Patterns of Autonomous and Conditional Neural Specification for the Brain and Ventral Nerve Cord in Annelida

ABSTRACT

Evolution of central nervous systems is a long‐debated topic. Similar mechanisms of conditional neural specification linked to dorsal‐ventral (D–V) axis formation have been used to support homology across several taxa. We tested for autonomous versus conditional neural specification in two distantly related annelids, Capitella teleta and Platynereis dumerilii, using blastomere isolations and ablations. Our results demonstrate autonomous specification of anterior neural tissue in isolated first‐quartet (1q) micromeres for both annelids. In both annelids, partial larvae derived from isolation of the somatoblast, micromere 2d, also resulted in the formation of trunk neural tissue (ventral nerve cord or VNC), suggesting that this cell has intrinsic neural fate. However, at least in C. teleta, we found evidence that additional external proneural and anti‐neural signals affect specification of trunk neural fate in 2d. Ablation of the vegetal macromeres at the 16‐cell stage (leaving 1q + 2q) resulted in partial larvae that failed to form trunk neural tissue although they did have anterior neural tissue and a D–V axis, suggesting that micromeres other than 2d may repress the intrinsic neural fate of 2d. Conversely, ablation of only three of four vegetal macromeres at the 16‐cell stage (e.g., leaving 1 macromere with 1q + 2q) “rescues” trunk neural fate, suggesting the presence of a vegetal proneural signal in all four quadrants. Taken together, these results imply that VNC fate is specified by a complex interplay of signaling that is decoupled from D–V axis formation in C. teleta. Our results suggest that the distinct neural specification mechanisms between the head and trunk may be conserved in Annelida and that annelid CNS specification may operate independently of D–V axis formation. These findings will broaden our understanding of how neural specification mechanisms evolved and diversified in Spiralia in comparison to other bilaterians.

Diagrams of blastomere isolations in Capitella teleta and Platynereis dumerilii showing autonomous head neural tissue (from 1q) and trunk neural tissue (from 2d) as well as a decoupling of D–V axis formation from trunk neural tissue in annelids.

Summary

• Isolating blastomeres of two annelids, Capitella teleta and Platynereis dumerilii showed:

• Complex patterns of neural specification for the brain and ventral nerve cord.

• Separate mechanisms of neural specification in the head and trunk.

• Neural specification is not coupled to D–V axis in the trunk.

1. Introduction

The central nervous system (CNS) is a key bilaterian innovation whose origins and intricacies are still debated. Until recently, most studies of the nervous system have sensibly focused on a few key lineages. With the advent of modern molecular tools, it has become feasible to widen our questions and investigations into less‐well‐known systems (Martín‐Durán ²⁰²³). Here we explore the developmental regulation of neural tissue in annelids by isolating and ablating blastomeres during early cleavage stages.

Across Bilateria, many different neural subtypes in varied circuit arrangements enable the full breadth of behaviors and physiological functions. The large diversity of nervous system arrangements and functions offer intriguing opportunities to compare the molecular and cellular mechanisms controlling their formation across animals (Martín‐Durán and Hejnol ²⁰²¹). CNS development is best understood in deuterostomes (e.g., vertebrates and hemichordates) and ecdysozoans (e.g., nematodes and arthropods). In vertebrates and flies, cell‐extrinsic factors involved in dorsal‐ventral (D–V) axis formation, including Bone Morphogenetic Protein (BMP) signaling and activation of the MAPK cascade, are responsible for the correct development of the CNS (Holley et al. 1995; Kuroda et al. ²⁰⁰⁴; Mieko Mizutani and Bier ²⁰⁰⁸; Wilson and Edlund ²⁰⁰¹). This has led to an assumption that D–V axis and neural specification are mechanistically linked across Metazoa. Less is known about how nervous systems are specified in the third major bilaterian clade, Spiralia (~Lophotrochozoa, e.g., annelids and mollusks) (Dunn et al. 2014; Martín‐Durán and Hejnol ²⁰²¹).

There are striking differences in how the CNS develops in spiralians. Recent studies suggest that neural specification in Spiralia, at least in the episphere (the larval tissue anterior to the prototroch—the anterior ciliary band—hereafter termed the “head”), may not require cell‐extrinsic factors. Instead, blastomere isolation data from the sedentary annelid Capitella teleta Blake et al. ²⁰⁰⁹ suggest that specification of anterior neural tissue, that is, the brain, can occur without external signals (Carrillo‐Baltodano and Meyer ²⁰¹⁷). Autonomous neural specification has only been reported in one other bilaterian, the ascidian Halocynthia roretzi (Minokawa et al. 2001), in which part of the nerve cord but not brain is autonomously specified. This is notably different to how neural fate is specified in vertebrates and insects. This raises questions about how the developmental mechanisms differ between the head and trunk CNS across Bilateria and how autonomous vs. conditional control of CNS development evolved. Furthermore, insights into divergent mechanisms of CNS specification across taxa can provide an additional line of evidence for the homologous vs. convergent evolution of the CNS in Bilateria.

Spiralians have a stereotypical form of embryonic cleavage called spiral cleavage that is thought to be ancestral for the group (Hejnol 2010). The invariant nature of spiral cleavage allows comparative analyses of fates, body plans, and molecular mechanisms controlling development (Freeman and Lundelius 1992; Henry ²⁰⁰²). Individual blastomeres can be identified and tracked in animals with unequal spiral cleavage; at the 4‐cell stage, the blastomeres are named A, B, C, and D, and each gives rise to a different embryonic quadrant. These four macromeres divide along the animal‐vegetal axis to generate the first tier of smaller cells on the animal side: the first‐quartet micromeres (1a, 1b, 1c, 1 d) or collectively 1q, while the cells on the vegetal side are first‐quartet macromeres (1A, 1B, 1C, 1D) or 1Q, (Figure 1A). The macromeres continue dividing asymmetrically to give rise to subsequent tiers of animal micromeres, e.g., 2q, 3q, and 4q, until the onset of gastrulation (Conklin 1897; Wilson ¹⁸⁹²). The stereotypical nature of spiral cleavage allows us to trace cellular lineages and hypothesize the specific cells responsible for tissues and organs (Figure 1C). We can then isolate specific cells or groups of cells to determine their role in downstream development and identify which cells are necessary or sufficient to produce specific structures such as the brain and ventral nerve cord (VNC).

Figure 1.

Fatemap. (A) Micromere identity at the stages where blastomere isolations were performed. (B) Ventral larval diagrams of stage 6 Capitella teleta and 72 h Platynereis dumerilii. (C) Simplified fate map of C. teleta modified from Meyer et al. (2010). Colors correspond roughly to germ layer: Violet: head ectoderm (1q); green: foregut (2a, 2b, 2c); purple: trunk ectoderm (2d); blue: mesoderm (3q); yellow: endoderm (3Q). pg, pygidium; PT, prototroch; TT, telotroch; VNC, ventral nerve cord. [Color figure can be viewed at wileyonlinelibrary.com]

Across all annelids and mollusks examined to date, the anterior ectoderm including the brain is formed by 1q (Figure 1C, purple, Ackermann et al. ²⁰⁰⁵; Clement ¹⁹⁶⁷; Damen and Dictus ¹⁹⁹⁶; Hejnol et al. ²⁰⁰⁷; Meyer et al. ²⁰¹⁰; Render ¹⁹⁹¹; Vopalensky et al. ²⁰¹⁹; Weisblat et al. ¹⁹⁸⁴). The cells responsible for forming the CNS in the trunk are more variable in spiralians. In annelids, the cell responsible for forming the VNC in the trunk (herein defined as the tissue between the prototroch and telotroch—the posterior ciliary band) is called the somatoblast, and is often the micromere 2d (Ackermann et al. 2005; Matveicheva et al. ²⁰²⁴; Meyer et al. ²⁰¹⁰; Shimizu and Nakamoto ²⁰¹⁵; Weisblat et al. ¹⁹⁸⁴). The source of the molluscan trunk neural tissue is less clear but is generally some combination of 2q micromeres, which suggests clade‐specific differences in the embryonic origin of CNS tissue in the trunk cross‐phyla. Other key fates of note include that mesodermal structures are generally formed by 4d, while endodermal structures are generally formed by 3Q macromeres. However, in the annelid, C. teleta, trunk mesoderm is formed by 3c, 3d, and 4d (Figure 1C, blue) while anterior or head mesoderm is formed by 3a (Figure 1C, Meyer et al. ²⁰¹⁰).

Previously, blastomere ablation experiments at the 8‐cell and 16‐cell stages in C. teleta suggested that the brain is specified autonomously while trunk neural tissue is specified conditionally (Carrillo‐Baltodano and Meyer ²⁰¹⁷). Partial larvae derived from 1q micromeres without the 1Q macromeres formed a “swimming head” with neural tissue expressing the pan‐neural marker Ct‐elav1 and immunoreactive for the neurotransmitter marker FMRF‐like but not serotonin (5HT). This suggests that 1q cells inherit maternal determinants that allow for rapid specification of the brain. Neural specification in the trunk was examined by ablating the 2Q macromeres at the 16‐cell stage, leaving 2d, along with the rest of the 1q + 2q micromeres (1q¹, 1q², and 2q micromeres = “2q animal cap”, herein 1q + 2q). These ablations resulted in partial larvae that expressed Ct‐elav1 in the head but not trunk, suggesting conditional specification of trunk neural tissue. However, when 1q + 2q micromeres were isolated along with the 2D macromere, which contributes to the left mesodermal band and endoderm (Figure 1C), Ct‐elav1 expression in the trunk was restored, suggesting that this macromere was able to provide extrinsic signals sufficient for neural specification in the trunk.

Here we continued this work by examining a second annelid, the errant polychaete Platynereis dumerilii (Audouin and Edwards 1833), to look at conservation of this developmental regulation. We also performed additional blastomere manipulations in C. teleta: isolating 2d alone, isolating different macromeres with 1q + 2q, and additional markers for 1q + 2q, focusing on whether ablation of blastomeres affected D–V axis formation. Most strikingly, we found that 2d alone is sufficient to produce trunk neural tissue in both annelids. We also found that in C. teleta, the capacity of 2d to form neural tissue is decoupled from D–V axis formation and is regulated by external proneural and anti‐neural signals from macromeres and micromeres, respectively. This, combined with our finding that isolated 1q micromeres form neural tissue in P. dumerilii, suggests that head neural tissue can be autonomously specified in both C. teleta and P. dumerilii, while 2d can also form trunk neural tissue in isolation. More broadly, this further supports the hypothesis that specification of neural tissue and the D–V axis are decoupled.

2. Results

2.1. Neural Specification

2.1.1. Individual 1q Micromeres Are Intrinsically Capable of Forming Anterior Neural Tissue in C. teleta

The 1q micromeres form the majority of the head ectoderm, including the brain (Figure 1, Meyer and Seaver ²⁰¹⁰). We previously demonstrated that 1q partial larvae were capable of autonomous neural tissue formation when isolated together (1q: 1a + 1b + 1c + 1d), but whether this applied to all quadrants individually was unclear (Carrillo‐Baltodano and Meyer ²⁰¹⁷). Here, we tested if individual 1q micromeres (1a, 1b, 1c, or 1d) can produce neural tissue in isolation. Partial larvae derived from isolated 1a, 1b, 1c, or 1d micromeres were all very similar; they showed an anterior‐posterior (A–P) axis with a posterior presumptive prototroch and no trunk or pygidium—the posterior‐most body region. This matches the “swimming head” phenotype of 1q isolates but with smaller sizes (Figure 2). Each quadrant produced a partial larva of a comparable size with multiple neural markers, including a single Ct‐elav1 ⁺ patch (1a: N = 4/4; 1b: N = 4/4; 1c: N = 7/7; 1d: N = 11/11), anti‐acetylated Tubulin (Ac‐Tub; 1a: N = 5/5; 1b: N = 7/7; 1c: N = 5/5; 1d: N = 3/4), and FMRF‐like immunoreactivity (FMRF‐LIR; 1a: N = 5/5; 1b: N = 6/7; 1c: N = 3/5; 1d: N = 2/4). This further supports the hypothesis that brain tissue is autonomously specified in C. teleta.

Figure 2.

Phenotypes of 1a, 1b, 1c, and 1d isolates from 8‐cell embryos in Capitella teleta. (A and B) 1a isolates. (C and D) 1b isolates. (E and F) 1c isolates. (G and H) 1d isolates. (A, C, E, G) Ac‐Tub. (A′, C′, E′, G′) FMRF‐LIR. (A′′, C′′, E′′, G′′) Hoechst nuclei. (A′′, C′′, E′′, G′′) Ac‐Tub (yellow), FMRF‐LIR (magenta), and Hoechst nuclei (cyan) merged fluorescent image. (B, D, F, H) Ct‐elav1 ISH. Images with the same letter are the same individual. Left‐most column: Isolation and phenotype diagrams. Top right numbers in figure panel: number of isolates with shown neural phenotype out of total. pt, prototroch; A, anterior; scale bar: 20 µm. [Color figure can be viewed at wileyonlinelibrary.com]

2.1.2. Micromere 2d (1D‐2D) Is Intrinsically Capable of Forming Trunk Neural Tissue in C. teleta

The entirety of the ectoderm in the trunk and pygidium including the VNC as well as the two ciliary bands, the posterior telotroch and ventral neurotroch, arise from the descendants of the 2d micromere (Figure 1, Meyer and Seaver ²⁰¹⁰). We could not isolate 2d from an intact embryo in C. teleta, possibly because 2d is a very large cell that contacts all other cells in the 16‐cell embryo (Amiel et al. 2013). Instead, we were able to isolate 2d in a stepwise manner, by first isolating its parent cell (1D), and then removing its larger, yolk‐filled sibling cell 2D after division. Partial larvae derived from 1D‐2D were generally elongate with an A–P axis; they had no head features but there was a pronounced pygidium on one end (Figure 3). The pygidium was characterized by the presence of cilia on one end of the elongated partial larvae (N = 27/33), but the cilia were not clearly arranged into a ciliary band, that is, the telotroch (Figure 3H′′–J′′). Sometimes patches of cilia were also present in the trunk (N = 13/33), which may represent a partial neurotroch that was not clearly localized along a D–V axis. A lack of phalloidin staining suggests an absence of muscles, and therefore mesoderm (N = 16/16), as expected from the fate map. FMRF‐LIR (N = 7/7; Figure 3B), 5HT‐LIR (8/9; Figure 3D), Ac‐Tub neural signal (8/9; Figure 3D′), and Ct‐elav1 (N = 23/26; Figure 3F–H) were detected in the 1D‐2D partial larvae, showing the presence of neural tissue. Neural axons generally ran the length of the trunk but were not clearly localized along a D–V axis. Clusters of cells expressing Ct‐elav1 were generally seen throughout the trunk but they were not confined to any one side of the animal, i.e., the ventral side (Figure 3F–H), and could not be easily categorized. This is surprising as 2d is thought to act as the D–V organizer, at least for the head (Amiel et al. 2013). These results suggest that 2d is capable of producing neural tissue in the absence of other blastomeres, including those that generate mesoderm but without a D–V axis.

Figure 3.

Phenotypes of control and 1D–2D (2d) isolates from 16‐cell embryos in C. teleta. (A, D, F) Control animals (B, E, H–J) 1D‐2D partial larvae. (A, B) FMRF‐LIR. (A′, B′) FMRF‐LIR (magenta) and Hoechst nuclear (cyan) merged fluorescent images. (A′′, B′′) DIC images. (C, D) 5‐HT. (C′, D′) Ac‐Tub. (C′′, D′′) 5HT‐LIR (red), Ac‐Tub (yellow), and Hoechst nuclear (cyan) merged fluorescent images. (E, F, G, H) Ct‐elav1 images. (E, F, G) Three representative 1D–2D partial larvae showing the range of Ct‐elav1 phenotypes. (E′–G′) nuclei. (E′–G′′) Ac‐Tub. Top row: Isolation and phenotype diagrams. Images with the same letter are the same individual. fg, foregut; nt, neurotroch; pg, pygidium; pt, prototroch; tnt, trunk neural tissue; tt, telotroch; A, anterior; L, left lateral; scale bars: 50 µm. [Color figure can be viewed at wileyonlinelibrary.com]

2.1.3. Any 2Q Macromere Can Rescue Trunk Neural Tissue in 1q + 2q Partial Larvae

We previously showed that separation of the first two micromere quartets, including 2d, from the macromeres at the 16‐cell stage (1q + 2q) resulted in elongated partial larvae with an anterior brain but no trunk neural tissue. Including the 2D macromere (1q + 2q + 2D), which generates mesoderm and endoderm on the left side of the trunk, rescued trunk neural fate in the trunk (Carrillo‐Baltodano and Meyer ²⁰¹⁷). 2D was chosen because the D quadrant is generally responsible for organizing signals in spiralians (Goulding and Lambert 2016; Henry and Perry ²⁰⁰⁸; Henry et al. ²⁰¹⁷; Seudre et al. ²⁰²²). To determine whether macromeres from other quadrants can induce neural fate in the trunk, we isolated 1q + 2q micromeres with either macromere 2A, 2B, or 2C by ablating three of the four macromeres at the 16‐cell stage. In C. teleta, descendants of macromeres 2A and 2C generate endoderm plus anterior mesoderm or right trunk mesoderm, respectively, while macromere 2B generates endoderm and not much, if any, mesoderm (Figure 1C). The resulting partial larvae were similar in phenotype to 1q + 2q + 2D; all showed a clear A–P axis and the vast majority exhibited a well‐defined D–V axis with a head, brain, foregut, stomodeum, prototroch, and telotroch (Figure 4). Although the brain Ct‐elav1 ⁺ domain appear lighter in the control (Figure 4A), this is likely an artifact from z‐stack imaging through the thicker tissue in control larvae. Each macromere was sufficient to restore trunk neural tissue as assayed by Ct‐elav1 expression, 1q + 2q + 2A (N = 17/22), 1q + 2q + 2B (N = 16/22), 1q + 2q + 2C (N = 23/28), or 1q + 2q + 2D (N = 16/16), but with varying degrees of efficiency. Assessing the size of the Ct‐elav1 ⁺ domain in the trunk and the presence of a ventral midline (determined by the presence of a gap between bilateral bands of Ct‐elav1 expression) suggests that 1q + 2q + 2A had the weakest trunk neural rescue; the majority of these partial larvae had only weak, short (~1 ganglion) regions of Ct‐elav1 (1.8 ± 1.4 SD, N = 6), and none showed a midline. The 1q + 2q + 2C had the next weakest rescue where only 60% showed evidence of a midline (N = 4/7) and there were 3.4 ± 1.9 SD ganglia (N = 8) as determined by Ct‐elav1 expression. The 1q + 2q + 2B partial larvae were 75% likely to show a midline (N = 10/13), and there were 4.7 ± 2.2 SD (N = 10) hemiganglia with Ct‐elav1. In contrast, all 1q + 2q + 2D partial larvae expressed Ct‐elav1, with ~90% of 1q + 2q + 2D (N = 16/18) showing a midline and an elongated Ct‐elav1 ⁺ expression band with 6.1 ± 1.9 SD ganglia (N = 18). Control animals had 8.5 ± 1.4 SD hemiganglia (N = 10) with Ct‐elav1 ⁺ expression, and all had a clear midline. While these details are not sufficient to tease apart the signaling differences among the 2Q macromeres, all four vegetal macromeres at the 16‐cell stage or their descendants can produce proneural signals and these signals are not specifically localized to any quadrant.

Figure 4.

Ct‐elav1 expression in Capitella teleta partial larvae including the 1q + 2q micromeres (ventral view). (A) Control larva. (B) 1q + 2q partial larva. (C) 1q + 2q + 2A isolate. (D) 1q + 2q + 2B isolate. (E) 1q + 2q + 2C isolate. (F) 1q + 2q + 2D isolate. (G) 1q + 2q + 3q isolate. Left‐most column: diagram of ablation. Top right numbers in figure panel: number of isolates with shown phenotype out of the total. Asterisk: position of the mouth opening. A, anterior; D, dorsal; fg, foregut; nt, neurotroch; pg, pygidium; pt, prototroch; tnt, trunk neural tissue; tt, telotroch; L, left lateral; scale bars: 50 µm. [Color figure can be viewed at wileyonlinelibrary.com]

2.1.4. 1q+2q+3q Partial Larvae Have Reduced Trunk Neural Tissue

We ablated the 3Q macromeres after birth of the 3q micromeres at the 32‐cell stage and tested if partial larvae arising from the 1q + 2q + 3q micromeres had trunk neural tissue. 1q + 2q + 3q partial larvae looked very similar to the partial larvae derived from 1q + 2q + 2Q except they had a narrower trunk and less trunk neural tissue (Figure 4G). They had a clear A–P axis with a head and pygidium, and most had a clear D–V axis with a neurotroch (N = 10/14) and a ventral midline, although only some showed evidence of trunk mesodermal bands (nuclear staining; N = 5/14). Ct‐elav1 was expressed in both the head (N = 39/39) and trunk (Figure 3G Ct‐elav1; N = 24/39), albeit only in less than 5 ganglia (N = 17/20). This suggests that signaling events are still occurring for neural fate specification after birth of 3q micromeres, leading to a reduction instead of repression of neural tissue in the trunk.

2.2. Axial Organization of 1q+2q Partial Larvae in C. teleta

While the partial larvae derived from 1q + 2q isolated at the 16‐cell stage do not express neural markers in the trunk, they do elongate and seem to form an A–P axis with a head, trunk, and pygidium. Although we previously reported a lack of a D–V axis, foregut, and mouth (Carrillo‐Baltodano and Meyer ²⁰¹⁷), a more detailed investigation proved otherwise. Based on recent findings that the 2d blastomere is likely the D–V axial organizer in C. teleta (Lanza and Seaver 2018, ²⁰²⁰), we assessed the expression of additional D–V and tissue‐specific markers in partial larvae generated by 1q + 2q micromeres (Figure 5).

Figure 5.

Gene expression in partial C. teleta larvae including the 1q + 2q micromeres. (A, C, E, G, I, K) Control larvae. (B, D, F, H, J, L) 1q + 2q partial larvae. (A and B) Ct‐foxA ISH at stage 6. (C and D) Ct‐bmp5‐8 ISH at stage 6. (E and F) Ct‐pax3/7 ISH at stage 6. (G and H) Ct‐gataB1 ISH at stage 4. (I and J) Ct‐myoD ISH at stage 6. (K and L) Ac‐Tub (yellow) and Hoechst (white) at stage 6. Left most: diagram of ablation. Top right numbers in figure panel: number of isolates with shown phenotype out of total. Asterisk: position of the mouth opening. A, anterior; D, dorsal; L, left lateral; fg, foregut; nt, neurotroch; pt, prototroch; tt, telotroch; scale bars: 50 µm. [Color figure can be viewed at wileyonlinelibrary.com]

2.2.1. 1q+2q Partial Larvae Have a D–V Axis

The ectodermal foregut is formed by cells from the 2a, 2b, and 2c lineages (Meyer et al. 2010), and should be present in 1q + 2q partial larvae. We tested this by examining Ct‐foxA, which is normally expressed in the ventrally‐positioned foregut at stage 6, immediately dorsal to the mouth (Figure 5A, Boyle and Seaver ²⁰⁰⁸). In 1q + 2q isolations, Ct‐foxA was expressed in the anterior region of the elongated trunk, posterior to the prototroch and on the same side as the mouth opening (N = 6/8), suggesting a D–V axis is formed in the absence of 2Q macromeres or trunk neural tissue (Figure 5G).

We used Ct‐bmp5‐8 as a dorsal marker because it is normally expressed on the dorsal side of the foregut as well as in two dorsolateral stripes along the trunk ectoderm (Figure 5B, Webster et al. ²⁰²¹). In the partial larvae derived from 1q + 2q isolated at the 16‐cell stage, Ct‐bmp5‐8 is expressed in both a ventral and dorsal ectodermal domain. The first is anterior, just posterior to the prototroch, and on the same side as the mouth opening (N = 18/19; Figure 5H); the second domain is opposite and more posterior, suggesting that this is the dorsal domain, although there is no sign of two separate dorsal stripes (N = 14/19). This expression pattern further supports D–V axis formation in partial larvae generated by the 1q + 2q micromeres.

We examined Ct‐pax3/7 expression as a D–V marker in the partial larvae as this gene is normally expressed in two ventrolateral ectodermal bands at stage 6 (Figure 5C and Seaver et al. ²⁰¹²). In partial larvae derived from the 1q + 2q micromeres, no Ct‐pax3/7 expression was observed (N = 10/10; Figure 5I). This could indicate a lack of ventrolateral ectoderm. However, Pax3/7 homologs are thought to be involved in later stages of neurogenesis in at least some spiralians (Navet et al. 2017), and the lack of Ct‐pax3/7 in the partial larvae also could result from a lack of neural specification leading to an absence of neurogenic markers.

We also used Ac‐Tub immunolabeling to look for ciliated cells that might represent remnants of the neurotroch, a ciliary band that normally runs along the ventral midline in the trunk (Figure 5F). We found that 1q + 2q isolates had patches of external cilia in the trunk that were on the same side as the mouth (N = 10/14; Figure 5L), supporting the presence of neurotroch cells and ventral, non‐neural ectoderm. These cilia are unlikely to be the external cilia that surround the mouth, and are instead true neurotroch cilia, as the cilia near the mouth in wild‐type larvae come from third‐quartet micromeres (Meyer et al. 2010), which were removed in our experiment. A re‐examination of the partial larvae derived from 1q + 2q micromeres in Carrillo‐Baltodano and Meyer (2017), showed that most of these larvae also had a mouth opening, although a foregut lumen was not detectable despite the presence of foregut markers (N = 42/52). Taken together, these molecular and morphological markers indicate that there is an ectodermal D–V axis but no trunk neural tissue in partial larvae derived from 1q + 2q micromeres in C. teleta.

2.2.2. 1q + 2q Partial Larvae Lack Endoderm and Mesoderm

At the 16‐cell stage in C. teleta, mesoderm and endoderm are largely generated by progeny of the macromeres, which are removed in 1q + 2q partial larvae (Figure 1C; Meyer et al. ²⁰¹⁰). In order to confirm that these partial larvae lack mesodermal and endodermal tissue, we examined two markers, Ct‐myoD and Ct‐gataB1 (Boyle and Seaver 2008). Ct‐gataB1 is most visible in the endoderm at stage 4 (Figure 5D), and when control animals were at stage 4, no Ct‐gataB1 expression was observed in partial larvae derived from 1q + 2q isolations (N = 10/12; Figure 5J), suggesting a lack of endodermal tissue in these animals. Interestingly, the stage 4 partial larvae were already elongated whereas control animals were still ovoid in shape, which may be due to the lack of the large yolky macromeres during gastrulation, possibly accelerating the process of convergent‐extension. Ct‐myoD is expressed in the left and right mesodermal bands and has a role in muscle development (Boyd and Seaver 2023). In stage 6 controls, Ct‐myoD was expressed in the posterior region of both mesodermal bands (Figure 5E), but in 1q + 2q partial larvae, there was no clear Ct‐myoD expression (N = 11/11; Figure 5K) suggesting a lack of mesodermal tissue as expected.

2.3. Blastomere Isolations in P. dumerilii Recapitulate the Pattern of Autonomous Neural Specification in C. teleta

To test whether the results in C. teleta (Sedentaria), can be generalized more broadly, we performed blastomere isolations in the distantly‐related annelid P. dumerilii (Errantia; Figure 6) (Struck et al. 2011; Weigert and Bleidorn ²⁰¹⁶). Errantia + Sedentaria comprise Pleistoannelida, and members of these clades diverged no later than the Early Ordovician (~490 Ma; Parry et al. ²⁰¹⁴). These groups have distinctly different lifestyles (errant vs. sedentary) and associated morphologies (e.g., swimming vs. reduced parapodia). Both C. teleta and P. dumerilii have unequal spiral cleavage with early blastomeres displaying very similar developmental fates (Ackermann et al. 2005; Meyer et al. ²⁰¹⁰; Vopalensky et al. ²⁰¹⁹). In P. dumerilii, isolated 1q micromeres also developed into “swimming heads”, similar to what we previously described for C. teleta (Carrillo‐Baltodano and Meyer ²⁰¹⁷). They all showed a clear A–P axis, with a posteriorly‐localized ciliary band that was likely the prototroch and anteriorly‐localized neural markers, but no eyes (N = 24/24; Figure 6D–F), which contrasts with formation of eyes in some 1q partial larvae in C. teleta. Specifically, 1q partial larvae had neurites and cilia labeled with Ac‐Tub (N = 24/24), FMRF‐LIR (N = 11/12), and Pdu‐elav1 expression (N = 2/2) but not 5HT (N = 0/12), as in C. teleta. These data suggest that P. dumerilii can also produce some types of anterior neural tissue autonomously.

Figure 6.

Phenotypes of Platynereis dumerilii isolates. (A, B, C) Control. (D, E, F) 1q isolates. (G, H, I) 2d direct isolates. (J, K, L) 1D‐2D isolates (2d indirect isolations). (A, D, G, J) Ac‐Tub. A′, D′, G′) FMRF‐LIR. (A′′, D′′, G′′, J′′) Hoechst nuclear. (A′′′, D′′′, G′′′) Ac‐Tub (yellow), FMRF‐LIR (magenta), and Hoechst nuclear (cyan) merged fluorescent image. (B, E, H, J′′′) 5‐HT (red in merged images) (B′, E′H′J′′′′) 5‐HT (red), Ac‐Tub (yellow), and Hoechst nuclear (cyan) merged fluorescent images. (C, F, I, L) Pdu‐elav1 ISH. Images with the same letter are the same individual. Top row: diagram of isolation. Top right numbers in figure panel: number of isolates with shown neural phenotype out of total. A, anterior; L, left lateral; pt, prototroch; tt, telotroch; scale bars: 20 µm. [Color figure can be viewed at wileyonlinelibrary.com]

Unlike in C. teleta, we were able to directly isolate the 2d micromere from 16‐cell P. dumerilii embryos. However, the 2d partial larvae did not elongate as they do in C. teleta and instead appeared more as an ovoid ball of cells with no external features: no chaetae, only disorganized cilia, and no distinguishable axes. Regardless, the 2d‐derived partial larvae in P. dumerilii developed neural tissue (Figure 6G–I), as shown by neurites and cilia labeled with Ac‐Tub (N = 16/18), FMRF‐LIR (N = 3/3), Pdu‐elav1 expression (N = 12/12), and unlike the head, some 5HT (N = 2/4). Although the neural tissue was sometimes confined to a single side of the embryo, this was not consistent enough to suggest a D–V axis. We also carried out 1D‐2D isolations for comparison and found that 1D‐2D partial larvae were generally indistinguishable from 2d partial larvae with no morphologically‐distinguishable axes or external features. These partial larvae also expressed the neural markers (Figure 6J–L) Ac‐Tub (N = 7/7), Pdu‐elav1 (N = 4/4), and 5HT (N = 3/5). This supports the idea that 1D–2D isolation can be used to indirectly isolate 2d in C. teleta as there were no phenotypic differences between the two isolations in P. dumerilii. It also suggests that the failure of 2d partial larvae to elongate was not due to differences in isolation procedures between C. teleta and P. dumerilii. Taken together, these data suggest that 2d also has the intrinsic capacity to form trunk neural tissue in P. dumerilii.

3. Discussion

3.1. Separation of Neural Specification and D–V Axis Formation in Annelida

It is generally considered that the specification of the D–V axis and the nervous system are mechanistically linked (Holley et al. 1995; Kuroda et al. ²⁰⁰⁴; Mieko Mizutani and Bier ²⁰⁰⁸; Wilson and Edlund ²⁰⁰¹). Thus, elucidating the molecular and cellular mechanisms controlling neural specification in a wide array of animal clades is crucial to reconstructing the evolutionary trajectory of nervous systems and their relationship to the D–V axis. Although there has been recent progress in comparing the molecular control of D–V axis formation across spiralian taxa, the control of neural specification in spiralians is relatively unknown. Various studies have demonstrated that blocking Activin signaling (Lanza and Seaver 2018, ²⁰²⁰) or deleting 2d (Amiel et al. 2013) in C. teleta results in a loss of the D–V axis but not head neural tissue. More recently, Carrillo‐Baltodano et al. (2026) suggest that this can be explained by developmental systems drift. They show that BMP does not play a role in D–V axis signaling in annelids with unequal spiral cleavage (C. teleta and P. dumerilii) but is involved in D–V axis formation in species with equal spiral cleavage (Owenia fusiformis and Spirobranchus lamarcki). Furthermore, their data across different annelids suggest that neither BMP nor Activin signaling have an anti‐neural role, implying that D–V axis formation and neural specification have different molecular underpinnings. Our data complement these findings. Isolated 1q + 2q partial larvae express non‐neural ectodermal gene markers in an ordered manner, Ct‐foxA ventrally and Ct‐bmp5‐8 dorsally, but not neural markers, suggesting the formation of the D–V axis in the ectoderm in the absence of trunk neural tissue. This suggests that in C. teleta, D–V organizer signaling is not responsible for restricting neuroectoderm to one region of the D–V axis. Instead, a combination of inherited factors plus additional vegetal proneural signals and anti‐neural signals from animal micromeres, which may not even originate from one quadrant, establish the neuroectoderm. This raises an intriguing question of how the brain and VNC in C. teleta become localized to one region along the D–V axis. Indeed, recent work by Knabl et al. (2025) suggests that the anti‐neural role of BMP in vertebrates and arthropods is not ancestral. If these two developmental features are ancestrally decoupled in Spiralia, it should be considered as a key piece of evidence to understand how central nervous systems evolved.

3.2. Autonomous and Conditional Neural Specification in Annelida

Our primary novel result is the clear presence of neural tissue in partial larvae derived from just a single cell—micromere 2d—in C. teleta (Sedentaria) and P. dumerilii (Errantia), demonstrating its intrinsic capacity to form neural tissue in the trunk. Furthermore, our data show that isolated 1q micromeres produce neural tissue in both species. These results suggest that head neural tissue can be specified autonomously via inherited determinants in annelids as individual or subsets of blastomeres. While these individual micromeres at the 8‐ and 16‐cell stages can produce neural tissue in isolation, we also showed a pattern of cell‐cell interactions during cleavages stages that affects neural fates.

For the brain, 1q partial larvae in C. teleta (Carrillo‐Baltodano and Meyer ²⁰¹⁷) and P. dumerilii lacked serotonergic neurons, suggesting that some neural subtypes in the brain require extrinsic signaling. Alternatively, it is also possible that the neural progenitor for serotonergic neurons in the brain originates from 2d; some 2d descendants are found in the dorsal brain and the circumesophogeal nerve ring (Meyer et al. 2010; Meyer and Seaver ²⁰¹⁰). In contrast, cell lineage data in P. dumerilii shows 1q descendants as the source of 5HT neurons in the apical organ (Vopalensky et al. 2019). In C. teleta 1q + 2q isolates do have serotonergic neurons in the brain (Carrillo‐Baltodano and Meyer ²⁰¹⁷), and preliminary data showed that 1q + 2d partial larvae of P. dumerilii also have serotonergic neurons in the brain (N = 2/3). We do not have the data to distinguish whether serotonergic neurons in the brain are conditionally specified from the 2d or 2q lineage, or whether they originate from 2d.

In the trunk, we show an interesting pattern of proneural and anti‐neural signaling from other blastomeres that are also involved in trunk neural specification. We are confident that 2d (1D–2D) partial larvae develop VNC tissue because trunk peripheral neurons only appear as a few cells of Ct‐elav1 expression (Meyer and Seaver 2009) rather than the large patches found here (Figure 3). Isolating micromere 2d with the other 16‐cell micromeres (1q + 2q partial larvae) blocks trunk but not head neural formation in C. teleta. Moreover, any single vegetal 2Q macromere, not just 2D, is sufficient to rescue trunk neural fate when isolated with 1q + 2q micromeres. Signaling from other blastomeres (pro‐ and anti‐neural) is thus necessary to maintain balance in neural fate specification in the trunk in complete larvae.

We did not identify the specific signaling molecules or centers for neural specification; however, our data suggest a few hypotheses in C. teleta. Firstly, micromeres 1q² and/or 2a–c produce anti‐neural signals for the trunk but not the head. 1q + 2q partial larvae consist of micromeres 1q¹, 1q², and 2q, and they generate head, but not trunk neural tissue. Neural fate may be largely autonomously determined in 1q¹ cells, making any anti‐neural signals from 1q² or 2q insufficient to alter their trajectory, explaining why anterior neural tissue was not affected. Alternatively, the specification of head and trunk neural tissue is different enough to isolate the brain from trunk anti‐neural signals. Interestingly, both 1q² and 2a–2c micromeres primarily produce ectodermal tissues, the prototroch and foregut, respectively, although micromeres 2a–2c also generate interstitial cells in the trunk, which were previously interpreted as ectomesoderm (Meyer et al. 2010). This would suggest that the anti‐neural signal comes from cells that generate non‐neural ectoderm.

Our data also suggest that, to varying degrees, 2Q macromeres at the 16‐cell stage produce a vegetally‐localized proneural signal for the trunk that is not restricted to the D‐quadrant. By adding any single 2Q macromere to the 1q + 2q micromeres, trunk neural tissue could be rescued. The timing of this signal is notable; isolating the 1q + 2q + 3q micromeres just 1 h later resulted in development of a small amount of trunk neural tissue. Thus, the balance of conditional signals at the 16‐cell stage have changed; the proneural signals have increased relative to the anti‐neural signals seen in the 1q + 2q partial larvae, allowing the formation of some trunk neural tissue. This could indicate that by the 32‐cell stage, the anti‐neural signal has decreased or the proneural signal from 2Q has already had a partial effect. This timing fits well with a key BMP signaling window, whereby exogenous BMP causes a radialized brain if it overlaps the birth the 3q micromeres (Webster et al. 2021). Surprisingly, the proneural signal is not localized to the D quadrant macromere, 2D, as any of the four macromeres can rescue trunk neural fate. This suggests a decoupling of neural fate from the organizing activity of the D quadrant found in many spiralians, at least in the trunk. In general, 1q + 2q + 2D partial larvae had the most robust trunk neural signal, which we predicted because of the general role of the D‐quadrant as organizers in spiralians (Goulding and Lambert 2016; Henry and Perry ²⁰⁰⁸; Henry et al. ²⁰¹⁷; Seudre et al. ²⁰²²), even though 2d is generally considered the organizer in C. teleta (Amiel et al. 2013). Macromeres 2A and 2C make sense as alternative trunk proneural signaling cells because of their role in mesoderm formation; however, 2A only contributes to the formation of mesoderm in the head (Meyer et al. 2010). Mesoderm generally plays an important role in inducing neural tissue in vertebrates and insects (Beddington 1994; Jones and Woodland ¹⁹⁸⁹; Kiecker and Niehrs ²⁰⁰¹; Lynch and Roth ²⁰¹¹). Most surprising was the neural rescue in 1q + 2q + 2B because macromere 2B does not generate mesodermal derivatives in C. teleta (Figure 1C; Meyer et al. ²⁰¹⁰). It is possible that the descendants of 2B instead play a role in early signaling, possibly including proneural signals, rather than directly producing mesoderm. The fact that trunk proneural signals can come from all four quadrants does raise some interesting questions about the details of neural signaling in spiralians, and especially the role of mesodermal signaling.

Intriguingly, we show that the mechanisms of neural specification in the head and trunk are different, something that is not regularly examined in the literature. Brain neural tissue, with the exception of 5HT+ neurons may be autonomously specified while VNC fate is affected by proneural and anti‐neural signals. This fits with the larger picture of larval evolution suggesting that there is a decoupling of head and trunk formation that may have facilitated larval and life cycle evolution (Carrillo‐Baltodano and Meyer ²⁰¹⁷; Martín‐Zamora et al. ²⁰²³). Indeed, even the definition and make up of a brain varies between groups (Nielsen 2005; Sumner‐Rooney and Sigwart ²⁰¹⁸). For example, the larval apical organ might be considered an ancestral eumetazoan brain, but is not homologous to the adult brain and is lost at metamorphosis (Moroz 2009; Nielsen ²⁰⁰⁵). If head and trunk neural tissue have different regulatory mechanisms, perhaps they should be treated differently when investigating the homology of CNSs across Bilateria (Martín‐Durán et al. ²⁰¹⁸; Nanglu et al. ²⁰²³). Ultimately, identification of the autonomous and conditional signals for head and trunk neural tissue specification in different species will help us understand if these two systems evolved in parallel or if there are common molecular mechanisms controlling both.

Whether this neural specification pattern is conserved across spiralians remains to be seen, but we think that this is unlikely considering the wide array of nervous system arrangements in Spiralia. Most of the data on neural development in Spiralia comes from work with BMPs; existing data on the role of BMP on neural tissue formation in spiralians shows a lack of developmental conservation (Lyons et al. 2020; Tan et al. ²⁰²²; Webster et al. ²⁰²¹). Overall, we have demonstrated patterns of both autonomous and conditional signaling during cleavage stages that regulate both head and trunk neural specification in two species of annelids.

3.3. P. dumerilii and C. teleta Partial Larvae Developed Differently

While the neural specification patterns for 1q and 2q isolates we tested were conserved between these two species, we were unable to make strong conclusions from additional blastomere isolations in P. dumerilii. Partial larvae resulting from isolated 1q + 2q micromeres or other late isolations were too difficult to interpret and were not included here. In particular, P. dumerilii partial larvae never had eyes or an elongated body shape. Without the elongated body shape, it was more difficult to identify larvae that did not gastrulate properly or to even determine an A–P axis. We believe this likely relates to key differences in how the blastomeres are held together at early cleavage; a more rigorous process was required to allow the separation of blastomeres, and even treatment controls—where embryos were treated for isolation, but no cells were removed—did not develop normally in P. dumerilii. Previous authors have suggested that blastomere isolations are not possible in P. dumerilii (Fischer and Dorresteijn 2004) or the related Alitta virens (Kostyuchenko and Dondua 2017) as everything but the trochoblasts cease differentiation. However, Costello (1945) successfully raised isolated blastomeres in Alitta succinea to trochophores using alkaline seawater. Carrillo‐Baltodano et al. (2026) did have some larvae elongate when only deleting the later blastomeres 2D, 3d, and 4d in P. dumerilii, which resulted in partial larvae with defined axes and external features. It is possible that more refined techniques would produce data that are easier to interpret, but we think it is possible that P. dumerilii are more affected by blastomere manipulation because they are planktonic, as compared to the brooded C. teleta larvae.

3.4. Conclusions

In conclusion, by manipulating blastomeres at early cleavage stages, we showed that neural specification shows autonomous and conditional features that differ between the brain and VNC of two annelids, C. teleta and P. dumerilii. We further showed that neural specification is not coupled to D–V axis organization in the trunk of C. teleta. We hypothesize that these processes occur independently in other annelids as well. Uncoupling of neural and D–V axis specification, as well as decoupling head and trunk neural specification could have permitted the evolvability of different types of nervous systems and life cycles within annelids, explaining the large diversity of nervous systems in this group. It is only by expanding into a broad diversity of animals that we can gain a fuller understanding of the diversity, development, and evolution of nervous systems.

4. Materials and Methods

4.1. Animal Care and Embryo Collection

A colony of Capitella teleta Blake et al. ²⁰⁰⁹ adults were cultured as previously reported (Grassle and Grassle 1976; Seaver et al. ²⁰⁰⁵). Embryos were dissected from the broods of females and cultured in ASW with 60 µg/mL penicillin and 50 µg/mL streptomycin (ASW + PS) (Carrillo‐Baltodano and Meyer ²⁰¹⁷; Meyer et al. ²⁰¹⁰).

Platynereis dumerilii (Audouin and Edwards 1833) were maintained at the Marine Biological Laboratory, as previously reported (Kuehn et al. 2019). Zygotes were produced by spawning mature worms as described previously. Briefly, mature male and female worms were assembled into a small glass dish until spawning. Zygotes were then rinsed to prevent polyspermy and de‐jellied and left to develop until the appropriate stage.

4.1.1. Blastomere Isolations

In all species, embryos of the correct stage were washed three times with ASW + PS. After chemical dissociation (see below for species specifics), the unwanted blastomeres were removed from the rest of the embryo using an eyelash brush in a gelatin‐ or agarose‐coated dish. The partial embryos were raised individually in ASW + PS in gelatin‐ or agarose‐coated 24‐well dishes until a comparable stage was reached. The delicate embryos were only handled with eyelash brushes and mouth pipettes.

Isolations in C. teleta were performed following Carrillo‐Baltodano and Meyer (2017). Briefly, embryos at the selected cleavage stage were incubated in 4‐well gelatin‐coated dishes with Calcium and Magnesium Free Artificial Sea Water (CMF‐ASW) (Strathmann 1987) for 5 min, followed by incubation in 1% protease (Millipore‐Sigma cat# P6911) and 1% sodium thioglycolate (Millipore‐Sigma cat# T0632) in CMF‐ASW for 5 min at room temperature (~22°C). ASW + PS was changed daily or not at all in partial and whole‐embryo controls, with no changes in survival or health. Embryos were raised individually in ASW + PS for 6 days at room temperature until stage 6, when the major features, especially the nervous system, have formed (Seaver et al. 2005). The exception was for Ct‐gataB1, which is best expressed at stage 4 (~4 dpf).

In P. dumerilii, a stronger dose of thioglycolate was necessary to remove the egg envelope. Embryos were treated as above but on a rocker and with 1% thioglycolic acid (pH 8, Sigma T6750) instead of sodium thioglycolate. Embryos were raised for 72 h at 18°C in ASW + PS with no water changes to reach a comparable stage. In P. dumerilii, 72 h post‐fertilization is roughly comparable to stage 6 in C. teleta, where the major features of the brain and VNC have developed (Fischer et al. 2010).

4.1.2. Fixation

Animals were relaxed with a 1:1 mixture of ASW plus 0.37 M MgCl₂ for 5–12 min then fixed with 4% paraformaldehyde (stock 32% PFA ampules from Electron Microscopy Sciences, cat. 15714) in ASW for 30 min for immunolabeling or overnight for whole mount in situ hybridization (WMISH). Fixation was stopped with three washes in PBS.

4.1.3. Immunolabeling

After fixation, C. teleta larvae were immunolabeled following Meyer et al. (2015). Briefly, animals were permeabilized in PBS + 0.1% Triton‐X 100 (PBT), blocked in 10% goat serum, and then incubated in primary antibody in block overnight at 4°C. Animals were washed and then incubated with a secondary antibody and 0.1 µg/mL 33342 Hoechst (Millipore‐Sigma, cat. B2261) in block overnight at 4°C. Animals were then washed at room temperature with PBT, cleared in 80% glycerol in PBS, and mounted on slides for differential interference contrast (DIC) and fluorescent imaging. Primary antibodies are as follows: 1:600 rabbit anti‐serotonin (5HT; Millipore‐Sigma, cat #S5545) and 1:800 mouse anti‐acetylated tubulin (Ac‐Tub; clone 6‐11B‐1, Millipore‐Sigma, cat #T6793). Secondary antibodies were as follows: 1:300 goat anti‐mouse F(ab’)₂ FITC (Millipore‐Sigma, cat #F8521) and 1:600 sheep anti‐rabbit F(ab’)₂ Cy3 (Millipore‐Sigma, cat #C2306). 1:100 BODIPY FL‐Phallacidin (Life Technologies, cat #B607; stock concentration 200 Units/mL in methanol) was included with secondary antibodies.

In P. dumerilii, immunolabeling followed the above, except an additional proteinase K digestion step was added for permeabilization (2 min in 100 μg/mL proteinase K), followed by a 20 min post fixation in 4% PFA.

4.1.4. Whole Mount In Situ Hybridization (WMISH)

WMISH was conducted as described previously (Seaver et al. 2001). After fixation, animals were serially dehydrated in methanol and stored at −20°C. Animals were hybridized for a minimum of 72 h at 65°C with 1 ng/μL of each probe. Spatiotemporal RNA localization was observed using an NBT/BCIP color reaction. The color reaction was stopped using 3 washes of PBS + 0.1% Tween‐20. After WMISH, some animals proceeded to immunolabeling using 1:800 mouse anti‐acetylated tubulin (Millipore Sigma, cat #T6793) and 0.1 μg/mL Hoechst 33342 (Sigma‐Aldrich, cat. B2261) as in Carrillo‐Baldodano and Meyer (2017), cleared in 80% glycerol in PBS, and mounted on slides for DIC and fluorescent imaging.

DIG‐labeled anti‐sense RNA probes from Boyle and Seaver (2008) and Meyer and Seaver (2009) were used at a concentration of 1 ng/µL; 2 ng/µL of Ct‐bmp5‐8 RNA probe was used as in Corbet (2016).

4.1.5. Imaging

Immunolabeled and WMISH animals were imaged using DIC optics on an AxioImager M2 compound microscope (Zeiss) coupled with an 18.0 megapixel EOS Rebel T2i digital camera (Canon) or with an AxioCam MRm rev.3 camera (Zeiss) and Zen Blue software (Zeiss). DIC z‐stack images were rendered in Helicon Focus (HelSoft). Z‐projections were obtained from the ZenBlue stacks using Fiji (Image J2, NIH). Contrast and brightness of immunolabeling and WMISH images were edited in Photoshop CC, and figure panels were constructed using Illustrator CC (Adobe Systems Inc.).

Author Contributions

A.C.B., N.P.M., and N.B.W. conceived of the study. A.C.B., N.P.M., N.B.W., J.D.S., S.D. collected data. A.C.B. and N.B.W. prepared visualization. A.C.B. and N.B.W. wrote the manuscript. A.C.B., N.P.M., N.B.W., J.D.S., and B.D.O. edited the manuscript, and N.P.M. and B.D.O. provided resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.