Cysteinyl leukotriene type I receptor desensitization sustains Ca2+-dependent gene expression.

Receptor desensitization is a universal mechanism to turn off a biological response; in this process, the ability of a physiological trigger to activate a cell is lost despite the continued presence of the stimulus. Receptor desensitization of G-protein-coupled receptors involves uncoupling of the receptor from its G-protein or second-messenger pathway followed by receptor internalization. G-protein-coupled cysteinyl leukotriene type I (CysLT1) receptors regulate immune-cell function and CysLT1 receptors are an established therapeutic target for allergies, including asthma. Desensitization of CysLT1 receptors arises predominantly from protein-kinase-C-dependent phosphorylation of three serine residues in the receptor carboxy terminus. Physiological concentrations of the receptor agonist leukotriene C(4) (LTC(4)) evoke repetitive cytoplasmic Ca(2+) oscillations, reflecting regenerative Ca(2+) release from stores, which is sustained by Ca(2+) entry through store-operated calcium-release-activated calcium (CRAC) channels. CRAC channels are tightly linked to expression of the transcription factor c-fos, a regulator of numerous genes important to cell growth and development. Here we show that abolishing leukotriene receptor desensitization suppresses agonist-driven gene expression in a rat cell line. Mechanistically, stimulation of non-desensitizing receptors evoked prolonged inositol-trisphosphate-mediated Ca(2+) release, which led to accelerated Ca(2+)-dependent slow inactivation of CRAC channels and a subsequent loss of excitation-transcription coupling. Hence, rather than serving to turn off a biological response, reversible desensitization of a Ca(2+) mobilizing receptor acts as an 'on' switch, sustaining long-term signalling in the immune system.

Receptor desensitization poses a paradox: how can long-term responses be evoked if the receptor inactivates? This is a particularly acute problem in immune cells, where cell differentiation and clonal selection develop over hours in the continued presence of external cues.

Stimulation of RBL-1 cells with LTC4, acting exclusively on CysLT1 receptors[7,8] (Supplementary Fig. 1), led to cytoplasmic Ca2+ signals (Fig. 1a) followed by robust expression of c-fos at both mRNA (Fig. 1b-c)[4] and protein levels (Fig. 1d-e). Maximal activation of CRAC channels with thapsigargin led to a similar increase in c-fos expression (Fig. 1b-e). Both LTC4 and thapsigargin induce c-fos expression through the build-up of Ca2+ microdomains near open store-operated CRAC channels[4,5]. Thapsigargin led to a larger, more sustained Ca2+ signal than LTC4 (Fig. 1a, see also[4,5]) and the rate of Ca2+ entry through CRAC channels was ~ 2-fold more for thapsigargin than LTC4 (Fig. 1f), consistent with patch clamp recordings[7]. The similar increase in Ca2+-dependent c-fos expression to LTC4 and thapsigargin was therefore surprising, given the significant difference in CRAC channel activation.

Figure 1

CysLT1 receptor-dependent c-fos expression requires PKC

a, Averaged Ca2+ signals to LTC4 and thapsigargin are compared (>50 cells per graph). b, c-fos expression is compared between control (non-stimulated), 160 nM LTC4- and 2 μM thapsigargin-stimulated cells. Stimulus was present for 8 minutes. c, Histograms show averaged responses from 3 independent experiments. LTC4 and thapsigargin groups were different from control (p<0.001), but not from one another (p>0.3; Anova). d, Ca2+ entry rate was measured following readmission of Ca2+ to cells stimulated with LTC4 or thapsigagin in Ca2+-free solution (* denotes p<0.01). e, Cells stained with antibody against c-fos protein. f, Aggregate data are compared (n>20 per bar). Thapsigargin and LTC4 groups were different from control (p<0.001) but not from one another (p=0.11). g, G06983 (1 μM; 10 min pre-treatment) suppresses LTC4-induced c-fos expression. h, Histogram comparing the effects of PKC blockers. LTC4 control group (LTC4 in absence of PKC block) was different from the other groups (p<0.01). There were no significant differences between the other groups. G0 is G06983. i, Single-cell Ca2+ signals to LTC4 are compared for the conditions shown. j Averaged data is compared (> 45 cells for each condition). k, Histogram showing c-fos expression to thapsigargin in presence of PKC blockers. All thapsigargin-treated groups were significantly different from control (p<0.001) but were not significantly different from one another. l, Ca2+ signals to thapsigargin are unaffected by PKC block. m, Downregulation of PKC (PMA; 500 nM, 24 hours) reduces LTC4- but not thapsigargin-induced c-fos expression (data from 4 independent experiments). All stimulated groups were significantly different from control (p<0.01). For LTC4 the PMA group was different from the LTC4 control (p<0.01). For thapsigargin, the PMA groups was not different from the thapsigargin control (p=0.07). n, PKC downregulation alters the LTC4-evoked Ca2+ signal.

We considered various explanations for why CysLT1 receptor activation and thapsigargin evoked similar c-fos levels despite striking differences in the extent of CRAC channel activation. These included i) CysLT1 receptors tapped into a different signalling mechanism linking CRAC channel microdomains to c-fos expression; ii) Local Ca2+ entry through CRAC channels was larger following receptor activation because LTC4 hyperpolarized the membrane potential and iii) Cytoplasmic Ca2+ and protein kinase C (PKC) interacted synergistically to drive gene expression in response to CysLT1 receptor activation. Evidence against these possibilities is presented in Supplementary Figs 2-4. Instead, gene expression showed high sensitivity to Ca2+ entry, enabling CysLT1 receptor activation to couple effectively to c-fos transcription (Supplementary Fig. 5) as well as rapidity and high gain. Combined, this ensures efficient gene expression to bursts of CRAC channel activity following physiological levels of receptor stimulation.

Experiments described in Fig. 1g-h revealed an important role for protein kinase C (PKC) in receptor-dependent gene expression. The structurally distinct PKC blockers G06983 and calphostin C abolished c-fos expression (Figure 1g-h). PKC block had a dramatic effect on the Ca2+ signal evoked by agonist. Whereas cytoplasmic Ca 2+ oscillations were routinely observed with LTC4 (Figure 1i), the response was converted into a large, single, slowly decaying Ca2+ spike after PKC inhibition (Fig. 1i,j). Acute stimulation with PMA in the absence of LTC4 failed to induce significant c-fos expression (data not shown)[9], demonstrating that PKC activity per se was not sufficient to induce c-fos expression in these cells. The PKC inhibitors had no effect on thapsigargin-evoked c-fos expression (Fig. 1k) or cytoplasmic Ca2+ signals (Fig. 1l). Thapsigargin (2 μM) activates CRAC channels maximally (Supplementary Fig. 5) and, by blocking SERCA pumps (which can be located near CRAC channels[10]), reduces the decay of Ca2+ gradients radiating from the plasma membrane. It is possible that other non-receptor dependent stimuli, that raise local Ca2+ less effectively than 2 μM thapsigargin, might activate c-fos in a manner dependent on basal PKC activity, but this activity would be unusual in that it is not stimulated acutely by PMA in the presence of submaximal CRAC channel activation (Supplementary Fig. 4).

In RBL cells, exposure to the phorbol ester PMA for several hours downregulates several PKC isozymes[11]. Using this protocol, we found that c-fos expression was substantially reduced in response to CysLT1 receptor stimulation (Fig. 1m), whereas no significant reduction was seen when thapsigargin was used instead (Fig. 1m) or when inactive 4α-phorbol replaced PMA (data not shown). Similar to PKC blockers, the Ca2+ signal to LTC4 was prolonged after PKC downregulation (Fig. 1n). This prolonged Ca2+ signal did not reflect a change in Ca2+ clearance mechanisms (Supplementary Fig. 6); instead, it is characteristic of loss of receptor desensitization, particularly for CysLT1 receptors where desensitization is mediated predominantly by PKC[3] and prevention of desensitization leads to broader Ca2+ signals[12]. Inhibition of CysLT1 receptor desensitization is predicted to lead to greater InsP3 production and hence more extensive Ca2+ store emptying. Several findings are consistent with this. First, Ca2+ release to LTC4 lasted ~ 5 times longer when PKC was blocked than in control cells (Fig. 2a, expanded in inset). Second, the amount of Ca2+ remaining within the stores, measured as the ionomycin-sensitive Ca2+ response[13], was substantially less after activation of CysLT1 receptors in the presence of PKC block than in control cells (Fig. 2a). Third, InsP3 production, measured using the GFP-PH construct[14], increased to a greater extent when PKC was inhibited (Fig. 2b).

Figure 2

Gene expression to non-desensitizing CysLT1 receptors is rescued by preventing a cytoplasmic Ca2+ rise

a, Stimulation with LTC4 in the presence of G06983 evokes a more sustained Ca2+ release response, and this leads to more extensive store depletion (measured through the extent of Ca2+ release evoked by 5 μM ionomycin). Both LTC4 and ionomycin were applied in Ca2+-free external solution. Inset compares the kinetics of Ca2+ release. b, Cytosolic GFP-PH levels, a measure of InsP3 levels, rise when CysLT1 receptors are stimulated in the presence of G06983. c, Upper panel, loading cells with the Ca2+ chelator EGTA prevents loss of gene expression to agonist when PKC is blocked. Lower panel, aggregate data from five independent gels are summarised. d as in c, but calphostin C was used to block PKC instead. Lower panel, aggregate data from three independent gels are summarised.

Cytoplasmic Ca2+ inhibits CRAC channels through mechanisms of fast and slow inactivation[15]. The prolonged Ca2+ release evoked by LTC4 in the presence of non-desensitizing receptors could therefore inactivate CRAC channels to suppress agonist-evoked gene expression. In support of this, accumulation of the slow Ca2+ chelator EGTA in the cytoplasm rescued gene expression to CysLT1 receptor activation in the presence of PKC block (Fig 2c-d). Ca2+-dependent fast inactivation of CRAC channels is unlikely to contribute here because i) it is unaffected by the slow chelator EGTA[16,17], which reversed the inhibitory effects of PKC block (Fig. 2c-d) and ii) the rate and extent of fast inactivation was unaltered by CysLT1 receptor activation in the presence of PKC downregulation (Fig. 3a). Instead, Ca2+-dependent slow inactivation is likely to be the dominant mechanism because: i) it too is suppressed by cytoplasmic EGTA[18,19]; ii) the Ca2+-dependence of slow inactivation has a KD of ~0.5 μM and full block occurs at ~ 1 μM (Fig. 3b), which is similar to the peak Ca2+ rise evoked by LTC4 in the presence of PKC inhibitors or following downregulation of PKC (0. 87±01μM); iii) Ca2+-dependent slow inactivation develops with a time course similar to the duration of the prolonged Ca2+ rise seen to LTC4 following loss of PKC activity[18,19].

Figure 3

Ca2+-dependent slow inactivation underlies suppression of c-fos expression to non-desensitizing CysLT1 receptors

a, Ca2+-dependent fast inactivation is unaffected by non-desensitizing receptors (labelled –PKC). Cells were stimulated with LTC4 (160 nM) prior to breaking in with a pipette solution containing thapsigargin and buffered Ca2+ (140 nM) and fast inactivation was measured within 60 seconds of break-in. b, dependence of Ca2+-dependent slow inactivation on patch pipette Ca2+ concentration. c, Stimulation of non-desensitizing receptors with LTC4 prior to break-in significantly reduced the size of ICRAC that developed in response to dialysis with thapsigargin in weak buffer (0.2 mM EGTA). d, as c, but cells were dialysed with a pipette solution containing strong Ca2+ buffer (10 mM EGTA, 140 nM free Ca2+). e, Store-operated Ca2+ entry recovers partially by increasing the time interval between Ca2+ release and subsequent Ca2+ entry. f, c-fos expression to non-desensitizing receptor stimulation is rescued partially when Ca2+ entry occurs several minutes after Ca2+ release has reached completion. g, c-fos expression in human nasal mast cells after CysLT1 receptor activation is suppressed by PKC inhibition. h, Aggregate data is compared (12-17 cells per bar; 3 patients each).

If prolonged Ca2+ release to non-desensitizing CysLT1 receptors leads to slow inactivation of CRAC channels, then development of ICRAC to a subsequent stimulus should be impaired. Pre-activation of CysLT1 receptors reduced ICRAC evoked by thapsigagin but only in the presence of PKC block (Fig. 3c). No such inhibitory effect was seen when cells were dialysed with a strongly buffered Ca2+-containing pipette solution, which prevents the development of slow inactivation (Fig. 3d). Increasing the time between the termination of Ca2+ release and subsequent store-operated Ca2+ entry should enable some recovery from Ca2+-dependent slow inactivation and this should partially rescue gene expression. When Ca2+ influx was evoked a few minutes after Ca2+ release, significant, albeit incomplete, rescue of Ca2+ entry (Fig. 3e) and c-fos transcription (Fig. 3f) occurred in cells stimulated with LTC4 in the presence of PKC inhibition. Hence allowing CRAC channels time to recover from Ca2+-dependent inactivation results in partial rescue of agonist-driven gene expression.

Our attempts to express the PKC-insensitive CysLT1 receptor, in which S313, S315 and S316 had been mutated to alanines, were thwarted by the difficulty of expressing these receptors[3], although in a few cells we observed that Ca2+ oscillations to LTC4 were less frequent (3.1±0.5 versus 5.4±0.4, 4 and 6 cells, respectively) and the initial spike was a little broader (~1.25 fold) than mock-transfected cells.

To place our findings in a physiological context, we turned to the human nasal polyp, which is rich in mast cells[20]. The polyp and associated nasal mucosa are largely self-contained, providing an excellent ‘quasi in vivo human system’. Mast cells from polyps, acutely isolated from patients undergoing surgery, respond to LTC4 and express functional CRAC channels[7,8]. Stimulation with LTC4 activated c-fos protein expression in mast cells isolated from polyps (Fig. 3g) and this was reduced by pre-treatment with either calphostin C or G06983 (Fig. 3h). PKC inhibitors had no inhibitory effect when thapsigargin was used instead.

Western blots revealed the presence of Ca2+-dependent PKCα, β and ζ isozymes[11,21] but only faint expression of PKCδ and ε (Fig. 4a). Overnight PMA exposure significantly reduced PKCα and β expression, but not ζ (Figs 4a and b). Their weak expression made PKCδ and ε difficult to quantify. Confocal microscopic studies confirmed robust expression of PKCα, β and ζ (Fig. 4c), with barely detectable levels of δ and ε (data not shown). Overnight PMA exposure significantly reduced PKCα and β but not ζ at the cellular level (Fig. 4d). Knock down of PKCα using a targeted siRNA approach (Fig. 4e) resulted in a broadening of the first Ca2+ oscillation evoked by LTC4, indicative of less receptor desensitization, and fewer Ca2+ oscillations in each cell (Supplementary Fig. 7). Knockdown of PKCβ had a much weaker effect on the Ca2+ oscillations (Supplementary Fig. 7). Knockdown of PKCα or PKC α plus β simultaneously, but not PKCβ alone, reduced LTC4-driven c-fos expression to an extent similar to that seen in following overnight PMA treatment (Fig. 4f). Stimulation of non-G protein coupled FCεRI receptors in RBL-2H3 cells activates c-fos expression primarily through PKCδ and ε[22]. Although it is possible that these PKC isoforms also contribute to gene expression under our conditions, our results nevertheless suggest a major role for PKCα in G-protein coupled receptor desensitization, and thus coupling to the nucleus.

Figure 4

PKCα regulates CysLT1 receptor-driven c-fos transcription. a, Expression of PKCα, β and ζ (western blot) is shown in control cells and cells exposed to PMA for 24 hours. b, Quantification of data from 3 independent experiments as in a. c, Confocal microscopic images of PKC expression for the conditions shown. Cells were fixed before analysis. d, Quantification of images from experiments as in c. e, siRNA against PKCα or β significantly reduces corresponding protein expression. Left panel: DAPI staining of nuclei (left), GFP expression (indicating transfection; middle) and PKCα expression (right) after siRNA-mediated knockdown. Right panel, aggregate data from 4 experiments are depicted. Both siRNA groups were different from control (p<0.005). f, knockdown of PKCα, β and α + β on LTC4-dependent c-fos expression. Data are compared with mock-transfected cells. For comparison, 24 hour exposure to PMA is included. All treated groups were significantly different from the LTC4 control (black bar) group except siRNA β knockdown (p>0.1). α+β and 24 h PMA groups had p<0.01; α group had p<0.05.

Collectively, our findings reveal a counter-intuitive function for desensitization of a phospholipase C-coupled receptor. Rather than terminating a response, homologous receptor desensitization is essential for maintaining excitation-transcription coupling. Desensitization of CysLT1 receptors is mediated principally by PKC-dependent phosphorylation[3]. Prevention of receptor desensitization through either acute block or degradation of protein kinase C or after knockdown of PKCα all led to loss of Ca2+-dependent gene expression, despite potentiation of Ca2+ release to agonist. Mechanistically, the prolonged Ca2+ release phase accelerated Ca2+-dependent slow inactivation of CRAC channels, resulting in loss of Ca2+ entry. Because Ca2+ microdomains near open CRAC channels drive c-fos expression, the decline in CRAC channel activity abolishes excitation-transcription coupling. The interval between Ca2+ oscillations following CysLT1 receptor activation is ~ 25 seconds[4]. Since InsP3 has a short half-life in the cytoplasm (~ 1 second)[23], receptor desensitization will presumably lower InsP3 levels during the interspike interval. Store refilling will occur quickly and CRAC channel activity will be transient following CysLT1 receptor stimulation. The short duration of Ca2+ release and thus Ca2+ entry, determined by receptor desensitization, will ensure Ca2+-dependent slow inactivation does not develop, since this inhibitory mechanism requires a sustained Ca2+ rise for several seconds. It is therefore the kinetics of receptor desensitization and recovery from desensitization within a highly Ca2+ sensitive and high gain system that ensures bursts of store-operated Ca2+ entry occur that are sufficient for the activation of c-fos expression, without the build-up of the Ca2+-dependent slow inactivation pathway that would abolish the response.