Morphology of amphipod silk glands compared with other crustacean glands
Secretory cell nucleus quantitynext section
Dyopedos bispinis D1 and D2 secretory cells are uninucleate (Fig. 6A), as are the silk glands of the majority of other amphipod species (Jassa falcata, Jassa ocia (Bate, 1862), Crassicorophium crassicorne (Bruzelius, 1859), Ericthonius punctatus (Bate, 1857), and Ampithoe ramondi Audouin, 1826) (Nebeski, 1880)). However, in Ampithoe rubricata glands, binuclear secretory cells have also been observed together with uninucleate cells (Neretin, 2016). The basis glands in the pereopods of Crassicorophium bonellii are also binuclear according to the drawing by Kronenberger et al. (2012b).
Binuclearity is typically regarded as a prominent feature of lobed glands (Talbot and Demers, 1993) and is occasionally used as a diagnostic character (Kakui and Hiruta, 2014). However, the secretory cells in lobed glands contain one or two nucleuses, apparently independent of cell size (Gorvett, 1951). Furthermore, it was recently shown that binuclear cells, together with uninuclear cells, occur in isopod rosette glands (Vittori et al., 2012). Thus, variations in nuclear number are present in different types of crustacean tegumental glands but are currently only described within Percarida (Isopoda, Tanaidacea, Amphipoda) and not in Decapoda. It is likely that similar variations also occur in Mystacocarida (Elofsson and Hessler, 2005).
In Dyopedos bispinis, the proximal silk glands (D1) are covered with a lining layer (Fig. 6C, E and Fig. 7). We cannot unambiguously determine whether the lining is a part of the duct cell or an independent structure. Thus, we can only hypothesize as to its origin.
The presence of nuclei favours the independence of the lining cell from other gland cells (duct and secretory). The cellular layer covering the glands, which is classified as connective tissue, has been described in many crustaceans: decapod (Yong, 1932; Pugh, 1962) and isopod (Ide, 1891; Gorvett, 1946) rosette glands, cirripedian cement glands (Lacombe and Liguory, 1969), amphipod Ampithoe rubricata silk glands (Neretin, 2016) and isopod lobed glands (Ide, 1891; Gorvett, 1951), but the fallacy of this interpretation for lobed glands was observed in a study by Weirich and Ziegler (1997).
The lining could also represent a system of branching extensions of the duct (or intermediate) cell, and such systems have been described for intermediate cells in lobed (tricellular, Weirich and Ziegler, 1997) and rosette (Vittori et al., 2012) glands of isopods. In Dyopedos bispinis, this hypothesis is supported by the fact that, in some places, the lining comprises numerous cytoplasmic extensions (arrowheads, Fig. 7E) that deeply penetrate secretory cell invaginations (i, Fig. 7A-D), as in lobed glands. However, branches of intermediate cells in the isopod glands serve to collect secretions on the basal surface of secretory cells (Weirich and Ziegler, 1997). In contrast, the D1 gland cell secretion is excreted through the apical cell membrane, as evidenced by the presence of a definitely visible accumulation site (as, Fig. 6).
The lining might be considered an enlarged sheath cell, thus sharing a common origin with the duct cell. Sheath cells are typically observed in arthropod sensilla and at some stages of insect dermal gland development (Quennedey, 1998; Merritt, 2006).
The lining might also be a complex structure; for example, nervous terminations are likely located in this region as previously assumed for isopod lobed glands (Wägele, 1992). To understand the nature of the lining, more detailed electron microscopy investigations are required.
The observation of an axon-like structure (Fig. 6F) in contact with the secretory cell likely indicates the presence of nervous control of amphipod silk glands. Gland innervation has been demonstrated in the branchial rosette glands of the decapod Palaemonetes pugio Holthuis, 1949 (Doughtie and Rao, 1982), in the labrum glands of the cladoceran Daphnia obtusa Kurz, 1874 emend Scourfield, 1942 (Zeni and Zaffagnini, 1988), in cirripedian cypris larva glands (Okano et al., 1996), and in mystacocarid tegumental glands (Elofsson and Hessler, 2005). However, crustacean tegumental glands are not typically innervated (Talbot and Demers, 1993). For amphipod silk glands, innervation was shown for the first time in the present study.
Cellular composition of Dyopedos bispinis proximal amphipod silk glands
Each proximal gland (D1) contains some (up to 17) secretory cells and one duct cell with an intracellular duct. Insect glands contain duct cells with an intercellular duct, which are referred to as class 3 glands (Noirot and Quennedey, 1974) or dermal glands, in contrast with unicellular and tubular glands (Merritt, 2006). In contrast to typical insect class 3 glands (Noirot and Quennedey, 1974; Quennedy, 1998), the secretory cells of the D1 glands do not have microvilli in the ductule lumen. D1 duct cells also do not have a mesaxon, which is the cross connection between the inner and outer duct cell membranes and has been described in many insect and crustacean dermal glands (Lai-Fook, 1970; Martens, 1979; Lawrence and Staddon, 1975; Doughtie and Rao, 1982; Elofsson and Hessler, 1998; Lombardo et al., 2006; Vittori et al., 2012).
Dermal glands are common in Arthropoda and contain one to two consecutive duct cells and one or several secretory cells (Quennedy, 1998; Coons and Alberti, 1999; Hilken et al., 2005; Pekár and Šobotník, 2007, Müller et al., 2014). The dermal glands of the Crustacea contain one to two secretory cells, which are referred to as bicellular (Fig. 9A) or tricellular (Fig. 9B) depending on the quantity of duct cells (Talbot and Demers, 1993, Elofsson and Hessler, 1998; Zeni and Stagni, 2000). In Malacostraca, rosette dermal glands (Fig. 9C) are also widely distributed, and each are composed of two duct cells (in the narrow sense, duct and intermediate) and multiple (up to 40-50) secretory cells (Talbot and Demers, 1993).
The proximal silk glands (D1) of Dyopedos bispinis (Fig. 6 and Fig. 9F) contain some secretory cells and appear similar to rosette glands, but D1 glands, at least at first glance, have two significant differences. First, we have observed only one (not two) duct cell in each gland. Glands comprising single duct cells and several secretory cells have been described in Crustacea (Claus, 1879; Ide, 1891; Gorvett, 1946; Pugh, 1962), but there are no electron microscopy studies confirming this “bicellular” scheme. Thus, we propose that the boundary between the duct and intermedium cell might be difficult to detect on semi-thin sections and might not be captured on ultrathin sections due to the significant length of the duct (up to 2 mm).
Second, D1 glands markedly differ in shape from typical rosette glands. Rosette glands are typically globular; ductules from secretory cells coalesce at approximately one point (Johnson and Talbot, 1987; Alexander, 1989; Talbot and Demers, 1993, Fig. 9C). In contrast, D1 glands are strongly elongated, and ducts from secretory cells fall individually and sequentially into the main duct. However, occasionally, rosette glands might also be slightly elongated (Ide, 1891; Pugh, 1962) or even have practically tubular forms (recently described terrestrial hermit crab Coenobita spp. antennal glands, Tuchina et al., 2014). Dyopedos bispinis D1 glands even have more elongated, almost cord-like forms (Fig. 5B, 9F), and such gland morphology has not been incorporated into the Talbot and Demers (1993) classification of crustacean tegument glands. If typical rosette glands are similar to the typical acinar glands of metazoans (Ide, 1891), then Dyopedos bispinis D1 glands resemble rather tubular glands. We propose that such glands can be called pseudotubular.
Comparison with other arthropod silk glands
Glands similar in shape to D1 have been described within silk-producing systems in other corophiid amphipod species (Fig. 9D-E) (Nebeski, 1880; Neretin, 2016) and in the tanaid Heterotanais oerstedii (Krøyer, 1842) (Fig. 9G-H) (Blanc, 1884), all of which might be defined as pseudotubular. Building glands in other crustaceans suggest another structure; the amphipods Crassicorophium bonellii and Lembos websteri and the tanaid Phoxokalliapseudes tomiokaensis (Shiino, 1966) have lobed and typical rosette glands (Kronenberger et al., 2012b, Kakui and Hiruta, 2014), while callianassid shrimps (Decapoda) have only typical rosette glands (Dworschak, 1998). Cirripedian cement glands are multicellular but do not possess special duct cells (Lacombe and Liguori, 1969) and might be considered tubular gland variations (class 1 glands, according to the Noirot and Quennedey classification, 1974).
Silkworms and spiders, the most famous silk-producers, have classical tubular glands (Sehnal and Akai, 1990), but there is some evidence of homology between spider ampullate silk glands and sensilla (Hilbrant and Damen, 2015). In insects, dermal glands are homologous to sensilla (Quennedey, 1998; Merritt, 2006), and it can be assumed that spider silk glands passed through the dermal gland stage during evolutionary development. Futhermore, there is an assumption, that insect labial silk glands originate from dermal glands, but it is questionable (Kenchington, 1969; Sutherland et al., 2010).
Dermal (class 3) silk glands are common among insects (Sutherland et al., 2010); these glands are bicellular or tricellular (according to “crustacean” terminology) and typically numerous. However, multicellular dermal silk glands are unknown in insects (although secretory cells can be multinucleated, Nagashima et al., 1991). Thus, the pseudotubular silk glands of amphipods and, likely, tanaids are probably unique among arthropod silk glands as these structures are tubular in shape, although indeed dermal.
Pseudotubular gland origin
We propose that the pseudotubular glands of Dyopedos bispinis could originate from increasing secretion in conjunction with space limitations in narrow appendages. Multicellular gland enlargement could occur at the expense of elongation and growth towards the proximal part of the appendage. In the hermit crab, Coenobita spp., elongated antennal glands also probably appeared as a result of a great increase in secretion production volume (Tuchina et al., 2014).
Coenobita spp. antennal glands likely appear as modified rosette glands (Tuchina et al., 2014). Dyopedos bispinis D1 glands could also have evolved from typical rosette glands, particularly as these glands have been observed in amphipods (Schmitz, 1967, 1992).