A family of expressed antifreeze protein genes from the moth, Choristoneura fumiferana Daniel Doucet, Michael G. Tyshenko, Peter L. Davies and Virginia K. Walker Department of Biology, Queen's University, Kingston, Ontario, Canada The freeze-intolerant insect, Choristoneura fumiferana (spruce budworm), produces multiple antifreeze protein (AFP) isoforms for protection during t he overwintering stage. We now report the cloning of AFP genes from insects; Afp-Lu1 encodes a  9-kDa AFP isoform, and Afp-Iu1 encodes a  12-kDa AFP isoform. Both CfAFP genes have similar structures w ith a single 3- to 3.6-kb intron inter- rupting the coding region. The second exon of an additional CfAFP gene, 2.7a, encoding a new  9-kDa isoform, was found 3.7 k b upstream of Afp-Lu1 and demonstrates that some AFP family members are linked in tandem. This gene appears t o encode an AFP w ith 68±76% identity to p revi- ously isolated CfAFPs. With i ts eight Cys residues necessary for disul®de bonding and ®ve perfectly conserved ÔThr buttonÕ (Thr-Xaa-Thr) ice-binding motifs, it c an be modeled as a functional AFP. Southern blot analysis shows that there are  17 genes in this AFP family, w ith e ach of t he isofo rms represented by two to ®ve gene copies. Transcript accumu- lation from Afp-Lu1 and Afp-Iu1 (or closely related genes) was maximal during the overwintering stage, while 2.7a transcripts were only detected in ®rst instars, larvae that are normally found only in the summer. Contrary to expecta- tions, this dierential expression demonstrates that CfAFP gene family transcripts are primarily regulated during development, rather than by seasonally low temperatures. Keywords: antifreeze protein; gene family; cold stress; mo th development. Insects have developed m ultiple adaptations to changes in their environment. Environmental stress of a predictable nature such as seasonal d rought, h eat or cold may be endured by entering diapause, a physiological state charac- terized by low m etabolic activity [1]. Many species of insects prepare for overwintering by triggering t he diapause response, but freeze-intolerant species such as the spruce budworm, Choristoneura fumiferana, must a lso have some protection from freezing at subzero temperatures. Their diapause in the second larval instar is accompanied by the synthesis of antifreeze proteins [2] that depress the f reezing point of extracellular ¯uids and a llow the larvae to supercool [3]. Over the course of the winter, larvae also increase the synthesis of the low molecular mass cryopro- tectant, glycerol, from glycogen stores [4]. Antifreeze proteins (AFPs) a dsorb to microscopic ice crystals and p revent their g rowth, thus lowering the freezing point of solutions [5]. While this adsorption depresses the freezing point, it leaves the melting point unaffected, generating a thermal hysteresis. AFPs a re found in differen t organisms including ®sh, insects, plants, bacteria and fungi, but are best known in ®sh [6]. Four different types of antifreeze proteins (AFPs I to IV) and one antifreeze glycoprotein have been characterized in bony ®sh from t he icy seas of both the northern a nd southern hemispheres. At high concentrations they offer protection to the freezing point of seawater ()1.9 °C). AFPs isolated f rom beetles and spruce budworm are 10±100 times more active on a molar basis than the ®sh AFPs or antifreeze glycoproteins [2,7]. This p resumably re¯ects t he need for high TH in freeze- susceptible insects that have to cope with the unpredictable, lower temperatures of the terrestrial environment. The structures of the insect AFPs are unique and unlike the ®sh AFPs [8,9]; they are b helices with two rows of Thr residues down one side of the protein that make a good match to t he ice surface on both prism and basal planes. Thermal hysteresis activity increases with AFP concen- tration and bo th ®sh and insects are dependen t on high circulating AFP concentrations for full protection against freezing. In some ®shes, the demand for AFP has been largely met by increasing the number of AFP genes. Ocean pout and wolf®sh from cold coastal waters have  150 and 80 AFP gene copies, respectively [10,11]. O ther ®shes such as the s ea raven [12] and winter ¯ounder [13] a lso h ave multiple AFP g ene copies. I nsects may a lso use ge ne families f or AFP production. Numerous AFP isoforms differing slightly in primary sequence or m ass have b een found in the beetles, Tenebrio molitor and Dendroides c anadensis as well as in C. fumiferana [14±16]. In Tenebrio, Southern hybridizations with AFP gene probes showed numerous bands, re¯ecting the presence of a multiple copy AFP gene family [14]. In C. fumiferana, at least seven different AFP (CfAFP; previously designated sbwAFP) isoforms are known at the cDNA level. These d ifferent cDNAs e ncode p roteins o f two distinct size s,  9and 12 kDa. The larger variants c ontain a 30- or 31-amino-acid insertion creating two additional 15- or 16-amino-acid loops in the b helix [16]. The 9-kDa proteins are e ncoded by cDNAs possessing either a long ( 1000 nucleotides) or a short (  200 nucleotides) 3¢ UTR, while the 12-kDa CfAFPs are e ncoded by c DNAs w ith 3¢ UTRs of intermediate size ( 450 nucleotides). The existence of Cf AFP v ariants, differing in both protein s ize Correspondence to V. K. Walker, D ep artment of B iology, Queen's University, Kingston, Ontario, Canada K7L 3 N6. Fax: + 1 613 533 6617, Tel. + 1 613 533 6123, E-mail: walkervk@biology.queensu.ca Abbreviations: AFP, antifreeze protein. (Received 30 August 2001, accepted 19 October 2001) Eur. J. Biochem. 269, 38±46 (2002) Ó FEBS 2002 and message length strongly suggests that a gene family encodes these AFPs. This paper reports the cloning of AFP genes f rom a n insect. As w ell a s revealing additional i soform variation, the isoforms show an unexpected differential transcript accumulation pattern, especially during d iapause. MATERIALS AND METHODS Nucleic acid probes DNA probes for the detection and isolation of speci®c members of the CfAFP gene family were generated from cDNA clones 10 and 337 [16] or genomic sequences. A cDNA 10-speci®c probe encompassed the entire 414-bp coding region of the isoform. A probe speci®c for the 337 cDNA isoform was generated b y PCR with SPF and SPR primers as previously described [16]. A sequence 2.7-speci®c probe was generated with the SPF primer and primer 2.7R (5¢-TCAGGACACTACTTTCAC-3¢), located 58 nucleo- tides downstream of the putative termination codon. Genomic DNA isolation and Southern blots C. fu miferan a were obtained from t he Canadian Forest Service ( Sault Ste-Marie, O N, Canada). Genomic DNA was puri®ed from a single sixth instar larva by proteinase K digestion and phenol/chloroform extraction using the method of Blin & Stafford [17]. DNA samples (20 lg) were digested with the restriction enzymes HindIII, EcoRI, SstI, and BamHI (Gibco BRL, Burlington ON, USA), electro- phoresed in a 0.9% a garose gel, and subsequently trans- ferred to Hybond N nylon membrane (Pharmacia, Baie d'Urfe  , Quebec, Canada), following the manufacturer's protocol. Hybridization was carried in 0.25 M Na 2 HPO 4 / 7% SDS [18] or 5 ´ NaCl/Cit/0.5% SDS/5 ´ Denhardt's solution for a minimum of 12 h [19]. To examine the complexity of the gene family, the blot was hybridized with the 337 isoform probe and washed u nder low stringency ( 0.1 M Na 2 HPO 4 /5% SDS, 65 °C, 2 ´ 15 min). To detect speci®c hybridization signals for the 337, 10 and 2.7a probes, the blots were hybridized successively with each probe and washed under high stringency conditions (0.2 ´ NaCl/Cit/0.1% SDS, 65 °C 15 min). After washing, blots w ere exposed to a Phosphorimager screen (Pharmacia) at room temperature or to Biomax MS ®lm (Eastman Kodak, Rochester, NY, USA) at )80 °C. Genomic DNA library screening A C. fumiferana genomic library constructed in kDASH II (Stratagene, LaJolla, CA, USA) was plated, and 6.5 ´ 10 5 plaques were s creen ed using the 337 cDNA fragment probe. Filters were washed using the low stringency conditions described above. Positive clones were randomly picked, a nd their DNA isolated and d igested with HindIII, SstIand XhoI restriction enzyme s to generate restriction enzyme maps. DNA fragments, obtained by restriction digests or long PCR (Expandä Long Template PCR, Boehringer Mannheim, Laval, QC, USA) using vector primers, were subcloned into pBluescript (Stratagene) or pCR2.1 (Invi- trogen, Carlsbad, CA, USA), respectively. DNA encoding AFPs was sequenced on both s trands. Noncoding DNA was sequenced in one direction only. RNA isolation and Northern blots C. fu miferan a used for developmental studies were reared on an arti®cial diet [20]. All individuals were maintained at 23 °C under a 16-h light/8-h dark photoperiod regime, except for diapausing second instar larvae (L2) that were switched to 2 °C and constant darkness after 2 weeks. Eggs were collected 3±5 days after oviposition and ®rst instar (L1) samples w ere s acri®ced 48±51 h after egg hat ching. L2s were collected from che esecloth at various times a fter hibernaculum (cocoon) spinning and molting: after 1 and 2 w eeks at 2 3 °C and 1, 5, 10, 15 a nd 3 0 weeks af ter storage at 2 °C. Post-diapausing instars (L3±L6) were sacri®ced 24±48 h after molting. Animals were frozen in liquid n itrogen andstoredat)80 °C. Total RNA was isolated with Trizol (Gibco BRL) using a modi®cation of t he manufacturer's protocol [21]. RNA (10 lg) was loaded onto a 1.2% agarose/formaldehyde gel and transferred to Hybond N according t o established metho ds [ 19]. Blots w ere h ybridized successively with the appropriate AFP probes in 50% formamide/5 ´ NaCl/Cit/5 ´ Denhardt's solution/0.5% SDS and 100±150 lg of sheared salmon sperm DNA (Gibco BRL) for a minimum o f 12 h. T he blots were a lso probed with a n a-tubulin fragment from Drosophila mela- nogaster as a RNA loading c ontrol. High stringency washes were carried out in 0.1 ´ NaCl/Cit/0.5% S DS at 65 °Cfora minimum of 20 min. Detection of signal on X-ray ®lm was carried out as described for Southern hybridizations. RESULTS Isolation and characterization of C. fumiferana AFP genes When C. fumiferana genomic libraries were hybridized with the 337 isoform CfAFP probe, a total of 165 positive plaques were identi®ed. Of these, two were chosen and mapped using restriction enzymes (Fig. 1). One of the two, 2.7, h ad a 13.9-kb insert that contained the complete coding, upstream and downstream nucleotide sequence c orrespond- ing to a member of the long 3¢ UTR class of isoform. This gene, designated Afp-Lu1 (Long UTR 1) has six nucleotide substitutions in the coding region compared to 337 cDNA (three are nonsilent: Glu36 ® Asp, Thr67 ® Ser and Ser76 ® Leu) and two substitutions compared to another isoform of the long UTR class, 333 [16] (both substitutions are nonsilent, with Glu36 ® Asp and Ser84 ® Thr changes). It is possible that Afp-Lu1 is an allelic variant of either of these  9-kDa isoform proteins, or represents a closely related isoform. A single 3.6-kb intron, identi®ed by comparison with the cDNA sequence, is positioned after the ®rst nucleotide of t he Val16 codon in the signal sequence, and possesses conserved splice junction sequences found in Drosophila and in the C. fumiferana trypsinogen gene [22,23] (GenBank accession no. AF32 5859). Afp-Lu1 contains several kb o f 5¢ ¯anking DNA that was examined for potential regulatory regions. A putative TATA box was found at position )84 relative to the translation start site, and a putative CAAT box at position )116. The clone cont ained 6 kb of 3¢ ¯anking DNA, of which 1.4 kb contiguous to Afp-Lu1, were sequenced. This section was very similar to the 3¢ UTR of isoform 337, except for an insertion of an additional 517 nucleotides. No Ó FEBS 2002 Antifreeze protein genes in Choristoneura (Eur. J. Biochem. 269)39 splice jun ctions were found within this 3¢ UTR insertion indicating that it is not an intron. Sequencing also allowed the identi®cation of putative polyadenylation signals at positions 4548, 4650 and 4785. Upstream of the Afp-Lu1 gene a t )641 t here was a large i nverted repeat spanning about 500 bp (Fig. 1). Approximately 200 nonoverlapping nucleotides of this palindrome w ere s equenced and it appeared to be unrelated to any repetitive or transposable element in the GenBank database. Clone 2.7 also contained a sequence which, when conceptually translated, showed 87% identity with the second exon of Afp-Lu1. This sequence, tentatively desig- nated 2.7a (as i t should not be designated a formal gene symbol without the corresponding cDNA), was located  3.7 kb upstream from the translation start of Afp-Lu1. The sequence does not have the obvious features of a pseudogene such as stop codons or frameshifts. It has conserved C ys residues and putative ÔThr buttonÕ ice- binding motifs (Thr-Xaa-Thr; where X aa is any a mino acid) every 15±17 residues (Fig. 2A) a nd therefore probably encodes a functional CfAFP. With a predicted mass of 9320 Da, i t would be placed either in the long or short 3¢ UTR isoform class ( 1000 and  200 nucleotides, respectively) a nd indeed it shows similar amino-acid identity (68±73%) with most members of both classes. As four potential polyadenylation sites were found within 1.4 kb downstream of the stop codon at 845, 1102, 1304 and 1760, it has been provisionally placed in the long UTR isoform class b ased on its nucleic acid sequence and the potential size of its mRNA. Molecular modeling of t his new, putative isoform, based on the NMR-derived structure of the 337 isoform, is consistent with the 2.7 a sequence e ncoding a functional AFP (Fig. 2B). In deed, this gene would e ncode an isoform containing all ÔperfectÕ Thr-Xaa-Thr repeats. Of a dozen previously identi®ed AFP isoforms, all contain imperfect repeats of the ice-binding motifs, except for 2.7a. (Fig. 2A; M. G. Tyshen, unpublished results) The second phage p laque characterized from the g enomic library, clone 2.26, had a 13.9-kb insert which included the complete sequence for a gene of the intermediate size 3¢ UTR c lass of AFP ( Fig. 1). With t he exception of the longer coding region, the overall structure of the AF P-Iu1 (inter- mediate UTR 1) gene is very similar to Afp-Lu1. The open reading f rame is interrupted by a s ingle phase 1 intron at t he conserved V al codon (Val15 in Af pIu-1). At 3 kb, this intron is only 600 bp sm aller t han t hat o f Afp-Lu1 .Theinterme- diate UTR isoform class encodes AFPs that have two additional structural repeats (loops of the b helix), and thus have a 30-amino-acid insertion in the coding region. The coding sequence of the Afp-Iu1 (GenBank accession no. AF325857) matched the larger 12-kDa isoforms with 10 nucleotide differences when compared to cDNA 10 (four resulted in amino-acid changes: Asp13 ® Asn, Gln88 ® Arg, Tyr89 ® Phe and Asn101 ® Ser). A puta- tive TATA box was found at )79, but no evidence of a CAAT box could be found. AFP-Iu1 had the same polyadenylation signal, AATATA, found in the cDNA 10 isoform. Southern analysis In order to estimate the copy number of the cloned genes, probes representing cDNA 337 and 10 clones, as well as the newly discovered 2.7a sequence, were hybridized to blots of digested C. fumiferana genomic DNA (Fig. 3A,B). South- ern hybridization with the 337 probe, washed at low stringency, showed that the AFP gene sequence i s presen t in multiple copies. Washing conditions were chosen to detect hybridization equivalent to a T m  20 °C below that of routine washes with the 337 probe. Assuming a 1% mismatch per degree of difference [24], it was estimated that even the more divergent intermediate UTR class [16] would be detected. Examination of Fig. 3A demonstrates that bands corresponding to those seen with the interme- diate UTR class probe, 10, could also be seen in the blot washed at low stringency, verifying this procedure. After washing at low stringency, approximately 17 bands (consistent with the number of hybridizing plaques/genome in the library screens) of various intensities could be resolved in the SstI, HindIII and EcoRI digests of the DNA from individual animals. Reprobing the same blots with sequenc- es speci®c for each of the three genes and washing under conditions of high stringency yielded different overall banding patterns. As would be expected, however, the bands were a subset of those seen after low stringency washes of the heterologous probe (Fig. 3A). Southern blots using DNA isolated independently from ®ve animals (Fig. 3B) showed that there are at least two copies of the gene hybridizing with the 337 probe (one of which would correspond to Afp-Lu1) and two copies hybridizing t o the Fig. 1. Restriction maps of C. fumiferana genomic clones encoding the AFP genes Afp-Lu1 and 2.7a (A) and Afp- Iu1 (B). Horizontal black boxes below the long ho riz ontal lines indicate the co ding regions of the AFP genes, and horizontal thin lines, r egions sharing homology with cDNA untranslated regions. The dashed horizontal line for the gene 2.7a ind icate s the putative 3¢ UTR o f the ge ne, b ased on the t ranscript size estimation. The white box in the 3¢ end of Afp-Lu1 is a 517- nucleotide nonint ronic sequence that is not present in the 3¢ UTR of cDNA 337. Ve rtical arrows point to p utative polyadenylation sites and the s tars indi cate those that a re used based again on transcript size estimation. V-shape d t hin lines joining exons represe nt s pliced regions of the AFP gene s, and the inward pointing triangles i n clone 2.7 rep- resent a  50 0 bp inverted repeat. The restriction enzymes used were: H, HindIII; S, SstI; X, XhoI. 40 D. Doucet et al. (Eur. J. Biochem. 269) Ó FEBS 2002 2.7a gene probe. Hybridization to the cDNA 10 probe (corresponding to Afp-Iu1) shows that there are at least ®ve copies (Fig. 3A,B). Northern analysis The probes used f or South ern hybridizations were also us ed to study the pattern of expression of each isoform throughout C. fumiferana development. The sizes of tran- scripts corresponding to Afp-Lu1 and Afp-Iu1 are consistent with the size class of the previously cloned cDNA homo- logues, 337 and 10, respe ctively ( Fig. 4). However, North- ern blots probed w ith the 337 probe showed an additional, smaller mRNA of 1 kb as well as the expected transcript at 1.4 kb. There was a single 1-kb transcript detected with the cDNA 10 probe, although a faint band representing Fig. 2. Compendium of CfAFP sequences. (A) Amino-acid alignment of sequences retrieved from clones 2.7 (Afp- Lu1 and 2.7a)andclone2.26 (Afp-Iu1) as well as representatives of t he intermediate (cDNAs 10, 105 and 501), long (cDNAs 4, 337 and 333), and short (cDNA 339)UTRclasses of CfAFP c DNAs. Amino acids identical to Afp-I u1 a re r epresented by dots whil e d ashes indicate gaps in the alignment. Afp-Iu1 and c DNAs 10 , 104 and 501 each encode a longer ( 12-kDa) isoform than the other cDNA or gene s e quences, which encode  9-kDa AFPs. The shaded b oxes in the alignments in dicate the conserved putative ice-binding Ôthreonine buttonsÕ of the Thr-Xaa-Thr motifs. Residues in italics make up the signal peptide. (B) Molecular model of the putative mature AFP encoded by the 2.7a gene with Thr residues pointing upward, compared to the 337 isoform. Ó FEBS 2002 Antifreeze protein genes in Choristoneura (Eur. J. Biochem. 269)41 differentially processed transcripts or a related, uncloned isoform could be seen late in second instars (see Fig. 4, arrow). The strongest hybridization signals for both probes were detected in the larval diapausing stage, from 1 week to 30weeksaftertransferto2°C. Unexpectedly, howeve r, the transcripts that correspond to the Afp-Lu1 and Afp-Iu1 genes also accumulated to relatively high levels in ®rst instars and in diapausing second instars maintained at 23 °C. A faint Afp-Lu1 signal could even be detected in egg mRNA. No Afp-Iu1 signals were seen in eggs but low transcript levels were seen in the ®fth and sixth larval instars kept at 23 °C. In order to examine the expression of 2.7a,forwhichno cDNA has been previously described, a probe was s ynthe- sized encompassing the majority of the putative coding sequence, and a small portion of the 3¢ UTR. Northern analysis showed the accumulation of two transcripts, at 1.4 k b and 1 kb (Fig. 4). In contrast to the expression Low stringency 337 10 23.1 9.4 6.6 4.4 2.3 2.0 123 45 12 345 12 345 12 345 B Low stringency 337 10 ESH A 2.7a BESHB ESHB ESHB 2.7a 23.1 9.4 6.6 4.4 2.3 2.0 Fig. 3. Southern hybridization of three CfAFP sequences to restriction-digested C. fumiferana DNA. (A) DNA from a single s ixth instar larva w as digested with four restriction enzymes. The ®rst panel shows low stringency washes of the CfAFP 337 isoform probe after hybridization. Hybridization and high stringency wash es for probes of isoforms 337 (corresponding t o Afp-Lu1), 10 (to identify Afp-Iu1 )and2.7a are s hown in the second, third and fourth panels, respectively. DNA was digested with: S, SstI; E, EcoRI ; H, Hin dIII and B, BamH1. Molecular ma ss markers ( in kb) for the restriction fragments are indicated on the left. (B) Southern hybridization o f ®ve dierent individua l sixth instar DNA under the same conditions as in (A) above. Genomic DNAs were digested with the restriction enzyme SstI. 42 D. Doucet et al. (Eur. J. Biochem. 269) Ó FEBS 2002 pattern of the previous described genes, however, the two 2.7a mRNAs accumulated to higher levels in the ®rst instar than in the second instar where the abundance appeared to be independent of diapause status. No transcripts were seen at any other stage. As p reviously mentioned t here were four potential polyadenylation sites identi®ed in 2.7a. Assuming thatthegenehasa5¢ exon and 5¢ UTR of roughly t he same length as the other AFP genes (100±115 bp depending on the isoform), transcript sizes of  0.9, 1.1, 1.3 and 1.8 kb would b e expected, suggesting that t he ®r st and third sites are recognized. DISCUSSION AFP gene families Based on the discovery of several CfAFP isoforms differing in sequence, it was postulated i n an earlier s tudy [16] that members o f a multigene family would encode them. H ere w e have addressed this hypothesis b y cloning genomic frag- ments containing insect AFP genes. There are  17 unique loci, making t his a low abundance g ene f amily, a nd one that appears to be developmentally regulated. Gene families can often be found where high production levels of certain proteins must be achieved at a particular developmental stage [ 25±27]. Multiple g ene copies also appear t o be selected in response to environmental stress [28]. Even in Lepi- dopteran (moths) and Dipteran (¯ies) orders that have a compact genome [29], gene ampli®cation can occur if the selection p ressure is strong, such as i n the ampli®cation of esterase genes in the mosquito organophosphate insecticide resistance phenotype [30], the metallothionein gene dupli- cation in metal resistant Drosophila [31] and the magni®ca- tion of ribosomal DNA in bb ± mutants [32]. In ®sh genomes there may be even less constraint to increasing gene copy number by selection. Indeed, AFPs in ®sh a re often encoded by moderately sized multigene families. Ocean pout caught in the cold waters off the Newfoundland coast have  150 AFP genes, and those caught in a more southern latitude have  40 AFP genes [10]. This high gene dosage is presumably required to maintain temperature-appropriate serum AFP concentra- tions (20±25 mgámL )1 ) during the winter [33]. Analogously, multiple gene copies may have been sele cted in this C. fu miferan a population, collected from the northern boreal forest, to satisfy a similar demand for elevated levels Egg L1 Alpha- tubulin 2.7a 10 337 Probe 1w 2w 1w 5w 10w 15w 30w L3 L4 L5 L6 Pu Ad 1.5 Size (kb) 1.1 1.0 1.4 1.0 1.4 23°C2°C L2 (Diapause) Fig. 4. Expression of CfAFP isoforms 337, 10 and 2.7 a and a control gene (a-tubulin) during d evelopment. Eggs, all six larval instars ( indicated w ith the ÔLÕ p re®x a nd a number), p upal (Pu) a nd adult ( Ad) stages of the insects w ere tested f or e xpression by northern hybridization. S izes o f the AFP mRNAs (in kb ) are indicated at t he right. Second instars in diapause, 1 and 2 weeks (w) after hibernaculum spinning (L2, 23 °C) as well as 1, 5, 10, 15 and 3 0 weeks after transfer to cold storage (L2, 2 °C) were sa mpled. The arrow on t he is oform 1 0 p anel i ndicates a 1.4-kb transcript seen later during diapause. The 337 probe hybridizes with transcripts c orrespondin g to the Afp-Lu1 gene and the 10 probe hybridizes w ith transcripts corresponding to the Afp-Iu1 gene. Ó FEBS 2002 Antifreeze protein genes in Choristoneura (Eur. J. Biochem. 269)43 of AFP in the hemolymph of overwintering larvae. I nsect AFP is hyperactive compared to ®sh AFP, but nevertheless, selection for increasing gene copy number would be strong presumably because of the more extreme subzero terrestrial temperatures. The beetle, Tenebrio molitor, also has a hyperactive AFP and has 30±50 gene copies as detected by Southern hybridization [14]. Curiously then, T. molitor with at least twice the number of A FP genes i s a domestic s pecies that overwinters in granaries at more moderate temperatures than C. fumiferana, which undergoes diapause at the tips of coniferous tree branches in the boreal forest. Diapausing C. fumiferana, however, as w ell as synthesizing A FPs, spin a silk hibernaculum, which may prevent inoculative freezing. During the winter they also increase the concentration of the cryoprotectant, glycerol, 10-fold and desiccate to 40% of the prediapause water content [4]. Taken together, these adaptations may explain the impressive ability of C. fumiferana to survive temperatures of )30 °Corlower. The existence of a C. fumiferana AFP s equence upstream of the gene Afp-Lu1 on clone 2.7 provides the ®rst direct evidence that some AFP genes are tightly linked in this insect. However, v ery l arge arr ays of tandem genes a s f ound in ®sh type I and III AFPs [11,13], and hypothesized for Tenebrio AFP [ 14], are unlikely. Southern blots indicate a somewhat smaller A FP gene family (Fig. 3A,B), and because the hybridization signals were distributed between several, nonidentical large f ragments of DNA, it is likely that at least some of the  17 genes of the family are spaced several kb ap art (as fo r Afp-Lu1 and 2 .7a), e ven though they may be linked. In both Afp-Lu1 an d Afp-Iu1, the majority of the sequence is taken up by a single, intervening sequence of at least 3 kb. This is a relatively large intron for species with a compact genome like C. fumiferana, and often such large introns are characteristic of d evelopmentally regulated g enes [34,35]. In addition, a large palindrome of  500 bp (of which  200 b p at each extremity was sequenced) was found 641 bp upstream of Afp-Lu1 in clone 2.7 (Figs 1A and 2A). There are many examples of palindromic sequences or inverted repeats associated with tandem amplicons in other systems [36±38] and it is possible that the CfAFP-associated sequence could promote the rear- rangement of the DNA as well as mediate gene conversion events in the AFP gene cluster. AFP gene expression Transcripts corresponding to Afp-Lu1 and Afp-Iu1 accu- mulate in the second larval instar. This is consistent with the detection of CfAFP isoforms in larval extracts of the same stage, 1 2 weeks after storage at 2 °C [2,17];. Although AFP would be obviously required during the obligate diapause where subzero temperatures are nor- mally experienced, AFP messages are not diapause-speci®c because approximately equivalent levels were detected in ®rst instars, which in the wild, are exposed to the high temperatures of mid and late summer. Temperature, and constant darkness, did not affect tran script accumulation of Afp-Lu1 or Afp-Iu1 either, as no difference could be detected when diapausing larvae were transferred to the 2 °C incubator. This was also apparent after the transfer of L2s, kep t at 2 °C for 10 weeks, to 15 °C for 1 w eek; this treatment had no effect on transcript accumulation (not shown). Expression in other stages was low and isoform-speci®c, as exempli®ed by the transcript corre- sponding to Afp-Lu1 found in eggs and the Afp-Iu1 transcripts found in third, ®fth and sixth instars. The cloned AFP genes are thus developmentally regulated. Although synthesis of cryoprotectants in some insect species appears to be controlled by temperature or photoperiod [4], a number of Ôstress genesÕ such as the AFP genes from T. molitor [39], the heat shock protein gene, hsp70, from the ¯ esh ¯y, Sarcophaga crassipalpis [40], and the AFP genes studied here, appear to be developmentally regulated but sometimes with enhanced transcript accumulation during cold or desiccation stress [39]. Northern analysis of the 2.7a sequence, upstream of Afp- Lu1 con®rmed that it was transcribed a nd therefore not a pseudogene but transcript accumulation was different than for Afp-Lu1 and Afp-Iu1. As transcript levels for 2.7a were low in L2s, it is not surprising that an isoform correspond- ing to this gene w as not re covered in plaque lifts of a cDNA library made from s econd instars [ 16]. The reason for this differential pattern of expression in these AFP genes, however, is unclear as the insects would not normally encounter subzero temperatures f or extended periods after hatching from the egg in late summer [41]. I t i s possible t hen, that this gene (tentatively grouped with the long 3¢ UTR class) encodes a protein that accumulates in early second instars to protect against late summer frosts. Two differently sized transcripts were seen for each of Afp-Lu1 and 2.7a. As Souther n blots showed that there were two d ifferent gene copies, t hese different transcript lengths could b e encoded by distinct loci. It must be noted, however, that transcript diversity can also be generated by alternative polyadenylation [42] and several sites containing the canonical sequence AATAAA, or a single nucleotide variant of it, were found downstream of both Afp-Lu1 and 2.7a genes, followed within 30 bp by the ÔCAÕ dinucleo- tides necessary for primary transcript cleavage and poly(A) attachment [43]. AFP gene regulation by the hormonal regime preceding and during diapause would be a ttractive, as Afp-Lu1 and Afp-Iu1 appear to be similarly r egulated, but these A FP gene family members have no obvious regulato ry elements in upstream or intron regions. It is thus apparent that C. fumiferana evolved multiple AFP gene copies as part of a strategy to survive extreme winter temperatures. Naively, one might assume that the copy number was increased during evolution simply to provide for an increased t itre of hemolymph AFP. I t now appears that not only was copy number increased, but differences in protein structure, most obviously represented by larger proteins encoded by the intermediate 3¢ UTR class, as well as subtle differences in gene regulation seen during development, may all be part of a complex solution to selective forces exerted by seasonal environmental stress. ACKNOWLEDGEMENTS Dr Michael Kuiper is thanked for generously modelin g the 2.7a protein. One of two genomic libraries used was a gift from Dr Donal Hickey (University of Ottawa). The r esearch w as supported by s cholarships from the Fonds pour la Formation des Chercheurs et l'Aide a Á la Recherche and t he Ontario Gov ernment to D. D. and a grant from the 44 D. Doucet et al. (Eur. J. Biochem. 269) Ó FEBS 2002 National Science and Engineering R esearch Council of Canada to V. K. W. P. L. D. is funded by Canadian Institutes of Health Research. REFERENCES 1. Tauber, M.J., Tauber, C.A. & Masaki, S. (1986) Seasonal Adap- tations of Insects. O xford University Press, New York. 2. Tyshenko, M.G., Doucet, D., D avies, P.L. & Walker, V.K. (1997) The antifreeze potential of the spruce budworm t hermal hysteresis protein. Nat. Biotec hnol . 15, 887±890. 3. Duman, J.G., Xu, L., N even, L .G., Tursman, D. & Wu, D.W. (1991) Haemolymph prot eins involve d in insect su bzero-tem per- ature tolerance: ice nucleators and antifreeze proteins. In Insects at Low T emperatures. ( Lee, R E & Denlinger, D .L., eds), pp. 94±127. Chapman & Hall, New York, USA 4. Han, E N . & Bauc e, E. (1993) Physiological changes and t he cold hardiness of sp ruce bud worm larvae, Ch oristo neura f umiferan a, during pre-diapause and diapause development under laboratory conditions. Can. Entomologist 125, 1043±1053. 5. Raymond, J.A. & DeVries, A .L. (1977) Adsorption i nhibition as a mechanism of freezing resistance in polar ®sh es. Proc.NatlAcad. Sci. USA 74, 2589±2593. 6. Ewart, K.V., Lin, Q. & He w, C.L. (1 999) Structure, function and evolution of antifreeze proteins. Cell Mol. Life Sci. 55, 271±283. 7. Graham, L.A., Liou, Y.C., Walker, V .K. & Davies, P.L. (1997) Hyperactive antifreeze p rotein from beetles . Natu re 38 8 , 727±728. 8. Graether,S.P.,Kuiper,M.J.,Gagne,S.M.,Walker,V.K.,Jia,Z., Sykes, B.D. & Davies, P.L. (2000) Beta-helix structure and ice- binding properties of a hyperactive antifreeze protein from an insect. Nature 406, 325±328. 9. Liou, Y.C., Tocilj, A., Davies, P .L. & Jia, Z. (2000) Mimicry of ice structure b y surface hydroxyls and water of a beta-helix antifreeze protein. Nature 406, 322±324. 10. Hew, C.L., Wang, N .C., Joshi, S., Fletcher, G.L., Scott, G.K., Hayes, P.H., Buettner, B. & Davies, P.L. (1988) Multiple genes provide the basis for antifree ze protein d iversity and dosage in t he ocean pout, Macrozoarces americanus. J. Biol. Chem. 26 3, 1 2049± 12055. 11. Scott, G .K., Hayes, P .H., Fletcher, G .L. & Davies, P .L. (1988) Wolsh antifreeze protein genes are primarily organized as tan- dem repeats that each contain two genes in inverted orientation. Mol. Cell Biol. 8, 3670±3675. 12. Hayes, P.H., Scott, G.K., Ng, N.F., Hew, C .L. & Davies, P.L. (1989) Cystine-rich type II antifreeze protein precursor i s i nitiated from the third AUG codon of its mRNA. J. Biol. Chem. 264, 18761±18767. 13. Scott, G.K., Hew, C.L. & Davies, P.L. (1985) Antifreeze protein genes are tandemly linked and clustered in the genome of the winter ¯ounder. Proc. Natl Acad. Sci. USA 82, 2613±2617. 14. Liou, Y.C., Thibault, P., Walker, V.K., Davies, P.L. & Graham, L.A. (1999) A complex family of highly heterogeneous and internally repetitive hyperactive antifreeze p roteins from t he beetle Tenebrio molitor. Biochemistry 38, 11415±11424. 15. Andorfer, C.A. & Duman, J.G. (2000) Isolation and c haracter- ization of cDNA clones encoding antifreeze proteins of the pyrochroid beetle Dendroides canadensis. J. Insect Physiol. 46, 365±372. 16. Doucet, D., Tyshenko, M .G., Kuiper, M.J., Graether, S .P., Sykes, B.D., Daugulis, A.J., Davies, P.L. & Walker, V.K. (2000) Struc- ture±function relationships in spruce budworm antifreeze protein revealed by isoform diversity. Eur. J. Biochem. 267, 6082±6088. 17. Blin, N. & Staord, D.W. (1976) A g eneral method for isolation of high molecular weight D NA fro m eukaryotes. N ucleic A cids Re s. 3, 2303±2308. 18. Church, G.M. & Gilbert, W. (1984) Genomic sequencing. Proc. Natl Acad. Sci. USA 81, 1991±1995. 19. Sambrook, J., F ritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring. Harbor Laboratory Press, Cold Spring Harbor, New York. 20. McMorran, A. (1965) A synthetic diet for the spruce budworm, Choristoneura fumiiferana (Clem.) (Lepidoptera: Tortricidae). Can. Entomologist 97, 58±62. 21. Chomczynski, P. & Mac key, K. (1995) Short technical reports. Modi®cation of the TRI reage nt proced ure for isolation of RNA from polysaccharide- and proteoglycan-rich sources. Biotech- niques 19, 942±945. 22. Mount,S.M.,Burks,C.,Hertz,G.,Stormo,G.D.,White,O.& Fields, C. (1992) Splicing signals in Drosophila:intronsize, information c onten t, and conse nsus s equ ences. N ucleic A cids Res. 20, 4255±4262. 23. Wang, S., Young, F. & Hickey, D.A. (1995) Genomic organiza- tion and expression of a trypsin gene from the spruce budworm, Choristoneura fumiferana. Insect Bioc hem. Mo l. Biol . 25, 899±908. 24. Meinkoth, J . & Wahl, G. (1984) H ybridization of nucleic a cids immobilized on solid supports. Anal. Biochem. 138, 267±284. 25. Kravariti, L., Lecanidou, R. & Rodakis, G.C. (1995) Sequence analysis of a small early chorion gene subfamily inter- sperse d within the late gen e locus in Bombyx mori. J. Mol. Evol. 41, 24±33. 26. Payant, V., Abukashawa, S., Sasseville, M., Benkel, B.F., Hickey, D.A. & David, J. (1988) Evolutionary conservation of the chromosomal con®guration and regulation of amylase genes among eight species of the Drosophila melanogaster species sub- group. Mol. Biol. Evol. 5, 560±567. 27. McKenna, M.P., Hekmat-Scafe, D.S., Gaines, P . & Carlson, J.R. (1994) Puta tive Dr osophila ph eromone-binding proteins e xpressed in a s ubregion of the olfactory system. J. Bio l. Chem. 269, 1 6340± 16347. 28. Dunkov, B.C., Rodriguez-Arnaiz, R., Pittendrigh, B., rench- Constant, R.H. & Feyereisen, R. (1996) Cytochrome P450 gene clusters i n Drosophila me lanogaster. Mol. General Genet. 251, 290± 297. 29. Tyshenko, M.G. & Walker, V.K. (1997) Towards a reconciliation of the introns early or late views: triose phosphate isomerase g ene s from insects. Biochim. Biophys. Acta 1353, 131±136. 30. Mouches, C ., Pasteur, N., Berge, J.B., Hyrien, O ., Raymond, M., Saint Vincent, B .R., de Silvestri, M. & Georghiou, G.P. (1986) Ampli®ca tion of an esterase gene is resp on sibl e for insecti- cide resistance in a California Culex mosquito. Science 233, 778±780. 31. Otto, E., Young, J.E. & Maroni, G. (1986) Structure and expression of a tandem duplic ation of the Drosophila metallothi- onein gene. Proc. Natl Acad. Sci. USA 83, 6025±6029. 32. Komma, D.J. & Endow, S.A. (1986) Magni®cation of the r ibo- somal genes in female Drosophila melanogaster. Genetics 114, 859±874. 33. Fletcher, G.L., Hew, C.L., Li, X., Haya, K. & Kao, M.H. (1985) Year-round pr esence of high lev els of plasma antifreeze peptides in a temperate ®sh, ocean pout (Macrozoarces americanus). Can. J. Zool. 63, 488±493. 34. Hooper, J.E., P erez-Alonso, M., Bermingham, J.R., Prout, M., Rocklein, B.A., Wagenbach, M., Edstrom, J.E., de Frutos, R. & Scott, M.P. (1992) Comparative s tudies of Drosophila Antenna- pedia genes. Genetics 132, 453±469. 35. Hatton, A.R., Subramaniam, V. & Lopez, A.J. (1998) Generation of alternative Ultrabithorax iso forms and stepwise removal of a large intron by resplicing at exon-exon junctions. Mol. Cell 2, 787±796. 36. Krawinkel, U., Zoebelein, G. & Bothwell, A.L. (1986) Palindro- mic sequences are associated with sites of DNA b reakage during gene conversion. Nucleic. Acids Res. 14, 3871±3882. 37. Ford, M. & Fried, M. (1986) Large inverted duplications are associated with gene ampli®cation. Cell 45, 425±430. Ó FEBS 2002 Antifreeze protein genes in Choristoneura (Eur. J. Biochem. 269)45 38. Mouches, C., Pauplin, Y., Agar wal, M., L emieux, L., Herzog, M ., Abadon, M., Beyssat-Arnaouty, V., Hyrien, O., Saint Vincent, B.R. & Georghiou, G.P. (1990) Characterization of ampli®cation core and esterase B1 gene responsible for insecticide resistance in Culex. Proc.NatlAcad.Sci.USA87, 2574±2578. 39. Graham, L.A., Walker, V.K. & Davies, P.L. (2000) Develop- mental and environmental regulation of antifreeze proteins in the mealworm beetle Tenebrio molitor. Eur. J. Biochem. 26 7, 6452± 6458. 40. Rinehart, J.P., Yocum, G.D. & Denlinger, D.L. (2000) Develop- mental upregulation of inducible hsp70 transcripts, but not cognate form, during pupal diapause in the ¯ esh ¯y, Sarcophaga crassipalpis. Insect Biochem. Mol. Biol. 30, 515±521. 41. McGugan, B.M. (1954) Needle-mining habits a nd larval instars of the spruce budworm. Can Entomologist 86, 439±454. 42. Edwalds-Gilbert, G., Veraldi, K.L. & Milcarek, C. (1997) Alter- native poly ( A) site selection in complex transcription units: means to an end? Nucleic Acids Res. 25, 2547±2561. 43. Zhao, J., H yman, L. & Moore, C. ( 1999) F ormation of mRNA 3¢ ends in eukaryotes: mechanism, regulation, and interrelationships with other s teps in mRNA syn thesis. Microbiol. Mol. Bio l. Rev. 63, 405±445. 46 D. Doucet et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . it is likely that at least some of the  17 genes of the family are spaced several kb ap art (as fo r Afp-Lu1 and 2 . 7a) , e ven though they may be linked. In. Ser). A puta- tive TATA box was found at )79, but no evidence of a CAAT box could be found. AFP-Iu1 had the same polyadenylation signal, AATATA, found in the