J. Mol. BioZ.
(1991)
222,
553-565
Characterization of the Memory Gene Dunce of Drosophila melanogaster Yuhong Qiu’, Chun-Nan Chenlt, Tom Malone2, Liz Richter’ Steven K. Beckendorf’ and Ronald L. Davis’ f ‘Department of Cell Biology Baylor College of Medicine Houston, TX 77030, U.S.A. 2Department of Molecular Biology University of California Berkeley, CA 94720, U.S.A. (Received
6 May 1991; accepted 1 August
1991)
The dunce (dnc) gene of Drosophila melanogaster encodes CAMP phosphodiesterase (PDEase) and is required for learning/memory and female fertility. The gene is structurally complex, demonstrated in part by Northern blotting experiments which detected multiple RNAs ranging in size from 4.2 to 9.6 kb (1 kb = lo3 bases or base-pairs). To characterize these RNAs and to understand their sequenceheterogeneity, we isolated and analyzed 29 new and independent cDNA clones representing the dnc RNAs. Restriction mapping, hybridization analysis and sequence determination of these cDNA clones and the corresponding genomic exons resolved these into six different classes. Exons defined by the cDNA clones are distributed over more than 148 kb of genomic DNA, with some exons being used alternatively among the RNAs. The RNAs are transcribed from at least three initiation sites; two of these were mapped by parallel S,-nuclease and primer extension experiments. In addition, some of the heterogeneity is generated by using varying lengths of a 3’-untranslated trailer sequence. Altogether, the results indicate that the size and sequence heterogeneity of dnc transcripts results from transcription initiation at multiple sites, alternative splicing, and processeswhich generate different 3’ ends. The existence of multiple protein products is suggested by the alternative use of exons which code for portions of the open reading frame. The protein variation potentially includes N-terminal differences coded for by transcript-specific 5’ exons and internal differences arising from the optional inclusion of a 39 base-pair exon and from the alternative use of two 3’ splice sites separated by six base-pairs. Expression of a cDNA clone in yeast containing a large portion of the open reading frame produced CAMP PDEase activity identical in properties to the Drosophila enzyme affected by the dnc mutation. The results suggest that the remarkable structural complexity of dnc may reflect an intricate control of the spatial and/or temporal expression of various isoforms of CAMP PDEase. Keywords: dunce; gene structure; intronic gene; learning; memory
normal flies after training in associative or nonassociative learning situations and they exhibit alterations in courtship behaviors that are dependent upon prior experience (for a review, seeDudai, 1988). In addition, dnc mutant females are sterile, owing to a somatic cell requirement of the product for egg deposition and a germ-line requirement for normal development of the embryo (Bellen et al., 1987; Bellen & Kiger, 1988). Thus, the normal function of the gene is required for normal reproduction as well as normal behavior.
1. Introduction The dunce (dnc) gene is one of several different Drosophila genes known to code for molecules involved in conditioned behavior. Flies carrying mutant forms of the gene forget more rapidly than t Present address:Department of Biochemistry and Biophysics. tJniversity of California, San Francisco, CA 94143, U.S.A. $ Author to whom addressed. 0013-2X3(i/91/230553~13
all correspondence
$03.00/O
should
be
553 0 1991 Academic
Prrss Limited
554
Y. Qiu et al.
The gene’s product is the enzyme, cAMP phosphodiesterase (PDEaset), and mutant flies exhibit a reduction in the activit’y of this enzyme leading t’o an elevated CAMP content (Byers et al., 1981; Chen et al., 1986; Davis & Kiger, 1981; Kauvar, 1982). Thus, cAMP metabolism appears to be intimately involved in behavioral plasticity, a conclusion also reached from the elegant work with the sea snail, ApZysia eaEifornica (Byrne, 1987; Goelet et al., 1986). Although the steps following CAMP involvement in the biochemical pathway(s) required for normal learning/memory in flies are unknown, it is assumed that CAMP-dependent protein phosphorylation, cA,MP-dependent gene expression, and/or CAMPgated ion channels are likely to be involved. Despite the clear identification of dnc as the struct)ural gene for cAMP PDEase. previous molecular studies hinted at, an unexpected complexity of the gene. An extensive series of RNA blotting experiments using probes from t,he dnc genomic region identified an array of at least six large (4.2 to 9.6 kb) and overlapping poly(A)+ RNAs as dnc transcripts (Davis & Davidson, 1986). These RNAs are extremely rare, existing in the adult fly at an abundance level of no more than five parts per million of the polyadenylated RNA fraction (Chen et al., 1986; Davis & Davidson, 1986). Several cDNA clones representing the RNAs were selected from oligo(dT)-primed and primer extension cDNA libraries and sequenced, which revealed the introni exon organization of a portion of one dnc transcription unit, and a portion of the conceptual amino acid sequence of the protein product (Chen et al., 1986, 1987). In addition, they demonstrated the existence Sgs-4, within a of several other genes, including 79 kb intron of dnc (Chen et al., 1987). The studies presented here were designed t’o elucidate the basis for the RNA het’erogeneity and to rontribute to our general understanding of the gene’s structure and promoter regions. Tn particular, we wished to know whether the marked RKA heterogeneity reflected alternative processing to produce PDEase isoforms, or whether the heterogeneity reflected alternative untranslated sequences which might confer differences in stability or translational efficiency. The former possibility seemed particularly attractive since rats have at least four genes which encode different CAMP-specific PDEases (Colicelli et al., 1989; Davis et al., 1989: Swinnen et al., 1989), and since dnc is the only Drosophila locus that codes for this class of enzymes (Davis & Kauvar, 1984). The possible biological need for multiple enzymes might be accomplished by alternative splicing of protein-coding exons. Through sequence analysis of additional cDNA clones and the corresponding genomic exons, we demonstrate that the transcript heterogeneity arises by transcription from at least three start sites as
t Abbreviations used: PDEase, phosphodiesterase: kb, lo3 bases or base-pairs; bp. base-pair(s); ORF. open reading frame.
well as alternative splicing of exons contribut)ing t’o the open reading frame. In addition. differential processing of 3’ ends produces transcripts that vary in length by as much as 3 kb.
2. Materials and Methods (a)
Isolation
oj
RX.4
Qnd
tlS.4
blotting
Total RKA was prepared from Canton-S adult flips according to either the met,hod described b,v Labarca & Paigen (1977) or procedures used previously (Davis & Davidson. 1986). Polyadenylatrd RNA was select,ecl b? I or 2 passages over an oligo(dT)-c~ellulosr (Collaborative Research, type 3) column as described by Maniat,is rf nl. (1982). RNA blots were prepared and probed as described (Davis & Davidson. 1986). except t,hat single-stranded probes (see below) were used. III some caases. the stringency of the hybridization and washing was rrducrd because of the high A + T ront,ent of a particular axon (exon 10) or the short homology predicated between the probe and RSA (exon 4). Size markers for RiKA blots were RKA ladders purchased from Mhesda Researcbh Laboratories.
One primer extension cDXA library was c,onstruc+trd from 5 pg of poly(A)+ RNA by priming the first strand synthesis using M-MLV reverse transcriptase (Bethesda with a unique 20-me1 Research Laboratories) (.5’-L4GTCt~L4GCTCCTCGATTGTG-3’). whose srquencsr is complementary t*o a portion of exon 6 (Fig. 2). The se~~ntl strand was made according to the method described b> Gubler & Hoffman (1983) without, the use of Escherichin coli DEA ligase. The cDKA was then methylated. bluntended. and EcoRT linkers (New England Biolabs) were added followed by EcoRT digestion. To frac%ionate the resulting cDNA and to remove excess linkers. the restrict,ion reaction was chromatographed on a Bio-gel A-5Om (BioRad) csolumn according to Huynh et ~2. (19%). Fractions containing cDNA greater t,han 500 bp were pooled and ligated to LgtlO arms. The librar? was packaged in vitro and screened without amplification. Two additional cDNA libraries in igt 11 were scrrened using probes representing 5’ portions of different cI)?r’As. B total of 1.7 x lo6 phage were screened from a Canton-S head cD?JS library obtained from M. Robash (Brandris). Three independent clones were recovered. each more than once. We also screened (2 x lo6 phage) a (‘antorl-8 head cDNA library made by P. Salvaterra (Beckman Research Institute). Sine independent clones were recovered. most of them more than once. The inserts of’t,he (sDK.4 clones were subcloned int)o plasmids for sequencing.
Genomic and cDNA sequencing was performed with thr dideoxy chain termination method using either the Klenow fragment (Pharmacia) or phage T7 DXL4 polymerase (U.S.B.), following the manufacturer’s conditions or our own modified conditions. Most of the sequence of genomic regions from co-ordinates + 21.2 t,o +489 was obtained from random phage Ml3 subclones of apprtrpriate fragments generated by the sonication method described by Deininger (1983). Tn some cases. fragments to be sequenced were cleaved with selected restriction enzymes and the subfragments subcloned into Ml 3 vectors in a shotgun fashion. Alternatively. progressive
Dunce deletion subclones were generated by a simplified ExoIII deletion method described by Henikoff (1984). A total of 70,500 residues were sequenced from co-ordinates +21.2 to +469, such that each of the 23,607 residues was sequenced 3.0 times, on average. We connected the sequences of these fragments by sequencing through the junction restriction sites, except for the Hind111 site at co-ordinate +3%9 and the EcoRI site at +41.3. Sequences for cDNAs were obtained either by the simplified ExoII deletion method or by sequencing subcloned restriction fragments of the original cDNA clones. (d) S,
555
Structure
poly(A)+ RNA for primer extension experiments, or with an equal amount of yeast transfer RNA as a control. The products of S, nuclease or primer extension mapping were sized using DNA sequencing ladders or labeled pBR322 HpaII restriction fragments. (e) Expression
Expression of dnc cDNA clones in yeast and CAMP PDEase assays were as described (Henkel-Tigges & Davis. 1990).
experiments, primer extension analysis, and single-stranded probes
n&ease
3. Results and Discussion
S, nuclease mapping of the 5’ ends of dnc transcripts was performed according to the procedure described by Berk & Sharp (1977). Primer-extension analysis was carried out as described by Laughon et al. (1986). 32Plabeled probes used for Si nuclease mapping, primer extension and RNA blotting experiments were generat’ed by priming DNA synthesis across an insert in Ml3 (Burke, 1984), truncating the product at an appropriate restriction site and isolating the single-stranded probes on denaturing polyacrylamide or agarose gels. Radiolabeled DNA was hybridized with 10 pg of poly(A)+ RNA from Canton-S adult flies for 8, nuclease analysis, 30 pg of
z-40
experiments
(a) New cDNA
clones de&e dnc transcripts
six classes of
Three cDh’A clones previously characterized and named ADCl, ADC6 (Chen et al., 1986) and ~863 (Chen et al., 1987) overlap to define a major portion of one type of dnc RNA designated as class I (Fig. 1). More specifically, these cDN.4 clones contained a portion of exon 1, and exons 2 through 13. In addition, they defined the 79 kb intron between exons 2 and 3 that houses several other genes, including Sgs-4, Pig-l and Indnc (Fig. 1).
kb
MI I)
2.7 2.8 lndnc
I 4I I II 11111 B
3
5 67
“9’9,12
ATG
stop
ATG
stop
13
IA
(7)
f stop
ATG
(15) ATG
III ATG ELA
ATG
stop I II
f
(1)
stop I II
f
(2)
3
(4)
stop I
Figure 1. Schematic representation of the dnc genomic region and dnc RNAs. An arbitrary co-ordinate system, extending in this Figure from -50 to +47, has been described (Davis & Davidson, 1984), but omits 7.3 kb of sequence contained in a transposable element inserted in the Canton-S strain between co-ordinates 2 and 5. The position of exon 69 at approximately -90 is estimated, since the region between exons 69 and 1 has not been completely cloned. Dunce exons are represented as numbered boxes and intronic genes by arrows. These are not drawn to scale. The intronic genes shown include Sgs-4, several Pig genes and the genomic region coding for a 2.0 kb RNA (Indnc). Splicing patterns of 6 different classes of RNAs are illustrated (IA to IVB). Filled portions represent sequences contained in cT)NA clones recovered. Open portions represent sequences predicted to be present in RNAs from RNA blotting, S, nuclease, and primer-extension experiments. The number in parentheses to the right of each class is the number of independent clones recovered and assigned to each class, with 7 clones representing both IA and IB since these could not be distinguished. The putative open reading frame in each class is shown by ATG and STOP. tss, transcription start site.
556
Y. Qiu et al.
In order to understand the sequence heterogeneity of dnc RNAs detected by RNA blotting experiments, we carried out detailed molecular analyses of many additional cDNA clones. A primer extension cDNA library constructed using a synthetic oligonucleotide in exon 6 (Chen et al., 1987) provided 17 additional clones beyond the initial isolate ~863 upon screening with probes representing exons 3 and 5. This primer was chosen based on the observation that sequence information of exon 6 exists in most, if not all, of the dnc RNAs (Fig. 6). Many of these clones were sequenced and others mapped extensively with restriction endonucleases to determine overlaps. S1 nuclease and primer extension experiments demonstrated that none of these cDNA clones extended to transcription start sites (not shown). Therefore, oligonucleotides representing the 5’ portion of several of the cDNA clones as well as ORF probes were used to screen two cDNA libraries made from head poly(A)+ RNA. Twelve additional and independent clones were obtained. These clones were also subjected to comparative restriction mapping, hybridization analysis to genomic clones, and sequence analysis along with the hybridizing genomic fragments to complete a detailed picture of the intron/exon structure and RNA splicing patterns. From these results, and the results from 6, nuclease and primer extension experiments presented below, several conclusions regarding the structure of dnc, its RNA transcripts and its protein products can be drawn. First, the cDNA clones fall into six structurally distinct and overlapping classes (Fig. 1). The clones in each class were identical in restriction map and sequence? except that some extended further to the right and/or left than others. Second. all of the t’ranscriptional units are large and one is exceptionally large, extending over more than 148 kb. The exons contained within the cDNA clones were identified by hybridization with genomic clones from a chromosomal walk (Davis & Davidson, 1984), representing sequencesfrom - 50 to + 47. Exon 69 resides outside the region covered by the chromosomal walk and was localized by screening a genomic library with the 5’ portion of class I cDNAs. Examination of the restriction map of genomic clones containing 0.9 suggested that these sequenceswere just to the right of the Notch locus. This was confirmed by subsequent hybridization experiments, which showed that exon 0.9 maps approximately 10 kb from the 3’ end of Notch (Kidd et al., 1986; Wharton et al., 1985). Since the Notch locus residesat band 3C7, dunce must extend from approximately chromomere 3C8 through 3D4, a region of nine salivary gland chromosome bands. Third, the intron defined by exons 99 and 1 is very large, although probably not as large as that defined by exons 2 and 3. A precise estimate of the size of the intron is not possible, since genomic clones covering the region have not yet been isolated. However, partial walks from the Notch
locus (S. Kidd $ M. Young, personal communication) and from exon 1 (Y. Qiu 8r.R. Davis, unpublished results), indicate that the intron must exceed 41 kb. In addition, this intron is like that between exons 2 and 3 in that it contains several genes unrelated to dnc. At least three genes, one a Sgs-like gene (Furia et al., 1990) and two others similar to Pig-l in spatial and temporal expression pattern, and named Pig-2 and Pig-3, reside just to the left of exon 1 (S. Beckendorf, unpublished results). Fourth, the RNA sequence heterogeneity occurs within 5’ and g-untranslated regions as well as within the protein coding regions of the transcripts (see below). Five of the RNA classes contain different exons coding for RNA leader sequences. These sequence differences open the possibility of transcript-specific stability or translatability determinants. The sequence of all exons defined by the cDNA clones, along with some sequence of the flanking introns, is presented in Figure 2. Several features of the sequence are to be noted. First, the splice junctions defined by the clones conform to consensus splice sequences(Mount, 1982). Second, exons 1. 2,3 and 2.8 are punctuated with short opa repeat,s (CAA/G: Wharton et al., 1985). These and otfher similar triplet repeats are common within Drosophila transcriptional units but t,heir signiticance remains unclear. Third, a,ddit,ional ORFs that show a codon bias expected for Drosophila protrincoding sequencesexist within dnc RNA molecules upstream from those that code for cAMP PDEase (see below). One of these is specific to class I1 transcripts, being contained within exon 2.3, and could encode a protein molecule of 77 residues (Fig. 2). A search of the Protein Sequence Databa,sr of the National Biomedical Research Foundation using the conceptual translation sequence of t,his ORF revealed no significant similarity with known proteins. It has been postulated that additional ORFs such as these might code for proteins serving regulatory roles (Kozak. 1987). (1)) The dnc proteins
The CAMP PDEase ORF was reported to extend from exon 8 to exon 13 (Chen et al.. 1986, 1987), predicting a protein molecule of 40,000 J$. However, several sequencing errors were discovered from the analysis of the new cDNA clones, one of which when corrected shifted the prior reading frame to extend it into sequences previously thought to be a non-coding leader. The first ATG defining the ORF for each of t,he various transcript classes is located as shown in Figure 1. If each ATG were used in witlo, four classes of protein product with distinct N-terminal sequenceswould result. In addition to the potential N-terminal heterogeneity, exon 4 is used optionally for class IV transcripts, predicting the presence/ absence of a cassette of 13 additional amino acid residues in some of the protein isoforms. And. one case of alternative 3’ splice site choice was detected
Dunce Structure among the cDNA clones which also predicts protein product variation. For the 25 clones which covered the exon 5-6 splice, the 3’ splice site bordering exon 6 in 11 of the clones resides six bases to the left of that in the other 14. Some protein products should therefore carry two additional amino acid residues. This alternative 3’ splice choice was found in cDNA clones representing classes I and II, but not for classes III and IV. In all, a total of seven possible protein products are predicted from the cDNA clones. The calculated mass for class I, II, III, IVA and IVB products are 79,000, 85,000, 71,000, 77,000 and 76,000 M,, respectively. The conceptual translation sequences of the various ORFs are shown in Figure 3. It is note-
557
worthy that all of the predicted proteins contain the PDEase “conserved domain” of approximately 260 amino acid residues found in many different types of PDEase. This domain has been assigned a catalytic function, since proteolytic fragments of other types of PDEase that contain this core exhibit full catalytic activity (Stroop et al., 1989; Carbonneau, 1990). (c) Expression
of a
dnc cDNA
clone
in
yeast
To verify that the cDNA sequences are capable of coding for authentic CAMP PDEase activity we expressed two different cDNA clones in yeast. One clone contained 43 residues of exon 5, exon 6 to 12,
Fig. 2.
Y. Qiu et al.
Figure 2. Sequence of the dnc exons and portions of flanking introns. Exons are boxed, listed in upper ease. and numbered as for Fig. 1. Only a small portion of exon 13 is shown (see Fig. 7). The sequence complementary to the primer used in the primer extension library construction is boxed within exon 6. The opa repeats are underlined. The I’stT site in rxon 27 used for primer extension is double underlined. The 6 residues missing in some cDNA clones due to an alternative 3’ splice site at the left border of exon 6 are marked with dots. An unidentified ORF begins wit,h residue 75 ot exon 2.3 and ends at residue 308 of this exon. The CAMP PDEase open reading frames are shaded. The first ATG resides in exon 2, 2.3, 3 and 2.8 for class I, IT. III and IV, respectively (Fig. I). The transcription initiation site within rxon 044 is marked by an arrow
and exon 13 to 275 residues past the stop codon (Fig. 2). The clone was inserted into the yeast expression vector, pADNS, which uses the ADH promoter to drive expression (Colicelli et al., 1989), and the construct was transformed into the yeast, strain lODAB (Colicelli et al., 1989). This strain contains insertional disruptions in both of its PDEase genes so that no endogenous PDEase activity can be detected. Assay of the CAMP PDEase activity in lODAB cells containing the cDNA expression construct revealed the presence of a CAMP-specific PDEase activity with a K, value of was not inhibited by 4.9( kO.3) PM. The activity 1 miv-cGMP nor was any cGMP PDEase activity detected. Furthermore. the expressed CAMP PDEase activity was not inhibited by rolipram
(Ki > 100 PM), a drug with antidepressant properties known to inhibit the rat homolog of dnc hut not the fly CAMP PDEase (Henkel-Tigges & Davis. 1990). These characteristics are like t)he actkit? found in flies and affected by dnc mutations (Davis & Kauvar. 1984), which shows the exons 5 to 13 contain all sequence information necessary for PDEase activity. A second cDNA clone, extending more 5’ through exon 3 also expressed a (aAMPspecific PDEase activity in this assay. (cl) Transcription
initiation sites for class i und II RNAs
In order to determine the 5’ boundary 5’ exon defined by each class of cDNA.
of the most S, nuclease
Dunce Structure ClawI
f:
Cl&&s
IX:
559
WQLSMSALGLQQSSSXLXSKSAlfTIEBiKSSSAGt4RTQLTL
(40)
S6GFLAPPGIBfIFILSIIHAPPGLSDblLXRAQGRSPLSF
(80)
RISFPGSDSDLFG
Class
xvr
t4vcsFccccYNPxtxsPss
(18)
All:
FDVE~GQGARSPLEGGSPSAGLVLQl!ZLPQitBESFLYRSDS
(40)
DFE~SPASWSRlSSIASE
SEGEDLIVT
(80)
PFAQILASLRSVRl?ELLS
BQSSSASRS
(=oi
GIPPGAPLSQGEEAYTRLATDTI~ELDWCLDQLETIQTSR
(160)
SVSD?4ASLKPXRMLXIELSHFSESSRSG~QISEYICSTFL
(200)
DKQQEFDLPSLRVED~PELVAANAAAGQQSAGQYARSRSP
(240)
RGPPMSQISGVXRPLSXTNSFTGERLPTFGVETPRENELG
1280)
PPPSGVDENPQEl3ktfBFQVTLEESDQBNLhELBEGDgSGE
(640)
ETTTTGTTQTTAhSALBAGGGGGGGGGB8AFRTGGCQNQPQ
(6=‘t
HGGM* Figure 3. Predicted amino acid sequences of dnc PDEases. The translation of the sequence from the 5’ end of exon 3 to the stop codon in exon 13 is labeled All. The K-terminal sequences unique to the classes are also listed. The first Met residue downstream from the in-frame stop codon in class III is underlined and the 13 amino acid residue cassette encoded by the optionally used exon 4 is double underlined. The shaded region is that homologous to many PDEases and known as the “conserved domain”. The two amino acid residues eliminated in some PDEase forms due to 3’ splice site choice are boxed at residues 100 and 109 in the common region.
protection experiments were performed. For class I transcripts, a 2.2 kb single-stranded and uniformly radiolabeled genomic DNA fragment, extending from a synthetic primer (close to the BstYl site) in exon 0.9 to an upstream Hind111 site (Fig. 4), was hybridized to adult poly(A)+ RNA. The resulting duplexes were treated with different amounts of S, nuclease and the protected fragments sized by denaturing gel electrophoresis (Fig. 4). One protected band measured 130( f2) residues from the SpeI site in exon @9; a second measured 138( +2). A second probe with a different right endpoint. identified the
same left endpoints (not shown). To determine whether these sites are transcription initiation sites, primer extension experiments were carried out. A 184 base single-stranded and uniformly labeled primer with its 3’ end at the SpeI site was hybridized to adult poly(A)+ RNA and extended with reverse transcriptase. One product extended to the same site marked by the smaller S, protected product at 130() 1) residues from the SpeI site. This suggests that this position is a transcription start site. A second product was about 150 residues longer than the first. Another primer with its 3’ end
Y. Qiu et al.
560
t
S, nucleaee (a
Primer
1
extention (5)
I H I
R 6
I kb RS
I BSa
-
Figure 4. S, nuclease and primer extension analysis for class I cDNAs. (a) Nuclease S, and (b) primer-extension products after fractionation on 5 ye polyacrylamide/urea gels. P, intact probe for (a); it migratesat the top of the gel and is not included in the portion of the gel shown. A+, (a) probe or (b) primer hybridized with poly(A)+ RKA prior to digestion or extension. t, control lane with an equal mass of yeast tRNA substituted for poly(A)+ REA. The sequencing ladders show the actual sequence of exon 99 primed with the same primer used to prepare the probe or the primer extension product. The small arrow in (a) identifies the second band which is 8 residues longer than the first one. The arrow in (b) indicates the second band which is about 150 residues from the first one. A schematic drawing of the genomic region of exon 99 is shown at the bottom. The filled box, exon 99; thick arrow, the probe for (a); thin arrow, the primer for (b). H, WindIII; B, B&YI; R, EcoRI; So, SalI; S, &eI.
at the BstY 1 site also generated these two products (not shown). These data suggest that a transcription initiation
site resides
eight
bases more
3’ than
the
left end of exon 0.9, and that the left end is spliced to at least one upstream exon of 150 bases. The sequencein this region is consistent with this interpretation. A consensus3’ splice site is found at the left border of exon 0.9. The sequence at the indicated transcription initiation site (TCAGTT) conforms nicely with a sequencefound at the beginning of many Drosophila RNAs (TCA(G/T)T(C/T); Haynes et aZ., 1990). The RNAs using this transcrip-
tion initiation site apparently exist at an abundance level similar to those containing the upstream exon(s), since the two S,-protected bands are of similar intensity (Fig. 4). Similar experiments suggest that, exon 2.7 is a transcription start site. A single-stranded and uniformly labeled genomic fragment, bridging the terminal nucleotide of the longest class III cDNA clone was used for S, nuclease analysis (Fig. 5). A 389( & 5) bp fragment was protected. This maps a splice site or a transcription start site on the sequence as shown in Figure 2, 17( +5) residues from a PstT site (CTGCAG) in exon 2.7 (Fig. 2). Two other S, nuclease experiments (not shown) using probes with different right endpoints also mark this site. A single-stranded and uniformly labeled primer with its 3’ end at the PstI site in exon 2.7 (Fig. 5) was hybridized to adult poly(A)+ RNA and extended with reverse transcriptase. This produced an extension product of 17(f2) bp (Fig. 5). A second primer with a different 3’ end was also extended to this point (not shown). Thus. t,hese experiments identify a transcription start site at approximately 17 residues to the left of the PstT site. The sequence at the transcription start site, CCAGTG, shows a limited relationship, if any, to the transcription start site sequences of other Drosophila genes. The sequence just upstream from exon 0.9 is unusually rich in A and T residues, with an A + T content of 80% in the first 100 bp. However. no obvious promoter sequence homologies are present in the region upstream from this exon or exon 2.7. Similar experiments were performed for exons 2.3 and 2%. The S, nuclease experiments for these (Chen, 1987) identified the 5’ ends of these exons at’ accept,ors sequences matching consensus splice (Fig. 2) with primer extension products extending well beyond these sites. These results indicate that exons 2.3 and 2.8 splice to unidentified upstream exons although exon 2.8 may splice to 2.7 (see below). Screens of our existing cDNA libraries and construction of new primer extension cDNA libraries have failed to identify further upstream sequences. (e)
RNA
blotting exon-speci$c
experimen,ts probes
with
We performed RKA blotting analysis using probes specific for each of the identified exons to relate the RNA speciesidentified on the blots to the different classesof cDNA clones. Probes specific for exons 1 and 2 detected RNA molecules of 9.5 kb and at least one speciesof 7.2 kb (Fig. 6: Chen et ab., 1987). These are designated as class I dnc transcripts.
The
exon
0.9 probe
consistently
produces
RNA blots of low quality detecting a diffuse band of 7.2 kb, and a less-convincing one of 95 kb. An exon 2.3 probe detected 9.5 and 7.0 kb RNAs (class IT the 9.5 kb transcript. transcripts). However, detected by the exon 2.3 probe migrates slightly faster than the one detected by the exon 1 and 2
Dunce Structure
P
A+
t
561
M
3QQw “5
Primer
extension (b)
sa/ I I
PSll I
Figure 5. S, nuclease and primer extension analysis for class III and IV cDNAs. (a) S, nuclease and (b) primer extension products of exon 2.7 fractionated on 6% polyacrylamide/urea gels. P, intact probe for (a); A+, (a) probe or (b) primer hybridized with poly(A)+ RNA prior to digestion or extension. t, control lane with an equal mass of yeast tRNA substituted for poly(A)* RNA. M, size markers. A schematic drawing of the genomic region of exon 2.7 is shown at the bottom. The filled box, exon 2.7; thick arrow the probe for (a); thin arrow the primer for (b).
562
Y. Qiu et al.
0.9
I
2
2.3
2.7
2.8
3
4
5
6
7
8
9
IO
II
12
13
9.6
;:4” ,;:g m6.7 5.0 4.2
Figure 6. RNA blotting experiments using exon-specific probes. Single-stranded probes were prepared from cU,UL\ OI genomic clones and used to probe adult poly(A)+ RNA blot strips. Axons represented by the probes arc indicated at the top of the blot strips. The probe representing exon 11 overlapped exon 10 by 11 bases. but this is insufhcient homology to hybridize to exon 10 sequences. Some bands are extremely faint, due to the low abundance of the R?VTAs and/or the amount, of homology between RNA and probe. These are marked with a, filled circle.
probes. Therefore, there appear to be at least two different transcripts of about 9.5 kb. Probes specific for exons 2,7 and 2.8 both hybridize to 9.6, 7,4, 7.0. 6.7, 5.0 and 42 kb RNAs. The similarity in the hybridization patterns obtained with these probes suggests that exon 2.8 may be spliced to exon 2.7 in the majority of the transcripts (Fig. 1). Since the 5’ end of exon 2.7 appears to be a transcription start site, the 7.0 kb transcript identified with the exon 2.7 and 2.8 probes must be distinct from the 7.0 kb transcript visualized with the exon 2.3 probe. Therefore, there exist at least two distinct 7.0 kb transcripts. Because of its small size (39 bp), a probe for exon 4 barely detects any transcripts. However, faint bands are observed in the 7.0 t,o 7.2 and 9.5 to 9.6 kb size range, consistent with the contribution of this exon to class I and class II RNAs. Probes specific for exons 3, 5 to 12 and the 5’ portion of exon 13 show similar or identical patterns of hybridization to RNAs of 95, 7.2, 7.0, 5.0 and 42 kb. In addition, many if not all of these hybridize to RNAs of 9.6, 7.4 and 6.7 kb, although the signals in most cases are either partially obscured by hybridization to other RSAs of similar sizes or are very faint. However, these three R’NA transcripts are clearly detected with other exon probes, such as 2.8 and those representing portions of exon 13 (Fig. 7). These data are consistent with the idea that all classes of dnc RNAs utilize exons 3, 5 to 12 and the 5’ portion of exon 13; and that no major alternative splicing occurs between these exons. In summary, the use of exon-specific probes identifies a minimum of ten dnc R&As: 9.6, 7.4, 7.2, 6.7.
56 and 4.2 kb RNAs, and two each of 9.5 and 7.0 kb. Several other RNAs are also detected by t’hc exon-specific probes as shown in Figure 6. For example, probes specific for exons I. 2. 2.3, 2.X and 3 all detect Rh’As smaller than those noted abovcb. Some of these could be due to artifactual hybridization, since they are detected by RNAlike as well as RNA complementary probes (Chen c)t al.. 1987). However, some of the more distinct ones (i.e. thr (b6 kb RNA with exon 2 and 3.3 kb R?iA with exon 2.8) could be processing products, the products of regional transcription or t*he transcripts of other genes related in sequence to portions of dn,c. (f) The dnc tramcripts
possess dijfuen,t
J’ ends
Sane of the cI)NA clones identified to date contains a poly(A)+ tail. However, previous R?\;A blot analysis (Davis &’ Davidson, 1984) suggested that the dnc transcripts end at different sites within the +41*3 to +469 genomic region (Fig. 7). To define more precisely where the dnc transcripts end. we performed additional RNA blotting experiments using single-stranded probes representing subfragments of this region. A probe representing the entire +41.3 to +43.5 region hybridized to all of the dnr RNAs (Fig. 7), but probes representing smaller portions of the 43 to 47 region hybridize t)o successively fewer of the dnc RNAs proceeding from left to right. These results establish that the larger dnc RNAs tend to have longer 3’-untranslated tails. more than with the 9.519.6 kb RNAs containing 48 kb of 3’-untranslated sequence (Fig. 7).
Dunce
41 1
42 1
43 I
R
Structure
(1991)
222,
553-565
Characterization of the Memory Gene Dunce of Drosophila melanogaster Yuhong Qiu’, Chun-Nan Chenlt, Tom Malone2, Liz Richter’ Steven K. Beckendorf’ and Ronald L. Davis’ f ‘Department of Cell Biology Baylor College of Medicine Houston, TX 77030, U.S.A. 2Department of Molecular Biology University of California Berkeley, CA 94720, U.S.A. (Received
6 May 1991; accepted 1 August
1991)
The dunce (dnc) gene of Drosophila melanogaster encodes CAMP phosphodiesterase (PDEase) and is required for learning/memory and female fertility. The gene is structurally complex, demonstrated in part by Northern blotting experiments which detected multiple RNAs ranging in size from 4.2 to 9.6 kb (1 kb = lo3 bases or base-pairs). To characterize these RNAs and to understand their sequenceheterogeneity, we isolated and analyzed 29 new and independent cDNA clones representing the dnc RNAs. Restriction mapping, hybridization analysis and sequence determination of these cDNA clones and the corresponding genomic exons resolved these into six different classes. Exons defined by the cDNA clones are distributed over more than 148 kb of genomic DNA, with some exons being used alternatively among the RNAs. The RNAs are transcribed from at least three initiation sites; two of these were mapped by parallel S,-nuclease and primer extension experiments. In addition, some of the heterogeneity is generated by using varying lengths of a 3’-untranslated trailer sequence. Altogether, the results indicate that the size and sequence heterogeneity of dnc transcripts results from transcription initiation at multiple sites, alternative splicing, and processeswhich generate different 3’ ends. The existence of multiple protein products is suggested by the alternative use of exons which code for portions of the open reading frame. The protein variation potentially includes N-terminal differences coded for by transcript-specific 5’ exons and internal differences arising from the optional inclusion of a 39 base-pair exon and from the alternative use of two 3’ splice sites separated by six base-pairs. Expression of a cDNA clone in yeast containing a large portion of the open reading frame produced CAMP PDEase activity identical in properties to the Drosophila enzyme affected by the dnc mutation. The results suggest that the remarkable structural complexity of dnc may reflect an intricate control of the spatial and/or temporal expression of various isoforms of CAMP PDEase. Keywords: dunce; gene structure; intronic gene; learning; memory
normal flies after training in associative or nonassociative learning situations and they exhibit alterations in courtship behaviors that are dependent upon prior experience (for a review, seeDudai, 1988). In addition, dnc mutant females are sterile, owing to a somatic cell requirement of the product for egg deposition and a germ-line requirement for normal development of the embryo (Bellen et al., 1987; Bellen & Kiger, 1988). Thus, the normal function of the gene is required for normal reproduction as well as normal behavior.
1. Introduction The dunce (dnc) gene is one of several different Drosophila genes known to code for molecules involved in conditioned behavior. Flies carrying mutant forms of the gene forget more rapidly than t Present address:Department of Biochemistry and Biophysics. tJniversity of California, San Francisco, CA 94143, U.S.A. $ Author to whom addressed. 0013-2X3(i/91/230553~13
all correspondence
$03.00/O
should
be
553 0 1991 Academic
Prrss Limited
554
Y. Qiu et al.
The gene’s product is the enzyme, cAMP phosphodiesterase (PDEaset), and mutant flies exhibit a reduction in the activit’y of this enzyme leading t’o an elevated CAMP content (Byers et al., 1981; Chen et al., 1986; Davis & Kiger, 1981; Kauvar, 1982). Thus, cAMP metabolism appears to be intimately involved in behavioral plasticity, a conclusion also reached from the elegant work with the sea snail, ApZysia eaEifornica (Byrne, 1987; Goelet et al., 1986). Although the steps following CAMP involvement in the biochemical pathway(s) required for normal learning/memory in flies are unknown, it is assumed that CAMP-dependent protein phosphorylation, cA,MP-dependent gene expression, and/or CAMPgated ion channels are likely to be involved. Despite the clear identification of dnc as the struct)ural gene for cAMP PDEase. previous molecular studies hinted at, an unexpected complexity of the gene. An extensive series of RNA blotting experiments using probes from t,he dnc genomic region identified an array of at least six large (4.2 to 9.6 kb) and overlapping poly(A)+ RNAs as dnc transcripts (Davis & Davidson, 1986). These RNAs are extremely rare, existing in the adult fly at an abundance level of no more than five parts per million of the polyadenylated RNA fraction (Chen et al., 1986; Davis & Davidson, 1986). Several cDNA clones representing the RNAs were selected from oligo(dT)-primed and primer extension cDNA libraries and sequenced, which revealed the introni exon organization of a portion of one dnc transcription unit, and a portion of the conceptual amino acid sequence of the protein product (Chen et al., 1986, 1987). In addition, they demonstrated the existence Sgs-4, within a of several other genes, including 79 kb intron of dnc (Chen et al., 1987). The studies presented here were designed t’o elucidate the basis for the RNA het’erogeneity and to rontribute to our general understanding of the gene’s structure and promoter regions. Tn particular, we wished to know whether the marked RKA heterogeneity reflected alternative processing to produce PDEase isoforms, or whether the heterogeneity reflected alternative untranslated sequences which might confer differences in stability or translational efficiency. The former possibility seemed particularly attractive since rats have at least four genes which encode different CAMP-specific PDEases (Colicelli et al., 1989; Davis et al., 1989: Swinnen et al., 1989), and since dnc is the only Drosophila locus that codes for this class of enzymes (Davis & Kauvar, 1984). The possible biological need for multiple enzymes might be accomplished by alternative splicing of protein-coding exons. Through sequence analysis of additional cDNA clones and the corresponding genomic exons, we demonstrate that the transcript heterogeneity arises by transcription from at least three start sites as
t Abbreviations used: PDEase, phosphodiesterase: kb, lo3 bases or base-pairs; bp. base-pair(s); ORF. open reading frame.
well as alternative splicing of exons contribut)ing t’o the open reading frame. In addition. differential processing of 3’ ends produces transcripts that vary in length by as much as 3 kb.
2. Materials and Methods (a)
Isolation
oj
RX.4
Qnd
tlS.4
blotting
Total RKA was prepared from Canton-S adult flips according to either the met,hod described b,v Labarca & Paigen (1977) or procedures used previously (Davis & Davidson. 1986). Polyadenylatrd RNA was select,ecl b? I or 2 passages over an oligo(dT)-c~ellulosr (Collaborative Research, type 3) column as described by Maniat,is rf nl. (1982). RNA blots were prepared and probed as described (Davis & Davidson. 1986). except t,hat single-stranded probes (see below) were used. III some caases. the stringency of the hybridization and washing was rrducrd because of the high A + T ront,ent of a particular axon (exon 10) or the short homology predicated between the probe and RSA (exon 4). Size markers for RiKA blots were RKA ladders purchased from Mhesda Researcbh Laboratories.
One primer extension cDXA library was c,onstruc+trd from 5 pg of poly(A)+ RNA by priming the first strand synthesis using M-MLV reverse transcriptase (Bethesda with a unique 20-me1 Research Laboratories) (.5’-L4GTCt~L4GCTCCTCGATTGTG-3’). whose srquencsr is complementary t*o a portion of exon 6 (Fig. 2). The se~~ntl strand was made according to the method described b> Gubler & Hoffman (1983) without, the use of Escherichin coli DEA ligase. The cDKA was then methylated. bluntended. and EcoRT linkers (New England Biolabs) were added followed by EcoRT digestion. To frac%ionate the resulting cDNA and to remove excess linkers. the restrict,ion reaction was chromatographed on a Bio-gel A-5Om (BioRad) csolumn according to Huynh et ~2. (19%). Fractions containing cDNA greater t,han 500 bp were pooled and ligated to LgtlO arms. The librar? was packaged in vitro and screened without amplification. Two additional cDNA libraries in igt 11 were scrrened using probes representing 5’ portions of different cI)?r’As. B total of 1.7 x lo6 phage were screened from a Canton-S head cD?JS library obtained from M. Robash (Brandris). Three independent clones were recovered. each more than once. We also screened (2 x lo6 phage) a (‘antorl-8 head cDNA library made by P. Salvaterra (Beckman Research Institute). Sine independent clones were recovered. most of them more than once. The inserts of’t,he (sDK.4 clones were subcloned int)o plasmids for sequencing.
Genomic and cDNA sequencing was performed with thr dideoxy chain termination method using either the Klenow fragment (Pharmacia) or phage T7 DXL4 polymerase (U.S.B.), following the manufacturer’s conditions or our own modified conditions. Most of the sequence of genomic regions from co-ordinates + 21.2 t,o +489 was obtained from random phage Ml3 subclones of apprtrpriate fragments generated by the sonication method described by Deininger (1983). Tn some cases. fragments to be sequenced were cleaved with selected restriction enzymes and the subfragments subcloned into Ml 3 vectors in a shotgun fashion. Alternatively. progressive
Dunce deletion subclones were generated by a simplified ExoIII deletion method described by Henikoff (1984). A total of 70,500 residues were sequenced from co-ordinates +21.2 to +469, such that each of the 23,607 residues was sequenced 3.0 times, on average. We connected the sequences of these fragments by sequencing through the junction restriction sites, except for the Hind111 site at co-ordinate +3%9 and the EcoRI site at +41.3. Sequences for cDNAs were obtained either by the simplified ExoII deletion method or by sequencing subcloned restriction fragments of the original cDNA clones. (d) S,
555
Structure
poly(A)+ RNA for primer extension experiments, or with an equal amount of yeast transfer RNA as a control. The products of S, nuclease or primer extension mapping were sized using DNA sequencing ladders or labeled pBR322 HpaII restriction fragments. (e) Expression
Expression of dnc cDNA clones in yeast and CAMP PDEase assays were as described (Henkel-Tigges & Davis. 1990).
experiments, primer extension analysis, and single-stranded probes
n&ease
3. Results and Discussion
S, nuclease mapping of the 5’ ends of dnc transcripts was performed according to the procedure described by Berk & Sharp (1977). Primer-extension analysis was carried out as described by Laughon et al. (1986). 32Plabeled probes used for Si nuclease mapping, primer extension and RNA blotting experiments were generat’ed by priming DNA synthesis across an insert in Ml3 (Burke, 1984), truncating the product at an appropriate restriction site and isolating the single-stranded probes on denaturing polyacrylamide or agarose gels. Radiolabeled DNA was hybridized with 10 pg of poly(A)+ RNA from Canton-S adult flies for 8, nuclease analysis, 30 pg of
z-40
experiments
(a) New cDNA
clones de&e dnc transcripts
six classes of
Three cDh’A clones previously characterized and named ADCl, ADC6 (Chen et al., 1986) and ~863 (Chen et al., 1987) overlap to define a major portion of one type of dnc RNA designated as class I (Fig. 1). More specifically, these cDN.4 clones contained a portion of exon 1, and exons 2 through 13. In addition, they defined the 79 kb intron between exons 2 and 3 that houses several other genes, including Sgs-4, Pig-l and Indnc (Fig. 1).
kb
MI I)
2.7 2.8 lndnc
I 4I I II 11111 B
3
5 67
“9’9,12
ATG
stop
ATG
stop
13
IA
(7)
f stop
ATG
(15) ATG
III ATG ELA
ATG
stop I II
f
(1)
stop I II
f
(2)
3
(4)
stop I
Figure 1. Schematic representation of the dnc genomic region and dnc RNAs. An arbitrary co-ordinate system, extending in this Figure from -50 to +47, has been described (Davis & Davidson, 1984), but omits 7.3 kb of sequence contained in a transposable element inserted in the Canton-S strain between co-ordinates 2 and 5. The position of exon 69 at approximately -90 is estimated, since the region between exons 69 and 1 has not been completely cloned. Dunce exons are represented as numbered boxes and intronic genes by arrows. These are not drawn to scale. The intronic genes shown include Sgs-4, several Pig genes and the genomic region coding for a 2.0 kb RNA (Indnc). Splicing patterns of 6 different classes of RNAs are illustrated (IA to IVB). Filled portions represent sequences contained in cT)NA clones recovered. Open portions represent sequences predicted to be present in RNAs from RNA blotting, S, nuclease, and primer-extension experiments. The number in parentheses to the right of each class is the number of independent clones recovered and assigned to each class, with 7 clones representing both IA and IB since these could not be distinguished. The putative open reading frame in each class is shown by ATG and STOP. tss, transcription start site.
556
Y. Qiu et al.
In order to understand the sequence heterogeneity of dnc RNAs detected by RNA blotting experiments, we carried out detailed molecular analyses of many additional cDNA clones. A primer extension cDNA library constructed using a synthetic oligonucleotide in exon 6 (Chen et al., 1987) provided 17 additional clones beyond the initial isolate ~863 upon screening with probes representing exons 3 and 5. This primer was chosen based on the observation that sequence information of exon 6 exists in most, if not all, of the dnc RNAs (Fig. 6). Many of these clones were sequenced and others mapped extensively with restriction endonucleases to determine overlaps. S1 nuclease and primer extension experiments demonstrated that none of these cDNA clones extended to transcription start sites (not shown). Therefore, oligonucleotides representing the 5’ portion of several of the cDNA clones as well as ORF probes were used to screen two cDNA libraries made from head poly(A)+ RNA. Twelve additional and independent clones were obtained. These clones were also subjected to comparative restriction mapping, hybridization analysis to genomic clones, and sequence analysis along with the hybridizing genomic fragments to complete a detailed picture of the intron/exon structure and RNA splicing patterns. From these results, and the results from 6, nuclease and primer extension experiments presented below, several conclusions regarding the structure of dnc, its RNA transcripts and its protein products can be drawn. First, the cDNA clones fall into six structurally distinct and overlapping classes (Fig. 1). The clones in each class were identical in restriction map and sequence? except that some extended further to the right and/or left than others. Second. all of the t’ranscriptional units are large and one is exceptionally large, extending over more than 148 kb. The exons contained within the cDNA clones were identified by hybridization with genomic clones from a chromosomal walk (Davis & Davidson, 1984), representing sequencesfrom - 50 to + 47. Exon 69 resides outside the region covered by the chromosomal walk and was localized by screening a genomic library with the 5’ portion of class I cDNAs. Examination of the restriction map of genomic clones containing 0.9 suggested that these sequenceswere just to the right of the Notch locus. This was confirmed by subsequent hybridization experiments, which showed that exon 0.9 maps approximately 10 kb from the 3’ end of Notch (Kidd et al., 1986; Wharton et al., 1985). Since the Notch locus residesat band 3C7, dunce must extend from approximately chromomere 3C8 through 3D4, a region of nine salivary gland chromosome bands. Third, the intron defined by exons 99 and 1 is very large, although probably not as large as that defined by exons 2 and 3. A precise estimate of the size of the intron is not possible, since genomic clones covering the region have not yet been isolated. However, partial walks from the Notch
locus (S. Kidd $ M. Young, personal communication) and from exon 1 (Y. Qiu 8r.R. Davis, unpublished results), indicate that the intron must exceed 41 kb. In addition, this intron is like that between exons 2 and 3 in that it contains several genes unrelated to dnc. At least three genes, one a Sgs-like gene (Furia et al., 1990) and two others similar to Pig-l in spatial and temporal expression pattern, and named Pig-2 and Pig-3, reside just to the left of exon 1 (S. Beckendorf, unpublished results). Fourth, the RNA sequence heterogeneity occurs within 5’ and g-untranslated regions as well as within the protein coding regions of the transcripts (see below). Five of the RNA classes contain different exons coding for RNA leader sequences. These sequence differences open the possibility of transcript-specific stability or translatability determinants. The sequence of all exons defined by the cDNA clones, along with some sequence of the flanking introns, is presented in Figure 2. Several features of the sequence are to be noted. First, the splice junctions defined by the clones conform to consensus splice sequences(Mount, 1982). Second, exons 1. 2,3 and 2.8 are punctuated with short opa repeat,s (CAA/G: Wharton et al., 1985). These and otfher similar triplet repeats are common within Drosophila transcriptional units but t,heir signiticance remains unclear. Third, a,ddit,ional ORFs that show a codon bias expected for Drosophila protrincoding sequencesexist within dnc RNA molecules upstream from those that code for cAMP PDEase (see below). One of these is specific to class I1 transcripts, being contained within exon 2.3, and could encode a protein molecule of 77 residues (Fig. 2). A search of the Protein Sequence Databa,sr of the National Biomedical Research Foundation using the conceptual translation sequence of t,his ORF revealed no significant similarity with known proteins. It has been postulated that additional ORFs such as these might code for proteins serving regulatory roles (Kozak. 1987). (1)) The dnc proteins
The CAMP PDEase ORF was reported to extend from exon 8 to exon 13 (Chen et al.. 1986, 1987), predicting a protein molecule of 40,000 J$. However, several sequencing errors were discovered from the analysis of the new cDNA clones, one of which when corrected shifted the prior reading frame to extend it into sequences previously thought to be a non-coding leader. The first ATG defining the ORF for each of t,he various transcript classes is located as shown in Figure 1. If each ATG were used in witlo, four classes of protein product with distinct N-terminal sequenceswould result. In addition to the potential N-terminal heterogeneity, exon 4 is used optionally for class IV transcripts, predicting the presence/ absence of a cassette of 13 additional amino acid residues in some of the protein isoforms. And. one case of alternative 3’ splice site choice was detected
Dunce Structure among the cDNA clones which also predicts protein product variation. For the 25 clones which covered the exon 5-6 splice, the 3’ splice site bordering exon 6 in 11 of the clones resides six bases to the left of that in the other 14. Some protein products should therefore carry two additional amino acid residues. This alternative 3’ splice choice was found in cDNA clones representing classes I and II, but not for classes III and IV. In all, a total of seven possible protein products are predicted from the cDNA clones. The calculated mass for class I, II, III, IVA and IVB products are 79,000, 85,000, 71,000, 77,000 and 76,000 M,, respectively. The conceptual translation sequences of the various ORFs are shown in Figure 3. It is note-
557
worthy that all of the predicted proteins contain the PDEase “conserved domain” of approximately 260 amino acid residues found in many different types of PDEase. This domain has been assigned a catalytic function, since proteolytic fragments of other types of PDEase that contain this core exhibit full catalytic activity (Stroop et al., 1989; Carbonneau, 1990). (c) Expression
of a
dnc cDNA
clone
in
yeast
To verify that the cDNA sequences are capable of coding for authentic CAMP PDEase activity we expressed two different cDNA clones in yeast. One clone contained 43 residues of exon 5, exon 6 to 12,
Fig. 2.
Y. Qiu et al.
Figure 2. Sequence of the dnc exons and portions of flanking introns. Exons are boxed, listed in upper ease. and numbered as for Fig. 1. Only a small portion of exon 13 is shown (see Fig. 7). The sequence complementary to the primer used in the primer extension library construction is boxed within exon 6. The opa repeats are underlined. The I’stT site in rxon 27 used for primer extension is double underlined. The 6 residues missing in some cDNA clones due to an alternative 3’ splice site at the left border of exon 6 are marked with dots. An unidentified ORF begins wit,h residue 75 ot exon 2.3 and ends at residue 308 of this exon. The CAMP PDEase open reading frames are shaded. The first ATG resides in exon 2, 2.3, 3 and 2.8 for class I, IT. III and IV, respectively (Fig. I). The transcription initiation site within rxon 044 is marked by an arrow
and exon 13 to 275 residues past the stop codon (Fig. 2). The clone was inserted into the yeast expression vector, pADNS, which uses the ADH promoter to drive expression (Colicelli et al., 1989), and the construct was transformed into the yeast, strain lODAB (Colicelli et al., 1989). This strain contains insertional disruptions in both of its PDEase genes so that no endogenous PDEase activity can be detected. Assay of the CAMP PDEase activity in lODAB cells containing the cDNA expression construct revealed the presence of a CAMP-specific PDEase activity with a K, value of was not inhibited by 4.9( kO.3) PM. The activity 1 miv-cGMP nor was any cGMP PDEase activity detected. Furthermore. the expressed CAMP PDEase activity was not inhibited by rolipram
(Ki > 100 PM), a drug with antidepressant properties known to inhibit the rat homolog of dnc hut not the fly CAMP PDEase (Henkel-Tigges & Davis. 1990). These characteristics are like t)he actkit? found in flies and affected by dnc mutations (Davis & Kauvar. 1984), which shows the exons 5 to 13 contain all sequence information necessary for PDEase activity. A second cDNA clone, extending more 5’ through exon 3 also expressed a (aAMPspecific PDEase activity in this assay. (cl) Transcription
initiation sites for class i und II RNAs
In order to determine the 5’ boundary 5’ exon defined by each class of cDNA.
of the most S, nuclease
Dunce Structure ClawI
f:
Cl&&s
IX:
559
WQLSMSALGLQQSSSXLXSKSAlfTIEBiKSSSAGt4RTQLTL
(40)
S6GFLAPPGIBfIFILSIIHAPPGLSDblLXRAQGRSPLSF
(80)
RISFPGSDSDLFG
Class
xvr
t4vcsFccccYNPxtxsPss
(18)
All:
FDVE~GQGARSPLEGGSPSAGLVLQl!ZLPQitBESFLYRSDS
(40)
DFE~SPASWSRlSSIASE
SEGEDLIVT
(80)
PFAQILASLRSVRl?ELLS
BQSSSASRS
(=oi
GIPPGAPLSQGEEAYTRLATDTI~ELDWCLDQLETIQTSR
(160)
SVSD?4ASLKPXRMLXIELSHFSESSRSG~QISEYICSTFL
(200)
DKQQEFDLPSLRVED~PELVAANAAAGQQSAGQYARSRSP
(240)
RGPPMSQISGVXRPLSXTNSFTGERLPTFGVETPRENELG
1280)
PPPSGVDENPQEl3ktfBFQVTLEESDQBNLhELBEGDgSGE
(640)
ETTTTGTTQTTAhSALBAGGGGGGGGGB8AFRTGGCQNQPQ
(6=‘t
HGGM* Figure 3. Predicted amino acid sequences of dnc PDEases. The translation of the sequence from the 5’ end of exon 3 to the stop codon in exon 13 is labeled All. The K-terminal sequences unique to the classes are also listed. The first Met residue downstream from the in-frame stop codon in class III is underlined and the 13 amino acid residue cassette encoded by the optionally used exon 4 is double underlined. The shaded region is that homologous to many PDEases and known as the “conserved domain”. The two amino acid residues eliminated in some PDEase forms due to 3’ splice site choice are boxed at residues 100 and 109 in the common region.
protection experiments were performed. For class I transcripts, a 2.2 kb single-stranded and uniformly radiolabeled genomic DNA fragment, extending from a synthetic primer (close to the BstYl site) in exon 0.9 to an upstream Hind111 site (Fig. 4), was hybridized to adult poly(A)+ RNA. The resulting duplexes were treated with different amounts of S, nuclease and the protected fragments sized by denaturing gel electrophoresis (Fig. 4). One protected band measured 130( f2) residues from the SpeI site in exon @9; a second measured 138( +2). A second probe with a different right endpoint. identified the
same left endpoints (not shown). To determine whether these sites are transcription initiation sites, primer extension experiments were carried out. A 184 base single-stranded and uniformly labeled primer with its 3’ end at the SpeI site was hybridized to adult poly(A)+ RNA and extended with reverse transcriptase. One product extended to the same site marked by the smaller S, protected product at 130() 1) residues from the SpeI site. This suggests that this position is a transcription start site. A second product was about 150 residues longer than the first. Another primer with its 3’ end
Y. Qiu et al.
560
t
S, nucleaee (a
Primer
1
extention (5)
I H I
R 6
I kb RS
I BSa
-
Figure 4. S, nuclease and primer extension analysis for class I cDNAs. (a) Nuclease S, and (b) primer-extension products after fractionation on 5 ye polyacrylamide/urea gels. P, intact probe for (a); it migratesat the top of the gel and is not included in the portion of the gel shown. A+, (a) probe or (b) primer hybridized with poly(A)+ RKA prior to digestion or extension. t, control lane with an equal mass of yeast tRNA substituted for poly(A)+ REA. The sequencing ladders show the actual sequence of exon 99 primed with the same primer used to prepare the probe or the primer extension product. The small arrow in (a) identifies the second band which is 8 residues longer than the first one. The arrow in (b) indicates the second band which is about 150 residues from the first one. A schematic drawing of the genomic region of exon 99 is shown at the bottom. The filled box, exon 99; thick arrow, the probe for (a); thin arrow, the primer for (b). H, WindIII; B, B&YI; R, EcoRI; So, SalI; S, &eI.
at the BstY 1 site also generated these two products (not shown). These data suggest that a transcription initiation
site resides
eight
bases more
3’ than
the
left end of exon 0.9, and that the left end is spliced to at least one upstream exon of 150 bases. The sequencein this region is consistent with this interpretation. A consensus3’ splice site is found at the left border of exon 0.9. The sequence at the indicated transcription initiation site (TCAGTT) conforms nicely with a sequencefound at the beginning of many Drosophila RNAs (TCA(G/T)T(C/T); Haynes et aZ., 1990). The RNAs using this transcrip-
tion initiation site apparently exist at an abundance level similar to those containing the upstream exon(s), since the two S,-protected bands are of similar intensity (Fig. 4). Similar experiments suggest that, exon 2.7 is a transcription start site. A single-stranded and uniformly labeled genomic fragment, bridging the terminal nucleotide of the longest class III cDNA clone was used for S, nuclease analysis (Fig. 5). A 389( & 5) bp fragment was protected. This maps a splice site or a transcription start site on the sequence as shown in Figure 2, 17( +5) residues from a PstT site (CTGCAG) in exon 2.7 (Fig. 2). Two other S, nuclease experiments (not shown) using probes with different right endpoints also mark this site. A single-stranded and uniformly labeled primer with its 3’ end at the PstI site in exon 2.7 (Fig. 5) was hybridized to adult poly(A)+ RNA and extended with reverse transcriptase. This produced an extension product of 17(f2) bp (Fig. 5). A second primer with a different 3’ end was also extended to this point (not shown). Thus. t,hese experiments identify a transcription start site at approximately 17 residues to the left of the PstT site. The sequence at the transcription start site, CCAGTG, shows a limited relationship, if any, to the transcription start site sequences of other Drosophila genes. The sequence just upstream from exon 0.9 is unusually rich in A and T residues, with an A + T content of 80% in the first 100 bp. However. no obvious promoter sequence homologies are present in the region upstream from this exon or exon 2.7. Similar experiments were performed for exons 2.3 and 2%. The S, nuclease experiments for these (Chen, 1987) identified the 5’ ends of these exons at’ accept,ors sequences matching consensus splice (Fig. 2) with primer extension products extending well beyond these sites. These results indicate that exons 2.3 and 2.8 splice to unidentified upstream exons although exon 2.8 may splice to 2.7 (see below). Screens of our existing cDNA libraries and construction of new primer extension cDNA libraries have failed to identify further upstream sequences. (e)
RNA
blotting exon-speci$c
experimen,ts probes
with
We performed RKA blotting analysis using probes specific for each of the identified exons to relate the RNA speciesidentified on the blots to the different classesof cDNA clones. Probes specific for exons 1 and 2 detected RNA molecules of 9.5 kb and at least one speciesof 7.2 kb (Fig. 6: Chen et ab., 1987). These are designated as class I dnc transcripts.
The
exon
0.9 probe
consistently
produces
RNA blots of low quality detecting a diffuse band of 7.2 kb, and a less-convincing one of 95 kb. An exon 2.3 probe detected 9.5 and 7.0 kb RNAs (class IT the 9.5 kb transcript. transcripts). However, detected by the exon 2.3 probe migrates slightly faster than the one detected by the exon 1 and 2
Dunce Structure
P
A+
t
561
M
3QQw “5
Primer
extension (b)
sa/ I I
PSll I
Figure 5. S, nuclease and primer extension analysis for class III and IV cDNAs. (a) S, nuclease and (b) primer extension products of exon 2.7 fractionated on 6% polyacrylamide/urea gels. P, intact probe for (a); A+, (a) probe or (b) primer hybridized with poly(A)+ RNA prior to digestion or extension. t, control lane with an equal mass of yeast tRNA substituted for poly(A)* RNA. M, size markers. A schematic drawing of the genomic region of exon 2.7 is shown at the bottom. The filled box, exon 2.7; thick arrow the probe for (a); thin arrow the primer for (b).
562
Y. Qiu et al.
0.9
I
2
2.3
2.7
2.8
3
4
5
6
7
8
9
IO
II
12
13
9.6
;:4” ,;:g m6.7 5.0 4.2
Figure 6. RNA blotting experiments using exon-specific probes. Single-stranded probes were prepared from cU,UL\ OI genomic clones and used to probe adult poly(A)+ RNA blot strips. Axons represented by the probes arc indicated at the top of the blot strips. The probe representing exon 11 overlapped exon 10 by 11 bases. but this is insufhcient homology to hybridize to exon 10 sequences. Some bands are extremely faint, due to the low abundance of the R?VTAs and/or the amount, of homology between RNA and probe. These are marked with a, filled circle.
probes. Therefore, there appear to be at least two different transcripts of about 9.5 kb. Probes specific for exons 2,7 and 2.8 both hybridize to 9.6, 7,4, 7.0. 6.7, 5.0 and 42 kb RNAs. The similarity in the hybridization patterns obtained with these probes suggests that exon 2.8 may be spliced to exon 2.7 in the majority of the transcripts (Fig. 1). Since the 5’ end of exon 2.7 appears to be a transcription start site, the 7.0 kb transcript identified with the exon 2.7 and 2.8 probes must be distinct from the 7.0 kb transcript visualized with the exon 2.3 probe. Therefore, there exist at least two distinct 7.0 kb transcripts. Because of its small size (39 bp), a probe for exon 4 barely detects any transcripts. However, faint bands are observed in the 7.0 t,o 7.2 and 9.5 to 9.6 kb size range, consistent with the contribution of this exon to class I and class II RNAs. Probes specific for exons 3, 5 to 12 and the 5’ portion of exon 13 show similar or identical patterns of hybridization to RNAs of 95, 7.2, 7.0, 5.0 and 42 kb. In addition, many if not all of these hybridize to RNAs of 9.6, 7.4 and 6.7 kb, although the signals in most cases are either partially obscured by hybridization to other RSAs of similar sizes or are very faint. However, these three R’NA transcripts are clearly detected with other exon probes, such as 2.8 and those representing portions of exon 13 (Fig. 7). These data are consistent with the idea that all classes of dnc RNAs utilize exons 3, 5 to 12 and the 5’ portion of exon 13; and that no major alternative splicing occurs between these exons. In summary, the use of exon-specific probes identifies a minimum of ten dnc R&As: 9.6, 7.4, 7.2, 6.7.
56 and 4.2 kb RNAs, and two each of 9.5 and 7.0 kb. Several other RNAs are also detected by t’hc exon-specific probes as shown in Figure 6. For example, probes specific for exons I. 2. 2.3, 2.X and 3 all detect Rh’As smaller than those noted abovcb. Some of these could be due to artifactual hybridization, since they are detected by RNAlike as well as RNA complementary probes (Chen c)t al.. 1987). However, some of the more distinct ones (i.e. thr (b6 kb RNA with exon 2 and 3.3 kb R?iA with exon 2.8) could be processing products, the products of regional transcription or t*he transcripts of other genes related in sequence to portions of dn,c. (f) The dnc tramcripts
possess dijfuen,t
J’ ends
Sane of the cI)NA clones identified to date contains a poly(A)+ tail. However, previous R?\;A blot analysis (Davis &’ Davidson, 1984) suggested that the dnc transcripts end at different sites within the +41*3 to +469 genomic region (Fig. 7). To define more precisely where the dnc transcripts end. we performed additional RNA blotting experiments using single-stranded probes representing subfragments of this region. A probe representing the entire +41.3 to +43.5 region hybridized to all of the dnr RNAs (Fig. 7), but probes representing smaller portions of the 43 to 47 region hybridize t)o successively fewer of the dnc RNAs proceeding from left to right. These results establish that the larger dnc RNAs tend to have longer 3’-untranslated tails. more than with the 9.519.6 kb RNAs containing 48 kb of 3’-untranslated sequence (Fig. 7).
Dunce
41 1
42 1
43 I
R
Structure
