Non–small cell lung cancer (NSCLC) is the leading cause of cancer-related deaths worldwide and results from a complex interaction between carcinogen exposure and inherent susceptibility. Despite its prevalence, genetic factors that predispose to the development of lung cancer remain elusive. Inbred mouse models offer a unique and clinically relevant tool to study genetic factors that contribute to lung carcinogenesis due to the development of tumors that resemble human adenocarcinoma and broad strain-specific variation in cancer incidence after carcinogen administration. Here, we set out to investigate whether strain-specific variability in tumor immunosurveillance contributes to differences in lung cancer. Using bone marrow transplantation, we determined that hematopoietic cells from lung cancer–resistant mice could significantly impede the development of cancer in a susceptible strain. Furthermore, we show that this is not due to differences in tumor-promoting inflammatory changes or variability in immunosurveillance by the adaptive immune system but results from strain-specific differences in natural killer (NK) cell cytotoxicity. Using a newly discovered congenic strain of mice, we show a previously unrecognized role for strain-specific polymorphisms in the natural killer gene complex (NKC) in immunosurveillance for carcinogen-induced lung cancer. Because polymorphisms in the NKC are highly prevalent in man, our data may explain why certain individuals without obvious risk factors develop lung cancer whereas others remain resistant to the disease despite heavy environmental carcinogen exposure. Cancer Res; 72(17); 4311–7. ©2012 AACR.
Investigations of mouse and human models of cancer have significantly advanced our understanding of the innate and adaptive immune responses to malignancies (1). Despite the fact that non–small cell lung cancer (NSCLC) remains the leading cause of cancer-related deaths, only limited progress has been made in understanding mechanisms regulating its immunosurveillance. Quantitative trait loci (QTL) mapping using genome-wide association studies and linkage analysis of lung cancer–susceptible and -resistant strains of mice and families with hereditary lung cancer has led to the identification of several polymorphisms that contribute to this disease (2–5). Such a “global approach,” however, has not led to the identification of genetic differences responsible for immunologic susceptibility to lung cancer. In fact, almost all genes identified by QTL mapping function in a tumor cell–intrinsic manner and can suppress or promote growth of stromal cells both in vitro and in vivo (6). Thus, the study of immunologic processes that contribute to lung cancer susceptibility and resistance may facilitate the design of rational immunotherapy strategies for this disease.
In this communication, we show that bone marrow–derived cells of lung cancer–resistant mice impede the development of urethane-induced tumors. Unlike the case for other malignancies (7), we show that natural killer (NK) cells rather than T lymphocytes play a critical role in immunosurveillance for urethane-induced lung cancer. We further define that strain-specific differences in the NK gene complex (NKC) contribute to resistance to NSCLCs.
Materials and Methods
Animals and reagents
Mice consisted of 6- to 12-week-old males purchased from Jackson Laboratories except for 129/SvEv, which were purchased from Taconic. 129/SvEv.B6-NKC Rag2−/− mice were identified and maintained at Washington University (St. Louis, MO). Microsatellite typing was conducted as previously described (8). Urethane (Sigma) was injected intraperitoneally at 3 mg/g in 3 separate injections over the course of 6 days. Mice were euthanized 4 months after injection of urethane.
Immunodepletion was accomplished with antibodies for CD4 (GK1.5), CD8 (YTS169.4), NK1.1 (PK136), or isotype controls at an initial dose of 500 μg intraperitoneally and a weekly dose of 250 μg intraperitoneally (BioXcell). Such treatment depletes respective cell populations in both lungs and secondary lymphoid organs by flow cytometry (data not shown; ref. 9). Lungs were digested in type II collagenase and stained in saturating concentrations of fluorochrome-conjugated antibodies (eBioscience) for 30 minutes on ice.
Bone marrow chimeras were created by injecting 1 × 107 T-cell–depleted donor bone marrow cells intravenously after lethal irradiation of the host with 10 Gy. These chimeras were injected with urethane 3 months after bone marrow transplantation and are designated as donor genotype → recipient genotype.
Lungs were fixed in 10% formalin followed by 70% EtOH. Tumors were counted and measured by a co-author (X. Lin), blinded to the experimental condition. The individual tumor volume was calculated as (mm3) = 4/3πr3. Tumors were graded on a 4-stage grading system on hematoxylin and eosin (H&E) slides with typical adenomatous hyperplasia as grade I, tumors with round to ovoid uniform nuclei as grade II, moderately pleomorphic nuclei with occasional mitotic figure as grade III, and markedly pleomorphic nuclei with nucleoli, increased mitoses, with or without necrosis and with or without glandular architecture as grade IV (10).
In vitro cytotoxicity assays
51Chromium release was conducted by incubating the target cells with 100 μCi sodium-51-chromate (PerkinElmer) for 1 hour. Effector cells were isolated from secondary lymphoid organs for NK cells and bone marrow for macrophages, monocytes, and neutrophils (Miltenyi Biotec). Specific lysis was expressed as (experimental release − spontaneous release)/(maximum release − spontaneous release) ×100% with 0% specific lysis as lowest expressed value. Blockade of NKC receptors was accomplished using antibodies manufactured in house (by W.M. Yokoyama) at 30 μg/mL.
Analysis was conducted by t test, Mann–Whitney U test, ANOVA, and χ2 analysis as appropriate, based on consultation with statistician Dr. Kathryn Trinkas from the Biostatistics Core of Washington University in St. Louis and Siteman Cancer Center.
Results and Discussion
Bone marrow–derived cells contribute to strain-specific susceptibility to lung cancer
Urethane, or ethyl carbamate, is found in most, if not all, fermented foods and is considered by the International Agency for Research on Cancer as a group 2A human carcinogen for lung cancer (11). Consistent with previous reports (6), we found that the incidence, number, size, and overall burden of tumor vary in the highly susceptible AJ, moderately susceptible 129/SvEv, and resistant C57BL/6 (B6) mice following urethane injection (Fig. 1A). To determine whether bone marrow–derived cells contribute to this process, we next created bone marrow chimeras between the highly susceptible AJ and resistant B6 mice and injected them with urethane. We postulated that the drastic differences in tumor burden between these 2 strain combinations would more accurately allow us to determine whether hematopoietic cells contribute to strain-specific susceptibility to carcinogenesis. We noted that the tumor burden was significantly lower in AJ mice receiving B6 bone marrow than in syngeneic transplants (Fig. 1B). Similar, albeit less pronounced, differences were evident in B6 recipients of AJ bone marrow (Fig. 1C). The fact that AJ hematopoietic cells did not completely revert B6 mice to the susceptible phenotype extends our previous reports that strain-specific differences in nonhematopoietic stromal cells contribute to lung cancer susceptibility (2). Our data, however, now identify a previously unknown contribution of hematopoietic cells to lung carcinogenesis.
Minimal inflammatory changes are evident after treatment with urethane
Inflammation can enhance the development of lung cancer (12). No evidence of pulmonary inflammation, however, was evident after urethane injection by H&E staining (Fig. 2A) or flow cytometric analysis of lung-resident CD45+ hematopoietic cells in AJ, 129/SvEv, or B6 strains of mice (Fig. 2B). No differences in accumulation or activation of neutrophils, macrophages, or inflammatory monocytes were evident as well (Fig. 2B). These data support the notion that urethane can lead to the promotion and induction of lung cancer independent of inflammation. On the basis of these data, we considered it unlikely that differences in susceptibility to urethane-induced lung cancer were due to tumor-promoting inflammatory changes.
The innate rather than the adaptive immune system plays a critical role in immunosurveillance for urethane-induced lung cancer
On the basis of reports showing the critical role of T lymphocytes in eliminating both chemically induced and spontaneous tumors (1, 7), we next set out to determine whether the adaptive immune system of resistant mice can prevent the development of lung cancer. To this end, we injected wild-type B6, B6Rag−/− and B6 athymic nude mice with urethane and analyzed them 4 months later for lung cancer. Surprisingly, we saw no differences in cancer incidence between wild-type and Rag−/− mice (4 of 10 vs. 4 of 13 mice, respectively; P = 0.68) with similar numbers of tumors per mouse and identical overall tumor burden. B6 nude mice seemed especially resistant to the development of lung cancer with only 1 of 13 mice developing tumors (Fig. 3A). To evaluate the possibility that the adaptive immune response may control malignant progression rather than the incidence or tumor burden, individual tumors were histologically graded on a 4-stage scale as previously described (10). No differences in tumor grade were evident between immunocompetent and immunodeficient animals (Fig. 3A). Similarly, T-cell depletion did not impact tumor burden in susceptible AJ mice (Supplementary Fig. S1). Taken together, these data indicate that components of the adaptive immune system, such as T and B lymphocytes, play essentially no role in strain-specific differences in urethane-induced lung cancer.
Our findings are in stark contrast to certain other models of primary carcinogenesis, such as 3-methylcholanthrene (3-MCA)-induced fibrosarcoma, where both innate and adaptive immunity play a key role in controlling the growth and shaping the immunogenicity of cancer (7, 9). Such discrepancies may be due to the unique tolerogenic properties of the lung, as direct antigen administration into this organ can lead to systemic tolerance rather than priming of the adaptive immune response (13).
To investigate the role of the innate immune system, we next depleted NK cells in a cohort of B6 mice by weekly injection of anti-NK1.1 antibody for 4 months after urethane administration. Mice treated in this fashion showed 100% tumor incidence with a significantly larger tumor burden and a more advanced tumor grade than controls (Fig. 3B). Tumors developing in anti-NK1.1–treated mice had a similar size to those treated with control IgG. This finding is consistent with the notion that NK cells are able to eliminate malignant cells early in their transformation, but not once a large tumor has developed. While we recognize that PK136 treatment also depletes other NK1.1+ cells, such as NKT cells, B6 nude and Rag−/− mice that are deficient in NKT cells remain relatively resistant to urethane-induced lung cancer (Fig. 3A). Nevertheless, to corroborate our findings in an in vitro model, we isolated NK cells from AJ, 129/SvEv, and B6 mice and tested lung-cancer directed cytotoxicity in a 51chromium release assay. B6 NK cells were significantly more efficient in lysing both Lewis lung carcinoma (LLC), derived from the B6 strain, and LM-2 lung cancer cell line, derived from a urethane-induced AJ tumor, than AJ and 129/SvEv NK cells (Fig. 3C). Little to no lung cancer–specific cytotoxicity was evident when using other cells of the innate immune system, such as neutrophils or macrophages and monocytes (Supplementary Fig. S2). Collectively, our in vitro and in vivo findings support the notion that resistance to lung cancer in B6 mice is at least partially due to NK-mediated immunosurveillance.
Strain-specific differences in chromosome 6–linked natural killer gene complex contribute to immunosurveillance for lung cancer
The fact that NK cells from highly susceptible AJ and moderately susceptible 129/SvEv mice (Fig. 1A) show an identical level of lung cancer–specific cytotoxicity (Fig. 3C) indicates that factors other than NK cells contribute to urethane-induced lung cancer development as well. At the same time, these data reaffirm that when compared with B6 mice, defects in NK cytotoxicity may contribute to the development of lung cancer in AJ and 129/SvEv strains. The relationship between AJ and 129/SvEv mice is difficult to infer as these 2 strains are virtually unrelated, possess different major histocompatibility antigens and immunologic characteristics (14). Furthermore, AJ mice show several NK cell–specific defects in peripheral function (15) and are highly susceptible to lymphotoxic substances including urethane (16). NK cells from 129 mice, on the other hand, show only limited defects and are able to lyse other targets, such as YAC-1 (17), better then AJ mice (Supplementary Fig. S3). Taken together, these data suggest that another factor may account for the deficiency of lung cancer–specific cytotoxicity in 129/SvEv and AJ NK cells.
NK cell function is regulated by inhibitory and activating receptors that bind target cell ligands to stimulate (or not) NK cell killing. The NKC on distal chromosome 6 contains several clusters of genes encoding the Nkrp1, Ly49, CD94, and NKG2 families of receptors (18). We have recently shown using array-based comparative genomic hybridization and sequence-tagged site analysis by microsatellite markers that different strains of mice can be grouped on the basis of their NKC (8, 19). It is in this region of their genome that the AJ and 129 strains show similarity and differ from B6. To evaluate this further, we compared lung cancer–specific NK cytotoxicity of Balb/c mice, which share NKC homology to AJ and 129, and C57BL/10 mice, which share homology to the B6 strain. Similar differences in lung cancer–specific cytotoxicity were evident between these strains (Supplementary Fig. S4), further implicating strain-specific polymorphisms in the NKC locus and lung cancer–specific cytotoxicity (19). Nevertheless, as these mice differ in other parts of their genome as well, a more rigorous model is required to link the NKC locus to lung cancer susceptibility.
Recombination activating gene-deficient (Rag2−/−) mice, originally generated from a 129-strain ES cell (20), have been backcrossed to both B6 and 129/SvEv by Taconic Farms. Recently, we discovered that one substrain of a Rag2−/− mouse on the 129/SvEv background was incompletely backcrossed. Using microsatellite analysis, we determined that this substrain was congenic with B6 for the NKC genomic region while otherwise containing the near complete genome of the 129/SvEv strain (Fig. 4A; Supplementary Fig. 5). The Pas-1 locus, which contains Kras, is also located on chromosome 6 and is therefore genetically linked to the NKC. By both microsatellite and single-nucleotide polymporphism analysis, however, the Pas-1 locus of this mouse was of 129 origin (Supplementary Figs. S5 and S6). The microsatellite data showing B6 origin of the NKC were confirmed by fluorescence-activated cell-sorting analysis using NK1.1 as a B6-specific antigen and PCR to amplify the 129/SvEv and B6 alleles of the Nkrp1c gene (data not shown). This mouse, designated as 129/SvEv.B6-NKC Rag2−/−, thus represents a unique animal model containing the genome derived from the 129 strain of mice, which are susceptible to lung cancer (6), but the NKC derived from the resistant B6 strain. Importantly, both 129/SvEv and 129/SvEv.B6-NKC Rag2−/− mice display identically high susceptibility to 3-MCA tumorigenesis and 3-MCA–induced fibrosarcomas from both strains display identically high levels of immunogenicity (ref. 7; R.D. Schreiber, unpublished observations).
To investigate the impact of NKC polymorphism on urethane-induced lung cancer, we injected fully backcrossed 129/SvEvRag2−/− and 129/SvEv.B6-NKC Rag2−/− mice with urethane and evaluated lung cancer by necropsy 4 months later. Similar to the finding in the B6 strain (Fig. 3A), no differences in the number of tumors or total tumor burden were detectable between wild-type and Rag2−/− mice on a pure 129 background (3.9 ± 0.7 vs. 3.4 ± 0.6 mm3, respectively; P = 0.54; Figs. 1A vs. 4B). In stark contrast, 129/SvEv.B6-NKC Rag2−/− mice had a significantly lower tumor burden than mice with 129/SvEv NKC (Fig. 4B). NK cells from 129/SvEv.B6-NKC Rag2−/− mice also showed superior lung cancer–specific cytotoxicity in vitro compared with NK cells from 129/SvEvRag2−/− mice (Fig. 4C). These data further indicate that NK cells from the resistant B6 strain of mice, and specifically the B6 NKC alleles, form an efficient barrier to the development of lung cancer. To evaluate what receptors in the B6 NKC are critical for eradication of lung cancer, we repeated the in vitro chromium release experiments using a panel of blocking antibodies. Blockade of NKG2D led to near-complete inhibition of cytotoxicity (Fig. 4D) whereas blockade of other receptors had little to no effect on cytotoxicity. Such data also suggest that NKG2D plays a critical role in lung cancer recognition and imply that polymorphisms in this activating receptor, or other inhibitory receptors that may prevent NKG2D signaling, may contribute to strain-specific differences in lung cancer cytotoxicity.
In contrast to the adaptive immune system, receptors of the innate immune system are encoded in the germline. Such receptors, especially those in the NKC locus, manifest extreme genetic polymorphism that includes both allelic variations within a single gene and structural genomic variation of total gene content (19). While such genetic diversity is most likely shaped by evolutionary pressure brought on by infectious diseases, the impact of this genetic diversity on cancer is unknown. We now show that NK cells, rather than cells of the adaptive immune system, play a critical role in immunosurveillance for urethane-induced lung cancer and show, for the first time, that allelic variation in the NKC of otherwise immunocompetent animals accounts for susceptibility or resistance to lung cancer. Our data may explain why certain individuals develop lung cancer despite minimal risk factors or carcinogen exposure.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
The work was supported by ATS/Lungevity Foundation, the Alvin Siteman Cancer Center Internal Research Grant by The American Cancer Society, NIH (KO8CA131097) and Biostatistics Core (P30 Ca091842), the Rheumatic Diseases Core Center NIH (P30 AR48335), NIH 1R01HL094601, the Barnes Jewish Research Foundation, The American Association for Thoracic Surgery, Advancing a Healthier Wisconsin, Thoracic Surgery Foundation for Research and Education, and the generous support of Charlotte and Sheldon Rudnick. R.D. Schreiber is supported by NIH/NCI CA43059, CA141541 and the Cancer Research Institute. W.M. Yokoyama is an investigator of the Howard Hughes Medical Institute
The authors thank Sanjay Murala, Jessica Archambault, and Michelle Becker-Hapak for their expert technical assistance.
- Received March 12, 2012.
- Revision received June 13, 2012.
- Accepted June 18, 2012.
- ©2012 American Association for Cancer Research.