Correlation between Plasma Membrane Potential and Second Messenger Generation in the Promyelocytic Cell Line HL-60”

The effects of plasma membrane depolarization on cytosolic free calcium ([Ca”+]i) and inositol 1,4,5-tris- phosphate (Ins( 1 ,4,5)P3) generation were investigated in the human promyelocytic cell line HL-60 differen- tiated with either dimethyl sulfoxide or retinoic acid into neutrophil-like cells. Increases in [Ca’+]i and ac- cumulation of Ins(1,4,5)Ps were triggered by two chemoattractants fMet-Leu-Phe and leukotriene Bq. Plasma membrane potential was depolarized by isoosmotic substitution of NaCl with KCl, by the pore-form- ing ionophore gramicidin D, or by long term treatment with ouabain. Both Ca2+ mobilization from intracellu- lar stores and Ca2+ influx across the plasma membrane were reduced by prior depolarization of plasma membrane potential regardless of the procedure employed to collapse it. Agonist-induced generation of Ins(l,l,B)- Pa was also reduced in parallel in pre-depolarized HL-60 cells. The present findings provide further evidence sug- gesting that plasma membrane potential can be an important modulator of agonist-activated second messenger generation in myelocytic cells.

of NaCl with KCl, by the pore-forming ionophore gramicidin D, or by long term treatment with ouabain.
Both Ca2+ mobilization from intracellular stores and Ca2+ influx across the plasma membrane were reduced by prior depolarization of plasma membrane potential regardless of the procedure employed to collapse it. Agonist-induced generation of Ins(l,l,B)-Pa was also reduced in parallel in pre-depolarized HL-60 cells.
The present findings provide further evidence suggesting that plasma membrane potential can be an important modulator of agonist-activated second messenger generation in myelocytic cells.
Receptor binding on the plasma membrane of several nonexcitable cells, among which are neutrophils, causes changes in the plasma membrane potential (l-5). The physiological role of these plasma membrane potential variations during cell activation has remained dubious however. Recently, evidence has accrued that plasma membrane potential in nonexcitable cells might have a regulatory role in the homeostasis of [Ca"]i,' as it has been shown by several laboratories that membrane depolarization reduces agonist-induced Ca*+ inflow across the plasma membrane whereas hyperpolarization has the opposite effect (6-8). The inhibition of Ca*+ influx could be due, at least in part, to a decreased Ca2+ electrochemical gradient, but it cannot be excluded that the membrane pathways for Ca2+ influx in nonexcitable cells are "voltage-modulated," although not "voltage-gated" (8). In a previous study, we showed that, in human neutrophils, prior depolarization of plasma membrane potential reduced not only Ca2+ influx but also Ca2+ release from intracellular stores activated by the chemotactic peptide fMet-Leu-Phe (6). Conversely, incubation of the neutrophils with the ionophore valinomycin, that hyperpolarizes the plasma membrane potential, potentiated both Ca*+ release and Ca2+ influx triggered by this same agonist (6). In the same study, we postulated that reduced mobilization of Ca*+ from the intracellular stores in depolarized neutrophils might depend on a decreased agonist-triggered generation of Ins(1,4,5)P3.
In the present investigation, we have directly tested this hypothesis by assessing the role of plasma membrane depolarization on [Ca*+]i homeostasis and agonist-dependent Ins(1,4,5)Pz formation in the human promyelocytic cell line HL-60 differentiated either with DMSO or retinoic acid. Two stimuli (fMet-Leu-Phe and LTBJ, acting on different receptors, were applied, and three different procedures (incubation in high KC1 buffer, gramicidin D, or ouahain) were used to depolarize the plasma membrane potential in intact cells.

MATERIALS AND METHODS
Reagents and their sources were as follows: N-formyl-L-methionyl-L-leucyl-L-phenylalanine (fMet-Leu-Phe), N-t-BOC-L-methionyl-Lleucyl-L-phenylalanine ( Phe, but much less LTB( receptor-mediated responses, while the contrary is true for retinoic acid-differentiated HL-60 cells. Inositol phospholipids and inositol phosphates were labeled as previously described (12). the cells being incubated from day 5 to day 7 in RPM1 incubation medium; (ii) by treatment with the pore-forming ionophore gramicidin D; or (iii) by incubation with the inhibitor of the Na+/K+ ATPase ouabain. In order to study the effect of depolarization on Ca2+ redistribution from intracellular stores without interference from Ca*+ influx, the experiments of Fig. 1 to Fig. 7 were performed in Ca'+-free media supplemented with 1 mM EGTA. medium containing myo-[2-"Hlinositol (1.5 &i/ml).

Measurements of Inositol
Phosahates-Cells (35 X 106/0.5 ml) were warmed for 5 min at 37 "C before stimulation, and incubations were terminated by the addition of ice-cold 10% (v/v) trichloroacetic acid. Samples were kept on ice 5-10 min and centrifuged (800 X g X 10 min); supernatant was washed three times with a 5-fold excess of diethyl ether. The ether-washed and neutralized extracts were analyzed by HPLC, essentially as previously described (10). [3H] Ins(1,4,5)P3 serving as position and recovery standard was purchased from Amersham International, United Kingdom. In some experiments, inositol phosphates were separated by stepwise elution from small Dowex (anion exchange) columns as previously described (10, 11). Measurement of Cytosolic Free Calcium and Ca" Influx Determinotion-Fura-loading was performed at 37 "C as described previously (12), at a final fura-2/AM concentration of 2 PM. After the loading procedure, the cells were kept at room temperature. Just before use, a sample of the cell suspension was centrifuged to remove BSA and resuspended in the indicated medium. Fluorescent measurements were performed with a Perkin-Elmer fluorimeter (LS 3, Perkin-Elmer Cetus Instruments, Norwalk, CT) as described (12). Excitation and emission wavelengths for fura-measurements were 340 and 505 nm, respectively. The values of fura-fluorescence were not expressed as absolute [Ca*+]i values because the exact number of differentiated (responsive) cells may vary in different cell preparations (11).
Determination of Ca*+ influx in fura-2-loaded cells was assessed as previously described (12,13) using the quenching properties of Mnzf on fura-fluorescence, assessed at 360 nm excitation wavelength (isosbestic point). Ca*+ influx rates were deduced from the initial slope produced upon Mn" addition (100 PM) to f'Met-Leu-Phestimulated cells and expressed as A of total fura-signal (in %/min), as described (12).

Membrane Potential
Measurements-Membrane potential changes were measured with the lipophilic fluorescent dye bisoxonol as previously described for human neutrophils (6). DMSO-or retinoic aciddifferentiated HL-60 cells and dye concentration in these experiments were 4 X lO'/ml and 100 nM, respectively.
Ouabain Treatment-The cells (2 x 107/ml) were preincubated for 5 min at 37 "C in Ca*+ medium containing BSA (O.l%), before the addition of ouabain (100 PM). The cell suspension was gently shaken every 10 to 15 min to prevent sedimentation and clumping and kept at 37 "C throughout the incubation period. Under these conditions after 8 h in the presence of ouabain, trypan blue exclusion reached 6 it: 2% (mean -C SD., n = 4 in two separate experiments); i.e. very similar to that of controls also kept at 37 "C for the same period in the absence of ouabain. Intracellular Ca*+ stores (assessed by the addition of ionomycin to cells incubated in Ca*+-free medium) were slightly depleted after 8 b in Ca*+ medium both in the presence (8.5 1?: 3%) and in the absence (8 ? 2%) of ouabain (mean k S.D., n = 5 in two separate experiments).
Just before use, a sample of the cell suspension was centrifuged to remove BSA and resuspended in fresh medium still containing ouabain. Fura-loading was performed during the last hour preceding the use of the cells and did not interfere with the ouabain treatment (as assessed by membrane potential changes). In control experiments, ouabain pretreatment did not alter the fura-loading, as judged by comparing the intracellular fura-concentrations in ouabain-treated and untreated cells (14).
Presentation of Data-Unless otherwise indicated, typical experiments are shown which are representative of at least four similar experiments.

Depolarization of Plasma Membrane Potential Decreases
[Ca2+]i Release from Intracellular Stores-Ca*+ release from intracellular stores induced by fMet-Leu-Phe and LTB, was measured in HL-60 cells differentiated with DMSO (to induce expression of f&let-Leu-Phe receptors) or retinoic acid (to induce expression of LTB, receptors) (11). Plasma membrane potential was depolarized according to three different procedures: (9 by isoosmotic substitution of NaCl with KC1 in the Following a 5-min preincubation at 37 "C in Ca*+-free, EGTA-supplemented NaCl (traces a, b, d, e, g, h, j, and k) or KC1 (140 mM) medium (traces c, f, i, and 1). fura-a-loaded cells were stimulated by various concentrations of fMLP or LTB,. In traces b, e, h, and k, gramicidin D (1 1~) was added 3 min before agonist stimulation.
at the excitation wavelength of 360 nm (isosbestic point), indicating an effect on fura-fluorescence independent of [Ca'+]i (data not shown).
At lo-' M fMet-Leu-Phe, the inhibition by gramicidin D was about lOO%, while that observed in 140 mM KC1 ranged between 60 and 80%. However, while the inhibition by gramicidin D was relieved at high agonist concentrations (compare trace h with g and k with j), that by high KC1 persisted both for fMet-Leu-Phe and LTB, (compare trace i with g and 1 with j). Fig. 2 shows however that, quite unexpectedly, Ca*' mobilization by the Ca2+ ionophore ionomycin as well was reduced in cells incubated in high KC1 media (Fig. 2, compare b with c) but unaffected by gramicidin D pretreatment (Fig.  2, compare a with c). Given that ionomycin-induced Ca*' redistribution should bypass any receptor-linked step, the simplest explanation is that incubation in high KC1 reduces the total Ca*+ content of intracellular stores. In control experiments, the ionomycin-sensitive Ca*' stores were shown to be proportionally reduced by 15 f 4, 26 + 7, and 53 + 5% (mean k SD., n = 3-5 in three separate experiments) when DMSO-differentiated HL-60 cells were preincubated for 5  Values are mean f S.D., n = two to five determinations in two separate experiments and expressed as percent inhibition of the maximal increase in [Ca'+], triggered by the addition of 10-s M fMLP (84 + 2% of fura-2 saturation). min in the presence of 50,100, and 140 mM KCl, respectively. Fig. 2 also shows that, in agreement with previous reports (15,16), high KC1 medium sensitizes the cells to low doses of fMet-Leu-Phe. Indeed, HL-60 cells depolarized with 140 mM KC1 in the presence of gramicidin D still weakly responded with a rise in [Ca'+]i to lo-' M fMet-Leu-Phe (Fig. 2e), while an aliquot of the same batch of cells depolarized with 100 nM gramicidin D in Na' medium failed to respond to the same concentration of the chemotactic peptide (Fig. 2d). The residual Ca2+ mobilization by fMet-Leu-Phe in KC1 + gramicidin D-treated cells was not due to incomplete depolarization since membrane potential was not appreciably different in the two experimental conditions (not shown). These observations indicate that incubation in high KC1 media has effects on [Ca*+]i homeostasis in HL-60 cells unrelated to depolarization of plasma membrane potential and make ionic substitution unsuited for the study of the effect of depolarization on receptor-triggered [Ca"]i changes.
Although gramicidin D had neither of the undesirable effects of KCl, it cannot be excluded that this ionophore too has effects on Ca2+ homeostasis unrelated to changes in membrane potential. Therefore, we examined the relationship DMSO-differentiated HL-60 cells loaded with fura-(2 PM) were sequentially stimulated by low9 and lob6 M WLP. Prior to fMLP stimulation, cells were pretreated as follows: a, Me$O for 3 min; b, ouabain (100 wM) for 3 min; c and d, ouabain (100 pM) for 4 h. Ionomycin (500 nM) was added in c. Inset, bisoxonol fluorescence changes; open arrow refers to the addition of the cells (4 X lO'/ml) to bisoxonol (100 nM) containing Ca*+-free medium; cells were pretreated with ouabain (100 NM) for 3 min (a) or for 4 h (b). Gramicidin D (1 PM) was added when indicated. The traces shown were obtained the same day using the same batch of fura-e-loaded cells previously or acutely treated with ouabain as described under "Materials and Methods"; similar data were obtained in two additional experiments.
between the concentrations of gramicidin D needed to depolarize the plasma membrane potential and those which inhibited the [Ca2+]i rise. As shown in Fig. 3A, gramicidin D fully depolarized HL-60 cells at a concentration of 250-500 nM. Increasing the gramicidin D concentration up to 5 FM had no further effect on membrane potential. Fig. 38 shows that inhibition of the [Ca2+li rise induced by fMet-Leu-Phe was also maximal at about 250 nM gramicidin D. In agreement with previous observations, the inhibition was progressively smaller at higher fMet-Leu-Phe concentrations (6). The EDho for gramicidin D inhibition of the [Ca'+]i rise was however independent of the met-Leu-Phe concentrations. Similar results were observed with LTB, (not shown).
HL-60 cells possess a Na+/K' pump similar to that found in other cell types (17). Binding sites for ouabain, a powerful blocker of the Na'/K+ ATPase, have been demonstrated to be present in both DMSO-and retinoic acid-differentiated HL-60 (18,19). Cells were thus depolarized using ouabain treatment. Ouabain had no short term effect on either plasma membrane potential (Fig. 4, inset a) or fMet-Leu-Phe (Fig. 4, compare a and b) and ionomycin (not shown)-induced Ca2+ redistribution.
On the contrary, as shown in Fig. 4d Plasma membrane depolarization (0) and [Ca'+], changes (A) were assessed following various times of incubation in the presence of ouabain (100 PM). Plasma membrane potential values are expressed as percentage of maximal depolarization induced by gramicidin D (1 PM) assessed in parallel in cells not previously treated with ouabain. [Ca'+], changes are expressed in percent of maximal fura-fluorescence saturation following the addition of 10e9 M fMet-Leu-Phe (72 + 2% at, time 0). Data shown were obtained the same day using the same batch of fura-2-loaded cells kept at 37 "C and continuously or not exposed to ouabain..Similar data were obtained in two different experimental days.
(mean + SD., n = 5 in three separate experiments) reduction of the [Ca2']i rise induced by lo-' M fMet-Leu-Phe, whereas the [Ca"']i rise caused by ionomycin or 10m6 M fMet-Leu-Phe was unaffected (Fig. 4, c and d). The inset of Fig. 4 shows that after 4 h of ouabain treatment (b), membrane potential was about 68 + 2% depolarized Both gramicidin D and ouabain treatment, besides depolarizing the plasma membrane potential, also affect the cytoplasmic concentration of Na+ and K+. It cannot be excluded therefore that the inhibitory effect on Ca2+ mobilization is due to perturbation of the Na'/K' concentration rather than to depolarization of the plasma membrane. We mimicked the effects of gramicidin D and ouabain on the intracellular Na+/ K' concentration by treating HL-60 cells simultaneously with nigericin, a proton/K+ exchanger, and monensin, a proton/ Na+ exchanger.
This treatment is expected to affect the intracellular Na' and K' concentrations without major effects on membrane conductance.
Nigericin (1 NM) induced a weak membrane potential depolarization, 25-30% when compared to full depolarization induced by gramicidin D (1 PM), and subsequent addition of monensin did not cause further depolarization.
In nigericin + monensin-treated HL-60 cells differentiated with DMSO or retinoic acid, the [Ca'+]i rise induced by lo-' M met-Leu-Phe or LTB+ respectively, was unaffected (data not shown).
To further rule out the possibility that inhibition of the [Ca'+], transients by gramicidin D was due to Na' overloading, we tested the effects of gramicidin D in HL-60 cells incubated in a medium where Na+ was replaced by Cs+, which is highly Plasma Membrane Potential, [Ca"+Ji,and Im(1,4,5 permeable through the gramicidin D pore (20). As shown in Fig. 6, in Cs+-containing medium, gramicidin D induced a large depolarization and Ca2+ release stimulated by submaximal fMet-Leu-Phe concentrations was completely inhibited (Fig. 6, compare truces a and b).
We have previously shown that in mature human neutrophils the dose dependence for f'Met-Leu-Phe-induced [Ca2+li rises was about 1 log unit shifted to the left compared to the fMet-Leu-Phe dose dependence for plasma membrane depolarization (6). This observation was confirmed in the present study; the relationship between [Ca*+]i rises and plasma membrane depolarization caused by increasing concentrations of fMet-Leu-Phe or LTB, in HL-60 cells is shown in Fig. 7, A and B. As shown, a rise in [Ca*+]i was detectable at agonist concentrations unable to depolarize the plasma membrane potential. Whereas the calculated EDso for Ca*+ mobilization induced by fMet-Leu-Phe in DMSO-and by LTB, in retinoic acid-differentiated HL-60 cells were 8 X 10-l' M and 10-l' M, respectively, those calculated for agonist-induced plasma membrane depolarization were 5 X lo-' M and lo-' M, respectively. Fig. 7 also shows that predepolarization of plasma membrane inhibited the [Ca*+]i rise completely only at the fMet-Leu-Phe concentrations that did not themselves depolarize plasma membrane potential (Fig. 7A)  Fura-2-loaded cells pretreated with ouabain (100 pM) for 4 h at 37 "C! (open symbols) were stimulated by increasing concentrations of fMLP in Ca*+-free medium; control cells were also kept at 37 "C for 4 h. The trichloroacetic acid extract of 3.5 x 10' cells stimulated by various concentrations of fMLP for 30 s was separated on a Dowex column as described under "Materials and Methods." [Ca*+]; rise values are mean + S.D., n = three to four determinations in three separate experiments and expressed as percent inhibition from maximal fura-saturations reached upon 1OW6 M fMLP (80 + 2%). Ins(1,4,5)Ps values, assessed in parallel using the same batch of cells, are expressed as percentage of maximum response elicited by fully stimulatory fMLP concentration (10e6 M) in the absence of ouabain (0): 846 + 13 dpm, mean + S.D. of duplicate determinations; substracted basal values (control unstimulated cells) were 240 + 12 dpm. Similar data were obtained in one separate experiment. pattern of inhibition was observed with LTB, in retinoic aciddifferentiated cells (Fig. 7B).
Inhibition of Mn2+ influx, unlike the effect on Ca2+ release from intracellular stores, was present all over the range of chemotactic peptide concentrations tested (Fig. 8) Fig. 9, A and B). Similar results were obtained with LTB, as a stimulus in retinoic acid-differentiated cells (not shown). Fig. 10 shows also that ouabain treatment caused a large inhibition of Ins(1,4,5)P3 generation. However, consistent with the results on plasma membrane potential and [Ca2+]i rises shown in Figs. 4 and 5, the effects of ouabain on Ins(1,4,5)Pa generation were smaller than that of gramicidin D (compare Fig. 9A with 10).

DISCUSSION
It is increasingly appreciated that a single membrane surface receptor may be linked to the generation of multiple intracellular signals, often with both stimulatory and inhibitory activity on cell responses. This generates a complex interplay of intracellular messages that dictate whether a given cell response will eventually be generated or aborted. Chemotactic receptors in myelocytic cells are good examples of receptors associated to the generation of both stimulatory and inhibitory signals. In this cell type, activation of the phosphoinositide turnover, mobilization of intracellular Ca2+, and increased influx of extracellular Ca2+ are thought to represent activatory signal, while the cyclic AMP increase has inhibitory functions (23). The physiological meaning of the changes in plasma membrane potential triggered by chemotactic substances is still puzzling (4,9,24); these cells are not excitable and do not possess voltage-gated Ca2+ channels (25, 26).
In a previous paper (6), we suggested that, in neutrophils, plasma membrane depolarization might represent a feedback mechanism devised to dampen agonist-induced increases of [Ca*+]i. This hypothesis was mainly based on the effects of the pore-forming ionophore gramicidin D, which inhibited both Ca2+ influx and Ca*+ mobilization. Regulation of Ca*+ entry into nonexcitable cells by plasma membrane potential is now a widely recognized phenomenon (6-8) and there is general agreement that depolarization decreases, while hyperpolarization increases, Ca*+ influx through agonist-activated Ca*+ channels, contrary to excitable cells where depolarization opens voltage-gated Ca*' channels.
There are a number of ways to depolarize plasma membrane potential in intact cells. All of them however will also alter other cellular parameters (intracellular cation and anion distribution, pH, receptor number, and affinity, etc . . . ), and, in order to make a cause-effect relationship between depolarization and [Ca'+]i rises, one should exclude that the inhibitory effect is due to one of these side effects. Isoosmotic substitution of NaCl with KC1 is the most classical procedure to depolarize plasma membrane potential; this treatment however proved not to be a suitable means in our cell model to test the effect of plasma membrane depolarization since high KC1 (a) depletes intracellular Ca'+ stores and (b) increases the affinity of fMet-Leu-Phe receptor for the agonist (6, 9). Roberts et al. (9) have previously shown complex changes on neutrophil physiology induced by high KCl, in addition to increases in fMet-Leu-Phe receptor number and affinity (see also Ref. 6). Gramicidin D, on the other hand, appears in our model the most efficient depolarizing agent, and its inhibitory effects can most likely be attributed to membrane potential collapse. This latter conclusion is based on the following evidence: (a) the dose dependence inhibition of [Ca*+]i rises by gramicidin D is superimposed on the dose dependence of gramicidin D-induced depolarization; (b) when maximal depolarization is obtained, a further increase in gramicidin D concentration has no additional effects on [Ca'+]i rises, and the EDso for gramicidin D inhibition is independent of fMet-Leu-Phe concentration; (c) the effect of gramicidin D is observed for both fMet-Leu-Phe in DMSO-and LTB, in retinoic acid-differentiated HL-60 cells, thus excluding a specific inhibitory effect on fMet-Leu-Phe receptor (27). An alternative to gramicidin D was long term ouabain treatment. Ouabain proved to be a convenient tool for depolarizing HL-60 cells without obvious undesirable side effects. In fact: (a) long term treatment with ouabain causes plasma membrane potential depolarization and inhibits [Ca"+], rises induced by both chemotactic ligands and (b) the extent of plasma membrane depolarization parallels its effects on Ca*+ release triggered by the agonist. In particular, the degree of inhibition of [Ca*+]i rises by ouabain treatment was very similar to that caused by doses of gramicidin D which caused equivalent depolarization of the plasma membrane potential.
Treatment of cells with gramicidin D and ouabain alters profoundly the intracellular Na' and K' concentration by causing extensive Na' overloading and K' depletion. However, the observation that gramicidin D also inhibits the [Ca'+], rises caused by low fMet-Leu-Phe concentrations in Cs' medium is against a major inhibitory role of Na' overloading, although under these conditions K' depletion still occurs. Therefore, we cannot exclude that the effects of gramicidin D on [Ca*+], homeostasis are, at least in part, due to intracellular K+ deprivation. On the other hand, treatment of HL-60 cells with nigericin plus monensin, which causes a large decrease in intracellular K+ concentration3 is without significant effect on [Ca*+]i transients induced by fMet-Leu-Phe or LTB.,.
It should be stressed that the effect of membrane potential collapse on Ca2+ release from stores and Ins(1,4,5)P3 formation might be specific for myelocytic cells. It remains to be seen whether the observed effect may be generalized to other agonists and whether it might also play an important regulatory role in other cellular systems.
We can only speculate about the mechanism by which depolarization inhibits Ins(1,4,5)P3 formation and thus Ca*+ mobilization. Changes in plasma membrane potential might easily affect protein mobility, the lipid bilayer organization, or the activity of various transmembrane enzymes. Possible mechanisms for the observed effect might thus include changes in the accessibility of phosphatidylinositol (4,5)-bisphosphate to phospholipase C or uncoupling an important regulatory subunit, such as a G protein, from the phospholipase C.
Finally, membrane potential collapse also reduced Ca2+ influx triggered by fMet-Leu-Phe. Unlike the effect on Ca2' mobilization, but similar to the effect on Ins(1,4,5)P3 production, the effect on Ca2+ influx is observed at all concentrations of agonist tested. This inhibition can be explained by the reduction of the driving force for Ca*+ entry across the plasma membrane, although we cannot exclude that either the chemotactic peptide-operated Ca2+ channels are voltage-modulated, although not voltage-gated, or that the inhibition of Ins(1,4,5)P3 generation might be responsible for the inhibition of Ca*+ influx.