Control of Alternative Splicing of Potassium Channels by Stress Hormones

See allHide authors and affiliations

Science  17 Apr 1998:
Vol. 280, Issue 5362, pp. 443-446
DOI: 10.1126/science.280.5362.443


Many molecular mechanisms for neural adaptation to stress remain unknown. Expression of alternative splice variants of Slo, a gene encoding calcium- and voltage-activated potassium channels, was measured in rat adrenal chromaffin tissue from normal and hypophysectomized animals. Hypophysectomy triggered an abrupt decrease in the proportion of Slo transcripts containing a “STREX” exon. The decrease was prevented by adrenocorticotropic hormone injections. InXenopus oocytes, STREX variants produced channels with functional properties associated with enhanced repetitive firing. Thus, the hormonal stress axis is likely to control the excitable properties of epinephrine-secreting cells by regulating alternative splicing of Slo messenger RNA.

Stressors including cold exposure, hypoglycemia, and physical constraint trigger adaptive changes in catecholamine- and peptide-secreting chromaffin cells of the adrenal medulla. Rapid stress-induced increases in transcription of the epinephrine-synthesizing enzyme phenylethanolamine-N-methyltransferase (PNMT) result from direct interaction of receptor-bound glucocorticoid stress hormones with glucocorticoid response elements in the promoter (1). Glucocorticoids also regulate transcription of voltage-gated K channel genes in cardiac and pituitary cells (2). In chromaffin cells, large-conductance “BK” calcium- and voltage-gated K channels are particularly prominent, participating in action potential repolarization and driving the brief afterhyperpolarization. Pronounced differences in the repetitive firing properties of chromaffin cells have been attributed to variations in kinetics and voltage dependence of BK channel gating in these cells (3, 4).Slo is the only known gene encoding BK channels, and alternative splicing may underlie much of the existing BK functional diversity (5, 6).

Hypophysectomy (pituitary removal) in rats represents a robust and reproducible experimental nullification of hormonal stress axis function and has confirmed results with other approaches demonstrating extensive stress-related plasticity in chromaffin cell phenotype (7). We used a quantitative reverse transcription polymerase chain reaction (RT-PCR) strategy (8) to measure in parallel the effects of hypophysectomy (9) on adrenal medullary levels of PNMT and total rat Slo (rSlo) mRNA (without regard to splicing variation). PNMT mRNA was reduced to 8.5 ± 1.1% (±SEM;N = 4) of normal levels by hypophysectomy, whereas total Slo mRNA was not detectably altered over 10 weeks after hypophysectomy.

Five alternative splice sites have been identified in the COOH-terminal half of mammalian Slo genes (5, 10,11). Using RT-PCR on adrenal medulla, we found no inserts at sites I, III, or IV (Fig. 1A). However, primers bracketing sites I to III yielded two distinct bands. Subcloning and sequencing indicated that the lower molecular weight band contained only products with no insert at site II (configuration referred to as ZERO). Two related variants of similar size composed the upper band, referred to as STREX-1 and -2 (stress axis–regulated exons). The STREX variants share a 174–base pair (bp) exon (10, 11), with STREX-2 having an additional 9-bp exon (Fig. 1B).

Figure 1

Structure of mammalian Slo mRNAs. (A) Roman numerals indicate five alternative splice sites downstream of the six transmembrane domains common to voltage-gated K+ channels (boxes). PCR primers Slo1 and Slo2 spanning a non–alternatively spliced junction were used to measure total Slo transcript amounts in adrenal medullary RNA. Slo3 and Slo4 revealed inserts at site II only. (B) Three variants at site II. ZERO had no insert between the flanking exons. Nucleotide sequence of STREX-1 dictates the substitution of P for L, followed by the insertion of 58 residues. The sequence of STREX-2 dictates the insertion, after L, of 61 residues, the downstream 58 of which are shared with STREX-1 at nucleotide and amino acid levels (C). Boxed segments demarcate midcodon splice boundaries providing the most parsimonious explanation for the observed sequences. In STREX-2, a 9-bp exon is presumably inserted before the 174-bp exon in STREX-1. Insertion of this 9-bp exon alone can account for a variant identified in other tissues (6) having residues IYF following L. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; R, Arg; S, Ser; T, Thr; V, Val; and Y, Tyr.

The relative intensities of STREX and ZERO PCR product bands were very consistent between reactions from the same sample, providing an internal calibration system for measuring relative abundance across treatment groups (12). The ratio of STREX to ZERO forms was markedly reduced by hypophysectomy (Fig.2). In six unoperated animals, the ratio varied from 0.59 to 0.83 (mean ± SEM = 0.74 ± 0.02) (Fig. 2B). In 14 rats hypophysectomized 15 days earlier, the ratio varied from 0.13 to 0.39 (mean = 0.25 ± 0.005). Thus, ranges did not overlap, and means differed very significantly (P < 0.00002). In a separate series of reactions, we measured a decrease in the abundance of STREX relative to total RNA using STREX-specific primers and a STREX-derived calibration template (P < 0.0002). No other variants between the obligatory primer sites were detected in adrenal medulla by us or others (10). Because total Slo transcript was not changed by hypophysectomy, the change described likely reflects a decrease in the absolute abundance of STREX with an accompanying increase in ZERO.

Figure 2

Stress hormone–related changes in expression of Slo splice variants. (A) Agarose gel electrophoresis of RT-PCR products obtained with primers Slo3 and Slo4 on RNA from adrenal medullas of three rats for each group. The lower band represents the 450-bp product from ZERO, and the upper band is a composite of unresolved STREX-1 and -2 products (624 and 633 bp). STREX band intensity is much reduced at 15 days after hypophysectomy (H), compared with age- and sex-matched normal rats (N). ACTH injections beginning at hypophysectomy (H + A) maintain STREX variants at near normal levels, although implanted corticosterone pellets (H + C) do not. (B) Mean values (± SEM) of ratios of upper to lower gel band intensities (STREX to ZERO products), measured from digitized images of ethidium bromide–stained gels. Sample sizes are above bars. H was significantly below N (P < 0.00002) and H + A (P < 0.00001). (C) Serum corticosterone (CORT) concentrations at death, measured by RIA. P< 0.0005 for H compared with N; P < 0.02 for H + A compared with N. (D) Adrenal corticosterone concentrations. ACTH injections but not corticosterone implants were effective in raising corticosterone concentrations locally (P < 0.0006 for H + A compared with H alone). (E) Time course of the decline in STREX/ZERO abundance (top) after hypophysectomy parallels that for PNMT/actin abundance (bottom). PNMT and STREX were normalized by means for unoperated rats in parallel reactions.

The STREX to ZERO ratio declined rapidly after hypophysectomy (Fig.2E). By 11 days, it was 42.5% of normal (P = 0.002), and by 32 days, it was 11.5% of normal, indistinguishable from the value at 70 days. Thus, the time course of STREX decline was very similar to that for PNMT mRNA.

Pituitary influence on chromaffin cells is generally indirect, with adrenocorticotropic hormone (ACTH) acting on cortical cells to stimulate the synthesis of glucocorticoids, which then act on chromaffin cells. Subcutaneous injections of ACTH (4 U/100 g) (13) were given daily or on alternate days beginning at surgery to determine whether the effects of hypophysectomy could be prevented by ACTH replacement. Fifteen days after surgery, ratios of STREX to ZERO products were indistinguishable from normal animals and significantly higher than those of vehicle-injected or uninjected hypophysectomized animals (Fig. 2B).

Attempts to bypass ACTH by directly manipulating corticosteroid concentrations in the blood were ineffective in controlling STREX variant expression. Neither dexamethasone injections 81 days after surgery (10 mg kg−1 day−1 for 4 days) nor implantation of corticosterone pellets (the major rat adrenal glucocorticoid under ACTH control) at surgery raised or maintained STREX levels significantly above hypophysectomized levels. One explanation is that injections and implants, although restoring serum corticosterone concentrations, cannot match the much higher local concentrations to which chromaffin cells are exposed in cortico-medullary venous sinuses within the gland (14). To address this question, we used radioimmunoassay (RIA) (15) to measure serum corticosterone concentrations and total gland corticosterone content. Normal and hypophysectomized rat sera contained 312 ± 22 and 20 ± 6 ng/ml, respectively (Fig. 2D). Corticosterone was roughly 100-fold more concentrated in adrenal tissue than in serum (Fig. 2D). Corticosterone implants (100 mg of pellet implanted subcutaneously at surgery) raised serum concentrations to 121 ± 14% of normal but raised adrenal content to only 10 ± 2% of normal. By contrast, ACTH injections raised serum concentrations to 529 ± 113% and adrenal content to 61 ± 7% of normal. They also prevented the cortical atrophy normally caused by hypophysectomy. Thus, ACTH injections maintained high local concentrations (with supernormal secretory rates) apparently required to maintain near normal STREX levels after hypophysectomy.

We estimated the relative abundance of STREX and ZERO transcripts in native tissue by correcting for differences in PCR amplification efficiencies, assuming similar RT efficiencies. To measure amplification efficiencies, we constructed a plasmid containing both ZERO and STREX-2 isoforms, ensuring a one to one template ratio. Amplification yielded band intensity ratios (STREX to ZERO) of 0.64 ± 0.2 (STREX efficiency = 97% of ZERO efficiency per cycle). The ratio was negligibly affected by template amount, even when 100-fold greater than in tissue samples, indicating amplification within the linear range. Extrapolating from relative efficiencies, the STREX to ZERO template ratio in normal rats was 1.16. For hypophysectomized and ACTH-injected hypophysectomized rats, this ratio was 0.47 and 1.17, respectively, at 15 days. With the assumption that no other variants at site II occur in significant proportions in chromaffin cells, STREX variants made up about 53, 32, and 54% of total Slo transcripts in the respective groups. By 70 days, STREX transcripts declined to 15% of the total.

To determine whether STREX exons affect the functional properties of Slo channels, we constructed Slo expression plasmids differing only at site II, having the form of STREX-1, STREX-2, or ZERO (no inserts at sites I, III, and IV) (16). Currents in inside-out patches pulled from STREX cRNA-injected Xenopus oocytes (17) activated at voltages ∼20 mV negative to those in ZERO-injected oocytes (Fig. 3) (10, 18). At 10 μM intracellular Ca2+ concentration ([Ca2+]i), the half-activation voltage (V 0.5, from Boltzmann fits) was −2.33 ± 0.30 mV for 25 STREX patches and 20.5 ± 1.44 mV for 8 ZERO patches. The steepness of voltage dependence did not differ (17.1 ± 0.5 mV/e-fold change for STREX, 16.2 ± 1.2 for ZERO). In addition, STREX speeds activation and slows deactivation at a given test potential. Slower BK deactivation has been linked to enhanced repetitive firing in chromaffin cells. Time constants of tail current kinetics (+90 to −80 mV) averaged 2.31 ± 0.06 and 0.85 ± 0.06 ms for 18 STREX and 7 ZERO patches, respectively.

Figure 3

Functional differences in Slo splice variants. (A) STREX-2 channels activate at more negative voltages than ZERO channels. Representative currents activated from a holding potential of −100 mV in inside-out patches from cRNA-injected Xen- opus oocytes. [Ca2+]i = 100 μM. (B) Conductance-voltage (G-V) relations for patches in (A). G values were normalized by the maximal conductance (G max) in each patch and fit with Boltzmann functions. (C) AverageG-V curves for 25 STREX and 8 ZERO patches (STREX-1 and -2 not different; [Ca2+]i = 10 μM). Curves were generated from independently averagedV 0.5 and steepness parameters from individual fits. Horizontal SEM bars for V 0.5 are barely detectable. Steepness varied little within or between groups. (D) STREX channels deactivate more slowly, as seen in tail currents for ZERO and STREX-2 patches [Ca2+]i= 100 μM. I, current. (E) Mean time (τ) constants (± SEM) from exponential fits to tails with [Ca2+]i = 10 μM.

Studies of firing properties and BK channels in native rat chromaffin cells (3, 4) suggest that STREX exons enhance repetitive firing. Chromaffin cells subdivide into two types, those spiking continuously with a sustained current injection and those spiking only once or twice. BK channels in the former exhibit substantially slower deactivation kinetics. This slower kinetics enhances repetitive firing by augmenting the afterhyperpolarization, facilitating recovery from inactivation of sodium and calcium channels. Both more negative and faster activation will further augment the afterhyperpolarization by increasing BK openings during the brief action potential. It has been proposed that chromaffin cells secreting epinephrine rather than norepinephrine fire repetitively (3). Because the pattern of glucocorticoid receptor expression maps onto that for PNMT expression (19), we hypothesize that STREX is differentially expressed in epinephrine- and norepinephrine-secreting cells.

Our observation that stress hormones affect Slo mRNA composition suggests that hypophysectomy and more natural stress system perturbations will alter chromaffin BK channel protein composition and cellular excitability. For PNMT, protein levels and enzyme activity are not always commensurate with mRNA fluctuations, as translational and later stages are subject to additional regulatory controls (20). However, even small changes in STREX expression producing subtle changes in chromaffin cell excitability could have pronounced effects on the secretion of catecholamines and other products (including corticosteroids through reciprocal interactions with the adrenal cortex) (21), with potentially far-reaching consequences in humans. Among the functions likely to be affected to some extent are cardiovascular, digestive, metabolic, immune, and mental functions (22). Changes in chromaffin BK channels in response to hormonally communicated stress represent another dimension of stress-related plasticity in adrenal tissue.

Tissue-specific and developmental regulation of alternative splicing is well established for many genes, with factors and mechanisms being worked out. Reports of dynamic regulation of splicing patterns in adult tissues are rare (23). Exons in complex modular proteins such as ion channels often comprise discrete functional units, and dynamic hormonal control of exon selection provides a unique dimension for regulating the often critical functional nuances of the whole protein.

  • * To whom correspondence should be addressed. E-mail: dpm9{at}


Stay Connected to Science

Navigate This Article