A new K+ channel beta subunit to specifically enhance Kv2.2 (CDRK) expression.

Cloned K+ channel beta subunits are hydrophilic proteins which associate to pore-forming alpha subunits of the Shaker subfamily. The resulting alphabeta heteromultimers K+ channels have inactivation kinetics significantly more rapid than those of the corresponding alpha homomultimers. This paper reports the cloning and the brain localization of mKvbeta4 (m for mouse), a new beta subunit. This new beta subunit is highly expressed in the nervous system but is also present in other tissues such as kidney. In contrast with other beta subunits, coexpression of the mKvbeta4 subunit with alpha subunits of Shaker-type K+ channel does not modify the kinetic properties or voltage-dependence of these channels in Xenopus oocytes. Instead, mKvbeta4 associates to Kv2.2 (CDRK), a Shab K+ channel, to specifically enhance (a factor of up to 6) its expression level without changing its elementary conductance or kinetics. It is without effect on another closely related Shab K+ channel Kv2.1 (DRK1). Chimeras between Kv2.1 and Kv2. 2 indicate that the COOH-terminal end of the Kv2.2 protein is essential for its mKvbeta4 sensitivity. The functional results associated with the observation of the co-localization of mKvbeta4 and Kv2.2 transcripts in most brain areas strongly suggest that both subunits interact in vivo to form a slowly-inactivating K+ channel. A chaperone-like effect of mKvbeta4 seems to permit the integration of a larger number of Kv2.2 channels at the plasma membrane.

Voltage-gated K ϩ channels are involved in a considerable number of physiological functions including neuronal integration, cardiac pacemaking, and hormone secretion. In excitable cells, they contribute to set the membrane resting potential and determine the frequency and duration of action potentials. Electrophysiological studies have identified several subtypes of voltage-gated K ϩ channels with different kinetics, voltage-dependences, and pharmacological properties (1,2). The purification and functional reconstitution in lipid bilayers of the ␣-dendrotoxine sensitive K ϩ channels from rat brain has initially revealed that this class of channels is made of the assembly of ␣ subunits (70 -80 kDa) associated with smaller ␤ subunits (38 -42 kDa) (3)(4)(5). On the other hand, molecular biology techniques have now lead to the identification of more than 20 mammalian genes encoding K ϩ channels ␣-subunits and cloning has also recently led to new structural and functional information concerning hydrophilic K ϩ channel ␤ subunits (Kv␤) (for a review, see Ref. 6). The Kv␤ 1 family has 3 members: Kv␤ 1.1 , Kv␤ 1.2 , and Kv␤ 1.3 , which are alternative spliced products from the same gene, their calculated molecular masses are 44 -47 kDa (7)(8)(9)(10)(11)(12)(13). The other ␤ subunits, Kv␤ 2.1 and Kv␤ 3.1 , have molecular masses of 41 and 45.5 kDa, respectively (13,14). Sequence alignments indicate that Kv␤ subunits have a common conserved core (over 85% amino acids identity) and variable amino termini (40 -100 amino acids long). Kv␤ 1.2 and Kv␤ 1.3 have been cloned from heart and Kv␤ 1.1 , Kv␤ 2.1 , and Kv␤ 3.1 from brain. These auxiliary ␤ subunits specifically associate to large pore-forming ␣-subunits of the Kv1 (Shaker) subfamily to modify their functional properties. For example, the coexpression of Kv␤ 1.1 or Kv␤ 3.1 subunits with Kv1.1 or Kv1.4 ␣ subunits results in an increase of their inactivation kinetics. The proposed mechanism is a fast N-type inactivation. The ␤ subunits have an "inactivation ball" in the NH 2 -terminal part of their structure which is thought to induce a rapid inactivation by occluding the internal mouth of the pore (6,(13)(14)(15). However, Kv␤ 1.2 and Kv␤ 2.1 subunits have also been reported to increase the rate of inactivation of Kv1.4 and Kv1.5 channels, also they do not seem to contain an inactivation ball (11,12). ␤ subunits not only change channel kinetics, they can also change voltage-dependences of ␣ subunits expressed in Xenopus oocytes as recently shown for the co-injection of Kv␤ 1.3 with Kv1.5 (10). This paper describes the cloning and localization of a new K ϩ channel ␤ subunit cDNA from mouse brain. This 249-amino acid ␤ subunit is the shortest of all the Kv␤ subunits cloned till now. It is highly expressed in the brain but it is also present in different tissues, particularly in kidney. Unlike other ␤ subunits, mKv␤ 4 is without any effect on members of the Kv1 (Shaker) K ϩ channel family but it specifically enhances the expression of Kv2.2 (CDRK) in Xenopus oocytes, without affecting the inactivation properties. Therefore, its main role appears to be that of a chaperone-like protein to direct more efficiently the Kv2.2 ␣ subunit to its physiological location at the plasma membrane.
In Situ Hybridization-All animals were killed by a transcardially perfusion with 0.9% NaCl followed by ice-cold 4% (w/v) paraformaldehyde, 0.1 M sodium phosphate buffer solution (PBS, pH 7.4). The dissected brains were postfixed in the same solution for 2 h and then immersed overnight at 4°C in a 20% sucrose, PBS solution. Brains were frozen on dry ice and stored at Ϫ70°C. Sagittal frozen sections (10 m) were cut on a cryostat (Microm) at Ϫ25°C, collected on 3-aminopropylethoxysilane-coated slides, and stored at Ϫ20°C until use. To check the specificity of the labeling, two oligonucleotides and two riboprobes were used and the expression of mRNA transcripts compared.
In Vitro Transcription, Preparation of Xenopus Oocytes, and Electrophysiology-Capped cRNAs were synthesized in vitro from linearized plasmids by using the T7 RNA polymerase except for pBS-Kv␤ 4 for which the T3 RNA polymerase was used (Stratagene). Preparation of oocytes and mRNA injection have been previously described (17). In order to obtain similar current amplitudes, 12.5 ng of Kv2.2 and Kv2.2⌬C 248 cRNAs, 2.5 ng of chimeras Kv2.1/Kv2.2 and Kv2.2/Kv2.1 cRNAs, and 0.25 ng of Kv2.1 cRNA were injected into oocytes. The co-injected amount of mKv␤ 4 cRNA was always the same as that of the injected ␣ subunit cRNA. Microelectrode measurements and patchclamp recordings on oocytes were performed as described previously (17,27). In patch-clamp studies, the internal solution contained 140 mM KCl, 3 mM MgCl 2 , 5 mM EGTA, and 10 mM HEPES at pH 7.2. The external solution was standard ND96 (96 mM NaCl, 2 mM KCl, 2 mM MgCl 2 , 1.8 mM CaCl 2 , 5 mM HEPES at pH 7.4 with NaOH). To inhibit the activity of the stretch-activated channels in oocyte patches, 10 M GdCl 3 were added to the ND96 solution.
Statistical Analysis-The variability of the results was expressed as the standard error of the mean with N indicating the number of batches and n the number of oocytes contributing to the mean. Analysis of the differences was made with the Student's t test with a confidence limit as indicated (*, 0.05; **, 0.01; ***, 0.001; ****, 0.0001).
For immunoprecipitation experiments, infected cells were homogenized at 4°C during 1 h in a solubilizing buffer containing 50 mM Tris (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, and 0.1 mM iodoacetamide supplemented with 10 l of Pansorbin (Calbiochem). The insoluble material was removed by centrifugation at 12,000 ϫ g for 15 min at 4°C. Mouse anti-T7.Tag (Novagen), mouse anti-FlagM2 (Eastman Kodak Co.) monoclonal antibodies, or purified rabbit anti-Kv1.5 (Alomon labs, Israel) antibodies were added overnight at 4°C at a 100-fold dilution, then followed by addition of protein A immobilized on Sepharose CL-4B (Sigma) for 1 h at 4°C under slow rocking. Pellets were washed six times in solubilizing buffer. Immunoprecipitated proteins were resolved by SDS-polyacrylamide gel electrophoresis (10% polyacrylamide) and transferred onto nitrocellulose membranes (Hybond C-extra, Amersham). Blots were saturated with PBS supplemented with 0.1% Tween 20 and 5% lowfat dry milk for 1 h at room temperature then incubated overnight at 4°C with primary antibodies. After several washes with PBS containing 0.1% Tween 20, blots were revealed with purified goat anti-mouse or anti-rabbit horseradish peroxidase-conjugated secondary antibodies (1:10,000) (Jackson) for 1 h at room temperature followed by incubation with substrate for enhanced chemiluminescence detection (ECL, Pierce).

Cloning, Sequencing, and Tissue Distribution of Kv␤ Subunits-To
isolate new members of the K ϩ channel ␤ subunit family (Kv␤), oligonucleotide primers derived from the coding sequence of bovine Kv␤ 2 subunit (7) were used. A fragment of around 1 kb was first amplified with reverse transcriptase-PCR from mouse brain which had a high homology with the coding sequence of the bovine brain Kv␤ 2 cDNA. This fragment was labeled and used to screen a mouse brain cDNA library. Several positives clones were studied by restriction analysis and revealed the existence of three different populations of cDNA. The longest open reading frames derived from a 2.2-kb Kv␤ 1 and a 3.8-kb Kv␤ 2 cDNA encode for 401 and 367 amino acid proteins with calculated molecular masses of 44.7 and 41 kDa, respectively, in good agreement with the sizes determined by SDS-PAGE analysis for the corresponding in vitro translated products (46 and 42 kDa) (Fig. 1B). These two ␤ subunits are the equivalents in mouse of those previously cloned from rat brain (13). Our mouse clones were called mKv␤ 1 and mKv␤ 2 , they had 97 and 90% amino acid identity with ratKv␤ 1 and ratKv␤ 2 , respectively.
The sequencing of the third cDNA (1.9 kb) revealed a shorter open reading frame of only 747 base pairs. The 5Ј non-translated region contains 3 termination codons prior to the first initiation codon ATG. The predicted product consists of a 249amino acid sequence designated as mKv␤ 4 with a calculated molecular mass of 27.7 kDa which is again in good agreement with the molecular mass of 28 kDa obtained by SDS-PAGE analysis of in vitro translated products (Fig. 1B). Fig. 1A presents a protein alignment of the three cloned Kv␤ subunits. mKv␤ 4 has only 50% similarity with the other cloned ␤-subunits. The comparison clearly shows that mKv␤ 4 is the shortest of all, especially in the NH 2 -terminal portion where the most significant differences are found. In contrast with mKv␤ 1 , mKv␤ 4 does not contain in its NH 2 -terminal end the cysteine/ serine motif and the cluster of positively charged amino acids, that constitutes the inactivating ball (13). The high homology between mKv␤ 1 , mKv␤ 2 , and mKv␤ 4 starts at glycine 25 of the mKv␤ 4 sequence corresponding to glycine 173 in mKv␤ 1 . The global amino acid similarity in that part of the sequence is 83.5 and 85.3% compared to mKv␤ 1 and mKv␤ 2 , respectively. This high amino acid identity provides evidence that mKv␤ 4 belongs to the K ϩ channel ␤ subunit family.
The tissue distribution of the three subunits was investigated by Northern blot analysis ( Fig. 2A) using DNA fragments corresponding to the coding sequence of mKv␤ 1 , mKv␤ 2 , and mKv␤ 4 cDNA clones. The mKv␤ 1 probe hybridizes with a 3.5-kb transcript in the brain but this subunit is not significantly expressed in other tissues. Conversely mKv␤ 2 mRNA is highly expressed in brain but is also well detected in heart and lung and is moderately present in kidney and skeletal muscle. mKv␤ 4 is also highly expressed in brain but is also present in kidney where a 2.5-kb transcript is detected. A longer exposure of the autoradiogram revealed a lower expression of the same transcript in lung, skeletal muscle, and heart. A 4.3-kb transcript is present in kidney, lung, and brain where an additional mRNA of 6.3 kb is also found. In kidney, the mKv␤ 4 probe also hybridized to a short fragment of 1.3 kb. These various messenger species may correspond to different pre-mRNA transcripts or to isoforms of the mouse mKv␤ 4 gene. Fig. 2B shows a characteristic autoradiograph illustrating Autoradiographs were exposed for 17 h at Ϫ70°C except for glyceraldehyde-3phosphate dehydrogenase (GAPDH) for which exposure was 6 h. A blot was hybridized with a glyceraldehyde-3-phosphate deshydrogenase probe for relative quantification. The size of messengers (kb) are indicated by arrows. B, x-ray film autoradiographs illustrating expression patterns of mKv␤ 4 mRNAs in sagittal mouse brain sections following in situ hybridization with specific mKv␤ 4 probes. Scale bar, 1 mm. Abbreviations: Cx, cortex cerebral; Ce, cerebellum; Hi, hippocampus; OB, olfactory bulb. the differential distribution of mRNAs encoding mKv␤ 4 potassium channel subunit in sagittal sections of adult mouse brain obtained with both radiolabeled cRNA and synthetic oligonucleotides of the mouse Kv␤ 4 homologue. Expression of the mKv␤ 4 mRNA was ubiquitous throughout the brain, but was clearly concentrated in certain anatomically distinct regions corresponding to gray matter regions. Higher expression levels appeared in the neo-and allocortical regions, hippocampus, olfactory bulb, and cerebellum. In the olfactory bulb, intense labeling was evident in the granule cells. A strong signal was observed throughout the cell layers of the cerebral cortex. A uniformly high hybridization signal was found in all the fields of the hippocampal formation. Transcripts were localized in the CA1-CA4 pyramidal cell layer as well as in the granule cells of the dentate gyrus. Very strong labeling in all lobules of the cerebellum was present and was confined to the granular layer and to the Purkinje cell layer. The molecular layer was very weakly labeled. No labeling was seen in the deep cerebellar nuclei. A diffuse hybridization signal was observed over most other regions including amygdaloid complex, thalamic nuclei, caudate-putamen, and globus pallidus of the basal ganglia, hypothalamus, midbrain, and brainstem (pons and medulla). The microscopic analysis of emulsion-dipped sections performed at higher magnification in the most strongly labeled brain areas (olfactory bulb, hippocampus, and cerebellum, data not shown) shows that the expression of the mKv␤ 4 potassium channel subunit correspond exactly to the specific expression of the Kv2.2 (CDRK) ␣ subunit in the brain (22,29,30).
Functional Expression of mKv␤ 4 -Rat Kv␤ 1 and Kv␤ 2 have been given a role in the regulation of Shaker type K ϩ channels (Kv1). In order to test for the functional role of mKv␤ 4 , this subunit was first coexpressed in Xenopus oocytes with members of the five most classical structural families of voltage-dependent K ϩ channel (Kv1 to Kv4, and IsK). Neither the peak currents recorded during voltage pulses to ϩ50 mV, nor the activation or inactivation time constants were significantly modified, in the presence of mKv␤ 4 , for Kv1.1 (RCK1), Kv1.2 (RCK5), Kv1.3 (HLK3), Kv1.4 (RCK4), Kv1.5 (RCK7), Kv3.4 (RAW3), or Kv4.1 (data not shown). Similarly no modification has been recorded for the expression of the other member of the Shab subfamily, Kv2.1 (DRK1) (Fig. 3A). In addition, mKv␤ 4 had no effect on K ϩ channel activity generated in Xenopus oocytes by hIsK (not shown) another type of membrane protein with a single transmembrane segment which generates slow K ϩ channel activity when expressed in oocytes (18). On the other hand, Kv2.2 (CDRK) expression was largely enhanced by co-injection of mKv␤ 4 . Fig. 3B shows the averaged peak currents recorded 4 days after the injection of the Kv2.2 cRNA alone or together with mKv␤ 4 cRNA. In these 19 oocyte batches mKv␤ 4 produced an average increase of more than 5-fold of the Kv2.2 expression (p Ͻ 0.0001). Fig. 3C shows the currentpotential relationship in the absence or presence of the subunit. The Kv2.2 activation threshold was unchanged in the presence mKv␤ 4 (21 Ϯ 2 and 21 Ϯ 1 mV (n ϭ 9), respectively). Fig. 3D shows recorded currents 4 days after cRNA injection. The Kv2.2 activation time constant (24.0 Ϯ 0.2 ms, n ϭ 17) was slightly diminished by the coexpression with the new ␤ subunit (21.3 Ϯ 0.3 ms, n ϭ 17). Fig. 4A shows the Kv2.2 currents recorded in outside-out patch in the presence or absence of the mKv␤ 4 subunit. The Kv2.2 channel slope conductance (14.7 Ϯ 1.1 pS, n ϭ 3) was unchanged in the presence of mKv␤ 4 (14.9 Ϯ 0.5 pS, n ϭ 3). However, in this particular series of experiments, the percentage of patches that contained at least one active channel was increased 3-fold in the presence of the ␤ subunit (Fig. 4B).
The influence of mKv␤ 4 on the time course of the functional expression of the Kv2.2 channel in Xenopus oocytes has been studied with the two-microelectrode technique over a period of 8 days after cRNA injection (Fig. 5A). When Kv2.2 was expressed alone the maximum current reached a peak about 2 days after injection and then stabilized. Expression of the Kv2.2 current in the presence of mKv␤ 4 was of course much higher and reached a steady state about 3 days after injection. In other series of experiments, Kv2.2 and mKv␤ 4 cRNAs were injected separately. In the particular series of experiments shown in Fig. 5B, a 3-fold increase of the Kv2.2 current level was observed when both cRNAs were injected at the same time, at the same place, with the same pipette. Injection of mKv␤ 4 1 day before Kv2.2 injection lead to more than a 2-fold increase of the K ϩ current. Conversely, the mean peak current was increased, but only slightly (a factor of 1.5) when Kv2.2 and the ␤ subunit were injected at the same time but with 2 separate pipettes or when mKv␤ 4 was injected 1 day after Kv2.2 injection.
One first possibility to explain the stimulatory effects of One particularly interesting aspect of the work is that mKv␤ 4 enhances specifically Kv2.2 (CDRK) expression and not Kv2.1 (DRK1) expression, the second member of the Kv2 potassium channel subfamily. Kv2.1 and Kv2.2 have a global amino acid identity of 61.2%. The NH 2 -terminal cytoplasmic domain and the hydrophobic core which contains six transmembrane segments have a 84.2% amino acid identity but the remaining COOH-terminal cytoplasmic portion starting at threonine 535 in Kv2.1 displays only 21.0% of identity. Therefore we suspected that this part of the sequence was important for the sensitivity of Kv2.2 to mKv␤ 4 . Fig. 6A represents chimera constructs between the Kv2.2 and the non-mKv␤ 4 sensitive Kv2.1 ␣ subunits. Since it was known that removal of the last 318 COOH-terminal acids (after Thr-535) encoded by Kv2.1 still allowed the expression of functional K ϩ channels (31), we deleted the corresponding region on the Kv2.2 protein by constructing Kv2.2⌬C 248 , a chimera which lacks the 248 last amino acids after Ser-554 and we then constructed the Kv2.1/ Kv2.2 and Kv2.2/Kv2.1 chimeras by exchanging the COOHterminal portions (see Fig. 6A). Fig. 6B shows the effects of mKv␤ 4 4 on Kv2.2 expression suggested a physical association between these two subunits. To test this possibility, both subunits were expressed in insect Sf9 cells using the baculovirus expression system (Fig. 7A) and their association was shown by co-immunoprecipitation (Fig. 7B). To allow the immunodetection of proteins, "tag" sequences were added to the coding sequences (FlagM2 for Kv2.2 and T7.TAG for mKv␤ 4 ). From Bac-mKv␤ 4 -infected cells, a major band was detected by anti-T7.TAG antibodies with the expected molecular mass of 29 kDa. Different bands ranging from 90 to 140 kDa were detected by anti-FlagM2 antibodies from Bac-Kv2.2-infected cells (Fig. 7A). The major band with an apparent molecular mass of 97 kDa corresponds to the Kv2.2 band detected in the brain (30). Other bands could correspond to different glycosylated forms. No signal was obtained from control Sf9 cells infected with Bac-IsK, a baculovirus designed to express the IsK protein (28). From Sf9 cells co-infected with Bac-mKv␤ 4 and Bac-Kv2.2, mKv␤ 4 and Kv2.2 were co-precipitated by anti-T7.TAG antibodies (anti-"tagged" mKv␤ 4 ), whereas Kv2.2 was not precipitated by anti-T7.TAG from Sf9 cells infected with Bac-Kv2.2 alone (Fig. 7B). Conversely, both proteins were coimmunoprecipitated by anti-FlagM2 antibodies (anti-tagged Kv2.2) from co-infected Sf9 cells (data not shown). These results clearly demonstrate the physical association of mKv␤ 4 and Kv2.2 subunits. It was shown recently that Kv␤ 1 interacts with Kv1 ␣ subunits via its conserved COOH terminus (32). The amino acid similarity between Kv␤ 1 and mKv␤ 4 in this particular region being about 85%, it was important to test whether mKv␤ 4 could interact with ␣ subunits of the Kv1 subfamily. A recombinant baculovirus referred to as Bac-Kv1.5 was constructed to express Kv1.5 in insect cells. From Bac-Kv1.5-infected Sf9 cells, anti-Kv1.5 antibodies detected a band whose apparent molecular mass of 67 kDa was in agreement with the expected size (20) (Fig. 7A). The association of Kv1.5 and mKv␤ 4 was tested by immunoprecipitation from cells infected with Bac-mKv␤ 4 and Bac-Kv1.5. Fig. 7C shows that anti-Kv1.5 antibodies coprecipitate Kv1.5 and mKv␤ 4 subunits and that mKv␤ 4 cannot be precipitated by anti-Kv1.5 antibodies in the absence of Kv1.5. The formation of mKv␤ 4 -Kv1.5 complexes was confirmed by co-immunoprecipitation reactions using anti-T7.TAG antibodies (anti-tagged mKv␤ 4 ) (data not shown). These results demonstrate that mKv␤ 4 and Kv1.5 ␣ subunit can associate in spite of the lack of any functional effect. DISCUSSION Expression studies have shown that many of the cloned ␣ subunits of voltage-dependent ion channels (Na ϩ , Ca 2ϩ , and K ϩ ) were functional on their own, when expressed in Xenopus oocytes, but that they then exhibited pharmacological and biophysical properties sometimes different from native channels. In the last years, many auxiliary subunits have been described (see for reviews, Refs. [33][34][35]. Their physiological role is a regulation of the function of the ␣ subunits to which they associate. So far, auxiliary subunits ␤ 1 and ␤ 2 have been cloned for voltage-sensitive Na ϩ channels (see Ref. 36, for review). One of their primary roles has been shown to stabilize Na ϩ channels expression at the plasma membrane resulting in a 5-10-fold increase in the number of active voltage-sensitive Na ϩ channels (36). Not only the functional effects of these ␤ subunits lead to an increased peak current, but it is also associated with a change of activation and inactivation kinetics and a modification of the voltage dependence of the inactivation process (see Ref. 33, for review). Voltage-sensitive Ca 2ϩ channels are formed by a complex of subunits designated as ␣ 1 , ␣ 2 , ␦, ␤, and ␥. The ␣ 1 subunit is the pore-forming protein. The other subunits have a regulatory function, they are essential for a solid expression of the ␣ 1 subunit of different types of channels such as cardiac L-type Ca 2ϩ channels, N-type Ca 2ϩ channels, or P-type Ca 2ϩ channels (see Ref. 33). The modulating effects of the ␤ subunits also results in a change of both the biophysical and pharmacological properties of the ␣ 1 subunit (36). Auxiliary subunits called ␤ subunits (Kv␤) have also been identified for voltage-sensitive K ϩ channels (3,37) as well as for the Ca 2ϩ -activated K ϩ channels (BK) (38).
Coexpression of some of the cloned Kv␤ subunits confers rapid A-type inactivation on non-inactivating delayed rectifiers Kv1 channels when expressed in the Xenopus oocyte (11)(12)(13)(14). K ϩ channel ␣ and ␤ subunits seem to use inactivating domains that have similar properties. A-type Kv channels have an inactivating domain ("␣ ball") in their NH 2 -terminal sequence that can rapidly obturate the internal mouth of the channel upon membrane depolarization (39). A ball and chain mechanism has been proposed to explain the NH 2 -terminal (N-type) inactivation of A-type ␣ subunits (40). Kv␤ 1 subunits also contain a "␤ ball" in their NH2-terminal end with structural similarities to the ␣ ball. Therefore, both ␣ and ␤ subunits seem to provide a "ball domain" for an inactivating N-type mechanism. The ␤ ball contains a critical cysteine residue which seems to be sensitive to the intracellular redox state as well as a cluster of positively charged amino acids (6). This paper describes a new member of the Kv␤ subunit family, mKv␤ 4 . With its 249 amino acids, it is the shortest of all the previously cloned subunits. It has no inactivating ball motif in its amino terminus, suggesting that it cannot modify the inactivation properties of non-inactivating or slowly inactivating or even rapidly inactivating K ϩ channels. Indeed, mKv␤ 4 did not modify the electrophysiological expression in Xenopus oocytes of a variety of Kv ␣ subunits such as Kv1.4 or Kv1.5 whose properties are altered by other previously cloned ␤ subunits.
It turns out that the sequence of mKv␤ 4 is very similar to the sequence of a rat Kv␤ 3 (14). However, a major difference exists between these two ␤ subunits at the NH 2 terminus. The NH 2terminal end of rKv␤ 3 has the ␤ ball motif (14), whereas the NH 2 -terminal end of mKv␤ 4 does not have it. It is therefore not very surprising that rKv␤ 3 , unlike mKv␤ 4 , changes the kinetics of a non-inactivating ␣ subunit such as the NH 2 -terminal deleted form of Kv1.4. Unlike mKv␤ 4 , rKv␤ 3 is not expressed in kidney or in hippocampus and dentate gyrus (14). mKv␤ 4 assayed in the Xenopus oocyte expression system had no effect on Kv1.1, Kv1. baculovirus-infected insect cells. The function of this ␤ subunit seems to facilitate the trafficking of this particular ␣ subunit toward the plasma membrane. It could even be that Kv2.2 is normally unable to go to the plasma membrane in the absence of mKv␤ 4 and that its successful expression is due to the presence, in oocyte, of an endogenous ␤ subunit analogous to mKv␤ 4 . This sort of situation has recently been described for GIRK1, one of the major G-protein regulated inward rectifiers which expresses at the oocyte membrane only because it forms an heterologous assembly with another endogenous Xenopus subunit (41).
That one subunit facilitates the access and integration of another subunit in the plasma membrane is not surprising and there are many other precedents of this situation in the channel field. For example, the amiloride-sensitive Na ϩ channel is formed of three subunits ␣, ␤, and ␥. In the colon (42,43), it is the association with ␤ and ␥ that brings the functional subunit ␣ to the plasma membrane. The ␣ subunit by itself, hardly integrates into the surface membrane. Another example is channel-inducing factor (44). This protein with a single transmembrane segment associate with K ϩ channel subunits intrinsic to the oocyte to reveal a slow K ϩ channel activity which is not detectable in its absence. The amplitudes of the Ca 2ϩ currents of the ␣ 1 subunits of class A, class B N-type, class D L-type, and class E brain Ca 2ϩ channels are also greatly increased by coexpression of ␣ 2 ␦ and ␤ subunits (see Ref. 33, for review).
The striking property of mKv␤ 4 is that it distinguishes between the two clones in the Shab subfamily Kv2.1 (DRK1) and Kv2.2 (CDRK). Not only Kv2.1 and Kv2.2 display very similar delayed rectifier properties, but they also have a very high sequence homology in the NH 2 -terminal cytoplasmic domain (86% identity), in core transmembrane domains (95% identity) as well as in the first 195-amino acid sequence of the COOH terminus immediately after the transmembrane domain (63% identity). However, the last amino acids in the COOH-terminal part of the COOH-terminal cytoplasmic domain are very different in the two channel structures. The fact that mKv␤ 4 selectively enhances the expression of Kv2.2 and not of Kv2.1 is a first indication that the COOH-terminal part of Kv2.2 could be a target for association with mKv␤ 4 and indeed removal of the COOH-terminal end of Kv2.2 or replacement of this COOHterminal end by the COOH termini of Kv2.1 abolishes the enhancing effect of mKv␤ 4 .
It was recently demonstrated that the COOH terminus of Kv␤ 1 associates to the NH 2 terminus of Kv1 ␣ subunits (15,32). Although mKv␤ 4 has a truncated COOH terminus relative to Kv␤ 1 , the amino acid similarity in this region is about 85%. Consequently, the observed association of mKv␤ 4 with Kv1.5 in infected insect cells is not surprising. However, the two types of interaction, mKv␤ 4 -Kv2.2 and mKv␤ 4 -Kv1.5, seem to be of a different nature. mKv␤ 4 probably associates to the NH 2 terminus of Kv1.5 as shown for Kv␤ 1 while it probably associates to the COOH terminus of Kv2.2 as suggested by the mutagenesis study presented above. The different types of associations with the 2 different families of Kv ␣ subunits could explain that the functional effect of mKv␤ 4 was only seen on Kv2.2 expression and was absent on the expression of Kv1.5 or other Kv1 ␣ subunits.
Kv2.1 and Kv2.2 have different distributions in the brain and the distribution of mKv␤ 4 coincides with the distribution of Kv2.2. High expression levels of mKv␤ 4 are present in hippocampus, olfactory bulb, and cerebellum. As for Kv2.2 (22), intense labeling for mKv␤ 4 was observed in the granule cells. As for Kv2.2 (22), transcripts of mKv␤ 4 are present in the pyramidal cell layer of the CA 1 -CA 4 region of hippocampus as well as in the granule cells of the dentate gyrus. As for Kv2.2 (22), a prominent signal associated with mKv␤ 4 is present in the cerebellum particularly in granular cells and not in the molecular layer. Moreover, as Kv2.2, mKv␤ 4 is mainly expressed in the brain, is expressed at a lower level in the kidney, and is not expressed at all in heart or in skeletal muscle while Kv2.1 is a very ubiquitous channel expressed at high level not only in the brain but also in the kidney, heart, and skeletal muscle. The coincidence of these distributions of Kv2.2 and mKv␤ 4 strongly suggests that in most places where it is produced, Kv2.2 associates with mKv␤ 4 to form a K ϩ channel. When they are present in the same neuronal cells, Kv2.1 and Kv2.2 are differently distributed (29,30). Kv2.2 seems to be predominantly localized diffusely in the soma and in fibers while Kv2.1 seems to be present mainly in soma and dendritic processes. It may well be that one prominent role of mKv␤ 4 is to help Kv2.2 find its normal functional localization in the neuronal cell.
Experiments presented in Fig. 5 clearly shows that the injection of Kv2.2 and mKv␤ 4 at the same time with the same pipette is required for an efficient increase of Kv2.2 expression. Clearly the two subunits have to assemble first before integration occurs at the surface membrane to form a K ϩ channel. Making the assembly difficult by injecting the two different subunits at the same time but at different location in the oocyte, or at two different times, is counter-productive for Kv2.2 expression.
To our knowledge, this is the first example of a ␤ subunit that physically associate to Kv2.2, a pore-forming subunit which does not belong to the Shaker (Kv1) subfamily. Association of mKv␤ 4 with Kv2.2 does not alter the kinetic properties but increase the ␣ subunit synthesis and/or stability. While this paper was submitted for publication, Shi et al. (45) reported an increase in Kv1.2 surface expression due to Kv␤ 2 expression. One chaperone-like effect of Kv␤ 2 is to increase stability of Kv␤ 2 -Kv1.2 complexes as suggested for mKv␤ 4 -Kv2.2. On the other hand, mKv␤ 4 is also able to bind Kv1.5 subunit without the alterations observed on Kv2.2 expression. The exact role and physiological consequence of this association remains to be determined in tissues expressing other Kv␤ subunits.