Appendage abnormalities in spiders induced by an alternating temperature protocol in the context of recent advances in molecular spider embryology

Oligomelic postembryos

Molecular embryological research has suggested alterations to normal gene expression that might account for some instances of appendage loss. Parental RNAi (pRNAi) studies, especially in P. tepidariorum, have revealed a range of abnormal phenotypes from knockdown of selected developmental genes (Oda & Akiyama-Oda, 2020), depending on the specific gene suppressed and on the degree of suppression of a given gene within different embryos. These phenotypes can include embryos exhibiting oligomely, though in some instances lethal abnormalities co-occur, indicating that widespread down-regulation of the targeted genes does not account for oligomelic postembryos like those in Figs. 2A–2D. For example, knockdown of the Notch-signaling-pathway component Delta in P. tepidariorum (Pt-Delta) (Oda et al., 2007) or the spider gap gene Pt-Sox21b.1 (Paese et al., 2018; Baudouin-Gonzalez et al., 2021) results in loss of leg-bearing segments, but this is accompanied by loss of all opisthosomal segments. More localized suppression, however, comparable to that achieved by embryonic RNAi (eRNAi) (Oda & Akiyama-Oda, 2020), cannot be ruled out in oligomelic postembryos.

As an aside, conspicuously lethal consequences of gene downregulation, as occur with knockdown of genes such as Pt-Delta and Pt-Sox21b.1, are potentially relevant to the high mortality that was observed in experimental E. atrica embryos. Equally lethal, though less conspicuous, is embryonic development that, superficially, proceeds essentially to completion without obvious defect, but the embryo nevertheless fails to hatch. Embryos like these were among the 40% of the experimental group that did not hatch. It is thus worth noting that fully developed embryos, not exhibiting defects but unable to hatch, were produced with high frequency when three transcription factors, Pt-foxQ2, Pt-six3.1, or Pt-six3.2, were individually suppressed by pRNAi (Schacht, Schomburg & Bucher, 2020). We hasten to emphasize, however, that there may well be many mechanisms by which this inability to hatch is produced, with lethal consequence, including some unrelated to abnormal gene expression. Here and elsewhere in this discussion, in noting similarities between phenotypes obtained by thermal shock and by RNAi of specific genes, the involvement of those genes in producing the thermally-induced abnormalities, while a possibility, is by no means assured and certainly no such definitive claim is intended.

More likely to be involved in appendage losses like those in Fig. 2 are genes that, when knocked down, result in oligomelic embryos able to survive hatching. Examples of two such genes, expressed during early embryogenesis within the period our thermal treatment is applied, are the gap gene hunchback (hb) (Schwager et al., 2009) and Distal-less (Dll), an appendage patterning gene that also plays an earlier gap gene role in spiders (Pechmann et al., 2011). pRNAi of hb in P. tepidariorum (Pt-hb) yielded postembryos missing the L2 leg pair or both L1 and L2 legs (Schwager et al., 2009), while, similarly, pRNAi of Pt-Dll produced postembryos lacking the L1 leg pair or both L1 and L2 legs (Pechmann et al., 2011; Setton et al., 2017; Setton & Sharma, 2018). These losses reflected loss of the segments on which the legs would have developed and it was only segments bearing walking legs that were so affected (Schwager et al., 2009; Pechmann et al., 2011), reflecting the distinction between segmentation of the head region, with its chelicerae and pedipalps, and that of the thorax region, with its four pairs of legs (Kanayama et al., 2011). If this also applies to E. atrica, then abnormal suppression of Ea-hb would not contribute to oligomely involving chelicerae and pedipalps (Figs. 2A, 2B, and 2D), but it could be a factor in spiders with missing legs (Figs. 2C and 2D). The same can be said for Ea-Dll suppression during its early involvement with prosomal segmentation (its gap gene role) (Pechmann et al., 2011). Later suppression of Dll in limb buds (Chen, Piel & Monteiro, 2016), whether preceded by early Dll suppression (pRNAi; Pechmann et al., 2011) or not (eRNAi; Schoppmeier & Damen, 2001; Pechmann et al., 2011), resulted in truncated appendages but not in any additional appendage loss.

There are, however, two confounding considerations where potential abnormal Ea-hb or Ea-Dll expression is concerned: (1) Though Schwager et al. (2009) did note left-right leg reduction asymmetry in Pt-hb pRNAi embryos, leg losses resulting from prosomal segment losses have usually been symmetric (Schwager et al., 2009; Pechmann et al., 2011), whereas the alternating temperature treatment applied in this study has often yielded asymmetric (Figs. 2C and 2D), as well as symmetric (e.g., Jacuński, Templin & Napiórkowska, 2005), leg oligomely. (2) We have not been able to determine which legs specifically have been missing in oligomelic postembryos, even after examining leg neuromeres in histological sections (Jacuński, Templin & Napiórkowska, 2005; Napiórkowska, Napiórkowski & Templin, 2016b), and therefore we do not know if leg losses have been consistent with Ea-hb or early Ea-Dll suppression.

Regarding (1), if, speculatively, Ea-hb or Ea-Dll is inhibited by our alternating temperature protocol (at this point possibilities only), such inhibition might be more localized, random, and asymmetric than that often resulting from pRNAi. Indeed, unilaterally oligomelic E. atrica with corresponding unilateral losses of leg nerves and ganglia indicate that thermally-induced disturbances result in losses of hemisegments more often than of full segments (Jacuński, 1983; Jacuński, Templin & Napiórkowska, 2005; Napiórkowska, Napiórkowski & Templin, 2016b). This is reminiscent of asymmetric prosomal appendage shortening that has been induced in C. salei by knockdown of Cs-Dll using eRNAi (Schoppmeier & Damen, 2001) and of seven-legged postembryos that have occasionally resulted from Pt-Dll pRNAi, indicating loss of a single L1 hemisegment (Setton et al., 2017). Schoppmeier & Damen (2001) noted the median furrow (ventral sulcus) that divides the right and left halves of the embryonic germ band (Foelix, 2011; Wolff & Hilbrant, 2011), and the seemingly independent development of the two halves, as a possible explanation for such asymmetric phenotypes.

Regarding (2), future studies could explore a strategy used by Pechmann et al. (2011) for ascertaining the identity of missing legs: for oligomelic postembryos able to molt successfully to at least 1^st instars, the number and arrangement of slit sense organs on the sternum, compared to control spiders, should help identify the missing legs and provide an alternative to histological sectioning for indicating if symmetric/asymmetric oligomely of legs is accompanied by loss of an entire segment/hemisegment, as previously suggested based on histology (Jacuński, 1983; Jacuński, Templin & Napiórkowska, 2005; Napiórkowska, Napiórkowski & Templin, 2015, 2016b).

We should also note that, unlike RNAi experiments, in which the gene targeted by treatment is known, genes most directly impacted by application of the alternating temperature protocol may be cofactors, upstream regulators, or downstream targets of genes discussed here as being potentially perturbed by the protocol, rather than directly affecting expression of the candidate gene itself. For example, pRNAi of the transcription factor Sp6-9 in P. tepidariorum (Pt-Sp6-9) has been observed to reduce or eliminate Pt-Dll expression (Königsmann et al., 2017; Setton & Sharma, 2018) as well as eliminate expression of the segment polarity gene Pt-engrailed-1 (Pt-en-1) in the L1 and L2 segments (Setton & Sharma, 2018), similar to the effect of Pt-Dll pRNAi on Pt-en-1 expression (Pechmann et al., 2011). Resulting phenotypes included embryos missing these two segments and, so, also the legs that would form on them (Königsmann et al., 2017; Setton & Sharma, 2018). Thus, in this example, defects consistent with inhibited Ea-Dll expression could, hypothetically, arise via thermally-induced direct disruption to a different member of the same gene network, namely Ea-Sp6-9 (Königsmann et al., 2017; Setton & Sharma, 2018). Also, genes most directly affected may vary among embryos depending on, e.g., the exact timing of a temperature switch in relation to an embryo’s stage of development. It is also worth repeating that thermally-induced perturbations to normal gene expression might have abnormal spatial or temporal components in addition to, or rather than, quantitative aberrations.

On first consideration, missing pedipalps, as in Figs. 2B and 2D, could suggest disturbance to the normal expression of the Hox gene labial (lab), specifically the paralog lab-1 (lab-B in Schwager et al., 2017), first expressed at Stage 4 in P. tepidariorum (Pechmann et al., 2015). Its knockdown by pRNAi can result in postembryos lacking pedipalps, though, unlike leg losses that are due to loss of the corresponding prosomal segments, the pedipalpal segment is retained (Pechmann et al., 2015). On the other hand, like the above pRNAi-induced leg losses, pedipalp loss as seen in Pt-lab-1 pRNAi postembryos has been symmetric (Pechmann et al., 2015), whereas the alternating temperature treatment more often results in asymmetric pedipalp oligomely in E. atrica (Figs. 2B and 2D), suggesting a potential localized disruption to Ea-lab-1 expression. However, an abnormal postembryo like that shown in Fig. 2B, in which the site of a missing pedipalp is adjacent to a greatly reduced chelicera (labeled ‘a’), does not support this suggestion if we assume a shared genetic cause for both anomalies (this assumption is by no means certain). This is because expression of lab-1 (or any of the Hox genes) is not involved in specifying chelicera morphology (Pechmann et al., 2010).

An alternative explanation that might encompass both defects has not yet emerged from functional studies in spiders. The gene dachshund-2 is expressed proximally in both chelicerae and pedipalps, but the only noted phenotypic consequences of its knockdown by pRNAi in P. tepidariorum are malformed patellae in the walking legs (Turetzek et al., 2015). Two paralogs of extradenticle (exd-1, exd-2) and homothorax-1 (hth-1) are also expressed proximally in pedipalps and chelicerae (Prpic & Damen, 2004; Pechmann & Prpic, 2009; Turetzek et al., 2017), but exd has not been the subject of RNAi experiments in spiders, or any chelicerates (Nolan, Santibáñez-López & Sharma, 2020), and among chelicerates hth function has only been examined by eRNAi in the harvestman Phalangium opilio Linnaeus, 1758 (Sharma et al., 2015). However, studies in insects and spiders indicate that exd-1 and hth-1 of spiders are functionally linked (Hth-1 required for translocation of Exd-1 into the nucleus), such that knockdown of either gene would likely produce similar, though not identical, phenotypes (Sharma et al., 2015; Turetzek et al., 2017; references therein). Phenotypes resulting from knockdown of the single-copy hth in P. opilio (Po-hth) included homeotic transformations of chelicerae and pedipalps to leg identities, appendage truncation, and fusions between chelicerae and pedipalps, though, importantly, apparently not pedipalp oligomely (the results do, however, state “The labrum and/or some appendages also failed to form” (among Class I phenotype embryos) (Sharma et al., 2015) without elaboration). Interestingly, like the aforementioned defect asymmetry observed in Cs-Dll eRNAi C. salei embryos (Schoppmeier & Damen, 2001), a high incidence of asymmetric defects was also obtained with Po-hth eRNAi P. opilio embryos (Sharma et al., 2015). As mentioned, asymmetric defects are likewise often obtained by the alternating temperature protocol. These three examples demonstrate that aberrations on one side of the germ band do not necessarily affect the other side (Schoppmeier & Damen, 2001) and suggest limited, non-global perturbations to gene expression or other developmental processes.

Postembryos with schistomely or in ‘Other abnormalities’ category

Appendage development relies on differentiation along proximal-distal (P-D), dorsal-ventral (D-V), and anterior-posterior (A-P) axes, the last especially little studied in spiders. Genes involved with establishing these axes may be susceptible to thermally-induced abnormal expression, resulting in limb malformations. For example, a key player in establishing the D-V axis is the gene FoxB, encoding a forkhead box transcription factor that is ventrally expressed within appendages (Heingård, 2017; Heingård et al., 2019). Its knockdown in P. tepidariorum by pRNAi resulted in greatly reduced hatching success and altered expression of downstream genes that normally show ventral (wingless (Pt-wg/Wnt1), Pt-H15-2), dorsal (optomotor-blind (Pt-omb)), and distal (decapentaplegic (Pt-dpp)) expression within appendages, resulting in ‘dorsalized’ legs and pedipalps (Heingård, 2017; Heingård et al., 2019). Such Pt-FoxB pRNAi embryos that were able to hatch successfully and progress to the 1^st stadium exhibited distally crooked legs and pedipalps, comparable to some postembryos included in our ‘Other abnormalities’ category (Table 1). This category also included postembryos with significantly shortened appendages, a phenotype that has also been observed in mildly affected Pt-Sp6-9 pRNAi embryos and postembryos, and has included asymmetric defects (Königsmann et al., 2017; Setton & Sharma, 2018).

Appendage bifurcation, i.e., schistomely (Figs. 2E and 2F), in postembryos might also be considered in terms of erroneous expression of genes modeling the appendage axes, with schistomely representing distal duplication of the P-D axis (Cotoras, de Castanheira & Sharma, 2021). Though functional data (e.g., RNAi) are lacking in chelicerates (Cotoras, de Castanheira & Sharma, 2021), expression data in P. tepidariorum for dpp (Akiyama-Oda & Oda, 2003) and wg/Wnt1 (Janssen et al., 2010), among other evidence from spiders and other arthropods (Pechmann et al., 2010), have been consistent with dpp and wg/Wnt1 expression early in spider appendage development initiating a gene cascade that generates the P-D axis (Prpic et al., 2003). In legs and pedipalps, three distinct domains of expression establish the P-D axis via expression of Dll distally, dachshund-1 (dac-1) medially, and exd-1/hth-1 proximally (Prpic & Damen, 2004; Pechmann et al., 2010). Disturbances in the normal expression of dpp, wg/Wnt1, or their downstream targets caused by thermal shocks may result in a duplication of the P-D axis. In a report of cheliceral schistomely in the spider Tetragnatha versicolor Walckenaer, 1841, Cotoras, de Castanheira & Sharma (2021) hypothesized that the defect could be replicated by introducing ectopic Dpp and Wg/Wnt1. The schistomely shown in Fig. 2E, at the distal end of a leg, suggests perturbations that included direct or indirect abnormality in Dll expression while the more proximal schistomely indicated in Fig. 2F, on a noticeably wider appendage than the normal legs, potentially represents abnormal expression of dpp, wg/Wnt1, and dac-1 (among other possibilities), the latter’s expression coincident with the trochanter and femur (Abzhanov & Kaufman, 2000; Prpic et al., 2003; Prpic & Damen, 2004).

Postembryos exhibiting pedicel polymely

Arguably the most interesting cases from the perspective of evolutionary/developmental biology involve two individuals with an appendage on the pedicel (first segment of the opisthosoma, O1; in spiders, coincident with somite VII) that are presented in Figs. 3A and 3B. Appendages do not usually form on the O1 segment in spiders and such defects are rare even among E. atrica subjected to alternating temperatures as embryos. Within this segment, the principal Hox genes expressed are the two paralogs of Antennapedia (Antp) (Damen et al., 1998; Khadjeh et al., 2012; Schwager et al., 2017). Knockdown of Antp-1 in P. tepidariorum (Pt-Antp-1) by pRNAi has demonstrated that it is responsible for repressing the development of legs on the O1 segment (Khadjeh et al., 2012). At its most severe, this down-regulation of Pt-Antp-1 resulted in sufficient de-repression of leg development in O1 that 10 walking legs formed; the usual eight plus a pair on the pedicel that were like the former morphologically and in lateral placement except a little shorter and thinner (Khadjeh et al., 2012; replicated by Setton & Sharma, 2018). Expression of the genes that establish the P-D axis in legs (Pt-exd-1, Pt-hth-1, Pt-dac-1, Pt-Dll) was nearly identical between the ectopic O1 legs and normal L1–L4 legs. Moreover, expression of the Hox genes Deformed-A (Pt-Dfd-A) and Sex combs reduced-B (Pt-Scr-B; paralogs as designated in Schwager et al., 2017) within the 10 legs indicated that the ectopic legs on O1 were not homeotic copies of any of the normal walking legs, but they were instead true O1 segment de-repressed legs (Khadjeh et al., 2012).

It is of interest that Khadjeh et al. (2012) obtained not only severely affected postembryos with a pair of complete legs on the pedicel following knockdown of Pt-Antp-1, but in more moderately affected individuals they observed only short leg-like projections on the pedicel. Further, in a triple pRNAi experiment (to suppress Pt-Antp-1 and two other Hox genes), they obtained two postembryos with an incomplete appendage on just one side of the pedicel. They attributed this asymmetric (“mosaic”) phenotype to the lesser quantity of each dsRNA that could be injected when attempting to inhibit three genes simultaneously, resulting in less effective suppression of Pt-Antp-1. This range of outcomes is again reminiscent of the results obtained when alternating temperatures are applied to embryos of E. atrica, where appendages may form on the pedicel symmetrically or only on one side (Fig. 3), and these appendages may exhibit little or considerable development, from a short, unsegmented projection to a segmented, essentially complete leg (Jacuński, 1971, 1984; Jacuński & Templin, 1991; this study). This suggests that the alternating temperature protocol has the potential to disturb, to varying extent, normal expression of Ea-Antp-1 or associated up- or downstream genes in the O1 segment.

There is a long history of embryological observations on spiders that indicates an ancestry in which appendages were present on somite VII (e.g., Korschelt & Heider, 1890; Jaworowski, 1896; Janeck, 1909; Yoshikura, 1954; Yoshikura, 1955; Wolff & Hilbrant, 2011). Principally, this is indicated by a small, short-lived protuberance or patch, sometimes explicitly interpreted as an incipient limb bud, appearing on each O1 hemisegment when the opisthosomal limb buds develop. These transient O1 limb buds apparently do not form in all spider taxa (Dawydoff, 1949), however, as they have not been noted in some detailed embryological studies (Montgomery, 1909; Holm, 1940; Rempel, 1957; Mittmann & Wolff, 2012; Pechmann, 2020). It is notable that putative limb buds on O1 have been observed in Heptathela (Yoshikura, 1954; Yoshikura, 1955), a member of the basal Mesothelae, as well as in several members of the derived araneomorph RTA clade, to which E. atrica belongs (Wheeler et al., 2017).

Considering that small, transitory protrusions (potential appendages) may appear on the pedicel (O1) segment in embryonic spiders, and that by use of targeted gene suppression (pRNAi) it is possible to obtain appendages on the pedicel with the structure of walking legs that nevertheless have their own O1 identity (Khadjeh et al., 2012), it might be worth reconsidering whether somite VII, the pedicel, is indeed the first segment of the opisthosoma, as it is usually described, rather than the last segment of the prosoma. This thought is stimulated by another result obtained by Khadjeh et al. (2012); that limb repression also occurs as a normal part of development in the O2 segment (somite VIII), but when the genes that redundantly promote this repression (Pt-Antp-1, Ultrabithorax-1 (Pt-Ubx-1)) are suppressed by double pRNAi, the ectopic appendages that form on O2 appear far more vestigial than the legs induced to form on O1. This may reflect less effective overall de-repression in O2 because of the repression redundancy present in O2, not shared by O1, but it could also conceivably reflect an early euchelicerate ancestry in which appendages on somites VII and VIII differed substantially in morphology, with those on VII more limb-like and those on VIII more plate-like, suggestive of a border between tagmata. Such a difference in appendage morphology has been interpreted for the Devonian euchelicerate Weinbergina and is also seen in extant Xiphosurida (horseshoe crabs) (Dunlop & Lamsdell, 2017).

Applying Lamsdell’s (2013:4) definition of a tagma as “…a distinct and discrete morphological region that comprises a series of equivalently modified appendages that constitute a unit of specific form…or sometimes function…”, the traditional view of the O1 segment as part of the spider opisthosoma seems appropriate. Both the normally legless condition of the pedicel and the maneuverability it imparts to the rest of the opisthosoma (Dunlop & Lamsdell, 2017) suggest a form and function more in keeping with those of the opisthosoma. In addition, during spider embryogenesis, the germ band initially divides into the prosomal segments and a posterior ‘segment addition zone’ (SAZ) from which the opisthosomal segments, including O1, subsequently derive in anterior-to-posterior sequence (Schwager et al., 2015). These differing paths to segmentation in the two tagmata also favor an opisthosomal identity for the O1 segment.

On the other hand, Lamsdell (2013) and Dunlop & Lamsdell (2017) acknowledge that establishing borders between tagmata can be difficult because the ends of a tagma and their associated appendages may differ substantially from the rest of the tagma. The border between prosoma and opisthosoma, with somite VII’s questionable affiliation, is given as a prime problematic example (Dunlop & Lamsdell, 2017). They review evidence from fossil and extant chelicerates that supports a chelicerate groundplan in which somite VII is prosomal, as suggested by Stürmer & Bergström (1981). This possibility is further supported by the potential for appendages with leg-like morphology to develop on the spider pedicel, whether induced by application of pRNAi or alternating temperatures, and, along with transitory limb bud formation on the O1 segment in some spiders, suggests loss of somite VII appendages present in basal euchelicerate ancestors of arachnids (Dunlop & Lamsdell, 2017). Thus, an interpretation of atavism for appendages developing on the pedicel in teratological spiders (Jacuński, 1971, 1984; Jacuński & Templin, 1991) remains valid. Also noteworthy is the observation that, in some chelicerates, walking leg segments (all or just L4), as well as the opisthosomal segments, are derived from the SAZ and, in one known instance (a mite), O1 segmentation precedes that of L4 (reviewed in Schwager et al., 2015). Thus, it seems the mechanism of segmentation during embryonic development does not necessarily provide a reliable means for assigning segments to tagmata in a way that agrees with morphological/functional regions.

Summary and future directions

By applying alternating temperatures during early spider embryogenesis, we obtained high embryo mortality, changes in number, size, and shape of appendages or their podomeres, and formation of appendages on the pedicel; a body segment (O1 = somite VII) on which appendages are not normally found in spiders. Thus, by using appropriate methods, abnormalities can be induced that potentially reflect certain ancestral traits present in basal (eu)chelicerates, including possibly atavistic appendages on segment O1. This type of developmental abnormality has a bearing on the question of the tagma to which somite VII belongs, prosoma or opisthosoma, with implications tied to chelicerate phylogeny.

Based on recent research on genes that determine the formation of segments and appendages, we suspect that at least some of the observed developmental defects arising from our alternating temperature protocol are the result of blocked or otherwise aberrant expression of relevant genes, including Hox genes. Atypical expression may potentially include spatial and temporal, as well as quantitative, deviations from normal. Though the possible involvement of specific genes as discussed above is speculative, it is one step toward the goal of testing hypotheses that attribute specific anomaly types to disturbances affecting specific genes. For example, by identifying hb as a candidate gene that may have its expression distorted by the alternating temperature protocol, potentially resulting in oligomely (as discussed above), the expression of hb over time may be compared between experimental and control embryos to ascertain if the former exhibit notable deviations in expression (e.g., asymmetric expression) compared to the latter.

Modified versions of the alternating temperature protocol can also be investigated that intentionally attempt to disrupt expression of a specific gene and/or increase defect frequency; for example, by narrowing the window of treatment and exploring the application of an abrupt temperature switch at specific times relative to the height of expression for a given gene and given site(s) within embryos. This could lead to the establishment of a protocol that is able to induce certain types of anomalies with greater regularity, reducing numbers of embryos that would need to be screened for defects.