Fcgamma receptor I activation triggers a novel Ca2+-activated current selective for monovalent cations in the human monocytic cell line, U937.

Previous reports have suggested that receptors for immunoglobulin G (IgG), FcγRs, directly activate a nonselective cation channel (Young, J. D.-E., Unkeless, J. C., Young, T. M., Mauro, A., and Cohn, Z. A. (1983) Nature 306, 186-189; Nelson, D. J., Jacobs, E. R., Tang, J. M., Zeller, J. M., and Bone, R. C. (1985) J. Clin. Invest. 76, 500-507). To investigate the mechanisms underlying membrane conductance changes following human high affinity (FcγRI) receptor activation, we have used the human monocytic cell line U937 and combined conventional whole cell patch-clamp recordings with single cell fura-2 Ca2+ measurements. Using a K+-free internal solution, antibody cross-linking of IgG-occupied FcγRI activated an inward current at negative potentials, whose amplitude and time course mirrored the concomitant rise in intracellular Ca2+. Current-voltage relationships, obtained under different ionic conditions, revealed a monovalent cation-selective conductance that, under physiological conditions, would result in Na+ influx. Noise analysis of current recordings indicated a single channel conductance of 18 picosiemens and a mean opening time of 4.5 ms. This current was also activated by rises in intracellular Ca2+ induced by ionomycin (3 μM) or thapsigargin (1 μM). Addition of the Ca2+ chelator 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid to the intracellular medium abolished any channel activation by ionomycin, FcγRI, or the low affinity receptor, FcγRII. These results demonstrate that FcγRI activation triggers a novel Ca2+-activated channel selective for monovalent cations and that neither FcγRI nor FcγRII can directly activate a channel.

Human receptors for the constant, or Fc, region of immunoglobulin G (IgG), 1 Fc␥Rs, play a central role in linking the cellular and humoral arms of the immune system and trigger a number of downstream events including endocytosis, phagocytosis, superoxide generation, and cytokine release (for reviews, see Refs. [1][2][3]. Three closely related classes of Fc␥Rs have been identified: a high affinity (Fc␥RI) and two low affinity (Fc␥RII and Fc␥RIII) forms, each of which has different tissue distribution, structure, and affinity for IgG (1). Both Fc␥RI and Fc␥RII are constitutively expressed on cells of monocyte/macrophage lineage including the human monocytic cell line U937, which has commonly been used as a model in which to study Fc receptor signaling (4). Fc␥RI is a 72-kDa protein comprising three extracellular immunoglobulin (Ig)-like domains of the C2 set, a single transmembrane-spanning region, and a short cytoplasmic tail with no known signaling motifs (5), whereas the 40-kDa class II receptor contains only two Ig-like extracellular domains and has a cytoplasmic region containing tyrosine kinase activation motifs (6). Signaling by both receptors is thought to involve mainly, although not exclusively, aggregation of tyrosine kinase activation motifs leading to recruitment and activation of a number of soluble tyrosine kinases and the subsequent initiation of various signaling pathways, for example phospholipase C␥ activation (7).
Several reports suggest that Fc␥Rs could directly couple to and activate a nonselective cation channel. Indirect measurements of mouse macrophage membrane potential using [ 3 H]tetraphenylphosphonium ion accumulation indicated that Fc␥R activation triggered an initial Na ϩ -dependent depolarization followed by a prolonged hyperpolarization in part attributable to K ϩ efflux (8). Subsequent recordings from planar lipid bilayers, into which Fc␥R-containing proteoliposomes (again from mouse macrophages) had been incorporated (9), suggested that the receptor was tightly associated to and directly activating a nonselective cation channel. More recently, in studies using human macrophages, antibody cross-linking of Fc␥Rs (10) and Fc receptor-mediated phagocytosis of opsonized particles (11) have been shown to trigger channel activation and inward currents. In this study we have combined conventional whole cell patch-clamp recording and single cell fura-2 Ca 2ϩ measurements of U937 cells to examine the conductance changes following Fc receptor activation and to determine the mechanism of channel activation. Furthermore, we have investigated the possible functional consequences of channel activation using measurements of membrane potential and intracellular [Na ϩ ]. Preliminary results from this study have appeared in abstract form (12).
Electrophysiology-Whole cell patch-clamp experiments were carried out using an Axopatch 200A patch-clamp amplifier (Axon Instruments, Foster City, CA). Pipettes were pulled from borosilicate glass tubing (Clark Electromedical Instruments) and had filled resistances of 2-3 megohms. Electronic compensation of capacitance currents and series resistances (which were between 10 and 30 megohms) was performed. Membrane currents during voltage ramps were filtered at 2 kHz and sampled at 10 kHz using Axon Instruments hardware and pCLAMP6 software (Axon Instruments). Currents were also acquired continuously at 37 kHz (filtered at 5 kHz) by a VR-10B digital data recorder (Instrutech Corp.). Liquid junction potentials were measured by reference to a 3 M KCl bridge and correcting computations made. Cells were resuspended in standard external saline containing (in mM): 145 NaCl, 5 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 HEPES, 10 glucose (pH 7.35, Tris). The internal solution contained (in mM): 150 KCl, 1 MgCl 2 , 10 HEPES (pH 7.35, Tris). To minimize the contribution of K ϩ currents, K ϩ was replaced internally by Cs ϩ , externally by Na ϩ , and 10 mM tetraethylammonium chloride was added externally. For low Cl Ϫ solutions, all Cl Ϫ , except that added with divalent salts, was replaced by either aspartate or gluconate. For solutions containing no monovalent cations, these were substituted by NMDG ϩ . High Ca 2ϩ external solution contained (in mM): 110 CaCl 2 , 10 HEPES (pH 7.35). Highly buffered internal solution contained (in mM): 80 cesium gluconate, 20 Cs 4 BAPTA, 5 NaCl, 0.2 Na 2 GTP. For simultaneous fura-2 fluorescence experiments, 0.1 K 5 fura-2 was added to the patch solution. In nystatin-perforated patch experiments, the pipette contained (in mM): 100 KCl, 40 K 2 SO 4 , 1 MgCl 2 , 10 HEPES (pH 7.35, Tris). Cross-linking antibodies and ionomycin were applied from a nearby pipette (150 m from the cell) using a pressure injection system (PLI-100, Medical Systems).
Fc␥ Receptor Activation-To activate Fc␥RI, cells were first loaded, for 15 min, with polyclonal human IgG (10 M). The addition of goat anti-human IgG antibody (0.2 mg/ml) was then used to cross-link and thus activate IgG-loaded Fc␥RI. Since Fc␥RI alone can bind monomeric IgG with significant affinity, this established cross-linking method only results in Fc␥RI activation (for review, see Ref. 13). To activate Fc␥RII specifically, cells were preloaded with the mouse IgG1 anti-Fc␥RII monoclonal antibody, AT10 (10 M) followed by subsequent addition of goat anti-mouse IgG1 antibody (0.2 mg/ml).
Noise Analysis-Whole cell currents used for both nonstationary fluctuation analysis and spectral analysis were obtained at a holding potential of Ϫ28 mV with 140 mM sodium aspartate, K ϩ -free external solution, and cesium aspartate internally. Recordings were filtered at 1 kHz (through an eight-pole Bessel filter) and acquired at 5 kHz. For fluctuation analysis, the mean current and its variance were calculated for 200-ms segments taken once or twice every second before, during, and after the addition of cross-linking antibody. Spectral analysis, using Origin (Microcal, MA) software, involved averaging the fast Fourier transform of 10 4,096-point segments taken during channel activation and subtracting the average fast Fourier transform of 10 taken before the addition of cross-linker. The subtracted power spectrum was fitted by a single Lorentzian function, where S(0) is the zero frequency asymptote, and fc is the corner frequency. The total variance (V) was calculated from this backgroundcorrected spectrum by the equation Fura-2 Fluorescence Measurements-Single cell fura-2 fluorescence measurements were made using a Cairn Spectrophotometer system. Excitation light passed through a spinning filter wheel assembly containing four 340 nm and two 380 nm bandpass excitation filters. Emitted light was selected by two (400 -600 nm) dichroic filters and further filtered by a 485 nm long pass gelatin filter and a 600 nm dichroic mirror. The combined output from all 340 and 380 nm excitation filters provided a 340/380 nm ratio for each revolution of the filter wheel. The signal was then averaged to give a ratio value every 67 ms. Background and cell autofluorescence were subtracted from the signal to give fura-2 fluorescence. [Ca 2ϩ ] i was calculated according to Grynkiewicz et al. (14) using a K d for fura-2 of 135 nM.

Cross-linking Fc␥RI Triggers Activation of Two Ionic
Currents and Release of Intracellular Ca 2ϩ -Under pseudophysiological ionic conditions of 145 mM NaCl externally and 150 mM KCl pipette solution, the addition of goat anti-human antibody to IgG-loaded cells held at Ϫ20 mV resulted in the generation of a large transient outward current (current density of 23.7 Ϯ 5.6 pA/picofarad; n ϭ 3). Voltage ramps from Ϫ80 to ϩ60 mV applied every 9 s during the experiment (shown plotted against time in Fig. 1A) revealed the development of a current, entirely outward over this voltage range, which displayed a curvilinear current-voltage relationship. The current reached a peak approximately 60 s after the addition of cross-linking antibody and returned to resting levels after 2-3 min. Since the size and duration of this current mirrored the Fc␥RI-induced rise in [Ca 2ϩ ] i , shown in Fig. 1B and reported previously in U937 cells (17), and this current could be abolished by substitution of internal K ϩ by Cs ϩ (data not shown), we concluded that it was due to the activation of calcium-activated K ϩ channels known to be present in many monocytic cell types (18).
Under conditions that minimized the contribution of both K ϩ and Cl Ϫ to whole cell current recordings (145 mM sodium aspartate, 10 mM tetraethylammonium chloride externally, and 150 mM cesium aspartate internally; see "Experimental Procedures"), the addition of goat anti-human IgG antibody generated a small inward current in 23/30 cells held at Ϫ40 mV (current density 1.148 Ϯ 0.52 pA/picofarad, n ϭ 12), which mirrored the concomitant rise in [Ca 2ϩ ] i (Fig. 1B). No current activation or Ca 2ϩ rise was observed in cells not preloaded with polyclonal human IgG (n ϭ 2, data not shown). Leak-subtracted current-voltage relationships obtained during and after current activation are shown in Fig. 1C and demonstrate the development of a linear ("ohmic") conductance with an E rev (where E rev ϭ a reversal potential), under these ionic conditions, of Ϫ16 mV.
Activation Mechanism of Nonselective Cation Current-To determine whether this current was activated directly by Fc␥RI or required an increase in [Ca 2ϩ ] i , 20 mM Cs 4 BAPTA was added to the pipette solution to buffer any [Ca 2ϩ ] i rise. Under these conditions, the addition of goat anti-human antibody both failed to trigger a [Ca 2ϩ ] i rise ( Fig. 2A) and also prevented the activation of any whole cell current (n ϭ 11), as shown by the leak-subtracted current-voltage curves (Fig. 2B). Under identical conditions, specifically cross-linking Fc␥RII using monoclonal antibody (see "Experimental Procedures"), again failed to activate a current (n ϭ 9). In separate fluorescence experiments, specific Fc␥RII cross-linking was shown to trigger a [Ca 2ϩ ] i rise in intact cells (data not shown). Leak-subtracted current-voltage relationships obtained 60 and 180 s following Fc␥RII cross-linking are shown in Fig. 2C. These results indicate that neither Fc␥RI nor Fc␥RII can directly activate a nonselective current in U937 cells.
To confirm that a rise in [Ca 2ϩ ] i was necessary and sufficient to activate a nonselective current, we examined the effect of a 10-s application of the Ca 2ϩ ionophore, ionomycin (3 M), to whole cell current recordings with Na ϩ aspartate externally and Cs ϩ aspartate internally. In 6/6 cells held at Ϫ40 mV, ionomycin activated an inward current (current density 2.53 Ϯ 1.375 pA/picofarad). The leak-subtracted current-voltage curves, generated by voltage ramps applied every 5 s following the addition of ionomycin, are shown in Fig. 2D and demonstrate the development of a nonselective conductance that, at high levels of current activation, displays some inward rectification. Since these recordings were obtained in the presence of 5 mM KCl, one possibility was that this observed rectification was due to a small K ϩ influx through Ca 2ϩ -activated K ϩ chan-nels which, despite the presence of internal Cs ϩ (blocking K ϩ efflux), would still be predicted under these conditions. This conclusion was supported by a similar experiment, carried out in the absence of any external K ϩ (Fig. 2E), where the activated current showed a linear (ohmic) current-voltage relationship in  the voltage range Ϫ80 to ϩ60 mV.
An indication of the affinity of Ca 2ϩ binding and the level of cooperativity involved in activation of this current was obtained by plotting the degree of current activation (under ionic conditions identical to those for Fig. 2E) as a percentage of maximum against [Ca 2ϩ ] i , as determined by simultaneous fura-2 fluorescence measurements. The resultant plot (Fig. 2F) was fitted by a modified Hill equation with a dissociation constant (K d ) of 278 nM and a Hill coefficient (n) of 4.1 Ionic Selectivity-The ion selectivity of this Ca 2ϩ -activated conductance was investigated in a series of ionic substitution experiments (Fig. 3 A-C). Voltage ramps, applied throughout the experiments, were used to generate leak-subtracted current-voltage relationships of the Fc␥RI-induced current as described above. When external monovalent cations were substi-tuted with the impermeant ion, NMDG ϩ , an outward current developed in response to a 10-s addition of either cross-linking antibody (current n ϭ 8; Fig. 3A) or 3 M ionomycin (n ϭ 12, data not shown). Substitution of both internal and external monovalent cations with NMDG ϩ in the presence of 5 mM Ca 2ϩ externally resulted in no detectable conductance change (n ϭ 6; Fig. 3B), suggesting a conductance permeable to monovalent cations and with little or no permeability to Ca 2ϩ . To define the level of permeability to divalents, 110 mM CaCl 2 was used externally with NMDG ϩ aspartate internally. The addition of 3 M ionomycin (n ϭ 5; Fig. 3C) or cross-linking antibody (n ϭ 3; data not shown) failed to generate any detectable inward current. Ionomycin also failed to activate a current when the external solution was changed to 110 mM BaCl 2 (n ϭ 4, data not shown), further indicating no significant permeability to divalent cations.
Single Channel Properties-The mean single channel conductance of this current was estimated by nonlinear fluctuation analysis. The variance and mean current were calculated for 200-ms segments taken once every 0.5 or 1 s during low levels of channel activation (Fig. 4A). The variance is shown in Fig.  4A, and the variance plotted against the mean current in Fig.  4B. Although the relationship of current variance with mean current over the entire range of opening probabilities is best described by a binomial distribution (19), at low levels of channel activity (i.e. when channel opening follows a Poisson distribution), it is approximately linear with a slope equal to the mean single channel current (19,20). A straight line, fitted to the data in Fig. 4C by linear regression), gave a single channel current of 218 fA. With a holding potential of Ϫ28 mV and a reversal potential under these ionic conditions, of Ϫ16 mV, this corresponded to a unitary conductance of 18 pS (n ϭ 2). This may be an underestimate since this method is known to generate lower values than direct single channel recording (21).
We went on to use spectral analysis to determine the mean channel opening time. The background corrected spectrum (Fig. 4C) was well fitted by a single Lorentzian function with a corner frequency of 35 Hz. This corresponded to a single open state with a mean opening time of 4.5 ms. The total variance calculated from this background-corrected spectrum (1.37 pA 2 ) agreed well with the variance of the mean current, obtained by fluctuation analysis, in the same experiment (1.1 pA 2 ), confirming that the power spectrum obtained was dominated by noise attributable to the nonselective cation channel.
Fc␥RIand Ca 2ϩ -triggered Na ϩ Influx-One expected consequence of activation of this monovalent-selective channel under physiological conditions will be to cause an influx of Na ϩ . We examined the magnitude of changes in [Na ϩ ] i using population fluorescence recordings of cells loaded with the Na ϩ indicator, SBFI (Fig. 5A). Cross-linking Fc␥RI triggered a slow rise in [Na ϩ ] i reaching a peak of 18.7 Ϯ 3.1 mM (n ϭ 3) after approximately 4 min and returning to basal levels after 20 -25 min. To assess whether this Na ϩ influx would result in a membrane depolarization, current-clamp recordings were made under nystatin whole cell configuration. Fig. 5B shows a typical experiment. Resting membrane potentials of Ϫ21.0 Ϯ 8.29 mV (n ϭ 7) were recorded. These values compare with previous measurements in monocytes and macrophages of between Ϫ15 and Ϫ56 mV (10; for review, see Ref. 18). Following a short delay, the addition of cross-linking antibody resulted in membrane hyperpolarization to Ϫ64.7 Ϯ 6.07 mV (n ϭ 7) lasting 2-3 min.

DISCUSSION
The present study demonstrates that Fc␥RI cross-linking triggers a Ca 2ϩ -activated cation channel, highly selective for monovalent over divalent ions, with a unitary conductance of FIG. 3. Ionic selectivity of the Fc receptor-evoked current observed in K ؉ -free salines. Panels A-C, ramp-generated current-voltage relationships of the Fc-and intracellular Ca 2ϩ -evoked current under various ionic conditions. The relationships were corrected for background currents by subtraction of currents observed prior to either Fc␥RI receptor (panels A and B) or ionomycin (panel C) stimulation. The major ions present were: panel A, NMDG gluconate (external) and cesium gluconate (internal); panel B, NMDG gluconate (external and internal); panel C, CaCl 2 (external) and NMDG gluconate (internal). The external CaCl 2 concentrations were 2, 5, and 110 mM, respectively, in panels A, B, and C. 18 pS. In addition, we have shown that neither the high affinity (Fc␥RI) nor low affinity (Fc␥RII) forms of the IgG receptor can directly activate a nonselective cation channel in U937 cells. There have been numerous reports of various types of Ca 2ϩactivated nonselective cation (CAN) channels in a wide variety of cells types (for review, see Ref. 22). These channels show considerable variation in unitary conductance (18 -45 pS), mean opening time (0.5-930 ms), and Ca 2ϩ sensitivity for activation (50 nM-1 mM). In addition, there seems to be a division between CAN channels showing some permeability to divalent as well as monovalent cations and those that show no detectable permeability to Ca 2ϩ (22). Thus the Ca 2ϩ -activated channel found in U937 cells appears to belong to this last group. A Ca 2ϩ -activated channel with similar selectivity for monovalents and single channel conductance (22 pS) has been reported in neuroblastoma cells (23); however, the Ca 2ϩ sensitivity for activation (K d of 1 M) and mean single channel opening time (50 -200 ms) vary considerably from those observed for the channel in U937 cells (K d for Ca 2ϩ activation: 278 nM; mean opening time: 4.5 ms).
The properties of the Ca 2ϩ -activated monovalent cation channel in U937 cells are similar to those of Fc receptoroperated cation channels reported by other groups. The direct Fc receptor-operated channel reported from lipid bilayer studies (9) has similar ionic selectivity (low Ca 2ϩ permeability) and a slightly larger single channel conductance (50 pS). In single cell studies where Fc receptor cross-linking, by antibody (10) or opsonized particle (11), has been shown to trigger a current attributable to Na ϩ influx, intracellular Ca 2ϩ changes were not monitored or prevented. This raises the possibility that channel activation was via a rise in [Ca 2ϩ ] i and not directly receptor-triggered. Inside-out patch recordings from mouse macrophages, excised after Fc␥R-evoked channel activity had been observed in a cell-attached configuration (24), showed the presence of a nonselective cation channel with a 35-45-pS single channel conductance whose opening could be modulated by [Ca 2ϩ ] i . Neither Ca 2ϩ activation in the absence of receptor cross-linking nor Ca 2ϩ permeability was assessed; however, it seems likely that this channel is similar to the one reported in this study.
Whole cell current recordings indicated that activation of Ca 2ϩ -dependent K ϩ channels was the dominant ionic conductance change following Fc␥RI stimulation of U937 cells under pseudophysiological conditions (see Fig. 1A). This accounts for the prolonged hyperpolarization of more than 45 mV observed in current-clamp recordings of membrane potential (Fig. 5B). The lack of observable depolarization can be explained by both the opposing action of K ϩ current through Ca 2ϩ -activated K ϩ channels and the fact that the resting potential (Ϫ21 mV) is very close to the reversal potential for this Ca 2ϩ -activated cation channel (Ϫ16 mV). Differences in resting potentials and variations in the relative density and/or differences in Ca 2ϩbinding affinities of CAN and Ca 2ϩ -activated K ϩ channels may explain the initial depolarization and subsequent hyperpolarization observed in mouse macrophages (8) and the transient outward current followed by a sustained inward current reported in human alveolar macrophages (10) following Fc␥R activation. Indeed, earlier microelectrode studies reporting action potentials in human monocyte-derived macrophages (25) may also be explained along similar lines.
The possible physiological role for this CAN channel remains unclear. One predicted consequence of channel activation would be to cause a Na ϩ influx that would be enhanced under conditions of membrane hyperpolarization. Fluorescence measurements of SBFI-loaded U937 cells (Fig. 5A) revealed an Fc␥R-triggered [Na ϩ ] i rise of 10 -20 mM. The contribution of CAN channel activation to this [Na ϩ ] i can be estimated by integration of Fc␥RI-evoked currents generated under the ionic conditions in Fig. 1B and scaled to holding potentials of Ϫ75 mV. This is the potential observed in current-clamp experiments following cross-linking antibody addition (Fig. 5B). Using this method, the total Na ϩ influx through CAN channels following Fc␥RI activation was obtained and provided an estimated [Na ϩ ] i increase, for an 8-m cell, of 31.6 Ϯ 3.3 mM. This value is considerably greater than peak Na ϩ concentrations obtained by SBFI fluorescence measurements (18.7 mM) and indicates that the CAN channel can account for most of the observed Na ϩ influx. However, it also suggests that some Na ϩ efflux must take place. Further studies will be required to assess the contributions of other influx pathways, such as Na ϩ /Ca 2ϩ or Na ϩ /H ϩ exchangers, to this [Na ϩ ] i rise. Increases in [Na ϩ ] i have been reported to alter cytosolic pH (26) and osmolarity (27) required for cytoskeletal rearrangement in a variety of cell types, modulate G protein receptor coupling (28), and alter K ϩ channel activity (29). This raises the possibility that one or more downstream events initiated by Fc␥R activa-tion may require or be modulated by rises in [Na ϩ ] i .
In conclusion, we have shown that in U937 cells, Fc receptor aggregation does not activate a conductance directly but triggers a Ca 2ϩ -activated cation channel, selective for monovalents, which contributes to Fc␥R-mediated Na ϩ influx.