Role of NK cell subsets in organ-specific murine melanoma metastasis

Abstract

Tumor metastasis plays a major role in the morbidity and mortality of cancer patients. Among solid tumors that undergo metastasis, there is often a predilection to metastasize to a particular organ with, for example, prostate cancer preferentially metastasizing to bones and colon cancer preferentially metastasizing to the liver. Although many factors are thought to be important in establishing permissiveness for metastasis, the reasons for organ-specific predilection of each tumor are not understood. Using a B16 murine melanoma model, we tested the hypothesis that organ-specific NK cell subsets play a critical role in organ-specific metastasis of this tumor. Melanoma cells, given intravenously, readily colonized the lungs but not the liver. NK cell depletion (either iatrogenically or by using genetically targeted mice) resulted in substantial hepatic metastasis. Analysis of NK cell subsets, defined by the differential expression of a combination of CD27 and CD11b, indicated a significant difference in the distribution of NK cell subsets in the lung and liver with the mature subset being dominant in the lung and the immature subset being dominant in the liver. Several experimental approaches, including adoptive transfer, clearly indicated that the immature hepatic NK cell subset, CD27+ CD11b-, was protective against liver metastasis; this subset mediated its protection by a perforin-dependent cytotoxic mechanism. In contrast, the more mature NK cell subsets were more efficient at reducing pulmonary tumor load. These data indicate that organ-specific immune responses may play a pivotal role in determining the permissiveness of a given organ for the establishment of a metastatic niche.

Figure 1. NK depletion results in establishment of liver tumor nodules and increased pulmonary tumor load.

B6 mice were given control Ig or PK136 on day −3 and day 0 (i.p.) followed by i.v. injection of B16 tumor on day 0. Mice were euthanized on Day 14 and tumor nodules counted; there were five mice per group. The increase in tumor nodules in the lung was significant at p = 0.019 (indicated by *). The increase in liver tumor nodules was significant at p = 0.005 (indicated by **). The design of the experiment was such that all mice groups were processed simultaneously. A visual depiction of a similar experiment is in (B) where the bottom row shows livers from NK-depleted mice.

Figure 2. Intravenous inoculation with B16F1 melanoma results in the establishment of liver tumor nodules in NK cell deficient mice.

2×10⁵ melanoma cells were injected i.v. into (A) wild-type, (B) SCID or (C) γc/RAG 2 KO mice (all on a C57BL/6 background). Twelve days later, lungs and livers were excised and tumors enumerated. There were five mice per group.

Figure 3. Cytolytic machinery but not IFN-gamma mediates protection from liver tumor colonization.

(A) Perforin KO mice or wild-type B6 were injected with 3×10⁵ B16F1 i.v. and euthanized 12 days later. The difference in tumor nodules in lungs was not significant while the difference in liver nodules was highly significant (p = 0.004). (B) IFN-γ/IL-12 KO or wild-type B6 mice were injected with B16F1 i.v. and euthanized 12 days later. The differences were not significant (C) Ly49A Tg or wild-type littermate control mice were injected with B16F1 i.v. and euthanized 12 days later. The number of pulmonary tumor nodules was significantly increased in Ly49A Tg mice as compared to WT (p = 0.018). The design of the experiments was such that all groups in each panel were examined simultaneously.

Figure 4. Arrested maturation of NK cells exacerbates lung tumor engraftment, but has minimal impact on liver tumor load.

Wild type or Ly49A transgenic mice were treated with control mouse immunoglobulin, or depleted of NK cells using mAb PK136. 3×10⁵ B16F1 melanoma cells were injected i.v., and mice were euthanized 14 days later for enumeration of tumor burden. Ly49A Tg mice develop more pulmonary tumor nodules than wild type littermates. Elimination of NK cells does not significantly increase the number of tumor nodules in the lung. In contrast, Ly49A Tg mice develop a similar number of liver tumor nodules as wild type controls, indicating that the immature hepatic NK cells confer substantial protection against tumor establishment in this organ.

Figure 5. Organ-specific NK cell subset distribution in WT and Ly49A transgenic mice.

Mononuclear cells were isolated from the spleens, livers and lungs of wild-type (upper row) and Ly49A transgenic mice (lower row) and analyzed by flow cytometry. Software gates were set to identify NK cells (NK1.1+ CD3–), which were then examined for the expression of the maturation markers CD27 and CD11b.

Figure 6. Adoptive transfer of NK cells protects NK cell deficient mice from tumor establishment.

γc/RAG2 mice were injected i.v. with 2×10⁵ sort-purified NK cells from the spleens or livers of wild-type C57BL/6 mice. Two weeks later, recipients were inoculated intravenously with 2×10⁵ melanoma cells. Fourteen days later lungs and livers were excised and tumors enumerated. Six recipients were used per group.

Figure 7. Adoptive transfer of liver NK cell subsets confers variable levels of protection against liver tumor establishment.

γc/RAG 2 KO mice were injected i.v. with 5×10⁴ sort-purified NK cells of the indicated subset phenotypes from the livers of wild-type C57BL/6 mice. Recipients were then inoculated intravenously with 3×10⁵ melanoma cells. Fourteen days later lungs and livers were excised and tumors enumerated. Five recipients were used per group.

Figure 8. Liver mononuclear cells from perforin deficient mice do not confer protection from liver tumor colonization in NK cell deficient recipients.

γc/RAG 2 KO mice or wild type control mice were injected intravenously with 1×10⁶ liver MNC from wild type or perforin deficient C57BL/6 mice. Four days later, recipients were inoculated intravenously with 2×10⁵ B16F1 melanoma cells. Fourteen days later livers were excised and tumors nodules were enumerated.

Figure 9. Lysis of B16 melanoma by hepatic NK cell subsets.

B6 mice were given CpG i.p, their livers were harvested 48 hrs later, a mononuclear cell preparation obtained and labelled with CD5, NK1.1, CD27 and CD11b. Software gates were set on NK cells and the resulting four subsets of CD27 and CD11b dual label were sorted and used in a 4 hr ⁵¹Cr-release assay against B16 melanoma targets. Lytic units were calculated with one lytic unit defined as the number of effector cells needed to effect 30% specific lysis. The values on the X-axis are lytic units per 1×10⁶ effector cells. The lytic potential of the CD27^hi/CD11b^lo subset was statistically significant when compared to the CD27^loCD11b^lo (p = 0.03) and the CD27^loCD11b^hi (p = 0.028) but not when compared to the CD27^hiCD11b^hi subset.