Profiling the Phospho-status of the BKCa Channel α Subunit in Rat Brain Reveals Unexpected Patterns and Complexity*S

Molecular diversity of ion channel structure and function underlies variability in electrical signaling in nerve, muscle, and non-excitable cells. Protein phosphorylation and alternative splicing of pre-mRNA are two important mechanisms to generate structural and functional diversity of ion channels. However, systematic mass spectrometric analyses of in vivo phosphorylation and splice variants of ion channels in native tissues are largely lacking. Mammalian large-conductance calcium-activated potassium (BKCa) channels are tetramers of α subunits (BKα) either alone or together with β subunits, exhibit exceptionally large single channel conductance, and are dually activated by membrane depolarization and intracellular Ca2+. The cytoplasmic C terminus of BKα is subjected to extensive pre-mRNA splicing and, as predicted by several algorithms, offers numerous phospho-acceptor amino acids. Here we use nanoflow liquid chromatography tandem mass spectrometry on BKCa channels affinity-purified from rat brain to analyze in vivo BKα phosphorylation and splicing. We found 7 splice variations and identified as many as 30 Ser/Thr in vivo phosphorylation sites; most of which were not predicted by commonly used algorithms. Of the identified phosphosites 23 are located in the C terminus, four were found on splice insertions. Electrophysiological analyses of phospho- and dephosphomimetic mutants transiently expressed in HEK-293 cells suggest that phosphorylation of BKα differentially modulates the voltage- and Ca2+-dependence of channel activation. These results demonstrate that the pore-forming subunit of BKCa channels is extensively phosphorylated in the mammalian brain providing a molecular basis for the regulation of firing pattern and excitability through dynamic modification of BKα structure and function.

Molecular diversity of ion channel structure and function underlies variability in electrical signaling in nerve, muscle, and non-excitable cells. Protein phosphorylation and alternative splicing of pre-mRNA are two important mechanisms to generate structural and functional diversity of ion channels. However, systematic mass spectrometric analyses of in vivo phosphorylation and splice variants of ion channels in native tissues are largely lacking. Mammalian large-conductance calcium-activated potassium (BK Ca ) channels are tetramers of ␣ subunits (BK␣) either alone or together with ␤ subunits, exhibit exceptionally large single channel conductance, and are dually activated by membrane depolarization and intracellular Ca 2؉ . The cytoplasmic C terminus of BK␣ is subjected to extensive pre-mRNA splicing and, as predicted by several algorithms, offers numerous phospho-acceptor amino acids. Here we use nanoflow liquid chromatography tandem mass spectrometry on BK Ca channels affinity-purified from rat brain to analyze in vivo BK␣ phosphorylation and splicing. We found 7 splice variations and identified as many as 30 Ser/Thr in vivo phosphorylation sites; most of which were not predicted by commonly used algorithms. Of the identified phosphosites 23 are located in the C terminus, four were found on splice insertions. Electrophysiological analyses of phosphoand dephosphomimetic mutants transiently expressed in HEK-293 cells suggest that phosphorylation of BK␣ differentially modulates the voltage-and Ca 2؉ -dependence of channel activation. These results demonstrate that the pore-forming subunit of BK Ca channels is extensively phosphorylated in the mammalian brain providing a molecular basis for the regulation of firing pat-tern and excitability through dynamic modification of BK␣ structure and function. Molecular & Cellular Proteomics 7:2188 -2198, 2008.
Ion channels are membrane proteins responsible for electrical signaling in nerve, muscle, and non-excitable cells (1). The diversity in electrical properties of different cell types, or of the same cell type at different developmental stages or physiological conditions, is defined not only by expression and subunit composition of distinct ion channels, but also by posttranscriptional and posttranslational modifications of their component subunits (1,2). Alternative splicing of pre-mRNA to yield changes in primary structure and protein phosphorylation to alter folding and charge are fundamentally important mechanisms to generate structural and functional diversity of ion channel proteins (3)(4)(5)(6)(7)(8). To date, however, systematic investigations of in vivo phosphorylation and splicing by direct analyses of ion channel proteins are largely lacking (7).
Mammalian BK Ca (large-conductance calcium-activated potassium, also termed K Ca 1.1, Maxi-K, or Slo1) 1 channels are unique potassium-selective channels that are dually activated by two independent physiological signals: intracellular Ca 2ϩ concentration and transmembrane voltage (9 -12); thereby playing a powerful integrative role in the regulation of electrical excitability through coupling of calcium signaling and cellular excitability. In mammalian central neurons, BK Ca channels underlie the repolarization and fast after-hyperpolarization of action potentials (13,14), shape dendritic Ca 2ϩ spikes (15), and control neurotransmitter release at presynaptic terminals (16,17). BK Ca channels also play key roles in other diverse physiological processes such as contractile tone of various types of smooth muscle cells, frequency tuning of auditory hair cells, hormone secretion, and innate immunity (9 -11).
BK Ca channels are robustly expressed in central neurons throughout most regions of the mammalian brain (18). Unlike many mammalian potassium channels, a single gene (Slo1, KCNMA1) encodes all BK Ca channel ␣ subunits (BK␣). However, native BK Ca channels display a broad range of functional properties that differ between different cells (9,10), at different stages of development (19,20), and under different physiological conditions (21). In addition to assembly with tissuespecific auxiliary ␤ subunits (␤1-␤4), diversity in the physiological properties of BK Ca channels can be generated by extensive pre-mRNA splicing and phosphorylation of BK␣.
BK Ca channels are tetramers of the pore-forming, voltage-, and Ca 2ϩ -sensing BK␣ either alone or in association with regulatory ␤ subunits. BK␣s are 125-140 kDa polypeptides containing seven transmembrane segments (S0 -S6), a short extracellular N-terminal domain, and a large cytoplasmic C terminus (22) (Fig. 1). This C-terminal domain comprises Ͼ70% of the total protein and contains four hydrophobic segments (S7-S10) and sequences similar to RCK (Regulating Conductance of K ϩ ) domains (23)(24)(25), and a string of Asp residues known as the "Ca 2ϩ bowl" (26).
BK␣ contains ϳ200 serine and threonine residues that can be potentially phosphorylated by cellular protein kinases (supplemental material). Modulation of native and cloned BK Ca channels by protein kinases is well established (27)(28)(29)(30). In smooth muscle, protein kinases PKA, protein kinase G, and protein kinase C play an important role in BK channel-mediated regulation of contractility (reviewed in Ref. 8). In central neurons, both enhancement and inhibition of BK Ca channel activity by PKA have been observed (28,31,32). However evidence for direct phosphorylation and a systematic analysis of BK Ca channel phosphorylation in native tissue are lacking. Alternative splicing has been extensively studied at the mRNA level in different species and tissues, and more than 20 BK␣ splice variants have been identified (2,33,34). However, no information on expression of the polypeptide products of these alternatively spliced mRNAs is yet available at the protein level.
To help understand the molecular structure and diversity of BK Ca channels in the central nervous system, we have immunopurified BK␣ from rat brain and taken an unbiased approach using nanoflow tandem mass spectrometry (nano-LC MS/MS) to systematically analyze their in vivo phosphorylation sites and splice variants. We have identified extensive in vivo phosphorylation of native BK␣ and examined the functional consequence of phosphorylation at a number of identified sites by electrophysiological analyses of phospho-and dephosphomimetic BK␣ mutants transiently expressed in HEK-293 cells.

EXPERIMENTAL PROCEDURES
Affinity Purification of BK Ca Channels from Rat Brain-Two different sets of affinity purifications of BK Ca channels were used in this study. One set employed monoclonal antibody-based immunopurification from rat brain membranes prepared from freshly isolated adult whole rat brains as described (35) and solubilized by 1% Triton X-100 or 1% dodecyl-maltoside. BK␣ was affinity-purified using the monoclonal antibody L6/60 (termed anti-BK␣_1) (36,37) immobilized on protein G agarose beads. L6/60 binds within amino acid residues 729 -930 of mouse BK␣ (UniProt/Swiss Prot accession number Q08460). Phosphatase inhibitors (10 mM NaF, 1 mM sodium orthovanadate, 5 mM sodium pyrophosphate) and protease inhibitors were used throughout the procedure. The other set of affinity purifications used two polyclonal antibodies with plasma membrane-enriched protein fractions prepared from adult rat brains as described previously (38,39). Almost complete solubilization (Ͼ95% as judged from densitometric estimates of Western-probed solubilisate versus pellet) was achieved with 1-1.25 ml of ComplexioLyte 48 (including protease inhibitors; Logopharm GmbH, Freiburg, Germany) per mg membrane protein.
1.5 ml of solubilisate was incubated with 20 g immobilized rabbit polyclonal antibodies raised either against amino acid residues 1184 -1203 (termed anti-BK␣_2, gift from Dr. Hans-Guenther Knaus, University of Innsbruck) or 1184 -1200 (termed anti-BK␣_3, Alomone Labs, Jerusalem, Israel) of the mouse BK␣ (UniProt accession number Q08460). Bound proteins were eluted with Laemmli buffer (dithiothreitol added after elution). Aliquots from each step were analyzed on Western blots with equivalent fractions (as in Fig. 1B, lanes 1, 2 of each affinity purification and supplemental Fig. S4) to ensure quantitative purification and recovery of intact BK␣ protein.
In-gel Digestion and Enrichment of Phosphopeptides-Affinity-purified proteins were separated by SDS-PAGE and either silver-stained (in the absence of cross-linkers) or visualized with Coomassie G-250. Bands containing BK␣ (as validated by Western analysis) as well as complete lanes were separately excised and washed thoroughly with 50% acetonitrile in 25 mM ammonium bicarbonate. In-gel digestion was carried out essentially as described by (40). After reduction and alkylation of Cys residues using dithiothreitol and iodoacetamide, gel pieces were washed, dehydrated, and subsequently swollen with ammonium bicarbonate buffer containing 10 ng/l trypsin (Promega, Madison, WI) and incubated for ϳ16 h at 37°C. Digested peptide mixtures were extracted, dried in a speed vacuum concentrator, and finally redissolved in 0.2-0.5% formic acid (protein samples obtained from anti-BK␣_1 eluates) or trifluoroacetic acid (protein samples obtained from anti-BK␣_2 and anti-BK␣_3 eluates).
The peptide mixtures were either directly used for MS analyses or further processed to enrich the phosphopeptides. Phosphopeptide enrichment and tandem mass spectrometric analysis were performed essentially as described previously (41,42) with a few modifications. Briefly, a slurry of titanium dioxide beads precoated with 2,5-dihydrobenzoic acid was prepared by mixing 10 g of titansphere TiO 2 beads (GL Sciences) with 20 l of 30 mg/ml 2,5-dihydrobenzoic acid (Fluka) in 80% acetonitrile. 5 l of this 2,5-dihydrobenzoic acid/TiO 2 slurry was added to the acidified peptide mixtures, extracted from in-gel digests. The peptide mixtures were shaken for 30 min at 1000 rpm at 4°C, and then spun down in a microcentrifuge. The pelleted TiO 2 beads were washed twice with 30% acetonitrile in 3% trifluoroacetic acid, and peptides were eluted with 15% NH 4 OH in 40% acetonitrile (pH Ͼ 10.5). Finally, the eluates were dried in a speed vacuum concentrator for 20 min at 45°C and reconstituted in 8 l of 2% acetonitrile in 0.1% trifluoroacetic acid to prepare them for LC-MS.
Mass Spectrometry, Data Processing, and Analysis-BK␣ affinitypurified with the monoclonal antibody anti-BK␣_1 was analyzed with an LTQ ion trap or LTQ-FT hybrid mass spectrometer (Thermo-Fisher, San Jose, CA) connected to a Waters UltraPerformance LC system (Waters, Milford, MA). Peptide samples were concentrated on a Waters Symmetry C18 280 m ϫ 20 mm nanoAcquity trap column at a loading flow rate of 15 l/min. Peptides were then eluted from the trap and separated by a Waters 100 m ϫ 100 mm UltraPerformance LC column using a 90 min gradient of 2-80% buffer B (buffer A, 0.1% formic acid, buffer B, 95% acetonitrile, 0.1% formic acid), and sprayed into an LTQ or LTQ-FT ion trap mass spectrometer through a nanoelectrospray source. The MS survey scan was acquired using the LTQ or Fourier transform ion cyclotron resonance mass analyzer and then the top four ions in each survey scan were subjected to automatic low energy collision-induced dissociation for MS/MS scans.
MS/MS spectra were extracted using the program Extract_msn v.4.0 of the Bioworks software v.3.3 (Thermo Finnigan) with default parameters and interpreted with Mascot v.2.2 (Matrix Science, London, UK) and Sequest/Bioworks v.3.3 search engines by searching against UniProt/Swiss-Prot database (release 50.8, subset Rodents, 18284 protein entries) supplemented with all known and artificially spliced variants of rat BK␣ (rSlo), based on the rat genomic sequence and known splicing variants from mRNA or expressed sequence tag sequence of other mammalian species. Database searches were performed with a peptide mass tolerance of 2 Da (LTQ) or 20 ppm (LTQ-FT), MS/MS tolerance of 0.4 Da, and strict tryptic specificity (cleavage after lysine and arginine) allowing one missed cleavage site; carbamidomethylation of Cys was set as fixed modification, whereas oxidation (Met), N-acetylation, N-pyroglutamine formation, and phosphorylation (Ser, Thr, Tyr) were considered as variable modifications. Peptide ions with a Mascot ion score of Ͻ20 were manually checked for validation. For database search with Sequest/Bioworks, peptide sequences from the search result were filtered out with criteria of correlation value (Xcorr) Ͼ1.5, 2.0, and 3.0 for singly, doubly, and triply charged peptide ions, respectively, and ⌬Cn (difference in correlation with the next higher Xcorr) Ͼ0.08. Each MS/MS spectrum exhibiting possible phosphorylation was manually checked and validated based on the existence of a 98 Da mass loss (H 3 PO 4 ; phosphopeptide-specific collision-induced dissociation neutral loss) for both precursor and fragmented ions.
The peptide mixtures from BK␣ affinity-purified with the polyclonal antibodies anti-BK␣_2 and anti-BK␣_3 were separated by online high pressure liquid chromatography and directly electrosprayed into an LTQ-Orbitrap hybrid mass spectrometer as described (41). The instrument was operated in data-dependent mode to automatically switch between full scan MS and MS/MS acquisition. All full scans were acquired with a resolution of 60,000 at m/z ϭ 400 by the Orbitrap detector system using automatic internal lock-mass recalibration in real-time (43). From each full scan up to five peptide ions with charge states Ն2 were selected for fragmentation by multi-stage activation (multistage activation or pseudo MS 3 ) (44). All fragment ion spectra were recorded with the LTQ detectors.
MS/MS peak lists were extracted from the raw MS files by in-house written software Raw2msm v.1.10 (43) using default parameters (intensity-weighing the parent ion m/z over the LC elution profile and keeping top 6 most intense fragment ions per 100 m/z units), and searched by Mascot v.2.2 against a concatenated forward (target) and reversed (decoy) version of the IPI rat database v.3.25 supplemented with standard observed contaminants such as porcine trypsin and human keratins (total number of protein sequences: 82886). Carbamidomethylation of cysteine residues was set as a fixed modification, whereas oxidation (Met), N-pyroglutamine formation, and phosphorylation (Ser, Thr, Tyr) were considered as variable modifications. Full tryptic specificity (cleavage after lysine and arginine) was required and up to three missed cleavages were allowed. The initial mass tolerance in MS mode was set to 5 ppm and MS/MS mass tolerance was 0.5 Da. The resulting Mascot html-output file was linked to the raw MS files and loaded into the MSQuant software only accepting the highest-scoring peptide sequence for each MS/MS spectrum and requiring a minimum Mascot score of at least 10. To minimize the false discovery rate all peptide identifications were filtered by thresholds on peptide length, mass error, and Mascot score. We fixed the thresholds and accepted peptides based on the criteria that the number of forward hits were at least 200-fold higher than the number of reversed hits, which gives an estimated false discovery rate of less than 1% (p Ͻ 0.01). To pinpoint the actual phosphorylated amino acid residue within the identified phosphopeptides in an unbiased manner, we calculated the localization probabilities of all Ser and Thr phosphorylation sites using the post-translational modification score algorithm as described (41).
Electrophysiological Recordings and Data Analysis-All recordings were performed at room temperature in the inside-out patch-clamp configuration. Data were acquired with an Axopatch 200A patchclamp amplifier (Axon Instruments, Inc.) in resistive feedback mode and were low pass filtered at 10 kHz with its 4-pole Bessel filter. An ITC-16 hardware interface (Instrutech) and Pulse acquisition software (HEKA Elektronik) were used to sample the records at 20-s intervals. Capacitive transients and leak currents were subtracted by a P/5 protocol at holding potentials of Ϫ120 mV or Ϫ150 mV (for recordings with 103 M intracellular Ca 2ϩ ). Four to eight consecutive current series were averaged to increase the signal to noise ratio. Normalized conductance-voltage relationships were fitted with single Boltzmann functions:  40 M fresh (ϩ)-18-crown-6-tetracarboxylic acid (18C6TA) was added to the internal solution just before recording. Inside-out patches were continuously perfused with internal solution using a sewer pipe flow system (DAD-12; Adams and List Assoc., Ltd.); computer-controlled switches allowed for complete solution exchange at excised patches in Ͻ1 s.

MS Analyses of Rat Brain BK␣: Primary Sequence Coverage and Splice
Variations-Two independent proteomic analyses of BK␣ affinity-purified from membrane preparations of total rat brain form the basis of this study. One set of analyses used BK␣ immunopurified with a BK␣-specific mouse monoclonal antibody binding near the S9 region (anti-BK␣_1) (37), the other used BK␣ immunopurified with two different rabbit polyclonal antibodies (anti-BK␣_2, anti-BK␣_3) recognizing short peptide sequences within the C-terminal tail region (38). All three antibodies were highly effective in affinity purifications fully depleting the source material of BK␣ (Fig. 1). After separation on SDS gels, BK␣ proteins were in-gel digested with trypsin, and the resultant tryptic peptides were analyzed by nanoflow liquid chromatography tandem MS (nano-LC MS/ MS) using a linear ion trap or high resolution hybrid mass spectrometers (LTQ, LTQ-FT, or LTQ-Orbitrap).
Based on the high yield of our affinity purifications, mass spectrometry retrieved Ն60 BK␣-specific peptides, which is almost the complete set obtainable under our experimental conditions. Together, these peptide fragments cover ϳ70% of the BK␣ amino acid sequence as encoded by the 27 exons considered constitutive (34) (Fig. 2A). The non-covered regions are hydrophobic segments of the transmembrane core, and a short segment of the cytoplasmic C terminus for which no peptides could be detected because of the unfavorable mass of the resultant tryptic peptides (molecular weight of detectable peptide fragments between 740 and 3000). In addition, MS analysis retrieved a number of peptides that correspond to entries in a custom database of BK␣ splice variants, and demonstrate expression of these isoforms of BK␣ in the rat brain (supplemental Table S2). Fig. 2B illustrates the amino acid sequences of the peptide fragments that define these variations from the constitutive BK␣ primary sequence. Together, MS analyses identified seven insertions at three distinct splice sites in the C terminus of BK␣ as well as an extension at the N terminus. At the first site, insertions of either three (IYF, one letter code) or 61 amino acids (termed Strex) (46) were found (corresponding to the alternative usage of exons 22 or 23). At the second site, immediately proximal to a domain termed "Ca 2ϩ bowl", an additional stretch of 27 residues encoded by exon 29 (AKPGKLPLVSVNQEKNSGTH-ILMITEL) was identified. At the C terminus, MS analyses determined four different termini ranging from 8 to 61 residues (resulting from alternative usage of the 3Ј part of constitutive exon 33 and the non-constitutive exons 34 and 35 (34)), whereas at the N terminus it was an extension of 65 amino acids resulting from alternative starts of translation. Together, these results verified alternative splicing of BK␣ at the protein level and determined which splicing events were used in rat brain.
Extensive Phosphorylation of BK␣ in Rat Brain-Next we investigated the in vivo phosphorylation status of BK␣ by MS analyses of tryptic peptide fragments with and/or without enrichment for phosphopeptides. As illustrated in Figs. 2-4, mass spectrometry identified a total of 30 serine/threonine (Ser/Thr) phosphorylation sites on BK␣, 24 of which were found on the constitutive sequence, and six were located on splice extensions. Representative MS/MS spectra of two unique peptides for identified phosphorylation sites Ser(P)-854, Ser(P)-855, Ser(P)-859, and Thr(P)-965 are shown in location of the single phosphorylation site within the tryptic peptide could not be unambiguously determined and could be on either of the two neighboring Ser/Thr residues.

FIG. 2. In vivo phosphorylation sites and splice inserts of rat brain BK␣ identified by MS analysis.
A, amino acid sequence of the constitutive form of BK␣ (Swiss-Prot accession number Q62976 -2; last 8 residues from splice insert are not included) together with the identified phospho-Ser/Thr residues (highlighted in blue) and the splice insertion sites (marked by colored triangles). Dot-free Ser/Thr residues denote unambiguous phosphorylation sites, whereas dots underneath Ser/Thr residues mark ambiguous sites. B, amino acid sequence of the identified N-terminal extension and splice variants. Peptides identified by MS analysis are shown in red, those not identified in MS analyses are in gray. Name of each splice insert is given in bracket on the right side of the amino acid sequence. Horizontal bars denote hydrophobic segments S0 -S10.

Differential Modulation of BK Ca Channel
Gating by Phosphorylation-Next we individually mutated 16 of the identified phospho-Ser/Thr residues on BK␣ to Ala and Asp attempting to mimic phospho and dephospho states, respectively (48). These mutations were made in the hSlo-HF1 plasmid (45) encoding the constitutive human BK␣ (accession AAB65837) with the C-terminal splice insert C-ERL; hSlo1 is identical to rat BK␣ except for five residues (human versus rat: A86V, I619V, K631R, S639P, and Q1093P).
Electrophysiological properties of wild type and mutant BK Ca channels were determined in transfected HEK-293 cells under basal conditions (i.e. in the absence of any further stimulus) by patch-clamp recordings in inside-out configuration at different values for [Ca 2ϩ ] i . Similar to previous studies (45), activation of WT hSlo-HF1 channels was strongly dependent on [Ca 2ϩ ] i with half-maximal activation (V 1 ⁄2) occurring at a membrane potential of 149 Ϯ 3 mV (mean Ϯ S.E.) at Ca 2ϩ -free conditions (n ϭ 9), 97 Ϯ 2 mV at 1.1 M Ca 2ϩ (n ϭ 11), 14 Ϯ 2 mV at 10.1 M Ca 2ϩ (n ϭ 12), and Ϫ35 Ϯ 2 mV at 103 M Ca 2ϩ (n ϭ 12; Fig. 5 and Table I). Similar to WT channels, all phosphosite mutants gave rise to robust BK Ca currents, suggesting that none of the phosphosite mutations Phosphorylation Profile of the BK Ca Channel ␣ Subunit appreciably affected assembly and trafficking of channels in our heterologous expressions. Alterations in activation gating resulting from introduction of a phosphomimetic negatively charged Asp residue was observed for Ser(P)-855 ( Fig. 5B and Table I). Compared with the dephospho-mimicking S855A mutant, the V 1 ⁄2 value of S855D displayed a 28 mV shift toward hyperpolarizing potentials, from 152 Ϯ 2 mV (n ϭ 8) to 124 Ϯ 3 mV (n ϭ 9) under Ca 2ϩ -free conditions and of 24 mV, from 102 Ϯ 1 mV (n ϭ 8) to 78 Ϯ 2 mV (n ϭ 9) at 1.1 M Ca 2ϩ . Smaller shifts (13 and 15 mV) were observed at higher [Ca 2ϩ ] i (10.1 and 103 M, respectively). All other Ala/Asp mutants failed to exert significant changes on channel activation (i.e. shifts of V 1 ⁄2 were smaller than 10 mV) (Table I). However, a role of these sites in modulating channel activity in other BK␣ backbones, or in response to specific signaling events, cannot be excluded. Thus, previous studies on bovine BK␣ showed that mutating the site corresponding to Ser-869 (identified here as Ser(P)-869) eliminated the 35 mV hyperpolarizing shift in channel activation observed upon treatment of excised patches with purified PKA (49). Moreover, it should be kept in mind that aspartate mutations may fail to mimic the phosphorylated state because of fundamental differences between the carboxylate anion and the phosphate di-anion and trigonal planar versus tetrahedral geometry, respectively (50 -53). DISCUSSION This work presents the first comprehensive MS analysis on the impact of phosphorylation and alternative splicing of BK Ca channels in rat brain. As central findings, our results identify an unanticipated large number of 30 Ser/Thr residues phosphorylated under in vivo conditions and a total of seven splice variations at three distinct sites. Together, these results demonstrate an unexpected level of complexity introduced into the primary structure and properties of BK␣ by posttranslational modification and alternative splicing.
Analysis of in Vivo Phosphorylation by Mass Spectrometry-For investigation of in vivo phosphorylation of the BK␣ polypeptide, we undertook a proteomic analysis based on a combination of affinity purification of appropriately solubilized FIG. 4. Localization of identified phosphosites and splice variants on rat brain BK␣. Membrane topology of BK␣ together with localization of the identified phosphosites and splice insertions. The constitutive form of BK␣ contains the transmembrane core (S0 -S6), the pore region, the hydrophobic intracellular segments S7-S10, the Ca 2ϩ bowl, and the RCK domains (25,65,66). Insertion sites of sequence stretches generated by alternative splicing or alternative start of translation are indicated by triangles with the same color coding and names as in Fig. 2. Phosphosite Ser(P)-855 exhibiting a marked effect on channel gating is highlighted in red.  Table I). BK␣ with nano-LC MS/MS analyses (38,48). Using this approach, we identified a total of 30 Ser/Thr phosphosites, a number that by far exceeds the average number of phosphosites detected in a recent proteomic study on proteins in the soluble fraction of culture cells (41) and even that found for affinity-purified Kv2.1 (48), another ion channel particularly known for its pronounced phospho-regulation (54). In addition both number and pattern of the MS-identified phosphosites markedly differed from the results obtained with computer algorithms, which were widely used in protein/ion channel research and were based on consensus sites of specific protein kinases. As illustrated in Fig. 6, although these algorithms predicted quite a number of phosphosites on the BK␣ sequence, only a minor subset coincided with the Ser(P)/Thr(P) detected by mass spectrometry on BK Ca channels isolated from native tissue. In other words, in this case prediction algorithms yielded a significant number of sites that were not phosphorylated under physiological conditions and failed to identify those actually targeted by cellular kinases (Fig. 6B). As further shown in Fig. 6B, the number of "false positives" may be reduced by combining the results of several computer algorithms. However, this strategy comes at the cost of losing many of the de-facto phosphorylation sites (Fig. 6B). Taken together the proteomic approach used here appears largely superior in providing comprehensive phospho-analysis of ion channel proteins to conventional (consensus-site based) strategies, although it may still underestimate the actual number of phosphosites present on BK␣.
With respect to localization of the identified phosphosites in the BK␣ primary structure (22), the bulk of identified Ser(P)/ Thr(P) are clustered in three regions: the linker region between S8 and S9 near the Strex site, the region close to the Ca 2ϩ - bowl and the C-terminal tail (Fig. 4). Two of the identified sites, Ser(P)-869 and Ser(P)-1081, have been previously identified as sites targeted by protein kinases PKA and PKG, respectively (30,55). The remaining 28 phosphosites represent a de novo description. Residues Ser-1098 and Ser-1101 reported as protein kinase C sites (49) could not be verified in our MS analyses because of the low mass of the respective tryptic peptide(s). The striking extent of BK␣ phosphorylation raises questions as to its physiological role in regulating neuronal BK Ca channels. Obviously, phosphorylation at some sites is used to modify BK Ca channel gating properties, as demonstrated by the significant V 1 ⁄2 shift observed for certain phospho/dephospho-mimicking mutants, for example S855A/S855D (Fig. 5). Other properties of BK Ca biology regulated through the phosphorylation state may include association with interacting proteins, trafficking to the cell surface, turnover/endocytosis of cell surface channels, targeting to distinct subcellular compartments, etc. (11). The significance of phosphorylation de-tected for the two residues in the extracellularly located Nterminal extension remains unclear at present.
Analysis of Alternative Splicing by Mass Spectrometry-To date, expression of BK␣ splice variants has been studied only at the mRNA level (2,33,34), and there are only a few reports using mass spectrometry to systematically analyze variants of extensively spliced proteins (56). The high yield of BK␣ obtained from our affinity purifications allowed for comprehensive MS analysis of splice variations excluding, however, those variants with tryptic fragments too small for mass spectrometry, such as the four residue insert SRKR encoded by the 3Ј-truncated form of exon 19 or variants providing peptide fragments too low in mass for MS detection. By searching MS/MS spectra of immunopurified BK␣ against a custom BK␣ splice variant database, we have identified seven splicing inserts at three different C-terminal sites as well as an N-terminal extension. To our knowledge, this is the first evidence to show that the IYF splice variant and the three C-terminal variants, C-ERL, C-SSP, and C-DEC are present in the mammalian brain. Alternative inclusion and exclusion of the seven identified splice inserts at three splice sites can form as many as 24 different splicing variants with distinct biophysical properties, or that are altered in other aspects of BK Ca channel biology. For example, expression of Strex-containing BK Ca channels is under hormonal control (46), yielding channels more sensitive to hypoxia (57), and that display a significant hyperpolarizing shift (ϳ20 mV) in V 1 ⁄2 and considerably slower rates of deactivation when compared with the insertless form (58). The Ca27 splice variant yields channels with an increased activation rate and modified Ca 2ϩ cooperativity (59). The C-DEC splice variant generates an extended C terminus resulting in BK␣ with enhanced retention of newly synthesized BK Ca channels in the endoplasmic reticulum, which finally results in decreased cell surface expression (60,61). BK Ca channels are robustly expressed in mammalian central neurons (62,63) and are predominantly localized to axons and presynaptic terminals (17,37) but are also present in certain neuronal somata and dendrites (18,64). In rat brain, BK␣ splice variants may be differentially expressed in distinct cell types or different compartments of the same neuron. A region-specific distribution of the insertless counterpart of Ca27 splice variant has been observed in rat brain, mRNA is predominantly enriched in cerebellum, whereas Ca27 is abundant in all brain regions (59). * This work was supported, in whole or in part, by National Institutes of Health Grant NS34383 (to J. S. T.). This work was also supported by the Deutsche Forschungsgemeinschaft Grant SFB 746, TP16 (to B. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www. mcponline.org) contains supplemental material, supplemental Figs. S1-S4, and supplemental Tables S1-S4.
¶ ¶ To whom correspondence may be addressed: Section of Neurobiology, Physiology and Behavior, College of Biological Sciences, 196 Briggs Hall, University of California, One Shields Ave., Davis, CA 95616-8519. E-mail: jtrimmer@ucdavis.edu.