A Novel Class of Ligand-gated Ion Channel Is Activated by Zn 2 (cid:1) *

In mammals, the superfamily of “Cys loop,” ligand-gated ion channels (LGICs), is assembled from a pool of more than 40 homologous subunits. These subunits have been classified into four families representing channels that are gated by acetylcholine, serotonin, (cid:2) -aminobu-tyric acid, or glycine. By searching anonymous genomic sequence data for exons that encode characteristic motifs of the channel subunits, we have identified a novel LGIC that defines a fifth family member. Putative exons were used to isolate a full-length cDNA that encodes a protein of 411 amino acid residues. This protein (ZAC) contains all of the motifs that are characteristic of Cys loop channel subunits but cannot be assigned to any of the four established families on the basis of sequence similarity. Genes for ZAC are present in human and dog but appear to have been lost from mouse and rat genomes. Transcripts of ZAC subunits were detected in human placenta, trachea, spinal cord, stomach, and fetal brain. Transfection of human embryonic kidney cells with ZAC subunit cDNA caused expression of spontaneous current. By screening with a broad range of potential agonists and antagonists, we determined that tubo-curarine inhibits the spontaneous current whereas Zn 2 (cid:1) activates the expressed receptors. The absence of Zn 2 (cid:1) activated channels in rats and mice may explain why this fifth

In mammals, the superfamily of "Cys loop," ligandgated ion channels (LGICs), is assembled from a pool of more than 40 homologous subunits. These subunits have been classified into four families representing channels that are gated by acetylcholine, serotonin, ␥-aminobutyric acid, or glycine. By searching anonymous genomic sequence data for exons that encode characteristic motifs of the channel subunits, we have identified a novel LGIC that defines a fifth family member. Putative exons were used to isolate a full-length cDNA that encodes a protein of 411 amino acid residues. This protein (ZAC) contains all of the motifs that are characteristic of Cys loop channel subunits but cannot be assigned to any of the four established families on the basis of sequence similarity. Genes for ZAC are present in human and dog but appear to have been lost from mouse and rat genomes. Transcripts of ZAC subunits were detected in human placenta, trachea, spinal cord, stomach, and fetal brain. Transfection of human embryonic kidney cells with ZAC subunit cDNA caused expression of spontaneous current. By screening with a broad range of potential agonists and antagonists, we determined that tubocurarine inhibits the spontaneous current whereas Zn 2؉ activates the expressed receptors. The absence of Zn 2؉activated channels in rats and mice may explain why this fifth member of the LGIC superfamily has evaded detection until now.
Ion channels that are gated by acetylcholine, serotonin, ␥-aminobutyrate, and glycine are a superfamily of homologous neurotransmitter receptors (nACh, 1 5-HT 3 , GABA A/C , and glycine receptors, respectively). Each of these four receptor families is composed of multiple receptor subtypes derived from distinctive assemblies of five homologous subunits (1,2). Although a wide variety of subunit genes has been identified in mammalian genomes (42 in human), each subunit can be assigned to one of the four receptor families on the basis of sequence similarity and, in most cases, functional activity. Many of these subunits confer unique properties to recombinant receptors in which they assemble, properties that have often been poorly defined in prior studies of native tissues (3,4). Consequently, the identification of novel receptor subunits continues to reveal unexpected properties of this receptor superfamily.
Most of the known subunits were identified by traditional methods of cross-hybridization or expression cloning. More recently, data bases of ESTs have been used to identify more distantly related homologues through computational searches for characteristic structural motifs (5)(6)(7). However, all of these methods have inherent limitations that can hinder the identification of a complete gene family. This is particularly relevant for subunits that are distantly related, that confer unexpected properties, and that are expressed at low levels in discrete locations or for short time periods during development. For such subunits, identification may only be possible through the analysis of a completely sequenced genome. Previously, during the early stages of sequencing the human genome, we used the draft sequence data to identify an elusive subunit of 5-HT 3 receptors (8). Here we have used a similar approach to uncover a very distinctive member of the subunit superfamily. This subunit cannot be assigned to any of the known receptor families on the basis of sequence similarity or functional properties. It appears to represent an old and distinct branch of the receptor superfamily that displays unique functional properties. Moreover, it also appears to be the first example of a subunit from this receptor superfamily that is expressed in human tissues but has been lost from at least some rodent species.

EXPERIMENTAL PROCEDURES
Isolation of the Human ZAC Subunit cDNA-A consensus peptide sequence of subunits for 5HT 3 and nACh receptors was used to search the nr and htgs data bases using the TBLASTN algorithm (www.ncbi. nlm.nih.gov/blast/). Homologous peptide fragments were identified within the six-frame translation of a human genomic DNA fragment (GenBank™ accession number AC018665). Oligonucleotide primers were designed from the genomic sequence to amplify the 5Ј-and 3Јflanking sequences from fetal brain and spinal cord cDNA libraries using the Marathon system (Clontech). Amplification at 95°C for 45 s, 60°C for 60 s, and 72°C for 2 min was performed for 35 cycles using the XL-PCR system (PerkinElmer Life Sciences). Reaction products were purified from agarose gels and sequenced directly. The open reading frame of the new subunit cDNA (termed ZAC) was amplified from a spinal cord cDNA library as described above using primers containing nucleotides 1-21 (sense) and 1266 -1289 (antisense) of the ZAC subunit cDNA sequence (GenBank™ accession number AF512521). The cloned product was sequenced over its entire length to ensure that no mutations had been introduced. Sequence alignments were generated by the ClustalW program from the MacVector package of sequence analysis software (Oxford Molecular Group). A cladogram was constructed using the neighbor-joining method with pairwise distances measured by absolute differences and gaps ignored. The bootstrap consensus was generated using 1,000 replications.
Northern Blot Analysis-Samples of ϳ2 g of poly(A) ϩ mRNA (Clontech) underwent electrophoresis on a 1.2% formaldehyde agarose gel, were transferred to nylon membranes, and were hybridized with an * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF512521-AF512523.
Cloning of Orthologous Genomic Fragments from Mouse, Rat, and Dog-Fragments of genomic DNA were amplified using primer pairs that correspond to the 3Ј-ends of genes encoding the mouse, rat, and dog orthologues of the human galanin receptor 2 and KIAA1067 genes. For mouse and rat, these were 5Ј-CACCCGCACTTCCCAACTGCACA and 5Ј-CTCATCATGGTGCCTGGTCAGGTA (nucleotides 1-23 and 5146 -5169 of GenBank™ accession number AF512522). For dog, they were 5Ј-CCGACGGTTAATGCGACCTGAGGA and 5Ј-GGAGCAGGTGGGC-GACATGATCGA (nucleotides 1-24 and 5846 -5869 of GenBank™ accession number AF512523). Aliquots of DNA (50 ng) from mouse, rat (Clontech), or dog (Novagen) were amplified using the XL-PCR system (Applied Biosystems). Reaction products were purified from agarose gels and sequenced directly.
Cell Culture and Transfection-HEK cells were grown in Dulbecco's modified Eagle's medium, supplemented with 10% calf serum, 100 IU/ml penicillin, and 100 g/ml streptomycin. When cells approached confluence, they were seeded into 35-mm diameter dishes and transfected with cDNAs encoding the human ZAC subunit (in pcDNA1.1/amp) and GFP (in pCDM8). Cells were transfected using calcium phosphate precipitation (6). Cells were used 24 -44 h after transfection.
Electrophysiology-The whole-cell patch-clamp technique was used to record currents from HEK cells. The bath was continuously perfused . Junction potentials were nulled prior to each experiment. Their inappropriate compensation was ignored in graphs of current-voltage relationships, but equilibrium potential values in the text were corrected. Cells were clamped at Ϫ60 mV unless otherwise stated. Drugs were applied either by pressure ejection from modified micropipettes or by bath perfusion. Experiments were performed at 22-24°C. Acquisition and Analysis of Data-Currents were amplified (Axopatch 200A, Axon Instruments), low pass-filtered at 1 kHz, and digitized (Digidata 1320, Axon Instruments, Foster City, CA) for acquisition onto the hard drive of a personal computer. Currents were averaged, superimposed, and measured using pCLAMP software (Axon Instruments). Zn 2ϩ concentration-response data were obtained by prolonged (2 s) pressure ejection of randomized agonist concentrations from low resistance pipettes as described previously (10). Zn 2ϩ -activated currents often exhibited run-up. To compensate for this, 1 mM Zn 2ϩ was applied before each concentration of Zn 2ϩ . The amplitudes of the Zn 2ϩ -activated currents were subsequently normalized to the cur- rent elicited by the prior application of 1 mM Zn 2ϩ . Graphs of concentration-response relationships were fitted using the logistic function as described previously (10). Current-voltage relationships were analyzed by averaging at least two currents recorded at each holding potential. Individual current-voltage relationships were plotted, and a linear fit to points either side of current reversal yielded the equilibrium potential. All data are expressed as the arithmetic mean Ϯ S.E., and statistical comparisons were made using the Student's t test.

RESULTS
Cloning of the Human ZAC Subunit cDNA-Searches of the draft human genomic sequence with a consensus sequence of conserved residues revealed exons of a novel LGIC subunit within unannotated fragments of a bacterial artificial chromosome clone from chromosome 17q23. The exon sequences were used to amplify flanking exons and a contiguous cDNA containing the complete open reading frame of the subunit. The subunit forms Zn 2ϩ -activated channels (see below) and was termed ZAC. The ZAC subunit cDNA encodes a polypeptide of 411 amino acid residues (Fig. 1A). It has a signal sequence, a Cys-Cys motif, four predicted transmembrane domains, and several invariant residues that underpin the conserved secondary structure of the Cys loop LGIC superfamily (11). However, the sequence of the ZAC subunit is not closely related to any known subunits of this superfamily (maximum of 15% amino acid identity with 5-HT 3A and nAChR_␣7 subunits; Fig. 1A). This is much lower than the level of identity between known subunits of 5-HT 3 receptors (30 -40%) or nACh receptors (25- FIG. 2. Spontaneous currents mediated by ZAC are blocked by TC. A, I spont recorded from an HEK cell expressing recombinant ZAC appeared rapidly upon establishing the whole-cell configuration at Ϫ60 mV. Uncompensated capacitive currents appear during whole-cell recording in response to Ϫ5 mV steps. Leak currents were negligible in cells expressing GFP alone. B, TC (100 M), applied with pressure for 2 s, reduced I spont in cells expressing ZAC but had no effect on cells expressing GFP alone. C, the TCinhibited current exhibits rectification. The TC-inhibited current amplitude was plotted against holding potential. Data points are average current amplitudes from three recordings normalized to the amplitude of currents recorded at Ϫ60 mV. Vertical bars represent Ϯ S.E.

FIG. 3. Zn 2؉ activates ZAC.
A, leak subtracted Zn 2ϩ (1 mM)-activated currents, recorded from a cell expressing ZAC, with intracellular and extracellular solutions containing equal K ϩ and Na ϩ concentrations (ϳ140 mM), respectively. Traces are averages of two currents recorded at each potential. The graph illustrates that I Zn exhibited pronounced outward rectification with equilibrium potentials dependent on the concentration of intracellular K ϩ . I Zn were recorded from cells expressing ZAC with electrode solutions containing either 140 mM K ϩ (open circles) or in which K ϩ was substituted by equimolar NMDG ϩ (70 mM; filled circles). B, a graph illustrating the timedependent run-up of Zn 2ϩ (1 mM for 2 s)activated currents. Data points fitted with a sigmoid function were I Zn amplitudes at each time point normalized to the maximum I Zn recorded from each cell (n ϭ 9). The inset graph illustrates a lack of I spont run-up recorded from the same cells. I spont amplitudes were normalized to the maximum I spont recorded from each cell. Representative current traces recorded from a single cell are shown in the right panel.
90%). Consequently, the ZAC subunit appears to represent a distinct species that diverged from the common ancestral subunit of the present day nACh and 5-HT 3 receptors (Fig. 1B).
A search of ZAC subunit cDNA and genomic sequences against human ESTs (www.ncbi.nlm.nih.gov/dbEST/index. html) revealed a surprisingly large number (Ͼ150) that contained exons of the ZAC subunit gene (encoding residues 181-411; Fig. 1A). However, almost all of these ESTs contain intronic sequences and are derived from mRNA that is transcribed in the opposite direction to the ZAC subunit mRNA. These ESTs represent transcripts of the KIAA1067 gene, which is located downstream of the ZAC subunit gene in the opposite orientation (see below, Fig. 6). The 3Ј-untranslated region of the KIAA1067 mRNA is complementary to both protein-coding and 3Ј-untranslated regions of the ZAC subunit mRNA and terminates within intron 5 of the ZAC subunit gene. Transcripts of the KIAA1067 gene are ubiquitous and abundant in human tissues. Reverse transcription/PCR, with primers from the first and last exons of the ZAC subunit gene, yielded a 1.27-kb ZAC subunit cDNA from prostate, thyroid, trachea, fetal whole brain, spinal cord, placenta, and stomach but not from adult whole brain, heart, liver, spleen, or kidney cDNA (data not shown). The presence of ZAC subunit mRNA (1.5 kb) in placenta and stomach was confirmed by Northern analysis, using an antisense riboprobe that was derived from the 5Ј-end of the ZAC subunit cDNA (Fig. 1C).
ZAC Subunits Form Homomeric Channels That Open Spontaneously-After transfection with ZAC and GFP cDNAs, HEK cells displayed spontaneous currents (mean amplitude ϭ Ϫ180 Ϯ 76 pA) immediately after achieving the whole-cell configuration at Ϫ60 mV. These were not seen in recordings from cells expressing GFP alone ( Fig. 2A). This suggests that ZAC subunits form channels that can open in the absence of agonist. The existence of spontaneous current (I spont ) enabled us to screen for compounds that have either positive or negative intrinsic activity.
In the absence of an agonist, we exploited the I spont to test antagonists of related LGICs for negative activity at ZAC. When applied by bath, strychnine (100 nM), bicuculline methiodide (10 M), ␣-bungarotoxin (10 g/ml), mecamylamine (10 M), and ondansetron (1 nM) had no effect on I spont (n Ն 2). By contrast, TC (100 M), a relatively nonselective inhibitor of nACh and 5-HT 3 receptors, inhibited I spont when applied to the bath or by local pressure ejection (n ϭ 6, Fig. 2B). The I spont that was blocked by TC had an equilibrium potential of Ϫ2 Ϯ 3 mV (n ϭ 3, Fig. 2C). Currents were larger at positive potentials than at corresponding negative potentials, indicating that I spont is outwardly rectifying (Fig. 2C).
Activation of ZAC Channels by Zn 2ϩ -Zn 2ϩ -activated currents (I Zn ) had an equilibrium potential (E Zn ) of Ϫ5 Ϯ 1 mV (n ϭ 6) and outward rectification similar to that of I spont (Figs. 2C and 3A). We used alternative electrode solutions to examine the ionic selectivity of ZAC. Substitution of half the intracellular K ϩ with N-methyl-D-glucamine (70 mM), caused a right shift in the current-voltage relationship (Fig. 3A), producing a positive Zn 2ϩ equilibrium potential (2 Ϯ 3 mV, n ϭ 3). This demonstrates a role for intracellular K ϩ ions in I Zn . Substitution of half of the intracellular K ϩ Cl Ϫ with K ϩ -gluconate (70 mM) had no significant effect on Zn 2ϩ equilibrium potential (n ϭ 4), demonstrating that the channels have negligible Cl Ϫ permeability.
There was little correlation between the amplitudes of I Zn and I spont in cells expressing ZAC. Furthermore, run-up of I Zn was frequently observed during an experiment without a corresponding change in the amplitude of I spont (Fig. 3B). These observations suggest that Zn 2ϩ activates channels that are closed prior to its application. In cells expressing ZAC, Zn 2ϩ caused a concentration-dependent activation of currents, with a threshold of ϳ30 M and an EC 50 of 540 Ϯ 9 M (Fig. 4A). As with the I spont , I Zn was insensitive to bath-applied ␣-bungarotoxin (10 g/ml), mecamylamine (10 M), ondansetron (1 nM), and ketamine (100 M; Fig. 4B). By contrast, TC caused a concentration-dependent inhibition of I Zn (Fig. 5) with an IC 50 of 6.6 Ϯ 0.8 M (n ϭ 4).
Absence of Functional ZAC Subunit Genes in Rodents-We were unable to amplify fragments of an orthologous gene from mouse or rat genomic DNA using degenerate primers designed from the ZAC amino acid sequence (data not shown). More surprisingly, searches of the draft mouse and rat genomic sequences (www.celera.com and www.ncbi.nlm.nih.gov) failed to yield fragments of orthologous genes. The human ZAC gene is located between genes encoding the galanin2 receptor and KIAA1067 (Fig. 6A). Consequently, we amplified and sequenced the region between the orthologues of these two genes from mouse, rat, and dog genomic DNA. For all four species, there is detectable homology between the DNA sequences of this region. This is most apparent where the human sequence represents an exon of the ZAC gene (Fig. 6B). However, for mouse and rat, most of the fragments that are homologous to exons of the human ZAC subunit gene contain numerous insertions, deletions, and substitutions that introduce in-frame stop codons (Fig. 6B). Only a footprint of the ZAC gene remains FIG. 6. A functional ZAC gene is absent from orthologous loci in rodents. A, the region of human chromosome 17q23 encoding the galanin2 receptor (GALR2), ZAC subunit, and the human orthologue of Exo70 (KIAA1067) was aligned with the orthologous region of mouse chromosome 11. Exons are indicated by black rectangles. Mouse exons that are homologous to human exons, but which contain insertions and deletions that generate in-frame stop codons, are indicated in gray. B, the sequence of exon 5 from the human ZAC subunit gene is aligned with orthologous sequences from mouse, rat, and dog genomic DNA. For the human sequence, exon 5 is in uppercase, inside brackets. Nucleotides that are conserved between human and other species are shaded. C, alignment of the deduced amino acid sequences for the human and dog ZAC subunits (hZAC subunit and dZAC subunit, respectively). Conserved residues are shaded.
in these species, and they are no longer capable of expressing a functional ZAC subunit. In contrast, dog genomic DNA from this region encodes a full-length ZAC subunit that displays 84% amino acid conservation with the human ZAC subunit (Fig. 6C). DISCUSSION The ZAC subunit displays all of the structural motifs that are common to the superfamily of Cys loop LGICs. However, its distinctive sequence and function prevent its classification within any of the four established receptor families. Sequence analysis indicates that a common ancestral gene gave rise to the present day ZAC, nACh, and 5-HT 3 receptor subunit genes. Since they diverged, a functional ZAC subunit gene has been retained by some mammals (e.g. human and dog) but has been lost by others (e.g. mouse and rat). Most human genes, including those for all previously identified LGIC subunits, have rodent orthologues. However, there is evidence for a differential loss of a limited number of genes from either rodent or human genomes (e.g. Refs. 14 and 15). With respect to ZAC, this raises the possibility that humans and rodents differ in some fundamental cell signaling mechanisms. To date, the properties of endogenous LGICs have been studied predominantly in rodent cells. An absence of ZAC from mice and rats may explain why its unusual properties have not been described previously.
ZAC exhibits TC-sensitive spontaneous activity. Some nACh and GABA A receptor subtypes are also spontaneously active, either in their natural environment (16) or when expressed in recombinant systems (17). Persistent GABA A channel openings attenuate cellular excitability and can contribute to the maintenance of resting membrane potential (18). Spontaneous nACh channel activity may participate in embryonic muscle development (16). The physiological role of spontaneous ZAC remains to be determined. However, the ability of TC to block this activity will be useful for characterizing the properties of native ZAC in tissue preparations.
Zn 2ϩ causes a concentration-dependent activation of ZAC. Zn 2ϩ can also modulate the activity of related LGICs, although, in most cases, it inhibits channel function (19 -22). However, at the glycine receptor, its effect is biphasic, potentiating glycineevoked responses at low concentrations and inhibiting them at higher concentrations (20). Zn 2ϩ may activate ZAC through an action at the putative N-terminal loop B ligand-binding domain. Subunits of each LGIC family display characteristic loop B residues that may reflect a role in agonist selectivity (11). Notably, the relevant loop B residues in the ZAC subunit (Leu-177 and Asn-179) are distinct from the corresponding residues of related subunits. This also suggests that ZAC represents a distinct class of Cys loop LGICs. It is also possible that Zn 2ϩ activates ZAC through an allosteric site that is unrelated to the loop B motif. This would be analogous to the direct activation of GABA A receptors by general anesthetics (23).
A relatively high concentration of Zn 2ϩ (Ͼ30 M) is required for activation of ZAC. Zn 2ϩ is concentrated in various tissues, including the forebrain, testes, and neuroendocrine cells (24,25). In the hippocampus, pituitary, and pancreatic ␤-cells, Zn 2ϩ is concentrated in vesicles at high concentrations (Ͼ1 mM; Ref. 25). During vesicular release, it is possible that micromolar concentrations of Zn 2ϩ are present within the synaptic cleft. However, high affinity binding to extracellular albumin is likely to reduce free Zn 2ϩ concentrations to the low nanomolar range within a short distance from its source. Native ZAC may be located close to a vesicular source of free Zn 2ϩ . Indeed, fetal brain is among a variety of human tissues that express transcripts of the ZAC subunit. Expression of ZAC subunit mRNA was detectable in several other human tissues. However, the absence of a functional ZAC subunit gene in rats and mice precludes the use of these rodents for more detailed localization studies. To resolve the distribution of ZAC expression at the cellular level, it will be necessary to perform such studies on tissues from human, dog, or other non-rodent species.
It is possible that, in its native environment, the affinity of ZAC for Zn 2ϩ is increased by a cofactor or by post-translational modification. The observed run-up of I Zn indicates an increased ability of Zn 2ϩ to activate the channel during the course of an experiment. Future experiments will examine whether this phenomenon is caused by post-translational regulation. It is also possible that Zn 2ϩ is not the physiological agonist for ZAC. However, with the knowledge that Zn 2ϩ behaves as an agonist and TC as an inhibitor of spontaneous gating, these agents will be useful tools for characterizing the properties of native ZAC receptors in tissues from human and other non-rodent species.