Selective activation of cAMP-dependent protein kinase type I inhibits rat natural killer cell cytotoxicity.

The present study examines the expression and involvement of cAMP-dependent protein kinase (PKA) isozymes in cAMP-induced inhibition of natural killer (NK) cell-mediated cytotoxicity. Rat interleukin-2-activated NK cells express the PKA α-isoforms RIα, RIIα, and Cα and contain both PKA type I and type II. Prostaglandin E2, forskolin, and cAMP analogs all inhibit NK cell lysis of major histocompatibility complex class I mismatched allogeneic lymphocytes as well as of standard tumor target cells. Specific involvement of PKA in the cAMP-induced inhibition of NK cell cytotoxicity is demonstrated by the ability of a cAMP antagonist, (Rp)-8-Br-adenosine 3′,5′-cyclic monophosphorothioate, to reverse the inhibitory effect of complementary cAMP agonist (Sp)-8-Br-adenosine 3′,5′-cyclic monophosphorothioate. Furthermore, the use of cAMP analog pairs selective for either PKA isozyme (PKA type I or PKA type II), shows a preferential involvement of the PKA type I isozyme, indicating that PKA type I is necessary and sufficient to completely abolish killer activatory signaling leading to NK cell cytotoxicity. Finally, combined treatment with phorbol ester and ionomycin maintains NK cell cytotoxicity and eliminates the cAMP-mediated inhibition, demonstrating that protein kinase C and Ca2+-dependent events stimulate the cytolytic activity of NK cells at a site distal to the site of cAMP/PKA action.

Natural killer (NK) 1 cells constitute a distinct lineage of lymphocytes (1) with the innate ability to specifically recognize and kill certain normal allogeneic leukocytes (2,3), tumor cells, and virally infected cells (4,5) by direct cell-mediated cytotoxicity. The presence of both activating and inhibitory NK cell receptors have been implicated in regulation of NK cytolysis. As formulated in the "missing self" hypothesis (6), a no-kill signal is induced by specific interaction of "self" MHC class I molecules with inhibitory NK receptors. Therefore, the presence of intracellular signaling pathways down-regulating or inhibiting the cytotoxic response may be important in preventing killing of normal autologous cells. In contrast to the "self" MHC class I-induced inhibition of natural killing, activating receptors have not been characterized to the same extent. However, both foreign MHC antigens on normal allogeneic cells and non-MHC antigens expressed on tumor cells have been demonstrated to interact with yet not fully characterized activating NK cell receptors (7)(8)(9)(10). The present view is that NK cytotoxicity is a result of a fine balance between negative and positive signals induced through distinct sets of receptors inhibiting or activating the cytolytic activity (11).
The second messenger cAMP regulates a number of cellular processes (12,13). With the exception of certain ion channels, directly regulated by cAMP (14,15), all known effects of cAMP are mediated via activation of cAMP-dependent protein kinase (PKA). The inactive PKA holoenzyme is a tetramer consisting of two regulatory (R) and two catalytic (C) subunits (16). Two major types of PKA, PKA type I and PKA type II, display different biochemical properties due to differences in their R subunits (RI and RII, respectively) (16). Molecular cloning has revealed even greater heterogeneity of PKA subunits, and at present four regulatory (RI␣, RI␤, RII␣, and RII␤) and three catalytic subunits (C␣, C␤, and C␥) are known (for references, see Ref. 17). Each RI or RII subunit contain two cAMP binding sites, termed site A and site B (18), and the C subunits will be released only when both of these sites are occupied by cAMP. By using site-selective cAMP analogs, a preferential activation of either PKA type I or PKA type II is possible (19 -22). Thus, by employing isozyme-selective pairs of cAMP analogs, correlation between activation of a distinct isozyme of PKA and biological responses can be obtained.
The inhibitory effect of cAMP and PKA activation on proliferation of B and T lymphoid cells is well documented (23)(24)(25). Interestingly, this regulation seems mainly to be mediated through the PKA type I isozyme. PKA type I is necessary and sufficient to confer cAMP-dependent inhibition of lymphocyte proliferation induced through the antigen receptor of T cells (26). A distinct redistribution and colocalization of PKA type I to the antigen receptors in both T cells (27) and B cells (28) is observed, supporting the notion of an isozyme-specific effect of PKA type I. Whether such direct interactions of PKA type I to receptors are limited to "mitogenic" receptors and proliferative responses or may be a general phenomenon of all antigen receptors is not known. In this context, we found it of interest to examine the effects of cAMP and a possible role of specific isozymes of PKA on NK cell cytotoxicity. This would clarify whether PKA type I might also negatively influence acute cytotoxic and not only proliferative responses. Prostaglandin E 2 and forskolin have previously been shown to exert an inhibitory effect on the cytolytic activity of NK cells toward certain tumor target cells (29 -31). However, the effect of cAMP in NK allorecognition and the possibility that cAMP and specific PKA isozymes may be implicated in modulation of the killer activatory signaling pathway have not been investigated.
We have therefore addressed the effects of cAMP in both allorecognition and tumor killing and the presence and functional significance of PKA and PKA isozymes in rat IL-2-activated NK cells. The results show a restricted isozyme expression and a distinct function of the PKA type I isozyme, RI␣ 2 C␣ 2 , in regulating the cAMP-induced inhibition of NK cell cytotoxicity both toward allogeneic and tumor targets. Furthermore, the observation that activation of PKC and perhaps other Ca 2ϩ -dependent events counteract the cAMP-induced inhibition of NK cell function indicates a site of action of PKA proximal to PKC.  (32) were dissolved in phosphate-buffered saline to stock concentrations of 0.01-0.1 M. Concentrations were determined by OD measurements using the respective absorption maxima and extinction coefficients for each cAMP analog. 8-AHA, 8-CPT, and N 6 -Bnz were obtained from Sigma, whereas 8-MA, 8-Pip, N 6 -Phe, (S p )-cBIMPS, (R p )-8-Br, and (S p )-8-Br were obtained from BioLog Life Science Institute (Bremen, Germany). All solutions of cAMP analogs were sterile-filtered before storage and use. Stock solutions of forskolin (10 mM; Calbiochem), prostaglandin E 2 (PGE 2 , 2.8 mM; Sigma), 12-O-tetradecanoylphorbol-13-acetate (TPA, 1 mM; Sigma), and ionomycin (1.3 mM; Sigma) were prepared in ethanol.
Target Cells-Concanavalin A-activated blasts were generated as described (8). Lymphoprep-separated spleen cells were adjusted to 2 ϫ 10 6 cells/ml and cultured for 3 days in RPMI 1640 supplemented with 5% FCS, 5% normal rat serum, 2 mM glutamine, 1 mM sodium pyruvate, 50 M 2-mercaptoethanol, and 5 g/ml of concanavalin A. The mouse T cell lymphoma YAC-1 and the chemically induced mouse mastocytoma P815 were maintained in RPMI 1640 supplemented with 10% FCS. For studies using "allogeneic" combinations, we employed effector NK cells and target concanavalin A blasts from rat strains differing at the MHC only. Previous experiments have shown that in these circumstances PVG or PVG.1U NK cells recognize MHC products of PVG.1AV1 target cells. The term allogeneic subsequently refers to these strain combinations.
In Vitro Cytotoxicity Assay-The standard 4-h 51 Cr release assay for measuring NK cell cytotoxicity has been described in detail previously (8). Briefly, target cells were incubated with 3.7 MBq/ml of Na 2 51 CrO 4 (Amersham Corp.) in RPMI 1640 with 25 mM Hepes and 10% FCS for 1 h at 37°C in an atmosphere of 5% CO 2 /humidified air, washed three times, and added to effector cells in U-bottomed 96-well plates. The effector:target ratio (E:T) was always 50:1 unless otherwise stated. Cell-free supernatants were harvested with a Titertek harvesting system (Skatron, Lierbyen, Norway). Spontaneous release (SR) of [ 51 Cr] in the absence of effector cells was usually 5% of the total incorporation.
Triplicate values at each effector:target cell ratio were included in the analysis and the intra-assay variation was routinely below 5%. The specific 51 Cr release was usually 40 -60% at an effector:target ratio of 50:1, and this represents almost 100% killing of target cells, since some radioactivity is irreversibly bound and remains particulate after lysis.
Treatment of IL-2-activated NK Cells in the Cytotoxicity Assay with Chemicals-IL-2-activated effector cells were distributed in 96-well plates and treated in triplicate with different concentrations of cAMP analogs (diluted in RPMI 1640 with 10% FCS). The cAMP analogs were added 30 min (room temperature) before the target cells, and the cytotoxicity assay was performed as above. Forskolin and PGE 2 were added 10 min before target cells. In experiments combining cAMP analogs with TPA or ionomycin, the phorbol ester and the calcium ionophore were added 5 min before target cells (i.e. 25 min after cAMP analogs). All concentrations refer to the final concentration of compound in wells.

Identification of PKA Subunits in Rat IL-2-activated NK
Cells-The expression of PKA subunits in rat NK cells was assessed in cultures of highly purified IL-2-activated NK cells containing more than 98% NKR-P1 ϩ ,CD3 Ϫ cells (33). Northern analyses demonstrated mRNAs for RI␣, RII␣, and C␣ isoforms of PKA, and Western blotting revealed immunoreactive protein for RI␣, RII␣, and C (data not shown). The distribution between soluble and particulate PKA activity was 2:1, and ion exchange chromatography revealed characteristic peaks of phosphotransferase activity and [ 3 H]cAMP binding identified as PKA type I (70%) and PKA type II (30%) as well as some free RI (data not shown).
NK Cell-mediated Killing of Normal Allogeneic Lymphoblasts Is Inhibited by cAMP-Analysis of effects of cAMP analogs and agents that elevate cAMP on NK cell lysis of normal MHC-mismatched allogeneic lymphoblasts revealed a reproducible inhibition of NK cell cytotoxicity. To examine effects of elevated endogenous cAMP levels, IL-2-activated NK cells were treated with either PGE 2 (10 M) or forskolin (100 M) (Fig.  1A). Furthermore, for comparison, some cells were also treated with a cAMP analog (8-CPT, 300 M). Fig. 1 shows that treatment with PGE 2 and forskolin reduced cytolysis from 55 to 16% specific 51 Cr release, whereas the effect of 8-CPT completely abolished NK-mediated cytolysis (Fig. 1A). To ascertain that the observed effects of cAMP analogs were due to modulation of effector cells and not target cells, 8-CPT-pretreated effector cells were washed to remove the analog prior to addition to the target cells ( Fig. 1B; 8-CPT, effector). As shown, similar results were obtained as when the cAMP analog was present during the whole experiment (8-CPT, both). In contrast, preincubation of target cells with 8-CPT was without effect (8-CPT, target). This showed that the observed inhibitory effects of cAMP on natural killing were due to modulation of the effector and not the target cells.
To ensure that cAMP analogs to be employed in the subsequent studies or breakdown products of these analogs would not influence the 51 Cr release from target cells, the spontaneous release of 51 Cr from target cells in the absence of effector cells was measured in the presence of high doses of all analogs employed in the study. Spontaneous release was between 5 and 10% of total 51 Cr incorporated, and no effect of any of the cAMP analogs on the level of spontaneous release was observed (data not shown). By trypan blue exclusion, the viability of both effector and target cells was constantly Ͼ90% and not affected by treatment with any of the cAMP analogs.
Effect of cAMP on Cytotoxicity at Different Ratios between Effector (E) and Target (T) Cells: Concentration-dependent Effects of cAMP- Fig. 2A shows a typical titration curve of NK cytotoxicity toward allogeneic target cells with graded numbers of effector cells. Similarly, the concentration-dependent effect of 8-CPT was tested at the same five different E:T ratios, and the respective concentrations of 8-CPT needed to reduce cytotoxicity by 50% (IC 50 ) were calculated. 8-CPT inhibited NK cell-mediated cytolysis of allogeneic target cells in a concentration-dependent manner. The IC 50 values for cAMPdependent reduction of NK cell alloreactivity were almost identical at all E:T ratios tested (Fig. 2B). Thus, the inhibitory effect of cAMP on natural killing was not appreciably influenced by the E:T ratio, and further experiments were performed at an E:T ratio of 50:1 in order to maximize differences without saturating the assay.
The Inhibitory Effect of cAMP on NK-mediated Cytotoxicity Is Specific and Operates through PKA-To further investigate the involvement of PKA in cAMP-mediated inhibition of NK cell cytotoxicity, we used a cAMP antagonist ((R p )-8-Br) to reverse cAMP-dependent activation of PKA. IL-2-activated NK cells were treated with increasing concentrations (0 -1000 M) of the cAMP agonist (S p )-8-Br alone or pretreated with increasing concentrations of the complementary antagonist (R p )-8-Br (0 -1000 M) before the addition of a fixed concentration (1000 M) of (S p )-8-Br, and their abilities to lyse allogeneic lymphoblasts were measured (Fig. 3). As expected, a concentration-depend-ent reduction of alloreactivity was observed with the cAMP agonist (S p )-8-Br with an IC 50 of ϳ750 M (additional data not shown). The inhibitory effect of (S p )-8-Br was almost completely reversed by pretreatment of the NK cells with the cAMP antagonist (R p )-8-Br (Fig. 3). The observation that the cAMPmediated inhibition of NK cell cytotoxicity could be blocked by a cAMP antagonist with selectivity for PKA type I (37) clearly testifies to the fact that the effect of cAMP is mediated through PKA.
PKA Type I, but Not PKA Type II, Inhibits NK Cell Cytotoxicity toward Allogeneic Target Cells-To further investigate the possible involvement of PKA isozymes in the regulation of NK cell cytotoxicity, we used isozyme-selective cAMP analog pairs to preferentially activate PKA type I or PKA type II, as described previously (19 -22). As shown in Fig. 4A binds with higher affinity to the cAMP binding site B in both PKA type I and type II, but when used in higher doses it occupies both cAMP binding sites and produces a complete inhibition of NK-mediated cytotoxicity. When 8-MA was combined with a "priming" concentration of 8-Pip (2 ϫ 10 Ϫ4 M), a distinct shift in the apparent IC 50 of 8-MA was observed. At this "priming" concentration, 8-Pip selects the A site only of PKA type I and thus complements 8-MA in the preferential activation of the type I isozyme but not type II. No such synergism was observed when a PKA type II-selective analog pair was used ( Fig. 4B; 8-CPT with high affinity for the B site only of PKA type II combined with N 6 -Bnz selecting the A site of both isozymes). Taken together, this synergistic effect with an analog pair preferentially activating PKA type I, but not with an analog pair selectively activating PKA type II, indicates that PKA type I is sufficient to mediate inhibition of NK cell cytotoxicity.
To establish that the PKA type I-synergistic effects observed in Fig. 4 (upper panels) were not unique to the particular combinations of cAMP analogs used, a panel of site-selective analogs were employed and combined in different ways. These results are summarized in Table I 50 values partly reflect the different affinities for the cAMP binding sites, but they also reflect differences in membrane permeability and metabolic degradation within cells (21). IC 50 values for cytolysis inhibition curves by the different analogs in the presence of priming analog were calculated separately and designated IC 50 * (Table I). The degree of synergism in each combination is presented as the ratio of IC 50 /IC 50 *. These experiments revealed a clear synergism using type I-selective analog pairs but no such synergism using type II-selective analog pairs. Whereas the ratio IC 50 /IC 50 * was close to unity in type II combinations, the PKA type I analog pairs gave ratios between 4 and 5. These findings provide additional strong evidence for an important role for PKA type I as a general modulator in controlling the lytic machinery in NK cells after receptor activation.
Cyclic AMP-dependent Inhibition of NK-mediated Killing of Tumor Targets Is Directed through PKA Type I-Having analyzed the effect of cAMP/PKA in MHC class I-associated natural killing of normal allogeneic lymphoblasts (Fig. 4, upper panels, and Table I), we also wanted to see if the modulatory role of PKA type I in NK cytotoxicity could be extended to encompass MHC class I-independent recognition of tumor targets. As targets we used a mouse T cell lymphoma YAC-1 (Fig.  4, middle panels) and a chemically induced mouse mastocytoma, P815 (Fig. 4, lower panels), both sensitive to IL-2-activated NK cells. Using the same technique as in the allogeneic system, IL-2-activated NK cells were treated with different combinations of site-selective analog pairs preferentially activating PKA type I (left panels) or PKA type II (right panels), and their reactivities against either YAC-1 or P-815 were examined. A concentration-dependent inhibition of cytolysis of tumor target cells by the cAMP analogs 8-MA and 8-CPT was observed with both cell lines. As illustrated, synergism was only observed when a cAMP analog pair (8-MA and 8-Pip) activating PKA type I was used, giving IC 50 /IC 50 * ratios of 5.8 and 4 for YAC-1 and P815, respectively (Fig. 4, C and E). The analog pair 8-CPT and N 6 -Bnz, complementing each other in the activation of PKA type II, gave no such synergistic effect, with IC 50 /IC 50 * ratios of 1.03 and 1,00 (Fig. 4, D and F). Thus, the preferential activation of PKA type I appears necessary and sufficient in the regulation of NK cell-mediated killing both of normal allogeneic lymphocytes and of tumor cells.
Activation of PKC by TPA and Ionomycin Reverses the Inhibitory Effect of cAMP on NK Cell Cytotoxicity-Protein kinase C and Ca 2ϩ have key roles in the intracellular signaling of cells, and activation of the enzyme has a mitogenic effect on B and T cells. It has also been shown that this Ca 2ϩ -/phospholipid-dependent enzyme plays an important role in activating NK cell cytotoxicity (38). We therefore examined whether the phorbol ester TPA activating PKC, a calcium ionophore (ionomycin), or a combination of these influenced the cAMP-induced inhibition of NK-mediated killing of allogeneic lymphoblasts. As shown in Fig. 5, 100 M 8-CPT strongly inhibited NK cell-mediated lysis, and neither TPA nor ionomycin alone were able to reconstitute cytotoxicity. In contrast, the combined treatment with TPA and ionomycin almost completely reversed the 8-CPT-induced inhibition (Fig. 5A), suggesting that the direct activation of the Ca 2ϩ /PKC signaling pathways renders the NK cell insensitive to the effect of cAMP/PKA. The combined treatment of NK cells with TPA and ionomycin also allowed killing of syngeneic target cells (Fig. 5B), and also in this case, the TPA/ionomycininduced cytolysis was insensitive to inhibition by cAMP. This indicates that combined treatment with TPA and ionomycin short-circuits both the effect of the killer inhibitory receptors, normally preventing killing of self targets, and cAMP. DISCUSSION We have addressed the presence and functional significance of cAMP-dependent protein kinase in rat NK cells. Although both PKA type I and type II isoenzymes are present in NK cells, PKA type I is necessary and sufficient for mediating cAMP-dependent inhibition of NK cytotoxicity both toward allogeneic MHC class I-mismatched leukocytes and tumor target cells. Furthermore, treatment with TPA and ionomycin reverses cAMP-mediated inhibition of NK cytolysis, indicating that the Ca 2ϩ /PKC and cAMP signaling pathways have inverse roles in regulating the cytolytic activity of NK cells and that the targets for PKA may be proximal to activation of Ca 2ϩ /PKC.
The ␣ isoforms RI␣, RII␣, and C␣ are ubiquitously expressed in almost all cells and tissues (for references, see Ref. 17), and these are also the primary subunits of PKA in NK cells and constitute approximately 70% PKA type I and 30% PKA type II isozyme. It could be argued that the observation of a PKA type I-specific effect is due to the high levels of PKA type I and that putative PKA type II-specific effects could be masked by the type I enzyme. However, the isozyme-specific effects exerted by both the type I and type II isozymes appear to be more dependent on specific anchoring than on the actual amounts present within the cell (22,26,27,39,40). PKA type I-mediated inhi-bition of receptor-induced B and T cell proliferation is accomplished by a distinct redistribution and localization of PKA type I to the antigen receptor complexes in these cells. PKA type II, on the other hand, localizes mainly to the centrosomes in lymphocytes and is therefore more likely to serve a completely different function in these cells (27,41). Thus, PKA type I may play an important role in all lymphoid cells as a negative regulator of receptor-induced cell proliferation and function in lymphoid cells. It is interesting that PKA type I also has been found at the initiation site of phagocytosis in neutrophilic granulocytes (42), suggesting that PKA type I may be implicated in specific recognition-associated function in all leukocytes.
Characterization and cloning of inhibitory receptors for MHC class I molecules on NK cells have been reported both in rodents and in humans (43,44). Upon ligation with the appropriate "self" MHC class I molecule, the cytotoxic response is inhibited and target cell lysis is prevented (6). Thus, the NK receptors for MHC class I molecules may have evolved to recognize certain class I motifs within the individual with the mode "recognition: no kill." Interestingly the inhibitory function of these receptors involves tyrosine phosphorylation of the receptor and recruitment of tyrosine phosphatases (45,46). Hence, it seems that the inhibitory receptors act by recruiting proteins that can uncouple the activation signal. Activation of NK cells may thus proceed by a release from inhibition or by   (3) 1.0 II a The potency of each analog to inhibit NK cell-mediated cytoxicity by 50% is designated IC 50 (mean Ϯ S.E.). b For experiments using analog pairs that together occupy both the A and B sites of a particular isozyme, one analog (priming analog) was used at a concentration (priming concentration) that would occupy only the one cAMP binding site it has the highest affinity for. Priming concentrations for each analog were selected as approximately 10% of the IC 50 value. The priming concentrations used gave no inhibitory effect alone. The priming concentration of 8-Pip was chosen on the basis of our own experiments giving synergism and previous results in other cell systems (26). As illustrated in Fig. 4, the priming concentration of 8-Pip (200 M) complemented site B-selective analogs without any inhibitory effect alone. The combination of N 6 -Phe/(S p )-cBIMPS was selected for maximal activation of PKA type II (32,37).
c The apparent IC 50 in combination with different priming analogs is designated IC 50 * (mean Ϯ S.E.). d The degree of synergistic activation by each analog pair is indicated by the ratio IC 50 /IC 50 *. specific recognition by certain activating molecules. There is evidence for such activating receptors both in rodents (7-9) and humans (10). Furthermore, this is supported by the notion that alloreactive NK cells in the rat, employed in this study, appear to be regulated both by negative and positive signals from classical (RT1.A) and nonclassical (RT1.C) MHC class I molecules on target cells (33,47). Recent results indicate that the heterotrimeric G proteins G o and G z may be involved in regulating alloreactivity in rat NK cells (48). It is therefore tempting to speculate that G proteins (G z ) acting on adenylyl cyclase may represent a link between receptor-mediated recognition and the intracellular cAMP pathway in NK cells. Cyclic AMP and PKA type I may also be regulated by the level of signaling through ␤-adrenergic and prostaglandin E receptors.
Other reports have focused on the role of PKA-mediated phosphorylation of the small G proteins in regulating NK activity. Proteins of the Rho family are suspected to be involved in controlling the organization of the cytoskeleton, and RhoA is thought to act on actin filaments, thereby altering cellular conformation and regulating movement of the NK cell from one target to another (49,50). PKA-mediated phosphorylation of RhoA serves to detach RhoA from the membrane toward the cytosol. RhoA may certainly be a substrate for a tethered PKA type I involved in modulation of signaling through the killer activatory receptors at the effector:target cell contact sites.
Interestingly, cAMP has been shown to exert a concentrationdependent inhibition of phosphoinositide turnover induced by activating receptors on NK cells (51). This could indicate that PKA, directly or indirectly, may act at the level of phospholipase C-␥1. The ability of PKA to phosphorylate phospholipase C-␥1 and modulate tyrosine phosphorylation-dependent activation of phospholipase C-␥1 in vitro is compatible with this notion (52). Our results, showing that Ca 2ϩ and PKC activation blocks PKA-mediated inhibition of natural killing, confirm that PKC and calcium mobilization are important in maintaining the chain of events leading to cytolytic activation of NK cells. Furthermore, the fact that combined treatment with TPA and ionomycin makes NK cells refractory to the inhibitory effects of cAMP suggests a site of action for PKA proximal to PKC activation and suggests that PKA-dependent phosphorylation of proteins in close proximity to the surface membrane receptors may be involved. That PKC activation by TPA and ionomycin also allows killing of syngeneic target irrespective of PKA activation supports this assumption.
In conclusion, we report that PKA type I negatively regulates both MHC class I-dependent and -independent NK cytotoxicity. This serves to establish an inhibitory signaling pathway whereby cAMP through PKA type I can regulate cytotoxicity initiated through killer activatory receptors. The characterization of receptor molecules expressed by alloreactive rat NK cells will hopefully aid in clarifying a possible connection between receptors on NK cells, their downstream signaling pathways, and the molecular targets for PKA type I.