Glycoconjugates mark a transient barrier to neural crest migration in the chicken embryo. R. A. Oakley1, C. J. Lasky2, C. A. Erickson3 and K. W. Tosney1,2,*.
103
Development 120, 103-114 (1994) Printed in Great Britain The Company of Biologists Limited 1994
Glycoconjugates mark a transient barrier to neural crest migration in the chicken embryo R. A. Oakley1, C. J. Lasky2, C. A. Erickson3 and K. W. Tosney1,2,* 1Neuroscience Program and 2Biology Department, The University of Michigan, Ann Arbor, MI 48109; 3Department of Molecular and Cell Biology, The University of California, Davis, CA 95616, USA
*Author for correspondence
SUMMARY We report that two molecular markers correlate with a transient inhibition of neural crest cell entry into the dorsolateral path between the ectoderm and the somite in the avian embryo. During the period when neural crest cells are excluded from the dorsolateral path, both peanut agglutinin lectin (PNA)-binding activity and chondroitin-6sulfate (C6S) immunoreactivity are expressed within this path. Both markers decline as neural crest cells enter. Moreover, both markers are absent after an experimental manipulation that accelerates neural crest entry into this path. Specifically, dermamyotome deletions abolish expression of both markers and allow neural crest cells to
enter the dorsolateral path precociously. After partial deletions, dermatome remnants remain. These remnants retain PNA and C6S labeling and impede migration locally. Local glycoconjugate expression thus correlates directly with avoidance responses. Since both PNA-binding activity and C6S expression also typify inhibitory somitic tissues, molecules indicated by these markers (or co-regulated molecules) are likely to inhibit both neural crest and axon advance.
INTRODUCTION
enter this path precociously as though an inhibition were removed (Erickson et al., 1992). Therefore, regardless of whether crest cells are specified to take only the dorsolateral path, the dorsolateral path must transiently possess inhibitory properties. These inhibitory properties, while transient, may not be unique. Similar inhibitory properties have been experimentally documented in three tissues next to paths, the posterior sclerotome (Keynes and Stern, 1984; Rickmann et al., 1985; Tosney, 1987, 1988a), perinotochordal mesenchyme (Newgreen et al., 1986; Pettway et al., 1990; Tosney and Oakley, 1990) and pelvic girdle precursor (Tosney and Landmesser, 1984). The inhibitory function of these tissues correlates with the expression of two molecular markers. All three tissues selectively express peanut agglutinin lectin (PNA)-binding activity and chondroitin-6-sulfate (C6S) immunoreactivity (Oakley and Tosney, 1991). The coincidence between inhibitory function and marker expression inspired the hypothesis that similar cues guide both crest cells and peripheral axons (Tosney and Oakley, 1990). The dorsolateral path differs from the other inhibitory tissues in being only transiently inhibitory. The same area first inhibits and later permits crest cell advance. We therefore asked whether the previously identified markers for inhibition were expressed in the dorsolateral path transiently. They are. We then asked whether the surgery that releases the inhibition also abolishes marker expression in the path. It does. Last, we addressed the nature of the inhibition itself. We found that the
In the trunk of mouse and avian embryos, neural crest cells emigrate from the neural tube into a space where two paths diverge, a ventral path through the somite and a dorsolateral path between the somite and the ectoderm (Weston, 1963; Derby, 1978; Le Douarin, 1982). Despite apparently equal access to both paths, crest cells initially migrate only along the ventral path and subsequently enter the dorsolateral path a day later (Weston and Butler, 1966; Derby, 1978; Serbedzija et al., 1989, 1990; Erickson et al., 1992). Why do neural crest cells delay entering the dorsolateral path? Perhaps only one subpopulation is able to colonize this path and it emerges late. Late-emerging crest cells have been suggested to enter the dorsolateral path exclusively (Serbedzija et al., 1989, 1990) where they form a single derivative, melanocytes (DuShane, 1935; Dorris, 1939; Weston, 1963). However, these cells are unlikely to be specified solely for the dorsolateral path since the melanocyte precursor is bipotential. It can form melanocytes or Schwann cells, it migrates along both paths, and it can form melanocytes even after migrating ventrally (Cowell and Weston, 1970; Nichols et al., 1977; Stemple and Anderson, 1992; see Ciment, 1990; Weston, 1991). Moreover, even fully differentiated melanocytes can migrate ventrally (Erickson et al., 1980). A second explanation for the delay, that the early path inhibits migration of all crest cells, is supported by experimental evidence. After dermamyotome removal, crest cells
Key words: inhibition, neural crest migration, PNA-binding activity, chondroitin sulfate, melanocyte
104
R. A. Oakley and others
inhibition acts locally, apparently eliciting avoidance on contact. Moreover, local inhibition coincides with local marker expression. These glycoconjugates thus mark inhibitory function both in space and in time. MATERIALS AND METHODS Sectioning For cryotomy, embryos processed as below were infiltrated in 5% sucrose in phosphate-buffered saline (PBS) for 1-24 hours and then in 15% sucrose (in PBS) for 18-24 hours at 4°C. After incubation in a solution of 7.5% gelatin (Sigma, 300 bloom) and 15% sucrose (in PBS) for 4-6 hours at 37°C, embryos were transferred to a block containing the same solution, oriented, and incubated for 24 hours at 4°C. Blocks were rapidly frozen in isopentane on dry ice. Serial, 10 µm frozen sections were cut using a Reichert cryostat, collected on gelatin-subbed slides and viewed with epifluorescence on a Nikon Optiphot microscope equipped with a Nikon UFX camera. DiI labeling We used DiI (1,1-dioctadecyl-3,3,3′,3′-tetra-methylindocarbocyanine perchlorate; Molecular Probes) to label premigratory neural crest cells in chick embryos at stages 15 to 24 (Hamburger and Hamilton, 1951) as described in Serbedzija et al. (1990). DiI was prepared as a 0.05% solution in 20% ethanol in 0.3 M sucrose solution, filtered at 5 µm, and pressure-injected into the lumen of the neural tube. After 12 or 18 hours, embryos were fixed overnight in 4% paraformaldehyde with 0.25% glutaraldehyde in PBS, washed with PBS and sectioned as above. Sections were not coverslipped and were photographed immediately after sectioning. Immunocytochemistry Crest cells within the dorsolateral path were previously presumed to lack antigenicity for HNK-1 (Rickmann et al., 1985; Loring and Erickson, 1987; Serbedzija et al., 1989). However, these cells do label brightly with HNK-1 if embryos are incubated in the antibody before sectioning (Erickson et al., 1992), the procedure used here. We fixed stage 17 to 24 embryos for 4 hours in 4% paraformaldehyde with 0.5% cetylpyridinium chloride in PBS. Embryos were cut into fragments 5-7 segments in length, washed, rinsed once in 0.02 M glycine, incubated for 3 hours in 3% bovine serum albumin (BSA) in PBS and incubated overnight at 4°C in HNK-1 antibody (hybridoma ascites; 1:250 in PBS with 3% BSA). Except where noted, all washes were in three changes of PBS and all incubations were at room temperature. Unbound antibody was washed out by incubating 4 hours in 0.5% BSA in PBS. Specimens were immersed overnight in secondary antibody (TRITC-conjugated rabbit anti-mouse IgM; 1:50 in 0.5% BSA in PBS; Jackson Immuno-Research), incubated 4 hours in 0.5% BSA, washed, postfixed overnight in 0.4% paraformaldehyde in PBS to crosslink bound antibody, washed, rinsed in 0.02 M glycine, washed and sectioned as above. HNK-1 applied as above labels crest cells in the dorsolateral path effectively, but penetrates deeper tissues poorly (see Loring and Erickson, 1987). The limited staining of ventral derivatives hindered our ability to detect surgeries that had depleted crest cells. For this reason, and for aesthetics, we also applied HNK-1 antibody to frozen sections of embryos labeled as above. We co-localized HNK-1 immunoreactivity and PNA-binding as follows. Rather than using a secondary antibody to amplify the lectinbinding signal (as in Oakley and Tosney, 1991), we used PNA lectin directly conjugated to FITC (Vector). All reagents were diluted in Hepes-buffered saline (HBS: 10 mM Hepes, 0.15 M NaCl, 0.1 mM CaCl2). Sections were rehydrated in PBS to remove gelatin, incubated for 20 minutes in HBS with 1% BSA, incubated for 90 minutes in primary antibody (HNK-1 hybridoma ascites; 1:250 in HBS with 1%
BSA), washed, blocked for 20 minutes in HBS containing 1% BSA and 10% normal rabbit serum, incubated for 60 minutes in a secondary antibody-lectin mixture (rabbit anti-mouse IgM-TRITC, 1:50; PNA-FITC, 1:200 in blocking solution), washed and coverslipped in immunofluor (ICN). The distribution of C6S epitopes was revealed using monoclonal antibody 5/6/3B3 (ICN; Couchman et al., 1984). Stage 17-24 embryos were fixed and sectioned as above. Sections were postfixed in cold 95% ethanol, rehydrated in PBS, digested with chondroitinase ABC (0.2 units/ml) in 0.1 M Tris-acetate buffer with 1% BSA for 1 hour at 37°C, washed, incubated for 20 minutes in PBS with 1% BSA, incubated for 60 minutes in 5/6/3B3 (1:50 in PBS with 1% BSA), washed, blocked as above, incubated in secondary antibody (rabbit anti-mouse-IgM-FITC, 1:50 in block), washed and mounted in immunofluor. Embryonic surgeries We deleted thoracic and upper lumbar dermamyotomes from stage 17 embryos as in Tosney (1987). In brief, we slit ectoderm posterior to the deletion site and removed 1-5 dermamyotomes using a small micropipette. We confined deletions to posterior somites where crest cells had yet to penetrate sclerotome. We fixed at stages 19-23. Using the relevant protocols above, 41 embryos were double-labeled with HNK-1 and PNA, 21 were labeled with HNK-1 antibody alone and 15 were labeled with 5/6/3B3 alone. Dermatome remnants were identified by their distinct epithelial morphology. Reconstructions and quantitation To compare migration on control and operated sides with accuracy, we used alternate sections to reconstruct the dorsolateral paths as flattened, dorsal projections (see Fig. 6). In the absence of dermatome, we included an HNK-1-positive cell if it lay within 30 µm of the ectoderm. This criterion successfully distinguishes crest cells in the dorsolateral path from those in sclerotome (Erickson et al., 1992). Maps portray 3-6 segments each from 27 operated embryos at stages 20-22. In 17 operated embryos, we also recorded the position and intensity of PNA-binding activity. Patterns were confirmed by observation in 35 additional embryos. To quantitate migration, we drew a grid over each dorsal map, with each square of the grid representing 100 µm on a side (Fig. 1). We then counted the crest cells within each square. Since each grid unit also has a third dimension of approximately 30 µm and represents a volume of the path, we expressed quantitative data as cells per unit volume. We assessed statistical significance using the Student’s t-test.
RESULTS Because HNK-1 has until recently been questioned as an effective marker for neural crest cells in the dorsolateral path, we used DiI to confirm when crest cells enter this path at hindlimb levels. We examined the progression of crest cells labeled with DiI at successive stages and fixed 12 or 18 hours later. Our results confirm that entry is delayed (as shown with DiI at wing levels; Serbedzija et al., 1989) and that HNK-1 accurately reveals migration in this path (Erickson et al., 1992). We also confirm that crest cells colonize their derivatives in roughly ventral to dorsal order (Weston and Butler, 1966; Serbedzija et al., 1989). For example, DiI-labeled cells first invade the sympathetic region at stage 19 but do not invade the dorsolateral path until stage 21, nearly a day after they began to migrate ventrally (Table 1). We document these results because they differ from a previous report (Serbedzija et al., 1989) in one important respect. We do not find an exclusive deployment to any par-
Glycoconjugates and neural crest cells
105
Fig. 1. One segment from a dorsal map shows how we quantitated migration. We drew a grid of squares over each dorsal map and counted the crest cells within each square. Each square was scaled to represent 100 µm on a side. Each row of four squares along the medial to lateral axis thus divides a path into four successive units (bottom scale) that represent the proximal to distal progression along a path. Each square represents a unit volume of the path since the dorsal maps represent, in planar view, a depth of approximately 30 µm. Each row of squares was classified according to the extent and position of the deletion (right), as described in the text. Maps do not include the dorsolateral path between somites where crest cells advance more slowly (Erickson et al., 1992).
ticular derivative at any given stage (Table 1). For example, neural crest cells that emigrate early contribute to both sensory and sympathetic derivatives. Cells labeled at stages 16 or 17 lie both in the sympathetic region and in the condensing sensory ganglion (Fig. 2A,B). Similarly, cells that emigrate late invade the dorsolateral path at stage 21 but continue to contribute to sensory ganglia (Fig. 2C). Labeled cells are found at both sites even after injections at late stages (Fig. 2D). Crest cells are thus unlikely to be specified for different destinations according to the time they emigrate. Transient expression during the delay To determine if delayed invasion correlates with specific glycoconjugate expression, we compared patterns of neural crest migration and PNA-binding activity. To assure that different labeling intensities were due to developmental changes rather than to protocol variations, the sections shown in Fig. 3 were processed together using the same reagents and were photographed and printed using identical procedures. The panels for each stage show hindlimb and forelimb levels from the same embryo. Forelimb levels are developmentally advanced and display the same sequence of events that hindlimb levels display later. PNA-binding molecules are transiently expressed in a temporal pattern that correlates with the transient delay in migration. PNA binds extensively to the dorsolat-
eral path before crest cells enter and then declines as crest cells enter. At stage 18 (Fig. 3A,B), PNA intensely labels the dorsolateral path at the hindlimb level but has already begun to decline at the forelimb level. At this stage, crest cells are found only along the ventral path at both levels. By late stage 19 (Fig. 3C,D), PNA binding in the dorsolateral path has declined only slightly at the hindlimb level where crest cells remain confined to the ventral path. In contrast, PNA binding has declined markedly at the forelimb level where crest cells have begun to
Fig. 2. Embryos injected with DiI at successive stages and fixed 18 hours later. Curved arrows indicate crest cells along the ventral path; arrows indicate crest cells along the dorsolateral path (between lines). (A) Crest cells labeled at stage 16 have migrated ventrally to sympathetic and sensory ganglia positions but have yet to enter the dorsolateral path at stage 19. (B) Crest cells labeled at stage 17 lie within ventral roots (v) and sensory ganglia (g) but none have entered the dorsolateral path at stage 20. (C) Crest cells labeled at stage 18 still contribute to sensory ganglia but also begin to enter the dorsolateral path at stage 21. (D) Crest cells labeled at stage 19 still contribute to sensory ganglia while others advance within the dorsolateral path at stage 22. These and subsequent micrographs are fluorescence micrographs of frozen cross-sections. Scale bars, 100 µm; n, neural tube.
106
R. A. Oakley and others Fig. 3. Double label shows that delayed entry into the dorsolateral path correlates with transient expression of PNA-binding sites. At each stage, the developmentally younger hindlimb (left; A,C,E,G) and forelimb (right; B,D,F,H) levels are shown from the same embryo. Arrows: crest cells (orange fluorescence) in the dorsolateral path (between lines). Curved arrows: crest cells in the ventral path. Arrowheads: PNA-binding activity (green fluorescence) in the dorsolateral path. (A,B) Stage 18. Crest cells have yet to enter the dorsolateral path at either level. PNA-binding activity is high in the dorsolateral path at the hindlimb level (A) and has begun to decline at the forelimb level (B). (C,D) Stage 19. Crest cells have yet to enter the dorsolateral path at the hindlimb level (C) where PNA-binding activity remains high, but are beginning to explore the most proximal path at the forelimb level (D) where PNAbinding activity is greatly diminished. (E,F) Stage 21. Crest cells begin to enter the dorsolateral path at hindlimb levels (E) where PNAbinding activity has declined markedly, although patches of bright staining remain (arrowhead). An isolated crest cell (curved arrow) is intercalated between myotome (m) and dermatome (d). Crest cells have entered and migrated along the dorsolateral path at the forelimb level (F) where PNA-binding activity is barely detectable. (G,H) Stage 22. Crest cells migrate along the dorsolateral path at hindlimb levels (G) where PNA is greatly diminished except in the distal path (arrowhead). At the forelimb level (H), PNA-binding activity is barely detectable and crest cells have progressed distally along the dorsolateral path. Scale bar, 100 µm. g, sensory ganglia; n, neural tube.
invade the dorsolateral path. By stage 21 (Fig. 3E,F), PNA binding has declined at the hindlimb level as crest cells enter the path, although patches of intense staining remain and give the path a mottled appearance. At the forelimb level, PNA binding is barely detectable as crest cells advance distally. By stage 22 at both levels (Fig. 3G,H), crest cells advance distally and PNA binding is barely detectable except in the far distal reaches of the path and in a thin line along the epithelial dermatome. As PNA binding first begins to decline, crest cells may abortively inspect the dorsolateral path. For instance, during stage 20 at the hindlimb level, isolated cells occasionally lie within
the most proximal border of the path (e.g., Fig. 6A) and intercalated within the medial tip of the dermatome (not shown). Such exploratory cells were not seen earlier when PNAbinding activity was higher. Crest cells can thus enter this path before overt invasion, but do not advance distally. Therefore the path does not inhibit migration absolutely, but only inhibits relative to other choices. C6S epitopes are also transiently expressed in the dorsolateral path before crest cell entry. The temporal changes in C6S expression are essentially identical to those documented for PNA-binding sites. At stages 18 and 19 when all crest cells migrate ventrally at the hindlimb level, C6S immunoreactivity
Glycoconjugates and neural crest cells
107
Table 1. Neural crest migration assessed with DiI at the hindlimb level n
Stage injected
Stage fixed
Time elapsed
dNT
sNT
DAS/ DRG
VR
Sym
DLP
3 3
16 16
18 19
12 hours 18 hours
+++ +++
+++ +++
+ +++
− +
− +
− −
3 3
17 17
19 20
12 hours 18 hours
+++ +++
+++ +++
+++ +++
+ +++
+++ +++
− −
7 4
18 18
20 21
12 hours 18 hours
+++ +++
+++ +++
+++ +++
+ +
− −
− +
4 5
19 19
21 22
12 hours 18 hours
++ ++
++ ++
+ +
+ +
− −
+ +++
5 3
20/21 20/21
22/23 24
12 hours 18 hours
+ +
+ +
+ +
− −
− −
+ +
3
23/24
25
12 hours
−
−
−
−
−
−
The location of neural crest cells labeled by injecting DiI into embryos that were fixed 12 or 18 hours later. Pluses indicate the relative abundance of labeled cells in each position. Minus indicates no labelled cells. Examples are shown in Fig. 2. Abbreviations: dNT: dorsal to the surface of the neural tube sNT: space between the somite and the neural tube DAS: dorsal-anterior sclerotome DRG: dorsal root ganglion VR: ventral roots and spinal nerves Sym: sympathetic ganglion region DLP: dorsolateral path
intensely labels the dorsolateral path (Fig. 4A,B). By stage 20, the staining intensity has declined dramatically with only patches of immunoreactivity remaining (Fig. 4C). By stage 22, C6S immunoreactivity is faint and restricted to the distal reaches of the path and the dorsal surface of the epithelial dermatome (Fig. 4D). Transient expression of both C6S immunoreactivity and PNA-binding activity thus correlates with transient inhibitory function in the dorsolateral path. Dermamyotome deletions abolish glycoconjugate expression and allow precocious migration Following complete dermamyotome deletions, inhibitory function and glycoconjugate expression are lost coordinately. Crest cells enter precociously and both PNA-binding and C6S immunoreactivity are lost, as shown in cross sections at the hindlimb level (Fig. 5). On the control side at stage 21, patchy regions of PNA-binding activity remain and crest cells have just begun to enter the path. In contrast on the operated side, dermatome is absent, PNA-binding activity is undetectable and crest cells have migrated precociously (Fig. 5A,B). Similarly, C6S immunoreactivity is undetectable in the dorsolateral path where dermatome is absent (Fig. 5C). Dorsal maps graphically convey the relation between migration and PNA-binding activity after surgeries (Fig. 6). On control sides, crest cells migrate distally as PNA-binding declines with development. On operated sides, the path is PNA negative in regions where dermatomes are absent. Crest cells penetrate these regions precociously. For instance at stage 20 (Fig. 6A), crest cells on the control side explore the most proximal portion of the path as the local intensity of PNAbinding diminishes. In contrast, crest cells on the operated side invade the PNA-negative regions where dermatomes are absent. Even at stage 21 (Fig. 6B) when crest cells on the control side have invaded the path, crest cells on operated sides penetrate farther distally in PNA-negative regions. The difference in distal penetration becomes less extreme by stages 21.5 and 22 when PNA-binding has diminished on control sides
(Fig. 6C,D). Even at these later stages, remnants of dermatome retain PNA-binding activity and slow local migration. The elimination of glycoconjugate expression does not, however, correlate tightly with the onset of precocious migration. PNA-binding activity is abolished rapidly, but crest cells do not enter the path immediately. Paths are PNA negative by stage 19, just 10-12 hours after the surgery, and yet crest cells fail to enter until stage 20 (n=7, not shown). A previous study suggested that migration into this path might be controlled by a gateway effect in which only the medial path inhibits entry (Erickson et al., 1992). This argument was based on an impression that surgeries hastened only the entry of cells. In the present study, we analyzed migration quantitatively to determine how surgeries altered both the entry and the distal advance of crest cells. This analysis shows that regions of complete deletion enhance both entry and advance. At stage 20, when only a few crest cells have begun to explore the path on the control side, significantly more crest cells populate the first 200 µm of the path on the deleted side (Fig. 7A). Moreover, cells precociously enter the more distal path in greater numbers. By stages 21 and 22, the most advanced crest cells have penetrated far distally on both sides. However on the operated sides, the number of distal cells is significantly greater (Fig. 7B,C), showing that crest cells also advance precociously within the distal pathway. Deletions do not recruit abnormally large densities of cells. Instead, crest cells on both sides progressively reach a maximum density of 16-20 cells per unit volume as they travel distally. Control and operated sides both reach this density in the first 100 µm of the path by stage 21 (Fig. 7B). By stage 22, crest cells on the operated sides have achieved this maximum density in both the first and second 100 µm units of the path. Crest cells that colonize precociously thus do so at normal densities. PNA-positive remnants inhibit advance locally Partial deletions let us characterize inhibitory function in
108
R. A. Oakley and others
Fig. 4. Delayed migration correlates with transient expression of C6S epitope. Micrographs portray hindlimb levels and are matched for exposure time. Arrowheads indicate C6S immunoreactivity in the dorsolateral path (between lines). (A) At stage 18, C6S immunoreactivity is high in the dorsolateral path. (B) At stage 19, C6S immunoreactivity begins to decline in the path. (C) By stage 20, C6S immunoreactivity has declined markedly although patches of bright staining (arrowhead) remain. (D) At stage 22, as neural crest cells migrate within the path, only weak C6S immunoreactivity remains and it is confined to a small strip lining the medial edge of the epithelial dermatome (d) and the far distal path (arrowhead). Scale bar, 100 µm. n, neural tube.
Fig. 5. Complete dermamyotome deletion allows precocious migration and eliminates both glycoconjugates in the dorsolateral path (between lines). Double arrowheads demark deletion sites. A,B show the same section at a stage 21 hindlimb level. (A) On the control side (right), a single crest cell (arrow) explores the most proximal path. In contrast, on the left side where dermatome is absent, crest cells (arrows) entered the path precociously and migrated distally. (B) On the control side, patchy PNA-binding activity (arrowhead) remains but on the operated side PNA-binding is undetectable in the path. (C) Similarly, C6S immunoreactivity (arrowheads) remains intense on the control side (right) of a stage 20 embryo but is abolished on the left, where dermatome is absent. Scale bar, 100 µm. n, neural tube.
detail. Moreover, since dermatome remnants retain both PNAbinding activity (Figs 6, 8) and C6S immunoreactivity (not shown), we were also able to determine how closely glycoconjugate expression correlates with inhibition. Local inhibition at the entrance to the path suffices to prevent precocious entry as shown by cross-sections that contrast medial and lateral deletions (Fig. 8). In the absence of the medial dermatome, crest cells enter the path precociously and penetrate to the edge of the PNA-positive lateral remnant
Glycoconjugates and neural crest cells
109
Fig. 6. Dorsal maps show the distribution of crest cells and PNA-binding activity in the dorsolateral path on operated (left) and control (right) sides at low thoracic/upper lumbar levels. On operated sides, dotted lines indicate segment boundaries and continuous lines indicate dermatome remnants. Three levels of PNA-binding activity are portrayed by shading. Dark: relatively high binding typical of the period when PNA-binding first begins to decline (as in Fig. 3C). Light: diminished and patchy binding (as in 3E). White: greatly diminished or undetectable binding (as in 3H). Maps omit the rim of binding on epithelial dermatome. Remnants generally displayed more PNAbinding activity than complementary control regions. (A) At stage 20, PNA-binding is beginning to decline on the control side as neural crest cells begin to explore the proximal path. On the operated side, many crest cells have precociously entered and advanced distally in regions with complete deletions. (B) On the control side at stage 21, PNA-binding activity has diminished and become patchy throughout much of the path as crest cells enter. Crosses indicate crest cells intercalated within dermatome or myotome. On the operated side, crest cells have advanced far distally in regions of complete deletions which are PNA negative but few lie dorsal to dermatome remnants which retain high PNA-binding activity. (C) On the control side at stage 21.5, PNA-binding activity remains only in the most distal path. While a few crest cells have advanced distally, the majority lie proximally. On the operated side, crest cells have reached similar distal positions but have more heavily colonized proximal and central regions. (D) On the control side at stage 22, PNA-binding activity is undetectable except at the distal edge of the path. Cells now colonize the central path in modest numbers. In contrast, on the operated side, many crest cells have colonized both medial and central regions and a few have nearly reached the distal edge of the path. Scale bar, 100 µm. Anterior is toward the top.
(Fig. 8A,B). In contrast, in the presence of a PNA-positive medial remnant, crest cells fail to enter the path precociously (Fig. 8C,D). Despite the equivalent size of these deletions, only the medial deletion allows premature entry. Premature entry is thus not caused by the surgical insult alone or by long-distance effects. Moreover, entry is unrelated to the size of the remnant, suggesting that inhibition is not due to a threshold of some inhibitory substance produced by a minimum volume of tissue. Instead, PNA-binding activity at the path’s entrance correlates with delayed entry. To determine whether lateral as well as medial portions of the path inhibit advance, we quantitatively analyzed deletions that left remnants at different sites. Remnants are usually irregular and fail to align with the unit volumes that we used to quantitate migration (Fig. 1). To assure that a unit would lie completely within a particular class of deletion, we divided the path roughly into thirds along the medial to lateral axis. For
example in medial deletions, only the medial one-third of the dermatome was absent and the remnant did not intrude into the most medial unit of the path (cartoon in Fig. 9A). In all cases, the remaining dermatome extended over the full anteriorposterior extent of the row and did not afford crest cells an open avenue distally. This analysis shows that all remnants inhibit migration locally, regardless of their position (Fig. 9). After a medial deletion, migration is enhanced only proximally (Fig. 9A). Similarly, deleting both medial and central dermatome confers additional migratory advantage. Crest cells penetrate farther distally than on control sides. Moreover, the density of crest cells increases in the central region, which is now free of dermatome (Fig. 9B). The larger size of combined medial and central deletions is insufficient to explain enhanced migration. Central and lateral deletions of equivalent size fail to stimulate migration (Fig. 9C). Inhibition is therefore local. The local
110
R. A. Oakley and others
A. Stage 20 and 20.5 20 prox
mean # cells
16 -
dis
*
a
b
c
d
12 operated side control side
8-
*
4-
a
b
ns x c
ns xx d
successive unit volumes
n=7
B. Stage 21 and 21.5 20 ns
mean # cells
16 -
* 12 8-
*
nature of the inhibition is particularly evident when both medial and lateral dermatome is absent but central dermatome Figure 7 remains (Fig.et9D). Oakley al. Migration is enhanced only in the region of the medial deletion; migration is not enhanced laterally, beyond the intervening remnant. Migration is thus accelerated only when the crest cells have physical access to a deletion site. While dermatome remnants inhibit migration locally, the inhibition is relative rather than absolute. Even remnants with high levels of PNA-binding activity do not prohibit migration completely. For instance, at stages 20-21.5 remnants bind PNA extensively and yet a few crest cells lie in the path between the remnant and the ectoderm (Fig. 6A-C). To analyze the relative inhibition, we compared numbers of crest cells in both regions. We selected remnants of one size class (30-50 µm wide) from stages in which crest cells had progressed far enough distally to encounter the remnant. We drew a mirror image of each remnant on an adjacent region of complete deletion to assure that we compared equivalent volumes (cartoon in Fig. 10). Significantly fewer cells lay dorsal to these remnants (Fig. 10). Thus, crest cells more heavily colonize adjacent, PNAnegative regions, as though they turned to avoid PNA-positive regions.
4-
* a n=24
b
c
successive unit volumes
C. Stage 22 and 22.5 20 -
ns
*
mean # cells
16 -
*
12 8-
*
4-
a n=8
b
c
DISCUSSION
d
d
successive unit volumes
Fig. 7. Numbers of advancing crest cells are compared in complete dermamyotome deletions and controls. Graphs show the mean number of crest cells in each successive unit of the path. Cartoon in A indicates the borders of successive units (lines between a,b,c,d) along the proximal (prox) to distal (dis) axis of paths. Black bars: operated sides; stippled bars: control sides. An “x” indicates that no cells had entered the unit in any samples. (A) During stage 20, significantly more crest cells have advanced into the path on operated sides. (B) On both sides during stage 21, the most advanced crest cells have migrated to similar distal extents and crest cells have reached their maximum density in the most proximal compartment (a). However, on the operated side, significantly more cells colonize the three distal compartments (b,c,d). (C) During stage 22 on operated sides, crest cells have reached maximum density in the two most proximal compartments (a,b) and significantly more cells lie in the three distal compartments (b,c,d). Error bars: SEM. Asterisks indicate P
