A major issue in current biology is the evolution of the metazoan body plan. Widespread developmental processes, such as the establishment of the antero-posterior polarity or the formation of the mesodermal layer, are conducted by highly conserved groups of regulatory genes, whereas morphological variations and novelties often result from divergences (duplication, mutation) in gene sequences and/or in regulatory networks (Force et al.1999; Gibert2002). These networks often involve many homeobox genes acting as transcription factors regulating gene expression during developmental patterning or cell differentiation. Studying how they act and which feature they control has become a powerful way of understanding organism complexity. Among those genes,engrailed is one of the most relevant example demonstrating the evolvability and plasticity of gene function during evolution. This transcription factor was first shown inDrosophila to be a key gene in the establishment of segment polarity (Kornberg1981; Fjose et al.1985), in neurogenesis (Patel et al.1989), and in appendage development (Raftery et al.1991). Highly conserved in protostomes and deuterostomes,engrailed orthologues show similar roles in other arthropods (Patel et al.1989; Abzhanov and Kaufman2000), in annelids (Wedeen and Weisblat1991; Seaver and Kaneshige2006), in echinoderms (Lowe and Wray1997; Byrne et al.2005; Yaguchi et al.2006), and in chordates (Joyner1996; Holland et al.1997). Extensive comparisons among taxa suggest that neurogenesis is likely the ancestral function ofengrailed and that subsequent recruitments have increasedengrailed contributions (Patel et al.1989; Gibert2002). In molluscs, however, there is no strong data for the involvement ofengrailed in neural development. Instead,engrailed is expressed in cells at the margin of the future shell (protoconch) in a wide range of molluscs: in a bivalve (Transenella tantilla, Jacobs et al.2000), in gastropods (Ilyanassa obsoleta, Moshel et al.1998;Patella vulgata, Nederbragt et al.2002), in a scaphopod (Antalis entalis, Wanninger and Haszprunar2001), and in a polyplacophoran (Lepidochitona caverna, Jacobs et al.2000). Fromengrailed role during shell development in molluscs, Nederbragt et al. (2002) proposed that its ancestral function is the formation of a compartment boundary as an alternative to the neurogenic hypothesis.

Among these studies on molluscs, a major group is lacking that could bring a larger and more complete overview of the molluscan taxa and of the lophotrochozoans in general. The cephalopods have not yet been investigated forengrailed, although they exhibit numerous derived traits among molluscs. As cephalopods undergo a direct development without a trochophore larval stage, investigatingengrailed in a cephalopod provides a direct insight at the role of this gene in definitive organogenesis. In cephalopods, the typical molluscan foot is modified into eight to ten prehensile appendages (arms and tentacles) and a funnel. Both this funnel and a muscular mantle are implicated in jet propulsion. High capacities of cognition and reaction are permitted by a compact brain, by a peripheral nervous system allowing rapid nervous transmission (giant axons, stellate ganglia), and by sensory structures, such as complex camerular eyes (Boletzky1988). Whereas one of the molluscan characteristic is an external shell secreted by the mantle, coleoid cephalopods, comprising all extant cephalopods except nautiluses, possess an internal shell, which may be regressed and sometimes absent.

Two paralogues ofengrailed have been found inNautilus pompilius, a cephalopod with an external shell (Wray et al.1995), whereas noen cognate was identified yet in the squidLoligo, a coleoid whose shell is internal and not calcified (chitinous gladius). As a consequence, Wanninger and Haszprunar (2001) have correlated the presence ofengrailed with that of an external shell. In this paper, we present the first report of anengrailed gene expression pattern in a coleoid cephalopod, the cuttlefishSepia officinalis. We show that it is expressed in the shell-forming cells in early stages of organogenesis. This supports the role ofengrailed in molluscan shell formation to organisms with an internal shell. Moreover, we show thatengrailed is also expressed in cephalopod-specific organs: the funnel and arms, the optic vesicle, and eye lids. We, here, identified anengrailed homologous in aLoligo species that adds further lines of evidence for a widespread recruitment ofengrailed in morphological novelties and diversification of metazoan body plan.

Collection ofSepia embryos

During spring and summer (April to September), fertilized eggs were laid by captiveS. officinalis females in the biological stations of Luc-sur-mer (France) and Banyuls-sur-mer (France). From egg batches, individual eggs were detached and embryos were taken out by removing the numerous surrounding envelopes using forceps in sea water. Then, embryos were visually staged using Lemaire’s (1970) system forS. officinalis. As we focused on organogenesis, embryos at stages 16 to 25 were selected for immunochemistry. After chorion removal, embryos were fixed for 1 h in 4% paraformaldehyde (PFA) at room temperature, washed in phosphate-buffered saline (PBS), gradually dehydrated in methanol, and stored at −20°C. Until stage 16, the chorion and embryo are in intimate contact, and an additional fixation for 1 h in PFA 4% was performed to strengthen the chorion and reduce rupture hazard.

Cloning ofengrailed gene homeodomain

Genomic DNA ofS. officinalis was extracted from adult brain using the DNeasy Tissue kit (Qiagen, Valencia, CA, USA). mRNAs ofLoligo vulgaris (collected in Caen, France) were extracted from stage 30 embryos using Tri Reagent (MRC, Cincinnati, OH, USA), then converted into cDNA by the Omniscript reverse-transcriptase (Qiagen). Initial amplification primers for polymerase chain reaction (PCR; Eng-2: 5′-GACAAGCGRCCDMGVACVGCNTT-3′: KPPRTAF; Eng-3: 5′-ATCAAGCTTWTTYTKRAACCANAYYTTNAYYTG-3′: QIKIWFQN) were used to amplify a 106-bp homeodomain fragment inS. officinalis. Primers Eng-2 and Eng-4 (5′-TGRTTRRTANARNCCYTGNGCCAT-3′:MAQGLYN) were used to amplify a 189-bp homeodomain fragment inL. vulgaris. PCR conditions were: 95°C for 5 min + (95°C for 1 min; 45°C for 1 min; 72°C for 1 min) for five cycles + (95°C for 1 min; 50°C for 1 min; 72°C for 1.5 min) for 30 cycles + 72°C for 10 min. PCR products were cloned into TOPO vector (Invitrogen, Carlsbad, CA, USA) sequenced by Genome Express (Meylan, France) and analyzed with the BioEdit software (Ibis Therapeutics, Carlsbad, CA, USA) and GenBank BLASTn (BLAST, basic local alignment search tool).

Whole-mount immunochemistry

Theengrailed protein (En) was detected by using a monoclonal antibody Mab4D9 (Developmental Studies Hybridoma Bank, University of Iowa, USA) raised against a portion of theDrosophila En as primary antibody (Patel et al.1989). It has been shown to selectively bind anengrailed protein in some molluscs: a gastropod and a chiton (Moshel et al.1998; Jacobs et al.2000), and we, therefore, assumed that it detected an engrailed-like protein inSepia. Embryos were incubated in hydrogen peroxide (3% in pure methanol) for 1 h to inactivate endogenous peroxidases; they were preincubated in blocking solution (PBS 1× + bovine serum albumin 1% + Triton X-100 0.1%) for 1 h at room temperature and incubated with the Mab4D9 (1:200) in blocking solution overnight at 4°C (without Ab in controls). After five washes of 30 min in polybutylene terephthalate (PBT; PBS 1× + Triton X-100 0.1%), embryos were preincubated again for 1 h at room temperature in blocking solution and incubated with the 1:500 biotin-conjugated universal secondary antibody (ABC Kit, Vector Laboratories, Burlingame, CA, USA) overnight at 4°C. After four washes of 30 min each in PBT, embryos were incubated with streptavidin-conjugated horseradish peroxidase (ABC kit) for 30 min, rinsed three times for 10 min in PBT, and MAb4D9-binding sites were finally revealed using 3,3-diaminobenzedine (DAB-nickel kit, SK-4100, Vector Laboratories Burlingame, CA, USA), a colored substrate of peroxidase (dark brown). After 10–15 min, the coloration process was stopped in PBT, and embryos were fixed with PFA 4%–PBT.

By PCRs on genomic DNA ofS. officinalis, we detected a singleengrailed gene (Fig. 1, accession number: AM114934), and we were able to similarly confirm the presence of twoen paralogues inN. pompilius (accession numbers: AM114935, AM114936). Using cDNA ofL. vulgaris, we identified the first engrailed cognate in a squid (accession number: AM422130). As expected, the En homeodomain is highly conserved among cephalopods, coleoid sequences being more similar between each other than with those ofNautilus (nautiloid; Fig. 1). For immunostaining, we used the 4D9 antibody, raised against theDrosophila Engrailed protein but able to recognize many other Engrailed homologues depending on the residue present at position 40 of the homeodomain (Patel et al.1989). En proteins with G or S are detected, whereas some residues, such as R, T or Q, are known to prevent 4D9 binding (Patel et al.1989; Wedeen and Weisblat1991). TheS. officinalis andL. vulgaris homeodomains are alone to exhibit a cysteine at position 40 (Fig. 1, arrow). It cannot be excluded that Mab4d9 recognizes anengrailed-like protein or a cognate ofengrailed, not yet sequenced inSepia. Recent molecular studies, however, showed that this antibody recognizes an engrailed protein in molluscs (Moshel et al.1998; Jacobs et al.2000). The staining we obtained inSepia, especially in the shell sac (see below), corresponds to that found in other molluscs, which suggests that 4D9 is able to detect the cysteine at position 40 or, at least, anengrailed-like protein.

Predicted amino-acid sequence deduced from the fragment ofengrailed homeodomain inS. officinalis (PCR primers areunderlined). Additional partial sequences of engrailed homeodomain are shown in other cephalopods, other molluscs, and inDrosophila. The residue at position 40 (arrow head) is crucial for the 4D9 antibody-specific recognition of Engrailed. Accession numbers:S. officinalis AM114934;L. vulgaris: AM422130;E. scolopes AF181095;N. pompilius A AM114935;N. pompilius B AM114936 ;Lepidochitona caverna U21675;P. vulgata AF440097;I. obsoleta A :U23432 B :U23433;Cadulus fusiformis CFU23153;Dentalium eboreum U23154;Placopecten magellanicus U23213;Transennella tantilla U23212;Crassostrea virginica U23214;D. melanogaster M10017

Full size image

S. officinalis develops directly with no larval intermediate and no metamorphosis. Organogenesis proceeds during 2–3 weeks, from stage 16 to hatching at stage 30, eventually resulting in adult anatomy (Boletzky et al.2006). Zygote cleavage gives rise to a disk-shaped embryo at the animal pole of the egg, whereas the vegetal pole is made of a thin layer of ‘extra-embryonic’ ectoderm cells that covers the yolk. After a disk-shaped phase where prospective organs start delineating (Fig. 2, first column), the embryo expands as all organs gain volume (Fig. 2, second column). Yet, the oral (or anterior) pole of the future adult lies at the periphery of the embryo (arm crown and mouth), whereas the future aboral (or posterior end of the adult) pole is central (mantle, gills, funnel). The final adult arrangement is reached at stage 21, where the whole embryo straightens: Eyes, mouth, and the arm crown are then located at the yolk side and the visceropallium (visceral mass, palleal cavity, and surrounding mantle) at the opposite side (Fig. 2, third column).

S. officinalis organogenesis summed up into three major phases visually based on the embryo shape.a–c Apical views of the animal pole.d–f Lateral views of the animal pole (d right side,e andf prospective ventral side). First column (a,d,g): following cleavage and gastrulation, the embryo remains disk-shaped until stage 19. Second column (b,e,h): from stage 20 to 22, the body parts expand, yet in a two-dimensional pattern. Third column (c,f,i): from stage 21, the embryo straightens up and acquires the adult orientation, most visible organs being in place and almost achieved.a1,a2,a3,a4, anda5: arms 1 to 5;ssa shell sac aperture;e: eye;f fin;fp funnel pouch;ft funnel tube;g gill;ma mantle;me mantle edge;mo mouth;sse shell sac edge;ssa shell sac aperture;st statocyste;y yolk. Scale bar: 500 μm. Orientation (relating to adult animal in the so-called physiological orientation):L left;R right;V ventral;D dorsal;AO aboral;O oral

Full size image

In all molluscs, the shell is an ectodermal product of the mantle. It holds true for the internal shell of cephalopods, in which the embryonic mantle invaginates and delineates a circular inner ‘shell sac’ (Boletzky et al.2006). InS. officinalis, as this cavity grows in size during stages 17 to 19, its aperture at the mantle surface decreases in size and becomes a small pore at stage 20 (Fig. 2h). Near the closure point, fins develop from stage 20. The enclosed ‘shell sac’ starts producing both the organic matrix of the shell and a periostracum at stage 21 and the calcite mineralization starts at stage 24 (Spiess1972). In this central area of the embryo (i.e., the aboral pole), the En protein was first detected at stage 16 in both the mantle and the whole shell sac area, at the same time, as these two structures appeared (Fig. 3a). At stage 17, En immunostaining was restricted to both the shell sac and mantle edges (Fig. 3b). At stage 18, 4D9-positive cells were only found at the posterior, prospective ventral edge of the mantle (Fig. 3c). Staining then vanished and was no more detected at stage 19.

Results of 4D9 immunostaining in the aboral pole of the embryo, showing both the developing mantle and shell.a At stage 16, En is located in the shell area (before invagination), in the mantle and in both funnel tube and pouch;b at stage 17, En is located at the edge of the shell, at the edge of the mantle and in both funnel tube and pouch;c at stage 18, engrailed is located at the edge of the mantle in both funnel tube and pouch;c’ a focus on the funnel pouch showstwo parallel lines of 4D9-positive cells (black arrows) included in a larger band of 4D9-positive cells (double arrow). Scale bar: 100 μm. Orientation and abbreviations: cf. Fig. 2

Full size image

Embryonic arms, supposed to be foot-derived structures, emerge as a peripheral crown at stage 15 (Fig. 2a,g). As the whole embryo expands and straightens upward above the yolk, arms become regularly arranged around the mouth at the oral end of the animal (Fig. 2f,i). They first appear as small buds and then grow as cylinders with suckers located on their oral surface starting from stage 22. Suckers are present all along the arms, except for tentacles (derived arms 4) showing suckers at their extremities only (club). From stages 17 to 19, En immunostaining sequentially appeared in arm buds (Fig. 4e). At stage 17, the prospective dorsal arms (1 and 2) started producing Engrailed (Fig. 4a,b), soon followed by arms 4 from stage 18 (Fig. 4c). At late stages 18 and 19, only arms 2 and 4 expressed En with high intensity, whereas arms 3 and 5 were less strongly colored (Fig. 4d). From stages 19 to 23, no expression was detected; then, a thin and linear En staining briefly appeared along the aboral side of arms 4 at stage 24 (Fig. 4f), which get restricted to the distal future ‘club’ at stage 25 (Fig. 4g).

Arm development inS. officinalis from stages 17 to 25 and En location revealed by 4D9 immunostaining.a,b Stage 17, Engrailed is located in arm 1 (not shown) and arm 4 (squared ina andenlarged inb);c early stage 18 where arms 1, 2, and 4 are stained ;d late stage 18, arms start growing as cylinders;e synopsis table of engrailed location in arms 1 to 5 from stage 16 to stage 25. Staining in the whole arm bud is indicated by acircle filled in black or gray (depending on the intensity), whereas discrete staining is indicated by aline;f stage 24, arms 4 exhibit En within an aboral line along the proximo-distal axis;g stage 25, theengrailed expression becomes restricted to the distal part of arm 4, where suckers are present (dotted line). Scale bar: 100 μm. Orientation and abbreviations: cf. Fig. 2

Full size image

Also deriving from the molluscan foot, the adult funnel is an unpaired organ made of two structures, a tube and a pouch, developing from distinct areas in the embryo. The funnel tube arises from two ventral bands that early separates from the arm crown at stage 16, close to arms 4 and 5 (Fig. 2, first column). At stage 20, they join as a cylinder that eventually closes at stage 22 (Fig. 2, second column). The funnel pouch, essentially providing the retracting muscles, develops from two narrow bands along the lateral mantle edges (Fig. 2, first column) that later connect the funnel tube (Fig. 2, second column). At stages 16 and 17, both tube and pouch areas showengrailed-expressing cells (Fig. 3a,b). The most noticeable change occurred at stage 18 when suddenly two intensely stained parallel lines included within a larger band of 4D9-positive cells appeared in the pouch area (Fig. 3c,c’). Subsequently, the funnel pouch differentiates and appears as two ‘walls’ running along the mantle edge (Fig. 2g,h) andengrailed stopped being expressed.

In cephalopods, successive ectodermal folds lead to the formation of the eye. At stage 16, the invagination of an ectodermic thickening in the cephalic area yields the optic vesicle, the edge of the remaining pore is sutured at stage 19 (Fig. 5a). The inner hemisphere of this vesicle, which eventually differentiates as the retina (Lemaire and Richard1978), showed 4D9-reactive cells from stage 20. The En protein was first detected as one spot of contiguous cells (stage 20), then as two anterior and posterior spots (stage 21, Fig. 5c), and later as an equatorial ring parallel to the head surface (Fig. 5d,e). At stage 21, the outer hemisphere develops the lens, soon encircled by the iris, a second ectodermal fold. A third and last fold arises at stages 25–26 yielding the lid. At stage 24, the head mass exhibits ridges, called ‘arm pillars’, somewhat prolonging arms 2, 4, and 5. From stages 25 to 26, the ‘arm pillars’ 2 (prospective dorsal face) and 4 (prospective ventral face) expand laterally toward the eyes, enclose them, and form the upper and the lower elements of the lids, respectively. During this process, 4D9-reactive cells were detected as rows at the lateral margin of arm pillars 2, 4 (Fig. 5d,e), and 5 (Fig. 5f), and staining persisted until they joined and formed the two lids at late stage 25 (Fig. 5g,h). Finally, the ventral lid produces a thin secondary cornea that fuses with the dorsal lid and isolates the lens from external environment. This constitutes the last developmental process of the so-called myopsid eyes, present in some Decabrachia, includingS. officinalis andL. vulgaris.

Eye development inS. officinalis from stages 17 to 25 and Engrailed location revealed by 4D9 immunostaining.a scheme of the eye placode invagination at stage 17;b stage 20, a large spot of 4D9-positive cells can be detected on the floor of the freshly closed optic vesicle (empty arrow head);c stage 21, Engrailed is located in a thick ring within the optic vesicle (empty arrow heads);d,e stage 24, En-positive cells are present in a thin ring within the eye (empty arrow heads) and at the lateral margin of arm pillars 4 and 5 (black arrow heads);f ventral view of thed individual showing En at the lateral margin of the arm pillar 2 (black arrow head);g,h stage 25, arms pillars 2 and 4, respectively, form the ventral and dorsal lid folds, at the margin of which En-positive cells are present (black arrow heads);i diagram of the eye elements at stage 25.a1,a2,a3,a4,a5: arms 1 to 5;ap arm pillar;dl dorsal lid fold;i iris;ov optic vesicle;r retina;le lens;vl ventral lid fold. Scale bar: 100 μm

Full size image

Engrailed and compartment boundaries

Theengrailed expression patterns that we detected in the developing shell sac are consistent with observations made in previous studies on molluscs (Moshel et al.1998; Jacobs et al.2000; Wanninger and Haszprunar2001; Nederbragt et al.2002). As En was no longer detected in the shell sac when shell formation started (at around stage 21), we provide additional evidence that En in molluscs is not required for the shell formation (i.e., the skeletogenesis, Jacobs et al.2000) but rather for delimiting the shell compartment boundary, as proposed by Nederbragt et al. (2002). Thisengrailed role, molecularly based on cell signaling together with the secreted Dpp protein, is similar to that observed in the formation of the antero-posterior boundary in arthropod parasegments (Gritzan et al.1999) and of the dorso-ventral boundary in vertebrate limbs development (Hidalgo1998). Our study contrasts with previously studied molluscs in thatS. officinalis has no trochophore larval stage and no external shell. Therefore, we show that the role ofengrailed also concerns adult shell development and can be extended to those molluscs whose shell sac invaginates leading to an internal shell.

The detection of En at the mantle edge in the first stages of organogenesis is a novel data among molluscs. Given that the mantle edge and the shell edge are two distinct boundaries in cephalopods,engrailed could have been co-opted inS. officinalis to set up the future mantle edge boundary in addition to the shell edge boundary. This would add a novel example of boundary delimitation triggered by the action of En. However, as the mantle edge secretes the shell edge in molluscs with an external shell, these two boundaries overlap, and it cannot be excluded that a ‘mantle delimitation’ function also exists in previously studied molluscs, not distinguishable from the ‘shell delimitation’ function.

Engrailed and cephalopod-derived structures

Another striking result is thatengrailed turned out to be expressed in structures specific to cephalopods, such as eye lids and the funnel and arms, derived from the foot, suggesting a recruitment ofengrailed for the development of major morphological novelties representing cephalopod synapomorphies.

The linear expression patterns observed in both the funnel and lid suggest thatengrailed has been recruited for the delineation of novel compartment boundaries, similar to that found in mantle and shell. The delineation of two parallel lines of 4D9-positive cells predates the emergence of the funnel pouch that stands from stage 19 as a ‘wall’ at the surface of the embryo. Each line could correspond to an ectodermal boundary that later separates the funnel pouch from the mantle at one side and from the cephalic region at the other side. Thisengrailed expression stopped as the funnel pouch thickening visually appeared (stage 19). This suggests thatengrailed is required for the establishment of these boundaries, but not for their maintenance. Along the arm pillars 4 and 2, lines of 4D9-positive cells predate the migration and the formation of the dorsal and ventral elements of the lids. Once again, this supports thatengrailed has been co-opted to delineate the boundary of a novel structure.

We observed two distinct periods of expression of engrailed in the arms: sequential expression in arm buds during the first period (stages 17 to 19) and a linear expression in arms 4 during the second period (stages 24 to 25). The sequential expression ofengrailed in the five pairs of arms reminds one of the unexpected expression patterns of Hox genes found inEuprymna scolopes (Lee et al.2003). Besides their conserved role in setting up the antero-posterior axis, seven Hox genes were shown to be sequentially expressed in arms of this sepiolid species. InS. officinalis, future tentacles are the only arms that express En at stages 24 and 25, in a linear pattern at their aboral side. This line of expression decreased in length and became restricted to the tentacular clubs, where suckers are exclusively located. As the other arms possess suckers all over their oral side, this differential expression ofengrailed along the proximo-distal axis of arms 4 may be somehow connected to the tentacle-specific distribution of suckers. Further studies are thus required to (1) investigate whetherengrailed is involved in the differentiation of tentacles vs arms and (2) show at the histological level whetherengrailed expression is located at nervous structures or not.

Engrailed and neurogenesis

Immunostaining with Mab4D9 did not revealed any expression ofengrailed in the nervous system. In annelids,engrailed is expressed in the peripheral neurons of leech embryos (Wedeen and Weisblat1991) and in various nerve cells (bilateral pairs and apical tuft sensorial cells) of theChaetopterus trochophore larva (Seaver et al.2001). In molluscs,engrailed expression has only been found in sensorial cells near the apical tuft in thePatella trochophore larva (Nederbragt et al.2002) and in the ladder-like nervous system ofLepidochitona (Jacobs et al.2000). It has been assumed that loss of anengrailed role in neurogenesis predates the Cambrian radiation of molluscs, including cephalopods (Gibert2002). We thus postulate thatengrailed has not been recruited either for ganglia condensation, which characterizes cephalopods within molluscs, or for the innovations of their peripheral nervous system (e.g., the stellate ganglia). Alternatively,engrailed might be involved at earlier stages of neurogenesis.

Successive detections of the En protein in the optic vesicle (Fig. 5b–e) suggest a potential involvement ofengrailed in the development of the camerular eye ofS. officinalis. Interestingly, the Engrailed protein has been shown to possess a paracrine activity and to be involved in guiding and linking the retinal axons to the optic tectum of vertebrate brains (Friedman and O’Leary1996; Logan et al.1996; Brunet et al.2005). This result provides additional evidence about the fascinating convergence between eyes of cephalopods and vertebrates (Harris1997). A putative co-factor could be thePax6 gene required in eye development in a large variety of organisms including cephalopods (Quiring et al.1994; Tomarev et al.1997; Hartmann et al.2003; Gehring2005). Interestingly, the pattern of 4D9-reactive cells we observed in the eyeS. officinalis is somewhat similar to that ofPax6 observed inL. opalescens (Tomarev et al.1997) andE. scolopes embryos (Hartmann et al.2003). As in chicken, Engrailed down-regulates the expression ofPax6 (Plaza et al.1997);Pax6 might also be a target gene of Engrailed in the regulatory pathway of eye development in cephalopods. Further studies are currently conducted to assess this question. and aS. officinalis fragment ofPax6 has been recently sequenced (accession number: AM422131).

Engrailed and evolutionary issues

Assuming thatengrailed in molluscs possesses a unique function (skeletogenesis) in a unique structure (the shell), Wanninger and Haszprunar (2001) suggested that the lack of a mineralized shell inLoligo (squid) was a valuable explanation for a lack of anengrailed cognate inLoligo species (Wray et al.1995). Our identification of the firstengrailed cognate in a squid and the previous characterization ofengrailed inE. scolopes (with a rudimentary chitinous gladius) definitely obliterate this assumption. Besides, numerous studies now provide evidence for a widespread evolvability and plasticity ofengrailed, which casts doubt on any reasoning based on the absence of a single trait. InS. officinalis,engrailed is expressed during organogenesis of derived structures shared by cuttlefishes and squids: the shell sac before shell formation, the funnel, the ten arms and the four lid elements. As a consequence, the presence ofengrailed in cephalopods is not linked to the presence of a calcified shell, and theengrailed expression pattern inL. vulgaris should be similar to that observed here inS. officinalis.

From an evolutionary perspective, we think that the loss of the external shell, which provides a physical protection against predators, could not occur without any earlier adaptations, anatomical, neural, or behavioral, allowing a better protection. The evolutionary success and diversification of the cephalopod group might result from awin–win evolutionary mechanism: Reduction in the external shell drawbacks (weight and volume) may have facilitated the expansion of both mantle and foot and their recruitment for new functions, such as prey capture or escape, which consequently decreased predation pressure and thus allowed further shell reduction. Recruitment ofengrailed in the development of novel structures as we detected in the present paper may have played an important role in this evolutionary scenario.