The pineal gland (epiphysis) is locatedat the dorsal edge of the diencephalonin the zebrafish. The conserved functionof this organ in all vertebrates is thesynthesis and secretion of melatonin, ahormone that regulates a variety of cir-cadian and circannual physiologicalprocesses (Arendt, 1995; Falcon, 1999;Klein, 2004). Melatonin levels are highat night and low during the day, as aconsequence of regulated transcriptionand stability of serotonin-N-acetyl-transferase (AANAT), the rate-deter-mining enzyme of melatonin synthesis.
In zebrafish and certain other non-mammalian vertebrates, the melatoninproducing cells of the pineal gland arephotoreceptors that can rhythmicallyproduce melatonin for several days inisolation, reflecting the presence of anautonomous circadian clock pacemakerwithin these photoreceptor cells (Ber-nard et al., 1997; Begay et al., 1998;Falcon, 1999). Therefore, the fish pinealphotoreceptor cell is a valuable modelsystem to study circadian function, pho-todetection, and melatonin production.
In addition, the zebrafish pinealgland is the first site where neurogen-
esis occurs, being apparent at approx-imately 24 hours post fertilization(hpf; Chitnis and Kuwada, 1990; Wil-son and Easter, 1991). The existenceof neuronal cells in the pineal glandwhich send projections to the brainmakes this tissue more heterogeneousas compared to the pineal gland ofmammals (Masai et al., 1997). Theneuronal patterning surrounding thepineal gland is regulated by the ho-meobox transcription factor floatinghead (flh) and by masterblind (mbl),which encodes the negative regulatorof wnt signaling Axin. Flh�/� ze-
Additional Supporting information may be found in the online version of this article.1Laboratory of Molecular Genetics, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda,Maryland2Unit on Biologic Computation, Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, Bethesda,Maryland3Department of Neurobiology, George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, Israel4Section on Neuroendocrinology, Program on Developmental Endocrinology and Genetics, Eunice Kennedy Shriver National Institute ofChild Health and Human Development, NIH, Bethesda, Maryland†Dr. Chen’s present address is Advanced Biomedical Computing Center, NCI-Frederick, MD.*Correspondence to: Reiko Toyama, Bldg. 6B, Room 420, 9000 Rockville Pike, Bethesda, MD 20852.E-mail: [email protected]
DOI 10.1002/dvdy.21988Published online 29 May 2009 in Wiley InterScience (www.interscience.wiley.com).
brafish show reduced neuronal pro-duction in the pineal gland, whereasmutations in mbl increase the numberof pineal neurons throughout the dor-sal forebrain (Masai et al., 1997). Fur-thermore, the basic helix loop helix(bHLH) transcription factors achaete/scute homologue 1a (ascl1a) and neu-rogenin1 (ngn1) act downstream of flhto regulate neurogenesis (Cau andWilson, 2003). The molecular mecha-nisms of pineal gland developmentand function beyond these initialsteps of neurogenesis have not beenfully explored. Recently, pineal devel-opment and its relationship to brainasymmetry has received considerableattention (Gamse et al., 2003, 2005;Halpern et al., 2003; Aizawa et al.,2005, 2007; Hendricks and Je-suthasan, 2007; Kuan et al., 2007a,b).Asymmetry depends on the lateralityof the parapineal and is controlled byNodal signaling (Concha et al., 2000,2003; Liang et al., 2000).
Gene profiling of the pineal gland ofthe chicken and rat have identifiedmany genes that are highly expressedin the pineal gland, show night/daydifferences, or both (Humphries et al.,2002; Bailey et al., 2003, 2008; Fuku-hara and Tosini, 2008). Here we re-port the results of global transcrip-tome analysis of the zebrafish pinealgland taken from day 3 larvae toadults. Many highly abundant tran-scripts that had not been previouslyreported to be present in this tissuehave been identified. Furthermore,many highly expressed genes werefound to dynamically change their ex-pression levels during development.These data provide a broad basis forfurther molecular analysis of pinealgland development and physiology.
RESULTS
As a first step in data analysis, allpineal gland data were averaged in-cluding data obtained during the dayand night and at all five developmen-tal stages; brain data were treatedsimilarly. Probe sets were selectedwith the following criteria: P value�0.05 and pineal/brain signal ratio �5. Among the total 15,503 probe setsin the globally averaged data pool, 94met these criteria. Of these, 43 probesets have been annotated (http://www.affymetrix.com/analysis/index.affx),
and nearly half of them (21 probe sets)correspond to genes known to behighly expressed in the pineal gland,including aanat2 (Gothilf et al., 1999),floating head (Talbot et al., 1995), ex-tra-ocular rhodopsin (Mano et al.,1999), phosducin (accession numberXM_677731 ), Crx (Liu et al., 2001),and otx5 (Gamse et al., 2002). Expres-sion of GFP in the pineal gland of thetransgenic fish used in this study wasconfirmed. These observations providea first-level indication that our dataeffectively discriminate betweengenes that are differentially expressedin the pineal gland and the brain. Fur-thermore, principle component analy-sis of individual repeats indicated thatthe data are of high quality (notshown).
Genes were considered to be highlyexpressed in the pineal gland relativeto brain if the probability of a differ-ence was � 0.05. Setting the absolutedifference at greater than threefoldchanges the number of selected genes(more accurately probe sets) shown inTable 1. The number of genes highlyexpressed in the pineal gland identi-fied in this manner is much higher inadults than in larvae, possibly reflect-ing functional maturation of the tis-sue. Similar numbers of transcriptswere enriched in the samples collectedduring the day and night. Probe setsselected at a threefold criterion in thethree larval stages or the two adultstages partially overlapped, as shownin Figure 1. Approximately 20–35% ofthe transcripts detected at each stagewere not detected at other stages,while approximately 60% of probe setsoverlapped in RNA samples of 3month and 1- to 2-year-old zebrafish.The total number of nonoverlappingprobe sets at larval stages was 128during the day and 150 at night, while1,018 and 1,017 nonoverlapping probesets were selected in adult day andnight samples, respectively. Approxi-mately 60% of probe sets identified ashighly expressed in the larval pinealgland were also found to be enrichedin adult tissue.
We next compared the results of dayand night analysis. Among the 128
genes highly expressed in the larvalpineal gland relative to the brain atday and the 150 genes highly ex-pressed at night, 83 were expressedboth at day and night. Because morethan 1,000 probe sets were selected aspineal enriched transcripts in adultsamples under these criteria, we ap-plied more stringent criteria to selectgenes. Increasing the required pineal/brain difference from three- to fivefoldwhile maintaining all other condi-tions, the numbers of nonoverlappingprobe sets highly expressed in adult (3month and 1–2 years old) pinealglands were 322 and 365 at day andnight, respectively; among these, 197were expressed during both day andnight.
The 50 annotated genes most highlyenriched in the pineal gland in at leastone of the stages studied are pre-sented in Table 2 (larvae) and Table 3(adult). More complete information onthe genes selected by our criteria atlarval and adult stages are listed inSupplementary Table S1 (3� enrich-ment, which is available online) andSupplementary Table S2 (5� enrich-ment). These lists include many moregenes highly expressed in the pinealgland than previously identified.Many of these genes code for proteinsinvolved in photoreceptor signaltransduction pathways including Gproteins and cGMP specific phospho-
TABLE 1. Number of GenesEnriched in the Pineal Gland asCompared to Brain (P/B), Listedat Different Enrichment Ratiosa
aSamples were analyzed at midday andmidnight at each developmentalstage. The following criteria were usedfor the selection of probe sets. Averageminimum signal � 200; P value�0.05.
1814 TOYAMA ET AL.
diesterase. We also identified a ho-molog of unc119, rbp4 (retinol bindingprotein 4), zic family members, andiroquois homeobox proteins as tran-scripts that are pineal enriched ascompared to the brain. The unc119homolog was studied further (see be-low).
At least two habenula-specificgenes, cadps2 and leftover, were alsoidentified as pineal enriched genes(Table 2, #32 and #36, respectively).Because habenular nuclei are locatedvery close to the pineal gland at larvalstages, this may be due to contamina-tion of the pineal sample with habe-nular tissue. As an independent ap-proach to test the quality of theanalysis, we picked a random series ofunannotated clones and analyzedtheir expression pattern by in situ hy-bridization. Twelve of 19 clones ana-lyzed showed enhanced expression inthe pineal gland relative to the sur-rounding brain (Supp. Fig. S1, whichis available online), while the remain-ing clones, whose expression levels
were moderate to weak in the microar-ray analysis, were not detected by insitu hybridization (Supp. Fig. S1).
Analysis of Genes HighlyExpressed in the PinealGland
To obtain functional insights into thecategories of genes highly expressedin the pineal gland, Gene Ontology(GO) analysis was done using humanhomologs, as described in the Experi-mental Procedures section. Humanhomologs could be identified for 43–50% of zebrafish probe sets annotatedfor the Affymetrix Genechip. Approx-imately 50% of zebrafish probe setsare not well annotated (referred to as“transcribed loci” or “hypothetical pro-teins”) and were not assigned humanhomologues. Similar GO analysisdone using the available zebrafishgene names was less informative(data not shown).
A summary of the GO analysis at Bi-ological Process level 4 of genes highlyexpressed in the pineal gland is shownin Figure 2. For this analysis, we se-lected at threefold enrichment, nonre-dundant probe sets for the three larvalstages and for the two adult stages,both at day and night (Fig. 1), and con-verted these groups of genes to theirhuman homologs. The most highly rep-resented GO terms are related to pho-toreception (e.g., “visual perception”and “detection of light stimulus”), bothin larval and adult stages. Terms forneuronal development and function(“neurotransmitter metabolism,” “cen-tral nervous system development,” and“transmission of nerve impulse”), “ni-trogen compound biosynthesis,” and“aromatic compound metabolism” wereenriched at larval stages only, while GOterms related to programmed cell deathand signal transduction (“intracellularsignaling cascade” and “regulation ofsignal transduction”) were found only atadult stages (Fig. 2). An additional GOanalysis at Molecular Function level 4showed that the terms “retinal,” “retinolbinding,” and “G-protein coupled photo-receptor activity” were found promi-nently in all four samples (Supp. Fig.S2); again these terms suggest enrich-ment of genes related to visual percep-tion.
Pathway analysis, as described inthe Experimental Procedures section,identified “phototransduction path-way” as the most likely representedpathway in all larval and adult sam-ples (Fig. 3). “Axonal guidance signal-ing” was also enriched in adult butless so in larval samples. Several sig-naling pathways were enriched onlyat adult stages, and among them “p53signaling” was particularly interest-ing because of its involvement in pro-grammed cell death which was foundto be enriched in the GO analysisabove (Fig. 2).
These analyses indicate that pinealgene expression is enriched in classesof genes among which those involvedin phototransduction are most promi-nent, consistent with the function ofthe fish pineal gland as a photorecep-tor organ. Furthermore, significantdifferences were found between genesets enriched in larval and adult pi-neal glands.
Fig. 1. Overlap of pineal-enriched genes at different stages. Venn diagrams of genes enriched inthe pineal gland threefold or higher compared with the brain at different stages; larval and adultstages are shown separately. The total number of nonoverlapping enriched genes is 128 (larvae,day), 150 (larvae, night), 1,018 (adult, day), 1,017 (adult, night).
ZEBRAFISH PINEAL GLAND TRANSCRIPTOME 1815
TABLE 2. Top 50 Pineal Gland Enriched Genes at Larval Stagesa
Probe set Gene titleGenesymbol P/B
1 Dr.352.1.S1 Floating head flh 66.42 Dr.9835.1.S1 Guanine nucleotide binding protein (G protein), gamma transducing activity
polypeptide 1gngt1 57.5
3 AFFX-Dr-U43284-1_s GFP GFP 47.74 Dr.10292.1.S1 Retinol binding protein 4, like rbp4l 37.45 Dr.9876.1.S1 Guanine nucleotide binding protein (G protein), gamma transducing activity
polypeptide 2gngt2 30
6 Dr.9829.1.S1 Phosphodiesterase 6G, cGMP-specific, rod, gamma pde6g 26.57 Dr.9853.1.A1 Phosphodiesterase 6A, cGMP-specific, rod, alpha pde6a 22.58 Dr.9908.1.A1 Similar to ENSANGP00000004777/LOC557454 21.79 Dr.12451.1.S1 Retinal pigment epithelium-specific protein a rpepa 17.910 Dr.5738.1.S1 Similar to interphotoreceptor retinol-binding protein/LOC563355 17.711 Dr.12451.2.A1 Retinal pigment epithelium-specific protein a rpepa 15.212 Dr.9899.1.S2 Guanine nucleotide binding protein (G protein), alpha transducing activity
polypeptide 1gnat1 12.7
13 Dr.8142.1.S1 Arylalkylamine N-acetyltransferase aanat2 11.314 Dr.9871.1.A1 Recoverin rcv1 9.515 Dr.8099.1.S1 Extra-ocular rhodopsin exorh 9.416 Dr.12762.1.A1 Phosducin 2 / similar to Pdc2 protein/LOC100007784 pdc2 917 Dr.12469.1.S1 Arrestin 3, retinal (X-arrestin), like arr3l 918 Dr.19931.1.S1 Tryptophan hydroxylase 1 (tryptophan 5-monooxygenase) tph1 819 Dr.20586.1.A1 Similar to agouti related protein 2/LOC796595 7.620 Dr.15967.1.A1 Tryptophan hydroxylase 1 (tryptophan 5-monooxygenase) tph1 7.421 Dr.9845.2.A1 ADP-ribosylation factor-like 3, like 2 arl3l2 7.122 Dr.9841.1.A1 Phosphodiesterase 6C, cGMP-specific, cone, alpha prime pde6c 6.723 Dr.11240.1.A1 Similar to gefiltin/LOC555251 6.424 Dr.14052.1.A1 Tryptophan hydroxylase 2 (tryptophan 5-monooxygenase) tph2 6.325 Dr.24898.1.S1 Chromosome 20 open reading frame 149 homolog (human) c20orf149 6.326 Dr.11286.1.S1 Phospholipase A1 member A pla1a 6.227 Dr.11085.1.A1 Retinaldehyde binding protein 1, like rlbp1l 628 Dr.5167.1.A1 Similar to lambda-recombinase-like protein/LOC100004795 629 Dr.12902.1.A1 Cytochrome P450, family 11, subfamily B, polypeptide 2 cyp11b2 5.630 Dr.9899.1.S1 Guanine nucleotide binding protein (G protein), alpha transducing activity
polypeptide 1gnat1 5.6
31 Dr.10689.1.S1 zic family member 2 (odd-paired homolog, Drosophila) b zic2b 5.532 Dr.6658.1.A1 Ca2�-dependent activator protein for secretion 2 cadps2 5.433 Dr.14325.1.S1 Cone-rod homeobox crx 5.434 Dr.11305.1.A1 Guanylate kinase 1 guk1 535 Dr.12592.1.S1 Guanylate cyclase activator 1A guca1a 4.936 Dr.17145.1.S1 Potassium channel tetramerization domain containing 12.1/similar to
leftover/LOC796664kctd12.1 4.9
37 Dr.16367.1.A1 Similar to centrin/LOC795513 4.638 Dr.9845.1.S1 ADP-ribosylation factor-like 3, like 2 arl3l2 4.539 Dr.16724.1.A1 Transcribed locus, strongly similar to NP_001076421.1 si:ch211-221n23.1 4.440 Dr.26347.1.A1 Pyrophosphatase (inorganic) pp 4.341 Dr.26319.1.A1 Dopa decarboxylase ddc 4.142 Dr.284.2.A1_a Orthodenticle homolog 1 otx1 4.143 Dr.22887.1.A1 Similar to zinc finger protein Zic6/LOC796374 444 Dr.13970.1.S1 ADP-ribosylation factor-like 4a arl4a 445 Dr.4807.1.S2 zic family member 2 (odd-paired homolog, Drosophila), a zic2a 3.946 Dr.12624.1.S1_a Iroquois homeobox protein 1, b irx1b 3.947 Dr.10724.1.S1 Eomesodermin homolog a eomesa 3.848 Dr.1730.1.A1 Similar to complement control protein factor I-B/LOC557557 3.649 Dr.9881.2.A1 Guanine nucleotide binding protein (G protein), alpha transducing activity
polypeptide 2gnat2 3.6
50 DrAffx.2.17.S1 Gamma-aminobutyric acid (GABA) B receptor, 1 gabbr1 3.5
aValues of P/B fold change for all six individual analyses (3d day, 5d day, 10d day, 3d night, 5d night, 10d night; see Supp. TableS1 for individual values) were averaged.
1816 TOYAMA ET AL.
TABLE 3. Top 50 Pineal Gland Enriched Genes in Adult Stagesa
Probe set ID Gene titleGenesymbol P/B
1 Dr.10292.1.S1 Retinol binding protein 4, like rbp4l 211.42 Dr.12592.1.S1 Guanylate cyclase activator 1A guca1a 203.53 Dr.12469.1.S1 Arrestin 3, retinal (X-arrestin), like arr3l 176.74 AFFX-Dr-U43284-1_s GFP GFP 169.75 Dr.12451.1.S1 Retinal pigment epithelium-specific protein a rpepa 162.56 Dr.9908.1.A1 Similar to ENSANGP00000004777/LOC557454 154.67 Dr.9853.1.A1 Phosphodiesterase 6A, cGMP-specific, rod, alpha pde6a 130.38 Dr.9841.1.A1 Phosphodiesterase 6C, cGMP-specific, cone, alpha prime pde6c 1289 Dr.9871.1.A1 Recoverin rcv1 125.210 Dr.11305.1.A1 Guanylate kinase 1 guk1 11911 Dr.8099.1.S1 Extra-ocular rhodopsin exorh 110.812 Dr.9899.1.S1 Guanine nucleotide binding protein (G protein), alpha transducing activity
polypeptide 1gnat1 101.8
13 Dr.5738.1.S1 Similar to interphotoreceptor retinol-binding protein/LOC563355 94.714 Dr.9899.1.S2 Guanine nucleotide binding protein (G protein), alpha transducing activity
40 Dr.4833.1.S1 Annexin A1c/zgc:86853 anxa1c 21.741 Dr.19801.1.A1 Similar to fatty acid binding protein 1b/LOC795525 21.442 Dr.994.1.S1 Claudin b cldnb 21.343 Dr.24943.1.S1 Similar to Desmoglein 2/LOC560026 21.344 Dr.11283.1.A1 Solute carrier family 25 (mitochondrial carrier; phosphate carrier),
member 3, likeslc25a3l 21.1
45 Dr.25140.7.A1_a Tumor-associated calcium signal transducer/zgc:110304/ proteinLOC791868/similar to pan-epithelial glycoprotein/LOC100000093
aAdult stages (see Supp. Table S2 for individual values) were analyzed as described in Table 2.
ZEBRAFISH PINEAL GLAND TRANSCRIPTOME 1817
Fig. 2. GO term analysis of pineal-enriched genes at biological processes (BP) level 4. GO terms with P values � 0.1 (larvae) and � 0.05 (adult) wereselected. Nonoverlapping genes were considered (Fig. 1). D, day, and N, night. The width of each slice indicates the number of genes in a given GOterm. The numbers next to the colored squares indicate the GO terms, arranged by P values (small to large), corresponding to the numbers inparentheses in the table below. These P values indicate the probability of finding this GO term occupancy by chance, with a blank space indicatingthat the GO term was not enriched in the pineal gland. The color scheme starts at 12 o’clock and proceeds clockwise.
Developmental changes in pineal geneexpression were identified by search-ing for genes that exhibited a � three-fold difference between the highestand lowest signal value among the fivestages examined (3 days, 5 days, 10days, 3 months, and 1–2 years), hav-ing a signal � 200 in at least onestage, and P value � 0.05. This anal-ysis does not consider enrichment inthe pineal relative to the brain. Usingthese criteria, 2,370 probe sets were
selected from daytime pineal glanddata as changing during development.Cluster analysis was performed withthese probes, and the set was dividedinto four subclusters (Fig. 4). In broadterms, subsets A and B contain geneswhose expression levels are low dur-ing larval stages but increased atadult stages, while subsets C and Dshow the opposite profile. GO analysisof clusters A and C was done, whilesubsets B and D contain too few genesfor this analysis. Subsets A and C,which contained 1,191 and 733 ze-brafish probe sets, respectively, wereconverted to 552 and 299 human ho-
mologs as indicated above. The resultsof GO analysis at Biological Processlevel 5 and Molecular Function level 3are shown in Figure 5 and Supple-mentary Figure S3, respectively.
Similar analysis with the pinealgland “night” samples produced es-sentially similar results (Supp. Figs.S4, S5, S6).
In subset A, GO terms related tophototransduction were considerablyenriched (Fig. 5; Supp. Figs. S3, S5,S6). Furthermore, terms related tocell death are also represented (Fig. 5;Supp. Fig. S5). These GO terms werefound in genes enriched in the adultpineal gland, reflecting physiologicalfunction rather than development ofthe pineal gland. In contrast, termsrelated to transcription were highlyenriched in subset C (Fig. 5; Supp.Figs. S3, S5, S6). Likewise, GO termssuch as “brain development” and“neuron differentiation” were found ingenes enriched at larval stages (Fig. 5;Supp. Fig. S5). These terms clearlyreflect developmental changes that oc-cur during these stages.
To illustrate the classes of genesthat represent the different categorieswe chose GO terms that were mostsignificantly represented in each sub-set. For each of these terms, we showthe genes included as hierarchicalclusters in Figures 6 (subset A) and 7(subset C). It is apparent that groupsof genes included in certain GO termsare co-regulated at different stages inthe pineal gland.
Analysis of Genes ThatExhibit Night/DayExpression Differences inthe Pineal Gland
The mid-day and mid-night gene ex-pression profiles of the pineal glandwere compared at different develop-mental stages. Although this ap-proach does not necessarily reflect cir-cadian clock control of gene expressionand is likely to miss changes that donot peak at midday or midnight, itprovides a useful approach to charac-terizing daily changes in gene expres-sion.
Probe sets were selected using thefollowing criteria: P value �0.05 andnight and day signal ratio �1.5 or 2(Fig. 8). Approximately 250 geneswere identified that showed at least
Fig. 3. Canonical pathways enriched in the pineal gland. Pathways enriched in all four samples(larvae day [D] and night [N], adult day [D] and night [N]) are shown at the top. The pathways shownat the bottom are enriched mainly in the adult pineal gland.
ZEBRAFISH PINEAL GLAND TRANSCRIPTOME 1819
�1.5-fold differences in expression be-tween day and night in four of the fivestages tested. There was no signifi-cant difference in the number of genesselected between larvae at d5 and d10and adults at 3 months and 1–2 years,however, a higher number of genesexhibited day/night differences at lar-
val stage d3. The biological meaning ofthis result remains to be elucidated.
The probe sets showing day/nightdifferences in at least four of the fivedevelopmental stages are listed in theSupplementary Table S3. aanat2 wasselected as a gene highly expressed atnight, as expected. Another gene
highly expressed at night is bHLH do-main containing class B, an inhibitorof the BMAL:CLOCK circadian tran-scription activator. The circadianclock gene per2 and rbp4 (retinal bind-ing protein 4), which was also identi-fied as highly expressed in the pinealgland compared with the brain, wereselected as being highly expressedduring the day. These results furtherstrengthen the validity of the microar-ray analysis and the genes identifiedhere.
unc119 Homolog Is HighlyEnriched in the PinealGland
Unc119 homolog (Dr. 9908, Supp. Ta-ble S1) was identified as a gene ofspecial interest because it is highlyexpressed in the pineal gland at bothday and night compared with thebrain at all stages we examined, andmight play a role in pineal gland de-velopment and/or function. Becausethis is the third family member ofunc119 found in the zebrafish, we re-fer to this gene as unc119c.
Unc119c shares 45.7% amino acididentity with the previously publishedzebrafish Unc119 (Manning, 2004)and is clearly most diverged fromUnc119 homologues in other species(data not shown). In situ hybridiza-tion experiments revealed thatunc119c is expressed in the pinealgland at 72 hpf (Fig. 9A). The tran-script is not found at the 20 somitestage or earlier (data not shown).
Recently, ADP-ribosylation factor-like protein 2 (ARL2) has been identi-fied as an interacting protein of a hu-man homologue of UNC119, HRG4(Kobayashi et al., 2003). Of interest,members of the arl gene family, arl3l1(ADP-ribosylation factor-like protein3 like 1) and arl3l2, were highly en-riched in the pineal gland relative tobrain in our microarray data. arl3l2 isexpressed in the zebrafish pinealgland, as visualized by in situ hybrid-ization (Fig. 9B), suggesting a possibleinteraction between Unc119c andArl3l2.
To test the possibility that Unc119cand Arl312 interact, we performedco-immunoprecipitation studies withFLAG-tagged Unc119c and myc-tagged zebrafish Arl3l2 in culturedcells. Arl3l2 specifically interacted
Fig. 4. Hierarchical clustering of genes whose expression levels change during the day in thepineal gland during development. Genes were selected using the following criteria: averagemaximum signal �200; lowest-to-highest signal ratio � 3; P value �0.05. Low expression: green;high expression: red. Samples were grouped into four subsets, A–D. 1, 3 days; 2, 5 days; 3, 10days; 4, 3 months; 5, 1–2 years.
1820 TOYAMA ET AL.
Fig. 5. GO term analysis of subsets A and C of Figure 4 at biological processes (BP) level 5. GO terms which P values �0.05 were selected. Thenumbers next to the colored squares correspond to the numbers in the table below; see also legend to Figure 2.
ZEBRAFISH PINEAL GLAND TRANSCRIPTOME 1821
with Unc119c (Fig. 9C). This suggeststhat these two proteins may be part ofa common functional pathway in thezebrafish pineal gland.
DISCUSSION
Genes Highly Expressed inthe Pineal Gland
In this study, we have systematicallyexamined the gene expression profilesin the larval and adult zebrafish pi-neal gland. To identify genes whoseexpression is enriched in the pinealgland, we used brain tissue withoutpineal gland and without eyes as ref-erence. Eyes were removed from con-
Fig. 6. Clustering of genes in the selected GO terms shown in Figure 5, subset A. Each row of the heatmap (see Fig. 4 for explanations) corresponds to one probe set. Affymetrix probe set ID numbers andgene symbols are shown on the left. 1, 3 days; 2, 5 days; 3, 10 days; 4, 3 months; 5, 1–2 years.
Fig. 7. Clustering of genes in the selected GO terms shown in Figure 5, subset C. Each row of theheat map corresponds to one probe set. Affymetrix probe set ID numbers and gene symbols areshown on the left. 1, 3 days; 2, 5 days; 3, 10 days; 4, 3 months; 5, 1–2 years.
Fig. 9. unc119c and arl3l2 are co-expressedin the pineal gland and interact with each other.A,B: Spatial expression of zebrafish unc119c(A) and arl3l2 (B). Both panels are shown asdorsal views of 3 dpf larvae. In situ hybridizationwas carried out as described (Toyama and Da-wid, 1997). C: Co-immunoprecipitation of ze-brafish Unc119c and Arl3l2 transfected intoHEK293 cells. The antibody used for precipita-tion is indicated by IP, and the antibodies usedfor blotting are shown on the right side of thepanels. Myc-tagged Arl3l2 co-precipitated withFlag-Unc119c (white arrowhead in the toppanel, shown above immunoglobulin lightchain.), but Myc-tagged nucleolin (Myc-Nucleo)did not. Black arrowheads indicate Myc-taggedArl3l2 and Nucleolin (used as negative control)in the middle panel.
1822 TOYAMA ET AL.
trol samples because the zebrafish pi-neal gland is a photoreceptive organand, therefore, similar molecularpathways may be active in eyes andpineal gland.
While there is significant overlapbetween genes highly expressed in thepineal gland at larval and adultstages, we considered the two groupsof data separately for GO analysis.This approach revealed the enrich-ment of developmentally specific GOterms which would be obscured bycombining all the data. As result, wefound that GO terms for neuronal de-velopment and function were specifi-cally enriched at larval stages.
Throughout this microarray analy-sis, we obtained evidence for manygenes that are highly expressed in thepineal gland relative to brain. Atpresent, nearly half of the probe setsare not well annotated, being referredto as “transcribed locus” or “hypothet-ical protein.” Future expansion of theEST database and completion of thezebrafish genome will promote theanalysis of these genes. Nevertheless,we generated a list of many annotatedgenes which have not been reportedpreviously as expressed in the pinealgland. Individual analysis of some ofthese genes, and of genes that may be
annotated in the future, offer the op-portunity to identify new players inpineal development and function.Some of these newly identified pinealgenes may relate to the establishmentof asymmetry in the pineal and sur-rounding epithalamic region (Harriset al., 1996; Concha and Wilson, 2001;Gamse et al., 2003; Snelson et al.,2008). This proposal is supported bythe finding that otx5, which is en-riched in the pineal gland, also exhib-its asymmetric expression in the ze-brafish epithalamus.
The finding of a highly expressedtranscript in the pineal gland providesreason to characterize unannotatedgenes because such highly expressedgenes are likely to be functionally im-portant. In addition to the unc119 ho-mologue (see below), another interest-ing novel gene is the zebrafishhomolog of agrp which corresponds toprobe set Dr. 20586.1.A1 (Supp. TableS1). AgRP is a key hypothalamic reg-ulator of ingestive behavior in mam-mals and zebrafish (Song et al., 2003).An immunocytochemical study in ze-brafish using an antibody against hu-man AgRP revealed unexpected AgRPimmunoreactivity in the pineal gland,leading to the suggestion that this wasdue to the expression of the gene in
the pineal gland termed AgRP2 (For-lano and Cone, 2007). Here, we inde-pendently reached the same conclu-sion based on our microarray analysis.The pineal-enhanced expression of theAgRP2 homolog suggests a possibleconnection between the circadian os-cillator in the zebrafish pineal to in-gestive behavior, an attractive direc-tion for further research.
Previous Microarray Studieson Gene Expression in theVertebrate Pineal Gland
Three groups have previously pub-lished results of microarray analysisof the rat pineal gland (Humphries etal., 2002; Bailey et al., 2008; Fuku-hara and Tosini, 2008). Fukuhara etal. found that approximately 2% of atotal of 8,000 genes on the microarrayshowed rhythmic expression in the pi-neal gland, consistent with the find-ings by Humphries et al. (approxi-mately 3% out of 1,176 genes). Baileyet al. found that approximately 4% ofthe genes (�600 of �13, 000) exhibitgreater than twofold change on anight/day basis. Our data identifiedapproximately 500 genes which areup-regulated at day or night in theadult zebrafish pineal gland (approxi-mately 3% of all genes on the microar-ray). Our finding agrees with previousobservations (Humphries et al., 2002;Bailey et al., 2008; Fukuhara andTosini, 2008) that only a limited num-ber of genes exhibited day/night dif-ferences in their expression level.
All previous analyses were donewith adult rat pineal gland and foundthat more genes were up-regulated atnight as compared to those up-regu-lated during the day (47 vs. 13; Fuku-hara et al., 2008). Likewise, Bailey etal. found approximately twofold moregenes that were elevated at night(Bailey et al., 2008). In contrast, wefound similar numbers of genes up-regulated at day and at night. Fur-thermore, we identified many moregenes showing day/night differencesin expression, which had not been re-ported previously. Only few genes pre-viously reported to display a diurnalrhythm of expression were identifiedin our analysis. One of the genes char-acterized by Humphries et al. (2002)as a nocturnal up-regulated gene, Id-1(inhibitor of DNA binding and differ-
Fig. 8. Number of genes highly expressed at day or night in the pineal gland. Samples wereanalyzed at each developmental stage separately. The following criteria were used for the selectionof probe sets: Average minimum signal �100; P value �0.05. Dark gray: D/N�1.5; black: D/N�2;light gray: N/D�1.5, white: N/D�2.
ZEBRAFISH PINEAL GLAND TRANSCRIPTOME 1823
entiation), did not show significantday/night changes in its expression inthe zebrafish pineal gland. These dif-ferences may be caused by species dif-ferences or by incomplete annotationof the zebrafish microarray.
Possible FunctionalInteraction Between Unc119cand Arl3l2
The unc119 gene is a new class of neu-ral gene that shares conserved se-quences in all metazoans examined.Although this gene is highly con-served from worms to human, its ex-pression pattern shows two extremes.In invertebrates (C. elegans and Dro-sophila), this gene is expressedthroughout the nervous system, andunc119 knockout causes damage toneurons widely distributed in the ner-vous system (Maduro et al., 2000;Knobel et al., 2001). In mammals (hu-man, mouse, rat), this gene was orig-inally identified as HRG4 (human ret-inal gene 4) (Higashide et al., 1996), agene that is expressed specifically inthe photoreceptor synapse. A trunca-tion mutation of HRG4 is associatedwith late-onset cone-rod dystrophy inhumans, and transgenic mice contain-ing the same mutation, develop late-onset retinal degeneration (Koba-yashi, 2000). HRG4 knockout in themouse causes severe damage to theretina (Ishiba et al., 2007). In ze-brafish, one unc119 homolog has beenidentified (unc119) with an expressionpattern similar to that seen in inver-tebrates, i.e., expression throughoutthe central nervous system; its knockdown results in disorganized neuralarchitecture (Manning et al., 2004). Asecond unc119 homolog (unc119b) wasfound in the zebrafish genome but itsexpression pattern was not character-ized (Manning et al., 2004). Here, weshow the existence of a third unc119homolog (unc119c) that exhibits en-hanced expression in the pineal gland.
Although the function of UNC119 isnot clearly understood, the recentfinding that ARL2 (ADP-ribosylationfactor-like protein 2) interacts withHRG4 suggests a possible function forUNC119 in the retina. ARL2 is a gua-nine nucleotide binding protein andmay play a role in microtubule assem-bly (Radcliffe et al., 2000). ARL2 in-teracts with BART (binder-of-ARL2),
enabling it to enter mitochondria andbind ANT-1 (adenine nucleotidetransporter), which is thought to beinvolved in apoptosis (Mori et al.,2006). Therefore, a truncation muta-tion of HRG4 may lead to mitochon-drial ANT-1-mediated retinal degen-eration by apoptosis through ARL2.
Here, we demonstrate the coexpres-sion of unc119c and arl3l2, one of theclosest homologs of arl2, as well as thephysical interaction between Unc119cand Arl3l2. To our knowledge, this isthe first evidence suggesting anUnc119-Arl interaction in the pinealgland. Recently, Veltel et al. reportedthe formation of a ternary complex be-tween Arl3, its GAP RP2 (retinitis pig-mentosa 2), and HRG4 (Veltel et al.,2008). Also, mouse Unc119 interactswith the synaptic ribbon specific pro-tein RIBEYE at photoreceptor ribbonsynapses (Alpadi et al., 2008). Thepreviously reported functions forUNC119 are all related to its expres-sion in the retina. However, Bailey etal. also identified Unc119 as a highlyexpressed gene in the pineal glandand the retina relative to other tissues(Bailey et al., 2008), and our findingsindicate that a distinct tissue-specificUnc119 homolog may play a role inphotoreceptor cell function in the pi-neal gland. The high expression ofUnc119 in photoreceptors might be aconserved feature in vertebrates, pos-sibly based on similar functions.Whether the existence of an addi-tional unc119 homolog is a unique fea-ture of the zebrafish is not known, butit should be remembered that the ze-brafish pineal gland is located super-ficially and contains photoreceptorcells, in contrast to the situation inhigher vertebrates. Zebrafish may,therefore, have gained an additionalunc119 gene that is functional outsideof the retina.
EXPERIMENTALPROCEDURES
Collection of Larvae andAdult Pineal Glands andRNA Preparation
Adults and larvae were kept under a14-hr-light/10-hr-dark cycle. Pinealglands were isolated manually usinga fluorescence dissection microscope,guided by green fluorescent protein
(GFP) fluorescence, from larval (3 days,5 days, and 10 days) and adult (3 monthand 1–2 years) transgenic zebrafish inwhich expression of the GFP gene isdriven by the aanat2 promoter (Gothilfet al., 2002). For comparison, brain tis-sue from which the pineal gland andeyes had been removed was also col-lected (referred to as “brain” from hereonward). The tissues were collected di-rectly in QIAzol (Qiagen). Five to nineadult or 10 to 38 larval pineal glands,and 2 to 3 adult or 2 to 5 larval brainswere pooled to yield one sample. Threeto five samples were collected at middayand midnight at each developmentalstage.
Total RNA was prepared using theRNeasy Lipid Tissue Mini Kit (Qia-gen) and biotin-labeled cDNA wasgenerated using the Ovation Biotinsystem kit (NeuGen)
Microarray Analysis andData Processing
The Affymetrix GeneChip ZebrafishGenome Array was hybridized andprocessed using the standard Af-fymetrix protocol. Altogether, we col-lected 20 types of samples: five timepoints (3 days, 5 days, 10 days, 3months, and 1–2 years), two organs(pineal gland and brain), and twosampling times (day and night). Foreach type of sample, tissue was ob-tained and processed three to fivetimes. After hybridization, microarraychips were scanned, quantitated, andnormalized by GCOS (Affymetrix). Alldata were submitted to the NCBIGEO database as series GSE13371.There are 15,617 probe sets on a ze-brafish Affymetrix microarray chip,including 114 hybridization controls;data for the remaining 15,503 probesets were subjected to further statisti-cal using JMP, the statistical Discov-ery Software (http://www.jmp.com/).Unless otherwise indicated, three tofive replicates for each sample typewere averaged, and probe sets show-ing differences with a P value smallerthan 0.05 were considered further.
Gene Ontology (GO) Analysis
To take advantage of the more com-plete annotation of the human ge-nome, we converted lists of selectedzebrafish genes to those of their hu-
1824 TOYAMA ET AL.
man homologs, using the WEB site of theZebrafish Gene and Microarray Annota-tion Project (ZGMAP) (Children’s Hospi-tal Boston, http://134.174.23.160/zfACA/hash/cumulative_expanded.aspx). GOterm analysis was performed using theDAVID Bioinformatics Resources, NI-AID/NIH, with the default level of P �0.05 to select genes to be included, unlessnoted otherwise.
Pathway Analysis
Ingenuity Pathways Analysis soft-ware (http://www.ingenuity.com/prod-ucts/pathways_analysis.html) (ver.5)was used to identify canonical path-ways most likely to be active in thepineal gland. Selected zebrafish geneswere converted to a list of humangenes as indicated above before thisanalysis.
Immunoprecipitation
HEK-293 cells were transfected withpcDNA3Unc119c-Flag, pcDNA3Arl3l2-Myc, and pcDNA3Nucleolin-Myc. Pro-teins were extracted and precipi-tated with anti-Flag antibodies (Sigma,F3165) coupled to anti-mouse IgG aga-rose beads (Sigma, A6531).
Proteins were detected with anti-Myc (Sigma, C6594) or anti-Flag(Sigma, F3165) antibodies.
ACKNOWLEDGMENTSWe thank Mark Rath for excellenttechnical assistance. This work waspartially supported by a grant fromthe United States-Israel BinationalScience Foundation (BSF), Jerusalem,Israel, and by the Intramural Re-search Program of the NICHD, NIH.
REFERENCES
Aizawa H, Bianco IH, Hamaoka T, Mi-yashita T, Uemura O, Concha ML, Rus-sell C, Wilson SW, Okamoto H. 2005.Laterotopic representation of left-rightinformation onto the dorso-ventral axisof a zebrafish midbrain target nucleus.Curr Biol 15:238–243.
Aizawa H, Goto M, Sato T, Okamoto H.2007. Temporally regulated asymmetricneurogenesis causes left-right differencein the zebrafish habenular structures.Dev Cell 12:87–98.
Alpadi K, Magupalli VG, Kappel S, KoblitzL, Schwarz K, Seigel GM, Sung CH,Schmitz F. 2008. RIBEYE recruitsMunc119, a mammalian ortholog of theCaenorhabditis elegans protein unc119,
to synaptic ribbons of photoreceptor syn-apses. J Biol Chem 283:26461–26467.
Arendt J. 1995. Melatonin and the mam-malian pineal gland. London: Chapmanand Hall.
Bailey MJ, Beremand PD, Hammer R,Bell-Pedersen D, Thomas TL, CassoneVM. 2003. Transcriptional profiling ofthe chick pineal gland, a photoreceptivecircadian oscillator and pacemaker. MolEndocrinol 17:2084–2095.
Bailey MJ, Coon SL, Carter DA,Humphries A, Kim JS, Shi Q, Gaildrat P,Morin F, Ganguly S, Hogenesch JB,Weller JL, Rath MF, Moller M, Baler R,Sugden D, Rangel ZG, Munson PJ, KleinDC. 2008. Night/day changes in pinealexpression of �600 genes: central role ofadrenergic/cAMP signaling. J Biol Chem284:7606–7622.
Begay V, Falcon J, Cahill GM, Klein DC,Coon SL. 1998. Transcripts encoding twomelatonin synthesis enzymes in the te-leost pineal organ: circadian regulationin pike and zebrafish, but not in trout.Endocrinology 139:905–912.
Bernard M, Klein DC, Zatz M. 1997. Chickpineal clock regulates serotoninN-acetyltransferase mRNA rhythm inculture. Proc Natl Acad Sci U S A 94:304–309.
Cau E, Wilson SW. 2003. Ash1a and Neu-rogenin1 function downstream of Float-ing head to regulate epiphysial neuro-genesis. Development 130:2455–2466.
Chitnis AB, Kuwada JY. 1990. Axonogen-esis in the brain of zebrafish embryos.J Neurosci 10:1892–1905.
Concha ML, Wilson SW. 2001. Asymmetryin the epithalamus of vertebrates. J Anat199:63–84.
Concha ML, Burdine RD, Russell C, SchierAF, Wilson SW. 2000. A nodal signalingpathway regulates the laterality of neu-roanatomical asymmetries in the ze-brafish forebrain. Neuron 28:399–409.
Concha ML, Russell C, Regan JC, Tawk M,Sidi S, Gilmour DT, Kapsimali M,Sumoy L, Goldstone K, Amaya E, Kimel-man D, Nicolson T, Grunder S, GompertsM, Clarke JD, Wilson SW. 2003. Localtissue interactions across the dorsal mid-line of the forebrain establish CNS later-ality. Neuron 39:423–438.
tral innervation of the midbrain target.Development 132:4869–4881.
Gamse JT, Thisse C, Thisse B, HalpernME. 2003. The parapineal mediates left-right asymmetry in the zebrafish dien-cephalon. Development 130:1059–1068.
Gothilf Y, Coon SL, Toyama R, Chitnis A,Namboodiri MA, Klein DC. 1999. Ze-brafish serotonin N-acetyltransferase-2:marker for development of pineal photo-receptors and circadian clock function.Endocrinology 140:4895–4903.
Gothilf Y, Toyama R, Coon SL, Du SJ, Da-wid IB, Klein DC. 2002. Pineal-specificexpression of green fluorescent proteinunder the control of the serotonin-N-acetyltransferase gene regulatory re-gions in transgenic zebrafish. Dev Dyn225:241–249.
Halpern ME, Liang JO, Gamse JT. 2003.Leaning to the left: laterality in the ze-brafish forebrain. Trends Neurosci 26:308–313.
Hendricks M, Jesuthasan S. 2007. Asym-metric innervation of the habenula inzebrafish. J Comp Neurol 502:611–619.
Higashide T, Murakami A, McLaren MJ,Inana G. 1996. Cloning of the cDNA for anovel photoreceptor protein. J Biol Chem271:1797–1804.
Humphries A, Klein D, Baler R, Carter DA.2002. cDNA array analysis of pinealgene expression reveals circadian rhyth-micity of the dominant negative helix-loop-helix protein-encoding gene, Id-1.J Neuroendocrinol 14:101–108.
Ishiba Y, Higashide T, Mori N, KobayashiA, Kubota S, McLaren MJ, Satoh H,Wong F, Inana G. 2007. Targeted inacti-vation of synaptic HRG4 (UNC119)causes dysfunction in the distal photore-ceptor and slow retinal degeneration, re-vealing a new function. Exp Eye Res 84:473–485.
Klein DC. 2004. The 2004 Aschoff/Pitten-drigh lecture: theory of the origin of thepineal gland—a tale of conflict and reso-lution. J Biol Rhythms 19:264–279.
Knobel KM, Davis WS, Jorgensen EM,Bastiani MJ. 2001. UNC-119 suppressesaxon branching in C. elegans. Develop-ment 128:4079–4092.
Kobayashi A, Higashide T, Hamasaki D,Kubota S, Sakuma H, An W, Fujimaki T,McLaren MJ, Weleber RG, Inana G.2000. HRG4 (UNC119) mutation foundin cone-rod dystrophy causes retinal de-generation in a transgenic model. InvestOphthalmol Vis Sci 41:3268–3277.
Kobayashi A, Kubota S, Mori N, McLarenMJ, Inana G. 2003. Photoreceptor synap-tic protein HRG4 (UNC119) interactswith ARL2 via a putative conserved do-main. FEBS Lett 534:26–32.
Kuan YS, Gamse JT, Schreiber AM, Halp-ern ME. 2007a. Selective asymmetry in aconserved forebrain to midbrain projec-tion. J Exp Zoolog B Mol Dev Evol 308:669–678.
ZEBRAFISH PINEAL GLAND TRANSCRIPTOME 1825
Kuan YS, Yu HH, Moens CB, Halpern ME.2007b. Neuropilin asymmetry mediatesa left-right difference in habenular con-nectivity. Development 134:857–865.
Liang JO, Etheridge A, Hantsoo L, Rubin-stein AL, Nowak SJ, Izpisua Belmon-teJC, Halpern ME. 2000. Asymmetricnodal signaling in the zebrafish dien-cephalon positions the pineal organ. De-velopment 127:5101–5112.
Liu Y, Shen Y, Rest JS, Raymond PA, ZackDJ. 2001. Isolation and characterizationof a zebrafish homologue of the cone rodhomeobox gene. Invest Ophthalmol VisSci 42:481–487.
Maduro MF, Gordon M, Jacobs R, PilgrimDB. 2000. The UNC-119 family of neuralproteins is functionally conserved be-tween humans, Drosophila and C. el-egans. J Neurogenet 13:191–212.
Manning AG, Crawford BD, WaskiewiczAJ, Pilgrim DB. 2004. unc-119 homologrequired for normal development of thezebrafish nervous system. Genesis40:223–230.
Mano H, Kojima D, Fukada Y. 1999. Exo-rhodopsin: a novel rhodopsin expressedin the zebrafish pineal gland. Brain ResMol Brain Res 73:110–118.
Masai I, Heisenberg CP, Barth KA, Mac-donald R, Adamek S, Wilson SW. 1997.floating head and masterblind regulateneuronal patterning in the roof of theforebrain. Neuron 18:43–57.
Mori N, Ishiba Y, Kubota S, Kobayashi A,Higashide T, McLaren MJ, Inana G.2006. Truncation mutation in HRG4(UNC119) leads to mitochondrial ANT-1-mediated photoreceptor synaptic andretinal degeneration by apoptosis. InvestOphthalmol Vis Sci 47:1281–1292.
Radcliffe PA, Garcia MA, Toda T. 2000.The cofactor-dependent pathways for al-pha- and beta-tubulins in microtubulebiogenesis are functionally different infission yeast. Genetics 156:93–103.
Snelson CD, Santhakumar K, HalpernME, Gamse JT. 2008. Tbx2b is requiredfor the development of the parapinealorgan. Development 135:1693–1702.
Song Y, Golling G, Thacker TL, Cone RD.2003. Agouti-related protein (AGRP) isconserved and regulated by metabolicstate in the zebrafish, Danio rerio. Endo-crine 22:257–265.
Talbot WS, Trevarrow B, Halpern ME, MelbyAE, Farr G, Postlethwait JH, Jowett T,Kimmel CB, Kimelman D. 1995. A ho-meobox gene essential for zebrafish noto-chord development. Nature 378:150–157.
Toyama R, Dawid IB. 1997. lim6, a novelLIM homeobox gene in the zebrafish:comparison of its expression patternwith lim1. Dev Dyn 209:406–417.
Veltel S, Kravchenko A, Ismail S, Witting-hofer A. 2008. Specificity of Arl2/Arl3signaling is mediated by a ternary Arl3-effector-GAP complex. FEBS Lett 582:2501–2507.
Wilson SW, Easter SS, Jr. 1991. Stereotypedpathway selection by growth cones of earlyepiphysial neurons in the embryonic ze-brafish. Development 112:723–746.
