The scape organ and the cone- and peg-shaped sensilla
The antennal scape organ was first mentioned by Verhoeff [13], but Fuhrmann’s [10] description in Scutigera coleoptrata gave a first overview on the external morphology and basic insights into its anatomy (compare Additional file 1A). It was described in a variety of scutigeromorph species: S. coleoptrata [10,11,12,13], Seychellonema gerlachi ([20], their Fig. 7B), Parascutigera peluda [18], and in several other species covering all three families (see character matrices in [16, 19, 34]). Thus, the scape organ is a groundplan feature of Scutigeromorpha [15, 35]. In S. coleoptrata, Fuhrmann [10] described the oval opening with a dimension of 15–20 µm with a depth of 30 µm, and counted about 20 cone-shaped sensilla inside the cavity. Sombke et al. [11] depicted only two of them in a pseudomaturus (14–16 mm body length, not sexually mature [1]). In this study, we found 7 to 15 cone-shaped sensilla in various adult specimens. As centipedes regularly molt throughout their adulthood [1, 36] these quantitative variations may be correlated with age and, thus, the size of the scape organ. Fuhrmann [10] detected receptor cells contributing to the cone-shaped sensilla (“Sinneszellen”, see Additional file 1D), the distal parts of the outer dendritic segments immediately below the cones (“Terminalstrang der Sinnesorgane”), and the stretched interstitial epidermal cells forming collars around the sensilla. However, there was no report yet on the three fields of peg-shaped sensilla flanking the scape organ. As they are minute and lack typical traits of sensillum shafts, they are hardly detectable (compare Fig. 1B in [11]). The discovery of peg-shaped sensilla at the base of the antennae of S. coleoptrata once again indicates that it is necessary to backup scanning electron microscopic data with transmission electron microscopy. As with this case, TEM studies should be considered mandatory not only to support the SEM-based definition of sensillum types but also to fully disclose the structural diversity of arthropod sensilla and epidermal exocrine glands (for latter aspect see also [37]).
Cone- and peg-sensilla are devoid of a socket. The shaft is smooth in cone-shaped sensilla (similar to larger antennal sensory cones, compare [11]) and ribbed in peg-shaped sensilla (similar to small antennal sensory cones, compare [11]). The terminal pores (= tip-pore) of both types are concealed by secretion. The recto-canal epidermal glands surrounding the scape organ secrete an extremely electron dense, presumably hygroscopic layer covering the tips of cone-shaped sensilla. These glands represent a three-cell variant of recto-canal epidermal glands described by Müller et al. [38, 39]. In peg-shaped sensilla, the terminal pores are clogged by a polymorphic substance. Although the origin of this secretion is still uncertain, it seems reasonable that it is produced by epidermal glands present around the sensillar fields at the scape. The fine structural organization only gradually differs in cone- and peg-shaped sensilla. Both exhibit a similar bending of the sensory apparatus (outer dendritic segments, sensillum lymph space, and dendritic sheath) that is surrounded by a voluminous extracellular space (Fig. 9B, F, I). Receptor cells are always biciliated and enwrapped by three sheath cells. The first (proximal) sheath cell includes an elaborated system of microtubules and cytoplasmic fibers embedded in a highly electron-dense structure that resembles a scolopale (Fig. 9H, I), which is more elaborated in cone-shaped sensilla. In both types of sensilla, the short outer dendritic segments of receptor cell 2 (rc2) are connected to the first (proximal) sheath cell with the scolopale-like structure. Differences are the presence of two cells with long outer dendritic segments (rc1) in cone-shaped sensilla vs. only one in peg-shaped sensilla (compare Figs. 4H and 9E, H), the extraordinary thickness of the dendritic sheath below the cuticle in peg-shaped sensilla (Figs. 8B, 9G, I), and a much lower electron-density of the inner receptor lymph (compare Figs. 5D, E, and 8F, H). Concerning their function, a detection of shaft deflection can be excluded. Likewise, a chemoreceptive (olfactory or gustatory) function as proposed by Fuhrmann [10] is unlikely. Since receptor cells with long outer dendritic segments (rc1) terminate in structures strongly resembling tubular bodies abutting a plug pin of secretion, primary mechanisms of mechanical receptor stimulation known from thermo- and hygroreceptive sensilla of insects likely come into consideration. According to the spatial coherence of specific ultrastructures and comparisons with similarly organized ciliary systems in sensilla of arthropods others than centipedes, we propose that in cone- and peg-shaped sensilla of S. coleoptrata receptor cell(s) with long outer dendritic segments (rc1) most likely function as hygroreceptor(s). Moreover, the receptor cell with short outer dendritic segments (rc2), along with its interconnection to a scolopale-like structure in the first (proximal) sheath cell, represents a component found in scolopidia of crustaceans and insects, potentially suited for vibration or strain detection.
Hygroreceptive component—receptor cells with long dendritic outer segments (rc1)
The possibility that receptor cells with long outer dendritic segments (rc1) are involved in hygroreception is based on three structural characters identified by Yokohari [30, 40] in Periplaneta americana. (1) They are associated with short, cone- or peg-shaped cuticular shafts with inflexible sockets and no wall pores (a tip-pore may occur) that are similar to hygroreceptors of some hexapods and arachnids (e.g. ‘no-pore sensilla’ in Lepidoptera and Coleoptera [41] and ‘tip-pore sensilla’ in Cupiennius salei (Trechaleidae) [42]). (2) They possess paired, but unbranched outer dendritic segments. (3) Tubular bodies are present at the distal ends of the outer dendritic segments connecting to the inner face of the tip cuticle (see summary in [29]). Functional similarities to cone- and peg-shaped sensilla in S. coleoptrata concern the mechanisms of both mechano-electric transduction (mechanical deformation of the dendritic tip) and impulse generation (induction at the level of receptor cell somata). In hexapods, the tip cuticle (without a terminal pore, hence ‘np-sensilla’) serves as a connector for the tubular bodies, as has been observed in the sensillum capitulum of P. americana [30, 43,44,45], in the sensillum coelocapitulum of Apis mellifera [46], and in the sensilla coeloconica of Locusta migratoria [47] or Carausius morosus [31]. In blunt-tipped peg sensilla on larval antennae of Tenebrio molitor, two tubular bodies are nested in a terminal network of the dendritic sheath, which is subjacent to a central canal connected to a tip-pore [48]. Similar to T. molitor and C. salei, in cone- and peg-shaped sensilla of S. coleoptrata there is a terminal tip-pore, but it is completely or partly plugged by an electron-dense substance, which most likely represents secretion released by recto-canal epidermal glands (compare [38]). This secretion may be hygroscopic and expand if the animal enters moist areas. It is likely that, if moistened, the secretion plug may elongate and distort the subjacent tubular bodies, thus transducing a mechanical stimulus on these receptor cells. Thus, the hygroscopic material, which is most likely produced by the associated epidermal glands, may drive this hygro-mechano transformation. The deformation of the tubular body-like structures, as it is here assumed to be achieved by the swelling of overlaying secretion, distinctly differs from hexapod hygroreceptors where the transformation is achieved by deformation of the cuticle [29].
The possible triggering of hygroreceptors utilizing secretion plug systems seems to be a new finding in arthropod sensilla research. Regrettably, the literature is ambiguous with respect to the spatial coherence and functional role of so-called molting pores in hexapod hygroreceptive sensilla. In their review on invertebrate hygroreceptors, Altner and Prillinger [23] did not regard the molting pore as crucial for stimulus transmission. Reassessing drawings and TEM micrographs provided by other authors, led us to the conclusion that the molting pore is plugged by an electron-dense material that makes connection to the inner sensillum lymph space [27, 29, 31, 41, 49]. Biochemical properties aside, this system of hygroscopic material being in close contact to the tubular bodies, as revealed in Hexapoda, may be generally comparable to that in S. coleoptrata. In hexapods, swelling or shrinking of hygroscopic sensillum structures result in a deformation/distortion of and voltage changes across the tubular bodies of (type-1) receptor cells (compare [41, 50] and citations therein). In the hexapod model, the hygroscopic transducer is the cuticle, in S. coleoptrata it is secretion. Both are a functional morphological prerequisite of hygroreceptors operating according to the mechanical hygrometer model as defined by Tichy and Loftus [29]. Thus, we propose to add secretion plugging the tip-pore as a potential transducer. In the hygroreceptive tip-pore sensilla of the wandering spider Cupiennius salei, the terminal pore is not plugged, but humidity-induced changes are recorded by electrolyte concentrations in the receptor lymph (electrochemical hygrometer model) [29, 42, 51].
The presence of biciliated hygroreceptor cells (as in cone- and peg-shaped sensilla in S. coleoptrata) is uncommon in hexapods (so far only found in Collembola [28]), but has been found in some myriapod species associated with mechano- and probably hygroreception [5, 32]. Possessing two cilia instead of a single one may result in an increased sensitivity of a given hygroreceptor cell (compare [30]). However, a combination of hygroreceptive with thermoreceptive elements, present in many hexapods as ‘triads’ [23, 27, 29, 30, 52, 53], is absent in S. coleoptrata. In myriapods, hygroreception is not restricted to the antennal base. In another scutigeromorph species, Thereuonema tuberculata, humidity and mainly CO2 reception was electrophysiologically recorded as a function of the organ of Tömösváry (= postantennal organ), which is located between the compound eye and the antennal base on both sides of the head [54, 55]. There, dendritic outer segments of biciliated receptor cells branch distally and terminate close to a thin cuticle; tubular bodies are absent [55]. A similar organization of postantennal organs (also in association with epidermal glands) was described for Lithobius forficatus [56, 57] and for millipedes and symphylans [58, 59]. Here, receptor cells are likewise biciliated, but stimulus transduction is different from the one proposed for cone- and peg-shaped sensilla in S. coleoptrata.
Scolopidial component—receptor cells with short dendritic outer segments (rc2)
Typical features are a single biciliated receptor cell with short dendritic outer segments transforming distally into tubular bodies. They nest in cavities of the proximal region of the dendritic sheath that is connected to a scolopale-like structure in the first (proximal) sheath cell. The presence of tubular bodies indicates a mechanical pathway of stimulus transduction while the scolopale-like structure reveals close structural and functional resemblance with sensilla containing scolopidial elements. In Arthropoda, there are internal and external mechanoreceptors (see reviews [26, 60,61,62]). In terrestrial arthropods, external mechanoreceptors are either uni- or bimodal hair-like (trichoid) sensilla (including campaniform sensilla). The dendritic outer segments of their receptor cells attach to the proximal end of the sensillum shaft via a typical tubular body [26]. In crustaceans, external mechanoreceptors are exclusively equipped with a scolopale [63,64,65]. Internal mechanoreceptors described in Pancrustacea have been assigned to the class of scolopidia (reviews [26, 62, 66,67,68]). If aggregated, the scolopidia act as functional units in multimodular mechanoreceptive sense organs, usually termed chordotonal organs [62]. Hexapod and crustacean scolopidia as well as crustacean mechanoreceptive (scolopidial) hair sensilla are characterized by: (1) one to four bipolar (type-1) receptor cells, (2) one or several attachment cell(s) at the distal end, (3) one or several proximal glial cell(s) surrounding the proximal region of receptor cells, and (4) a peculiar structure, called the scolopale, which is located within the innermost sheath cell, called the scolopale cell [61,62,63, 67,68,69]. The scolopale consists of an electron-dense matrix of aggregated proteins traversed by actin filaments and longitudinally arranged microtubules. In hexapod scolopidia, the scolopale may extend and diverge into a cylindrical complex of distinct scolopale rods strengthening distal finger-like processes of the scolopale cell (e.g. [70]). The scolopale cell surrounds the inner sensillum lymph space and the inner dendritic segments of the receptor cells, which house prominent ciliary rootlets in their dendritic inner segments [26, 67]. The attachment cell(s) of hexapod and crustacean scolopidia encompass proximally an extracellular structure, the so-called (scolopale) cap or tube [67, 69]. The typology of hexapod and crustacean scolopidia is rather complex and refers to various ultrastructural features. These include proportions of the dendritic outer segments, without (type-1) or with (type-2) distal dilation. Additionally, it refers to number and appearance of receptor cells, or the extracellular structure associated with dendritic outer segment(s) and the scolopale cell. In the latter, dendritic tips can be firmly inserted in a scolopale cap (mononematic) or just surrounded by an electron-dense tube (amphinematic)) [69]. The scolopidial component of cone- and peg-shaped sensilla of S. coleoptrata shares some ultrastructural traits with mononematic, type-1 scolopidia: (1) the position of the scolopale cell (first sheath cell surrounding the ciliary region), (2) the scolopale-like structure that at least in part shows a rod-like packing, (3) short dendritic outer segments with a slightly swollen terminal structure that (4) are firmly anchored into an extracellular structure that is the dendritic sheath.
Interestingly, in hexapods and crustaceans, the scolopale cap has never been consistently termed as a dendritic sheath, but should be identified as such because it is not in contact with the surface cuticle, it lines the inner sensillum lymph space, and is most likely secreted by the scolopale cell (e.g. [67, 71, 72]). Following the typical sequence of sheath cells in cuticular sensilla of arthropods, the attachment cell would be homologous to the second (trichogen) sheath cell as it overlays the scolopale cap and contains microtubules in high abundances (e.g. [67, 69, 73]). However, some ultrastructural details such as local dilations in the dendritic outer segment(s) below the cap level, or apical processes of the scolopale cells containing scolopale rods and piercing the attachment cell, are unique characters of crustacean and hexapod scolopidia (e.g. [67, 69]). The presence of two cilia as well as the presence of axial/perpendicular microtubules are two unique features of the receptor cells in cone- and peg-shaped sensilla of S. coleoptrata.
Affinities of cone- and peg-shaped sensilla among arthropods
The tip-pore sensilla (antennal cone- and peg-shaped sensilla associated with the scape organ of Scutigera coleoptrata) described here represent a new type of bifunctional arthropod sensilla that combine hygroreceptive and scolopidial components associated with long (rc1) and short (rc2) dendritic outer segments. Similar cone- and peg-shaped sensilla in S. coleoptrata occur in moderate numbers on the antennal nodes, which were interpreted as proprioceptors because of their specific location [11]. Peg-shaped sensilla are also present at the base of locomotory legs in S. coleoptrata (Sombke et al. pers. obs.), which makes it likely that similar sensory structures in general could be involved in possible hygro- and pressure-reception at the base of appendages. It is unclear whether secretions also act as general transducers, but the epidermis of centipedes is generally rich in solitary epidermal glands [1, 37,38,39, 74]. Hence, the clogging of terminal pores of peg-shaped sensilla would be potentially provided by those glands.
Similar arthropod sensilla equipped with short cones, inflexible sockets and scolopale-like structures in the first sheath cell have been recorded only in combined hygro- and thermoreceptors of some hexapods that are summarized as ‘no-pore sensilla with inflexible socket’ [27, 30, 73, 75, 76]. They include two types of receptor cells: type-1 receptor cells with long dendritic outer segments (referred to as hygroreceptors) and type-2 receptor cells with short dendritic outer segments. The dendritic tips of the type-2 receptor cells often exhibit a strongly infolded or lamellated apical membrane, a pattern which is considered suitable for thermoreception (e.g. [28, 31]). However, lamellated dendritic tips are absent in cone- and peg-shaped sensilla of S. coleoptrata. The cytoskeletal elements of the scolopale equivalent in hexapod np-sensilla are thought to stabilize the inner (thecogen) sheath cell, which consequently protects the outer dendritic segments against mechanical forces [47, 73]. Frequently, a type-3 receptor cell is present with a short and thin dendritic outer segment that exhibits an undifferentiated cilium [28]. In the larval blunt-tipped peg sensillum of Tenebrio molitor this cilium terminates near a scolopale-like structure close to the dendritic sheath [70], which is strikingly similar to the receptor cell with short cilia (rc2) in cone- and peg-shaped sensilla of S. coleoptrata. In addition, blunt-tipped peg sensilla in T. molitor bear a terminal (molting) pore that is distinctly constricted and plugged with an electron-dense material [70]. Thus, these sensilla can be considered functional equivalents (except for thermoreception) to cone- and peg-shaped sensilla in S. coleoptrata. Admittedly, as hygroreceptive sensilla must have evolved after the independent conquest of land by myriapods and hexapods [77, 78], these systems are very likely convergent transformations of an already exsiting sensillar type. Scolopidia (sensilla with scolopidial elements) are discussed as a possible ancestral type of sensilla in Pancrustacea (e.g. [63, 68, 79, 80]). However, it is not clear whether scolopidia of crustaceans and hexapods evolved independently or may share a common evolutionary origin with bifunctional sensilla containing scolopodial components as described here for S. coleoptrata. At least mononematic scolopidia may arguably represent an ancestral mechanoreceptor system in Mandibulata that contained externally visible sensilla with scolopidial components. As outlined above, the scolopale cap may be homologized with the dendritic sheath and would be the result of transformation and compression. Within myriapods, penicillate millipedes (Polyxenus lagurus) possess bifunctional hygro- and thermoreceptive np-sensilla, termed sensilla coeloconicum and setiform sensillum S’ [32, 81, 82]. However, both types differ from the cone- and peg-shaped sensilla of S. coleoptrata by the presence of lamellated dendritic outer segments (indicative of thermoreception). Nevertheless, both types of sensilla possess similar biciliated receptor cells with short outer dendritic segments. The short outer dendritic segments in the sensillum coeloconicum of P. lagurus lack tubular bodies and get in close contact with the dendritic sheath near the socket of the sensilla. However, in the setiform sensillum S’, the tips of two cilia of one receptor cell terminate in tubular bodies tightly encompassed by the dendritic sheath and apical processes of the inner sheath cell. There, apical processes are strengthened by microtubules and were therefore termed as ‘ciliary-like structures’ [32, 81]. These cytoskeletal structures are probably equivalent to the ‘scolopale rod-like structures’ described in the sensillum capitulum of Periplaneta americana [43]. Rilling [83, 84] mentioned scolopidia-like sense organs in the mandible and maxilla of the centipede Lithobius forficatus, but based on drawings of methylene blue stained sections, a clear identification as scolopidia is not possible. Nonetheless, the presence of scolopidia-like components was described before (at least) in Diplopoda. Consequently, we propose them to belong to the groundplan of Myriapoda if not even of Mandibulata. The latter assumption would stand in line with reassessments of other anatomical characters, which were previously thought to be present in Pancrustacea alone [85,86,87,88].