Caenorhabditus elegans arrestin regulates neural G protein signaling and olfactory adaptation and recovery.

Although regulation of G protein-coupled receptor signaling by receptor kinases and arrestins is a well established biochemical process, the physiological significance of such regulation remains poorly understood. To better understand the in vivo consequences of arrestin function, we have examined the function of the sole arrestin in Caenorhabditis elegans (ARR-1). ARR-1 is primarily expressed in the nervous system, including the HSN neuron and various chemosensory neurons involved in detecting soluble and volatile odorants. arr-1 null mutants exhibit normal chemotaxis but have significant defects in olfactory adaptation and recovery to volatile odorants. In contrast, adaptation is enhanced in animals overexpressing ARR-1. Both the adaptation and recovery defects of arr-1 mutants are rescued by transgenic expression of wild-type ARR-1, whereas expression of a C-terminally truncated ARR-1 effectively rescues only the adaptation defect. A potential mechanistic basis for these findings is revealed by in vitro studies demonstrating that wild-type ARR-1 binds proteins of the endocytic machinery and promotes receptor endocytosis, whereas C-terminally truncated ARR-1 does not. These results demonstrate that ARR-1 functions to regulate chemosensory signaling, enabling organisms to adapt to a variety of environmental cues, and provide an in vivo link between arrestin, receptor endocytosis, and temporal recovery from adaptation.

Modification of behavior in response to environmental change is crucial to the survival of an organism. A wide variety of stimuli, including odorants, light, hormones, peptides, and neurotransmitters, are coordinated into specific behavior through the activation of G protein-coupled receptors (GPCRs) 1 (1). Receptor activation promotes GDP/GTP exchange on heterotrimeric G proteins, and the subsequent dissociation of G␣GTP and G␤␥ subunits mediates signaling via activation of downstream effector molecules (2). To ensure precise signal magnitude and duration, GPCR signaling cascades are tightly regulated by a complex process involving, in part, GPCR kinases and arrestins (3,4). GPCR kinases phosphorylate agonist-occupied GPCRs, leading to recruitment of cytosolic arrestin, which in turn sterically inhibits GPCR/G protein interaction and terminates signaling, a regulatory process termed desensitization (3). Many organisms adapt to stimuli upon prolonged exposure, and although it is often assumed that this physiological phenomenon is the manifestation of the molecular process of GPCR desensitization, little direct evidence supports this assumption. Attenuation of GPCR signaling represents a critical step in the ability of an organism to adapt to a constantly changing environment, and it also impacts the efficacy of pharmaceuticals that target GPCR-mediated signaling pathways. It is therefore of considerable interest to clarify the relationship between the molecular process of desensitization and the phenomenon of adaptation that occurs at the behavioral level of an organism.
In mammals there are two classes of arrestin, visual and non-visual. Visual arrestins are expressed selectively in rod (arrestin-1) and cone (arrestin-4) photoreceptor cells and function to quench phototransduction. Non-visual arrestins, arrestin-2 (␤arrestin-1) and arrestin-3 (␤-arrestin-2), are ubiquitously expressed and regulate the desensitization of many GPCRs (3). Non-visual arrestins also interact with additional proteins that function to regulate receptor trafficking and signaling (4,5). For example, the binding of the C-terminal region of non-visual arrestins to clathrin and the ␤-adaptin subunit of AP2 (␤ 2 -adaptin), the major protein components of clathrin-coated pits, appears critical for mediating GPCR endocytosis (6 -8). Non-visual arrestins also activate mitogenic signaling pathways by acting as scaffolds to form complexes between receptor and various Src family members and mitogen-activated protein kinases, thereby providing a link between classical GPCR, nonreceptor tyrosine kinase, and mitogen-activated protein kinase signaling pathways (4,9). A newly appreciated role for non-visual arrestins is in the regulation of receptors other than GPCRs, including various growth factor receptors (10,11) and the low density lipoprotein receptor (12). Thus, non-visual arrestins are versatile adaptor proteins that play a broad role in regulating receptor signaling and trafficking.
Much of the initial data delineating the role of arrestins in GPCR regulation emanated from early in vitro studies of rhodopsin and the ␤ 2 -adrenergic receptor (13,14). The visual system of Drosophila melanogaster provided the first in vivo data identifying a role for arrestin as a terminator of phototransduction and more recently as a trigger for apoptotic cell death and photoreceptor degradation (15,16). Although it is thought that non-visual arrestins play an important role in the regulation of receptor signaling in many organisms, there is a general lack of in vivo data to support this, mainly because of issues concerning functional redundancy and viability. For example, a loss of function in kurtz, the sole non-visual arrestin in Drosophila, results in lethality because of the loss of kurtz in the nervous system (17). In mice, knockouts of the individual nonvisual arrestins appear overtly normal with defects observed only when animals are challenged with agonist. Specifically, a loss of arrestin-2 leads to enhanced cardiac sensitivity to ␤-agonists, suggesting a role in ␤-adrenergic receptor desensitization (18), whereas a loss of arrestin-3 results in enhanced morphine analgesia and attenuation of antinociceptive tolerance, suggesting a role in -opioid receptor desensitization (19). A double knock-out of arrestin-2 and -3 results in embryonic lethality, implying that non-visual arrestins play an important role in mouse development (20). In this regard, a recent study in Zebrafish revealed a role for arrestin-3 in signaling through Smoothened, a receptor that regulates development through hedgehog signaling (21). Overall, these studies reveal that a complete loss of non-visual arrestin leads to severe developmental defects, whereas loss of individual arrestins primarily impact processes such as drug sensitivity and tolerance. Although these studies address the in vivo role of arrestins, the combination of functional redundancy and embryonic lethality in non-visual arrestin knock-out models has hindered mechanistic insight into the in vivo function of these proteins.
In an effort to define further the biological role of non-visual arrestins and begin to correlate structural features with in vivo function, we initiated studies in Caenorhabditis elegans. C. elegans is particularly useful for studying behaviors generated by GPCR pathways as G protein signaling regulates many behaviors, including egg laying, feeding, locomotion, and chemotaxis (22,23). Genomic analysis reveals the presence of ϳ1200 GPCRs in C. elegans, many of which are thought to encode chemoreceptors (24). Additional G protein signaling components include 20 G protein ␣ subunits, 2 G␤ subunits, 2 G␥ subunits, 13 regulator of G protein signaling proteins, 2 GPCR kinases, and a single arrestin (24,25).
Here we report the characterization of ARR-1, the C. elegans arrestin encoded by the arr-1 gene. ARR-1 is expressed throughout the nervous system, including various chemosensory neurons that detect volatile chemicals. Worms lacking ARR-1 display a number of defects in behaviors mediated by GPCR/G protein signaling, including defects in egg laying and odorant adaptation. Defects in odorant adaptation and recovery were rescued in transgenic animals expressing wild-type ARR-1. Most interestingly, adaptation, but not recovery, was rescued in transgenic animals expressing a C-terminally truncated ARR-1 that lacks residues that mediate binding to clathrin and ␤ 2 -adaptin. The phenotypes reported here suggest that ARR-1 plays a broad role in GPCR regulation and serves as an important mediator of C. elegans behavior, specifically egg laying and chemosensation. We also demonstrate the importance of binding motifs required for interactions between nonvisual arrestin and components of the endocytic machinery, providing the first in vivo data highlighting the importance of these interactions in physiological adaptation and recovery from prolonged stimulation of chemosensory pathways.
Isolation of arr-1 Mutant Alleles-The arr-1(vs96) mutant allele was isolated by screening a psoralen-mutagenized library representing ϳ960,000 haploid genomes for mutations over 6 of the 10 predicted ARR-1 exons by PCR using gene-specific primers. The arr-1(ok401) deletion allele was obtained from the Oklahoma Medical Research Foundation. Knock-out Group of the C. elegans Gene Knock-out Consortium. Genomic DNA was isolated from both arr-1 alleles, and the arr-1 coding and predicted untranslated regions were amplified by PCR and sequenced on both strands using internal primers to determine break points. Mutants were outcrossed to the N2 background at least eight times before use in behavioral assays.
Molecular Biology and Plasmid Construction-Subcloning and general DNA manipulations were performed as described (27). The fulllength arr-1 cDNA clone yk371g9 was obtained from Dr. Yuji Kohara and sequenced using internal primers. Cosmid F53H8.2 was obtained from the Sanger Centre, and the arr-1 genomic region was sequenced using internal primers.
For Northern blot analysis, total RNA was prepared from wild-type and arr-1(vs96) and arr-1(ok401) mutant animals by Trizol extraction (Invitrogen). RNA was subjected to electrophoresis and blotting using the NorthernMax system (Ambion) and then probed with a 32 P-labeled arr-1 cDNA.
The arr-1::GFP fusion construct was prepared by PCR using Expand polymerase (Roche Applied Science). ϳ6.2 kb of the arr-1 region was amplified from cosmid F53H8.2, including the entire genomic sequence encoding ARR-1 and 3 kb of sequence upstream of the predicted start site to serve as an endogenous promoter. BamHI and PstI sites were engineered into the primers and were used to insert the amplified product into the pPD95.77 GFP vector (gift from A. Fire). For animals overexpressing ARR-1, ϳ7.2 kb of arr-1 genomic sequence was amplified from cosmid F53H8.2 using Expand polymerase, including ϳ3.2 kb of arr-1 genomic coding sequence, ϳ3 kb of sequence upstream of the predicted start site to serve as an endogenous promoter, and ϳ1 kb of 3Ј-untranslated sequence. XhoI and PstI sites were engineered into the primers and were used to insert the amplified product into the pBluescript II SKϩ/Ϫ vector. The ODR-3::WT and ODR-3::1-368 constructs for the rescue experiments were generated by amplifying full-length and truncated (encoding residues 1-368) arr-1 cDNAs by PCR using primers that contained 5Ј NcoI sites for subcloning into vector pPD49.26, which contains ϳ2.7 kb of the odr-3 promoter (22), kindly provided by C. Bargmann.
Transgenic Animals-Germ line transformations were performed as described previously (28). The arr-1::GFP and arr-1 pBluescript II SKϩ/Ϫ constructs were injected at 80 ng/l as a mix with lin-15 DNA (50 ng/l) as co-injection marker into the gonads of lin-15(n765ts) animals. Transgenic lines were identified by rescue of the lin-15 multivulva phenotype at 20°C. The ODR-3::WT and ODR-3::1-368 constructs were injected at 80 ng/l as a mix with lin-15 DNA (50 ng/l) as co-injection marker into the gonads of arr-1(ok401);lin-15(n765ts) animals. Transgenic lines were identified by rescue of the lin-15 multivulva phenotype at 20°C. Multiple independent transgenic lines were established from both injections.
Antibodies and Immunohistochemistry-A glutathione S-transferase (GST) fusion protein containing the C-terminal 114 amino acids of ARR-1 was used to immunize rabbits. Antibodies were purified as described previously (29). Briefly, anti-ARR-1 antibodies from sera were bound to nitrocellulose containing a maltose-binding protein-ARR-1 fusion protein, eluted with 100 mM glycine-HCl, and then further purified by pre-absorbing against an acetone precipitate from a C. elegans arr-1(ok401) lysate. An anti-peptide antibody directed against residues 289 -304 of C. elegans ARR-1 (CPLLSNNKD-KRGLALD) was generated by New England Peptide, Inc. The antibody was affinity-purified on a peptide column and then incubated with an acetone precipitate from a C. elegans arr-1(ok401) lysate before use. C. elegans total protein lysates were prepared from washed worm pellets (wild-type, ok401, and vs96) by sonication with lysis buffer. For Western blot analysis, ARR-1 protein was visualized using the purified anti-ARR-1 antibodies, a horseradish peroxidase-coupled goat anti-rabbit secondary antibody (Bio-Rad), and chemiluminescent detection using ECL (Pierce).
Whole mounts of C. elegans for antibody staining were prepared as described (30). Purified anti-ARR-1 antibodies were used at a 1:100 dilution and were incubated with specimens overnight at room temperature. Secondary antibodies (1:100 dilution) were fluorescein isothiocyanate-conjugated mouse anti-rabbit. Secondary antibody incubations were for 1 h at room temperature. Images were collected by using confocal microscopy and MetaMorph software.
Egg-laying Assays-Egg-laying rates were assayed as described previously (29). Both assays were performed on staged adult animals that were isolated as late L4 animals and allowed to develop for 36 h at 20°C. To quantify the number of unlaid eggs, 30 staged adults were individually dissolved in 50 l of a 1% sodium hypochlorite solution, and the remaining eggs were counted using an inverted microscope (Zeiss Axiovert). To quantify the developmental stage of freshly laid eggs, 30 staged adults were transferred to a clean seeded nematode growth media plate and allowed to lay eggs for 30 min at 20°C. The developmental stage of each freshly laid egg was then determined by viewing under a high magnification dissecting microscope. Values represent the percent mean Ϯ S.E. of three separate experiments.
Chemotaxis, Adaptation, and Recovery Assays-Chemotaxis assays were performed as described in Bargmann et al. (31). Adaptation assays were performed as described in L'Etoile and Bargmann (32). Animals were incubated in S-Basal or S-Basal plus 800 l of isoamyl alcohol/ liter, 60 l of benzaldehyde/liter, or 100 l of diacetyl/liter for 1 h, washed, and then tested for chemotaxis toward a point source of adapting odorant (1 l of isoamyl alcohol (diluted 1:100 in ethanol), benzaldehyde (diluted 1:200) or diacetyl (diluted 1:1000) plus 1 l of 1 M sodium azide). Chemotaxis time course assays were performed as described in Bargmann et al. (31) with the exception that worms were allowed to chemotax toward 1 l of diluted odorant (isoamyl alcohol 1:100, benzaldehyde 1:200, or diacetyl 1:1000 in ethanol) or control (ethanol) point sources with no sodium azide for 120 min with a chemotaxis index calculated every 20 min.
To assess recovery from adaptation, two different assays were performed. The first is a modification of the standard adaptation assay. Briefly, animals were incubated in S-Basal or S-Basal plus 800 l of isoamyl alcohol/liter for 40 min. Animals were washed free of odorant and then allowed to recover in 1 ml of S-Basal for 40 min. Animals were then assayed for their ability to chemotax toward adapting odorant (1 l of 1:100 isoamyl diluted in ethanol plus azide). The second assay is a modification of the standard chemotaxis assay. Animals were allowed to chemotax toward 1 l of diluted odorant (1:100 isoamyl alcohol diluted in ethanol) or control (ethanol) point sources with no sodium azide on chemotaxis assay plates. The number of animals at each point source and the total number of animals on the plate were counted after 40 min, and then 1 l of adapting odorant (1 l of isoamyl alcohol diluted 1:100 in ethanol plus 1 l of 1 M sodium azide) was applied to the same chemotaxis assay plate near the original control point source. Animals were tested for chemotaxis toward the new adapting odorant for 1 h. The recovery index was then calculated by subtracting the number of animals at the original odorant point source from the number of animals at the new odorant point source and then dividing this number by the total number of worms on the plate.
Cell Culture and Transient Transfection-COS-1 and HEK293 cells were maintained in Dulbecco's modified Eagles's medium (Invitrogen) supplemented with 10% fetal bovine serum, 100 units/ml penicillin G, and 100 g/ml streptomycin sulfate. Transient transfection was done using FuGENE 6 (Roche Applied Science) according to the manufacturer's recommendations.
Receptor Internalization Assay-Internalization of FLAG-tagged ␤ 2adrenergic receptor was assayed by enzyme-linked immunosorbent assay as described previously with some minor modifications (8). Briefly, HEK293 cells transfected with pcDNA3-FLAG-␤ 2 AR and various arrestin constructs were split into poly-L-lysine-coated 24-well tissue culture plates after 24 h. The next day (48 h post-transfection), cells were treated with 10 M (Ϫ)-isoproterenol and 0.1 mM ascorbate at 37°C for 15 min, fixed with 3.7% formaldehyde in Tris-buffered saline (TBS) for 5 min at ambient temperature, and washed three times with TBS. Cells were then blocked with TBS containing 1% bovine serum albumin (TBS/bovine serum albumin) for 45 min, incubated with a primary antibody (M2 anti-FLAG conjugated with alkaline phosphatase (Sigma) diluted 1:2000 in TBS/bovine serum albumin) for 1 h, and washed three times with TBS. Colorimetric visualization of antibody binding was performed using an alkaline phosphatase substrate kit (Bio-Rad), and samples were read at 405 nm in a microplate reader using Microplate Manager software (Bio-Rad). The reading from cells that did not express FLAG-␤ 2 AR was used as a blank, and the percentage of surface receptor loss was determined by calculating the change in antibody-accessible FLAG-␤ 2 AR.
Analysis of ARR-1 Interaction with ␤ 2 -Adaptin and Clathrin-A GST-␤ 2 -adaptin appendage (residues 700 -937) fusion protein and a GST-clathrin terminal domain (residues 1-579) fusion protein were expressed and purified on glutathione-agarose previously as described (8). HA-tagged wild-type ARR-1 and ARR-1-(1-368) were expressed in COS-1 cells by transient transfection. ARR-1 extracts were prepared by lysing the cells by Polytron disruption in 20 mM Hepes, pH 7.4, 0.1 M NaCl, 0.02% Triton X-100, 10 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 0.02 mg/ml leupeptin, 0.2 mg/ml benzamidine. The extracts were centrifuged (19,000 rpm for 20 min), and the supernatants were frozen at Ϫ80°C until needed. Arrestin-containing lysates were incubated with 5 l of glutathione-agarose beads (containing ϳ10 g of bound GST or GST fusion protein) in binding buffer (20 mM Hepes, pH 7.2, 120 mM potassium acetate, 0.1 mM dithiothreitol, 0.1% Triton X-100) for 1 h at 4°C (in a total volume of 100 l). The beads were pelleted (1000 rpm, 5 min at 4°C) and washed three times with 0.5 ml of ice-cold binding buffer, and bound ARR-1 was eluted by boiling the beads in SDS sample buffer for 10 min. The samples were electrophoresed on a 10% polyacrylamide gel and transferred to nitrocellulose, and ARR-1 was detected by immunoblotting using anti-HA monoclonal antibody and horseradish peroxidase-labeled horse anti-mouse secondary antibody and chemiluminescence.

arr-1 Encodes a Non-visual-like Arrestin-The C. elegans
arrestin gene arr-1 encodes a protein of 435 amino acids. ARR-1 shares the highest degree of similarity with the Drosophila non-visual arrestin kurtz (70% amino acid similarity) and the two mammalian non-visual arrestins, arrestin-2 (65% similarity) and arrestin-3 (64% similarity) (Fig. 1A). The crystal structures of bovine arrestin-1 and -2 reveal that arrestins are composed of two major domains that are held together by a polar core of buried salt bridges (33,34). Alignment of ARR-1 with bovine arrestin-1, -2, and -3 indicates that ARR-1 contains many of the structural features found in arrestin family members ( Fig. 1B), including the polar core, a partially conserved clathrin-binding box, a ␤ 2 -adaptin binding domain, and two proline-rich motifs in the C-terminal region that may mediate SH3 domain binding (7,9,35). It has been suggested that disruption of the polar core by interaction with phosphorylated receptor promotes structural changes in arrestin resulting in an active conformation (34) and enhancing interactions with additional binding partners such as clathrin and ␤ 2 -adaptin (8). Based on the conservation of critical sequence features, it seems likely that C. elegans ARR-1 and mammalian non-visual arrestins will interact with similar target proteins and function to regulate receptor signaling and trafficking.
To start to dissect the function of ARR-1 in C. elegans, we characterized two mutant alleles of arr-1. We isolated mutant allele arr-1(vs96) by using a PCR-based method to screen a library of mutagenized C. elegans (36), whereas mutant allele arr-1(ok401) was obtained from the C. elegans Gene Knock-out Consortium. Both alleles are recessive, and sequencing of the coding regions shows each allele has a deletion within the predicted open reading frame. In arr-1(ok401) animals, the deletion begins in exon 3 and creates a premature stop after 67 codons ( Fig. 2A), suggesting that ok401 is a null allele because regions known to be essential for proper folding of arrestin are absent. In arr-1(vs96) animals, the deletion begins in exon 8 and ends in the 3Ј-untranslated region, encoding a protein that retains 368 of the 435 amino acids of ARR-1 ( Fig. 2A). This truncated ARR-1 lacks the C terminus containing the putative clathrin and ␤ 2 -adaptin-binding sites. In probing total RNA from wild-type and mutant animals with a radiolabeled fulllength arr-1 cDNA, an ϳ1.5-kb message was detected in wildtype worms (Fig. 2B). No message was detected in arr-1(ok401) RNA preparations, whereas a message of ϳ1.6 kb was detected in arr-1(vs96) (Fig. 2B). The slightly larger size of the message in arr-1(vs96) may be because of an alternative polyadenylation site being used in the 3Ј-untranslated region.
To evaluate ARR-1 expression in wild-type and mutant worms, a polyclonal antibody was raised against the C-termi- nal 114 amino acids (residues 322-435) of ARR-1. Western blot analysis detected a major protein band of ϳ60 kDa in a total protein lysate from wild-type animals, whereas no ARR-1 was detected in lysates from ok401 or vs96 mutant worms (Fig. 2C). Most interestingly, additional ARR-1 bands of ϳ70 and ϳ80 kDa were also detected in the wild-type lysates. These may represent ubiquitinated forms of ARR-1, as it has been shown that mammalian arrestin-3 undergoes ubiquitination by the E3 ubiquitin ligase, Mdm2 (4). The absence of detectable ARR-1 protein in the vs96 lysates suggests either that the protein is not expressed or that the antibody does not detect the truncated ARR-1. Thus, to clarify whether any C-terminally truncated ARR-1 is expressed in vs96 mutant worms, we generated an anti-peptide antibody directed against residues 289 -304 in ARR-1. This antibody effectively detects ARR-1-(1-368) expressed in COS-1 cells as well as ARR-1 in lysates from wildtype animals but does not detect any ARR-1 expression in vs96 mutant worms (data not shown). Thus, we conclude that neither ok401 nor vs96 mutant worms express significant levels of ARR-1.
ARR-1 Is Expressed throughout the Nervous System-To determine where ARR-1 is expressed, 3 kb of sequence upstream from the predicted translation start codon was used to drive the expression of full-length ARR-1 containing GFP at the C-terminal end (37) (Fig. 3A). In transgenic animals, ARR-1::GFP was strongly expressed throughout the nervous system (Fig.  3B), which consists of 302 neurons primarily concentrated in the head, along the nerve cords, and in the tail. ARR-1::GFP was expressed in the dorsal and ventral nerve cords and most or all neurons and their processes in the head (Fig. 3C) and tail (Fig. 3F). Specifically, ARR-1::GFP was highly concentrated in the amphid chemosensory neurons that mediate chemotaxic responses to volatile and water-soluble compounds via GPCR signaling pathways, including the AWA, AWB, AWC, ADL, and ASH neurons (Fig. 3D). It was also observed in the HSN neuron (Fig. 3E), which stimulates egg laying, another behavior regulated by GPCR signaling (38). ARR-1::GFP in neuronal cells appeared to be primarily cytoplasmic and did not exhibit a distinct subcellular localization.
We also used affinity-purified anti-ARR-1 to evaluate endogenous ARR-1 expression. Endogenous ARR-1 was shown to be concentrated in the nerve ring, amphid processes, and in the ventral nerve cord of wild-type animals (Fig. 4A). No specific staining was observed in arr-1(ok401) mutant worms (Fig. 4B), confirming the loss of ARR-1 in these animals and the specificity of the antibody. The broad expression of ARR-1 throughout the nervous system implies that ARR-1 plays a role in behaviors mediated by G protein signaling in C. elegans as G proteins are also broadly expressed within the nervous system (25). To investigate the functional role of ARR-1 in C. elegans, we initially examined egg laying and chemotaxis, both behaviors known to be mediated by G protein signaling (22,25).
arr-1 Mutations Affect Egg Laying-Genetic studies have demonstrated that egg laying, locomotion, and feeding are all behaviors mediated through the mutually antagonistic G␣ o (GOA-1) and G␣ q (EGL-30) signaling pathways (39). Wild-type behavior is achieved through a balance of activity of these two G protein pathways. The expression pattern of ARR-1 overlaps that of GOA-1 G␣ o and EGL-30 G␣ q , which are expressed in most or all neurons, suggesting that ARR-1 may play a role in regulating behaviors associated with these G proteins. To determine whether ARR-1 regulates G␣ o /G␣ q , we analyzed egglaying behavior in arr-1(ok401) and arr-1(vs96) mutants as well as in animals overexpressing ARR-1. ARR-1(OE) transgenic animals carry an arr-1 genomic clone on a high copy transgene and express ϳ3-fold higher levels of ARR-1 compared with wild-type worms based on Western blot analysis (data not shown).
Wild-type animals maintain a steady state of fertilized unlaid eggs and lay eggs about 2.5 h after fertilization, by which point the eggs have developed beyond the 9-cell stage but have not yet reached the "comma" stage where morphogenesis begins. goa-1 G␣ o mutant animals exhibit hyperactive egg-laying behavior (accumulating fewer unlaid eggs in the uterus and laying eggs prior to the 9-cell stage), whereas egl-30 G␣ q mutants exhibit defective egg-laying behavior (more eggs in the uterus and eggs laid at late stages of development). Both arr-1 mutants exhibited an egg-laying defective phenotype as evidenced by the accumulation of unlaid eggs (Table I). Although wild-type worms contained 13 unlaid eggs, this number increased to 23 Ϯ 1 in arr-1(ok401) and 18 Ϯ 2 in arr-1(vs96). ok401 mutants also laid a significant number of late-stage eggs as compared with wild-type animals (Table I) changes in the rate of egg production, as all three strains produced a similar number of eggs (Table I). In contrast, animals overexpressing ARR-1 exhibited an increased rate of egg laying, as best evidenced by an increase in the number of eggs laid prematurely (Table I). Although wild-type animals laid only 1 Ϯ 2% of their eggs by the 8-cell stage, ARR-1(OE) animals laid 42 Ϯ 10% of their eggs at the 1-8-cell stages. Most interestingly, ARR-1(OE) animals produced a brood of only 111 Ϯ 26 fertilized eggs, whereas wild-type animals produced 281 Ϯ 3 fertilized eggs over their adult life (Table I). Normarski photographs of ARR-1(OE) adult hermaphrodites reveal that their gonads are disorganized (data not shown), although the reason for this remains unclear at the present time.
These results suggest that ARR-1 plays a role in egg-laying behavior, most likely through the regulation of the mutually antagonistic G␣ o and G␣ q pathways that inhibit and stimulate egg-laying, respectively. It is plausible that ARR-1 regulates both G␣ o and G␣ q to maintain homeostasis and that these opposing effects largely cancel each other, explaining why the phenotypes observed in the arr-1 mutants and ARR-1 overexpressers are not as severe as those that have been observed in goa-1 or egl-30 mutants (40, 41). Because loss of ARR-1 partially inhibits egg laying, whereas ARR-1 overexpression has the opposite effect, ARR-1 may function to shift the balance of G␣ o and G␣ q signaling by uncoupling G␣ o signaling more effectively than it uncouples G␣ q .
ARR-1 Is Not Essential for Chemotaxis-The ability of an animal to sense and relay chemosensory signals into behavioral responses is mediated largely by olfactory neurons that express olfactory/chemosensory receptors. These receptors are members of the GPCR superfamily, as biochemical and genetic evidence has implicated G proteins in olfactory signal transduction (42). In C. elegans, 32 chemosensory neurons located in and around the nerve ring contact the environment through openings in the cuticle (43). A large family of worm "orphan" chemosensory receptors expressed in these neurons have been identified as GPCRs, although they share limited sequence similarity with each other and with mammalian odorant receptors (44). Although various alcohols, ketones, and aromatic compounds serve as attractants to C. elegans (31), the only GPCR that has been paired with an attractant is ODR-10, a receptor for the odorant diacetyl (45). Additional components that mediate chemosensation have been identified, including two chemosensory G␣ subunits (ODR-3 and GPA-2), two cyclic nucleotide-gated ion channel subunits (TAX-2 and TAX-4), and a guanylyl cyclase (ODR-1) (22, 32, 46). Additionally, a recent The animal was double-labeled with the red lipophilic dye DiD, which fills a subset of chemosensory neurons whose names are indicated with arrows. ADL (amphid, dual ciliated neuron) appears red as it is labeled with DiD, but in this animal showed no green ARR-1::GFP fluorescence. All other DiD-labeled cells appear in shades of yellow because of simultaneous red DiD and green GFP labeling. Also indicated is the AWC (amphid wing C) neuron, an amphid neuron that does not take up DiD, which senses the odorant isoamyl alcohol and can be identified by its position adjacent to ASH (amphid, single ciliated neuron). E, ARR-1::GFP expression in the vulva region of transgenic adults. The arrowhead points to the vulva. The large arrow indicates the HSN (hermaphrodite-specific neuron), and the small arrow points to the ventral nerve cord. F, ARR-1::GFP expression in the tail of a transgenic adult. Fluorescence is observed in many posterior neuronal cell bodies. study has also revealed an essential role for the GPCR kinase GRK-2 in chemosensation (47).
Localization experiments revealed that ARR-1 is expressed in many of the chemosensory neurons, including the ciliated amphid neuron AWC (Fig. 3). To explore a potential role for ARR-1 in chemosensation, we initially used a standard chemotaxis assay to investigate whether ARR-1 is involved in chemotaxis to isoamyl alcohol, a volatile compound detected by the AWC neuron (31). Worms were placed on an agar plate, and the chemotaxis index was calculated after 20, 40, and 60 min by subtracting the number of animals at the control spot from the number of animals at the attractant spot and then dividing this number by the total number of animals on the plate. Neither arr-1 mutant allele exhibited defects in chemotaxis toward isoamyl alcohol as compared with the wild type, indicating that a loss of ARR-1 does not affect the processes responsible for detecting isoamyl alcohol (Fig. 5A). In contrast, the rate of chemotaxis of ARR-1(OE) transgenic animals was delayed ϳ2fold compared with wild-type animals, suggesting that ARR-1 overexpression may regulate sensory input (Fig. 5A).
arr-1 Mutants Exhibit Defects in Adaptation-In mammalian olfactory systems prolonged odorant exposure results in a decrease in responsiveness to odorant, a phenomenon termed adaptation (48). Adaptation is also observed in the C. elegans chemosensory response; exposure to a specific odorant attenuates the ability of the worm to chemotax toward that same odorant but does not affect the ability to distinguish between odorants sensed by the same ciliated neuron (49). Because the properties of adaptation are analogous to GPCR desensitization, a process known to be mediated by arrestins, arr-1 mutant, and ARR-1(OE) animals was tested for defects in adaptation. Following a standard adaptation assay, animals were incubated in buffer or diluted in isoamyl alcohol for 1 h, washed free of odorant, and then exposed to isoamyl alcohol in a che-motaxis assay (32). Wild-type animals adapt to isoamyl alcohol, as illustrated by the decreased sensitivity toward odorant of adapted animals (black bars) compared with nonadapted animals (white bars), which remain sensitive to odorant (Fig. 5B). Both arr-1 mutant alleles exhibited defects in adaptation as compared with the wild type, whereas animals overexpressing ARR-1 appeared hypersensitive to isoamyl alcohol adaptation (Fig. 5B). The adaptation defects observed in the arr-1 mutants were similar to those observed in osm-9(ky10) mutant worms (Fig. 5B), which have a defect in OSM-9, a TRP-like channel involved in adaptation to isoamyl alcohol and butanone (50). adp-1(ky20) mutant worms, which are specifically defective in adaptation to benzaldehyde and butanone, were used as a control and adapted normally to isoamyl alcohol (Fig. 5B) (49). These results provide evidence that ARR-1 is required for adaptation to the AWC-sensed attractant isoamyl alcohol in C. elegans.
Although the standard chemotaxis assay includes sodium azide at the control and attractant sources so that animals are paralyzed when they reach these spots, we also performed a time course chemotaxis assay in the absence of sodium azide in an effort to assess adaptation temporally. In these experiments we calculated the chemotaxis index every 20 min over a 120min period, and we thereby examined the ability of the animal to move away from the source of attractant over time as they adapted (Fig. 5C, right panel). The accumulation of wild-type animals at a source of isoamyl alcohol was maximal at 40 min, with a chemotaxis index of ϳ0.8, and then decreased over time as animals moved away from the attractant (Fig. 5C), implying that these animals become adapted to isoamyl alcohol. arr-1(ok401) and arr-1(vs96) mutant worms also move to the source of attractant (maximal chemotaxis index of ϳ0.9 at 40 min), but in striking contrast to the wild type, these animals remain at the attractant for a prolonged period of time. These results confirm that a loss of ARR-1 interferes with normal adaptation behavior in these animals most likely because of attenuated desensitization of the isoamyl alcohol receptor. ARR-1(OE) worms also differed from the wild type as a reduced number of worms were at the source of attractant between 20 and 80 min (maximal chemotaxis index of ϳ0.5 at 40 min). It is possible that these worms either rapidly adapt and move away from the attractant or start to adapt before they even reach the main site of attractant. These time course chemotaxis assays reinforce our model that mutations affecting ARR-1 expression interfere with adaptation behavior and provide insight into the role of arrestin in the temporal control of adaptation.
To determine whether the defects in adaptation in arr-1(ok401) animals were because of the loss of ARR-1 in the chemosensory neurons, ϳ2.7 kb of the odr-3 promoter was used to drive expression of wild-type ARR-1 (ODR-3::WT) or C-terminally truncated ARR-1 (ODR-3::1-368) in the AWC neurons of arr-1(ok401) null mutants. odr-3 encodes the G␣ subunit ODR-3, which is specifically expressed in AWC and to a lesser extent in AWB, AWA, and ASH neurons (22). These transgenic strains were then tested for their ability to adapt to isoamyl alcohol in both the standard adaptation and time course chemotaxis assays. The arr-1(ok401) adaptation defect was fully rescued in both transgenic strains (Fig. 6, A and B), although a modest delay in the ability to move away from odorant was observed in ODR-3::1-368 transgenic animals (Fig. 6B). These results demonstrate that a loss of ARR-1 in chemosensory neurons is responsible for the observed adaptation defects in arr-1 mutant animals and that the C-terminal domain of ARR-1 is not required for adaptation to isoamyl alcohol.
To determine whether the adaptation defects observed for isoamyl alcohol in arr-1 mutant animals are also observed with FIG. 4. ARR-1 protein expression. A, anti-ARR-1 antibody staining of the head of a wild-type adult hermaphrodite. Fluorescence is observed in the nerve ring (large arrow), amphid process (arrowhead), and ventral nerve cord (small arrow). B, control immunostaining experiment: anti-ARR-1 antibody staining of the head of an arr-1(ok401) adult hermaphrodite prepared in parallel to A lacks any specific staining. additional odorants, we investigated the responses of arr-1(ok401), ARR-1(OE), and ODR-3::WT transgenic animals to benzaldehyde and diacetyl, volatile odorants that are sensed by the AWC and AWA neurons, respectively. Chemotaxis toward benzaldehyde was unaffected in arr-1 nulls (Fig. 7A, left panel), whereas chemotaxis toward diacetyl appeared to be delayed in arr-1 mutants (Fig. 7A, right panel). This suggests that ARR-1 may play a role in chemotaxis toward diacetyl, although it clearly does not play an essential role like GRK-2 (47). arr-1(ok401) mutant animals were next tested for defects in adaptation to benzaldehyde and diacetyl. Following the standard adaptation assay, significant defects in adaptation to benzaldehyde were observed in arr-1 mutants (Fig. 7B, left panel). Similarly, time course assays confirmed that arr-1(ok401) mutants are defective in adaptation to benzaldehyde as indicated by the reduced ability of these animals to move away from the attractant (Fig. 7C, left panel). This ability is rescued in ODR-3::WT transgenic animals, although these animals do not chemotax toward the attractant as effectively as wild-type animals, perhaps due to the overexpression of ARR-1 in AWC neurons (Fig. 7, A and C, left panel). The behavior of ARR-1(OE) transgenic animals in this assay suggests that the overexpression of ARR-1 enhances adaptation to benzaldehyde.
The loss of ARR-1 has a more modest effect on adaptation to diacetyl. Although the standard adaptation assay reveals no striking defect (Fig. 7B), a clear defect in adaptation is observed in the time course analysis of arr-1 mutants as compared with wild-type animals (Fig. 7C, right panel). This defect is partially rescued in the ODR-3::WT transgenic animals, as these animals move away from attractant but at a slower rate as compared with wild type. This may be due to a low level of ARR-1 expression in AWA neurons in these animals because the ODR-3 promoter drives expression primarily in the AWC and to a much lesser extent in the AWA (22). Taken together, these results demonstrate that ARR-1 regulates chemosensory adaptation in AWC neurons and also contributes to adaptation in AWA neurons.
ARR-1 Is Involved in Recovery from Adaptation-Adaptation is fully reversible in C. elegans; wild-type worms recover their ability to sense and chemotax toward an adapting odorant within 3 h of removal from odorant (49). To investigate whether ARR-1 contributes to the ability of worms to recover from odorant-induced adaptation, we developed two assays to measure recovery from adaptation. The first is a modification of the standard adaptation assay (32). Animals are incubated for 40 min in buffer or diluted isoamyl alcohol in a microcentrifuge tube, washed free of odorant, and then either analyzed immediately in a standard chemotaxis assay or allowed to recover in buffer for 40 min before analysis (Fig. 8A, right). Wild-type and ODR-3::WT transgenic animals effectively adapt to isoamyl alcohol (Fig. 8A, white versus black bars), and after the 40 min of recovery incubation, these animals largely regain their abil-ity to chemotax toward isoamyl alcohol, indicating that they have recovered from adaptation (Fig. 8A, black versus gray  bars). More interestingly, this recovery is not observed in ODR-3::1-368 transgenics. Although these animals adapt to isoamyl alcohol, they do not regain the ability to chemotax toward isoamyl alcohol after the 40 min of recovery incubation (Fig. 8A, black versus gray bars).
The second assay we developed attempted to assess recovery from adaptation in a temporal manner. In this assay, a fixed number of animals are initially exposed to a spot of isoamyl alcohol (A) or control (C) on an assay plate in the absence of sodium azide. After 40 min, a new attractant source (new A) is generated by placing fresh isoamyl alcohol plus sodium azide near the original control site. Animals are then allowed to chemotax for 1 h, and a recovery index is calculated by subtracting the number of animals at the original odorant (old A) from the number of animals at the new odorant (new A) and then dividing this number by the total number of animals on the assay plate (Fig. 8B, right). A positive recovery index indicates that more animals are at the new site versus the original site. In setting up this second chemoattractant gradient, we are evaluating the ability of animals to recover from adaptation to the initial attractant and then chemotax to a second spot of the same attractant.
The recovery index for wild-type animals was 0.44 Ϯ 0.06, demonstrating that many of the worms recovered from adaptation and were able to move to the new attractant after 1 h (Fig. 8B, left). In contrast, the recovery index for arr-1(ok401) mutant animals was negative (Ϫ0.22 Ϯ 0.03), indicating that these animals were unable to move to the new attractant. In principle, this defect could be due entirely to the inability of arr-1(ok401) animals to adapt and move away from the initial attractant (Fig. 5), a caveat of this assay. However, the ability to move to a new attractant was completely rescued by ARR-1 expression in the ODR-3::WT transgenic animals with a recovery index of 0.33 Ϯ 0.05 (Fig. 8B). This demonstrates that expression of ARR-1 in AWC neurons effectively restores both adaptation (Fig. 6) and recovery from adaptation. Most interestingly, the ODR-3::1-368 transgenics had a negative recovery index (Ϫ0.09 Ϯ 0.05) in this assay (Fig. 8B). Because expression of ARR-1-(1-368) restores normal adaptation (Fig.  6), these results suggest that the C-terminal domain of ARR-1 is specifically involved in mediating the recovery process. More importantly, virtually identical results were observed when benzaldehyde was used as the odorant. Wild type and ODR3::WT animals recovered their ability to chemotax to benzaldehyde, whereas ODR-3::1-368 transgenics do not (data not shown).
The C-terminal region missing in ARR-1-(1-368) contains putative binding motifs for clathrin and ␤ 2 -adaptin, interactions known to be involved in mediating GPCR endocytosis in mammalian cells (8). Receptor endocytosis is involved in resen- sitization of some GPCRs (51), and although this has not been linked to the physiological phenomena of recovery in behavioral assays, we hypothesized that a defect in receptor endocytosis is responsible for the recovery defect observed in the ODR-3::1-368 transgenic line. To address the potential and functional role of the C-terminal domain of ARR-1, we initially  -1(vs96), arr-1(ok401), ARR-1(OE), adp-1(ky20), and osm-9(ky10) animals to isoamyl alcohol. arr-1(vs96) and arr-1(ok401) are defective in adaptation, whereas ARR-1(OE) animals are more efficient in their ability to adapt as compared with the wild-type. Data shown represent the mean Ϯ S.E. for six trials totaling over 1800 animals for each strain. C, right panel, temporal chemotaxis assay; animals were tested for chemotaxis to a point source of odorant in the absence of sodium azide. Left panel, time course of chemotaxis for wild-type, arr-1 mutants vs96 and ok401, and ARR-1(OE) animals to a point source of isoamyl alcohol not containing sodium azide. Wild-type animals adapt and move away from attractant over time, whereas arr-1 mutants vs96 and ok401 do not. Data shown represent the mean Ϯ S.E. for three trials totaling over 300 animals for each strain.
performed in vitro GST-pull-down assays and found that ARR-1 binds to clathrin and ␤ 2 -adaptin, whereas ARR-1-(1-368) does not (Fig. 8C). Binding to clathrin is likely mediated by the partially conserved clathrin-binding motif present in ARR-1 (LIQLH; consensus motif is LXE, where is a bulky aliphatic residue), whereas ␤ 2 -adaptin likely binds to the FXXXR region in ARR-1 (Fig. 1B) (8). To investigate if ARR-1 can promote receptor endocytosis, we tested the ability of ARR-1 and ARR-1-(1-368) to enhance internalization of the ␤ 2 -adrenergic receptor. When co-expressed in HEK293 cells, ARR-1 promoted the internalization of the ␤ 2 -adrenergic receptor to levels comparable with those observed with co-expression of mammalian non-visual arrestins ( Fig. 8D and data not shown). In contrast, ARR-1-(1-368) was completely defective in promoting ␤ 2 -adrenergic receptor internalization, underscoring the importance of the C-terminal region of ARR-1 in this process (Fig. 8D). Our data suggest that in ODR-3::1-368 transgenics, ARR-1-(1-368) mediates the uncoupling of receptor from downstream signaling events that lead to adaptation (Fig. 6). However, the inability of ARR-1-(1-368) to interact with components of the endocytic machinery in ODR-3::1-368 transgenics likely results in a defect in receptor endocytosis and resensitization and a loss in recovery from adaptation. DISCUSSION ARR-1 Is Not Essential in C. elegans Development-Nonvisual arrestins participate in the processes that regulate GPCR signal transduction, including desensitization, signaling, and trafficking (3,4). The coordination of these events in response to a changing environment constitutes crucial elements of homeostasis; however, the in vivo mechanisms governing these processes remain elusive. To understand further the physiological role of non-visual arrestin and the potential mechanisms involved in mediating its in vivo function, we  -1(ok401), ARR-1(OE), and ODR-3::WT animals to benzaldehyde (left) and diacetyl (right). Chemotaxis to diacetyl is slightly delayed in arr-1(ok401) mutants as compared with wild type. Data shown represent the mean Ϯ S.E. for four trials totaling over 400 animals for each strain. B, adaptation of wild-type and arr-1(ok401) mutant animals to benzaldehyde (left) and diacetyl (right). arr-1(ok401) mutants were defective in their ability to adapt to benzaldehyde but not to diacetyl as compared with wild type. Data shown represent the mean Ϯ S.E. for four trials totaling over 400 animals for each strain. C, time course of chemotaxis for wild-type, arr-1(ok401), ARR-1(OE), and ODR-3::WT animals to a point source of either benzaldehyde (right) or diacetyl (left) not containing sodium azide. Wild-type animals adapt and move away from point source of benzaldehyde (right) over time, whereas arr-1(ok401) mutants do not. This defect in adaptation was rescued in the ODR-3::WT transgenic animals as compared with arr-1(ok401) mutants. Wild-type animals adapt and move away from point source of diacetyl (left), whereas arr-1(ok401) mutants do not. This defect in adaptation was partially rescued in the ODR-3::WT transgenic animals as compared with arr-1(ok401) mutants. Data shown represent the mean Ϯ S.E. for four trials totaling over 400 animals for each strain. characterized ARR-1, the single arrestin encoded by the C. elegans genome. ARR-1 is most similar to the D. melanogaster non-visual arrestin kurtz and retains motifs that mediate interactions with endocytic proteins and Src family members specific to mammalian non-visual arrestins (Fig. 1). arr-1 is not an essential gene in C. elegans as worms lacking ARR-1 are viable and do not exhibit obvious developmental abnormalities. In light of the lethal phenotypes observed in flies and mice lacking non-visual arrestins (17,20), the viability of arr-1 mutant C. elegans was somewhat surprising. However, it is recognized that the viability and fecundity of C. elegans is not as dependent on nervous system function as it is in flies and mammals. This is exemplified by the fact that mutations in several other genes associated with neural function are lethal in mice and flies but not in C. elegans (24). To characterize ARR-1 function, we isolated two mutant arr-1 alleles (Fig. 2). arr-1(ok401) appears to be a null as no transcript is detected by Northern blot analysis and no protein is detected by Western blotting. Similarly, no ARR-1 is detected in lysates from arr-1(vs96) mutant worms, although a transcript is made (Fig. 2). ARR-1 is broadly expressed throughout the nervous system, including the HSN and various chemosensory neurons, suggesting that it plays a role in mediating behaviors such as egg laying and olfactory chemosensation (Figs. 3 and 4).
ARR-1 Activity Affects Egg-laying Behavior-Previous studies have shown that G␣ o (GOA-1) and G␣ q (EGL-30) signaling regulate behaviors such as egg laying, locomotion, and feeding and that the balance between these two mutually antagonistic pathways determines the overall behavior of the animal (39,41). The localization of ARR-1 overlaps that of GOA-1 G␣ o and EGL-30 G␣ q , suggesting that it may impact these G proteinregulated behaviors. Moreover, the expression of ARR-1 in the serotonergic motor neuron HSN suggests that it could play a role in egg-laying behavior. We did not observe any locomotion defects in arr-1 mutant animals (data not shown); however, we did observe defects in egg-laying behavior. The loss of ARR-1 inhibits egg laying, whereas overexpression of ARR-1 enhances egg laying (Table I). Epistasis experiments suggest that ARR-1 is upstream or parallel to GOA-1 G␣ o because arr-1(ok401);goa-1(sa734) double mutants exhibit the hyperactive egg-laying phenotype of goa-1(sa734) worms (data not shown). These results suggest that ARR-1 uncouples G␣ o signaling more effectively than G␣ q ; however, the moderate phenotypes exhibited by the arr-1 mutants and transgenics imply that ARR-1 regulates the balance of egg laying rather than exhibiting specificity toward one G protein pathway versus another. Indeed, because there is a single arrestin in the C. elegans genome, it is reasonable to predict that ARR-1 will regulate multiple GPCR pathways. Moreover, an additional layer of regulation that controls egg laying is imposed by the RGS proteins EAT-16 and EGL-10 and might partially compensate for the disruption brought about by the loss of ARR-1 (29,39).
Olfactory sensory systems adapt to prolonged stimulation by modulating sensitivity to stimulant. Adaptation toward a variety of odorants and neurotransmitters has been studied extensively in C. elegans (49,54,55). These studies have identified several proteins that contribute to odorant adaptation including ODR-1 (guanylyl cyclase), EGL-4 (cGMP-dependent protein kinase), ADP-1, TBX2/TBX3 (transcription factor), and OSM-9 (TRP-like channel). The decreased response to odorants in these studies is reminiscent of the process of desensitization observed in cellular systems. Here we show that ARR-1 plays a significant role in olfactory adaptation in C. elegans. arr-1 null mutants exhibit defects in adaptation to AWC-sensed odorants isoamyl alcohol and benzaldehyde and to a lesser extent the AWA-sensed diacetyl (Figs. 5-7). The failure of arr-1 mutants to adapt suggests that the loss of arrestin significantly reduces the extent of receptor desensitization. The adaptation defects observed in arr-1 mutants toward these odorants is because of the loss of ARR-1 in AWC neurons, as animals overexpressing wild-type ARR-1 under the AWC-specific promoter ODR-3 are not adaptation-defective (Fig. 6). Differences in adaptation of arr-1 mutants to diacetyl observed between the standard and time course adaptation assays suggests that the time course may be better at detecting weak defects in adaptation. The extent of adaptation defects observed in arr-1 mutant animals to AWC-sensed odorants compared with AWA-sensed odorants could be because of the fact there are additional mechanisms involved in adaptation in AWA neurons. For example, chemosensation in the AWA neurons has been shown to require different signal transduction molecules than those in the AWC neurons (50). Overall, the phenotypes observed in these assays demonstrate that ARR-1 is involved in the adaptation of GPCR-mediated behavioral responses in C. elegans and provide clear evidence for a link between proteins such as arrestin that mediate the cellular/molecular process of desensitization and the physiological phenomenon of adaptation.
An important aspect of adaptation is the subsequent ability chemotaxis index representing adaptation (black bars) and recovery (gray bars) of wild-type and ARR-1 transgenic animals to isoamyl alcohol (C ϭ control, A ϭ adaptation, R ϭ recovery). No defects are observed in chemotaxis or adaptation to isoamyl alcohol. ODR-3::1-368 transgenic animals do not effectively recover from adaptation to isoamyl alcohol as compared with both wild-type or ODR-3::WT transgenic animals. Data shown represent the mean Ϯ S.E. for four trials totaling over 400 animals for each strain. B, right panel, modified recovery assay; animals were initially allowed to chemotax toward a point source of odorant (A) or control (C) on an assay plate in the absence of sodium azide. After 40 min, a fresh point source of odorant (new adaptation) plus sodium azide was generated on the same plate near the original control, and animals were tested for chemotaxis toward the new odorant for 1 h. Left panel, recovery index of wild-type and ARR-1 transgenic animals following adaptation to isoamyl alcohol represented by the animals at old adaptation subtracted from animals at new adaptation and divided by total number of animals. Both wild-type and ODR-3::WT transgenic animals recover from adaptation as represented by positive recovery indexes, whereas ODR-3::1-368 transgenics do not. Data shown represent the mean Ϯ S.E. for six trials totaling over 600 animals for each strain. C, binding of ARR-1 to GST-clathrin and GST-␤ 2 -adaptin. GST pull-down assays were performed using 10 g of GST-clathrin (top panel), GST-␤ 2 -adaptin (middle panel), or GST (bottom panel) bound to glutathione-agarose and 10 l of COS-1 cell lysates expressing HA-ARR-1 or HA-ARR-1-(1-368) as described under "Experimental Procedures." Bound ARR-1 was detected using an anti-HA monoclonal antibody. Full-length ARR-1 binds to both GST-clathrin and ␤ 2 -adaptin, whereas C-terminally truncated ARR-1 does not. The right half of each panel shows the relative expression levels of the ARR-1 constructs. D, internalization of ␤ 2 -adrenergic receptor in COS-1 cells overexpressing wild-type or C-terminally truncated ARR-1-(1-368). FLAG-␤ 2 AR and ARR-1 or ARR-1-(1-368) with or without an HA tag were co-expressed in COS-1 cells. Cells were incubated with or without 10 M isoproterenol for 15 min, and ␤ 2 -adrenergic receptor internalization was measured by enzyme-linked immunosorbent assay as described under "Experimental Procedures." Error bars represent the mean Ϯ S.E. from three independent experiments performed in triplicate.
of an organism to regain responsiveness to stimulus, a process referred to as recovery. Cell homeostasis is an important physiological result of recovery, as persistent adaptation leaves a cell unresponsive to its environment. The ability to recover from adaptation is reminiscent of the cellular process of resensitization that regulates GPCR responsiveness. In mammalian cells, arrestins play an important role in GPCR resensitization. Non-visual arrestins promote GPCR endocytosis, which in turn mediates receptor dephosphorylation and recycling to the cell surface, enabling the receptor to regain its responsiveness to stimuli (51). In C. elegans, recovery from adaptation was observed in wild-type animals (49) (Fig.  8, A and B). We hypothesized that this recovery might be analogous to the resensitization observed in mammalian cells and that ARR-1 might be involved. In an effort to address the role of ARR-1 in recovery, we expressed a C-terminally truncated ARR-1 in arr-1(ok401) worms using an odr-3 promoter that directs expression in AWC, AWB, AWA, and ASH chemosensory neurons. Expression of ARR-1-(1-368) effectively restores adaptation to isoamyl alcohol ( Fig. 6) but is unable to fully restore recovery from adaptation (Fig. 8, A and B). ARR-1-(1-368) lacks the region shown to be critical for interacting with clathrin and ␤ 2 -adaptin in mammalian non-visual arrestins, but retains the regions known to interact with receptor. Indeed, we were able to demonstrate that wild-type ARR-1 can bind to clathrin and ␤ 2 -adaptin and promote ␤ 2adrenergic receptor endocytosis in HEK293 cells, whereas ARR-1-(1-368) is defective in these assays (Fig. 8, C and D). Taken together, these results suggest an important role for ARR-1-promoted endocytosis in mediating the in vivo recovery from isoamyl alcohol-promoted adaptation.
Our results reveal an important role for ARR-1 in adaptation and recovery of GPCRs and provide a clear link between the cellular/molecular processes of desensitization and resensitization and the physiological phenomena of adaptation and recovery. The viability of arr-1 mutants in C. elegans opens up new avenues for researching the established functions of arrestin, the proposed roles for arrestin in receptor trafficking and signaling, and the discovery of new and undetermined behaviors involving arrestin activity.