NF-κB p50/p65 Affects the Frequency of Ly49 Gene Expression by NK Cells¹²³

Natural killer cells represent a critical component of the innate immune response to pathogens and tumors (1). They discriminate and eliminate cells harboring an incorrect self-MHC profile through the expression of receptors for MHC class I, such as CD94 and Ly49 molecules (2, 3, 4). The Ly49 family of MHC class I receptors is characterized by a high degree of variability between mouse strains (5). In fact, mapping studies of the Ly49 gene cluster performed on the C56BL/6 (B6) (6, 7, 8), 129/J (9, 10, 11), and BALB/c (12) genomes show distinct numbers of Ly49 genes in each mouse strain. In addition, developmental studies describe an orderly and sequential expression of the Ly49 molecules on NK cells (13, 14), with Ly49 expression on CD94⁺ NK cells representing the final step of the NK cell maturation process before acquisition of their physiological functions (15, 16). Finally, single-cell RT-PCR analysis of Ly49 expression reveals that each mature NK cell selectively expresses a subset of the complete Ly49 repertoire, the majority expressing one to four Ly49 molecules, indicating that the Ly49 genes are activated by a probabilistic mechanism (17, 18).

We have recently described a novel bidirectional Ly49 promoter, Pro1, that appears to control the expression of Ly49 genes through the probabilistic choice of either forward or reverse transcription (19, 20). Transcripts derived from Pro1 have been detected in liver and bone marrow (BM)⁵ NK cells and in the immature NK cell line LNK (CD3⁻IL-2Rβ⁺CD16⁺CD94⁺Ly49⁻) that lacks NK cell activity (21) but not in splenic NK cells or the mature CD3⁺NKRP1⁺Ly49⁺ NKT cell line EL4 (22), suggesting a role restricted to immature NK cells (19). Detailed analysis of the Ly49g Pro1 element revealed potential binding sites for several transcription factors, including two NF-κB sites. Interestingly, site-directed mutation of the NF-κB sites significantly decreases the Ly49g Pro1 transcriptional activity observed in the immature NK cell line LNK (20).

The mammalian NF-κB family is composed of five members: NF-κB1 (p105 and p50), NF-κB2 (p100 and p52), c-Rel, RelB, and RelA (p65), which form functional NF-κB dimers. These proteins share a Rel homology domain that mediates DNA binding, dimerization, and interaction with specific inhibitory factors, IκB-α, β, and ε (23). In most cell types, NF-κB is present as a NF-κB p50/p65 heterodimer in a complex with IκB-α. Activation of NF-κB requires phosphorylation-induced degradation of IκB by the tripartite IκB kinase (IKK) complex IKKα/IKKβ/IKKγ (NEMO). After NF-κB/p65 phosphorylation, the liberated active NF-κB p50/p65 dimer translocates to the nucleus where it regulates various genes. This process is referred to as the classical pathway of NF-κB activation. The NF-κB/p50 and NF-κB/p52 precursors (p105 and p100) can also form dimers with activating subunits, such as in the RelB-p100 complex, which constitutes an inactive NF-κB complex. Activation of this NF-κB complex is via a nonclassical pathway that requires phosphorylation-induced processing of the p100 precursor that involves IKKα and the NF-κB-induced kinase. The released active p52/RelB dimer enters the nucleus and binds to the DNA (23, 24). NF-κB exerts a pivotal role in both innate and adaptive immunity. Studies of mice lacking one NF-κB member or deficient in the NF-κB activation pathway implicate NF-κB in different stages of T and B lymphocyte development (25).

Although the steps leading to a mature NK cell from a NK cell progenitor are not well established, the studies of certain knockout mice strongly suggest a potential role for NF-κB in the process of NK cell development. The results presented here suggest that NF-κB p50/p65 regulates Pro1 activity and Ly49 gene activation.

C57BL/6 (B6) mice were generously provided by Dr. H. A. Young (National Cancer Institute (NCI)-Frederick, Frederick, MD). NF-κB/p50 null and NF-κB/p52 null mice fully backcrossed on B6 (11 generations) were generated as previously described (26, 27). Congenic B6 (CD45.1) were purchased from Taconic Farms. All mice were kept under specific pathogen-free conditions until use at 8–20 wk of age. Animal care was provided in accordance with the procedures outlined in A Guide for the Care and the Use of Laboratory Animals (National Institute of Health, publication no. 86-23, 1985).

YAC-1 and Ba/F3 cell lines were provided by Dr. J. R. Ortaldo and Dr. D. W. McVicar (NCI-Frederick) and were cultured in RPMI 1640 medium supplemented with 10% FBS, 2 mM l-glutamine, 100 IU/ml penicillin and, 100 μg/ml streptomycin. Rae-1γ-transfected Ba/F3 (Rae-1γ-Ba/F3) and m157-transfected Ba/F3 (m157-Ba/F3) cells lines were selected in puromycin (1 μg/ml) or G418 (4 mg/ml), respectively. The LNK cell line was grown in RPMI 1640 medium containing 10% FBS, 2 mM l-glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin, nonessential amino acids, 10 mM HEPES, 50 μM 2-ME, and 10,000 IU/ml or 200 IU/ml IL-2.

Forward and reverse Ly49g core promoter constructs in pGL3 were obtained as previously described (20). The CMV-p65 expression vector was provided by Dr. C. Kunsch (AtheroGenics, Alpharetta, GA). The phagemid vector pBK-CMV (Stratagene) was used as a control vector for the NF-κB/p65 transfection experiments. TranSilent small interfering RNA (siRNA) control or NF-κB/p65 vectors were purchased from Panomics.

LNK cells were split 1:3 the day before the transfection. A total of 5 × 10⁶ cells in serum-free RPMI 1640 medium was transfected with 10 μg of the Ly49g Pro1 reporter constructs by electroporation using an Electro Square Porator ECM 830 (BTX; Genomics) set at 250 mV with three pulses of 7 ms at an interval of 100 ms. The following plasmid DNA mixtures were added to the Ly49g Pro1 reporter construct to explore the role of NF-κB/p65 in its transcriptional activity: CMV-p65 and/or pK-CMV (10 μg total) or 10 μg of TranSilent siRNA control or NF-κB/p65 vectors. Each sample was added with 0.1 μg of the Renilla luciferase pRL-SV40 control DNA. Luciferase activity was assayed at 48 h using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instruction. The luciferase activity of the Ly49g promoter constructs was normalized relative to the activity of the Renilla luciferase produced by the pRL-SV40 control vector. Each experiment was performed at least three times.

Total cellular protein isolation was performed on 5 × 10⁶ cells/condition. Cells were lysed on ice for 30 min in 50 μl of Triton X-100 lysis buffer (25 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 1% Triton X-100) supplemented with protease inhibitors (Roche). The cell lysate was clarified by centrifugation at 15,000 × g in a refrigerated microcentrifuge. An alternative protein extraction was performed using Novex Tris-Glycine SDS Sample Buffer 2× (Invitrogen Life Technologies) diluted 1/2 in distilled water. Fifty microliters of 1× buffer was added per 5 × 10⁶ cells and mixed by pipetting. The sample was boiled and then spun at 15,000 × g to clear the lysate. For immunoblotting, 20 μg of whole cell lysate per condition was separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membranes were blocked for 1 h in blocking buffer (5% nonfat dry milk, 1× PBS, and 0.001% Tween 20), then probed with the appropriate primary Ab overnight at 4°C. The rabbit anti-p65/RelA (Research Diagnostics), rabbit anti-NF-κB2 p100 (Cell Signaling Technology), goat anti-NF-κB p50 (Santa Cruz Biotechnology), and mouse anti-β actin (Abcam) Abs were used at the following dilutions: 1/2,000, 1/1,000, 1/200, and 1/5,000. The bound Abs were detected with HRP-conjugated mouse anti-rabbit secondary Ab (Amersham Pharmacia), HRP-conjugated rabbit anti-goat secondary Ab, and HRP-conjugated goat anti-mouse secondary Ab (Zymed Laboratories) at, respectively, a 1/20,000 and 1/10,000 dilutions for 30 min at room temperature and developed using the SuperSignal West Pico Chemiluminescent Substrate System (Pierce) according to the manufacturer’s protocol.

The following mouse fluorochrome-conjugated mAb used in this study were purchased from BD Pharmingen: FITC-anti-Ly49A (A1), FITC-anti-Ly49C/I (5E6), FITC-anti-Ly49D (4E5), FITC-anti-Ly49G2 (4D11), FITC-anti-CD244 (2B4), and PerCP-anti-CD3ε (145-2C11). FITC-anti-CD49b (DX5), FITC-anti-Ly49C/I/F/H (14B11), FITC-anti-NKG2A/C/E (20d5), FITC-anti-CD11b (MAC-1, M1.70), PE-anti-CD94 (18d3), PE-anti-CD51 (α_v integrin, RMV-7), PE-anti-CD117 (c-kit, 2B8), PE-anti-CD122 (IL-2Rβ, 5H4), PE-anti-CD43 (eBioR2/60), and allophycocyanin-anti-NK1.1 (PK136) were obtained from eBioscience. Pacific Blue-anti-CD45.2 (Ly5.2, 104) and biotin-anti-CD45.1 (Ly5.1, A20) were purchased from Biolegend. Purified-anti-Ly49H (3D10) and PE-anti-Ly49A (YE148) mAb have been generously provided by Dr. J. R. Ortaldo (NCI-Frederick). The anti-Ly49H mAb was labeled with Alexa Fluor 488 using the mAb labeling kit from Molecular Probes according to the manufacturer’s protocol. Streptavidin-Pacific Orange conjugate was purchased from Invitrogen Molecular Probe.

Single-cell suspensions of mononuclear cells (MNC) were prepared from spleen, liver, and BM. Spleens and livers were removed from 8- to 20-wk-old mice and disrupted. BM cells were isolated by irrigation of the femurs and the tibias. Erythrocyte lysis was performed in ACK Lysing Buffer (Cambrex). Absolute numbers of cells were derived from the total lymphocyte count obtained from a Sysmex KX-21 automatic hemocytometer (Roche Diagnostics). Freshly isolated cells were then incubated for 1 h at 4°C with the indicated anti-NK cell receptor FITC- and PE-conjugated mAbs in combination with PerCP-anti-CD3ε and allophycocyanin-anti-NK1.1. The unbound mAbs were removed by washing twice with FACS buffer (1× PBS, 0.1% BSA, and 0.02% sodium azide). The stained cells were resuspended in FACS buffer and analyzed by four-color immunofluorescence using a FACSort flow (BD Biosciences). NK cells were defined as the CD3⁻NK1.1⁺ cells of the lymphocyte gate. The expression of each tested receptor was analyzed on this lymphocyte subset. Analysis of Ly49 expression on NK cells from NF-κB/p50 null and control chimeras was performed by adding to the previously described mAb mixture Pacific Blue-anti-CD45.2 and biotin-anti-CD45.1 mAbs. The CD45.1⁺ cells were detected with Pacific Orange-streptavidin conjugate. The stained cells were then analyzed by six-color immunofluorescence using a LSRI cytometer (BD Biosciences) with NK cells defined as the CD45.1⁻CD45.2⁺CD3⁻NK1.1⁺ cells of the lymphocyte gate. The acquisition and analysis of all data was performed with the WinMDI 2.8 (TRSI) and CellQuest (BD Biosciences) software, respectively.

NF-κB/p50 null and control hemopoietic chimeras were generated as followed. Briefly, BM cells (1–3 × 10⁶) from 5- to 6-wk-old B6 (CD45.2) or NF-κB/p50 null (CD45.2) mice were transferred into heavily irradiated (900 rad) 5- to 6-wk-old congenic B6 (CD45.1) recipient mice by i.v. injection. Six weeks later, splenocytes, BM, and liver cells were collected and the expression of Ly49 receptors on NK cells was monitored by flow cytometry as described above.

Freshly isolated MNC (2 × 10⁶) were stimulated with 1000 IU/ml IL-2 (BD Pharmingen) and 15 ng/ml IL-12 (R&D Systems), or with PMA (50 ng/ml; Sigma-Aldrich) and ionomycin (500 ng/ml; Sigma-Aldrich) in a 12-well flat-bottom culture plate (Costar) in RPMI 1640 containing 10% FBS and 10 μg/ml brefeldin A (Sigma-Aldrich). The cells were harvested 6 h later and stained for 1 h at 4°C with PerCP-anti-CD3ε and allophycocyanin-anti-NK1.1. The proportion of IFN-γ-expressing CD3⁻NK1.1⁺ NK cells was explored by intracellular staining using PE-anti-IFN-γ (eBioscience) and the Cytofix/Cytoperm Plus Kit (BD Pharmingen) according to the manufacturers’ instructions.

The cytotoxicity assay was performed with freshly isolated spleen MNC and the tumor cell targets described above using a standard 4-h ⁵¹Cr release assay. Fig. 4 shows the results obtained for an E:T ratio of 1:200.

Statistical analysis of wild-type (WT) and NF-κB/p50 null NK cell compartment distributions were obtained using GraphPad Prism software. Comparison of distributions was performed using a Mann-Whitney U test. A p < 0.05 was considered to be significant.

Detailed analysis of the Ly49g Pro1 element revealed two NF-κB binding sites. Moreover, site-directed mutation of the NF-κB sites significantly reduced the transcriptional activity of Ly49g Pro1 observed in the LNK cell line in both reverse and forward directions (20). The previous study indicated an important role of the NF-κB/p50 subunit for the activity of Ly49g Pro1 in LNK cells, and direct binding of NF-κB/p50 to the NF-κΒ binding site of the Pro1 element was demonstrated by chromatin immunoprecipitation analysis. Even though NF-κB functions as a NF-κB p50/p65 heterodimer in most cell types, the NF-κB/p65 subunit could not be detected in Triton X-100 lysates of LNK cells (Fig. 1,A, lanes 1). We therefore cotransfected the Ly49g Pro1 construct with increasing amounts of a NF-κB/p65 expression vector to evaluate the role of NF-κB/p65 in the activity of Ly49g Pro1. As shown in Fig. 1 B, this results in a dose-dependent inhibition of Ly49g Pro1 transcriptional activity in both forward and reverse directions.

Surprisingly, NF-κB/p65 expression was not detected by Western blot analysis in the LNK cells after transfection with the NF-κB/p65 expression vector, even though NF-κB/p65 transcripts were present in the cells (data not shown). Because increased levels of NF-κB/p65 protein were observed when the expression vector was transfected into 293T cells (data not shown), this observation could reflect a high rate of NF-κB/p65 degradation in LNK. Franzoso et al. (28) have reported that NF-κB/p65 can be COOH-terminally truncated by proteolysis during cellular extraction, thus preventing its detection by Western blot analysis. Indeed, when a cell lysate obtained with a SDS-based buffer was used, the NF-κB/p65 protein was detected (Fig. 1,A, lanes 2). In addition, after treatment with the proteosome inhibitor MG132, a high level of the NF-κB/p65 subunit was detectable as soon as 1 h after treatment with 50 nM MG132 (Fig. 1,C). Since the constitutive expression of NF-κB/p65 was established in LNK cells, the apparent ability of NF-κB/p65 to inhibit Pro1 activity was thus investigated by decreasing the expression of the NF-κB/p65 protein by using RNA interference. As expected, Fig. 1,D shows a specific increase in the forward and reverse transcriptional activity of the Ly49g Pro1 element in response to NF-κB/p65 siRNA. Similar results have been observed in the human Jurkat T cell and YT-Indy NK cell lines, both expressing constitutively high levels of NF-κB/p65 (data not shown). Fig. 1 E correlates the increased promoter activity with a concomitant down-regulation of NF-κB/p65 expression in LNK cells. These results indicate that NF-κB/p65 has a negative influence on the activity of the Pro1 element.

In the classical NF-κB activation pathway, NF-κB dimers are maintained inactive in the cytoplasm by IκB molecules, IκB-α associating with NF-κB p65/p50. Activation of NF-κB leads to IκB-α degradation by the proteosome. In parallel, new IκB-α is produced to block DNA binding by NF-κB and promote its nuclear export, thus controlling the reaction (29). It has been shown that in absence of IκB-α, NF-κB/p65 is directed to the proteosome for degradation (30). We investigated the presence of IκB-α, β, and ε proteins in LNK cells. Transcripts and proteins were detected in LNK cells for the three IκB members. In addition, the sequence of an IκB-α cDNA isolated from LNK cells was identical to mouse IκB-α (data not shown). These data suggest that the high level of NF-κB/p65 proteolysis in LNK is not due to a defect in the IκB-α pathway. Because treatment of LNK with the proteosome inhibitor MG132 allows the detection of the NF-κB/p65 by Western blot analysis (Fig. 1,C), we reasoned that MG132 treatment should mimic the effect observed when the NF-κB/p65 expression vector is cotransfected. Indeed, 10-h treatment of LNK cells with MG132 after transfection with Ly49g Pro1 shows the same profile of dose-dependent inhibition of Ly49g Pro1 transcriptional activity in both directions as observed with cotransfection of the NF-κB/p65 expression vector (Fig. 1,F). Because stimulation of NK cells with IL-2 induces NFκB activation, the high rate of NF-κB/p65 degradation observed in LNK cells could be the result of a feedback mechanism induced by culturing the cells in high IL-2 culture medium. Indeed, decreasing the IL-2 level to 200 IU/ml allowed the detection of a low level of NF-κB/p65 by Western blot analysis in a cell lysate processed with a Triton X-100-based buffer (Fig. 1 E). These data support the importance of the amount of NF-κB/p65 protein for the activity of the Ly49 Pro1 promoter.

To further investigate the role of NF-κB/p50 in Pro1 activity and Ly49 gene activation, the Ly49 repertoire was studied in NF-κB/p50 null mice. The percentage of cells expressing individual Ly49 proteins on freshly isolated NF-κB/p50 null NK cells (CD3⁻NK1.1⁺) was compared with that in WT mice (Fig. 2,A). To rule out the possibility that the low percentage of Ly49⁺ NK cells detected in the spleen was due to a trafficking defect, the Ly49 repertoire was investigated in the BM. Ly49 expression in the liver was also examined, because it contains more immature NK cells as compared with the spleen, and Pro1 activity was preferentially detected in the liver (19, 31). In the spleen and the BM of NF-κB/p50 null mice, a statistically significant decrease in the proportion of each Ly49 subset is observed and represents an average decrease of 26–43% in the frequency of NK cells expressing each receptor in the BM and 12–37% in the spleen (Fig. 2, A and B). No difference in the intensity of expression of the Ly49 molecules was observed (data not shown), indicating that the level of Ly49 protein expression on individual NK cells was not affected by the absence of NF-κB/p50. Although the decrease in the percentage of Ly49-expressing NK cells in the spleen appears to be independent of the inhibitory or activating function of the receptors, the severity of the expression defect is inversely correlated with the order in which they are expressed during NK cell development (Fig. 2 B and Ref. 14). WT mice exhibit a similar Ly49 repertoire on NK cells from BM and liver, supporting the hypothesis that the liver could behave as a reservoir of immature NK cells. In contrast, an increased percentage of mature Ly49⁺ NK cells was observed in the spleen as compared to the liver, consistent with previous studies that described liver NK cells as phenotypically more immature than splenic NK cells (31). A similar effect was observed between the spleen and the BM, suggesting adaptation of the Ly49 repertoire outside the BM.

NK cell development and maturation, including expression of Ly49 receptors, occur mainly in the BM and they require interaction with stromal cells (13, 14, 15, 16, 32). Therefore, the observed decrease in Ly49 expression on NK cells could be the result of an indirect effect of NF-κB/p50 deficiency in the stromal cells. To test this hypothesis, we generated hemopoietic chimeras by reconstituting congenic B6 (CD45.1⁺ lymphocytes) mice with BM cells from either control B6 or NF-κB/p50 null (CD45.2⁺ lymphocytes) mice. Fig. 2,C shows the Ly49 repertoire expressed on splenic CD45.2⁺CD45.1⁻ NK cells reconstituted in congenic B6 mice. A significant decrease in the proportion of each Ly49 subset was observed, supporting the data previously obtained in the NF-κB/p50 null mice, and represents an average decrease of 13–71% in the frequency of NK cells expressing each receptor in the BM and the liver and 25–49% in the spleen (Fig. 2 C and data not shown). The intensity of expression of the Ly49 molecules was not affected (data not shown).

Taken together, these data suggest that NF-κB/p50 is involved in the in vivo activation of Ly49 expression in NK cells. The decreased percentage of NK cells expressing Ly49 proteins in the spleen of NF-κB/p50 null mice reflects the decrease observed in the BM, in both NF-κB/p50-null and NF-κB/p50-null chimeric mice, thus excluding reduced trafficking as a mechanism for the altered Ly49 repertoire in the spleen, and suggests that NK cells in NF-κB/p50 null mice could represent a more immature phenotype than in WT mice.

There was no statistically significant difference in the total number of NK cells present in liver, BM, and spleen between WT and NF-κB/p50 null mice, suggesting that the decreased Ly49 expression on NF-κB/p50 null NK cells reflects the presence of a reduced Ly49-expressing NK cell compartment in NF-κB/p50 null mice. The study of NK cell development in vivo has determined five stages of maturation and suggests that CD94 expression precedes Ly49 in the normal development of NK cells in adult mice (16). We therefore investigated the expression of Ly49 vs CD94 on NK cells to explore a potential maturational defect of the NK cell compartment in NF-κB/p50 null mice. Fig. 3 (left panels) represents the expression of CD94 vs seven Ly49 molecules on CD3⁻NK1.1⁺ NK cells in WT or NF-κB/p50-null spleen and BM. The results show that the decreased expression of Ly49 on NK cells observed in NF-κB/p50 null mice correlates with a global decrease in the size of the CD94⁺ Ly49⁺ NK cell compartment (stages III–V). In addition, there is a statistically significantly increase in the number of NK1.1⁺ CD94⁻ Ly49⁻ NK cells (stage II). The first developmental stages in NK cell are dependent on the acquisition of CD122 (IL-2/IL-15Rβ) and CD94/NKG2, CD51 (α_v), CD11b (Mac-1), and CD43, respectively, in stages I and II. Analysis of the expression of these markers on splenic NK cells suggests an accumulation of immature NK cells from both stages I and II in the spleen (Fig. 3, top middle and right panels). Nevertheless, exploration of the expression of these molecules in the NK cell compartment of NF-κB/p50 null BM refutes this hypothesis (Fig. 3, bottom middle and right panels). Indeed, if the concomitant decrease of CD94/NKG2D expression with CD11b and CD49b (DX5) could represent an accumulation of “nonlytic” NK cells (stage II), the lack of decreased expression of stage III markers, such as CD43 and CD51, associated with a statistically significant decrease of CD122 and CD244 (2B4) expression within the CD3⁻NK1.1⁺ does not fit the established model of NK cell development. Additional analysis revealed a partial or full expression of CD2, CD45R, CD27, CD44, and IL-7R by this subset (data not shown), suggesting an accumulation of immature NK/T cells (CD3⁻NK1.1⁺) as previously described in these mice (26, 27, 33), and this pattern was also observed in the BM from NF-κB/p50 null chimeric mice. In this regard, these data do not correlate the decreased expression of Ly49 and NF-κB/p50 deficiency with a block in NK cell maturation.

Ly49 expression on NK cells appears before the major expansion stage in the BM, and fully mature NK cells express high Mac-1 along with the capacity to produce cytokines and exert cytotoxicity (16). In the NF-κB/p50 null mice, no defect in Mac-1 expression is observed in the spleen, with Mac-1^high NK cells representing the majority of splenic NK cells (Fig. 3), suggesting fully developed cytolytic abilities and cytokine production.

The susceptibility of target cells to NK cell lysis is modulated by Ly49 recognition. The decreased level of activating Ly49 expression on NK cells from NF-κB/p50 null mice should reduce NK cell cytolysis observed in the presence of the corresponding Ly49 ligands on target cells. The Ly49H activating receptor binds to the cytomegalovirus m157 protein and activates NK cells (34). We therefore tested the cytotoxicity of NF-κB/p50 null NK cells against m157-transfected Ba/F3 cells. As expected, the 45% decrease in Ly49H expression on splenic NK cells correlates with a 35% reduction of their cytolytic activity against m157-transfected Ba/F3 cells (Fig. 4,A, middle panels). Interestingly, NK cell cytolytic activity against YAC-1 cells was not significantly different when NF-κB/p50 null NK cells were compared with WT B6 NK cells. (Fig. 4,B, right panel). It has been proposed that natural killing of YAC-1 tumor cells is largely NKG2D dependent because they express high levels of NKG2D ligands (35, 36, 37, 38, 39). Rae-1γ represents one of the mouse NKG2D ligands and NK cells can lyse RMA or Ba/F3 cells transfected with Rae-1γ. No difference in NKG2D expression was detected on splenic NK cells between WT and NF-κB/p50 null mice (data not shown and Fig. 4,A, left bottom panel). As shown in Fig. 4 A (right bottom panel), splenocytes from both WT and NF-κB/p50 null mice lyse Rae-1γ-Ba/F3 cells to the same extent, consistent with the results obtained with YAC-1 cells. These data demonstrate that there is not an intrinsic defect in the lytic capacity of NK cells from NF-κB/p50 null mice and support the interpretation that NF-κB/p50 deficiency specifically affects the process of Ly49 gene activation, rather than the development of mature cytotoxic NK cells.

In addition to cytotoxicity, another major NK cell effector function is cytokine production. The ability of the NF-κB/p50 null splenic NK cells to become activated was also assayed by measuring the secretion of IFN-γ after stimulation for 6 h by IL-2 and IL-12 (Fig. 4 C). In the absence of stimulation, no production of IFN-γ was detected from either WT or NF-κB/p50 null NK cells. The effect induced by IL-2 and IL-12 stimulation was similar in both WT and NF-κB/p50 null NK cells. Identical levels of IFN-γ production were observed after PMA and ionomycin stimulation. Taken together, these data indicate that NK cell maturation is not defective in the NF-κB/p50 null mice, confirming that the principle defect in NF-κB/p50 null mice is a specific effect of NF-κB/p50 on the ability of NK cells to activate the expression of the Ly49 genes as predicted by the Pro1 model.

Although there is a decrease in the percentage of NK cells that express a given Ly49 in NF-κB/p50 null mice, there is not a complete loss of Ly49 expression, suggesting that other factors are involved in Ly49 gene expression. Indeed, site-directed mutation of Pro1 NF-κB binding sites does not completely abrogate Pro1 reverse and forward transcriptional activities (20). The NF-κB p52 and p50 subunits are closely related and both are expressed in the LNK cell line (Fig. 1,A). In addition, NF-κB/p52 homodimers can form transactivating complexes when associated with Bcl-3, an unusual member of the IκB family (40). Finally, Western blot analysis of the NF-κB members in NF-κB/p50 null splenocytes revealed an increase in the levels of NF-κB/p100 and NF-κB/p52 but not NF-κB/p65 (Fig. 5,A). We therefore explored the Ly49 repertoire in NF-κB/p52 null mice to investigate a potential role of NF-κB/p52 in the expression of Ly49 genes. No statistically significant defect in the percentage or the absolute number of NK cells was observed in the spleen and the BM of NF-κB/p52 null mice. In the liver, the decrease of the absolute number of NK cells is related to a decrease of the percentage of NK cells (data not shown). The global proportion of Ly49-expressing NK cells is similar to that found in WT for the spleen, liver, and BM (Fig. 5,B, top panels, and data not shown). In addition, the expression of Mac-1 on NK cells is similar to that of WT mice, because Mac-1^high NK cells are found predominantly in the spleen and are Ly49⁺ (Fig. 5,B, bottom panels, and data not shown). As shown in Fig. 5 C, absence of NF-κB/p52 does not induce any alteration of the expression of inhibitory Ly49 receptors on NK cells in the spleen, except for Ly49G. On the contrary, increased expression of the activating Ly49D and H receptors is statistically significant. These data suggest that NK cells in NF-κB/p52 null mice are fully developed, thus excluding the involvement of NF-κB/p52 in the regulation of Ly49 gene expression, at least in a dominant manner.

To be fully functional and thus complete their developmental process, NK cells require the expression of Ly49 receptors on their cell surface. Although the molecular signals governing the maturation of NK cell precursors are poorly understood, recent studies have implicated various transcription factors in the expression of Ly49 molecules and the developmental process of NK cells (32). In the current study, we show that the in vitro activity of the bidirectional Pro1 promoter previously associated with the control of Ly49 gene expression can be modulated by NF-κB/p65. Moreover, a reduction in the percentage of NK cells expressing Ly49s in NF-κB/p50 null mice was observed, supporting an in vivo role of NF-κB/p50 in the control of Ly49 expression via the Pro1 promoter.

Mutational analysis of the Pro1 switch indicated an absolute requirement for the AML-binding portion of overlapping AML/NF-κB sites for promoter activity in either direction (20). The predicted three-dimensional structure of an active Pro1 element would require DNA bending at the overlapping AML/NF-κB site to form either the forward or reverse transcriptional complexes (Fig. 6). The AML-1 protein has been implicated in DNA bending in the TCRβ promoter where direct binding of Ets to AML-1 augments DNA binding by both proteins and the formation of a ternary DNA-protein complex is facilitated by TCF or SRY DNA bending proteins (41). The specific base contacts made by AML-1 are remarkably similar to those used by NF-κB/p65, RelB, and c-Rel (42), indicating that the formation of a NF-κB p50-p65 complex on this element would effectively prevent binding of AML-1, since the NF-κB p50-p65 complex is extremely stable (43). The NF-κB p50/65 heterodimer produces only a slight bend (17 degrees) in DNA in the human IFN-β promoter (44). Therefore, the ability of NF-κB/p65 to inhibit Pro1 activity may be due to the formation of a stable NF-κB p50-p65 complex at this site which excludes AML-1 binding and prevents the formation of the appropriate DNA bend required for Pro1 activity (Fig. 6).

A role for NF-κB in NK cell development and/or Ly49 expression has been suggested by studies of mice deficient in other factors required for the NK cell maturation process. A crucial role for IL-15 in NK cell development and Ly49 expression has been established through studies using mice lacking IL-15 or its receptor (45, 46, 47). Indeed, IL-15Rα chain deficient mice show a reduction in the total number of NK cells with a quasi-absence of Ly49-expressing NK cells (45). Other factors, such as lymphotoxin (LT) α interfere with NK cell development, because mice deficient for LTα or its receptor LTβR show a reduction in the number of NK cells, and the percentage of cells expressing Ly49 proteins is decreased in the LTα^−/− mice (48). Interestingly, the signals triggered by LTα and IL-15Rα are NF-κB dependent (48, 49, 50), consistent with our data indicating that NF-κB is involved in Ly49 gene activation. More specifically, mice deficient in the NF-κB activation pathway such as phospholipase Cγ2 knockout (KO) mice showed a reduction of Ly49 expression on NK cells (51), thus strongly suggesting that NF-κB is involved in the activation of Ly49 gene expression.

Mice lacking other transcription factors required for NK cell development, such as GATA-3, T-bet, and IFN regulatory factor 2, are associated with a reduction of Ly49 expression, decreased NK cell numbers in the spleen, impaired liver cell homing, increased cell apoptosis, reduced secretion of IFN-γ with normal YAC-1 cell killing, and a Mac-1⁺ CD43⁺ expression pattern (52, 53, 54). In NF-κB/p50 null mice, analysis of the splenic NK cell compartment revealed a reduced frequency of Ly49 expression associated with normal killing and IFN-γ production. The Ly49⁻ NK cells possessed a Mac-1⁺ CD43⁺ phenotype that is associated with mature NK cells in WT mice. Taken together, these data suggest that NF-κB/p50 is not required for NK cell developmental stages other than the expression of the Ly49 genes. The functional defects associated with impaired NK cell maturation in the previously cited studies could be related to a direct effect of a defective NF-κB pathway, as shown in mice deficient for both IκB-α and IκB-ε which present a reduced number of NK cells in the spleen and BM, increased apoptosis, lack of IFN-γ secretion, normal YAC-1 cell killing, and a Mac-1⁺ CD43⁺ expression pattern (55).

In addition, there could be different effects on the NK cell maturation stage depending on which pathway of NF-κB activation is involved. Indeed, the decreased expression of Ly49 observed in LTα^−/− splenic and BM NK cells can be restored after injection of IL-15 or LTα, each triggering a different NF-κB activation pathway involving TNFR-associated factor 2 and p52/RelB, respectively (48). Analysis of NF-κB/p52 null mice did not reveal any major defect in Ly49 expression on NK cells, potentially excluding the alternative pathway of NF-κB activation in the regulation of Ly49 gene activation. Analysis of the NK cells in double KO p50/p52 mice could determine the implication of NF-κB/p52 in the regulation of Ly49 gene expression. Unfortunately, the double KO mice present a severe defect in lymphoid development, including disruption of the splenic architecture, with death of the pups around 3–4 wk (56), thus precluding any meaningful assessment of NK cell development. Along these lines, the severe NK phenotypes observed in mice deficient in factors upstream of NF-κB could also result from the alteration of other signals such as cell survival and proliferation. The NF-κB/p50 null mice have been reported to display functional defects in immune responses (27).

In this study, in vitro data suggest that NF-κB/p65 exerts a negative effect on Pro1 promoter activity. Pro1 activity in the LNK cell line is associated with decreased NF-κB/p65 levels, providing an explanation for the previous observation of NF-κB/p50 homodimer binding to the Pro1 NF-κB sites in LNK cells (20). Increased NF-κB/p65 expression induced by transfection of a NF-κB/p65 expression vector or MG132 treatment inhibits Pro1 activity. Perhaps it is increased NF-κB/p65 protein expression in mature NK cells that accounts for the silencing of the Pro1 promoter when the constitutive Ly49 Pro2 promoter is activated. Because the Pro1 stochastic switch is associated with gene activation in immature NK cells, rather than the stable expression of the Ly49s from the Pro2 promoter in mature NK cells, it is important that Pro1 is silenced in mature NK cells to maintain a stable Ly49 repertoire. The current observations support the hypothesis that during NK cell maturation, an undefined signal would lead to binding of NF-κB p50/p65 instead of the NF-κB/p50 homodimer, suppressing the activity of the Pro1 switch in mature NK cells to prevent the spontaneous activation of additional Ly49 genes in mature NK cells.

In this study, we show that NF-κB/p50 and NF-κB/p65 play opposing roles in Pro1 activity and are therefore both required for control of Ly49 expression by limiting the activity of Pro1 to immature NK cells, leading to the stable generation of a stochastic Ly49 repertoire in mature NK cells.

We thank Mernoosh Abshari, Deborah L. Hodge, Sarahjane Locke, Earl W. Bere, and Paul Wright for technical support and helpful discussions.

The authors have no financial conflict of interest.