Interactive effects of Na and K in killing by human natural killer cells

Experimental

Interactive

Cell Research 184 (1989) 99-108

Effects of Na and K in Killing Natural Killer Cells

L. C. SCHLICHTER’ Department

of Physiology,

University

by Human

and 1. C. MacCOUBREY of Toronto,

Toronto,

Ontario,

Canada M5.S lA8

Contact-mediated lysis by human natural killer cells is inhibited by a number of drugs that block the predominant K channel. In this study we have further examined the role of the K channel and the interactions between passive K and Na transport in killing. Low external Na-inhibited killing and inhibition were not due to reduced inward current through the Na channels in the target cell. A role for the Na/H antiport is suggested since amiloride inhibited killing in a dose-dependent manner that was competitive with external Na. Depolarizing the killer cell with elevated external K did not inhibit killing. On the contrary, high K0 reduced the inhibition caused by low Na, and by the K-channel blockers quinidine, verapamil, and retinoic acid. Hyperpolarizing the killer cell with low K0 or valinomycin inhibited killing. Valinomycin, which should prevent the depolarization caused by Kchannel block, did not reverse the effect of the blockers quinidine, verapamil, and 4aminopyridine. Hence, the primary role of the K channels during killing is not to maintain the negative membrane potential. On the contrary, depolarization may promote killing under conditions where killing is submaximal. 0 1989 Academic press. IK.

One of the earliest events following the interaction of cytotoxic T lymphocytes (CTL) with susceptible target cells is an increase in K+ efflux (86Rb) from the CTL, which has been postulated as a signal in the lytic sequence [I]. Using the patch-clamp electrophysiological technique, several laboratories have identified a predominant voltage-dependent K channel in different lymphocytes, including CTL, T lymphocytes, and natural killer (NK) cells [2-61. For mouse CTL the K+ current increased after conjugation to target cells under conditions (37°C external Ca”) that are necessary for the lethal hit [4]. Stronger support for a role for K channels in killing comes from our previous work on human NK cells [2, 3, 71. The Ca-dependent phase of killing includes exocytosis of cytotoxic granules that presumably contain both natural killer cell cytotoxic factors (NKCF) [8] and the pore-forming molecules (perforin or cytolysin) that promote target cell lysis [9, 101.During the Ca-dependent, programming-for-lysis phase, killing is inhibited by blocking the K channels with any one of several channel-blocking drugs that have distinctly different mechanisms of block [2, 71. These drugs inhibit release of NKCF but the action of previously secreted NKCF is not affected, suggesting that exocytosis is impaired and that the target cell is not affected by blockers. Further evidence that the K channels in the NK cell (not the target cell) are the essential ones comes from our previous experiments [7]. Two of the drugs tested ’ To whom correspondence and reprint requests should be addressed at Department of Pharmacology Merck Frosst Centre for Therapeutic Research, P.O. Box 1005, Pointe Claire-Dorval, Quebec, Canada H9R 4P8. 99

Copyright @ 1989 by Academic Press, Inc. All rights of reproduction in any form reserved 00 14-4827/89 $03.00

100 Schlichter and MacCoubrey (quinidine, verapamil) enter the cell, and then several hours of washing are required to reverse the block. When either target cells or NK cells were preincubated with a blocker, washed, and then combined in a killing assay, inhibition occurred only if the NK cells had been exposed to the drug. K channels in a wide variety of cells establish and maintain a negative resting potential. The lymphocyte resting potential is -50 to -70 mV and raising the external K concentration (K,J depolarizes the membrane, as predicted for a Nernst potential established by the K gradient across the membrane [II-161. Concordantly, blocking the channels in lymphocytes causes the cells to depolarize [15, 171. This, in turn, could affect other ion-transport systems, including channels, to alter ion fluxes and intracellular ion content. There is evidence that both K+ and Na+ fluxes are involved in lymphocyte functions including T cell activation and cell-mediated cytotoxicity. Na+ and K+ fluxes increase following mitogenic stimulation of T cells [ 181,accelerated K+ efflux from CTL is associated with delivery of the lethal hit [4], and low N% inhibits killing by mouse CTL [19] and the DNA synthesis in T cells that is stimulated by mitogens [20]. In both T cells and CTL, raising the K0 somewhat relieves the effects of low N%. In the present study we have further investigated the role of K and Na in killing by human NK cells, including effects of depolarizing or hyperpolarizing the NK cell in the presence or absence of K-channel blockers. To alter the membrane potential we have varied K0 or added the K+ ionophore valinomycin to the killing assay medium. Valinomycin is commonly used to hyperpolarize cells toward the Nernst potential for K and even low concentrations of the ionophore hyperpolarize lymphocytes to about -90 mV [12, 161.We found that killing by NK cells is inhibited by low N% and we have asked whether this Na dependence is due to Na channels, to an effect on membrane potential, or to Na/H exchange. We find that (1) low NQ inhibits killing and that a low-affinity Na/H exchanger, blockable by amiloride, may be important for NK cell function; (2) Na+ fluxes through classical voltage-gated Na channels such as those in the target cell are not essential; (3) depolarizing NK cells with high K0 does not itself inhibit killing, but does relieve the inhibition caused by low Na and by K-channel blockers; and (4) using valinomycin to hyperpolarize the cells inhibits killing and fails to reverse the effects of the K-channel blockers. MATERIALS

AND METHODS

Chemicals were purchased from Sigma (St. Louis, MO) and stock solutions were made as follows: quinidine, 4-aminopyridine (4-AP), verapamil, amiloride, and tetrodotoxin (TTX) in distilled water; valinomycin and all-frans-retinoic acid (RA) in DMSO. Control experiments showed that the final concentrations of DMSO in the experimental solutions (SO.1 %) had no effect on killing. The control assay medium was a mammalian Ringer’s solution containing (in rruV): 150 NaCl, 5 KCl, 2.5 CaCl*, 1 MgC12, 10 d-glucose, 10 Hepes, adjusted to pH 7.4 with NaOH. Bovine serum albumin (2.5 mg/ml) was added to ail assay solutions. Unless otherwise indicated, Na and K substitutions were made with equimolar n-methylglucamine chloride (nMG) so that Na and K could be varied independently. The Cl concentration was constant in all experiments. For Low-Na solutions KOH was used to adjust the pH. Cytotoxiciry assay. Peripheral blood mononuclear cells were prepared from heparinized normal human blood by density-gradient centrifugation on Ficoll-Hypaque (Pharmacia). NK cytotoxicity Reagenfs

and solutions.

101

NuO and KO in natural killing 140 130 120 110 100 F ‘Z 9 K

90 80 70

.-F ii

60

&

50 40 30 20 10 0

t 0

40

20

60

80

100

No concentration 0

control

+

120

140

160

(mM) omiloride

Fig. 1. Low external Na and amiloride inhibit killing. Killing of K562 target cells by human NK cells was determined in a standard “Cr-release assay in Ringer’s solution in which Na was replaced with n-methylglucamine (nMG, see Methods). Killing at each Na concentration is plotted as percentage of killing in normal Na, (150 n&I). Values represent the mean f SD of the number of experiments indicated near each point. (0) Control Na solutions; (+) Na-substituted solutions in the presence of 150 or 300 t.rJ4amiloride.

was assessed against the myeloid leukemia cell line K562 using a standard 4-h “Cr-release assay [ 1, 2,7, 81. Under these conditions, spontaneous killing of tumor cells can be attributed to NK cells since CTL require prior sensitization to their targets. Percentage specific lysis was defined as the percentage specific j’Cr released, or [(experimental release-spontaneous release)/(maximal release-spontaneous release)] x 100. Spontaneous “Cr release from isolated target cells was determined and maximal release was measured after the targets were incubated with 5 % (v/v) Triton X-100 detergent. Results are presented as mean +S.D. (n) in the text. Statistical comparisons used Student’s t test. Electrical recording. For electrical recording [2, 3, 71, a patch electrode of 3- to 8-MR resistance was pushed lightly against a cell and after a high-resistance seal (>lO GQ) formed, a small negative pressure was applied to the pipet interior to establish a whole-cell recording. Linear components of capacitive and leakage were analog-subtracted and the current was low-pass filtered at 3 kHz. Current and voltage traces were stored on floppy disks using a Nicolet storage oscilloscope and were later printed with an X-Y plotter. Recordings were always begun while cells were being continuously perfused with the control mammalian Ringer’s solution described above. The patch electrode contained (in m&f): 20 NaCl, 1 MgC&, 100 potassium aspartate, 15 d-glucose, 10 K Hepes, 5 K*EGTA, 4 MgATP, 0.1 CAMP, and 2 theophylline at pH 7.4. All electrical recordings were made at 2&22”C.

RESULTS Effect of Na on Killing Figure 1 shows the effects of replacing external Na with the bulky cation, nMG. Because the proportion of NK cells and their efficacy vary greatly between donors, the percentage of K562 cells killed at an effector : target (E : T) ratio of

102 Schlichter and MacCoubrey 40: 1 varied from 30 to 60%. At lower E : T ratios the percentage killing was reduced. Therefore, to combine data from different donors, results were first standardized by expressing killing as a percentage of the value of the control solution (150 mM Na) for each donor, and then averaging the results at each Na concentration. The upper curve shows that killing was maximal at 90-105 mM Na and declined both below 90 mM and at the highest Na concentration (150 mM). Compared with the maximal killing activity, killing was significantly inhibited (JYO.01, Student’s t test) at Na 5 to 75 mM and 150 mM. (The data for amiloride are described later.) These results suggest that Na+ influx plays a role in killing. Since voltage-dependent Na channels are present in the KS62 target cells used in this study [2], we tested whether ionic current through these channels was essential. Figure 2 shows that nMG did not carry current through the Na channel and the specific neurotoxin, tetrodotoxin (TTX), blocked these channels. Membrane currents were recorded from a K562 target cell using the whole-cell configuration of the patch-clamp technique. Fast-inactivating inward currents were evoked at voltages more positive than -50 mV. This current was carried by Na+ influx since (1) it was abolished when external Na+ was replaced with nMG (Fig. 2B) or with Ca*+ (not shown); (2) the current reversed direction to outward when the membrane potential was made more positive than the Nernst potential for Na+ (+44 mV in this example, Fig. 20); and (3) the Na+ current was abolished by tetrodotoxin (Fig. 2 C); at 6x 1O-7M TTX, more than 90% of the current was blocked. Nevertheless, TTX (3 X 10e9to 3 x low6 M) had nonsignificant effects (P>O.lO) on killing by NK cells in the standard “Cr-release assay. Relative percentage killing ranged from 90.8&5.6% (n=4) at 3~ IO-” M TTX to 100.4t5.4% (n=5) at 3x 10m6M TTX compared with killing in normal solution without TTX. Therefore, the dramatic inhibition of killing in low-Na Ringer’s was not related to Na+ influx through the voltage-gated Na channels in the target cells. A second route of Na entry into lymphocytes is via the Na/H exchanger [21-261 that is inhibited by the family of K-sparing diuretics typitied by amiloride. Figure 1 shows that amiloride (150-300 CLM) significantly inhibited killing (P
Nao and K. in natural killing

10

103

c--L

300

D 200

100

0

-100

-200

-300 -100

l---A -60

-60

-40

-20 Membrane

0

20

potential

(mV)

40

60

60

100

Fig. 2. Na+ currents recorded from a K562 target cell in the whole-cell, patch-clamp configuration. Between test pulses, the cell was held at -90 mV and then stepped for 30 ms to each test potential (first 16 ms shown). For clarity, not all currents are shown. Zero-current levels are shown by the dash before and after each set of currents. Vertical scale bars. pA; horizontal scales, 2 ms (A) Na+ currents evoked by steps from -50 to + 10 mV (IO-mV increments) in normal Ringer’s solution. (E) Currents at -50 to +lO mV in nMG solution (150 mM) substituted for Na,,. (C) Currents at -50 to 0 mV in the presence of 0.6 @4 tetrodotoxin (TTX). Current-versus-voltage relations for the Na+ current in the absence (0) or presence (+) of 0.6 mM TTX or in nMG solution (0).

104 Schlichter and MacCoubrey 60

50

40

30

20

I

10

0

y///I 0

I3

I 100

Na 15

+

I

I

200 Amiloride Nc 45

concentration 0

(FM) Na 75

I

I 300

A

I 400

1

No 105

Fig. 3. Dose dependence of inhibition of killing by amiloride. Amiloride was added to the following Na solutions: 15 mM (O), 45 nuI4 (+), 75 mM (O), and 105mM (A). For different Na, concentrations, nMG was substituted for Na. Percentage inhibition was calculated by comparison with killing in the absence of amiloride at the same Na concentration (see text). Values represent the mean of three experiments.

at low external Na concentrations, a result that is consistent with competition between Na and amiloride for an external site. At the highest Na concentration tested (105 mM), inhibition was increased relative to Na 75, as though a second component of amiloride sensitivity existed; however, we have not investigated this point further. Effect of Membrane

Potential

A further possibility is that low external Na (N%) inhibited killing by preventing a membrane depolarization sometime during killing, a depolarization that involved TTX-insensitive, Na-permeable channels. Several types of TTX-insensitive Na channels have been described in other cell types, including a noninactivating, voltage-dependent channel [27] and voltage-independent Na channels in various cells [28, 291. To test whether low N% inhibited killing by preventing depolarization we tried to reverse the effect of low NQ by depolarizing the cells with high Ko. Figure 4 shows that raising K,, progressively increased the killing when NQ was held constant and low (5 or 30 mM). The 100% value was percentage killing in Na 5, K 5. (Low-Na inhibition was evident in these experiments since raising N% from 5 to 1.50mM at 5 mM K increased killing more than twofold.) Significant relief of low-Na inhibition was seen at K. concentrations from 30 to 150 mM (P
Nao and K0 in natural killing

105

160

T--T-~-.-7--.r-0

20

r --7--.~T--40

0

60 K concentration Na 5 mM

60

T-.

7-

100

-

T1--T-~-‘T~ 120

-I 140

160

(mt.4) + Na 30 mM

Fig. 4. High K,, relieves low-Na inhibition. Relative killing at 5 mA4 Nq (0) and 30 mA4 N% (+) is plotted at each K,, concentration as a percentage of the value at Na 5 mM, K 5 n~44.Solutions were Ringer’s in which either Na or K was substituted with nMG. Values are the mean f SD of the number of experiments indicated near each point.

are permeable mainly to K. The highest killing attained (in Na 5, K 120) was about 87% of the control levels (in Na 150, K 5); therefore, reversal of low-Na inhibition was nearly complete. Raising K0 markedly relieved low-Na inhibition in 13/14 experiments. These data suggest that K0 itself or depolarization by high K0 can replace the need for Na+ To further test the effects of membrane potential we increased K0 in the presence of an optimal NQ concentration (90 mM). The average killing was 51*5% (n=6) at the normal K0 concentration of 5 mM and 5326% (n=6) at 60 rnM K, a significant difference (P>O. 10). Therefore, depolarization itself neither inhibited nor enhanced killing. Conversely, to test whether hyperpolarization affects killing we reduced K0 in one series of experiments and added the K ionophore, valinomycin, in another series of experiments. Compared with normal Ringer’s solution (Na 150, KS), complete removal of K0 reduced the percentage killing from 37.2*5.0% (n=4) to 18.9f3.8% (n=4), a highly significant difference (PO.lO). Killing was significantly (PcO.01) inhibited by 0.5 uM valinomycin whether it was added at the beginning of the assay (inhibition, 3OC4%, n= 15) or only during the Ca-dependent, programming-for-lysis stage (inhibition, 34f5 %, n=lO). These results show that hyperpolarization inhibits killing but depolarization does not. Because high K,, relieved the low-Na inhibition we then asked whether depolar-

106 Schlichter and MacCoubrey ization with high K0 could reverse the inhibition of killing produced by K-channel blockers. In the presence of quinidine (100 rnM) killing was inhibited by 72.9*9.7% (n=6) in 150 mM NaCl solution and 59.4*9.6% (n=6) in KC1 solution, a significant relief of inhibition (P
Nao and K0 in natural killing

107

concentrations and low Na,, inhibited killing by only 45-65%. Hence, there is a significant component of killing that is independent of N% and Na/H exchange. Electroneutral Na/H exchange has been observed in most types of lymphocytes and lymphoid cell lines examined. The activity of the Na/H exchanger increases following mitogenic stimulation and it becomes the main route of Na+ entry [30-321. Other roles proposed for the Na/H exchanger are in the regulation of cytoplasmic pH [23] and in cell-volume regulation [21, 221. We observed that the low-Na inhibition of killing was almost entirely reversed by increasing K0 to high concentrations. Similarly, the inhibition of CTL-mediated cytotoxicity by low Na was relieved by raising K,, [ 191, as was low-Na inhibition of mitogen-stimulated DNA synthesis [14]. It is not known whether the K+ ion itself or the resultant membrane depolarization produced this relief. However, in our experiments, this enhancement appeared only at suboptimal NQ concentrations (<75 mM). In contrast to the effects of high K, hyperpolarization produced by low K0 or by the K ionophore valinomycin inhibited NK cytotoxicity (present study), killing by CTL [ 16, 191, and proliferation of lymphocytes [33]. The inhibition we observed occurred mainly during the Ca-dependent, programming-for-lysis phase. Recent results suggest a possible mechanism for this inhibition by hyperpolarization. Gray and his co-workers [16] used photometric measurements of membrane potential and cytoplasmic free Ca2+ (CaJ to show that Ca increased in the CTL after exposure to a specific antigen to which it was presensitized. The Ca rise required external Ca and was attributed to Ca*+ influx through channels that are somewhat voltage sensitive. Hence, valinomycin hyperpolarized the CTL and inhibited the rise in Ca. A similar rise in Cq was observed following CTL conjugation to target cells [34] and it is likely to occur in NK cells as well. By inhibiting the Ca*+ rise, hyperpolarization might prevent the exocytosis of cytolytic granules from CTL and NK cells, thereby inhibiting killing during the Ca-dependent, programming-for-lysis phase. Our results show that inhibition of NK cytotoxicity by the K-channel blockers quinidine, verapamil, and retinoic acid [2, 3, 71 was not caused by the membrane depolarization the blockers produced. High KO, which depolarized the cells, did not itself inhibit killing. Rather, under conditions of low Na,,, high K,, enhanced killing. Moreover, providing an alternate pathway for K+ flux (valinomycin), which will reduce or abolish the depolarization caused by K-channel block, did not reverse the effect of the blockers. On the other hand, depolarization by high & partially relieved the inhibition by K-channel blockers. This relief cannot be explained simply as a depolarization-dependent relief of channel block since we have shown that these drugs (quinidine, verapamil, retinoic acid) block the K channel in a use-dependent manner that increases as the membrane is depolarized [2, 3, 71. It appears that high K0 can partly bypass the need for the K channels; however, the mechanism is not known and we are now trying to further elucidate how high & exerts its effects. In summary, we have shown that: (1) killing by human NK cells is inhibited by reducing Na,, and this effect may be related to a low-affinity Na/H exchanger that is blockable by amiloride; (2) depolarization by high K0 does not itself inhibit

108 Schlichter and MacCoubrey killing; therefore, the primary result of K-channel block is not due to depolarization of the cell; (3) depolarization or high K0 reverses the inhibition by low Na,, and partly relieves the block by quinidine, verapamil, and retinoic acid; whereas (4) hyperpolarization with valinomycin inhibits killing and does not reverse the effects of K-channel blockers. We are grateful for the excellent technical assistance provided by L. Karrass and for discussions with Dr. N. Sidell. This work was supported by grants from the Medical Research Council of Canada (MRC), the National Cancer Institute of Canada, the Bickell Foundation, and the University of Toronto. L. C. Schlichter is a Scholar of the Medical Research Council of Canada.

REFERENCES 1. 2. 3. 4. 5.

Russell, J. H., and Dobos, C. B. (1983) J. Zmmunol. 131, 1138-1141. Schlichter, L. C., Sidell, N., and Hagiwara, S. (1986) Proc. Nut/. Acad. Sci. USA 83, 451-455. Sidell, N., and Schlichter, L. (1986) Biochem. Biophys. Res. Commun. 138, 560-567. Fukushima, Y., Hagiwara, S., and Henkart, M. (1984) J. Physiol. 351, 645-656. DeCoursey, T. E., Chandy, K. G., Gupta, S., and Cahalan, M. D. (1984) Nature (London) 307, 465-468. 6. Matteson, D. R., and Deutsch, C. (1984) Nature (London) 307, 486-471. 7. Sidell, N., Schlichter, L. C., Wright, S. C., Hagiwara, S., and Golub, S. H. (1986) J. Zmmunol. 137, 1650-1658.

8. Wright, S. C., and Bonavida, B. (1982) .Z. Zmmunol. 129, 433-439. 9. Henkart, P. A. (1985) Annu. Rev. Zmmunol. 3, 31-58. 10. Podack, E. R. (1985) Zmmunol. Today 6, 21-27. 11. Deutsch, C. J., Holian, A., Holian, S. K., Daniele, R. P., and Wilson, D. F. (1979) J. Cell Physiol. 99, 7sC94.

12. Rink, T. J., Montecucco, C., Hesketh, T. R., and Tsien, R. Y. (1980) Biochim Biophys. Acta 595, 15-30. 13. Montecucco, C., Rink, T. J., Pozzan, T., and Metcalf, J. C. (1980) Biochim. Biophys. Acta 595, 65-70. 14. Deutsch, C., and Price, M. (1982) J. Cell Physiol. 113, 73-79. 15. Wilson, H. A., and Chused, T. M. (1985) J. Cell Physiol. 125, 72-81. 16. Gray, L. S., Gnarra, J. R., Russell, J. H., and Engelhard,V. H. (1987) Cell 50, 119-127. 17. Tsien, R. Y., Pozzan T., and Rink, T. J. (1982) Nature (London) 295, 68-71. 18. Segel, G. B., Simon, W., and Lichtman, M. A. (1979) J. C/in. Invest. 64, 834-841. 19. Russell, J. H., and Dobos, C. B. (1981) Eur. J. Zmmunol. 11, 840-843. 20. Deutsch, C., Price, M. A., and Johansson, C. (1981) Exp. Cell Res. 136, 359-369. 21. Grinstein, S., Clarke, C. A., and Rothstein, A. (1983) J. Gen. Physiol. 82, 619-638. 22. Grinstein, S., Cohen, S., and Rothstein, A., (1984) .Z. Gen. Physiol. 83,341-369. 23. Grinstein, S., Goetz, J. D. Furuya, W., Rothstein, A., and Gelfand, E. W. (1984) Amer. J. Physiol. 247, C293-C298.

24. Mills, G. B., Cheung, R. K., Cragoe, E. J. Jr., Grinstein, S., and Gelfand, E. W. (1986) J. Zmmunol. 136, 1150-l 154. 25. Rosoff, P. M., Stein, L. F., and Cantley, L. C. (1984) J. Biol. Chem. 259, 7056-7060. 26. Calalb, M. B., Stanton, T. H., Smith, L., Cragoe, E. J. Jr., and Bomsztyk, K. (1987) .Z. Biol. 27. 28. 29. 30. 31. 32. 33. 34.

Chem. 262, 3680-3684. Schlichter, L. (1983) Deu. Biol. 98, 47-59.

Yau, K.-W., McNaughton, P. A., and Hodgkin, A. L. (1981) Nature (London) 292, 502-505. Maruyama, Y., and Petersen, 0. H. (1982) Nature (London) 299, 159-161. Grinstein, S., Cohen, S., Goetz, J. D., Rothstein, A., and Gelfand, E. W. (1985) Proc. Nat/. Acad. Sci. USA 82, 1429-1433. Prasad, K. V. S., Severini, A., and Kaplan, J. G. (1987) Arch. Biochem. Biophys. 252, 515-525. Rosoff, P. M., Stein, L. F., and Cantley, L. C. (1984) J. Biol. Chem. 259, 7056-7060. Daniele, R. P., and Holian, S. K. (1976) Proc. Nutl. Acad. Sci. USA 73, 3599-3602. Poenie, M., Tsien, R. Y., and Schmitt-Verhulst, A.-M. (1987) EMBO J. 6, 2223-2232.

Received November 24, 1988 Revised version received April 4, 1989 F’nnted

in Sweden