Background
Natural Killer (NK) cells were discovered almost 40 years ago due to their ability to kill tumor cells with no prior sensitization [
1]. Since then, extensive knowledge has been gained about their instrumental role in tumor immunosurveillance [
2]. NK cells are capable of killing tumor cells via multiple mechanisms. The ability of an NK cell to kill another cell is controlled by a balance of activating and inhibitory receptors expressed on their cell surface that allow the NK cells to sense self versus damaged cells [
3]. Recently, it has been shown that in addition to preventing tumor formation, NK cells can eradicate large solid tumors [
4] and kill mammary cancer stem cells [
5]. Unfortunately, in several malignancies, including breast cancer, NK cell activity as well as expression of activating receptors, is often suppressed [
6,
7]. Tumors promote this down regulation by the secretion of molecules such as TGF β and IL-10 [
7-
9]. Recent reports suggest that NK cells within the tumor may actually support tumor growth [
10,
11]. Alterations in NK cell activity are reversible, as NK cells rapidly respond to their environment [
7,
12]. The ability to shift the NK cell phenotype from inhibition/tumor promotion to activation will be essential for the use of NK cells against cancer.
IL-15 is a cytokine that has effects on both the innate and the adaptive immune system. IL-15 promotes the differentiation, proliferation and activation of NK cells and the formation of a subset of memory CD8 T cells [
13-
15]. This has been confirmed in IL-15 TG mice which have increased activated NK cells and increased proportions of memory CD8 T cells, whereas IL-15 KO mice lack NK cells and have decreased CD8 T cells [
16,
17]. The ability of IL-15 to promote both NK cell and CD8 T cell responses has led to interest in IL-15 as a cancer immunotherapy. Most
in vivo studies investigating the effects of IL-15 have used subcutaneous engrafted or lung metastasis cancer models. For example, several studies found that IL-15 TG mice were resistant to engrafted tumor formation [
18,
19]. IL-15 has been administered by several routes and use of each of these methods has impaired tumor growth or metastasis [
20-
25]. The protection observed was either NK cell and/or CD8 T cell dependent [
18-
20,
22]. While many treatment strategies have been successful in engrafted and metastatic models, it is unknown if this will translate into a spontaneous epithelial cancer model where tumors initiate and grow alongside an intact tolerized immune system.
In this study, we crossed IL-15 KO and IL-15 TG mice with a spontaneous breast cancer model (MT) to create IL-15 KO/MT and IL-15 TG/MT mice. MT mice express the polyoma MT antigen under the mouse mammary tumor virus long terminal repeat [
26]. In MT mice, multifocal adenocarcinomas form and these frequently metastasize to the lung [
26]. The MT model on a C57BL/6 background is a good model of human breast cancer as tumor formation is sequential and goes from focal hyperplasia to mammary intraepithelial neoplasms to carcinoma
in situ and ends with multiple invasive tumors [
27,
28]. IL-15 KO/MT, MT and IL-15 TG/MT were followed for tumor formation and endpoint. We characterized the immune environment both systemically and intra-tumorally and determined the relative contribution of NK and CD8 T cells to the protection we observed in IL-15 TG/MT mice. Lastly, we confirmed that when human NK cells were exposed to a similar cytokine environment as was observed in IL-15 TG/MT tumors, they were capable of killing human breast tumor cells.
Methods
Animal models
Mice were bred and maintained in the McMaster Central Animal Facility in “clean” rooms with a 12 hour day/night schedule and standard temperature controls. Procedures were approved by the McMaster Animal Research Ethics Board and comply with the guidelines set out by the CCAC. MMTV-MT mice (Dr. Gendler, Mayo Clinic, AZ) were crossed to IL-15 KO (Taconic, Germantown, NY) and IL-15 TG mice (Dr. Caligiuri, Ohio State University, OH) to generate IL-15 KO/MT and IL-15 TG/MT mice (C57BL/6 background). C57BL/6 control mice were purchased from Charles River (Quebec, Canada).
Tumors
In the subcutaneous model, a MT cell line, established from a spontaneous MMTV-MT tumor (Mayo Clinic, Arizona), was injected (1 × 105) subcutaneously. Mice were monitored 3 times per week for tumor formation/endpoint. In the spontaneous model, mice were palpated weekly for tumor formation and endpoint (tumors >10 × 10 mm). To examine metastasis, lungs from each group of mice were harvested at 120 days of age, perfused with 2% paraformaldehyde, embedded and sectioned 2 times 100 μM apart. Haematoxylin and eosin (H&E) stained sections were scored as positive or negative for the presence of tumor cells.
Histology/immunohistochemistry
Tumors were excised from multiple mice per group and embedded in Tissue-Tek® OCT (Sakura) or fixed in 2% paraformaldehyde. Fixed sections were stained with H&E (n > 10 per group). Immunohistochemistry was performed on OCT sections for CD8α (PharMingen- #550281; 1:50) and CD4 (PharMingen- #550278; 1:50) using a Rat on Mouse Kit (Biocare Medical- #RT517H). Colour was developed using an AEC Chromogen substrate solution (Sigma). For quantitation, 5 random fields of view per section were counted in 5 mice per group (blinded). A TUNEL assay (ApopTag In Situ Apoptosis Detection Kit- Millipore) was also performed as per manufacturer’s instructions on 3 size matched tumors per group.
Flow cytometry
To generate single cell suspensions, tumors were excised, digested (3 mg/ml Collagenase A, 0.025 mg/ml DNase I (Roche) - 45 min., 37°C, with shaking), and filtered sequentially (70 μm, 40 μM). Spleens were also collected and a single cell suspension was created by squishing. Red blood cells were removed with ACK lysis buffer. Cells were incubated with anti-mouse CD16/32 (eBioscience- #14-0161-86) (1 in 100, 15 min, 4°C) and then stained for markers including: NK1.1, CD69, NKG2D, NKp46, CD8, CD4, CD3, CD44, PD-1, CD27, CD62L and IFNγ/Perforin (intracellular flow cytometry using BD-cytofix/cytoperm) (eBiosciences/BD Biosciences). For flow cytometry analysis of tumors, the first gate was drawn on singlets followed by CD45+ cells (leukocytes) to exclude tumor cells. Fluorescence minus one (FMOs) controls were performed for all experiments. Samples were run on the BD LSRII or CANTO flow cytometer and FlowJo (Tree Star, Ashland, OR) was used for analysis.
T cell stimulation
Cells were isolated from spleens and tumors as described above and CD8 T cells were isolated using a CD8 T cell selection kit (Stem cell- #18753). Purity was assessed by flow cytometric analysis and was >80% for tumors and >90% for spleens. 96 well plates were coated overnight with purified 1 μg/ml anti-CD3 (eBioscience- #16-0031-82) and 5 μg/ml anti-CD28 (eBioscience- #16-0281-82) antibodies at 4°C. The next day, plates were washed and 5 × 105 CD8 T cells/well were added. Supernatants were collected 48 hours later. In the case of intracellular flow cytometry, the same process was followed, but after 12 hours, GolgiStop (BD Biosciences) was added for 10 hours prior to flow staining.
ELISA/Cytokine analysis
IFNγ (DY485), TNFα (DY410E) and IL-12 p40 (DY499) ELISAs were performed using Murine DuoSet Kits from R&D Systems (Minneapolis, MN) as per manufacturer’s instruction. For IL-18 levels, 3 tumors per group (size matched) were homogenized and pooled in an equal volume and sent to Rules Based Medicine (RBM, Austin, TX) for multi-analyte protein analysis (RodentMAP® version 2.0).
NK/Tumor Cell Killing assay
Mouse
NK cells were isolated from the spleens of C57BL/6 mice (PE selection kit Stem cell- #18551 with NK1.1-PE, 2 μg/ml). The MT cell line was labelled with 5-6-carboxyfluorescein diacetate succinimidyl ester (CFSE, 5 μM) and incubated with the NK cells at various Effector:Target ratios (1:1, 5:1, 10:1) for 5 hours (MT cells alone = basal lysis). After 5 hours, 7-amino actinomycin D (7-AAD, BD Biosciences, 5 μl/tube) was added and flow cytometry was performed to quantify the percentage of CFSE and 7-AAD positive cells (dead MT cells). Specific lysis was calculated via the following formula:
$$ \%\ specific\ lysis = \frac{100 \times \left(\%\ sample\ lysis - \%\ basal\ lysis\right)}{100 - \%\ basal\ lysis} $$
Human
Lymphocytes were isolated from PBMCs with Ficoll-Paque Plus (StemCell- #07907) and NK cells were selected with a human NK cell enrichment kit (StemCell- #19055, >90% purity of CD56 + CD3- NK cells). The resultant NK cells were cultured for 16 hours in IL-2 (100 U/ml) or IL-12 (10ng/ml)/IL-15 (20ng/ml)/IL-18 (100 ng/ml)(Peprotech). Cells were washed 4X before being cultured with CFSE (5 μM) labelled MDA-231 (NCI-60 panel) cells at various E:T ratios (1:1, 5:1, 10:1) for 5 hours (MDA-231 cells alone = basal lysis). Cells were then stained with CD45-PE (BD Biosciences) antibody and immediately before flow cytometry, 7-AAD (BD, 5 μl/tube) was added to identify dead cells. FMO controls were included in each experiment. Flow cytometry analysis was performed to determine PE-CFSE + 7AAD+ cells and the above formula was applied.
Antibody depletion
For the subcutaneous model, mice were given 2 injections, one day apart of anti-NK1.1 (PK136 mouse IgG2, hybridoma HB191; ATCC) (200 μg/dose) antibody intraperitoneally. Two days later, MT cells were injected subcutaneously (and the 3rd dose of NK1.1 antibody was administered). NK1.1 depleting antibody was delivered every 3-4 days to maintain the depletion. In the spontaneous model, IL-15 TG/MT mice were given 2 doses of 200 μg anti-NK1.1 mouse IgG antibody or 100 μg of anti-CD8α mouse IgG (clone 2.43; ATCC) intraperitoneally one day apart starting at 4 weeks of age. Depletion was continued every 3-4 days for the NK1.1 and once per week for the CD8 antibody for the duration of the experiment. At endpoint, animals were sacrificed and spleens/tumors were examined for depletion (n ≥ 3). Tumors were fixed (2% paraformaldehyde) and H&E sections were prepared.
Statistical analysis
Statistics were performed in GraphPad Prism. T tests or One-way ANOVAs (Bonferonni’s post test) were performed, depending on the number of groups to be compared and error bars represent standard error of the mean (SEM). Survival curves were analyzed with the Log Rank test (Mantel-Cox).
Discussion
While IL-15 has been under investigation as a cancer immunotherapeutic for the last decade, investigation has focused on tumor models that do not closely mimic spontaneous tumor formation in humans. It is known that injecting tumor cell lines subcutaneously or intravenously is a useful, but rather artificial system in which it is easier to develop immune responses to the tumor. In addition, there is a lack of studies that have examined the impact of IL-15 on solid epithelial tumors such as breast cancer. To examine the role of IL-15 in a more relevant model we utilized an immunologically tolerant mouse model of spontaneous mammary tumor formation (MT) and examined tumor formation in the absence of IL-15 (IL-15 KO/MT) or with IL-15 overexpression (IL-15 TG/MT). Overall, IL-15 TG/MT mice had increased survival when compared to either MT or IL-15 KO/MT mice. In contrast IL-15 KO/MT mice had faster tumor formation and decreased survival when compared to MT or IL-15 TG/MT mice. These results are similar to the anti-tumor effects of IL-15 that have been observed in other engrafted and metastatic models of melanoma and colon cancer, but it is one of the first reports of this in a spontaneous model of breast cancer [
18,
19,
21].
IL-15TG/MT mice formed tumors, but these tumors had a very different phenotype than IL-15 KO/MT or MT tumors. This included large areas of cell death, a higher proportion of NK cells as well as increased CD8 T cell infiltration. Increased CD8 T cells or NK cells within the tumor is a positive prognostic factor in many human and mouse tumor types [
32,
36,
37]. In many cases though, NK cells within the tumor express inhibitory receptors instead of activating receptors and have low cytotoxicity [
7,
38]. To examine NK cell phenotype, we identified NK cells using flow cytometry as NK1.1 + CD3- cells. NK1.1 is commonly used as a marker for NK cells in C57BL/6 mice. It is a member of the NKRP1 receptor family and while it is thought to be an activating receptor its’ ligand is unknown [
39-
41]. The NK cells within IL-15 TG/MT tumors possessed higher levels of both activating receptors (NKG2D, NKp46) and other markers of activation (CD69). Recently, ligands for NKG2D and NKp46 were found to be expressed on human primary breast tumors and breast tumor cell lines [
9]. In addition, blockade of either NKp46 or NKG2D decreased the ability of NK cells to kill breast tumor cells that expressed ligands to these receptors [
9]. We also found that there was a higher percentage of CD27
high NK cells in IL-15 TG/MT tumors than in MT tumors. CD27 expression, in addition to CD11b expression, has been used to define mature mouse NK cells into subsets [
30,
42]. In progression from less mature to more mature: CD27
lowCD11b
low to CD27
highCD11b
low to CD27
highCD11b
high to CD27
lowCD11b
high [
42]. Importantly, CD27
high NK cells were found to have a higher degree of effector function, including cytotoxicity and cytokine production [
30]. More recently it was found that CD27 on human NK cells could also be used to define NK cell subsets [
43,
44]. In contrast to what has been found with mouse NK cells, human CD27+ NK cells have been associated with low cytotoxic activity and high ability to secrete cytokines [
43,
44]. Overall, NK cells found within IL-15 TG/MT tumors are likely more capable of killing breast tumor cells than those found in MT tumors.
We observed increased CD8 T cells within IL-15 TG/MT tumors, but CD8 T cells within the tumor are not always functional as they can be anergic and/or exhausted [
33,
35]. While IL-15 KO/MT tumor CD8 T cells had high levels of exhaustion markers (PD-1) and lacked IFNγ production, IL-15 TG/MT CD8 T cells had very low levels of PD-1 and produced large amounts of IFNγ. This is in contrast to another report that found that treatment with IL-15 in a metastatic model of colon carcinoma led to increased PD-1 expression on CD8 T cells in the spleen [
45]. It is likely that the short term administration of IL-15 or the model system used accounted for the discrepancies in our observations. In addition, a higher proportion of CD8 T cells in the IL-15 TG/MT tumors were CD44+ and CD62L
high, which are markers of central memory CD8 T cells. Central memory CD8 T cells are thought to be extremely effective in anti-tumor defence [
31,
46]. We also observed a high proportion of unique NK1.1+ CD8 T cells in IL-15 TG/MT tumors. This cell type has been previously identified as highly cytotoxic (high perforin/granzyme level) and able to produce large amounts of IFNγ [
47,
48]. In our model, while some of these cells produce IFNγ, they were not the major source. While we have not examined this here, it is interesting to speculate about whether these unique CD8 T cells developed in the tumor or migrated to the tumor from elsewhere. We do know that a low percentage of these cells can be found in IL-15 TG mice in other organs such as the spleen (data not shown), so they are not completely unique to the tumor environment. The ability of IL-15 to induce expression of other NK cell markers such as CD56 in human CD8 T cells has also been reported in a variety of models [
49,
50]. Thus, this effect does not appear to be limited to our mouse models or to one type of NK cell receptor.
In previous studies involving IL-15, protective effects were found to be NK cell or CD8 T cell dependent [
18,
19,
22]. In IL-15 TG/MT mice, NK1.1 positive (includes NK cells, NKT cells, NK1.1+ CD8 T cells) but not CD8 positive cells were the most important cells for increased survival and tumor destruction. The CD8 depletion was substantial but not complete in the tumor and the majority of CD8 T cells that remained in the tumor were NK1.1+, indicating that this cell type was resistant to depletion via this method. It has previously been reported that these cells may be resistant to activation-induced cell death and this may contribute to the inefficient depletion [
48]. Since CD3 + CD8 + NK1.1+ cells were removed by the NK1.1 depletion, we cannot rule out a role for this cell type in the tumor destruction of IL-15 TG/MT mice. The lack of contribution of CD8 T cells to increased survival was surprising due to the fact that they existed in such high numbers and were of the correct phenotype to fight cancer. To confirm this data, we performed a CD8 T cell adoptive transfer experiment. Despite the fact that CFSE labelled CD8 T cells were present in the spleen and tumor at endpoint (data not shown), there was no impact on survival from transfer of either IL-15 TG/MT, MT or IL-15 TG splenic CD8 T cells or IL-15 TG/MT tumor CD8 T cells after a subcutaneous primary MT tumor injection. It is possible that either this aggressive tumor formed too fast for the transferred CD8 T cells to have an impact or that the tumor rapidly lost MHC I expression to compensate for the presence of tumor specific CD8 T cells. Also, MT tumor formation has been found to be slightly different each time and different tumors may express different tumor antigens, some even lose expression of MT itself [
51]. Thus, it is possible that despite taking MT tumor cells to inject from multiple mice and CD8 T cells from multiple mice and pooling them, they may still not have been specific for that tumor. In terms of MHC I loss, a similar phenomenon may be occurring in the spontaneous model. It is also possible that IL-15 overexpression may induce non-specific proliferation of CD8 T cells, not tumor specific responses [
52]. This data indicates that overexpression of IL-15 can generate an anti-tumoral NK cell response that is effective at extending survival in the MT model.
Another promising finding revealed in this model was that overexpression of IL-15 appears to delay the formation of lung metastases. This observation, in a spontaneous model of breast tumor metastasis, strongly indicates that IL-15 has potential therapeutically to prevent metastasis. Previously, this has only been examined in injected models of metastasis. This may be very useful in a clinical setting in which metastasis is frequent and leads to significant increases in mortality.
Conclusions
IL-15 is a promising new cancer therapeutic that is well tolerated in primate models [
53]. It appears to be superior to IL-2 as it has lower toxicity, does not increase T regulatory cells and induces higher levels of NK cell and CD8 T cell effector responses [
53,
54]. Based on the success of IL-15 in animal models, the first clinical trials have begun (NCT01727076, NCT01021059, NCT01572593) in multiple tumor types including melanoma, renal cell carcinoma and non-small cell lung carcinoma patients. Recently, there has been renewed interest in NK cells as a target to activate in the fight against breast cancer. It has been shown that human breast cancer cells express activating ligands as well as death-inducing receptors- both of which NK cells use to correctly identify their target cells [
9,
55]. In fact, expression of NKG2D ligands in human breast cancer was associated with a significant beneficial outcome [
56]. It has also been established that NK cells are capable of eradicating a solid epithelial cancer (fibrosarcoma) and that they may also be able to target breast cancer stem-cell like cells [
4,
5]. Here, we found that when human NK cells were exposed to a similar cytokine environment to that found in IL-15 overexpressed MT tumors, they were highly capable of killing a triple negative breast cancer cell line. Other studies have reported that NK cells grown in IL-15, IL-12 and IL-18 are thought to display long term effector functions and may be memory-like NK cells [
57,
58]. These studies, along with our data, lends credence to the idea that stimulating innate immune cells such as NK cells can be effective clinically against breast cancer primary tumors and metastasis.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AG participated in the design of the study, carried out all experiments, analyzed and generated the figures and drafted the manuscript. MC and TK participated in carrying out the animal experiments and editing of the manuscript. AA conceived of the study and participated in its design. All authors read and approved the final manuscript.