Two Arylalkylamine N-Acetyltransferase Genes Mediate Melatonin Synthesis in Fish*

Serotonin N-acetyltransferase (arylalkylamine N-acetyltransferase, AANAT, EC 2.3.1.87) is the first enzyme in the conversion of serotonin to melatonin. Large changes in AANAT activity play an important role in the daily rhythms in melatonin production. Although a single AANAT gene has been found in mammals and the chicken, we have now identified two AANAT genes in fish. These genes are designated AANAT-1 andAANAT-2; all known AANATs belong to the AANAT-1subfamily. Pike AANAT-1 is nearly exclusively expressed in the retina and AANAT-2 in the pineal gland. The abundance of each mRNA changes on a circadian basis, with retinal AANAT-1 mRNA peaking in late afternoon and pineal AANAT-2 mRNA peaking 6 h later. The pikeAANAT-1 and AANAT-2 enzymes (66% identical amino acids) exhibit marked differences in their affinity for serotonin, relative affinity for indoleethylamines versusphenylethylamines and temperature-activity relationships. Two AANAT genes also exist in another fish, the trout. The evolution of two AANATs may represent a strategy to optimally meet tissue-related requirements for synthesis of melatonin: pineal melatonin serves an endocrine role and retinal melatonin plays a paracrine role.

Melatonin is synthesized in two vertebrate tissues, the pineal gland and the retina (1)(2)(3)(4)(5). Production of melatonin is almost always under control of a circadian clock, with the highest levels occurring at night (6 -10). The pineal gland produces circulating melatonin, which serves as an endocrine signal of the night period. This signal of night length is important in circadian and seasonal physiology (11,12). In contrast, retinal melatonin functions as a paracrine signal within the retina to enhance retinal functioning under low light conditions. This is done by shifting the positions of the rods and cones and regulating the turnover rates of the photoreceptive apparatuses of the rods and cones and surrounding pigment epithelium (5,13,14).
It has been generally thought that a single AANAT gene is expressed in both the pineal gland and retina. AANATs from various species are characterized by three highly conserved peptide sequences (C/c-1, D/c-1, and D/c-2; see Fig. 1) that have been proposed to be involved in arylalkylamine binding (15,17).
The single AANAT gene hypothesis is consistent with the results from studies in mammals and the chicken, but not with those from studies in teleosts (bony fishes). Biochemical investigations of the Northern pike indicate that pineal and retinal AANAT have distinctly different affinities toward arylalkylamine substrates and different temperature-activity relationships (18). Furthermore, mRNA analyses in both pike and trout indicate that a trout pineal-derived AANAT probe does not detect retinal AANAT mRNA (19). These observations are reconciled by the results presented here demonstrating the existence of a second AANAT gene in teleosts.

MATERIALS AND METHODS
cDNA Synthesis, Reverse Transcriptase PCR, and Genomic PCR-RNA was extracted from pike, trout, or sturgeon pineal glands and retinae as described below. cDNA was synthesized on oligo-dT magnetic beads (Dynal) as described previously (20). PCR was conducted in a 50-l volume containing 10 mM Tris (pH 8.3), 50 mM KCl, 1.5 mM MgCl 2 , 200 M dNTPs, 500 nM primers, and 0.5 units of ExTaq DNA polymerase (TaKaRa). The cycling regime was as follows: 94°C for 5 min; 30 -35 cycles of annealing at 55-60°C (55°C was used for primers from a species different from the cDNA; 60°C was used when primers and DNA were from the same species) and extension at 72°C for 1 min; and final extension at 72°C for 9 min. Conditions for amplification of genomic DNA were the same as for cDNA, except the annealing temperature was 58°C and 50 ng of DNA was used as template. PCR products were cloned into pGEM-T-Easy (Promega) for sequencing.
Determination of the Sequence of AANAT cDNAs-The initial step leading to identification of two pike AANATs (pAANATs) involved reverse transcriptase PCR generation of fragments of AANAT from sturgeon and trout pineal cDNA. A 179-bp sturgeon fragment (clone S12) was obtained using the sheep-based primers 562 and 563 (Table I).
A 138-bp trout fragment (clone T14/9) was then obtained using primer 562 and primer S02R, which was based on the sequence of the sturgeon clone S12. A longer trout fragment (clone T3; 184 bp) was obtained using primers T01R (based on trout clone T14/9) and D07L (a degenerate primer based on mammalian AANATs). Trout clone T3 overlapped with the sturgeon clone S12 by 78 bp and was 63% identical within the overlap region. Probes derived from these clones generated different patterns when used in Northern blot analysis: the sturgeon pinealbased probe (S12) detected multiple bands of mRNA in the trout and pike pineal gland and only a single band in the retina from these species, whereas the trout pineal-based probe (T3) only detected mRNA * 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  in the pike and trout pineal gland. Based on this, we hypothesized that two AANAT genes are expressed in trout and pike, with the one corresponding to S12 expressed preferentially in the teleost retina and the second corresponding to T3 expressed preferentially in the teleost pineal gland.
Sturgeon AANAT-derived primers S03L and S02R were used with pike retinal cDNA to obtain clone 11/15; primers derived from clone 11/15 were used to obtain a full-length sequence from pike retinal cDNA by 3Ј-RACE and 5Ј-RACE (see below). Trout AANAT-derived primers T04L and T05R were used with pike pineal cDNA to obtain clone 2/6; primers derived from clone 2/6 were used to obtain a full-length sequence from pike pineal mRNA by 3Ј-RACE and 5Ј-RACE.
3Ј-RACE was performed using pike AANAT-specific nested primers in conjunction with oligo-dT primers (GAATTCT 20 N). The initial round of PCR consisted of 10 cycles with the first specific primer only at 60°C annealing temperature, followed by the addition of oligo-dT primers and continuing the PCR at 50°C annealing temperature for 5 cycles then at 60°C for 20 cycles. An aliquot of this first reaction was used as a template for nested PCR employing 30 cycles at 60°C. To obtain sequence 3Ј of clone 11/15, immobilized pike retinal cDNA was amplified using nested primers (P05L and P06L) based on clone 11/15 sequence. To obtain sequence 3Ј of clone 2/6, pike pineal cDNA was amplified using nested primers (P01L and P02L) based on clone 2/6 sequence.
5Ј-RACE was performed by a modification of a previously described method using nested primers from the initial pike AANAT fragments and a primer corresponding to the amplimer ligated to the end of the cDNA as described (20). To obtain sequence 5Ј of clone 11/15, anchor oligonucleotide B was ligated onto immobilized retinal cDNA, and this cDNA was amplified using anchor primer AMPB and nested primers P32R and P30R, which were based on clone 11/15 sequence. To obtain sequence 5Ј of clone 2/6, anchor oligonucleotide A was ligated onto immobilized pineal cDNA, and this cDNA was amplified using anchor primer AMPA and clone 2/6-based primer P04R. A second round of 5Ј-RACE was done to further extend clone 2/6 using AMPA with nested primers P10R and P40R, which were based on the sequence of the product of the previous round.
The full-length sequence of AANAT expressed in the trout pineal gland was determined using a strategy similar to that used to determine the sequence of the AANAT mRNA expressed in the pike pineal gland. 3Ј-RACE used nested primers T03L and T04L. Two rounds of 5Ј-RACE were employed; the first used nested primers T01R and T05R while the second used T01R along with primer T10R, which was based on the product of the first round.
Multiple clones of all 3Ј-and 5Ј-RACE products were sequenced, and consensus sequences were generated. The contiguity of 3Ј-and 5Јflanking regions and the locations of intron/exon junctions were determined by PCR of genomic DNA. Clones containing the full ORFs were constructed as described below. Sequence analysis and data base searching were conducted using the GCG package from the University of Wisconsin (21).
Animal Culture-Pike (Esox lucius; 300 -1,500 g; male and female) were collected (mid-October to mid-March) from ponds of the Poitou-Charentes (France) and held in a commercial hatchery in small ponds supplied with river water under natural conditions. Pike were adapted for at least 24 h to the experimental photoperiod which reflected natural conditions (except dim red light was substituted for the dark period). Fish were killed by decapitation at selected times during 24-h cycles of white light/red light, constant white light, or constant red light conditions. Each lighting regime was set up and sampled during three sequential weeks. The entire experiment was conducted once in October and repeated in February; the natural lighting regime during both periods was approximately 11 h light/13 h dark. Tissues were obtained by rapid dissection and were frozen in liquid nitrogen for subsequent analysis.
Northern Blot Analysis-RNA extraction and Northern blot analysis were performed as described previously (19,22). Briefly, total RNA was extracted using a guanidine HCl-phenol method (TRIzol; Life Technologies, Inc.), fractionated on an agarose-formaldehyde gel, and transferred to a nylon membrane (Hybond-N, Amersham Pharmacia Biotech). Analytical probes for Northern blots were generated by PCR. The probe for AANAT-1 corresponds to bp 1009 -1628 in the 3Ј-untranslated region of pAANAT-1; the probe for AANAT-2 corresponds to bp 961-1590 in the 3Ј-untranslated region of pAANAT-2. Probes were labeled with 32 P by random priming. All blots were probed for both AANAT-1 and AANAT-2. For normalization, blots were stripped and probed with a random primed human ␤-actin cDNA (CLONTECH).

Expression of Glutathione S-Transferase-AANAT Fusion Proteins-
The full-length ORF of each pike AANAT cDNA was subcloned into the procaryotic expression vector pGEX-4T-3 (Amersham Pharmacia Biotech). The resulting expression clones were pRX50 for the form dominating in the retina and pPX36 for the pineal form. Both clones were expressed as glutathione S-transferase fusion proteins in BL21(DE3)-pLysS bacteria. Culture sonicates (2ϫ phosphate-buffered saline ϩ protease inhibitors ϩ 10 mM dithiothreitol) were diluted (0.5 M NaCl ϩ 50 mM sodium citrate ϩ 1 mM dithiothreitol) and assayed.
Enzyme Activity Analysis (23) ]acetyltyramine, a sample of the assay was mixed with ethanol containing the appropriate authentic carrier, the mixture was subjected to thin layer chromatography, products were located, corresponding gel was removed and product was extracted into scintillation fluid. Radioactivity was determined by routine methods.

Sequence Analysis
The deduced amino acid sequences of the two pAANAT clones are 66% identical and 76% similar (Fig. 1A). Both sequences are ϳ65-75% identical to known mammalian and avian AANAT clones (Fig. 1B). Both contain known conserved features of AANATs including two putative protein kinase A phosphorylation sites, a conserved putative ubiquitination site (K9), a histidine-rich stretch of residues proposed to be important in catalysis (residues 113-128) and two regions, described as motifs A and B, which characterize members of a large acetyltransferase superfamily referred to as the A/B or GNAT superfamily (15,17,22). Clone pRX50, from the pike retina, contains the three highly conserved peptides (C/c-1, D/c-1, and D/c-2) found in all previously published AANATs (17); this clone and the known AANATs will be designated members of a single AANAT-1 subfamily (15,(22)(23)(24)(25)(26)(27)(28). The corresponding peptides in clone pPX36, from the pike pineal gland, differ at 6 of 36 amino acids in these sequences. This clone, along with the trout pineal AANAT cloned here, will be considered to represent a second AANAT subfamily, designated AANAT-2. This segregation is also consistent with the results of computer-assisted phyloge- Analysis of genomic sequences indicates that the locations of the intron/exon junctions in the ORF of both pAANAT genes are identical to that of human, sheep, and chicken AANAT-1 (Fig. 1C, and Ref. 15). Genomic Southern blot analysis confirmed that each gene was present as a single copy and did not co-migrate following digestion with five different restriction enzymes. 3 This and the above results indicate that it is highly likely that AANAT-1 and AANAT-2 evolved from a common ancestor rather than resulting from convergent evolution or alternative splicing.

Tissue Distribution
Northern blot analysis reveals that pAANAT-1 and pAANAT-2 exhibit strikingly different expression patterns. pAANAT-1 is expressed at high levels only in the retina and pAANAT-2 only in the pineal gland. A trace amount of a 1.6kilobase pAANAT-2-positive transcript is also found in the ovary. Northern blot analysis did not detect expression of either gene in other tissues. Day/night differences in expression also occur in retina and pineal gland (discussed below).
pAANAT-2 is present in the pineal gland as 2.1-and 2.8kilobase transcripts; a single 2.2-kilobase pAANAT-1 transcript is found in the retina. The relative amounts of the two pAANAT-2 transcripts varied between samples for unexplained reasons not linked to experimental history or sample preparation, as judged from the actin mRNA signal. Although Northern blot analysis failed to reveal expression of pAANAT-1 2 E. Koonin (National Institutes of Health), personal communication. 3

Kinetic Characterization of Recombinant AANAT-1 and AANAT-2 Proteins
Temperature Dependence-Temperature-activity relationships of the recombinant enzymes were different. Maximal activity with 10 mM tryptamine for pAANAT-1 occurred at 37°C and that for pAANAT-2 at 18°C. The temperature profile of pAANAT-1 corresponds to that reported for pike retinal homogenates and that of pAANAT-2 to pike pineal homogenates (Ref. 18; Fig. 3A, inset), consistent with the strong expression of pAANAT-1 in the retina and of pAANAT-2 in the pineal gland (Fig. 2). It is interesting that recombinant pAANAT-1 and pAANAT-2 enzymes exhibit significant levels of activity at 0 to 4°C, as is also true of the native enzymes (18).
The substrate-product formation relationships of pAANAT-1 and pAANAT-2 also differ in that only pAANAT-1 generates a pattern where acetylation is inhibited at substrate concentrations above 0.3 mM with indoleethylamine substrates; this same pattern is observed at 4°C, as well as 37°C. Neither enzyme exhibits such a pattern with phenylethylamines. We failed to obtain evidence for product inhibition of pAANAT-1 with up to 1 mM N-acetylserotonin or melatonin. 3 This suggests that substrate inhibition is likely to explain the observed pattern.

Circadian Pattern of Abundance of Retinal pAANAT-1 and Pineal pAANAT-2
The abundance of AANAT mRNA transcripts changes on a 24-h basis in both the retina and pineal gland in animals maintained in a lighting environment providing 12 h of light and 12 h of darkness (Fig. 4). The amplitude of the retinal AANAT-1 mRNA rhythm is Ͼ20-fold and that of pineal pAANAT-2 mRNA is Ͼ8-fold. The phasing of the rhythms differs, with the retinal AANAT-1 mRNA rhythm peaking at dusk, 6 h prior to the middark period peak in pineal AANAT-2 mRNA.
Daily rhythms in the abundance of retinal pAANAT-1 mRNA and pineal pAANAT-2 mRNA persist under constant lighting conditions of dim red light or constant light, indicating they are not dependent upon light/dark transitions but are controlled by an endogenous circadian clock.

Two AANAT Genes Explains Paradoxical Results in Pike-
The existence of a second AANAT gene in pike resolves two paradoxical observations regarding AANAT in pike retina and pineal gland (18,19). First, the report that AANAT mRNA was undetectable in the pike retina is explained by the finding that the trout pineal-derived probe used in those studies was AANAT-2-selective and did not detect pAANAT-1 under the conditions used, in part reflecting only a 68% bp match. As reported here, pike retina preferentially expresses AANAT-1; AANAT-2 is not detectable in the retina by Northern blot analysis.
The second paradox explained by these findings is that the kinetics of AANAT in the pike pineal gland and retina are different. This is explained by the dominant expression of pAANAT-1 in the retina and pAANAT-2 in the pineal gland, and the inherent differences in the proteins encoded by pAANAT-1 and pAANAT-2. This is supported by the remarkable similarity of the published kinetic characteristics of pike retinal AANAT and recombinant pAANAT-1 and that of pineal AANAT and recombinant pAANAT-2. It is noteworthy that retinal AANAT activity and recombinant pAANAT-1 appear to have nearly identical temperature dependence and substrate specificity characteristics, as did pineal AANAT activity and recombinant pAANAT-2, despite the fact that native enzymes were studied in crude pineal or retinal homogenates and fusion proteins were studied in bacterial extracts. The remarkable similarities observed between recombinant pAANAT-1 and retinal AANAT versus recombinant pAANAT-2 and pineal AANAT provide strong indications that the differences seen are valid reflections of the AANAT proteins and are not caused by other proteins or factors in the assay. Further investigations with pure preparations of the expressed proteins are required to fully characterize and provide a molecular explanation for these differences.
Functional Significance of Two AANAT Genes-It is not unreasonable to consider that the retinal/pineal expression pattern of pAANAT-1 and pAANAT-2 reflects a fine-tuning of AANAT's regulatory role in melatonin synthesis in each tissue. As indicated above, rhythmic changes in melatonin synthesis are regulated by changes in AANAT activity. Recent studies have established that AANAT activity is modulated through control of enzyme protein levels by regulated proteolysis (16). With this type of regulation, changes in enzyme protein will be most effectively converted into changes in product formation if the K m value approximates the substrate concentration. This is consistent with our finding that pAANAT-1 is expressed in the retina, which typically has a relatively low concentration of serotonin and that pAANAT-2 is expressed in the pineal gland, FIG. 2. Tissue distribution of pAANAT-1 and pAANAT-2. Tissues were removed from an adult pike at 2400 h and frozen in liquid nitrogen. Total RNA was extracted, and each lane was loaded with 20 g. The same blot was probed sequentially for both pAANAT-1and pAANAT-2 and then for ␤-actin. Similar results were obtained with tissues taken at midday and also with a second series of midnight and midday tissues. Molecular size markers are indicated to the left. kb, kilobases.
which contains one of the highest concentrations of serotonin (29). The low K m of pAANAT-1 appears to best support melatonin synthesis if very low levels of serotonin are available, as is the case in the retina. This level of synthesis appears to be sufficient for local demands, and the higher K m of pAANAT-2 in the pineal gland is consistent with the higher concentration of serotonin there and the role of the pineal gland as a factory for circulating melatonin.
There may also be functional significance to the differences in substrate specificity between the proteins encoded by pAANAT-1 and pAANAT-2. The former acetylates indoleethylamines and phenylethylamines equally well, whereas the latter exhibits strong preference for indoleethylamines. This weak affinity of pAANAT-2 for phenylethylamines eliminates or minimizes acetylation as a factor influencing phenylethylamine metabolism in the pineal gland. Likewise, the strong affinity of pAANAT-1 for phenylethylamines suggests the enzyme might function in the retina to control phenylethylamine levels. This is of interest because 3,4-dihydroxyphenylethylamine (dopamine) is an important retinal neurotransmitter (5,9).
The remarkable activity-temperature dependence difference between these two enzymes raises two questions: one being the molecular basis of the difference, and the second being the physiological significance. It has been suggested that molecular adaptation of closely related enzymes in different fish species to different physiological temperatures can be achieved through modification of flexibility and mobility within the molecule (30). The recent publication of the crystal structure of ovine AANAT-1 may help determine whether sequence differences between pAANAT-1 and pAANAT-2 could result in sim- ilar effects (31). The physiological significance is difficult to explain but may relate to the fact that the Northern pike is a cold water fish; the steep rise in pAANAT-2 activity within the range of seasonal temperatures may be important in integrating environmental parameters.
Two AANAT Genes-Exception or the Rule?-The retinal/pineal expression pattern of AANAT-1 and AANAT-2 seen in pike is also present in trout because trout AANAT-1 was previously cloned from retina (28), and AANAT-2 was cloned from the trout pineal gland in this study. However, the strict pattern of expression of AANAT-1 in the retina and of AANAT-2 in the pineal gland is not conserved within or across classes. For example, AANAT-1 is expressed at high levels in both the retina and pineal gland of the chicken and some mammals (22,(25)(26)(27), and AANAT-2 is expressed at high levels in both the pineal gland and retina of zebrafish (19). Accordingly, speciesto-species variations in tissue expression of AANAT-1 or AANAT-2 might reflect unique species-and tissue-dependent requirements for acetylation of serotonin and perhaps other arylalkylamines. Such biochemical and genetic variation is consistent with the anatomical and molecular diversity of these tissues (15,(32)(33)(34) and the important phototransduction roles the pineal gland and retina play in survival.
The question of whether AANAT-2 is expressed in non-teleosts remains open at this time. Efforts to clone chicken and human AANAT-2 s have been unsuccessful so far. 3 Based on the degree of difference between pAANAT-1 and AANAT-2, it is not unreasonable that pAANAT-1 and pAANAT-2 existed prior to the emergence of teleosts; both might have been present in a vertebrate ancestor common to both teleosts and higher vertebrates. However, it is also possible that AANAT-2 appeared only after fish emerged from the vertebrate line that led to higher vertebrates.
Circadian Patterns of Expression of pAANAT-1 and pAANAT-2-The circadian expression of pAANAT-1 and pAANAT-2 is consistent with previous reports indicating that there are rhythms in AANAT mRNA in the chicken retina and in the pineal glands of the rat, pike, and chicken. This appears to reflect the role of endogenous clocks which determines that melatonin synthesis can be significantly elevated only during the night and not during the day. This improves the reliability of melatonin as an indicator of the night period. The clock is reset and modulated by light. However, as indicated in the Introduction, the clock is not the only mechanism controlling AANAT activity. An additional control mechanism involves light acting downstream of the clock to suppress AANAT activity; this apparently involves light-initiated proteasomal proteolysis of AANAT protein (16). This provides a mechanism through which the system can respond immediately to changes in photoperiod length that are not anticipated by the endogenous clock. This is of special significance in considering the importance of the different phasing of the rhythms of retinal AANAT-1 and pineal AANAT-2 mRNAs.
We suspect that the earlier peak in retinal AANAT-1 mRNA may reflect a requirement for maximal melatonin production early in the night period, perhaps to optimally promote dark adaptation. As a result, as soon as light levels fall below a threshold and retinal AANAT-1 proteolysis ceases, retinal