A Nongenomic Mechanism for Progesterone-mediated Immunosuppression: Inhibition of K⁺ Channels, Ca²⁺ Signaling, and Gene Expression in T Lymphocytes

Salts and other reagents were obtained from Sigma Chemical Co. (St Louis, MO) unless otherwise noted. Thapsigargin (TG) was obtained from Alexis Corp. (San Diego, CA). Progesterone (10 mM) and PMA (3 mM) stocks were prepared in DMSO.

B3Z cells and K897 cells were provided along with the antigenic peptide fragment SIINFEKL by Dr. N. Shastri (University of California, Berkeley, CA). B3Z cells are a murine, CD8⁺, T cell hybridoma with a known antigen specificity for OVA/K^b-MHC and containing a β-galactosidase reporter gene construct (lacZ) under the control of the NF-AT promoter (26). The corresponding antigen-presenting K897 cells had been transfected with K^b class I MHC as described (27). The human leukemia T cell line Jurkat E6-1 was obtained from the American Type Culture Collection (Rockville, MD). Human peripheral T lymphocytes were collected from venous blood of healthy volunteers and isolated using a Ficoll-Hypaque density gradient as described previously (28). A population of activated T cell blasts was prepared in culture by treating the resting cells with 10 μg/ml PHA (PHA-P; Difco Laboratories, Inc., Detroit, MI). Cell lines expressing the cloned K_V channels Kv1.1, Kv1.2, Kv1.3, Kv1.4, Kv1.5, and Kv3.1 and a voltage-gated Na⁺ channel hSKM1 were maintained in culture. All cells were grown in RPMI 1640 medium containing 10% heat-inactivated fetal bovine serum, 10 mM Hepes, 2 mM glutamine, 1 mM pyruvate, 50 μM 2-ME. Cells were cultured in 25-ml flasks (Costar Corp., Cambridge, MA) at 37°C, 5% CO₂ in a humidified incubator. For experiments on inward rectifier K⁺ channels, rat basophilic leukemia (RBL) cells were maintained in Eagle's MEM supplemented with 10% FCS and 1% l-glutamine. RBL cells were plated onto glass coverslips 1–2 d before use.

The fraction of B3Z cells expressing β-galactosidase was measured using flow cytometry (FACScan^®; Becton Dickinson, San Jose, CA) as described previously (13). In brief, 5 × 10⁴ cells in RPMI without serum were placed in individual wells of 24-well plates activated either by 1 μM TG plus 50 nM PMA or by antibodies to the CD3-ε subunit of the TCR complex. In the latter, wells were coated with 10 μg/ml anti–CD3-ε antibodies (PharMingen, San Diego, CA) overnight and rinsed briefly with PBS before use. Cells were activated in the incubator for a total of 4 h before being resuspended and loaded by osmotic shock with the fluorogenic substrate, fluorescein-di-β-galactopyranoside (FDG; Molecular Probes, Inc., Eugene, OR). The fluorescence of lacZ⁺ cells was at least fivefold greater than autofluorescence. The effect of progesterone on lacZ gene expression was quantified using a MUG (4-methylumbelliferyl β-d-galactopyranoside) assay (29). In brief, B3Z cells were plated at 10⁵ cells per well in 96-well plates, and the fluorescence produced by cell lysates in a solution containing 3 mM MUG was measured in a multi-well plate reader (CytoFluor Series 4000; PerSeptive Biosystems, Framingham, MA).

Membrane currents were measured using the whole-cell configuration of the patch-clamp technique (28, 30), and membrane potential was measured using the perforated patch method in current-clamp mode with nystatin to permeabilize the cells (31). An EPC-9 amplifier (HEKA, Lambrecht/ Pfalz, Germany) interfaced to a Macintosh Quadra 700 computer was used for pulse application and data recording. Membrane voltages were corrected for liquid junction potentials, and current recordings were corrected for leak and capacitative currents. Patch pipettes were pulled from Accu-fill 90 Micropets (Becton Dickinson, Parsippany, NJ) using a P87 micropipette puller (Sutter Instruments Co., Novato, CA). Pipettes were coated with Sylgard (Dow Corning Corp., Midland, MI) and heat polished to final resistances of 2–5 MΩ. Patch-clamp experiments were performed at room temperature (20–25°C). Unless otherwise indicated, the membrane currents were filtered at 1.5 kHz. Data analysis was performed using the program Pulse (HEKA). Mammalian Ringer contained (in mM): 160 NaCl, 4.5 KCl, 2 CaCl₂, 1 MgCl₂, 10 Hepes (pH 7.4; osmolality 290–310 mosmol/kg). The ionic composition of the pipette solutions used in the individual experiments is reported in the figure legends.

[Ca²⁺]_i was measured ratiometrically using fura-2, as described previously (13). In brief, cells were loaded with 3 μM fura-2/AM (Molecular Probes, Inc.) for 30–40 min at room temperature (20–25°C). The cells were then washed three times with RPMI/10% FCS and stored in the dark. Illumination was provided by a xenon arc-lamp (Carl Zeiss, Jena, Germany) and transmitted through a filter wheel unit (Lambda 10; Axon Instruments, Inc., Foster City, CA) containing 350- and 385-nm excitation filters. The filtered light was reflected by a 400-nm dichroic mirror through a 63× oil-immersion objective to illuminate cells. Emitted light >480 nm was received by a SIT camera (C2400; Hamamatsu Photonics, Bridgewater, NJ) and the video information relayed to an image processing system (Videoprobe; ETM Systems, Petaluma, CA). Full field-of-view 8-bit images, averaged over 16 frames, were collected at 350- and 385-nm wavelengths. Digitally stored 350/385 ratios were constructed from background-corrected 350- and 385-nm images. Single-cell measurements of [Ca²⁺]_i were calculated from the 350/385 ratios using the equation of Grynkiewicz et al. (32) and a K_d of 250 nM for fura-2. The minimum 350/385 ratio was measured in single cells after incubation for 10 min in Ca²⁺-free Ringer containing 2 mM EGTA. Maximum ratio values were obtained after perfusion with Ringer containing 10 mM Ca²⁺, 1 μM TG, and 10 μM ionomycin.

Numerical values for single-cell [Ca²⁺]_i traces were analyzed with Origin (Microcal Software, Inc., Northampton, MA). Statistical analysis was performed on data sets using Excel version 5.0 (Microsoft Corp., Redmond, WA). Data are reported as mean ± SD. Analysis of variance (ANOVA) or Student's t test was used to compare mean values. Pairs of means were considered statistically different if P < 0.05.

The murine T cell hybridoma, B3Z, recognizes an octapeptide fragment from ovalbumin (SIINFEKL) and expresses the lacZ reporter construct under transcriptional control of the NF-AT response element of the IL-2 promoter (26, 27). Several treatments that increase [Ca²⁺]_i lead to NF-AT–driven lacZ expression in B3Z cells (13), including TCR engagement or stimulation with TG, a specific inhibitor of the endoplasmic reticulum Ca²⁺- ATPase (33). By flow cytometry in the present series of experiments, cross-linking the TCR with anti–CD3-ε antibodies or stimulating with TG plus PMA produced lacZ expression in the majority of B3Z cells (60 ± 4%, n = 7; or 72 ± 10%, n = 3, respectively). Progesterone reduced NF-AT–mediated lacZ gene expression in a concentration- dependent manner, with an IC₅₀ value of 22 ± 2.1 μM in cells stimulated by TG/PMA (Fig. 1,A, filled circles). The progesterone antagonist RU 486 also inhibited gene expression with slightly lower potency (Fig. 1,A, open squares). Progesterone reduced lacZ expression when B3Z cells were stimulated by TG alone or with PMA, by immobilized anti–CD3-ε antibody, or by antigen presentation (Fig. 1 B). Thus, at concentrations normally obtained in the placenta (34, 35), progesterone inhibits NF-AT–mediated gene expression when driven by four treatments that increase [Ca²⁺]_i. Our results with NF-AT–driven lacZ reporter gene expression are consistent with levels of progesterone or RU 486 shown previously to inhibit activation of human T cells in vitro (8, 9).

Upon contact with K897 cells preloaded with SIINFEKL, B3Z cells responded with an initial Ca²⁺ transient from a resting [Ca²⁺]_i of 180 ± 86 nM to a peak of 2.5 ± 0.5 μM (n = 20), followed by sustained Ca²⁺ oscillations (Fig. 2,A). In the absence of preloaded antigen, K897 cells did not elicit Ca²⁺ signaling in B3Z cells (data not shown). Application of 50 μM progesterone reversibly suppressed antigen-induced Ca²⁺ oscillations (Fig. 2 A).

To determine if progesterone directly inhibits TCR-initiated signaling or interferes with costimulatory pathways, we activated the TCR complex by cross-linking CD3. Settling B3Z cells onto chambers coated with anti–CD3-ε antibodies elicited an initial Ca²⁺ transient followed by vigorous Ca²⁺ oscillations that continued for at least 40 min (Fig. 2,B). In the presence of 50 μM progesterone, most cells produced only the initial Ca²⁺ transient lasting 271 ± 178 s, or a transient followed by severely attenuated oscillations (n = 14; Fig. 2 C). These data demonstrate that progesterone blocks Ca²⁺ signaling after TCR engagement. Since the initial Ca²⁺ transient results from IP₃- mediated release of Ca²⁺ from intracellular stores, and the sustained Ca²⁺ signal requires Ca²⁺ influx, these data also suggest that progesterone inhibits Ca²⁺ influx but not the steps that lead to Ca²⁺ release from intracellular stores.

TG inhibits the Ca²⁺ reuptake pump, leading to depletion of the intracellular Ca²⁺ stores and Ca²⁺ influx while bypassing the initial steps of TCR signaling and IP₃ generation (36, 37). In resting human T cells, the addition of TG to the bathing solution increased [Ca²⁺]_i from 72 ± 8 nM to a stable plateau level of 1.2 ± 0.1 μM (n = 76; Fig. 3,A). Progesterone reversibly inhibited the sustained Ca²⁺ signal with an IC₅₀ value of 28 ± 2.7 μM (Fig. 3, A and B). Thus, progesterone blocks Ca²⁺ influx subsequent to emptying of the Ca²⁺ stores.

Ca²⁺ influx depends upon the opening of CRAC channels and upon an electrochemical gradient for Ca²⁺ entry across the cell membrane. The negative membrane potential supported by K⁺ channels provides the electrical component of the driving force favoring Ca²⁺ entry; membrane depolarization would reduce this driving force. By direct measurement during perforated-patch current-clamp recording, we found that progesterone depolarizes the membrane potential, E_m. K_V channels normally maintain the resting membrane potential of T lymphocytes near −60 mV (38). Fig. 4,A illustrates that membrane hyperpolarization accompanies TG stimulation (from −61 ± 1.2 mV, n = 3; to −78 ± 2.0 mV, n = 11). The hyperpolarization that follows TG stimulation is produced by activation of K_Ca channels during the rise in [Ca²⁺]_i, driving E_m towards the K⁺ equilibrium potential of ∼−80 mV. Progesterone (50 μM; application bar, Fig. 4,A) not only reversed the hyperpolarization, but also resulted in depolarization to an average of −41 ± 4.3 mV (n = 13, P < 0.0001). These results suggest that progesterone may block both K_V and K_Ca channels. If the reduction of [Ca²⁺]_i by progesterone is secondary to decreased K⁺ current, then restoration of a K⁺ flux across the cell membrane should reverse this effect of progesterone. We tested this hypothesis by using the K⁺ ionophore, valinomycin, to hyperpolarize E_m. In the experiment shown in Fig. 4 B, 30 μM progesterone reduced the plateau in [Ca²⁺]_i that followed TG stimulation. The subsequent addition of valinomycin in the continued presence of progesterone resulted in an increase of [Ca²⁺]_i. Thus, Ca²⁺-imaging and current-clamp experiments demonstrate that progesterone reduces the driving force for Ca²⁺ entry, an effect that can be reversed by reestablishing K⁺ efflux. These data suggest that progesterone may affect [Ca²⁺]_i and gene expression by blocking K⁺ channels.

Whole-cell recording with the patch-clamp technique permits direct measurement of several types of channel activity in T cells. A voltage step from a holding potential of −80 to +30 mV elicits rapidly activating K_V currents, which reach a peak and then slowly inactivate. K_V currents consist of two components, a predominant fraction encoded by homotetramers of the Shaker-related Kv1.3 gene and a much smaller fraction of unknown molecular identity (for a review, see reference 24). The two components can be readily distinguished by their biophysical and pharmacological properties. Kv1.3 channels are blocked by nanomolar concentrations of a scorpion toxin, charybdotoxin (CTX), and undergo cumulative (use-dependent) C-type inactivation during repetitive depolarization. The smaller, CTX-resistant current does not exhibit use-dependent inactivation. Application of progesterone rapidly and reversibly reduces K_V currents in human or B3Z T cells, with an IC₅₀ of 29 ± 2 μM for the peak K⁺ current (Fig. 5). Progesterone accelerates the decline of K⁺ current during a depolarizing pulse, and thus the block is more potent when evaluated at the end of a 200-ms pulse (Fig. 5,C, filled triangles). The apparent increased rate of channel inactivation (Fig. 5, A and B) suggests that progesterone may preferentially bind to and block the open or inactivated Kv1.3 channel. To determine if steady-state inactivation enhances the block by progesterone, we inactivated a significant fraction of Kv1.3 channels by decreasing the holding potential from −80 to −50 mV. As shown in Fig. 6,A, progesterone (30 μM) reduced the peak Kv1.3 currents more potently when the holding potential was −50 mV (70% block; see open circle in Fig. 5,C) than when the holding potential was −80 mV (45% block), demonstrating that channel inactivation effectively enhances the block of Kv1.3 currents by progesterone. During antigen-induced oscillations of [Ca²⁺]_i, Kv1.3 channels would undergo repetitive cycles of activation, inactivation, and recovery as the membrane potential fluctuates. Activation cycles can result in frequency or use-dependent inhibition if the interval between depolarizations is less than the time required for full recovery from inactivation; normally, pulse intervals of >20 s allow full recovery. In the presence of progesterone, Kv1.3 currents steadily declined during repetitive pulsing, because channel inactivation recovers 10-fold more slowly, resulting in accumulation of channels in the inactivated state (Fig. 6, B and C).

We also examined the effects of progesterone on the smaller component of K_V current by selectively blocking the Kv1.3 component with 100 nM CTX. At this dose, we expect a residual Kv1.3 current of only 2% that in the absence of CTX. However, in 11 cells examined in this series of experiments and in previous work (21), 5–20% of the K_V current remains in the presence of CTX. This residual CTX-resistant current does not undergo use-dependent inactivation during repetitive depolarizing pulses (Fig. 7,A). Fig. 7,B demonstrates that the larger, use-dependent, CTX-sensitive component of K_V current is blocked either by progesterone or by RU 486. As shown in Fig. 7 C, most of the CTX-resistant current is also blocked by 50 μM progesterone, whereas RU 486 at the same concentration is less effective. We conclude that progesterone blocks both the predominant Kv1.3 component and the CTX- resistant component of K_V current in T cells, and that RU 486 blocks primarily the Kv1.3 component.

K_Ca channels hyperpolarize the membrane potential of T cells during the [Ca²⁺]_i signal, effectively counteracting the depolarizing effects of Ca²⁺ influx. K_Ca channels are voltage-independent and highly sensitive to a rise in [Ca²⁺]_i, with half-activation at ∼400 nM and a steep [Ca²⁺]_i dependence, suggesting that at least four Ca²⁺ ions must bind to open the channel (for a review, see reference 24). Based upon biophysical characterization and its expression in T cells, it is likely that the K_Ca channel in T cells is encoded by the gene hKCa4 (39). Intracellular dialysis of B3Z or activated human T cells with solutions containing 1 μM Ca²⁺ activated a large K⁺ current that was evident in voltage ramps (Fig. 8,A). Progesterone blocked K_Ca channels with an IC₅₀ value of 113 ± 9 μM (Fig. 8,B). Thus, progesterone blocks the K_V currents more potently than the K_Ca currents in T cells. In contrast, the progesterone antagonist RU 486 was consistently more potent than progesterone in blocking K_Ca current (Fig. 8 C). These experiments demonstrate some degree of selectivity of the two components of K_V current and the K_Ca current for progesterone and RU 486.

The results from patch-clamp experiments suggested that progesterone, RU 486, and perhaps other steroid hormones might block several K⁺ channel types, albeit with rather low affinity. Therefore, we screened a panel of steroids on the K_V and K_Ca channels in T lymphocytes. Most of the compounds tested either had no effect or were less effective than progesterone in blocking K_V or K_Ca currents (Table 1). In addition, we screened several channel types, including both cloned and native channels expressed in a variety of cell lines, for block by progesterone, as summarized in Table 2. Progesterone blocks several Kv1 family members, including Kv1.3 expressed in lymphocytes as the predominant K_V current, as well as Kv3.1 expressed in the brain and in certain subsets of mouse thymocytes. In contrast, progesterone had very little effect on a cloned voltage-gated Na⁺ channel found in skeletal muscle, or on a strongly inward rectifying K⁺ channel found in RBL cells. We conclude that progesterone is a broad spectrum, low-affinity K⁺ channel blocker.

In further experiments on Jurkat T lymphocytes, we evaluated effects of progesterone on CRAC and Cl⁻ channels to determine if modulation of these channel types might contribute to the inhibition of Ca²⁺ signaling. During whole-cell patch-clamp recordings, intracellular dialysis with heavily buffered low-Ca²⁺ solutions passively depleted the intracellular Ca²⁺ stores and activated CRAC channels (Fig. 9,A). After maximal activation, 30 μM progesterone (Fig. 9,B, application bar) had no effect on the amplitude or the current-voltage characteristics (n = 6 cells). This experiment rules out direct CRAC channel block as a possible contributor to the inhibition of Ca²⁺ signaling by progesterone; instead, it appears that progesterone blocks Ca²⁺ signaling by inhibiting K⁺ channels, indirectly reducing the driving force for Ca²⁺ entry. Cl⁻ channels have also been implicated in lymphocyte signaling mechanisms by helping to maintain E_m during T cell activation (20, 21). However, superfusion of B3Z or Jurkat cells with 50 μM progesterone did not affect the amplitude or the current-voltage characteristics of Cl⁻ currents induced by cell swelling (Fig. 9, C and D).

In this report, we demonstrate by patch-clamp measurement that progesterone directly blocks K_V and K_Ca channels, but not Ca²⁺ or Cl⁻ channels in T lymphocytes. Furthermore, we show that K⁺ channel blockade is associated with membrane depolarization, inhibition of Ca²⁺ signaling, and a reduction of NF-AT–driven gene expression. Since NF-AT links activation of the TCR to IL-2 production, interruption of these signals would inhibit production of the major proliferative cytokine for T cells. We propose that K⁺ channel blockade provides a mechanism contributing to the immunosuppressive effects of progesterone.

The rapid onset and reversibility of K_V channel block by progesterone is incompatible with changes in mRNA or protein synthesis, suggesting that these effects are not mediated by the classical steroid receptor pathway (40). The progesterone antagonist RU 486 is nearly as potent as progesterone in blocking both K_V channels and gene expression, also implicating a nongenomic action of progesterone. Furthermore, channel inactivation enhances block by progesterone, suggesting that the state of the channel modulates the affinity of progesterone. In T cells, the direct block of K_V channels by dihydroquinolines and benzhydryl piperidines is also enhanced by inactivation (41, 42). In addition, progesterone blocks Kv1.3 channels exogenously expressed in a cell line (Table 2). Progesterone also blocks other K_V and K_Ca channels with low affinity. These data suggest that progesterone blocks K⁺ channels directly, rather than acting via the classical nuclear progesterone receptor pathway.

Channel blockade by progesterone is not without precedent. Previous studies demonstrated that progesterone blocks voltage-gated Ca²⁺ channels in smooth muscle cells and a variety of K⁺ channels in MDCK cells and hepatocytes (43–45). Several transmitter-activated channels are also suppressed by progesterone in the micromolar concentration range (46–49). In contrast to its effects on somatic cells, progesterone activates Ca²⁺ influx in sperm (50, 51). We found no evidence for progesterone-induced Ca²⁺ influx in T cells.

Our data provide the first evidence that an endogenous hormone may act as an immunosuppressant by blocking K⁺ channels. Inhibition of K⁺ channels has been shown to reduce IL-2 production and T cell activation in vitro (22, 23, 52). Moreover, recent studies demonstrated that the peptide scorpion toxin margatoxin, a specific blocker of Kv1.3 channels, inhibits delayed-type hypersensitivity reaction and reduces response to allogeneic challenge in vivo (25). The depolarization and reduction of the driving force for Ca²⁺ entry resulting from K⁺ channel inhibition are sufficient to account for the reduction of Ca²⁺ signals and NF-AT–driven gene expression. CRAC channels are inwardly rectifying, and a modest depolarization can reduce Ca²⁺ entry significantly, reducing the rise in [Ca²⁺]_i below the threshold for gene expression (for a review, see reference 24). At high concentrations, progesterone reduces Ca²⁺ signaling and gene expression almost to control levels, below a plateau level seen with 100 nM CTX (21, 53, and data not shown). Progesterone, although acting with low affinity, may reduce Ca²⁺ signaling and gene expression to a greater extent than CTX because progesterone also inhibits CTX-resistant K_V channels. The block of K⁺ channels by progesterone or RU 486 can also account for previous results showing that progesterone or RU 486 inhibits activation of human T cells in vitro (8, 9), as well as the reduction of the number of CD3⁺ cells in the placenta compared with maternal blood (11, 54).

During pregnancy, immunoregulatory mechanisms must operate locally at the placental interface and be readily reversible to preserve the systemic immune competence of the mother. Several mechanisms involving progesterone may contribute to fetal–maternal protection, including altered expression of MHC class I proteins in fetal tissue, altered T cell subsets, or elaboration of immunosuppressive factors (2). Biochemical measurements have estimated progesterone concentrations to be 20 μM within the placenta (34, 35); concentrations in the vicinity of trophoblasts producing progesterone must be even higher. Average progesterone levels found in the placenta would be sufficient to block lymphocyte K⁺ channels and thereby mediate a highly localized and reversible immunosuppression without compromising the maternal immune system. The affinity of progesterone for K⁺ channels ensures that this mechanism would only be effective in the region of potential contact between allogeneic cells, where progesterone is present at high concentrations.

We thank Dr. Patricia Schmidt for helpful discussions at the onset of this project, Dr. K. George Chandy for providing clones and a panel of stable cell lines, and Dr. Lu Forrest for tissue culture support.

This work is supported by National Institutes of Health grants GM41514 and NS14609 (to M.D. Cahalan), a Schroedinger Stipendium (to H.H. Kerschbaum), and a fellowship of the Alexander von Humboldt Foundation (to C. Eder).

[Ca²⁺]_i

intracellular free calcium concentration

CRAC

Ca²⁺ release–activated Ca²⁺

CTX

charybdotoxin

E_m

membrane potential

IP₃

inositol 1,4,5-trisphosphate

K_Ca

Ca²⁺-activated K⁺

K_V

voltage-gated K⁺

NF-AT

nuclear factor of activated T cells

TG

thapsigargin