The Prion Protein Modulates A-type K+ Currents Mediated by Kv4.2 Complexes through Dipeptidyl Aminopeptidase-like Protein 6*

Background: Prion protein (PrP) interacts with dipeptidyl aminopeptidase-like protein 6 (DPP6), but the functional significance was unknown. Results: PrP formed complexes with and impacted the function of potassium channels containing DPP6 and Kv4.2. Conclusion: PrP modulates voltage-dependent and kinetic properties of Kv4.2 channels. Significance: This could explain a phenotype of PrP knock-out mice and the effects of amyloid β oligomers. Widely expressed in the adult central nervous system, the cellular prion protein (PrPC) is implicated in a variety of processes, including neuronal excitability. Dipeptidyl aminopeptidase-like protein 6 (DPP6) was first identified as a PrPC interactor using in vivo formaldehyde cross-linking of wild type (WT) mouse brain. This finding was confirmed in three cell lines and, because DPP6 directs the functional assembly of K+ channels, we assessed the impact of WT and mutant PrPC upon Kv4.2-based cell surface macromolecular complexes. Whereas a Gerstmann-Sträussler-Scheinker disease version of PrP with eight extra octarepeats was a loss of function both for complex formation and for modulation of Kv4.2 channels, WT PrPC, in a DPP6-dependent manner, modulated Kv4.2 channel properties, causing an increase in peak amplitude, a rightward shift of the voltage-dependent steady-state inactivation curve, a slower inactivation, and a faster recovery from steady-state inactivation. Thus, the net impact of wt PrPC was one of enhancement, which plays a critical role in the down-regulation of neuronal membrane excitability and is associated with a decreased susceptibility to seizures. Insofar as previous work has established a requirement for WT PrPC in the Aβ-dependent modulation of excitability in cholinergic basal forebrain neurons, our findings implicate PrPC regulation of Kv4.2 channels as a mechanism contributing to the effects of oligomeric Aβ upon neuronal excitability and viability.

Prion diseases involve the structural conversion of the primarily ␣-helical, cellular prion protein (PrP C ) 5 to an infectious, ␤-sheet-enriched form, PrP Sc . The high degree of primary to tertiary structural conservation of mammalian PrP C (1, 2) leads to an expectation of an explicit phenotype in Prnp 0/0 mice, but, aside from a total resistance to prion disease, this is not the case (3,4). That said, within the inventory of subtle or disputed phenotypic changes in these mice are reports of impairment in GABA A receptor-mediated inhibition and long term potentiation (5)(6)(7)(8). The diversity of altered end point measures is, to a certain extent, paralleled in the large number of reported PrP Cinteracting proteins: the laminin receptor (9 -11), the neural cell adhesion molecule (12,13), stress-inducible protein-1 (14,15), and NMDA receptors (16,17) to name but a few (reviewed in Ref. 18).
In addition to previously described interacting proteins, dipeptidyl aminopeptidase-like protein 6 (DPP6; also known as DPPX) was identified using time-controlled transcardiac perfusion cross-linking while probing the PrP C interactome (19). DPP6 is an auxiliary subunit of pore-forming Kv4.2 channels (20,21) and, together with most K ϩ channel-interacting protein (KChIP) isoforms (22), DPP6 increases Kv4.2 trafficking to the cell surface and is required for the reconstitution of the properties of the native channel complex in heterologous cells (23). A type II transmembrane protein, DPP6 has a number of splice variants differing in the length of the cytoplasmic, N-terminal portion (DPP6-S; short) (24). KChIPs are a family of intracellular Ca 2ϩ -binding proteins with four major isoforms (1)(2)(3)(4) and at least 16 splice variants (25). Interactions between Kv4.2, KChIPs and DPP6 have been confirmed by proteomic analyses demonstrating the pull-down of KChIP1 to -3 in comparable ratios (26). Assembled Kv4 channel complexes mediate sub-threshold operation somatodendritic transient outward K ϩ currents (A-type K ϩ currents), which play important roles in the regulation of neuronal membrane excitability, somatodendritic signal integration, and long term potentiation (27,28). Given the interplay between Kv4.2 channels and DPP6 in neuronal function (29) and the ability to cross-link DPP6 and PrP C in vivo (19), we sought to delineate the nature and repercussions of a DPP6-PrP C interaction. To this end, we investigated the impact of PrP C upon the properties of A-type K ϩ currents mediated by Kv4.2 channel complexes derived from co-expression of Kv4.2, KChIP2, and DPP6-S (30,31).

EXPERIMENTAL PROCEDURES
Plasmid Construction-cDNAs encoding Kv4.2 (MMM1012-9498428 clone 30356567, pYX-Asc) and KChIP2 (MMM1013-7511937 clone 4503251, pCR4-TOPO) (Open Biosystems) were inserted behind the CMV and EF1␣ promoters, respectively, of pBud.CE4 with Quick Ligase (New England Biolabs) to create pBud.CE4.Kv4.2.KChip2. To construct pBud.DPP6-S.RFP (co-expression driven from the CMV and EF1␣ promoters, respectively), total RNA was isolated from half mouse brains using acid guanidinium thiocyanate/phenol/chloroform extraction (32) and mRNA-purified using the Oligotex mRNA minikit (Qiagen). cDNA synthesis was performed in the presence of SUPERase-In RNase inhibitor (Ambion) using 0.5 g of mRNA, Omniscript reverse transcriptase (Qiagen), and an oligo(dT) primer. PCR amplification of DPP6-S and DPP6-E was conducted using Pfu Turbo DNA polymerase (Invitrogen) with a nested PCR strategy and inserted between the BamHI and XbaI sites of the pcDNA3 mammalian expression vector (Invitrogen). A PstI site was added to the pcDNA3.DPP6-S for insertion into pBud.empty.GFP with Quick Ligase (New England Biolabs) after gel purification (QIAquick gel extraction kit (Qiagen)). GFP was replaced with RFP by digestion of pBud.DPP6-S.GFP and pBud.empty.RFP with NdeI and NheI, isolation of the desired fragments, and ligation. HA-tagged DPP6-S, DPP6⌬cyto, and DPP6 deletion mutants were generated by standard PCR-based techniques. The secreted DPP6 ectodomain construct was generated by fusing the DPP6 ectodomain to the PrP N-terminal signal sequence using an introduced BsrGI site. The Thy-1 plasmid (Thy-1.2 isoform) was generated by amplification of the Thy-1 open reading frame from the MGC:62652 cDNA clone by PCR and then insertion into pcDNA3. PrP A116V and M128V HD dup PrP were created using the QuikChange (Stratagene) site-directed mutagenesis procedure with Pfu Turbo DNA polymerase. Other PrP mutants and Doppel plasmids were generated as described previously (33). All plasmids used for transfection were enriched using the EndoFree Plasmid Maxi kit (Qiagen).
Transfections-HEK293T cells were maintained in DMEM with 10% FBS (Invitrogen). Transfections were performed with Lipofectamine 2000 (Invitrogen) according to the manufactur-er's instructions. Cells used for electrophysiological recording were transfected with the following plasmids (fluorescent proteins co-expressed): pBud.DPP6-S.RFP, pBud.Kv4.2/KChIP2, pBud.wtPrP.GFP, pBud.octa13PrP.GFP, and pBud.empty.GFP. These transfection mixtures included (as noted) siRNA directed against the 3Ј-non-coding region of human PRNP mRNA (Ori-Gene Trilencer-27 siRNA duplex; rGrGrCrUrUrArCrArArUr-GrUrGrCrArCrUrGrArArUrCrGTT) or a scrambled control siRNA. 24 h after transfection, cells were trypsinized and plated on coverslips for electrophysiological recording the following day. Cells used for complex formation assays were transfected as above using constructs based on a pcDNA plasmid vector backbone (with identities indicated in the figure legends) except for Fig. 3C, which used a pBud vector.
Generation of DPP6 Antibodies-Peptides were synthesized (containing an N-terminal cysteine for KLH conjugation), conjugated to maleimide-activated KLH (Pierce), and then injected into New Zealand White rabbits. 03J2 was raised against residues 507-522 (DKRRMFDLEANEEVQK), and 03K1 was raised against residues 788 -803 (QDKLPTATAKEEEEED). Polyclonal antibodies were precipitated from serum using ammonium sulfate and affinity-purified using the corresponding immunogenic DPP6 peptide conjugated to a SulfoLink column (Pierce).
Cell Surface Biotinylation Assay-The assay was performed according to the manufacturer's instructions (Pierce).
Immunoprecipitations-120 g of cell lysate was incubated overnight at 4°C with 0.7 g of ␣-HA antibody (Sigma) in a total volume of 50 l. 100 l of protein A/G-agarose beads were washed three times with radioimmune precipitation assay buffer containing protease inhibitor (Roche Applied Science) before adding the overnight incubation and were then incubated at room temperature for 2 h rotating end over end. After three washes with radioimmune precipitation assay buffer, the beads were washed a final time with water, and the complexes were eluted from the beads (and cross-links were reversed) by incubation in sample buffer at 95°C for 30 min.
Animal Husbandry-All animal protocols were in accordance with the Canadian Council on Animal Care and were approved by the Institutional Animal Care and Use Committees at the University of Alberta and the University of Toronto.
Mouse Lines and Preparation of Mouse Brain Homogenates-Prnp 0/0 mice (ZrchI strain) were maintained on a C57/B6 background. DPP6 df5J/Rw mice were a generous gift from John Schimenti and were maintained on a C3H background. Mice were perfused with saline, and half-brains were extracted and then either homogenized directly in nine volumes of 0.32 M sucrose with protease inhibitors (Roche Applied Science) or snap-frozen and stored at Ϫ80°C.
Western Blotting-Cells were lysed with radioimmuneprecipitation assay lysis buffer. Brains were homogenized in 0.32 M sucrose. A BCA assay (Pierce) was performed to determine protein concentration. Samples were boiled in loading buffer (with the exception of samples to be used for Kv4.2 detection due to protein aggregation), and were electrophoresed with either Tris-glycine gels (8 or 12%) or 4 -12% NuPAGE gradient gels (Invitrogen) and transferred to PVDF membranes (Millipore). Membranes were blocked with either 3% (Kv4.2) or 5% milk (all others except for SHA31 blots, which are not blocked) in TBS with 0.1% Tween (Fisher). Membranes were incubated overnight in the indicated primary antibody (DPP6 clones 03K1 and 03J2 (created in house), actin (Sigma), Kv4.2 clone K57/1 and KChIP2 clone K60/73 (NeuroMab), SHA31 (Spi-bio), 8B4 and 7A12 (generous gifts from Man-Sun Sy), HA (Sigma), and Thy-1 (graciously provided by Roger Morris)) and after washing were incubated in the appropriate HRP-conjugated secondary antibody for 2 h at room temperature (goat ␣-mouse or goat ␣-rabbit (Bio-Rad)), washed again, exposed to ECL (Pierce), and visualized with light-sensitive film (Fujifilm). Quantification was performed using ImageJ. The intensity of the ϳ191-kDa species detected by Western blotting was normalized to intensity of "monomeric input" PrP in the same gel lane and plotted using an arbitrary scale. p values were determined using Student's t test.
Immunocytochemistry-HEK293T cells were transfected with mouse PrP and HA-tagged DPP6-S (both on a pcDNA backbone) and an siRNA against endogenous human PrP. 24 h after transfection, cells were replated on glass coverslips and given 24 h to recover. After rinsing twice with PBS and fixation with 4% paraformaldehyde, cells were washed three times with PBS and incubated in Sha31 (1:5000) overnight at 4°C with rocking. Following three PBS washes, cells were blocked with 2% goat serum (Invitrogen) and incubated with goat ␣-mouse Alexa Fluor 594 (1:300; Invitrogen) for 2 h at room temperature. After washing and permeabilization with 0.2% Triton X-100, cells were blocked and incubated overnight in 1:500 ␣-HA (Sigma) as above, washed three times, blocked, and incubated for 2 h with 1:300 goat ␣-mouse Alexa Fluor 488 (Invitrogen) at room temperature. Nuclei were stained with 1 g/ml Hoechst and visualized using a Nikon Eclipse 90I motorized upright microscope (Nikon) and a CFI PL ϫ40/numerical aperture 0.75 lens (Nikon) using the following excitation/emission filter properties: 325-375/500 -575 nm with a 495-nm long pass filter (blue channel), 440 -510/475-575 nm with a 495-nm long pass filter (green channel), and 505-615/570 -720 nm with a 595-nm long pass filter (red channel). Images were acquired with a Retiga 2000R monocooled camera, fast 1394, using NIS-Elements AR advanced research software at room temperature.
Recordings from HEK293T Cells-Whole-cell recordings were always applied to two groups of HEK293T cells on the same day to minimize variation resulting from cell manipulation prior to recordings. Fluorescence-positive individual HEK293T cells were visualized (by way of the GFP and RFP encoded in the bigenic plasmids) and selected for recording under an Axioscope 2 Fs microscope (Zeiss) at ϫ60 magnification. Oxygenated external solution was bath-perfused at a rate of 1.0 ml/min. The external solution contained 140 mM NaCl, 2.5 mM KCl, 1.2 mM MgCl 2 , 1.5 mM CaCl 2 , 10 mM HEPES, and 10 mM glucose, pH 7.4. Recording pipette solution was composed of 140 mM potassium methylsulfate, 5 mM MgCl 2 , 10 mM HEPES, 2 mM Na 2 ATP, 0.2 mM NaGTP, and 0.5 mM EGTA, pH 7.3. Borosilicate glass capillaries (thin wall with filament, 1.5 mm; World Precision Instruments) were pulled with a Narishige (PP-83) puller to yield recording pipettes with resistances of 2-4 megaohms. Series resistance of 4 -7 megaohms in whole-cell configuration was compensated by 60%. Three different stimulation protocols in our experiments were employed to evoke Kv4.2-mediated A-type K ϩ currents, which were used for building relationship of voltage-dependent activation (activation curve), steady-state inactivation (inactivation curve), and rate of recovery time from steady-state inactivation (curve of recovery time). A-type K ϩ currents for making the activation curves were evoked for 300 ms, depolarizing membrane potential ranging from Ϫ60 to 50 mV following a conditioning pulse of Ϫ110 mV for 400 ms. The A-type K ϩ currents used to build the inactivation curves were evoked during a fixed depolarizing potential of 30 mV following 400-ms conditioning pulses of various membrane potentials, ranging from Ϫ110 to Ϫ20 mV. A-type K ϩ currents for plotting the curve of the recovery rate were elicited during a fixed 30-mV depolarizing membrane potential for 300 ms following a series of increasing time intervals of conditioning hyperpolarizing potential at Ϫ110 mV. All A-type currents presented were subject to subtraction of the outward currents (endogenous currents) evoked by a conditioning pulse of Ϫ20 mV from that evoked by any protocols mentioned above. The current signals were acquired at a bandwidth of 10 kHz using pClamp version 9.2 software and filtered with a 5-kHz low pass Bessel filter using an Axopatch 200B amplifier (Molecular Devices).
Data Analysis-Time to half-inactivation (t1 ⁄ 2 ) is the time at which 50% of peak amplitude is inactivated at the indicated voltage. The activation curve was obtained by plotting normalized conductance of peak amplitude of A-type K ϩ currents against its corresponding depolarizing membrane potential. The inactivation curve was built by plotting normalized peak amplitude of A-type currents against its corresponding conditioning membrane potential. The curve of recovery rate from steady-state inactivation was obtained by plotting the peak amplitude of A-type currents against its corresponding time intervals at a conditioning hyperpolarizing membrane potential of Ϫ110 mV. The time constant (t rec ) of recovery rate was measured by fitting the curve to a single exponential function. Average values were expressed as mean Ϯ S.E., and statistical significance was evaluated by means of the two-tailed Student's unpaired t test. The significance level for the t tests was set at p Ͻ 0.05.

Determinants of Complex Formation between PrP C and DPP6-
To facilitate our analyses of DPP6 and DPP6-containing protein complexes, two polyclonal peptide antisera, 03K1 and 03J2, were raised in rabbits (Fig. 1A). Western blot analysis of brain homogenate from WT and DPP6-deficient mice (DPP6 df5J/Rw ) (34) assessed specificity (not shown), revealing immunoreactive species of ϳ110 kDa in accord with glycosylation of fulllength DPP6 (ϳ91 kDa) (35). Immunohistochemistry demonstrated the widespread expression of DPP6 in the mouse brain (not shown), confirming previous reports (36). The crystal structure of recombinant DPP6 is dimeric (35), and protein species with an electrophoretic mobility consistent with dimers were observed in transfected cells and mouse brain homogenate after cross-linking and analysis by Western blot (Fig. 1, B and C). When N2a cells expressing endogenous PrP C were transfected with plasmids encoding DPP6 (DPP6-S and DPP6-E) (23), cross-linked, and immunoprecipitated with 7A12 (␣-PrP) and membranes were probed with 03K1, a novel species was observed with a mobility of ϳ190 kDa (Fig. 1D). These data indicate that PrP C is located in membrane subdomains that harbor dimeric DPP6 or that PrP and DPP6 exist in direct physical contact in a protein complex with a stoichiometry that totals to an M r of ϳ190,000. Reversal of cross-links yielded a species that co-migrated with full-length glycosylated DPP6-S. In a reciprocal analysis, following cross-linking, cells expressing HA-tagged DPP6-S yielded high M r species and, after reversal, a signal compatible with glycosylated PrP C (Fig.  1E). Co-immunoprecipitation was also achieved in WT mouse brain without cross-linking using either 8B4 (␣-PrP) or 03K1 for pull-down (Fig. 1F).
Subsequently, mutant alleles were used to map regions within PrP C and DPP6 required for association (Figs. 2A, 3A, and 4A). Deleting the intracellular, N-terminal portion of DPP6 ("DPP6⌬cyto") had no effect upon complex formation with PrP C (Fig. 2, A, B, and E). "secDPP6," where DPP6 residues 56 -803 are prefaced by the PrP N-terminal signal peptide, causing secretion from the cell, was not associated with complex formation with PrP C (Fig. 2, A, C, E, and F). Because this secDPP6 allele was readily detected in the culture medium, this implies a requirement for membrane association for complex formation rather than presence in the extracellular milieu. Also, anchoring to the membrane by a glycosylphosphatidylinositol moiety was not sufficient for complex formation because the assay failed to detect DPP6⅐Thy-1 complexes (37) (Fig. 2D).
To delineate the role of the DPP6 ectodomain (35) in complex formation with PrP C , we used a series of DPP6 deletions with an N-terminal HA tag (Fig. 3A). Control experiments confirmed that these DPP6 constructs were expressed at the cell surface (Fig. 3B), and following cross-linking and immunoprecipitation with an ␣-HA antibody, cross-links were reversed before blot analysis. PrP C was recovered in conjunction with all DPP6 deletion mutants tested (Fig. 3C), leading to the inference that DPP6 residues 56 -80 contribute to complex formation. This was investigated further by performing the cross-linking assay with an internally deleted form of DPP6 ⌬56 -81 (this deletion covers the same interval but extends one residue further to isoleucine 81 to maintain the length of the hydrophobic TM region). For DPP6 ⌬56 -81, the ability to form complexes was lower than for WT DPP6 (Fig. 3D), underscoring a role for the juxtamembrane region.
With regard to PrP, N-terminal deletions ⌬23-88 and ⌬32-121, which encompass most or all of the octarepeats, had subtle effects on complex formation (Fig. 4, A, B, and E). The contribution of an interval that encompasses the first ␤-strand of the PrP structure (38,39) to complex formation with DPP6 is illustrated by the performance of the ⌬32-134 PrP allele (Fig. 4, A  and B). We also created PrP alleles with more C-terminal deletion intervals; aside from theoretical caveats (see "Discussion"), these were also limited by lower expression levels than WT PrP and were not examined further. A survey of PrP alleles from genetic prion diseases defined a striking result associated with A, schematic representation of DPP6-S with the approximate locations of antibody epitopes indicated. DPP6-S is a type II membrane protein and is presented in the form, HOOC-ectodomain, TM domain, cytoplasmic tail-NH 2 , from left to right. B, transfected membrane-anchored DPP6 exists as a dimer in N2a cells. N2a cells were transfected with plasmids encoding the indicated proteins, cross-linked with 2% formaldehyde as indicated, lysed, and then analyzed by Western blotting with 03J2. Cross-linked dimeric DPP6 is the main species observed. C, DPP6 exists as a dimer in WT mouse brains. Brains from mice of the indicated strain were cross-linked in vivo using the timecontrolled transcardiac perfusion crosslinking procedure and analyzed by Western blotting 03J2. A signal at ϳ191 kDa is apparent, indicative of dimeric DPP6. D, immunoprecipitation (IP) of DPP6 by 7A12 (␣-PrP). N2a cells were transfected with DPP6-S, cross-linked with 2% formaldehyde, and lysed, and then immunoprecipitations were performed with either normal mouse serum (NMS) or 7A12. Following pull-down, a band reactive to DPP6 antibodies was observed, indicative of complex formation between PrP C and DPP6-S. After cross-link reversal, a signal was obtained compatible with full-length glycosylated DPP6 monomers. E, immunoprecipitation of PrP C by ␣-HA. N2a cells were co-transfected with HA-tagged DPP6-S and PrP C . Following crosslinking with 2% formaldehyde, high M r complexes are observed (arrowhead). PrP C ⅐DPP6 complexes were isolated by immunoprecipitation using ␣-HA, and then cross-links were reversed. F, reciprocal co-immunoprecipitation of DPP6 and PrP C from non-cross-linked mouse brain. Detergent-extracted wild-type mouse brain homogenates were subjected to immunoprecipitation with either 03K1 (left panels) or 8B4 (right panels).
the octarepeat region (33, 41) but not with mutations C-terminal to this position (42)(43)(44). octa13 PrP, an expansion of the octarepeats to a total of 13, was less efficient than WT PrP at complex formation with DPP6 (Fig. 4, A, C, and D). Densitometry revealed that octa13 PrP formed ϳ191-kDa complexes at a level of 23.3 Ϯ 16.4% that of WT PrP (100 Ϯ 12.4%; n ϭ 3, p Ͻ 0.002). As a preface to the functional studies described below, this result was confirmed in cross-linking studies of HEK293T cells (Fig. 4D). Furthermore, we confirmed that full-length HAtagged DPP6-S and PrP C co-localize at the cell surface of the HEK293T cells used for electrophysiology (Fig. 4F). Last, interactions between PrP C and DPP6 were found to take place in cis in a cell biological sense, as demonstrated by the lack of complex formation when cells expressing PrP C or DPP6 are co-cul-tured before cross-linking and lysis (Fig. 4G). Interestingly, Doppel, with a three-dimensional fold similar to that of the PrP C C terminus (40), can also form complexes with DPP6 isoforms (Fig. 5).
PrP C Modulates A-type K ϩ Currents Mediated by Kv4.2 Channel Complexes-Following co-expression of the components of the Kv4.2 channel complex (Kv4.2, KChIP2, and DPP6-S; Fig. 6A), whole-cell recordings were performed on transiently transfected HEK293T cells. Two series of A-type K ϩ currents mediated by the Kv4.2 channel complex were investigated in the presence and absence of exogenous PrP C (the isolate of HEK293T cells used here expressed endogenous human PrP C ; Figs. 6A, 8A, 9A, and 10A). As seen in Fig. 6B, an A-type outward K ϩ current was generated in response to depolarizing FIGURE 2. Membrane anchorage of DPP6 is required for complex formation with PrP C . A, schematic representation of DPP6 mutants. DPPX⌬cyto lacks all but three residues of the cytoplasmic domain, whereas secDPPX directs the secretion of the DPP6 ectodomain into the lumen/extracellular space. B-F, N2a cells were transfected with the indicated plasmids, cross-linked with 2% formaldehyde, and lysed, and complexes (arrows) were analyzed by Western blotting. B and C, the cytoplasmic domain of DPP6 is not required for complex formation with PrP C , whereas the secreted DPP6 ectodomain fails to form a complex with PrP C . D, no high M r complexes are observed between Thy-1 and DPP6. E, both DPP6⌬cyto and secDPP6 are capable of forming dimers. F, minimal amounts of wild-type DPP6-S are present in the conditioned medium, whereas large amounts of secDPP6 are found in this fraction. Accordingly, a smaller proportion of secDPP6 is found in the cell lysate.
potentials. Following a rapid rise to peak amplitude, the current rapidly decayed despite a continued depolarizing step command. In the presence of exogenous PrP C , the peak amplitude of the A-type K ϩ currents at 20 mV was larger (14.5 Ϯ 0.9 nA; average Ϯ S.E., n ϭ 17) than that mediated by the Kv4.2 channel complex in its absence (10.2 Ϯ 1.0 nA; n ϭ 21, p Ͻ 0.05) (Fig.  6C). The curve of voltage-dependent activation of A-type K ϩ currents mediated by Kv4.2 channel complexes in the presence (triangles) and absence (squares) of exogenous PrP C was created by plotting averaged normalized conductance (G/G max ) against corresponding depolarizing potential (Fig. 6D). There was no significant difference in the voltage-dependent activation between the two groups.
To establish voltage-dependent steady-state inactivation, A-type K ϩ currents were evoked by another voltage stimulus protocol (see "Experimental Procedures"). To highlight differences, we have shown two traces of A-type K ϩ currents mediated by the Kv4.2 channel complex alone (left) and with exogenous PrP C (right) at conditioning potentials of Ϫ110 and Ϫ60 mV, respectively (Fig. 6E). The ratio of normalized peak amplitudes of the currents evoked at conditioning potentials of Ϫ60 and Ϫ110 mV was 0.29 in the absence of PrP C versus 0.56 in its presence. The plot for voltage-dependent steady-state inactivation was obtained by plotting corresponding average normalized currents (I/I max ) against conditioning potential in the presence (triangles) and absence of exogenous PrP C (squares) (Fig.  6F). Significant differences between these two groups were noted at conditioning potentials of Ϫ50, Ϫ60, and Ϫ70 mV (n ϭ 10, p Ͻ 0.05).
To measure decay of A-type K ϩ current from peak amplitude to base line, we used the time at which 50% of peak amplitude was inactivated at a given depolarizing potential (half-inactivation time) to quantitatively describe the time course for inactivation (20). Two representative traces of A-type K ϩ currents evoked by a depolarizing potential of 50 mV in Fig. 6B are shown with an expanded time scale and correspond to the Kv4.2 channel complex with or without exogenous PrP C (Fig.  7A). The half-inactivation time was increased from 17.6 to 31.2 Transfected cells were treated with a commercial cell surface biotinylation kit, and avidin-captured proteins were run on a 4 -12% NuPAGE gel. Western blots were performed with an ␣-HA antibody. Lower M r bands are inferred to represent fragments that retain the HA tag but not the C-terminal epitopes that were used for polyclonal antibody generation. C, HA tag immunopurification assay to assess co-precipitation of PrP C . D, Western blot demonstrating a lack of complex formation between DPP6 ⌬56 -81 and PrP C . RK13 cells were transfected with PrP and the indicated form of DPP6, cross-linked with 0.4% formaldehyde, lysed, and analyzed using Sha31 (␣-PrP). Black arrow, ϳ191-kDa complex; open arrow, ϳ100-kDa complex.
ms by the presence of exogenous PrP C . On average, the halfinactivation time of A-type K ϩ currents mediated by Kv4.2 channel complexes in the presence of exogenous PrP C (n ϭ 8) was significantly longer than that mediated by the Kv4.2 channel complex alone (n ϭ 10) at all but two depolarizing potentials ( Fig. 7B; p Ͻ 0.05).
We noted above (Fig. 6F) for HEK293T cells expressing Kv4.2 complexes that when the conditioning potential applied is more positive than Ϫ40 mV, it is not possible to evoke an A-type K ϩ current by stepping to a depolarizing potential of 30 mV. This status can be referred to as complete voltage-dependent steady-state inactivation of Kv4.2 channels and can be  3, p Ͼ 0.3). D, HEK 293 T cells were co-transfected with DPP6-S and the indicated PrP plasmids and cross-linked with 0.4% formaldehyde. Intensity of the ϳ191 kDa band is expressed in arbitrary units: WT PrP, 100 Ϯ 8.3%; octa13 PrP, 6.7 Ϯ 3.7%. E, HEK293T cells were co-transfected with PrP ⌬32-121 and the indicated DPP6 plasmids and cross-linked with 2% formaldehyde; PrP ⌬32-121 was capable of forming complexes with both DPP6 isoforms. F, co-localization of PrP C and HA-DPP6 in HEK293T cells as assessed by immunofluorescent labeling with Sha31 (␣-PrP) and ␣-HA. Scale bar, 10 m. G, PrP C ⅐DPP6 complexes result from interactions occurring within the same cell. HEK293T cells were either co-transfected with PrP C and DPP6-S (in cis with respect to cellular disposition) or singly transfected with either PrP C or DPP6-S (in trans with respect to cellular disposition). 24 h post-transfection, cells were trypsinized and replated either alone (cis) or with PrP C and DPP6-S singly transfected cells mixed together (trans). After an additional 24-h incubation (at which time the cells were confluent), cells were cross-linked with 2% formaldehyde, and lysates were analyzed by blotting. PrP C ⅐DPP6 complexes (arrow) are only formed when PrP C and DPP6 are present in the same cell (i.e. co-transfected). DECEMBER 27, 2013 • VOLUME 288 • NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 37247 removed by first stepping to a conditioning hyperpolarizing potential (e.g. Ϫ110 mV) for a short duration. Consequently, to measure how fast A-type currents can recover from complete voltage-dependent steady-state inactivation, we recorded A-type K ϩ currents evoked by a different stimulus protocol (Fig. 7C). Representative traces of the resulting A-type K ϩ currents are shown for the presence and absence of exogenous PrP C (Fig. 7C) with conditioning potentials of Ϫ110 mV for 5 and 170 ms, respectively. The ratios of the amplitudes of the currents measured with a pulse duration of 5-ms versus 170 ms at conditioning potentials of Ϫ110 mV were 0.24 (without) and 0.58 (with exogenous PrP C ). The average recovery rate from steady-state inactivation was obtained by plotting the time of conditioning hyperpolarizing potential at Ϫ110 mV against corresponding normalized currents (I/I max ) (Fig. 7D). The recovery rate was significantly faster in the presence of exogenous PrP C (n ϭ 8) than in its absence (n ϭ 10) at the first two time intervals of 5 and 25 ms (p Ͻ 0.05). In both groups, recovery from steady state was virtually complete after 100 ms following a conditioning hyperpolarizing potential of Ϫ110 mV.

PrP C Modulates K ؉ Channels through DPP6
PrP C Complex Formation with DPP6-S Is Required for Modulation of Kv4.2 Channel Properties-To ascertain whether or not the effects of PrP C in this system are mediated through DPP6-S, we characterized the modulatory properties of PrP C upon Kv4.2-mediated A-type K ϩ currents in the absence of co-expressed DPP6-S (Fig. 8A). These A-type K ϩ currents showed smaller peak amplitudes, a right shift of the activation curve, a right shift of the steady-state inactivation curve, a longer half-inactivation time, and a slower recovery rate from steady-state inactivation. These changes in the properties of A-type K ϩ currents without DPP6-S are consistent with the previous findings observed by another group (24). However, exogenous PrP C did not have a significant effect upon any A-type K ϩ current properties in the absence of DPP6-S (Fig. 8,  B-F).
Next, we studied the impact of co-expression of octa13 PrP (Fig. 4A) with Kv4.2 complexes that include DPP6-S. Although the stimulus protocols for these studies remained the same as in previous recordings (with the exception of those used to inves-tigate recovery rate from steady-state inactivation; Fig. 7, C and D), we adjusted two aspects of the experimental design. First, the time interval for conditioning hyperpolarizing potential was reduced from the previous 20 ms to 5 ms to acquire more data points. Second, we used siRNA against human PrP mRNA to reduce the levels of endogenous PrP C (Fig. 9A); by this means, we reduced cross-talk from WT PrP C that could complicate the interpretation of data associated with transgeneencoded PrP. In this experimental context, WT PrP C was found to modulate the Kv4.2 channel complex in the same manner as our previous data (Fig. 9, B-F). However, the properties of A-type K ϩ currents in the presence of octa13 PrP showed no significant difference from those mediated by the Kv4.2 channel complex alone (Fig. 9, B-F, and Table 1). These data indicate that octa13 PrP is a loss of function with regard to modulation of Kv4.2, a finding that parallels its Ͼ75% loss in efficiency at forming cross-linked complexes with DPP6-S (Fig.  4, C and D).
In a final set of analyses, we further investigated the inference that residues 56 -80 of DPP6 are required for complex formation with PrP C . A-type K ϩ currents mediated by the Kv4.2 complex composed of Kv4.2, KChIP2, and DPP6 ⌬56 -81 showed smaller peak amplitudes, a right shift of the activation curve, a right shift of the steady-state inactivation curve, a longer halfinactivation time, and a slower recovery rate from steady-state inactivation (Fig. 10). These changes in the properties of A-type K ϩ currents are similar to those observed in the absence of DPP6-S (Fig. 8), indicating that DPP6 ⌬56 -81 is not fully functional. However, exogenous PrP C did not modulate the properties of these A-type K ϩ currents, further supporting the idea that this juxtamembrane region of DPP6 is required for complex formation with PrP C (Figs. 3 and 10).

DISCUSSION
PrP C and Ion Channels-PrP C is highly expressed in the central nervous system (45)(46)(47), and consequently, its role in regulating neuronal excitability is of great interest. Prior studies have centered upon perturbations in electrophysiological recordings made from brain slices of Prnp 0/0 mice (5-8, 16, 17, 48 -51). Here we have investigated the influence of PrP C upon Kv4.2 channel complexes by their reconstitution in HEK293T cells to study the currents produced by these complexes in isolation. These cells have been reported to have small endogenous delayed rectifier K ϩ currents (52). We found identical currents in our HEK 293T cell isolate (generally smaller than 200 pA) but did not detect A-type K ϩ currents, as previously reported by another group (53). Therefore, the A-type K ϩ currents produced here (and measuring in the nanoampere range) by reconstitution of the Kv4.2 channel complex are not contaminated by the presence of endogenous currents. More recently, expression of PrP ⌬105-125 in transfected cells has been reported to induce spontaneous non-selective, cationpermeable ion currents (54,55). In this instance, because of the striking nature of the effect observed in HEK293T cells, we undertook parallel studies with the same allele inserted in our expression vectors, but none displayed a spontaneous ionic current (n ϭ 10). To observe the currents, the cells were held at either Ϫ80 or ϩ80 mV for several min. Under these conditions, it would be unlikely to miss any spontaneous current activity (54,55). These experiments were performed both at 23 and 34°C and with pipette solutions containing either 0.5 mM EGTA or 10 mM EGTA (54,55). Although these data (not shown) did not support the concept of a solitary action of PrP for this particular allele under our designated experimental conditions, WT PrP C impacted the performance of co-expressed DPP6⅐Kv4.2 complexes, as elaborated below.
In prior analyses, PrP C has been implicated in the modulation of a variety of ion channels, including GABA A receptor/ channels (5,6,8), calcium-dependent K ϩ channels, and NMDA receptors/channels (16, 17, 48, 56, 57); stress-inducible protein-1-dependent intracellular Ca 2ϩ fluxes mediated by the ␣7 nic-otinic acetylcholine receptor (14,15,58); and an AMPA-dependent Zn 2ϩ reuptake phenomenon (59). Remarkably, these studies convey a diversity of mechanisms whereby PrP C modulates ion channels and neuronal excitability. For instance, enhanced and drastically prolonged NMDA-evoked currents in PrP C knock-out mouse neurons were the result of a functional up-regulation of NMDA receptors containing NR2D subunits (16). On the other hand, impaired and depressed Ca 2ϩ -dependent after-hyperpolarization potential in PrP C knock-out mouse neurons arises from an increased intracellular Ca 2ϩ buffering capability (48,57). In this case, the free Ca 2ϩ through influx of activated voltage-gated Ca 2ϩ channels is decreased and in turn depresses the after-hyperpolarization potential.

2-mediated A-type currents in HEK293T cells.
A, Western blots demonstrating the presence or absence of the channel components. B, a series of A-type K ϩ currents evoked by a stimulation protocol (inset) in cells transfected with the Kv4.2 channel complex with (right) and without exogenous PrP C (left). C, average peak amplitude of A-type K ϩ currents recorded at depolarizing potential of 20 mV with (n ϭ 21) and without exogenous PrP C (n ϭ 17). D, average voltage-dependent activation of A-type K ϩ currents with (triangles; n ϭ 8) and without (squares; n ϭ 10) exogenous PrP C . E, sample traces of A-type K ϩ currents evoked by a stimulation protocol (inset) with (right) and without exogenous PrP C (left). F, the curves of averaged voltage-dependent steady-state inactivation of A-type currents with (triangles; n ϭ 10) and without (squares; n ϭ 10) exogenous PrP C . *, significant difference at a given depolarizing or conditioning potential between the two groups (p Ͻ 0.05). Error bars, S.E.
Here, the modulation of Kv4.2 properties by PrP C requires interaction with DPP6-S and indicates that yet another, presently unknown, mechanism is employed. One possibility is increased trafficking of DPP6-S in the presence of PrP C . A larger current amplitude and a faster recovery time from steady-state inactivation can be attributed to an increase of DPP6-S at the cell surface (20). However, the modulation of the voltage dependence of inactivation and half-inactivation time by PrP C in this study counters the effect of DPP6-S. This suggests that the modulation of these channels by PrP C occurs, at least partially, in a manner distinct from trafficking. Before addressing the puzzle presented by the pleiotropic actions of PrP C , we will first consider molecular and mechanistic aspects of the PrP C /A-type current paradigm.
PrP C and the Kv4.2 Channel Complex-In genetic mapping of determinants necessary for PrP C -DPP6 interactions (Figs. [1][2][3][4][5], WT N-terminal sequences up to residue 121, which are considered natively unstructured in the unmetallated form of PrP (39,60), were not required. Deletions that begin to encroach on the C-terminal globular domain diminished the interaction (Fig. 4B). Although there are certain caveats concerning expression levels, chaperone interactions, and global folding that apply to the use of C-terminal deletions (61,62), as yet we have been unable to find a crucial, common segment of PrP that is required for complex formation. We infer that a natively structured PrP C globular domain (rather than a linear "epitope") is essential for complex formation. This is supported by the finding that Doppel also forms high molecular weight complexes with DPP6 (Fig. 5). For DPP6-S, the intracellular portion had no effect on complex formation with PrP C , but there was a requirement for anchoring to the cell membrane (Fig. 2). With progressive ectodomain deletions, we determined that residues 1-81 retained the ability to immunoprecipitate PrP C following cross-linking (Fig. 3, A-C). Because the 55 N-terminal residues of DPP6-S are found either on the cytoplasmic side of or spanning the membrane (and thus inaccessible to PrP C ), these data lead to an inference that the DPP6-S juxtamembrane region (residues 56 -80) either complexes directly with the globular domain of PrP or is retained in PrPenriched membrane domains by the action of an intermediary protein. This inference was bolstered by analyses of cells expressing an internal deletion (residues 56 -81) incorporated into full-length DPP6-S (Fig. 3, A and D). These findings have a potential parallel in analyses and complement data indicating that the extracellular portion of DPP6 is not necessary for modulation of Kv4.2 channel properties (63).
Membrane Protein Assemblies and Pleiotropic Actions of PrP C -To explain the curiously diverse actions of PrP C in different experimental paradigms, Linden et al. (64) hypothesized its action as a dynamic scaffold at the cell surface, one that not . Note that the time to half-inactivation is increased in the presence of PrP C . B, average half-inactivation time of A-type K ϩ currents recorded at different depolarizing potentials with (triangles; n ϭ 8) and without (squares; n ϭ 10) exogenous PrP C . C, traces of A-type K ϩ currents evoked by a stimulus protocol (bottom), consisting of conditioning hyperpolarizing potential of Ϫ110 mV for 5 and 170 ms, respectively, with (middle) and without (top) exogenous PrP C . Note that the normalized amplitude of the current recorded with a 5-ms conditioning hyperpolarizing potential to that with 170 ms is smaller in the absence of exogenous PrP C . D, average curves of recovery rate from steady-state inactivation were created by plotting normalized peak currents (I/I max ) against the corresponding time interval of conditioning hyperpolarizing potential in the presence (triangles; n ϭ 8) and absence (squares; n ϭ 10) of exogenous PrP C . *, significant difference at a given depolarizing potential or time interval of conditioning hyperbolizing potential in the two groups (p Ͻ 0.05). Error bars, S.E.
only can assemble membrane proteins in a cellular signaling microenvironment but also can impact or adjust function. Findings presented here seem compatible with this proposal as can the loss-of-function phenotype of the octa13 allele in electrophysiological assays (Fig. 9). Aside from the inefficiency of octa13 PrP at forming DPP6-S-containing complexes (Fig. 4, C  and D), our studies revealed that it is incapable of forming another type of complex. This effect is apparent in both RK13 and HEK293T cells (Fig. 4, C and D), where a complex of ϳ110 kDa containing WT PrP (indicated by open arrows) is absent for octa13 PrP. The ϳ110-kDa complex could represent a PrP C ⅐DPP6-S monomer complex or interaction with a different protein, as sug-gested by the failure to detect endogenous DPP6 in HEK293T cells (Figs. 6A, 9A, and 10A). In terms of potentially analogous effects for protein-protein interactions, it is notable that PrP with 14 octarepeats was loss of function for inhibiting ␤-cleavage of APP, whereas PrP C without the octarepeat region retained activity (65). Although a direct physical interaction between PrP C and the ␤-cleaving enzyme has been questioned (66), the concept of  . Expansion of PrP octarepeats is loss of function for effects upon Kv4.2-mediated A-type currents. A, Western blots demonstrating the presence or absence of channel components. B, average peak amplitude of A-type K ϩ currents at a depolarizing potential of 20 mV mediated by different Kv4.2 channel complexes, as indicated. C-F, averaged activation curves, steady-state inactivation curves, half-inactivation time, and recovery curves from steady-state inactivation of A-type currents mediated by Kv4.2 channel complexes. For data in C-F, for Kv4.2 complexes, n ϭ 13; for Kv4.2 channel complex plus PrP C , n ϭ 9; and for Kv4.2 channel complex plus octa13 PrP, n ϭ 12. Channel compositions in traces are denoted within insets. There is no significant difference between the properties of A-type K ϩ currents of the Kv4.2 channel complex alone and in the presence of octa13 PrP. Endogenous PrP c was reduced by application of siRNA directed against human PRNP. *, significant difference (group of Kv4.2 channel complex plus exogenous PrP C versus Kv4.2 channel complex plus octa13 PrP and/or Kv4.2 channel complex alone (p Ͻ 0.05)). Error bars, S.E. plasma membrane-signaling microdomains orchestrated by PrP C may have considerable merit.
PrP C , DPP6, and Neurologic Diseases-The Ͼ75% reduction in formation of ϳ191and ϳ110-kDa complexes by octa13 PrP is notable (Fig. 4, C and D) and serves as a useful control to measure against the performance of WT PrP C . However, two other GSS alleles tested here did not behave in the same way, and GSS pathogenesis is normally considered as "gain of function" due to misfolding of PrP. Whether the loss-of-function effect is due to an approximately one-third reduction in protein at the cell surface (50.1 Ϯ 7.5% versus 78.1 Ϯ 3.9% for octa13 versus WT, as measured by a biotinylation assay; not shown) is not clear. The electrophysiological observations made here apply to non-neuronal HEK293T cells, and it remains possible that excitable cells may behave differently, but given prominent expression of DPP6 and PrP C in the CNS, where they are located in close proximity (19), our findings do broach the question as to how modulation of A-type K ϩ currents by WT PrP C via DPP6-S might feature within a broader spectrum of neurological diseases. Although the verdict may still be out on the DPP6 locus as a significant risk factor for autism spectrum disorders (67,68) and ALS (69 -71), other possibilities remain. WT PrP C modifies the A-type K ϩ currents from reconstituted Kv4.2 channel complexes by increasing peak amplitude, shifting the voltage-dependent steady-state inactivation curve to the right (more positive membrane potential), slowing inactivation, and decreasing recovery time from steady-state inactivation. This overall impact of enhancement prompts two speculations. First, enhancement plays a critical role in the down-regulation of neuronal membrane excitability and is associated with a decreased susceptibility to seizures (27,28,72); interestingly, a reported phenotype of Prnp 0/0 mice is an increased vulnerability to drug-induced seizures (73,74). Second, our previous work has established that PrP C is essential for the modulation of neuronal excitability by A␤ oligomers in cholinergic basal forebrain neurons (75). Thus, the present findings implicate PrP C regulation of Kv4.2 channels as a mechanism that could contribute to the observed effects of oligomeric A␤ (and perhaps other types of protein aggregate assemblies (76)) on neuronal excitability and viability. FIGURE 10. PrP C fails to modulate Kv4.2-mediated A-type K ؉ currents reconstituted with DPP6 ⌬56 -81. A, Western blots demonstrating the presence or absence of channel components. B, average peak amplitude of A-type K ϩ currents at a depolarizing potential of 20 mV. C, average curves of voltage-dependent activation of A-type currents. D, averaged curves of voltage-dependent steady-state inactivation of A-type currents. E, averaged halfinactivation times of A-type currents. F, averaged curves of recovery rate from steady-state inactivation of A-type currents. For B-F, data are from HEK293T cells expressing Kv4.2, KChIP2, and DPP6 ⌬56 -81 (squares, n ϭ 8) versus cells expressing Kv4.2, KChIP2, and DPP6 ⌬56 -81 with exogenous PrP C (triangles; n ϭ 9). Endogenous PrP c was reduced by application of an siRNA directed against human PRNP. Error bars, S.E.

TABLE 1 The properties of Kv4.2-mediated A-type currents in HEK293T cells
Activation: V on , depolarizing membrane potential at which significant currents first become apparent; V 0.5 , depolarizing membrane potential, in which 50% conductance of A-type currents is activated, obtained by fitting the activation curve to a first order Boltzmann function; k, slope at midpoint of the curve. Inactivation: V 0.5 , conditioning membrane potential, in which 50% of the Kv4.2 channels were inactivated, obtained by fitting the inactivation curve to a first order Boltzmann function; K, slope at midpoint of the curve. Kinetics: t1 ⁄ 2 , the time at which 50% of peak amplitude is inactivated at the indicated membrane potential; rec , time constant of recovery curve, obtained by fitting the curve to a single exponential function.