MHC Class I Mosaic Mice Reveal Insights into Control of Ly49C Inhibitory Receptor Expression in NK Cells¹

The MHC class I molecules affect the development of CD8⁺ T lymphocytes and NK cells (1, 2, 3). This has most clearly been revealed by studies of MHC class I-deficient mice, in particular mice with a disruption in the gene coding for β₂-microglobulin (β₂m)⁴ (4, 5). MHC class I-deficient mice have low numbers of CD8⁺ T lymphocytes due to impaired positive selection (4, 5). Residual CD8⁺ T cells in these mice have an altered specificity, most likely as a consequence of selection on low levels of MHC class I ligands and/or an altered MHC class I-presented peptide repertoire (6, reviewed in 2). In contrast to CD8⁺ T cells, MHC class I-deficient mice have normal numbers of NK cells (7–9, reviewed in 1 and 3). As for CD8⁺ T cells, the specificity of the NK cells in MHC class I-deficient mice is altered. NK cells from MHC class I-deficient mice are tolerant toward MHC class I-deficient targets, while similar targets are readily killed by NK cells from wild-type mice (7–9, reviewed in 1 and 3).

NK cells recognize MHC class I molecules with specific receptors. Interaction between these receptors and MHC class I molecules on target cells results in an inhibitory signal that prevents killing of the target cells (10, 11, 12, 13). In the mouse, MHC class I-specific inhibitory receptors are dimeric C-type lectins that belong to the Ly49 family (14). Recent studies have shown that host MHC class I molecules influence the expression levels of Ly49 receptors on NK cells (15, 16, 17, 18, 19, 20, 21, 22, 23, 24). For example, it was recently demonstrated that the Ly49C inhibitory receptor, with a specificity for H-2K^b (25, 26), is expressed at high levels in β₂m^−/− mice of the H-2^b haplotype, but is expressed at significantly lower levels in corresponding wild-type mice and other strains of mice expressing H-2K^b (22, 23). Similar observations have previously been made with the Ly49A receptor (15, 16, 17), which has H-2D^d as a ligand (10). Interestingly, different levels of Ly49 inhibitory receptor expression on NK cells correlate with their specificity (27). However, relatively little is known about how MHC class I molecules affect expression of Ly49 receptors.

In an attempt to re-express a functional β₂m gene in β₂m^−/− mice, we generated a panel of founder mice with an unexpected mosaic expression of MHC class I molecules. This allowed us to study lymphocyte development in an environment where MHC class I-positive (MHC class I⁺) and -deficient cells had coevolved. The analysis of the influence of MHC class I on the expression of the Ly49C inhibitory receptor on NK cells turned out to be of particular interest. In the present study, we demonstrate that expression of the Ly49C inhibitory receptor on NK cells is a consequence not only of MHC class I expression in the environment but also by the expression of MHC class I molecules by the NK cells themselves.

For the generation of transgenic mice, a 14-kb XhoI fragment containing the β₂m gene (28) was used to produce transgenic mice as previously described (29), using β₂m^−/− mice as egg donors and fertile males. The plasmid containing the mouse β₂m gene cloned into pSV2gpt was a kind gift from Dr. P. Robinson (Harrow, U.K.) C57BL/6 (B6) and β₂m^−/− mice (both H-2^b) were bred and maintained at the Microbiology and Tumor Biology Center, Karolinska Institute, in line with institutional guidelines. β₂m^−/− mice (5) were originally obtained from The Jackson Laboratory (Bar Harbor, ME). All β₂m^−/− mice used were backcrossed to a B6 background.

For FACS analysis of splenocytes, erythrocyte-depleted spleen cells from individual mice were resuspended in RPMI 1640 medium supplemented with 10% FCS and loaded onto equilibrated nylon wool (Polyscience, Warrington, PA) columns. The loaded columns were incubated at 37°C for 1 h and the nonadherent cells were subsequently eluted from the column with RPMI 1640/10% FCS. Before staining, cells were washed and resuspended in 200 μl PBS with 1% FCS. For FACS analysis of peripheral blood leukocytes (PBL), approximately 60 μl heparinized blood was collected. Erythrocytes were lysed by three consecutive 5-min incubations in 200 μl of FACS lysis solution (Becton Dickinson, Mountain View, CA; diluted 1:10). Before staining, cells were washed and resuspended as above. Anti-H-2K^b (AF6-88.5), anti-H-2D^b (KH95), anti-CD8 (53-6.7), and anti-NK1.1 (PK136) mAbs were obtained from PharMingen (San Diego, CA). Anti-CD4 (KH-CD4) were from Immunokontakt (Frankfurt, Germany). The anti-Ly49C specific mab (4LO3311) has been described (30). All mAbs were used at a concentration of 10–20 μg/ml. Incubations with mAbs as well as streptavidin-conjugates (Life Technologies, Paisley, U.K.) for biotinylated Abs were done for 30 min on ice. Cells were analyzed by flow cytometry on a FACScan (Becton Dickinson) flow cytometer.

Organs or tissue sections were embedded in Tissue-Tek (Sakura Finetek USA, Torrance, CA) and frozen in liquid nitrogen. Frozen sections (6-μm) were cut, air-dried, fixed in ice-cold acetone, dried, and stored at −20°C. Before staining, the sections were rehydrated in Tris-buffered saline and then incubated with 0.1% BSA-C (Avrion, Wageningen, the Netherlands) for 15 min to block unspecific protein interactions. The sections were incubated over night with a FITC-conjugated anti-H-2K^b mAb (AF6-88.5), obtained from PharMingen, at a concentration of 10 μg/ml. For immunohistochemistry, the sections were washed and then incubated for 40 min with a rabbit anti-FITC anti-serum (V0403), obtained from Dako (Glostrup, Denmark), at a 1:300 dilution. After washing, samples were finally incubated for 40 min with a swine anti-rabbit anti-serum conjugated with horseradish peroxidase (P0217), obtained from Dako, at a 1:40 dilution. Horseradish peroxidase localization was revealed by incubating with a diaminobenzidine substrate for 10 min. The sections were lightly counterstained with hematoxyline before mounting. The images were captured in pseudo color using a digital fluorescence microscope (Leica DM-RXA; Leica, Nussloch, Germany) and processed in Adobe Photoshop (Adobe, Mountain View, CA) to modify colors and to enhance contrast and intensity.

Anti-H-2^d specific CTL were generated in bulk mixed lymphocyte cultures by using splenocytes from B6, β₂m^−/− and MHC class I mosaic mice as responders and irradiated splenocytes from DBA/2 mice as stimulators. Cytotoxicity was assessed in a standard ⁵¹Cr-release assay.

For generation of NK cells activated with rIL-2 (referred to as LAK cells), 25 × 10⁶ splenocytes were put in 10 ml α-MEM tissue culture medium supplemented with 10% FCS, 10 mM HEPES buffer, 2 × 10⁻⁷ M 2-ME, antibiotics (complete medium), and 1500 U/ml human rIL2 in 25-cm² tissue culture bottles at 37°C in 10% CO₂. After 4 days, nonadeherent and adherent cells were removed, washed once, and resuspended in complete medium without rIL-2 and stained for flow cytometry analysis.

β₂m^−/− mice of the H-2^b haplotype express severely reduced levels of H-2K^b and D^b at the cell surface (4, 5). In an attempt to re-express a functional β₂m gene in β₂m^−/− mice, we generated a panel of founder mice with an unexpected mosaic expression of MHC class I molecules. Briefly, 13 of 59 founder mice were typed positive for the β₂m transgene (data not shown). Flow cytometry analysis of PBL showed that 3 of 13 transgene-positive mice (founder mice 33, 34, and 53, respectively) expressed MHC class I molecules on all PBL, resembling the pattern observed in wild-type mice (Fig. 1). Ten of 13 transgene-positive mice had an unexpected mosaic pattern of MHC class I expression, with an overall equal distribution of MHC class I⁺ and -deficient cells among PBL (founder mice 7, 11, 29, 36, 37, 44, 47, 49, 51, and 57, respectively). Genetic and molecular analysis of these founder mice, and offspring from these mice, suggested a late integration of the transgene as the most likely explanation for the observed phenotype, leading to expression of the transgene in one of two cells at the two-cell stage of embryonic development. Offspring from the MHC class I mosaic mice, i.e., (MHC class I mosaic × β₂m^−/−)F₁ mice, were either entirely MHC class I⁺ or MHC class I-deficient, and no β₂m transgene product could be amplified by PCR from the MHC class I-deficient offspring (data not shown). Taken together, these mice fulfill all criteria of being natural β₂m⁺ and β₂m⁻ chimeras, and will from here be referred to as MHC class I “mosaic mice” when not else noted.

In all mice with an MHC class I mosaic expression pattern, the levels of MHC class I expression on MHC class I⁺ cells were equal to those of wild-type mice, and no difference in levels of expression of H-2K^b or D^b was observed (Fig. 1). The mosaic MHC class I expression pattern was readily detectable among purified subpopulations of T cells, B cells, and NK cells, as revealed by flow cytometry analysis of PBL and single-cell suspensions of lymphoid organs (data not shown) and by immunohistochemical (Fig. 2, A and B) analysis of different organs and tissues. Several mice were observed for >1 yr, and the MHC class I mosaic phenotype was preserved in all mice over time (data not shown).

All mosaic mice analyzed had normal numbers of CD8⁺ T cells in the periphery (Table I). Thus, the mosaic expression of MHC class I molecules was sufficient to restore positive selection of CD8⁺ T cells. CD8⁺ T cells in the mosaic mice were capable of mediating normal allospecific and MHC class I-restricted T cell responses (Fig. 3; data not shown). Previously published results have shown that the specificity of the residual CD8⁺ T cells selected in an MHC class I-deficient environment, such as in the β₂m^−/− mouse, differs from that of corresponding CD8⁺ T cells selected in wild-type mice in that the former CD8⁺ T cells readily kill target cells expressing normal levels of self-MHC class I molecules (2, 6). In light of this observation, it was of interest to note that the CD8⁺ T cells from the MHC class I mosaic mice failed to kill target cells merely as a consequence of expression of normal levels of self-MHC class I molecules and that the mice did not have any visible pathological changes as revealed by visual inspection of mice during autopsy or by microscopical examination of tissue sections (data not shown).

The MHC class I mosaic mice had normal numbers of NK cells. Given a mouse model with normal numbers of NK1.1⁺ cells and mosaic MHC class I expression, it became of particular interest to study the expression of the Ly49C receptor in these mice. This is the only known Ly49 inhibitory receptor with specificity for an H-2^b class I molecule. Splenic NK1.1⁺ cells from β₂m^−/− mice express high levels of Ly49C, while expression is markedly lower on NK1.1⁺ cells from B6 mice (Fig. 4 and Table II; 23 . When Ly49C expression was assessed on NK1.1⁺ splenocytes and PBL from the mosaic mice, it was observed that levels of Ly49C receptor expression were low on all NK1.1⁺ cells, resembling the expression levels of those observed in B6 mice (Fig. 4). However, detailed comparison indicated that levels of Ly49C expression on the MHC class I-deficient/NK1.1⁺ splenocytes from the mosaic mice were on average somewhat higher than the levels on the MHC class I⁺/NK1.1⁺ cells (Fig. 4 and Table II). Nevertheless, the levels of Ly49C expression were still significantly lower on the MHC class I-deficient/NK1.1⁺ population from the mosaic mice than the levels observed on corresponding MHC class I-deficient NK1.1⁺ NK cells from β₂m^−/− mice (Table II). These results suggest that, to a significant extent, environmental factors (most likely MHC class I⁺ cells), affect levels of Ly49C expression. This conclusion could not merely be drawn from a comparison of β₂m^−/− and B6 mice, because the different levels of Ly49C receptor expression in these mice could either be a consequence of MHC class I molecules expressed in the environment or associated with the levels of MHC class I molecules expressed by the NK cells themselves.

The indications of relatively higher levels of Ly49C expression on MHC class I-deficient/NK1.1⁺ splenocytes than on corresponding MHC class I⁺/NK1.1⁺ splenocytes from the same mosaic mouse (Table II) indicated a potentially interesting role for MHC class I molecules in the NK cells themselves in determining expression levels of the Ly49C receptor. This notion was substantiated in experiment where NK cells were cultured for 4 days in rIL-2-generating LAK cells. In LAK cell cultures, marked changes in expression of Ly49C in the MHC class I-deficient population of NK1.1⁺ cells from the mosaic mice were observed (Table III). These cells now expressed high levels of Ly49C, resembling those observed on NK1.1⁺ LAK cells from β₂m^−/− mice. In contrast, the Ly49C expression on corresponding MHC class I⁺/NK1.1⁺ cells from the same LAK cell culture was similar to what was observed in B6 LAK cell cultures (Table III). This result strengthens the notion that MHC class I molecules on the NK cells themselves may affect the expression of Ly49C molecules. The results further suggest that the environmental factors that may serve to repress Ly49C expression on MHC class I-deficient/NK1.1⁺ cells in vivo are lost in vitro during culture in rIL-2.

In the present study, we have examined lymphocyte development in MHC class I mosaic mice generated through a transgenic approach in β₂m^−/− mice. These mice had regained normal numbers of CD8⁺ T cells, which had a target cell specificity similar to that of wild-type mice. Consequentely, the mice did not develop any signs of autoimmunity. They also had normal numbers of NK cells. The analysis of NK cells in these mice provided interesting insights into the role of MHC class I molecules in regulating expression of the Ly49C inhibitory receptor. The present results demonstrate that calibration of expression of the Ly49C inhibitory receptor is a direct consequence of the expression of MHC class I molecules on surrounding cells, an observation in line with the initial observations made in studies of the regulation of Ly49A inhibitory receptor expression in bone marrow chimeric mice (15, 16). Furthermore, they demonstrate that expression of Ly49C is affected by MHC class I molecules expressed by the NK cells themselves.

Differential expression of Ly49 receptors in vivo may be important to ensure proper regulation of NK cells, i.e., the ability to detect cells expressing reduced levels of certain or all MHC class I gene products, yet allowing maintenance of self-tolerance (15, 16, 17, 27, 31). It may at first glance seem paradoxical that a certain receptor is expressed at lower levels in mice expressing the ligand for the receptor. One explanation for this observation has been provided by a “calibration model” (17, 32), which suggests that the expression of Ly49 receptors may be calibrated according to the expression levels of the appropriate ligand in the host. According to this model, expression of low levels of an inhibitory receptor would make the NK cell more prone toward recognizing the difference between MHC class I^high and MHC class I^low targets by only being able to receive an appropriate inhibitory signal from the MHC class I^high targets. In contrast, NK cells with high levels of expression of an inhibitory receptor would receive an inhibitory signal from both MHC class I^high and MHC class I^low targets, making such NK cells relatively less useful. A recent study comparing the reactivity of Ly49A^high and Ly49A^low NK cells supports this notion (27). In line with the reasoning above, it remains an open question how the MHC class I-deficient cells in the MHC class I mosaic mice manage to avoid NK cell-mediated rejection. It may be so that NK cell tolerance is also controlled at levels other than the expression of Ly49 receptors (discussed in Refs. 23, 33, 34).

At present, little is known about the molecular mechanisms affecting Ly49C receptor expression. It cannot be excluded that they operate on an post-transcriptional level (20, 21). The present study strengthens the notion that MHC class I molecules expressed in the environment are involved in calibration of expression of (at least some) Ly49 receptor(s) (15, 16, 17, 20, 21, 24) and demonstrates this to be the case for the Ly49C receptor. The present study also suggests that MHC class I molecules expressed by the NK cells themselves may affect expression of the Ly49C receptors. This appears particularly effective under circumstances when calibration due to environmental factors function less well, such as during 4-day cultures of NK cells in IL-2. It is important to note that these results are not specific for the Ly49C receptor. Kåse and collaborators have obtained similar results in their analysis of the expression of the H-2D^d-binding receptor Ly49A in a transgenic strain of B6 mice expressing an H-2D^d gene product in a mosaic pattern (35) . Taken together, through the analysis of MHC class I mosaic mice generated through a transgenic approach in β₂m^−/− mice, the present results suggest that expression of the Ly49C inhibitory receptor on NK cells is calibrated in response to expression of MHC class I molecules in the environment as well as to the expression of MHC class I molecules expressed by the NK cells themselves.

We thank Drs. A. Kåse, P. Höglund, and K. Kärre as well as Dr. C. Sentman for fruitful discussions and for communicating unpublished results; E. Nilsson for excellent help with microinjections; M. L. Solberg and M. Hagelin for technical assistance; and Drs. A. D. Diehl and L. Van Kaer for providing comments to the present manuscript.