-67-positive and Ki-67-negative cell nuclei were counted in 4 random fields per sample using the ObjectJ plugin in ImageJ (NIH), and the average percentage of Ki-67-positive nuclei was calculated for each sample. Each analyzed field contained at least 500 total cells. For CD31 and F4/80 labeling analysis, 6 random 200X fields per sample were imaged, DAB signal (for CD31) or AlexaFluor 594 signal (for F4/80) was thresholded in ImageJ, and the average percentage of CD31-positive or F4/80-positive area was calculated for each sample.

Quantitative PCR (qPCR)

RNA from tumor tissues or cells was extracted using TRI reagent (Sigma), and the cDNA was generated using an iScript cDNA Synthesis kit (Bio-Rad). qPCR was carried out using the KAPA SYBR FAST master mix (KAPA Biosystems) on a Light Cycler 480 (Roche). Expression values were normalized to the housekeeping gene 36b4/RPLP0. Primer sequences are listed in Table 1.

Table 1.

Primers used for qPCR gene expression analysis.

Gene	Forward Primer (5' - 3')	Reverse Primer (5' - 3')
ApoER2	CGGACAGCGACTTCACCT	CTTCTCGGCAGGACACTCTT
Arg1	GCAGAGGTCCAGAAGAATGG	AGCATCCACCCAAATGACAC
Cxcl2	AATCATCCAAAAGATACTGAACAAAG	TTCTCTTTGGTTCTTCCGTTG
Cxcl5	TGCGTTGTGTTTGCTTAACCG	CTTCCACCGTAGGGCACTG
Cxcl10	CCAAGTGCTGCCGTCATTTTC	GGCTCGCAGGGATGATTTCAA
Dab1	GATGAAGTGTCCGCAGCTC	GTGTTCTCCCTTGGAACGTG
IL-1b	GCCCATCCTCTGTGACTCAT	AGGCCACAGGTATTTTGTCG
IL-6	AGTTGCCTTCTTGGGACTGA	TCCACGATTTCCCAGAGAAC
IL-10	GCTCTTACTGACTGGCATGAG	CGCAGCTCTAGGAGCATGTG
IL-12	CCATCAGCAGATCATTCTAGACAA	CGCCATTATGATTCAGAGACTG
iNOS	CACCTTGGAGTTCACCCAGT	ACCACTCGTACTTGGGATGC
Mmp-9	GGACCCGAAGCGGACATTG	CGTCGTCGAAATGGGCATCT
Mrc1	ATTGTGGAGCAGATGGAAGG	TGAATGGAAATGCACAGACG
Reln	GTCACGGTCTACCTGCCACT	TCAATAGCCCAGGAATCTGC
Tgfb	TTGCTTCAGCTCCACAGAGA	TGGTTGTAGAGGGCAAGGAC
Tnfa	TGCCTATGTCTCAGCCTCTTC	GAGGCCATTTGGGAACTTCT
Vegfa	GGAGAGCAGAAGTCCCATGA	GGGGTACTCCTGGAAGATGTC
Vegfb	TCTGAGCATGGAACTCATGG	TCTGCATTCACATTGGCTGT
Vegfc	CAAGGCTTTTGAAGGCAAAG	TGCTGAGGTAACCTGTGCTG
Vldlr	TACCCTAGACGGAGCCAAGA	GTAAACAAAGCCCGACAACG

Western Blotting

Tumor tissue was homogenized in radioimmunoprecipitation (RIPA) buffer containing phosphatase inhibitor cocktail (Zmtech scientific). Protein samples (35 µg) were separated on 4–12% Bis-Tris SDS-PAGE gels (Life Technologies), and transferred onto Immuno-Blot PVDF membranes (Bio-Rad). Membranes were blocked in PBS with 5% skim milk and 0.1% Tween-20 for 1 hour, and incubated with primary antibodies overnight at 4°C, then with secondary antibodies for 30 min at room temperature. Protein bands were detected using the ECL Western blotting detection reagent (GE Amersham). Band intensities were quantified using ImageJ and relative protein levels were normalized to β-actin levels. The following antibodies were used: mouse anti-arginase-1 (1:4000, BD Biosciences), mouse anti-β-actin (1:10,000, Sigma), mouse anti-reelin (1:500, Millipore), peroxidase conjugated goat anti-mouse (1:3000, Cell Signaling).

Bone Marrow-Derived Macrophage (BMDM) Culture

Femurs and tibias were collected from 6–8 week old female mutant and wild type rl^Orl mice. BMDM cultures were prepared as described [25]. Briefly, the bone marrow was flushed and passed through 26G needles to generate a single cell suspension and the red blood cells were lysed in RBC lysis buffer. The remaining cells were counted, and equal numbers were plated into 6-well plates. Macrophages were differentiated over the course of 7 days in DMEM with addition of 20% FBS, 100 u/mL penicillin/streptomycin, 2 mM L-glutamine (Life Technologies), 0.5 mM sodium pyruvate (Life Technologies), and 5% conditioned media from CMG 14-12 cells, which contains macrophage colony-stimulating factor (M-CSF). For 4T1-conditioned medium studies, medium was collected from 4T1 cells after 48 hours of culture. BMDM were treated with 4T1-conditioned medium for 24 hours. To generate M1 polarized macrophages, BMDM were treated with 50 ng/mL IFN-γ (Life Technologies) and 100 ng/mL LPS (InvivoGen) overnight.

Results

Reelin influences mammary tumor growth and metastasis in vivo

Loss of reelin expression in breast tumors correlates with poor patient prognosis and survival [18]. In order to better understand the impact of reelin on the progression of breast cancer, we examined growth of 4T1-derived mouse mammary tumors in syngeneic reelin-deficient mice (rl^{Orl −/−}). Tumor development was initiated by injecting the mammary fat pads of female rl^{Orl −/−} mice or wild type rl^{Orl +/+} controls with 10⁵ 4T1 cells. Primary tumors grew more slowly in rl^{Orl −/−} mice (Fig. 1a), and weighed significantly less than tumors from rl^{Orl +/+} control mice 25 days after the 4T1 cell injection (Fig. 1b). Mice with mutations in the reelin gene are usually smaller in size than their wild type and heterozygous littermates [26], and the rl^{Orl −/−} mice used in our study also weighed significantly less than the rl^{Orl +/+} controls (Fig. 1c). However, tumor growth had no impact on the overall weight of tumor-bearing animals, as the body weight of rl^{Orl −/−} and rl^{Orl +/+} mice did not change significantly from day 1 to day 25 (Fig. 1c). The metastatic spread of 4T1 cells to the lungs was assessed by a micrometastasis assay. In rl^{Orl +/+} mice, the tumor cells on average comprised 0.36% of all cells in lung suspensions (Fig. 1d). We observed a significant reduction of the metastatic burden in the lungs of rl^{Orl −/−} mice (Fig. 1d, e).

Fig. 1. 4T1 mammary tumor growth and lung metastasis in rl^{Orl −/−} mice.

(a) Growth of primary tumors in rl^{Orl −/−} (n=10) and rl^{Orl +/+} (n=10) control mice. (b) Wet weight of primary tumors collected 25 days after 4T1 cell injection. (c) Total body weight of animals on the day of 4T1 injections (t = 0) and the day of sacrifice (t = 25 days). (d) Quantification of metastatic burden in the lungs. (e) Representative images of 4T1-derived metastatic colonies from the lungs of rl^{Orl +/+} and rl^{Orl −/−} mice used for quantification in (d). Bar = 5 mm. *P < 0.05, **P < 0.01, ***P < 0.001. Statistical significance determined using two-tailed, unpaired Student’s t-test.

To determine if the decrease in lung metastasis in rl^{Orl −/−} mice was correlated with the size of the primary tumor, metastatic burden was compared in mice bearing similar-sized tumors. For these studies, we allowed tumor growth to progress in some rl^{Orl −/−} mice until the tumors reached the equivalent size as those seen in wild type controls at 25 days after 4T1 implantation. Lung tissue samples were collected from rl^{Orl −/−} mice at 29 days after 4T1 injections and used for micrometastasis assays. Metastatic burden in these animals was not significantly different from wild type controls (p = 0.48), suggesting that metastatic capacity of 4T1 cells was not impeded, but that metastatic progression was correlated with tumor size (Additional Files: Fig. S1a, b). These results suggest that the absence of reelin from the host environment impedes growth of primary 4T1-derived tumors in vivo.

Reelin does not directly affect migration or growth of 4T1 cells in vitro

Next, we asked whether reelin could impact tumor growth via direct effects on 4T1 cells. Previous studies showed that primary mammary epithelial cells (MECs) normally express reelin, Dab1, and the reelin receptors ApoER2 and VLDLR [13]. We assessed the expression of the reelin pathway components in 4T1 cells by qPCR, and found that both reelin and Dab1 were significantly downregulated in comparison to normal MECs (Fig. 2a). 4T1 cells expressed higher levels of ApoER2, while expression levels of Vldlr were similar to those in primary MECs (Fig. 2a). Primary MECs respond to reelin by slowing their migration rate [13], and we hypothesized that this effect would be abolished in 4T1 cells because they lacked Dab1. As we expected, 4T1 cells did not respond to reelin in the conditioned medium from reelin-expressing HEK293T cells in transwell assays (Fig. 2b). The proliferation of 4T1 cells was also unaltered in the presence of reelin (Fig. 2c). Full-length reelin (400 kD), as well as two smaller reelin fragments (300 kD, 180 kD) resulting from protein processing were all present in the conditioned medium from reelin-expressing HEK293T cells (Fig. 2d), suggesting that these results were not due to lack of available reelin. Thus, despite the expression of canonical reelin receptors in 4T1 cells, reelin does not appear to directly modulate the growth or migration of 4T1 cells in vitro.

Fig. 2. Expression of reelin signaling pathway components and the effect of reelin on growth and migration of 4T1 cells.

(a) qPCR analysis of relative expression levels of reelin signaling pathway components in 4T1 cells. Gene expression levels are normalized to those in wild type mammary epithelial cells (MEC). (b) Transwell migration assay of 4T1 cells in the presence of reelin conditioned media (Reelin-CM) or controlled conditioned media (293T–CM). (c) Number of 4T1 cells grown in the presence or absence of reelin for 24 or 48 hours. (d) Western blot of reelin expression in conditioned media from reelin expressing HEK293T cells (Reelin-CM) or conditioned media from control cells (293-CM). 20 µg of total protein were loaded into each lane.

Primary tumors grown in rl^{Orl −/−} mice display alterations in blood vessel formation

Histological analysis of the primary tumors from rl^{Orl −/−} mice did not reveal gross morphological differences from rl^{Orl +/+} controls. In both cohorts, layers of tumor cells and infiltrating immune cells surrounded the necrotic tumor core (Fig. 3a, b). Cell proliferation, as assessed by Ki67 staining, was also not significantly different in rl^{Orl −/−} animals (Fig. 3c, d). Tumor angiogenesis was impeded, as significantly fewer CD31-positive blood vessels were observed in primary tumors from rl^{Orl −/−} mice (Fig. 3e, f). However, analysis of angiogenic gene expression in primary tumors by qPCR showed no differences in levels of vascular endothelial growth factors A, B, and C (Vegfa, Vegfb, Vegfc), or transforming growth factor β (Tgfb) (Fig. 4). Additionally, levels of the anti-angiogenic factor chemokine (C-X-C motif) ligand 10 (Cxcl10) were not significantly different (Fig. 4).

Fig. 3. Characterization of proliferation and angiogenesis in primary tumors from rl^Orl/ mice.

(a) H&E staining of primary tumors from rl^{Orl −/−} and rl^{Orl +/+} control mice. Asterisk - tumor cells, double arrowheads – necrotic core. Bar = 200 µm. (b) High magnification view of the areas outlined in (A). Arrow - infiltrating immune cells. Bar = 30 µm. (c) Ki-67 labeling of primary tumors from rl^{Orl −/−} and rl^{Orl +/+} mice. Bar = 30 µm. (d) Quantification of (c), rl^{Orl +/+} n=7, rl^{Orl −/−}: n=7. (e) CD31 labeling of primary tumors from rl^{Orl −/−} and rl^{Orl +/+}mice. Bar = 50 µm. (f) Quantification of (e), rl^{Orl +/+} n=7, rl^{Orl −/−}: n=6. ns - not significant, *P < 0.05. Statistical significance determined using two-tailed, unpaired Student’s t-test.

Fig. 4. Gene expression analysis of primary tumors from rl^Orl ^−/− mice.

qPCR analysis of angiogenic genes (Vegfa, Vegfb, Vegfc, Tgfb), chemokines (Cxcl10, Cxcl2, Cxcl5), and metalloproteinase Mmp-9 in primary tumors from rl^Orl ^−/− mice. Gene expression levels are relative to those in tumors from rl^{Orl +/+} controls. *P < 0.05. Statistical significance determined using two-tailed, unpaired Student’s t-test.

TAMs are altered in the absence of reelin

Progression of 4T1 tumors is marked by the expansion of the myeloid cell compartment, with subsequent infiltration by immune cells, such as tumor associated macrophages (TAMs), into the primary tumor microenvironment [27]. TAMs are recognized as major contributors to tumor angiogenesis [28]. Mice with the reeler mutation were previously reported to have deficits in the function of macrophages and T cells [29]. We hypothesized that the rl^{Orl −/−} environment may lead to alterations in the number or function of TAMs, which could contribute to the reduction in tumor angiogenesis. Consistent with this hypothesis, we saw increased levels of the chemokine Cxcl5 and the matrix metalloproteinase 9 (Mmp-9) mRNA in tumors from rl^{Orl −/−} mice (Fig. 4). CXCL5 and MMP-9 are factors secreted by TAMs and other myeloid cells [30], and we reasoned that their augmented expression may be indicative of elevated numbers of TAMs in the primary tumors from rl^{Orl −/−} animals. Labeling of primary tumor sections with α-F4/80 antibody, a marker of TAMs [27], showed an increase in the area occupied by F4/80-positive cells in tumors from rl^{Orl −/−} mice (Fig. 5a, b, c). Thus, the elevated levels of CXCL5 and MMP-9 in the tumors from mice that lack reelin may be due to an increase in TAM infiltration.

Fig. 5. Analysis of tumor associated macrophages in primary tumors from rl^Orl ^−/− mice.

(a) F4/80 labeling of primary tumors from rl^{Orl −/−} and rl^{Orl +/+} mice. Bar = 50 µm. (b) High magnification view of the areas outlined in (a). Bar = 20 µm. (c) Quantification of the area occupied by F4/80-positive cells, rl^{Orl +/+} n=6, rl^{Orl −/−}: n=6. (d) qPCR analysis of iNOS (M1), Arg1 (M2), Mrc1 (M2), and IL-10 gene expression in primary tumors from rl^{Orl −/−}mice, gene expression levels are relative to those in tumors from rl^{Orl +/+} controls. (e) Western blot of Arg1 and β-actin expression in primary tumors from rl^{Orl −/−} and rl^{Orl +/+}mice. (f) Quantification of band intensities in (e). *P < 0.05, **P < 0.01, ***P < 0.001. Statistical significance determined using two-tailed, unpaired Student’s t-test.

The activation state of TAMs determines their effect on cancer progression. M2-activated macrophages promote tumor angiogenesis and provide a variety of factors to elicit a sustained tumor-promoting Type 2 immune response. In contrast, abundance of M1-activated TAMs impedes angiogenesis and tumor growth, and results in a tumoricidal Type 1 immune response [31, 32]. To test whether reelin influenced the M1/M2 balance in TAMs, we examined the expression levels of several cytokines associated with M1 and M2 macrophages in the primary tumors from rl^{Orl −/−} mice and wild type controls (Fig. 5d). Absence of reelin from the host environment resulted in altered levels of these genes. Specifically, we observed increased levels of a hallmark M1 cytokine inducible nitric oxide synthase (iNOS), while the levels of arginase 1 (Arg1) and mannose receptor, C Type 1 (Mrc1) mRNA, two genes expressed by M2-activated macrophages [31, 33], were downregulated in primary tumors from rl^{Orl −/−} mice (Fig. 5d). Western blotting confirmed the reduction in Arg1 levels (Fig. 5e, f). Additionally, we observed higher levels of interleukin-10 (IL-10) in tumors from rl^{Orl −/−} mice (Fig. 5d). IL-10 is a broadly expressed cytokine with anti-inflammatory properties, and is produced in response to factors that promote a Type 1 immune response [34, 35].

To assess the effect of 4T1 cells on rl^{Orl −/−} macrophages, we cultured bone marrow-derived macrophages from rl^{Orl −/−} mice and wild type controls in the presence of 4T1-conditioned medium (Additional Files: Fig. S2). Similar to observations in primary tumors, levels of iNOS were slightly increased in rl^{Orl −/−} macrophages treated with 4T1-conditioned medium, while levels of Arg1 and Mrc1 were slightly decreased. However, IL-10 levels were not significantly different in rl^{Orl −/−} macrophages treated with 4T1-conditioned medium than in wild type control macrophages. Our results indicate that in the absence of reelin, TAMs may shift to an M1-like state and elicit a tumor-restrictive Type 1 immune response.

Absence of reelin skews the M1 activation state of bone marrow-derived macrophages

To test if macrophage activation was altered in the absence of reelin, we differentiated bone marrow-derived macrophages (BMDM) from rl^{Orl −/−} mice and wild type controls and polarized them to an M1 state with interferon-γ (IFNγ) and lipopolysaccharide (LPS) [36]. IFNγ and LPS treatment robustly induced the expression of M1-associated genes IL-1β, IL-6, IL-12, tumor necrosis factor-a (Tnfa), and iNOS (Additional Files: Fig. S3). Unstimulated rl^{Orl −/−} BMDM expressed higher levels of IL-1β compared to unstimulated wild type control BMDM, while the expression of other M1 genes was not altered (Fig. 6a). M1-polarized rl^{Orl −/−} macrophages expressed significantly higher levels of IL-6 and iNOS compared to wild type control M1 macrophages (Fig. 6b). In addition, levels of IL-10 were 3-fold higher in M1 macrophages from rl^{Orl −/−} mice compared to control M1 macrophages. In this model of M1 macrophage activation, IFNγ primes the macrophages for a rapid response to LPS, which triggers toll-like receptor (TLR) signaling and production of cytokines associated with the Type 1 immune response [37]. TLR signaling induces IL-10 expression in macrophages in order to negatively regulate the inflammatory response [34, 38, 39]. Our results suggest that the upregulation of IL-10 in rl^{Orl −/−} mice may be a part of a similar negative feedback mechanism. Thus, BMDM display a higher propensity towards M1 activation in the absence of reelin.

Fig. 6. Cytokine expression levels in baseline and M1-polarized BMDM from rl^Orl ^−/− mice.

(a, b) IL-1β (M1), IL-6 (M1), IL-12 (M1), Tnfa (M1), iNOS (M1), and IL-10 mRNA expression levels in (a) unstimulated BMDM and (b) M1-polarized BMDM from rl^{Orl −/−}mice, analyzed by qPCR. Gene expression levels are relative to those in the respective rl^Orl +/+ BMDM controls. *P < 0.05, **P < 0.01, ***P < 0.001. Statistical significance determined using two-tailed, unpaired Student’s t-test.

Discussion

The goal of the studies presented here was to better understand the role of reelin signaling in breast cancer growth and metastatic progression. We found that reelin deficiency delays primary tumor growth and lung metastasis. We showed that 4T1 cells do not lose their metastatic capacity in the absence of reelin. Previous studies reported that the size of the primary 4T1 tumor correlated with the extent of the metastatic burden [40, 41], and 4T1 cells metastasized primarily via blood vessels [42]. In agreement with these results, we found that the reduction in lung metastases in rl^{Orl −/−} mice is likely due to the smaller size and poor vascularization of the primary tumors.

Although reelin did not directly affect proliferation or migration of 4T1 cells in vitro, we cannot rule out the possibility that reelin affects 4T1 cells via alternative mechanisms. For example, absence of reelin may lead to alterations in the gene expression profile of 4T1 cells, resulting in a tumor-restrictive microenvironment. The reduction in CD31-positive labeling of primary tumors from rl^{Orl −/−} mice suggested a deficit in tumor angiogenesis. Surprisingly, we observed increased levels of the Mmp-9 transcript in primary tumors from rl^{Orl −/−} mice. MMP-9 activity is provided by myeloid cells in the tumor microenvironment and is required for tumor vasculogenesis [43]. However, the pro-angiogenic effects of MMP-9 are contingent on its activation via a proteolytic cascade [43]. The absence of reelin could affect the activation of MMP-9, but we cannot make definitive conclusions based on the available data. Alternatively, the deficit in vascularization of tumors in the absence of reelin could be due to alterations in endothelial cells, but past studies did not find a role for reelin in angiogenesis, and the vascular network of reeler mice is normal [12, 44, 45].

The increased levels of M1-specific iNOS and decreased levels of M2-specific Arg1 in primary tumors from rl^{Orl −/−} mice indicate the predominance of M1-activated TAMs that may impede tumor growth and angiogenesis. In addition, macrophages from rl^{Orl −/−} mice display increased propensity for M1 activation. These results suggest that reelin may affect activation of macrophages in the tumor microenvironment. The influence of reelin signaling on the development and function of the immune system has not been studied extensively. However, one study found functional deficits in macrophages and T cells from reeler ^−/− animals on a C57/B6 background [29]. These reeler mutants differ from our model because they completely lack reelin. In addition, peritoneal macrophages from these mice showed elevated production of IL1β when activated by LPS in vitro [46, 47]. Another group proposed a role for reelin in clustering of the serotonin transporter (SERT) in blood lymphocytes, as peripheral lymphocytes from reeler ^−/− mice displayed spreading of the SERT clusters [48]. However, we cannot rule out the possibility that other cell types in the tumor microenvironment, such as stromal fibroblasts, myeloid-derived suppressor cells or tumor-infiltrating lymphocytes, may also be affected by the loss of host-derived reelin, causing a delay in tumor progression. These possibilities can be addressed in future studies.

Previous studies on reelin signaling in the context of tumor progression focused mainly on the expression of reelin pathway components in the primary tumor tissue, or in cancer cell lines [14,18, 49–51]. For instance, downregulation of reelin was observed in primary breast tumors, and loss of reelin expression correlated with poor survival. In addition, reelin expression was epigenetically silenced in breast cancer cell lines [18]. It is therefore interesting that in our model, loss of functional reelin protein in the host environment delayed the growth of normally aggressive 4T1 tumors. However, our finding that the 4T1 cell line does not express reelin correlates with previously published results. The human triple-negative breast cancer cell line MDA-MB-231 was reported to have a low baseline expression of reelin [18]. Transfection of this cell line with the full-length reelin construct decreased the migratory and invasive potential of MDA-MB-231 cells [18]. Because reelin impedes the migration of epithelial cells, the results of previous studies imply that cancer cells may downregulate reelin expression in order to become more migratory and invasive. On the other hand, our studies demonstrate that reelin may be necessary for tumor-promoting responses of host immune cells, and therefore may play a dual role in cancer progression.

In summary, our findings provide evidence for additional non-neuronal functions of reelin. Our data demonstrate that absence of host-derived reelin delays mammary tumor growth and metastatic progression. Our observations offer insight into the importance of stromal reelin in breast cancer progression, and suggest a potential role for reelin in the development and function of the innate immune system.

Supplementary Material

10911_2017_9373_Fig7_ESM. Fig. S1: Metastasis of 4T1 cells in rl^{Orl −/−} and rl^{Orl +/+} mice bearing similar-sized tumors.

(a) Wet weight of primary tumors collected 25 days (rl^{Orl +/+}, n=8) and 29 days (rl^{Orl −/−}, n=9) after 4T1 cell injection. (b) Quantification of metastatic burden in the lungs. ns – not significant. Statistical significance determined using two-tailed, unpaired Student’s t-test.

10911_2017_9373_Fig8_ESM. Fig. S2: Cytokine expression levels in BMDM treated with 4T1-conditioned medium.

Cytokine expression levels in rl^{Orl −/−} and rl^{Orl +/+} BMDM treated with 4T1-conditioned medium for 24 hours. Gene expression levels are relative to those in rl^{Orl +/+} BMDM controls. *P < 0.05, ns – not significant. Statistical significance determined using two-tailed, unpaired Student’s t-test.

10911_2017_9373_Fig9_ESM. Fig. S3: Cytokine expression levels in rl^{Orl +/+} BMDM treated with IFNγ and LPS.

Cytokine expression levels in rl^{Orl +/+} BMDM treated with IFNγ and LPS. Gene expression levels are relative to those in unstimulated rl^{Orl +/+} BMDM controls.