ICln159 folds into a pleckstrin homology domain-like structure. Interaction with kinases and the splicing factor LSm4.

ICln is a multifunctional protein involved in regulatory mechanisms as different as membrane ion transport and RNA splicing. The protein is water-soluble, and during regulatory volume decrease after cell swelling, it is able to migrate from the cytosol to the cell membrane. Purified, water-soluble ICln is able to insert into lipid bilayers to form ion channels. Here, we show that ICln159, a truncated ICln mutant, which is also able to form ion channels in lipid bilayers, belongs to the pleckstrin homology (PH) domain superfold family of proteins. The ICln PH domain shows unusual properties as it lacks the electrostatic surface polarization seen in classical PH domains. However, similar to many classical PH domain-containing proteins, ICln interacts with protein kinase C, and in addition, interacts with cAMP-dependent protein kinase and cGMP-dependent protein kinase type II but not cGMP-dependent protein kinase type Ibeta. A major phosphorylation site for all three kinases is Ser-45 within the ICln PH domain. Furthermore, ICln159 interacts with LSm4, a protein involved in splicing and mRNA degradation, suggesting that the ICln159 PH domain may serve as a protein-protein interaction platform.

ICln is a small (Ϸ25 kDa), multifunctional protein involved in regulatory mechanisms as different as cell volume regulation and RNA splicing (1,2). Proteins involved in more than one regulatory mechanisms were recently defined as "connector hubs" (3) and play a pivotal role in cell function. Knock-out of such proteins is usually lethal. Accordingly, functional knockout of the ICln protein in mouse or nematode is lethal in the very early stages of development. No vital embryos can be obtained in either animal system (2,4), which suggests a key role of ICln in cell homeostasis.
During a hypotonic challenge, ICln migrates from the cytosol toward the cell membrane, suggesting a role for ICln in cell volume regulation (15). This regulatory mechanism is further substantiated by the finding that heterologous expression of ICln in Xenopus laevis oocytes is followed by the appearance of ion currents (1) resembling those elicited by swelling of mammalian cells (2,16). Furthermore, ICln-specific antibodies (5) or antisense oligodeoxynucleotides (17,18) leading to a specific knock-down of the ICln protein impair regulatory volume decrease (RVD) 1 in native cells. Conversely, overexpression of ICln in mammalian cells increases RVD currents during a hypotonic challenge (2,19,20). In bacteria, overexpression of ICln leads to an improved tolerance to hypotonicity, an effect that can be reversed by the extracellular application of nucleotides (21,22). The extracellular effect of nucleotides was also shown for heterologous expression of ICln in oocytes, and the putative binding site for nucleotides was identified (1,23). The above mentioned findings point toward a channel function of ICln (1). The successful reconstitution of purified ICln in artificial lipid bilayers demonstrated that ICln can indeed form ion conducting channels, the selectivity of which depends on the lipid composition of the membrane (23)(24)(25)(26).
Another well documented function of ICln is its role in splicing. In the cytosol, ICln forms two separate complexes with splicing factors. One complex (6S) contains ICln and the Sm heteromers D1⅐D2, D3⅐B/BЈ and E⅐F⅐G. The other complex is composed of ICln, the arginine methyltransferase PRMT5, MEP-50, and Sm proteins and is termed the methylosome. The current hypothesis on the role of ICln in the splicing process is that ICln sequesters newly synthesized Sm proteins and directs these proteins to the methylosome, a protein complex that methylates Sm and possibly LSm proteins, prior to the assembly of the cytoplasmic UsnRNA core particle by the survival of motor neurons complex (6 -8).
For both functions of ICln, e.g. ion transport as well as splicing, the structure of ICln is of utmost importance to better understand, on a molecular level, the regulatory mechanisms involved. The aim of the present study was to get insight into the protein-protein interaction of ICln based on the structure of its water-soluble form, which seems to connect multiple and different regulatory modules, a concept common for connector hubs in complex protein networks (3). Here, we show the NMR solution structure of ICln 159 , which demonstrates that ICln belongs to the pleckstrin homology (PH) domain superfold family of proteins.
PH domains are highly conserved in regulatory proteins involved in intracellular signaling. The best characterized physiological role of PH domains is to recruit and tether their host proteins to the cell membrane (27). PH domains are also known as substrates for different kinases (27). Indeed, it was shown that ICln can be phosphorylated by a CKI/CKII-like kinase (28). Sequence analysis predicts that ICln could also be phosphorylated by purified protein kinase C (PKC), cAMP-dependent protein kinase (PKA), and cGMP-dependent protein kinase type I␤ or II (PKGI or PKGII). Here, we show that ICln 159 interacts with PKC, PKA and, in addition, with PKGII, but not PKGI. The finding that ICln 159 interacts with LSm proteins, as well as kinases, suggests that the ICln PH domain may serve as connector hub.

MATERIALS AND METHODS
Cloning-Starting from human (Homo sapiens) and dog (Canis canis) cDNA, the open reading frames of full-length LSm4 and ICln, as well as of the different ICln truncations, were amplified by PCR using standard protocols. The PCR products were cloned in-frame into the mammalian expression vectors pECFP-C1, pEYFP-C1, pECFP-N1, pEYFP-N1 (Clontech) to produce fusion proteins suitable for fluorescence resonance energy transfer (FRET). For protein purification, ICln and ICln truncations as well as LSm4 were cloned into the bacterial expression vector pET3-His (23). The primer pairs and restriction sites used for amplification and cloning are given in the Supplemental Material.
Protein Purification-Purification and labeling of ICln and its truncations in E. coli BL21(DE3) for NMR, interaction, and bilayer experiments are described in detail in Refs. 23, 25, and 29. For the purification of hsLSm4, bacterial lysates were cleared by centrifugation and loaded onto a cation exchange column (bed volume, 8 ml, MacroPrep High S, Bio-Rad). After washing with 1 M NaCl in 25 mM K 2 HPO 4 , 1 mM dithiothreitol, 5% glycerol, pH 7.2, bound proteins were eluted by increasing NaCl to 2 M. Fractions containing LSm4 were pooled, concentrated, and applied to a Sephacryl S100 gel filtration column (bed volume 125 ml, Bio-Rad) equilibrated with 25 mM K 2 HPO 4 , 150 mM NaCl, 1 mM dithiothreitol, 5% glycerol, pH 7.2. Fractions containing LSm4 were pooled, concentrated, and stored at Ϫ70°C. Protein-protein interaction assays were performed in gel filtration chromatography equilibration buffer at 4°C.
After the membrane was formed and its stability was assessed, ICln 159 (500 ng/ml) was added to the cis and trans chambers. The experiments were made at 28°C, in a symmetrical solution composed of 100 mM KCl, 5 mM HEPES, pH 8.00. To determine the ion selectivity of reconstituted ICln 159 , we chose a gradient for the ions employed that had a higher salt concentration in the trans chamber (150 mM KCl, 5 mM HEPES, pH 8.00) and a lower concentration in the cis chamber (10 mM KCl, 5 mM HEPES pH 8.00). The chloride gradient (⌬PDCl) was measured after each experiment and used to calculate the pK/pCl value according to the Goldman-Hodgkin-Katz equation. Reversal potentials were determined graphically by interpolation and were normalized to a gradient of 15 (23,25). NMR Analysis-NMR data were acquired at 298 K on Varian Inova 800 and 500 mHz spectrometers equipped with pulse field gradient triple resonance probes (H/N/C). The NMR sample conditions were 1 mM protein, 25 mM K 2 HPO 4 , 150 mM NaCl in 90% H 2 O, 10% 2 H 2 O, pH 7.0. NMR data were processed and analyzed with NMRPipe (30) and NMRView (31). Backbone and side-chain signal assignments as well as relaxation experiments were obtained using standard triple resonance experiments (32). Three bond nitrogen-carbon ( 3 JC␥N) and carbonylcarbon ( 3 JC␥C) couplings were measured by quantitative J methods (33,34). Three-dimensional structures were generated by simulated annealing with the XPLOR-NIH package (35). NOE distance restraints were derived from three-dimensional 15 N/ 13 C-nuclear Overhauser effect spectroscopy-heteronuclear single quantum coherence (150 ms) crosspeaks, which were classified into strong, medium, weak (1.8 -2.2 Å͉1.8 -3.4Å͉1.8 -6Å). Hydrogen bonds within the ␣-helix and between adjacent ␤-strands were inferred from NOE patterns. Backbone torsion angle restraints were obtained from backbone secondary chemical shifts using the TALOS software package (36) and X1 restraints from experimental 3 JC␥N/ 3 JC␥C couplings. The geometric quality of the calculated structures was analyzed using PROCHECK-NMR (37).
In Vitro Phosphorylation-Full-length and mutant ICln (1 g/assay) were phosphorylated in vitro using purified catalytic subunit of PKA, purified PKC (both purchased from Promega), and recombinant Histagged human PKGI or PKGII in a buffer containing 20 mM Tris/HCl (pH ϭ 7.4) and 10 mM MgCl 2 , supplemented with 0.5 mM CaCl 2 , 0.6 g of diolein, and 6 g of phosphatidylserine (PKC), 2 mM ␤-mercaptoethanol (PKA and PKGI or PKGII), or 0.01% Triton X-100 and 20 M cGMP (PKGI or PKGII). Phosphorylation was started by the addition of 5 l of ATP (final concentration 120 M) containing 1 Ci of [␥-32 P]ATP (Amersham Biosciences) and terminated after 30 min (30°C) by the addition of SDS sample mix. Proteins were separated by SDS-PAGE.
Mass Spectrometry-Samples were digested overnight at 37°C with two different enzymes. Digestion with Staphylococcus aureus V8 Protease (Roche Applied Science) was performed in a 25 mM sodium phos- phate buffer (pH 7.8), and digestion with trypsin (Roche Applied Science, sequencing grade) was performed in a 10 mM NH 4 HCO 3 buffer (pH 8.9). Protein digests were analyzed using capillary high pressure liquid chromatography connected on-line to a LCQ ion trap instrument (ThermoFinnigan, San Jose, CA) equipped with a nanospray interface as published previously (38). The nanospray voltage was set at 1.6 kV, and the heated capillary was held at 170°C. Tandem mass spectrometry spectra were searched against the Madin-Darby canine kidney ICln sequence using TurboSEQUEST (BioWorks; ThermoFinnigan) with subsequent manual validation. Phosphorylation (ϩ80 Da) was searched on serine, threonine, and tyrosine residues as variable modification.
FRET-FRET and acceptor photobleaching techniques as well as the normalized FRET (NFRET) calculations employed in this study are described in detail in Ritter et al. (20).
Salts, Chemicals, and Drugs-All salts and chemicals were of pro analysis grade. Lipids were purchased from Avanti Polar Lipids.
Statistical analysis-All values are given as mean Ϯ S.E.. Data were tested for differences in the means by Student's t test after verifying normal distribution and equal variance within the data sets. A statistically significant difference was assumed at p Ͻ 0.05.

RESULTS
The NMR Structure of ICln-As shown in Fig. 1, the fulllength ICln protein from Madin-Darby canine kidney cells con-sists of 235 amino acids (aa) (1). Since this protein is prone to degradation (within days at room temperature), we decided to modify ICln by C-terminal truncation to circumvent problems with proteolysis during NMR data acquisition. Deleting the C-terminal 76 aa leads to an ICln protein comprised of the N-terminal 159 aa, which contains the putative transmembrane ␤-strands (1). This truncated form of ICln (ICln 159 ) proved to be very stable (for more than 1 week at room temperature), is monomeric in solution, and forms functional ion channels in lipid bilayers with a pK/pCl of 21.87 Ϯ 4.2 (n ϭ 12). The selectivity of these channels can, similarly to full-length ICln, be shifted toward chloride after the addition of 2 mM Ca 2ϩ (pK/pCl ϭ 7.86 Ϯ 1.6; n ϭ 7; Supplemental Materials, Supplement 1).
The NMR structure of ICln 159 exhibits a PH domain topology. It consists, as shown in Fig. 2a, of a pair of nearly orthogonal antiparallel ␤-sheets with strands ␤ 2 , ␤ 3 , and ␤ 4 forming one side of the ␤-barrel and strands ␤ 1 , ␤ 5 , ␤ 6 , and ␤ 7 forming the other side of the ␤-barrel (Supplemental Materials, Table  I). Near the C terminus, the ␤-barrel is capped by an ␣-helix that is formed by residues 117-133. The loops between ␤ 5 and ␤ 6 as well as ␤ 6 and ␤ 7 are highly mobile as evidenced by little or no deviation of their shifts from random coil values and the high 15 N T 2 values of the amides (Fig. 2b).
Phosphorylation of ICln by PKA, PKC, and PKGII-In platelets, pleckstrin, the archetypical PH domain protein, is the main substrate for PKC. As shown in Fig. 1, the sequence of ICln contains several putative phosphorylation sites for PKC, PKA, and PKGI or PKGII. Phosphorylation by a CKI/CKII-like kinase was shown by Sanchez-Olea et al. (28) and discussed as a potential regulatory mechanism (40).
Here, we show that ICln can be phosphorylated by PKA, PKC, and PKGII but not by PKGI (Fig. 4a). To pin down possible phosphorylation sites, we selectively mutated Ser-45 (exchanging serine by alanine; S45A) in ICln; the results are summarized in Fig. 4b. In ICln S45A , phosphorylation by PKA is abolished, and phosphorylation by either PKC or PKGII is markedly reduced, demonstrating that Ser-45 is the sole target for PKA and the prime target for PKC as well as PKGII. The serine 45 is located in the ␤ 3 -strand of the ICln PH domain (Fig. 1). To identify further phosphorylation sites of PKGII within the ICln PH domain, we subjected PKGII-phosphorylated ICln 159 to mass spectrometry. We searched for peptides that differ by 80 mass units (mass of a single phosphate group), and by doing so, we identified Ser-2 and Ser-93 as additional PKGII phosphorylation sites of ICln 159 . Interestingly, Ser-93 is located within the mobile loop between strands ␤ 6 and ␤ 7 of the ICln PH domain. No threonine or tyrosine phosphorylation was detected.
PKGII Decreases RVD Channel (RVDC) Activity-RVD is one of the functional modules in which ICln is involved. Since kinases (PKA and PKC) are able to modulate RVDC (41), we tested whether PKGII is also able to modulate endogenous RVDC. RVDC activity was measured in NIH3T3 fibroblasts after reducing the extracellular osmolality by 50 mosM. This maneuver leads to an increase of RVDC currents from 0.03 Ϯ 0.01 nA (n ϭ 9) to 1.04 Ϯ 0.19 nA (n ϭ 9) at a holding potential of ϩ40 mV (Supplemental Materials, Supplement 3). To measure the effect of PKGII on RVDC, we overexpressed PKGII in fibroblasts and tested the PKGII effect, after activation with cGMP, on the endogenous RVDC. The baseline transcription of PKGII in fibroblasts is very low, and fibroblasts are not the prime target for PKGII action (42). PKGII is primarily expressed in cells of the intestinal mucosa and the brain as well as epithelial cells in the kidney (42). For the overexpression of PKGII, we used an internal ribosome entry site construct, which allows the simultaneous expression of PKGII and GFP (as a transfection marker) as two separate proteins. The expression of GFP alone did not significantly alter RVDC currents (from 0.11 Ϯ 0.04 nA (n ϭ 9; isotonic) to 0.91 Ϯ 0.19 nA (n ϭ 9; hypotonic); Supplemental Materials, Supplement 3). The expression of PKGII, however, abolishes inward currents of endogenous RVDC and markedly attenuates currents in the outward direction (reduction from 0.91 Ϯ 0.19 nA (n ϭ 9) to 0.33 Ϯ 0.06 nA (n ϭ 9)), without altering the currents under isotonic conditions (Fig. 5a).
As described earlier, ICln is a substrate for PKGII; however, the reduced activity of RVDC followed by the expression and activation of PKGII could also be the result of the phosphorylation of a protein not related to ICln. To test for this hypothesis, we tried to determine the phosphorylation of ICln in PKGIIexpressing cells. We were not able to detect phosphorylated ICln under these conditions (data not shown), most likely because the membrane fraction of ICln is, as reported, very small (5), and therefore, the expected fraction of PKGII phosphorylated ICln probably below the sensitivity of our assay. Therefore, we used another approach to test whether ICln is able to interact with PKGII under in vivo conditions. As shown in Fig.  5b, ICln and PKGII indeed interact at the level of the cell ICln Binds to LSm4, a Component of the Splicing and RNA Degradation Machinery-Using the yeast two-hybrid system, we were able to identify ICln partner proteins involved in another functional module, namely RNA splicing, i.e. Sm proteins (B/BЈ, D1, D2, D3) as well as LSm proteins (LSm2 (SmX) and LSm4) ((10), and data not shown). Here, we show that ICln as well as ICln 159 are able to bind to LSm4 in vitro (Fig. 6, a-e).
The LSm4 protein has a molecular mass of 16.5 kDa and a calculated pI of 10.12 and is thus positively charged at physiological pH. In native PAGE, purified LSm4 does therefore not migrate into the gel. Consequently, LSm4 can only migrate into the gel, if it forms a complex with purified ICln 159 (Fig. 6b, lane  1) or full-length ICln (Fig. 6b, lane 3), both of which are negatively charged (pI 4.17 and pI 4.05, respectively) and thus able to enter the gel and migrate toward the anode (Fig. 6b, lanes  2 and 4). The results obtained with native PAGE were verified by gel filtration chromatography (Fig. 6, a, c, and e) and Western blotting (not shown) using LSm4-specific antibodies.
The interaction between ICln and LSm4 observed in vitro was also tested under in vivo conditions. For these experiments, fusion proteins of LSm4 and ICln with ECFP as well as EYFP were constructed for FRET measurements. For these experiments, the donor ECFP protein was fused to the N terminus of ICln (ECFP-ICln), and the acceptor EYFP protein was fused to the N terminus of LSm4 (EYFP-LSm4). In HEK293T cells co-transfected with ECFP-ICln and EYFP-LSm4, acceptor (EYFP-LSm4) photo-bleaching leads to an increase in donor (ECFP-ICln) fluorescence, demonstrating FRET and thus interaction between ICln and LSm4 in living cells (Fig. 6f). FRET can be detected in the cytosol and in the immediate vicinity of the cell membrane (Fig. 6f, inset). Exchanging donor and acceptor on the respective proteins did not alter the result in quantity and quality (Fig. 6f). However, fusing the fluorescing proteins (donor and acceptor) to the C termini of ICln, or both LSm4 and ICln, annihilated the FRET signal (Fig. 6f).
To identify the binding site for LSm4 on ICln, we performed gel filtration chromatography and native PAGE using LSm4 and two different ICln mutants. As shown in Fig. 6, c and d, using the ICln ⌬loop mutant did not prevent complex formation; however, deleting further C-terminal 25 aa from ICln 159 , leading to ICln 134 , did prevent LSm4 complex formation, suggesting that these 25 aa comprise the binding site for LSm4 (Fig. 6d). DISCUSSION ICln is a multifunctional protein involved in several different regulatory modules within the living cell. Ion transport is one of these modules since ICln is able to form ion channels by directly inserting from its water-soluble form into the lipid membrane. Such a mode for channel insertion is well known for water-soluble bacterial toxin channels (43). Further functional modules are the regulation of RNA splicing by the interaction of ICln with splicing factors (6 -8, 10, 40) and the regulation of the cytoskeleton by ICln (5,(11)(12)(13). The ability of ICln to interact with more than one regulatory module suggests that the water-soluble form of ICln acts as a connector hub (3) for different regulatory mechanisms. The aim of the present study was to identify the structure of the water-soluble form, i.e. the connector hub form of ICln. Based on the structure of ICln, we aimed to identify binding sites for partner proteins such as LSm4 or phosphorylation sites of different kinases. This experimental framework provides an important guidepost for further studies aimed to elucidate the regulation of protein networks in which ICln is involved.
In this study, we determined the structure of a deletion mutant of ICln, i.e. ICln 159 , which is water-soluble, able to form ion channels in lipid bilayers, binds to LSm4, and serves as a substrate for PKA, PKC, and PKGII. The reason for using ICln 159 was the instable nature of full-length ICln when used for the NMR analysis.
ICln 159 folds into a PH domain-like structure. PH domains are a large family of structurally homologous protein domains of moderate to low sequence homology and are found in many proteins involved in signal transduction (44 -46). The canonical structure of PH domains is a core seven-stranded ␤-sandwich capped by a C-terminal ␣-helix and three major interstrand loops that could serve as putative binding sites for proteinprotein interactions (46). As described here, the ␤ 6 -␤ 7 loop of ICln is particularly mobile and long when compared with other PH domains. Usually, this loop is built by 4 -5 aa. However, PH domains tolerate large insertions without effect on the core structure, e.g. one of the two PH motifs of phospholipase C␥ is split at the ␤ 3 -␤ 4 loop by three Src homology domains leading to an insertion of 258 aa (47). Accordingly, deletion of 15 aa within the ␤ 6 -␤ 7 loop did not alter the structure of ICln 159 , demonstrating that the length of the ␤ 6 -␤ 7 loop is not affecting the core structure.
Scrutinizing the aa sequence of ICln 159 for possible sites of phosphorylation and analyzing putative target serines and threonines for accessibility on the structural model reveals three sites of interest, i.e. Thr-34, Ser-45, and Ser-115. As shown by single point mutagenesis, Ser-45 is the main target for PKA and PKC, as well as PKGII. It is noteworthy that in contrast to other substrates of PKGII in vivo, i.e. cystic fibrosis transmembrane conductance regulator (CFTR) (48,49), ICln is able to discriminate between PKGI and PKGII in vitro (Fig.  4a). Three PKGII phosphorylation sites, Ser-2, Ser-45, and Ser-93, were identified by single point mutagenesis or mass spectrometry. As shown in Fig. 4b, Ser-45 is the main phosphorylation site for PKGII as well as PKC. Interestingly, Ser-93 is located exactly within the mobile loop connecting strands ␤ 6 and ␤ 7 , and preliminary experiments indicate that this loop is important for current formation in lipid bilayers. 2 Therefore, it is intriguing to speculate that phosphorylation of Ser-93 could modulate the ICln-induced current. This scenario is speculative; however, FRET experiments reported in this study suggest a interaction between ICln and PKGII at the cell membrane, in vivo. Additional experiments need to be done to clarify this issue.
ICln 159 binds to LSm4, which is most likely methylated by PRMT5 that also methylates Sm proteins (9,50,51). PRMT5 possibly binds to AD3 at the C terminus of ICln (51). The experiments summarized in this study show that the binding of ICln to LSm4 involves the second acidic domain (AD2; Fig. 1) situated immediately after the capping ␣-helix of the ICln PH domain, but not AD1, located in the ␤ 6 -␤ 7 loop of the ICln PH domain. It is interesting to note that within this acidic domain, there is a serine at position 142, which is phosphorylated by cytosolic extracts (40). This phosphorylation may potentially regulate Sm/LSm binding to ICln. However, this hypothesis needs to be tested.
In conclusion, the water-soluble form of ICln is a PH domainlike structure with an overall negative surface potential. The protein is a potential target for PKA, PKC, and PKGII phos- phorylation and binds strongly to LSm4, a positively charged protein that is part of the spliceosome and of the RNA degradation machinery. As summarized in Fig. 7, water-soluble ICln acts as a protein-protein interaction platform involved in different regulatory modules, which is a key feature of connector hubs in complex protein networks (3). By solving the structure of ICln 159 , by determining the binding site of ICln for LSm4, and by demonstrating that ICln can be phosphorylated by PKC, PKA, and PKGII on Ser-45 as well as Ser-2 and Ser-93 (PKGII), we provide a valuable guidepost for the systematic exploration of the involvement of ICln in such diverse regulatory mechanisms as cell volume regulation, RNA splicing, RNA degradation, and cytoskeletal reorganization.