The yeast reporter strain L40 (Vojtek et al., 1993) (MATa, trp1, leu2, his3, LYS::lexA-HIS3, URA3::lexA-lacZ) was transformed with pLexN-hEag1 by the lithium acetate method and grown on synthetic medium lacking tryptophan. After additional transformation with plasmid pVP16-3-cDNA [postnatal 8 rat brain cDNA library, (Okamoto and Sudhof, 1997)], double transformants were plated on synthetic medium lacking histidine, leucine, uracil, lysine, tryptophan and in the presence of the competitive inhibitor of the HIS3 protein 3-amino-1,2,4-triazole (3-AT). Positive colonies were picked after 4–6 days and tested for ß-galactosidase activity using plate assay. Plasmids from positive clones were rescued and transformed in E. coli strain HB101. E. coli cells were plated on leucine-lacking medium. Positive clones were further analyzed by yeast retransformation and DNA sequencing.

HEK 293 cells were washed 3 times with ice cold PBS, harvested and directly used for RNA purification using RNAeasy Mini Kit (Qiagen). First strand cDNA was produced using SuperScript (Invitrogen) with oligo-dT. For cloning, PIST was amplified in a standard PCR reaction using Taq DNA polymerase (NEB) and sense and antisense primers (see below). PCR products were analyzed by electropohoresis, fragments excised out of gel and purified using NucleoSpin Extract (Macherey-Nagel) and subcloned in pGEMT-easy (Promega) following manufacturer's instructions.

Mouse first strand cDNA was a generous gift from Dr. R. Ufartes (Ufartes et al., 2013).

To distinguish between isoforms, PCR products were obtained in a standard PCR procedure using 100 ng cDNA mouse first strand cDNA, and a set of 2 or 3 specific primers in the appropriate combinations to distinguish between isoforms (see below) and Biotherme Polymerase (Genecraft).

Primers for human PIST (accession numbers NM_001017408.2—PIST- and NM_020399.3—sPIST-) were:

5′ ATGTCGGCGGGCGGTCCATGC3′ and

For mouse PIST (accession numbers NM_001199272.1 and NM_053187.3):

5′GCAGAGGGCGCAACGACTT3′ and

For selective amplification of the previously described transcript variants, we used different forward primers (nPIST; NM_020399.3: 5′ CAAGGCAAAATTGTCTGTCCAC 3′; PIST, NM_001017408.2: 5′ CAAGGCAAAATTGGAAAGAGAAC 3′), and to distinguish between short and long forms, the following antisense primers were used:

Short 5′ CTCTCTGCTGGGAGAGTATAG 3′

HeLa cells were cultured in MEM+GlutaMax (Invitrogen) supplemented with 10% FCS (PAA), HEK293 cells in DMEM/F12+GlutaMax (Invitrogen) supplemented with 10% FCS and for stable cell line HEK293-K_V10.1-BBS [HEK-BBS, (Kohl et al., 2011)] with Zeocin (Invitrogen) at 5% CO₂ and 37°C.

Transfection was performed using Lipofectamine 2000 or Lipofectamine (Invitrogen) according to the manufacturer's instructions. K_V10.x-BBS Venus and CFP-PIST were generated by cloning into pcDNA3 or pECFP-N1 vectors. Empty vectors were used as controls.

Labeling of whole-cell K_V10.1-BBS was performed in cell lysates in buffer LP [20 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM MgCl₂, 1% Nonidet P40, protease inhibitors (Roche)] with α-bungarotoxin (BTX)-biotin conjugate (Invitrogen) in a final concentration of 0.2 μg/ml for 30 min on ice. To detect membrane and/or internalized K_V10.1-BBS, transfected HEK-BBS cells were incubated in media supplemented with α-BTX-biotin conjugate (Invitrogen) in a final concentration of 2.5 μg/ml and kept at room temperature for 10 min (membrane) or at 37°C for one hour (internalized). For internalized K_V10.1-BBS, cells were washed with ice-cold acid wash buffer (150 mM NaCl, pH 3.0) for 3 min to remove membrane labeling of K_V10.1-BBS. Twice washing with cold PBS removed residual α-BTX-biotin conjugate. Cells were then harvested and lysed with LP buffer for 20 min on ice. Insoluble fraction was removed by centrifugation at 18,000 × g at 4°C, and the supernatant used for ELISA or pull down experiments.

For pull down approaches, labeled K_V10.1-BBS was bound on streptavidin-coated magnetic beads (T1, Invitrogen) for 2 h at 4°C. Unbound protein was removed by washing 5 times with buffer LP supplemented with protease inhibitors (Roche). Bound protein was eluted at 70°C for 10 min using LDS sample buffer containing reducing agent (Invitrogen) and analyzed by SDS PAGE (Invitrogen) and western blotting.

Quantification of the amount of labeled K_V10.1-BBS was performed by ELISA. After labeling, total cell lysates (30 and 150 μg protein), were immobilized on streptavidin-coated plates (Pierce) and detected using a C-terminal monoclonal anti-K_V10.1 antibody [Ab33, 5 μg/ml (Hemmerlein et al., 2006)] and a polyclonal anti-mouse secondary antibody (Pierce, 1:500) coupled to peroxidase. ABTS (Invitrogen) was used as a substrate for development and detected in a Wallac Victor2 reader at 405 nm (reference 490 nm). Experiments were performed in duplicates.

For immunoprecipitation, rat brain lysates (800 μg total protein) were incubated overnight at 4°C with 2–5 μg of antibody [anti-K_V10.1 33/62 (Hemmerlein et al., 2006); anti-PIST (Millipore); non-specific mouse IgGκ 2b) in buffer A (0,5% Triton X-100, 25 mM TrisHCl pH7.5, 75 mM NaCl, 2.5 mM EDTA and protease inhibitors (Roche)]. Pull down was performed by adding 20 μl of Protein G/A-coated Sepharose beads (Calbiochem) for 2 h at 4°C under rotation. After washing five times with buffer A, bound protein was eluted at 70°C for 10 min using LDS sample buffer containing reducing agent (Invitrogen) and analyzed by SDS PAGE (Invitrogen) and western blotting.

Recordings on Xenopus oocytes were performed as described by Stuhmer (1998) using a Turbo TEC-10CD amplifier (NPI electronics). cRNA was synthesized using the mMessage mMachine kit (Ambion) according to the manufacturer's instructions. 0.1–1 ng cRNA per oocyte were microinjected 1–3 days prior to recording. Cells were kept at 18°C in ND 96 solution (96 mM NaCl, 2 mM KCl, 0.2 mM CaCl₂, 2 mM MgCl₂, 5 mM Hepes/NaOH, pH 7.5) supplemented with theophylline (0.5 mM) to avoid maturation of the oocytes. For voltage clamp recordings, pipettes had resistances ranging from 0.5 to 1.2 MΩ when filled with 2 M KCl. External solution (NFR) contained 115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl₂, 10 mM Hepes/NaOH, pH 7.2. Current was digitized at 5 kHz and filtered at 1 kHz. Currents were elicited by 1 s depolarizations from a holding potential of −100 mV to values ranging from +80 to −100 mV. Current amplitude was determined as mean value in the last 200 ms of the depolarization. To fit current-voltage relationships, we used an equation of the form: I(V)=Γ·(V−Vrev)(1+eV50−Vk)4 where Γ is the total conductance, V_rev the reversal potential, V₅₀ the potential of half activation per subunit and k the slope factor.

Data acquisition and analysis was performed with Pulse-PulseFit (HEKA Electronics) and IgorPro (WaveMetrics).

For confocal microscopy, HeLa cells were grown on glass coverslips and transfected with Venus-tagged K_V10.1 and CFP-tagged isoforms of PIST. Empty vectors served as control. 24 h after transfection, the Golgi marker BODIPY FL C₅ Ceramide (Invitrogen) was used according to manufacturer's instructions to visualize Golgi apparatus. Cells were fixed in 4% paraformaldehyde in PBS for 10 min at room temperature, washed thoroughly with HBBS and mounted with ProLong Gold Antifade Reagent (Invitrogen). Image acquisition was performed using a Zeiss LSM 510 Meta confocal microscope with a Zeiss 40x 1.2 Korr water immersion objective. Single channel gray scale images are presented using color tables according to the fluorophore used. Merged images were generated by combination of the channel images into a single RGB file. Only linear contrast enhancement and Gaussian blur were applied off-line to the images; analysis was carried out with Fiji (Schindelin et al., 2012).

Finding new interaction partners of K_V10.1is crucial to unravel the role of this ion channel both in the healthy brain and in tumors. To identify candidate interaction partners of K_V10.1, we performed yeast two-hybrid screens using a rat brain expression library. The C-terminus of K_V10.1 in fusion to the DNA binding domain of LexA transcription factor was used as a bait to screen a rat brain cDNA library fused to the activation domain of VP16 transcription factor. As previously reported (Ninkovic et al., 2012), out of 2 million screened clones, eight were positive for the HIS3 marker. One of them encoded most of the open reading frame of the murine PIST, including its PDZ domain (Neudauer et al., 2001).

We then performed co-immunoprecipitation studies to confirm the interaction. Precipitation of PIST using a specific antibody was able to pull down K_V10.1 out of rat brain lysates (Figure 1A). In reverse co-IP experiments, PIST co-precipitated K_V10.1 (Figure 1B). Non-specific isotype antibodies did not precipitate either of the two proteins.

In a second pull-down approach we used HEK293 cells stably expressing a modified K_V10.1 bearing an α-bungarotoxin (BTX) binding site [K_V10.1_BBS, (Kohl et al., 2011)]. Specific binding of a BTX-biotin conjugate to its binding site allows precipitation of K_V10.1_BBS from cell lysates. After transfection with CFP-labeled PIST, pull down of K_V10.1_BBS with BTX-biotin and immunoblotting using anti-PIST antibodies revealed two bands compatible with endogenous PIST as well as CFP-labeled PIST (Figure 1C, lane +Bb); no bands were detected if BTX-biotin was omitted (Figure 1C, lane −Bb). The fact that both endogenous and CFP-tagged PIST are detected argues in favor of the specificity of PIST detection. As a control, the same experiments using unlabeled K_V10.1_BBS did not precipitate PIST, also indicating specificity of the co-IP. Taken together, these results strongly indicate that PIST and K_V10.1 interact physically both in vitro and in vivo.

For further analysis we amplified and cloned the human PIST out of cDNA obtained from untransfected HEK293 cells, using 5′ and 3′ flanking primers. Interestingly, during the cloning process we systematically observed two bands, one of them at the expected size of full length PIST and a shorter one. Cloning and subsequent sequencing of the shorter band revealed that it corresponds to a PIST variant lacking the complete exon 6, which is the region coding for the majority of the PDZ domain indicating that this clone corresponds to a new splice variant of PIST. To further proof the existence of the new isoform, we designed primers to generate an amplicon containing the skipped exon, in order to discriminate between the longer and the shorter form of PIST. We were able to amplify fragments with the expected sizes out of cDNA from HEK293 cells (data not shown). To evaluate the tissue distribution of the splice variant we used mouse cRNA from different tissues. The variant (lacking in the mouse exon 5) was detected in all tissues tested except pancreas, where amplification of full-length PIST was also much weaker (Figure 2A). The signals for the short form were consistently weaker than those for full-length PIST.

In 2002, a second isoform of PIST was cloned and termed “neuronal PIST” (nPIST) because it is believed to be the main form in the brain. It differs from PIST by an 8 amino acid insertion in the coiled-coil region (Yue et al., 2002). The sequence we identified here codes for a protein that lacks the main part of the PDZ domain but bears the 8 amino acids of the neuronal form. We therefore asked if there is a fourth isoform, lacking the PDZ as well as the 8 amino acid insertion. PCR with primer pairs distinguishing between nPIST and PIST and producing different fragment length depending on the presence of the alternatively skipped exon was used to answer this question. Amplification of the expected fragments out of cDNA from HEK293 cells as well as from different murine tissues indicated that all four isoforms are expressed in nearly every tissue (Figure 2B). Weak or absent expression of nPIST in liver and pancreas correlated with no detection of the short form (nsPIST) of it, indicating that expression of the shorter form depends on the expression of the corresponding longer one (Figure 2B). The cartoon in Figure 2C depicts the domain structure of the four PIST isoforms. Taken together, our data suggests that two new short forms of PIST (from here on termed sPIST and nsPIST) exist and are expressed in nearly every tissue. According to our semi-quantitative approach, both seem to be less abundant than their corresponding longer form. sPIST and nsPIST sequences have been deposited under accession numbers KF420122 and KF420123, respectively.

PIST is known to form homodimers depending on the second coiled-coil motif, which bears also a leucine zipper (Neudauer et al., 2001). As this region is intact in every identified isoform, we tested if the shorter variants can form heterodimers with the longer ones. To this end, we transfected CFP-labeled PIST constructs and used a GFP-specific antibody to immunoprecipitate the extracts that were subsequently blotted using a PIST-antibody. Precipitation of CFP-tagged PIST and nPIST co-precipitated a band compatible with endogenous PIST (or nPIST), strongly suggesting that dimerization of PIST occurs in our system (Figure 3A). The major band that co-immunoprecipitated with CFP-tagged nsPIST was smaller than the one pulled by full-length forms and compatible with the size of an endogenous short isoform, suggesting that the short isoforms (or at least nsPIST) do occur at the protein level in vivo. Expression of sPIST-CFP was not clearly detectable even after immunoprecipitation, rendering interpretation of the results difficult with this isoform, even though an endogenous band was detected. We therefore excluded sPIST from transfection-based experiments and concentrated on the other three variants.

We cannot exclude formation of heterodimers between nPIST and PIST isoforms, because HEK293 cells express all four isoforms of PIST and in this approach we cannot resolve the eight residues difference between PISTs and nPISTs.

We next tested the different isoforms for their capability to bind K_V10.1. Precipitation of K_V10.1_BBS out of stably expressing HEK293 cells was able to co-precipitate over-expressed CFP-tagged PIST and nPIST (Figure 3B). In comparison to the long forms, co-precipitation of the tagged short isoform CFP_nsPIST was much less evident in this assay (Figure 3B). Since the short forms lack most of the PDZ domain, these data suggest an involvement of the PDZ domain of PIST in K_V10.1 binding. Interestingly, overexpression of the long isoforms resulted in higher intensity in bands compatible with the endogenous full-length PIST, while the shorter isoform rendered more intense bands corresponding to short isoforms (Figure 3B).

We next tested if there are functional consequences of the interaction between PIST and K_V10.1. As a model we used voltage clamp recordings in the Xenopus oocyte expression system and checked if the overexpression of any PIST isoform can influence K_V10.1-mediated current. Regardless of the concomitant overexpression of human PIST isoforms, macroscopic K_V10.1 currents showed very similar voltage dependence (Figure 4); in fact, the current-voltage relationship could be fitted using the same parameters used for K_V10.1 alone (Figure 4B). We did not appreciate differences in kinetics, and the main properties of K_V10.1 were conserved in the presence of PIST isoforms (not shown).

The total current amplitude, however, was influenced by PIST expression (Figures 4A,C). Over-expression of PIST was able to nearly double the amplitude of K_V10.1 mediated current (189 ± 33% of control levels). The difference was more significant the more depolarized the stimulus (reaching p < 0.0001 at +80 mV and above). Injection of nPIST and nsPIST did not affect the amplitude. These observations are compatible with an influence of PIST on the membrane expression of K_V10.1, similar to the one described for CFTR (Cheng et al., 2005). In the presence of PDZ and coiled-coil domain, the amplitude of K_V10.1 currents was increased. We observed no functional interaction with the insertion in the coiled-coil of “n” forms.

We previously stated that very poor and irregular expression of sPIST did not allow us to document physical interaction between this isoform and K_V10.1. However, since experiments in oocytes do not depend on transfection or transcription, which is performed in vitro previous to oocyte injection, we also tested if injection of synthetic cRNA encoding for sPIST affects the amplitude of the K_V10.1-mediated current. Surprisingly, sPIST coinjection induced a reduction in current amplitude to 55 ± 4% of control levels. Although they had opposite effects, coinjection of a mixture of PIST and sPIST together with K_V10.1, did not results on compensation of the effects, but rather sPIST showed a dominant-negative effect, in such a way that the current was reduced to the same extent as when sPIST was injected alone (Figure 4D).

A Golgi-resident protein like PIST might have a general effect on the expression level of ion channels in the oocyte expression system, enhancing transport mechanism toward the membrane. As a measure for the specificity of our observations, we tested the effect of PIST overexpression on K_V10.2, the closest relative of K_V10.1. PIST overexpression did not change voltage-dependence or kinetic parameters of K_V10.2. In contrast to K_V10.1, there was also no change in current amplitudes of K_V10.2-mediated currents (Figure 4E), pointing toward a highly specific effect of PIST expression on K_V10.1, but not K_V10.2 current amplitude.

Having shown that PIST is able to interact physically and functionally with K_V10.1, we next asked if both proteins co-localize in vivo. PIST is reported to reside mainly in the Golgi apparatus, depending on at least one coiled-coil region, but can also be pulled to the membrane in co-expression with its interaction partners (Charest et al., 2001; Yao et al., 2001; Wente et al., 2005b). To perform subcellular localization studies, we used CFP-tagged forms of PIST. As described in the literature, the longer forms of PIST were mainly located in the Golgi, the neuronal form being slightly more diffuse (Figures 5, 6A). nsPIST localized to discrete areas of the Golgi.

Co-expression of a Venus-tagged K_V10.1 led to a slight change in the localization of PIST only, a fraction of which showed expression at the membrane; this also resulted in a clearer membrane localization of K_V10.1 itself. Expressed alone, K_V10.1_Venus showed a typical distribution of this ion channel in the cell, where most K_V10.1 stays intracellular and only weak expression can be seen at the membrane (Figure 6B). PIST coexpression led to peripheral venus signal compatible with increased channel localization at the plasma membrane (Figures 6C,D). Overexpression of nPIST as well as its short form nsPIST did not change K_V10.1 localization in the cell (not shown). This data further supports the idea that a trafficking effect of PIST might be involved in the regulation of macroscopic current amplitude.

To get more quantitative data we made use of another system recently established in our lab, which is based on K_V10.1 bearing an α-bungarotoxin (BTX) binding site between transmembrane segments 3 and 4 (K_V10.1_BBS). Different labeling techniques with an α-BTX-biotin conjugate allow to distinguish between whole cell, membrane bound and internalized K_V10.1 out of stably expressing cells and to quantify this different fractions via ELISA (Kohl et al., 2011). Specific labeling of membrane K_V10.1_BBS is possible by incubating intact cells stably expressing K_V10.1_BBS at low temperature (to avoid internalization) with an α-BTX-biotin conjugate. Quantification of this fraction of membrane bound K_V10.1_BBS revealed that the overexpression of CFP-tagged PIST can indeed increase surface K_V10.1_BBS to 214 ± 28% of control levels (Figure 7). In agreement with findings in the oocyte expression system, the neuronal forms of PIST (nPIST and nsPIST) did not have any effect on membrane K_V10.1_BBS, as values were not different (94 ± 15% nPIST, 91 ± 16% nsPIST) from control levels.

Both increased expression levels and increased stability of plasma membrane K_V10.1_BBS could explain this observation. To distinguish between both possibilities, we examined the effect of overexpressing PIST isoforms on whole cell K_V10.1_BBS, labeled after lysis of the cells. Neither isoform of PIST had an effect on the amount of total K_V10.1_BBS (Figure 7), giving values of 96 ± 5% (PIST), 92 ± 3% (nsPIST) or 102 ± 9% (nPIST) in comparison to control levels, indicating that neither isoform of PIST affect total expression levels of K_V10.1_BBS. Therefore, they may be involved in the transport of this voltage gated ion channel to and from the plasma membrane.

The K_V10.1_BBS system easily allows for the quantification of internalized K_V10.1_BBS within a given time frame (Kohl et al., 2011). For this, we incubated living cells with α-BTX-biotin conjugate at 37°C and allowed internalization for one hour. Removal of the label of membrane-bound K_V10.1_BBS by acid wash afterwards leaves only the internalized labeled K_V10.1_BBS for quantification. Interestingly, neither isoform had an effect on the total amount of internalized K_V10.1_BBS (PIST 90 ± 9%, nsPIST 92 ± 13%, nPIST 98 ± 8%). As membrane expression of K_V10.1_BBS is clearly increased in response to PIST overexpression (Figure 7), one would expect more channels internalized if the rate of internalization is not affected by this protein.

Luis A. Pardo is shareholder at iOnGen AG. The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.