A Cysteine-rich Domain Defined by a Novel Exon in a Slo Variant in Rat Adrenal Chromaffin Cells and PC12 Cells*

(Received for publication, October 29, 1996, and in revised form, February 18, 1997)

Mitsuyoshi Saito , Carl Nelson , Lawrence Salkoff Dagger and Christopher J. Lingle Dagger §

From the Washington University School of Medicine, Departments of Anesthesiology and Dagger  Neurobiology and Anatomy, St. Louis, Missouri 63110

ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
FOOTNOTES
REFERENCES


ABSTRACT

cDNA libraries from rat chromaffin cells and PC12 cells were screened for homologs to the mouse mSlo gene that encodes a large conductance, calcium (Ca2+)- and voltage-activated potassium channel (BK channel). One Slo variant contained sequence encoding a cysteine-rich, 59-amino acid insert for a previously described site of alternative splicing. This insert is reminiscent of zinc-finger domains. The exon was found in RNA from pancreas, anterior pituitary, cerebellum, and hippocampus. Expression in Xenopus oocytes of a Slo construct containing this exon conferred a -30 to -20 mV shift of the conductance-voltage curve. A previously uncharacterized alternative splice junction near the C-terminal end of Slo was also identified. In contrast to BK channels in rat chromaffin cells, none of the Slo variants exhibited inactivation when expressed in Xenopus oocytes. PCR screening of chromaffin cell RNA failed to reveal a homolog of an accessory beta  subunit known to influence Slo channel function. Furthermore, a beta -subunit-dependent Slo channel activator, dehydrosoyasaponin I, was without effect on chromaffin cell BK current. The results argue that an accessory subunit may not be a required component of the native chromaffin cell BK channel.


INTRODUCTION

In contrast to the extended gene families of the voltage-gated K+ channel families, molecular diversity in large conductance, voltage- and calcium (Ca2+)-activated channels (termed BK1 channels) appears to arise largely as a function of alternative splicing of a single gene product encoded by the Slo locus both in mammals (1-3) and Drosophila (4, 5). However, considerable functional diversity in BK channels exists among different tissues (6), the origins of which remain to be fully explained. Rat chromaffin cells express either of two phenotypically distinct forms of BK current (7, 8). One, termed BKi current, exhibits rapid inactivation following current activation by depolarization and elevated intracellular calcium ([Ca2+]i), while the second current, termed BKs, exhibits the more usual sustained activation as a function of [Ca2+]i and voltage.

The primary purpose of the present work was to identify and functionally characterize alternatively spliced forms of Slo found in chromaffin cells and PC12 cells, a rat pheochromacytoma line. One Slo alternative splice variant was found in both chromaffin cells and PC12 cells and encodes a pattern of cysteine residues with similarity to cysteine-rich regions in a number of other proteins. Slo constructs containing this insert exhibit a 20-mV leftward shift in the apparent voltage dependence of gating at a given [Ca2+] relative to constructs lacking this exon. Another variant also found in chromaffin cells was alternatively spliced at the C terminus, but was functionally indistinguishable from mSlo. However, no variants so far identified account for the inactivating phenotype of some chromaffin cell BK channels. Thus, inactivation behavior of BKi channels may not be governed by the properties of a particular Slo alpha  subunit itself. We then investigated whether a homolog of a known accessory beta  subunit of Slo channels (9, 10) might be found in these cells. Using both functional criteria and PCR-based screens for a beta  subunit, our results indicate that such an accessory subunit is not a component of BK channels in chromaffin cells. These results do not eliminate the possibility that other accessory subunits, different from the previously characterized beta -subunit (9), are present in these cells.


MATERIALS AND METHODS

Slo Homologs from Rat Adrenal Chromaffin Cells and PC12 Cells

PC12 cDNA libraries (courtesy of Drs. Jim Boulter (Salk Institute) and N. Gautam (Washington University School of Medicine)) and a commercial mouse chromaffin cell cDNA library (CLONTECH) were screened at high stringency by using the entire mSlo coding sequence as a probe. The probe was labeled with [32P]dCTP. Eight independent cDNA clones were isolated from the PC12 library. Two cDNAs contained virtually complete coding sequences with the exception of some N-terminal sequence and initiation methionines. Variation in N-terminal sequences was investigated by PCR screening of the cDNA libraries with primers containing vector-specific and gene-specific sequence. One presumptive initiation methionine was obtained by this method, and this N terminus was verified in rat chromaffin cells using the RACE (rapid amplification of 5' ends) method (11).

Extraction of Total RNA and RT-PCR

Total RNA was extracted from freshly isolated tissues (50-200 mg) by using a modified Chomczynski and Sacchi method (TriZol reagent, Life Technologies, Inc.). The first strand cDNAs were synthesized from the total RNA by reverse transcriptase (SuperScript II, Life Technologies, Inc.) by use of oligo(dT) primers. cDNAs were then used either for evaluation of the presence of different alternatively spliced variants or for isolation of unknown 5' end variations using the RACE method.

By using nested primer sets encoding the leading and trailing sequences adjacent to splice junction 2 (1, 2), alternatively spliced exons for this site were amplified from various tissues. The first round, outside primer sequences were: TTTAGGATTTTTCATCGCAAGTGA (sense) and GTGAAACATTCCAGTGGAGTCGTA (antisense). The second round, inside primer sequences were: TACTGCAAGGCCTGTCATGATGAC (sense) and GTGAAACATTCCAGTGGAGTCGTACTT (antisense).

The RACE reactions utilized a gene-specific antisense primer close to the initiation methionine in the mSlo coding sequence. Rat chromaffin cell total RNA was reverse-transcribed with a gene-specific primer: CCGCCAGAGCAAGATGATGAAGA. Excessive primer was removed by GeneClean II (BIO 101 Inc.), and the poly(A) sequence was added at the 5' end of the cDNAs by terminal transferase (Life Technologies, Inc.) for 15 min at 37 °C. The primer sequences were as follows. Primer set I: TAACCCTCACTAAAGGGAACATCGATTTTTTTTTTTTTTTTT (sense) and gene-specific primer I TGAGCGCATCCATC (antisense); primer set II: GCGCAATTAACCCTCACTAAAGGGAACA (sense) and gene-specific primer II CCGCCAGAGCAAGATGATGAAGA (antisense).

Construction of Expression Vectors

Since none of the PC12 clones were full-length, the functional role of the unique splice site 2 exon was investigated by subcloning a 557-base pair HindIII-NheI piece encoding the PC12 Slo 59-amino acid insert into an mSlo expression vector. For convenience, we used segments of a mouse cDNA (mbr5 (Ref. 1)), which were known to encode amino acid sequence identical to the rat Slo channel. To this we added 5', 3', and alternative exon sequence as mentioned in the text. To make these constructs, we utilized two unique restriction sites, HindIII and NheI, which are conserved in both mouse (mSlo) and rat (PC12) sequences.

For expression studies, capped cRNA was prepared by run-off transcription with T3 RNA polymerase (mMessage mMachine, Ambion). Approximately 50 nl (1 µg/lambda ) of cRNAs were injected in stage IV Xenopus oocytes. The oocytes were incubated for 1-6 days in ND96 and maintained at 17 °C.

Preparation of Rat Chromaffin Cell Cultures

Rat chromaffin cell cultures followed standard procedures used in this laboratory (12-14). Recordings were made from cells maintained in culture for 2-10 days.

Electrophysiological Methods

The vitelline membrane from injected oocytes was removed with fine forceps in hypertonic saline. All recordings were made at room temperature. The oocytes were bathed in a standard extracellular solution (ND-96: 96 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM MgCl2, 5 mM HEPES, 2.5 mM sodium pyruvate) during patch formation. For inside-out membrane patches, the pipette solution contained 140 mM potassium methanesulfonate, 20 mM KOH, 2 mM MgCl2, 10 mM HEPES, adjusted to pH 7.00 with 5% methanesulfonic acid. Just prior to excision of inside-out patches, the solution bathing the patch was changed to a 0 Ca2+ solution. Cytosolic solutions of defined [Ca2+]i were prepared as described (15) with all [Ca2+] verified with a Ca2+-sensitive electrode (catalog no. 93-20, Orion). The predominant anion in these solutions was methanesulfonic acid. For outside-out recordings, the patch pipette solution was the 100 µM CaCl2 solution.

For recordings from inside-out patches from chromaffin cells, the cells were bathed with an extracellular saline (in mM: 140 NaCl; 5.4 KCl, 10 HEPES, 1.8 CaCl2, and 2.0 MgCl2, pH 7.4) during seal formation and bathed with a 0 Ca2+ saline (in mM: 140 KCl, 20 KOH, 10 HEPES, 5 EGTA, pH 7.0) during excision of the patch. The pipette saline contained the following: (in mM) 140 KCl, 20 KOH, 2 MgCl2, 10 HEPES, pH 7.0, adjusted with N HCl, with 200 nM apamin to block small conductance, voltage-independent, Ca2+-activated SK-type currents. For cytosolic solutions of defined Ca2+, the composition was identical to that used for the 0 Ca2+ solution, except that HEDTA was used to buffer 4 and 10 µM Ca2+ solutions. Estimates of free [Ca2+] were determined as described previously (7, 8).

For patches from either oocytes or from chromaffin cells, solution exchange and drug applications were accomplished with a multibarrel solution delivery system (7, 13). Charybdotoxin (CTX) and dehydrosoyasaponin I (DHS-1) were gifts from colleagues at the Merck Research Laboratories (Rahway, NJ). CTX obtained from Peninsula Laboratories was used in some experiments. All other chemicals were from Aldrich or Sigma. Osmolarity was measured by dew point (Wescor Osmometer) and adjusted within 3% (internal saline, 290; external saline: 305). In experiments with CTX or apamin, each agent was added directly to the extracellular saline without additional adjustments. Removal of inactivation from chromaffin cell BK channels was accomplished by brief exposure to 0.3-0.5 mg/ml trypsin as described (7).

Currents were recorded with an Axopatch 1C amplifier (Axon Instruments, Foster City, CA). Voltage protocols and data acquisition were done using programs from pClamp (Axon Instruments). Fitting of current waveforms or extracted data was done using a Levenberg-Marquardt search algorithm to obtain non-linear least-squares estimates of function parameters. Conductance-voltage curves were fit with a single Boltzmann function, G(V) = Gmax/(1 + exp(V50 - V)/k), where Gmax reflects the maximal conductance, V50 is the voltage of half-maximal activation, V is the activation potential, and k is the slope factor reflecting the voltage dependence of conductance.

Estimates of drug dissociation constants (Kd) were made by fitting simultaneously the entire time course of blockade and recovery at one or more drug concentrations to a first order blocking reaction (open left-right-arrow  blocked), with forward blocking rate given by f[D], where [D] is the concentration of blocking drug, and reverse blocking rate given by b. Data points were evaluated over two time regimes where t0 = 0 is the time of drug application and t1, the time of washout of drug (Equations 1 and 2).
<UP>For </UP>t<SUB>0</SUB><t≤t<SUB>1</SUB>,
I(t)=(I<SUB>0</SUB>−I<SUB><UP>ss</UP></SUB>)<UP>exp</UP>(<UP>−</UP>&lgr;t)+I<SUB><UP>ss</UP></SUB> (Eq. 1)
<UP>For </UP>t>t<SUB>1</SUB>,
I(t)=(I<SUB>0</SUB>−I<SUB>1</SUB>)<UP>exp</UP>(<UP>−</UP>b(t−t<SUB>1</SUB>)) (Eq. 2)
Io was the current level prior to drug application, Iss is [b/(b + f[D])] × Io and indicates the steady-state level of current during blockade by a given drug concentration, Ir is the empirically determined current at the end of the actual period of drug application, and lambda  is b + f[D]. The free parameters in the fit were b and f. I0 was set to the mean of values before drug application. The apparent Kd was defined by b/f[D]. Drug applications were considered suitable for fitting if recovery from drug blockade to about 80% of the initial current amplitude prior to drug application was achieved. When multiple drug applications were fit simultaneously, the set of current amplitudes for each application was first normalized to the control current amplitude just prior to the drug application. If a reasonable period of recovery from drug application was achieved, blocking rate constants and the resulting Kd were tightly constrained even by single applications of drug.


RESULTS

Slo Homologs in Rat Chromaffin Cells and PC12 Cells

cDNA libraries from PC12 cells were screened at high stringency for mSlo homologs. Eight unique clones were identified (PC1-PC8). Two cDNAs encompassed most of the Slo coding region, but lacked an initiator methionine and the carboxyl end. The amino acid sequence was identical to the mSlo sequence, mbr5 (1), differing only at an alternative splice site termed splice junction A by Butler et al. (1) or splice junction 2 by Tseng-Crank et al., (2). Since this site corresponds to the location of the second splice site in the mammalian Slo sequence, we will refer to this site as splice site 2. The extents of the eight cDNAs are diagrammed onto the mSlo sequence in Fig. 1A.


Fig. 1. Schematic representation of clones isolated from PC12 cDNA libraries relative to mSlo sequence and unique features of PC12 Slo homologs. In A, a schematic representation of the mSlo sequence and putative transmembrane-spanning sequences (S1-S10) are aligned with eight cDNA clones isolated from PC12 libraries. Clones PC1, PC3, and PC5 each contained sequence encoding a unique 59-amino acid insert, which was absent in clone PC8, which also spanned the splice junction. Mammalian splice sites 2 and 5 are indicated by triangles. In B, the deduced amino acid sequence for the 59-amino acid insert is shown on the top. Potential phosphorylation sites are numbered. Sites 1-5 meet consensus criteria for both cGMP-dependent protein kinase and PKC-mediated phosphorylation; sites 1 and 2 meet criteria for protein kinase A-mediated phosphorylation. On the bottom, a portion of the 59-amino acid insert (underlined) together with upstream residues (italicized) that together define a pattern of cysteines with similarity to cysteine-rich regions in other proteins is shown.
[View Larger Version of this Image (28K GIF file)]

Alternative Splice Variants

Two distinct alternative splice forms were identified for splice junction 2. PC8, which has the longest 5' end, does not have any insert at splice junction 2. A second form encodes a peptide of 59 amino acids and was found in three clones, PC1, PC3, and PC5. These 59 amino acids share no sequence homology with any portion of other known proteins. However, a pattern of repeated cysteines beginning just upstream of the splice junction and including the 59 amino acids shares similarity both to variants of phospholipase A2 (PLA2) and to zinc-finger domains of protein kinase C (PKC) and diacylglycerol kinase (DGK) (Table I). In PLA2, the cysteines form disulfide linkages, whereas, in PKC and DGK, the cysteines participate in defining a zinc-binding site. By PCR methods, both splice junction 2 variants were found in both rat and mouse adrenal chromaffin cells. In contrast to variation observed at splice junction 2, none of the clones had an alternative exon at the previously described splice junction B (1). The sequence encoding the 59-amino acid insert was confirmed three times on both nucleotide strands.

Table I. Alignment of Cysteine-rich regions

Similarity in pattern of cysteines among the Slo exon, PKC variants, diacylglycerol kinase, and PLA2 variants are diagrammed. In PKC and DG kinase, this region contributes to a zinc finger structure, while, in PLA2 variants, each cysteine participates in a disulfide bridge. Similarity in pattern of cysteines among the Slo exon, PKC variants, diacylglycerol kinase, and PLA2 variants are diagrammed. In PKC and DG kinase, this region contributes to a zinc finger structure, while, in PLA2 variants, each cysteine participates in a disulfide bridge.
Protein Cysteine repeat pattern Ref.

PLA2; bovine pancreas        Cys-1-Cys-14-----Cys-Cys-QTHDN-        Cys-9-Cys 16
PLA2; cobra venom Cys-14-Cys-1-Cys-14-----Cys-Cys-QVHDN-        Cys-9-Cys 17, 18
PLA2; water mocassin        Cys-1-Cys-14-----Cys-Cys-FVHD-         Cys-Cys-6-Cys 17
PLA2; bee venom His-12-Cys-1-Cys---6----Cys-------------24--------Cys 19
PKC (alpha , beta , gamma , delta ) His-12-Cys-2-Cys-13(14)-Cys-2-Cys-Cys-FVVHKR- Cys-7-Cys 20
Rat DG kinase His-12-Cys-2-Cys-12-----Cys-Cys-2-Cys-KYAVHQR-Cys-7-Cys 21
slo-zf His-11-Cys-1-Cys-14-----Cys-Cys-FD           -Cys-6-Cys a

a This study.

C-terminal Variation

Previously published mSlo and hSlo homologs have been shown to differ at the C terminus. Specifically, following Q1187 there is sequence divergence between mSlo and hSlo (1, 3). Similarly, one cDNA clone from PC12 cells (PC7) contained sequence which was identical to mSlo and hSlo through Q1187, but which diverged downstream (Table II). This appears to represent a previously unrecognized splice site in both rodents and humans. Since four other splice sites were previously described in mammalian Slo forms (2), this new splice junction should be considered splice site 5. 

Table II. Sequence variation of Slo C-terminal domains


1. ANRPNRPKSRESRDKQ NATRMTRMGQAEKKWFTDEPDNAYPRNIQI          NQYKSTSSLIPPIREVEDECU
2. ANRPNRPKSRESRDKQ  T         EKKWFTDEPDNAYPRNIQIEMSTHMANQINQYKSTSSLIPPIREVEDECU
3. ANRPNRPKSRESRDKQ NRKEMVYR
4. ANRQNRPKSRESRDKQ  KYVQEERLU
5. ANRQNRPKSRESRDKQ NNRRCWWFSKRQDIHQQKRNGLGMRRIMPIPETFKSSP

The deduced C-terminal amino acid sequence for: 1, mSlo (1); 2, Canis familiaris (GenBankTM accession number: U41001[GenBank]); 3, PC12 (present study) and mouse whole brain (22); 4, hSlo (human skeletal muscle (22); human smooth muscle (3); human brain (2); human myometrium (23)); and 5, Slo from rat smooth muscle (GenBankTM accession number: U55995[GenBank]) are compared.

N-terminal Variation

Since an initiator methionine was not found in any of the PC12 cDNA clones, PCR methods were used to isolate the 5' ends from RNA. Two sets of nesting primers were designed. Sense primers were specific to the cloning vector (pBlueScript, Stratagene); antisense primers were specific to PC12 cDNA sequence. From PCR screening, one potential 5' end was isolated with the following sequence.
<UP>MWISHQQGLSRFSTGTMK∧SSVHEPKMDAL...</UP>
<UP>S<SC>equence</SC></UP> 1
This sequence contains 18 unique amino acids followed by downstream residues (residues following the and ), which are identical to the previously described mbr8 (1). In addition, the RACE (rapid amplification of 5' ends) method was used to isolate unknown 5' ends. The RACE procedure revealed the same N-terminal sequence just described, in three independent trials. In contrast, a complete 5' end sequence contiguous with the 5' end of clone PC8 was never identified by this method. Clone PC8 contained unique N-terminal sequence, but no initiator methionine was identified. The 5' end of PC8 is very similar to the mouse mbr17 clone (1), as follows.

Although N-terminal variation has also been obtained from mouse cDNA libraries (1), genomic Slo nucleic acid sequences do not reveal a mechanism by which such N-terminal diversity might be generated, because intron consensus splice junctions are not present in this region (3).

Tissue Distribution of the 59-Amino Acid Insert

The tissue distribution of the alternative splice variant encoding the 59-amino acid peptide was examined by splice-site specific RT-PCR (Table III). RT-PCR was used to amplify mRNAs from various tissues using sense and antisense primers to regions flanking this splice junction. Resulting PCR products were separated on 2% agarose gels. An approximately 442-base pair fragment, which encoded the 59-amino acid exon, was abundant in PC12, chromaffin cells, pancreas, cerebellum, and pituitary. It was present but less abundant in hippocampus. Spinal cord, smooth muscle, and skeletal muscle failed to reveal an exon of comparable size. The smaller and larger splice fragments were detected in relatively equal abundance in chromaffin cells. Partial sequencing of the 442-base pair band obtained from hippocampus revealed complete identity with the PC12/chromaffin cell exon. The exon is similar, but not identical, to an exon also present in human pancreatic islets (24). PCR fragments of other sizes were not detected in these tissues, indicating that only two alternative forms may be expressed in these tissues.

Table III. Distribution of site 2 splice variants


Tissue 266 bpa 350 bpb 443 bpc

PC12 + 0 +d
Adrenal chromaffin cells + 0 +d
GH3 cells + 0 +
Pituitary gland + 0 +
Pancreas + 0 +d
Hippocampus + 0 +d
Cerebellum + 0 +
Spinal cord + 0 0
Smooth muscle + 0 0
Skeletal muscle + + 0

a Product would probably contain an insert of "IYF" or no amino residues.
b Product is likely to contain the insert: KVEARARYLKDPFMHKNATPNSPHVPKPV (1).
c Product is likely to contain the 59-amino acid insert (Fig. 1B).
d Partial sequencing has confirmed identity of product with the cysteine-rich insert.
e +, present; 0, absent.

Functional Properties of Slo Variants Found in PC12 and Chromaffin Cells

Functional Properties Conferred by Site 2 and Site 5 Alternative Exons

cDNAs containing sequence encoding the 59-amino acid insert were studied in a rat Slo background (mSlo and rat Slo peptides are identical; see "Materials and Methods") and also in association with the alternative N-terminal domains from PC12/chromaffin cell cDNA. Currents resulting from expression of various Slo constructs were studied in excised inside-out patches from Xenopus oocytes. Patch currents resulting from Slo channels containing the unique insert (Slo-zf; Fig. 2A) or from mSlo/rSlo (Fig. 2B) were elicited by voltage-steps in the presence of 1-100 µM Ca2+. In comparison to mSlo currents, channels containing the 59-amino acid domain were activated by more negative voltages at comparable [Ca2+]i. Specifically, the presence of the exon produces a -20 to -30 mV shift in the conductance-voltage relation, suggesting an increase in apparent Ca2+ sensitivity when compared with wild-type mSlo at a given membrane potential (Fig. 2, B and C). V50 values for activation of currents resulting from the expression vector containing both the 59-amino acid insert and the 25-amino acid N-terminal sequence (MWISH-slo-zf) were indistinguishable from those without the N-terminal sequence (Fig. 2C). Slo channels containing the 59 amino acid insert at splice junction 2 were also comparable to mSlo channels in terms of the ability of Mg2+ to shift activation to more negative voltages (data not shown). Although Mg2+ itself is unable to activate Slo channels, the addition of 10 mM Mg2+ to a solution containing 100 µM Ca2+ results in an approximately -60-mV shift in the V50 for activation, which is comparable to the shift caused by Mg2+ on mSlo channels (25).


Fig. 2. The conductance-voltage curves are shifted leftward for the Slo-zf variant relative to mSlo. In A, currents were elicited in inside-out patches from Xenopus oocytes by voltage steps to +60 mV from a holding potential of -40 mV for 0, 1, 4, 10, 30, and 100 µM cytosolic [Ca2+]. Currents on the left resulted from expression of a Slo-zf construct in the oocytes, while currents on the right resulted from mSlo expression. In B, the normalized conductance-voltage curves for 1, 10 and 100 µM Ca2+ are plotted from expression of mSlo (filled symbols; mean ± S.E. for four patches) and from expression of Slo-zf (mean ± S.E. for four patches). Conductances were calculated from peak current during 50-ms voltage-steps using 0 mV for the potassium equilibrium potential. The lines show the best fit of a single Boltzmann function (see "Materials and Methods") to each set of points. In C, the voltages of half-activation (V50 values) for conductance with different [Ca2+] are shown for mSlo, Slo-zf, and a Slo-zf construct also containing the extended N terminus described in the text. Both constructs containing the cysteine-rich region exhibit about a 20-30-mV greater apparent sensitivity to Ca2+.
[View Larger Version of this Image (32K GIF file)]

The functional consequences of the unique C-terminal exon found in chromaffin cell Slo homologs were also examined both with (Slo-zf-3') and without (Slo-3') the 59-amino acid insert. Currents were activated with 10 µM submembrane Ca2+ over a range of voltages. Neither Slo-zf-3' (16 of 16 patches) nor Slo-3' (5 of 5 patches) constructs result in inactivating currents (e.g. Fig. 3). The V50 for activation of Slo-zf-3' at 10 µM Ca2+ was -7.7 + 1.3 mV (2 patches; mean ± 90% confidence limit of fit), while for the V50 for Slo-3' was 42.8 ± 2.1 (4 patches). These values are similar to those for constructs containing the mSlo C terminus (e.g. Fig. 2C). Thus, the chromaffin cell 3' terminal does not appear to influence the activation properties of mSlo. Furthermore, for constructs containing the chromaffin cell C terminus, currents activated by 400-ms voltage-steps to +60 mV exhibited no inactivation, indicative that the 3' addition does not influence inactivation properties of Slo.


Fig. 3. Properties of other Slo constructs. In A, currents resulting from expression in Xenopus oocytes of the Slo-zf-3' variant and the Slo-3' variant lacking the 59-amino acid insert are shown. Currents from inside-out patches were activated by voltage-steps from -80 through +200 mV at 10 µM Ca2+. Conductances were calculated as described for Fig. 2 and plotted in B. V50 values for activation for the two variants (Slo-zf-3': mean of two patches; Slo-3': mean ± S.E. from four patches) were estimated from the fit of a single Boltzmann function (solid lines). Values are given in the text. As in Fig. 2, the presence of the 59-amino acid insert results in a greater apparent sensitivity to [Ca2+] than constructs lacking this domain, whereas differences in the C terminus do not affect apparent Ca2+ sensitivity. Neither construct results in any observable inactivation.
[View Larger Version of this Image (28K GIF file)]

To examine the basis for the shift in apparent Ca2+ sensitivity of the Slo-zf constructs, the time courses of current activation and deactivation were examined at 4, 10, 30, and 100 µM Ca2+ over a range of voltages. These were reasonably well fit by single exponential functions, although, in some cases, currents could be better described by two exponential components (26). In Fig. 4A, the time constants derived from single exponential fits to the current activation time course are plotted for mSlo and the Slo-zf construct for voltages positive to 0 mV. Over this range, Slo-zf currents activate somewhat faster than mSlo currents at identical [Ca2+]. Differences in time constants of current deactivation between Slo-zf and mSlo are somewhat more pronounced. In Fig. 4B, deactivation and activation time constants for a set of patches at either 4 or 100 µM Ca2+ are plotted for both mSlo and Slo-zf. Relative to mSlo currents, Slo-zf currents activate somewhat more rapidly and deactivate clearly more slowly at a given Ca2+ and voltage. The V50 for activation corresponds approximately to that voltage where the activation time constant and deactivation time constant at a given [Ca2+] are identical. As a consequence, differences in both activation time and deactivation time between mSlo and Slo-zf contribute to the negative shift in the V50 for activation of conductance, although effects on deactivation may be somewhat more important.


Fig. 4. Properties of the Slo-zf construct. In A, time constants of current activation for inside-out patches containing either mSlo (left panel; means ± S.E. for four patches) or Slo-zf (right panel; means ± S.E. for four patches) channels are plotted as a function of voltage for 4, 10, 30, and 100 µM Ca2+. Over these voltages and [Ca2+]i, Slo-zf activates more rapidly than mSlo. In B, the time constants of activation and deactivation are plotted as a function of command potential for either 4 and 100 µM. Deactivation time constants were measured from currents resulting from repolarization to various potentials following brief, 5-10-ms activation steps to +60 mV. Deactivation time constants (solid lines) are consistently slower for the Slo-zf construct, while activation time constants (dotted lines) are somewhat faster. With 100 µM Ca2+, the shift between mSlo and Slo-zf in the voltage at which the time constant of activation is identical to the time constant of deactivation can be clearly seen. At 4 µM Ca2+, difficulties in measuring activation and deactivation between 0 and 40 mV obscure this expected relationship. In C, 20 nM CTX was applied to an outside-patch containing Slo-zf channels. Currents (see inset; vertical calibration: 1000 pA; horizontal: 12 ms) were activated by 100 µM pipette [Ca2+] with repetitive voltage steps to +60 mV. The time course of reduction of current amplitude in the presence of CTX and recovery following washout from CTX was fit as described under "Materials and Methods" to yield a Kd for CTX blockade of 2.2 nM.
[View Larger Version of this Image (38K GIF file)]

Sensitivity to CTX

The inactivating variant of BK current (BKi) found in native rat chromaffin cells is approximately an order of magnitude less sensitive to blockade by CTX than the noninactivating variant (Kd~ 1-4 nM).2 We therefore tested whether the presence of the 59-amino acid insert might alter the sensitivity of Slo channels to CTX. Currents were recorded in outside-out patches activated by voltage-steps to +60 mV with 100 µM Ca2+ in the recording pipette. The time course of onset and recovery from CTX blockade was used to define the blocking efficacy. As shown in Fig. 4C, application of 20 nM CTX leads to an almost complete blockade of Slo-zf currents. Both the onset and recovery of blockade proceed with an approximately single exponential time course; the time course of the reduction and recovery in current amplitudes resulting from steps to +60 mV can be reasonably fit by the first-order blocking model described under "Materials and Methods." The estimated Kd for blockade by CTX was 2.2 nM at +60 mV, comparable to blockade of BKs currents by CTX in native rat chromaffin cells.2 Thus, the 59-amino acid insert is unlikely to account for the reduced CTX sensitivity of inactivating BK channels in native chromaffin cells.

Is an Accessory Subunit an Obligatory Part of Chromaffin Cell BK Channels?

Recently, a gene encoding an accessory beta  subunit that is associated with the BK channel of bovine arterial smooth muscle (10) has been cloned (9). This subunit enhances the apparent Ca2+ sensitivity of cloned mSlo channels (27) and hSlo channels (3, 28). Similar to its interaction with other Slo variants, the bovine beta  subunit, when coexpressed with Slo alpha  subunits containing the 59-amino acid insert, shifted the apparent Ca2+ dependence of current activation (six of six patches). For these six patches, the V50 for activation of conductance with 10 µM Ca2+ exhibited considerable variability, ranging from -48.6 to -109 mV. This compares to a V50 for activation at 10 µM of about -10 mV for Slo-zf constructs in the absence of a beta  subunit (Figs. 2C and 3B). This negative shift shows that the 59-amino acid insert does not reduce the ability of the Slo alpha  subunit to interact with the bovine beta  subunit. Similar beta -subunit-induced shifts were also observed when the unique chromaffin cell N-terminal domain was appended to the Slo-zf sequence.

A unique feature of the beta  subunit is that it permits the triterpenoid glycoside BK channel agonist (29), DHS-I, to shift the apparent Ca2+ dependence of activation (27). Similarly, the bovine beta  subunit confers sensitivity to DHS-I on Slo-zf channels (Fig. 5, A and B). These results would argue that, if BK channels in chromaffin cells are normally associated with an homologous beta  subunit, those channels should be sensitive to DHS-I. However, when chromaffin cells are exposed to 500 nM DHS-I, there is no indication that the sensitivity of BK channels to Ca2+ is altered (Fig. 5, C and D). This lack of effect of DHS-I was observed in seven patches containing BKi channels and one patch containing exclusively BKs channels (result not shown). We did note that in some patches there was a tendency for a slight enhancement of BK current in the presence of DHS-I at more negative activation voltages. This appeared to be associated with a longer apparent burst duration once a BK channel opened. However, this effect was insignificant in comparison to effects of DHS-I on Slo channels coexpressed with the beta  subunit.


Fig. 5. The Slo-zf construct, when coexpressed with the beta  subunit, is sensitive to DHS-I, while native chromaffin cell BK currents are insensitive to DHS-I. In A, ensemble currents were generated in inside-out patches bathed with 4 µM Ca2+ from oocytes expressing Slo-zf with the beta  subunit. The voltage protocol is shown on the top. During application of 200 nM DHS-I, conductance was almost maximally activated at -40 mV (squares in A and B) and substantial current activation persists at -120 mV. In B, estimates of conductance at different voltages were made from ensembles generated as in A and at other activation voltages. Solid curves represent single Boltzmann fits with V50 values of -67.3 mV (+DHS-I) and -13.3 mV (-DHS-I), respectively. The DHS-I point at -120 mV was not used for the fit, since the duration of the step to -120 mV was insufficient to reach a steady-state current level. DHS-I resulted in a large shift in the conductance-voltage curve in four of four patches in which the Slo-zf construct was expressed with the beta  subunit. A change in maximal conductance was not a consistent feature of the results. In three patches in which the Slo-zf construct was expressed without the beta  subunit, DHS-I had no effect. In C, ensemble average currents from an inside-out chromaffin cell patch were generated with steps to the indicated voltages (-20, +20, and +40 mV on the left, and -40, -20, and +40 on the right). On the left, the patch was briefly stepped to -120 mV to partially remove inactivation prior to the activation step. On the right, currents were elicited after removal of inactivation with 0.3 mg/ml trypsin. In both cases, currents were elicited in the absence (solid line) and following application of 500 nM DHS-I (line with open circles plotted every 15 digitized values). DHS-I produces a negligible effect on BK current. In D, the voltage dependence of conductance was determined from ensemble averages generated in the presence (open symbols) and absence (closed symbols) of DHS-I both before (circles) and after (squares) removal of inactivation. Before trypsin, the V50 values were 14.8 mV (-DHS-I) and 19.6 mV (+DHS-I). After trypsin, the V50 values were -2.8 mV (-DHS-I) and -12.2 mV (+DHS-I). This patch contained nine BKi channels.
[View Larger Version of this Image (32K GIF file)]

Coupled with the observation that the Ca2+ dependence of the Slo-zf homolog alone is similar to that of BK channels in normal rat chromaffin cells (14), the lack of effect of DHS-I suggests that association of endogenous BK subunits in chromaffin cells with the known type of beta  subunit does not occur. Alternatively, if chromaffin cell BK channels contain a homologous beta  subunit as an integral portion of the channel protein, this subunit does not influence the apparent Ca2+ sensitivity of the channel and this subunit must be insensitive to DHS-I.

Finally, rat chromaffin cell RNA was screened with degenerate PCR methods by using primers based on the cloned bovine beta  subunit to test for the presence of a beta  subunit. Two positive controls were used in these experiments, the bovine BK beta  cDNA clone (9) and first strand cDNA from rat smooth muscle (stomach). Amino acid stretches with low sequence degeneracy were chosen as primer sites. The first round of PCR was carried out with a primer set that has degenerate sequences at the 3' ends. Degeneracy was 8-fold for the sense primer, and 16-fold for antisense primer. The PCR reaction was diluted 50 times. One µl was used as a template for the second round non-degenerate PCR reaction. Using the above protocol, the cDNA from rat smooth muscle and cDNA from the cloned beta  subunit (control) generated a 245-base pair fragment. However, the cDNA from adrenal chromaffin cells and a negative control reaction without cDNA template did not generate any PCR products.


DISCUSSION

The major findings of interest in this study are as follows. First, a unique alternative splice variant that defines a cysteine-rich region is found in Slo variants isolated from rat chromaffin cells and some other tissues. Second, we have described a fifth potential alternative splice site in mammalian Slo. Third, several arguments suggest that a known accessory subunit of BK channels is probably not a component of BK channels in chromaffin cells.

Slo Alternative Splice Variants and Their Relationship to Native BK Currents in Chromaffin Cells

We have identified variants of Slo channels found in cDNA and RNA from rat chromaffin cells and PC12 cells. Two splice variants, one not previously noted in any other tissues, are present at a previously described splice junction (A in the terminology of Butler et al. (1); 2 according to Tseng-Crank et al. (2)). In addition, a fifth mammalian alternative splice junction near the C terminus was also found in chromaffin cell Slo variants, and a unique C-terminal sequence was identified.

The unique splice site 2 variant encodes a peptide with a pattern of repeated cysteine residues, which as discussed below, may serve to define a secondary structure critical for interaction of this region with some other molecule. The greater apparent Ca2+ sensitivity of Slo variants containing this 59-amino acid insert in comparison to mSlo channels increases the number of alternatively spliced Slo variants that have distinct functional properties (2, 30). As observed in other studies, alternative splice variation may be important in defining the ability of particular BK channels to open as a function of Ca2+ or voltage.

Normal adult rat chromaffin cells are known to express two phenotypic variants of BK channels (8). One form exhibits rapid inactivation (7), while the second form is relatively non-inactivating and similar to BK channels described in many other cell types (8). None of the various Slo constructs expressed in the present work exhibit inactivation, and the molecular basis for inactivation of BK channels in chromaffin cells remains undetermined. However, as a result of our work, a number of factors can now be eliminated as possible elements of the inactivation mechanism. First, a novel N-terminal sequence found in rat chromaffin cells does not result in BK channel inactivation. Second, the 59-amino acid insert found in chromaffin cell Slo variants does not confer inactivation on Slo currents when expressed in Xenopus oocytes. Third, an accessory subunit (Kvbeta 1; Ref. 31), which has been shown to produce inactivation in Shaker homologs, but not other mammalian voltage-dependent channels, is also unable to produce inactivation of Slo channels.3 Fourth, the C-terminal splice variant found in chromaffin cell Slo channels does not produce inactivation. The possibility that additional Slo variants, which may affect inactivation, remain to be identified in chromaffin cells cannot be completely excluded.

The sensitivity of inactivation to cytosolic trypsin argues that the inactivating domain is probably a cytosolic peptide (7). The slowing of inactivation during trypsin digestion argues that there are multiple inactivation domains per channel (32).4 Furthermore, inactivation is unaltered and stable following patch excision. The following explanations for BK inactivation remain to be explored. First, a currently unidentified N-terminal sequence in rat chromaffin cells may be competent to produce inactivation. Since there is no genomic basis for N-terminal variation in Slo, this possibility seems unlikely. Second, an unidentified homolog of the known accessory subunits may produce inactivation. Third, an unknown type of accessory subunit may produce inactivation. Fourth, unknown post-translational modifications specific to particular Slo variants or particular expression systems may confer inactivation properties. Work in progress hopes to address these possibilities.

It remains a possibility that the 59-amino acid insert, although not conferring inactivation itself, may be correlated with one of the two phenotypic variants of BK channel found in chromaffin cells, perhaps the inactivating variant. However, none of our present data allows us to make such a correlation. Nevertheless, the presence of this exon in chromaffin cells, PC12 cells, and pancreatic beta cells, all of which express inactivating BK channels (7, 32), is consistent with the proposal that Slo subunits containing the 59-amino acid insert are integral components of BKi channels. However, contrary to this hypothesis, this alternative exon is also found in GH3 cells in which inactivating BK channels have not, as yet, been observed. Therefore, the function of this splice variant may be unrelated to the inactivation properties of BK current, but critical to some other functional role characteristic of neurosecretory cell types.

A similar 60-amino acid peptide is encoded by an alternative splice variant found in human islet cells (24). The islet cell insert contains an additional upstream arginine and differs at two amino acids within the insert. This human pancreatic islet Slo variant has not as yet been functionally expressed.

Possible Role of the Cysteine-rich Domain

Conserved patterns of cysteine residues are a common feature of many proteins and are usually indicative of the formation of peptide loops that are critical in defining key structural features of the protein. In the case of zinc-finger domains, such cysteines along with key histidine residues form coordination foci for the zinc ion (20, 21, 33, 34). Formation of the zinc-finger loops thereby defines a secondary structure critical for interaction of the zinc-finger domain with other molecules. Such zinc-binding domains are known to mediate interactions of proteins with nucleic acids (35), with other proteins (34, 36), and with phospholipids (33). The role of the cysteines in PLA2 variants is somewhat different. In this case, disulfide linkages between particular cysteines helps to orient residues that are critical for Ca2+ binding and catalytic activity (17, 19). It has been proposed that the similarities in the cysteine-rich motif in PLA2 and PKC reflect similar functional roles (37, 38). The similarity of the cysteine pattern in the Slo-zf variant both to the known zinc-finger domains of PKC and DGK and to the cysteines defining disulfide linkages in PLA2s is provocative. Thus, irrespective of whether the cysteine residues in this Slo variant form a zinc-finger domain or participate in disulfide linkages, it seems likely that this region defines peptide loops anchored by the cysteines. A common feature of this region in PKC variants, DGK, and PLA2 is the ability to bind a metal ion and also a phospholipid. At present, we have no evidence that would allow us to infer whether a metal ion remains tightly associated with this region in the Slo-zf variant. However, as a guide to future investigation, we propose that this region defines a metal binding site, which may serve to mediate interactions with some other molecule, perhaps an unknown type of accessory subunit.

Accessory Subunits in Chromaffin Cell BK Channel Function

Previous work indicates that Slo peptides in bovine smooth muscle are associated with an accessory beta  subunit (10). The ability of this subunit to increase the apparent Ca2+ sensitivity of cloned Slo variants (3, 27, 28) has led to the view that the relatively greater sensitivity of some smooth muscle BK channels to Ca2+ relative to cloned Slo channels may result from the association of Slo peptides and beta  subunits. Furthermore, the coexpression of the Slo peptide and beta  subunits also confers a sensitivity to the diterpene, DHS-I (27).

In contrast to the situation in smooth muscle, the apparent Ca2+ dependence of cloned chromaffin cell Slo channels is similar to that previously described for native BK channels in chromaffin cells (8, 14). Although this could be a fortuitous consequence of the two different membrane environments, more simply it would suggest that an accessory beta  subunit is not a required component of chromaffin cell BK channels. Furthermore, although the cloned chromaffin cell Slo variants can interact with the bovine beta  subunit in a fashion similar to other Slo variants, native BK channels in chromaffin cells are insensitive to DHS-I. These experiments lead to either of two conclusions. Either a related beta  subunit is not an obligatory component of functional BK channels in chromaffin cells or, alternatively, there is a different beta  subunit variant in the chromaffin cells which does not confer sensitivity to DHS-I and which does not significantly alter the apparent Ca2+-sensitivity of native chromaffin cell BK channels relative to the sensitivity of the cloned channels when expressed in oocytes. Although we cannot exclude the possibility that unknown accessory subunits may interact with chromaffin cell Slo peptides, the results argue that a beta  subunit, similar to that found in smooth muscle, is not a required component of BK channels in all cells.


FOOTNOTES

*   This work was supported by Grants DK46564 (to C. J. L.) and NS24785 (to L. S.) from the National Institutes of Health and a Muscular Dystrophy Association grant (to L. S.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§   To whom correspondence should be addressed. Tel.: 314-362-8558; Fax: 314-362-8571; E-mail: clingle{at}morpheus.wustl.edu.
1   The abbreviations used are: BK, big K+ channel; [Ca2+]i, intracellular calcium; HEDTA, N-hydroxyethylenediaminetriacetic acid; CTX, charybdotoxin; DHS-I, dehydrosoyasaponin-I; PLA2, phospholipase A2; PKC, protein kinase C; DGK, diacylglycerol kinase; PCR, polymerase chain reaction; RT-PCR, reverse transcriptase-PCR; RACE, rapid amplification of 5' ends.
2   J. P. Ding and C. J. Lingle, manuscript in preparation.
3   M. Saito, unpublished results.
4   C. R. Solaro, J. P. Ding, Z. W. Li, and C. J. Lingle, manuscript in preparation.

REFERENCES

  1. Butler, A., Tsunoda, S., McCobb, D. P., Wei, A., and Salkoff, L. (1993) Science 261, 221-224 [Medline] [Order article via Infotrieve]
  2. Tseng-Crank, J., Foster, C. D., Krause, J. D., Mertz, R., Godinot, N., DiChiara, T. J., and Reinhart, P. H. (1994) Neuron 13, 1315-1330 [Medline] [Order article via Infotrieve]
  3. McCobb, D. P., Fowler, N. L., Featherstone, T., Lingle, C. J., Saito, M., Krause, J. E., and Salkoff, L. B. (1995) Am. J. Physiol. 269, H767-H777 [Abstract/Free Full Text]
  4. Atkinson, N. S., Robertson, G. A., and Ganetsky, B. (1991) Science 253, 551-555 [Medline] [Order article via Infotrieve]
  5. Adelman, J. P., Shen, K.-Z., Kavanaugh, M., Warren, R. A., Wu, Y.-N., Lagrutta, A., Bond, C. T., and North, R. A. (1992) Neuron 9, 209-216 [Medline] [Order article via Infotrieve]
  6. McManus, O. (1991) J. Bioenerg. Biomembr. 23, 537-560 [Medline] [Order article via Infotrieve]
  7. Solaro, C. R., and Lingle, C. J. (1992) Science 257, 1694-1698 [Medline] [Order article via Infotrieve]
  8. Solaro, C. R., Prakriya, M., Ding, J. P., and Lingle, C. J. (1995) J. Neurosci. 16, 6110-6123
  9. Knaus, H.-G., Folander, K., Garcia-Calvo, M., Garcia, M. L., Kaczorowski, G. J., Smith, M., and Swanson, R. (1994) J. Biol. Chem. 269, 17274-17278 [Abstract/Free Full Text]
  10. Garcia-Calvo, M., Knaus, H.-G., McManus, O. B., Giangiacomo, K. M., Kaczorowski, G. J., and Garcia, M. L. (1994) J. Biol. Chem. 269, 676-682 [Abstract/Free Full Text]
  11. Frohman, M. A. (1990) in PCR Protocols: A Guide to Methods and Application (Innis, M. A., Gelfand, D. H., Sninsky, J. J., and White, T. J., eds), pp. 28-38, Academic Press, San Diego
  12. Neely, A., and Lingle, C. J. (1991) J. Physiol. 453, 97-131 [Abstract]
  13. Herrington, J., Solaro, C. R., Neely, A., and Lingle, C. J. (1995) J. Physiol. 485, 297-318 [Abstract]
  14. Prakriya, M., Solaro, C. R., and Lingle, C. J. (1996) J. Neurosci. 16, 4344-4359 [Abstract/Free Full Text]
  15. Wei, A., Solaro, C., Lingle, C., and Salkoff, L. (1994) Neuron 13, 671-681 [Medline] [Order article via Infotrieve]
  16. Dijkstra, B. W., Drenth, J., and Kalk, K. H. (1981) Nature 289, 604-606 [Medline] [Order article via Infotrieve]
  17. White, S. P., Scott, D. L., Otwinowski, Z., Gelb, M. H., and Sigler, P. B. (1990) Science 250, 1560-1563 [Medline] [Order article via Infotrieve]
  18. Kelley, M. J., Crowl, R. M., and Dennis, E. A. (1992) Biochim. Biophys. Acta 1118, 107-115 [Medline] [Order article via Infotrieve]
  19. Scott, D. L., Otwinowski, Z., Gelb, M. H., and Sigler, P. B. (1990) Science 250, 1563-1566 [Medline] [Order article via Infotrieve]
  20. Kazanietz, M. G., Bustelo, X. R., Barbacid, M., Kolch, W., Mischak, H., Wong, G., Pettit, G. R., Bruns, J. D., and Blumberg, P. M. (1994) J. Biol. Chem. 269, 11590-11594 [Abstract/Free Full Text]
  21. Goto, K., Funayama, M., and Kondo, H. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 13042-13046 [Abstract/Free Full Text]
  22. Pallanck, L., and Ganetsky, B. (1994) Hum. Mol. Genet. 3, 1239-1243 [Abstract]
  23. Meera, P., Wallner, M., Jiang, Z., and Toro, L. (1996) FEBS Lett. 382, 84-88 [CrossRef][Medline] [Order article via Infotrieve]
  24. Ferrer, J., Wasson, J., Salkoff, L., and Permutt, M. A. (1996) Diabetologia 39, 891-898 [CrossRef][Medline] [Order article via Infotrieve]
  25. Solaro, C. R., Nelson, C., Wei, A., Salkoff, L., and Lingle, C. J. (1995) Biophys. J. 68, A30 (abstr.)
  26. DiChiara, T. J., and Reinhart, R. H. (1995) J. Physiol. 489, 403-418 [Abstract]
  27. McManus, O. B., Helms, L. M. H., Pallanck, L., Ganetzky, B., Swanson, R., and Leonard, R. J. (1994) Neuron 14, 645-650
  28. Wallner, M., Meera, P., Ottolia, M., Kaczorowski, R., Latorre, R., Garcia, M., Stefani, E., and Toro, L. (1995) Recept. Channels 3, 185-199 [Medline] [Order article via Infotrieve]
  29. McManus, O. B., Harris, G. H., Giangiacomo, K. M., Feigenbaum, P., Reuben, J. P., Addy, M. E., Burka, J. F., Kaczorowski, G. J., and Garcia, M. L. (1993) Biochemistry 32, 6128-6133 [Medline] [Order article via Infotrieve]
  30. Lagrutta, A., Shen, K.-Z., North, R. A., and Adelman, J. P. (1994) J. Biol. Chem. 269, 20347-20351 [Abstract/Free Full Text]
  31. Rettig, J., Heinemann, S. H., Wunder, F., Lorra, C., Parcej, D. N., Dolly, J. O., and Pongs, O. (1994) Nature 369, 289-294 [CrossRef][Medline] [Order article via Infotrieve]
  32. Lingle, C. J., Solaro, C. R., Prakriya, M., and Ding, J. P. (1996) in Ion Channels (Narahashi, T., ed), Vol. 4, pp. 261-301, Plenum Press, New York [Medline] [Order article via Infotrieve]
  33. Nishizuka, Y. (1992) Science 248, 607-614
  34. Diaz-Meco, M. T., Municio, M. M., Sanchez, P., Lozano, J., and Moscat, J. (1996) Mol. Cell. Biol. 16, 105-114 [Abstract]
  35. Evans, R. M. (1988) Science 240, 889-895 [Medline] [Order article via Infotrieve]
  36. Feuerstein, R., Wang, X., Song, D., Cooke, N. E., and Liebhaber, S. A. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 10655-10659 [Abstract/Free Full Text]
  37. Maraganore, J. M. (1987) Trends Biochem. Sci. 12, 176-177
  38. Redman, C., Lefevre, J., and MacDonald, M. L. (1995) Biochem. Pharmacol. 50, 235-241 [CrossRef][Medline] [Order article via Infotrieve]

©1997 by The American Society for Biochemistry and Molecular Biology, Inc.