Download Contact guidance of CNS neurites on grooved quartz: influence of

2905
Journal of Cell Science 110, 2905-2913 (1997)
Printed in Great Britain © The Company of Biologists Limited 1997
JCS3646
Contact guidance of CNS neurites on grooved quartz: influence of groove
dimensions, neuronal age and cell type
Ann M. Rajnicek1,*, Stephen Britland2,3 and Colin D. McCaig1
1Department of Biomedical Sciences, Institute of Medical Sciences, University of Aberdeen, Aberdeen AB25 2ZD,
2Department of Electronics and Electrical Engineering, University of Glasgow, Glasgow, G12 8QQ, Scotland, UK
3Postgraduate Studies in Pharmacology, School of Pharmacy, University of Bradford, Bradford BD7 1DP, UK
Scotland, UK
*Author for correspondence (e-mail: [email protected])
SUMMARY
We used an in vitro system that eliminates competing
guidance cues found in embryos to determine whether substratum topography alone provides important neurite
guidance information. Dissociated embryonic Xenopus
spinal cord neurons and rat hippocampal neurons were
grown on quartz etched with a series of parallel grooves.
Xenopus neurites grew parallel to grooves as shallow as 14
nm and as narrow as 1 µm. Hippocampal neurites grew
parallel to deep, wide grooves but perpendicular to shallow,
narrow ones. Grooved substrata determined the sites at
which neurites emerged from somas: Xenopus neurites
sprouted from regions parallel to grooves but presumptive
axons on rat hippocampal neurons emerged perpendicular
to grooves and presumptive dendrites emerged parallel to
them. Neurites grew faster in the favored direction of ori-
entation and turned through large angles to align on
grooves. The frequency of perpendicular alignment of hippocampal neurites depended on the age of the embryos
from which neurons were isolated, suggesting that contact
guidance is regulated in development. Collectively, the data
indicate that substratum topography is a potent morphogenetic factor for developing CNS neurons and suggest that
in addition to a role in pathfinding the geometry of the
embryo assists in establishing neuronal polarity. In the
companion paper (A. M. Rajnicek and C. D. McCaig (1997)
J. Cell Sci. 110, 2915-2924) we explore the cellular
mechanism for contact guidance of growth cones.
INTRODUCTION
topography, further suggesting that purely physical cues can
direct neurite growth within embryos.
In an effort to understand embryonic contact guidance events
in vitro many cell types, including fibroblasts (Elsdale and
Bard, 1972) and neurons (e.g. Ebendal, 1976) have been
cultured on parallel arrays of aligned collagen fibrils or on
parallel grooves manufactured in artificial substrata (Hirono et
al., 1988; Clark et al., 1990; Oakley and Brunette, 1993). In
most cases cells exhibit conventional contact guidance,
aligning parallel to the groove/ridge axis, but even apparently
simple contact guidance cues provide complex directional
information to cells. For example, fibroblasts migrate unidirectionally on aligned collagen matrices (Boocock, 1989)
and, depending on species and neuronal type, mammalian CNS
neurites grow either parallel or perpendicular to aligned neurite
bundles (Nagata and Nakatsuji, 1991) or artificial microstructures (Nagata et al., 1993).
Contact guidance of cells by physical contours of the substratum was recognized in the earliest days of tissue culture
(Harrison, 1914) yet surprisingly little is known about the
cellular events of contact sensing and their transduction into
directional growth, especially in neuronal growth cones. The
aim of the present study was to describe the directional effects
of substratum contours on the morphology of developing CNS
neurons as a first step towards elucidating the mechanism for
Contact guidance, the phenomenon by which the physical
shape of the substratum induces alignment or directional
growth of cells is involved in normal embryonic pattern
formation (e.g. Newgreen, 1989). Nervous system development is influenced strongly by geometric patterns present
within embryos during developmentally significant periods.
For example, radial glia that span the walls of the developing
brain are organized into regular arrays that foreshadow the
route of migrating embryonic neurons (Hatten, 1990). In
addition to a role in directional migration of cell bodies
aligned cells also appear to guide process outgrowth. In the
mammalian forebrain a ‘sling’ of subventricular cells forms a
scaffold that predicts the trajectory of the corpus callosum
(Schneider and Silver, 1990). Aligned channels that are subsequently invaded by pioneer axons exist in mouse (Silver and
Sidman, 1980) and chick (Krayanek and Goldberg, 1981)
optic stalk and dorsal retina. Similarly, the earliest fiber tracts
in the Xenopus spinal cord form by ingrowth of axons into
preexisting longitudinally oriented spaces between neuroepithelial cells of the neural tube (Nordlander and Singer,
1982a,b). Electron microscope studies of Xenopus embryos
(Roberts and Taylor, 1983) demonstrate that even on a small
scale growth cone morphology is influenced by substratum
Key words: Topographic guidance, Growth cone, Hippocampal
neuron, Xenopus neuron, Neuronal development, Substratum
2906 A. M. Rajnicek, S. Britland and C. D. McCaig
topography-induced growth cone guidance. In the companion
paper we explore the cellular mechanisms for contact guidance
of growth cones (Rajnicek and McCaig, 1997b).
MATERIALS AND METHODS
Substrate preparation
Microgrooved substrates were prepared from fused quartz using a
direct writing electron beam lithographic process to etch a series of
repeating, parallel grooves and ridges that were evenly spaced and had
nearly right angle edges (Fig. 1A). The technique offers submicron
precision in controlling grating periodicity beyond that attainable by
conventional photolithographic microfabrication (Clark et al., 1990;
Britland et al., 1996a). Eight experimental culture substrata, each
comprising three 5 mm × 5 mm blocks of gratings with identical
groove depths but different groove widths (1, 2 and 4 µm, respectively), were fabricated on one electron beam mask plate (Hoya,
Japan). Each 6.5 cm square plate was composed of a 2 mm thick fused
quartz base, a 100 nm chrome layer and 500 nm of EBR9 electron
beam resist. The grating pattern was designed in WAM (an in-house
microelectronic design package) and then translated into the BWL
beam writing language format used by a Leica EBG5 Beamwriting
machine which was operated with a 200 nm spot size at 50 kV. Beam
current was automatically determined by spot size and resolution.
Exposed resist was then developed using neat methyl iso-butyl ketone
at room temperature and the underlying chrome removed by wet
etching to reveal the quartz. The quartz base was then dry etched
(RIE80, Plasma Technology, UK) using CHF3 plasma at 29 sccm
(standard cubic centimeters per minute) flow rate and 15 mTorr and
100 W (rf) at 13.6 Mhz to give an etch rate of approximately 50
nm/minute. After removal of the remaining resist/chrome using
acetone/chrome etch the substrates were blanket etched in CHF3 for
30 seconds using the parameters outlined above. Finally the mask
plate was cut into 8 individual microscope slides using a diamond saw
and the etch depth measured using Dektak and Talystep surface
profilers. Two of the substrates were used for SEM and the remainder
were used repeatedly for experimental purposes.
Xenopus neurons were plated directly onto slides previously soaked
overnight in concentrated nitric acid followed by several rinses with
distilled water. Since these neurons prefer a minimal volume of growth
medium, spacers made from narrow (~1 mm wide) strips of no. 1
thickness coverglass were secured to the long edges of the slides with
MS4, a non-curing silicone compound (Dow Corning). Cells were
plated into a small pool of culture medium in the central trough. After
allowing ~30 minutes for the cells to adhere to the slide a coverslip that
covered the entire quartz slide was secured on top to minimize evaporation. Slides were maintained at room temperature in a humidified
atmosphere during the subsequent 6 to 9 hour growth period.
For hippocampal neuron experiments the acid-cleaned slides were
sterilized by UV irradiation and were then coated for 1 hour with 1
mg/ml poly-L-lysine (30-70 kDa; dissolved in sodium borate buffer,
pH 8.5, immediately before use). Slides were then rinsed with sterile
water and immersed grooved side up in a 60 mm Petri dish containing enough medium to completely submerge the slide (~4 ml). Cells
were plated onto the slide without an overlying coverslip.
For one experiment hippocampal neurons were plated onto polylysine-treated polystyrene replicas of etched surfaces (see photo in
Fig. 5) completely submerged in culture medium. This surface
comprised a series of adjoining 4 sided wells separated by narrow
(~2 µm) walls with defined heights ranging from 0.118 µm to
3.16 µm. Neurites that stepped up, crossed over, then stepped down
into the adjacent well were considered to have ‘crossed the step’.
Cell culture
Dissociated neurons from the neural tube of Xenopus laevis were
cultured using a slight modification of a previously described culture
method (Hinkle et al., 1981; McCaig et al., 1994). Neural tubes were
dissected from several stage 20 Xenopus neurulae and pooled in
calcium-magnesium-free Steinberg’s solution, pH 7.9, to disaggregate. The resulting cell suspension was triturated gently and then
plated into a pool of culture medium in the central region of the slides
described above. Culture medium was Steinberg’s solution (58 mM
NaCl, 0.67 mM KCl, 0.44 mM Ca(NO3)2, 1.3 mM MgSO4 and 4.6
mM Tris-HCl, pH 7.9) supplemented with 20% Leibowitz’s L-15, 1%
fetal bovine serum, 100 i.u./ml penicillin, 100 µg/ml streptomycin (all
from ICN Biochemicals Ltd, Irvine, Scotland). Data were collected
after 6 to 9 hours in culture at room temperature. Dissociated neurons
from the E16 (and E19 for studies of age dependence) rat hippocampus were cultured in Earle’s minimal essential medium (MEM, Gibco)
as described previously (Davenport and McCaig, 1993). After 3 hours
the MEM containing 10% heat inactivated calf serum was replaced
with MEM containing 10% Nu Serum V (Collaborative Research,
Lexington, MA, USA). Data were collected from live cells after 24
hours in culture at 36°C, 5% CO2. Hippocampal cells were used fresh
or were thawed from stocks frozen at −70°C in 8% DMSO (dimethylsulfoxide) by the method of Mattson and Kater (1988). Data from
fresh and frozen cells were combined because there were no differences in their morphologies or directional responses.
Data collection and analysis
Measurements were made from live video images of cells using Leica
Quantimet 500MC image collection and analysis software. For
Xenopus neurons data were collected from each neurite but in the case
of hippocampal neurons data were only collected from the longest
neurite on each cell, which is the presumptive axon (Dotti et al., 1988).
The angle of neurite orientation relative to the groove direction (θ)
Fig. 1. (A) Scanning electron micrograph of a quartz microscope
slide with grooves 520 nm deep and 2 µm wide. Note that the
grooves have approximately right angled sides, sharp corners and are
evenly spaced. The lower photo is a magnified view of the boxed
region. (B) Angle measurement protocol for orientation assay.
Angles were measured using computerized image collection and
analysis software that allowed lines to be superimposed onto video
images of cells. Images were always viewed with the groove/ridge
axis horizontal. The direction of neurite growth was determined by a
line connecting the neurite initiation site on the soma with the center
of the growth cone. All angles measured were between 0° and 90°.
Overall orientation for a population of neurites was determined by
categorizing the angle of growth as ‘parallel’ to the grooves if θ was
between 0° and 30° and ‘perpendicular’ if θ was between 60° and
90°. The mean percentage of neurites in each category was
determined by pooling data for several sets of 50 cells grown under
identical conditions.
CNS neurite orientation on grooves 2907
was on grooves 1,100 nm deep and 4 µm wide. Orientation was
not biased on flat quartz for either cell type. Most non-neuronal
cells in hippocampal cultures aligned parallel to grooves of all
dimensions. Grooved quartz slides were treated with polylysine before plating hippocampal neurons but not Xenopus
neurons so we tested the notion that differences in orientation
between the neuronal types was an artifact of polylysine
treatment. Two lines of evidence argue that the differences are
not related to polylysine treatment: (1) Xenopus neurites
oriented parallel to grooves even when the slides were treated
with polylysine (data not shown). The equivalent experiment,
in which hippocampal neurons are grown on untreated quartz
is impossible because polylysine treatment is required for hippocampal neuron differentiation. (2) Rat hippocampal neurites
aligned either parallel or perpendicular to grooves depending
on groove dimensions even though the substrates were always
treated with polylysine.
Parallel growth of neurites is not merely due to physical constraint within the grooves because individual growth cones
often span several groove and ridge repeats (Figs 3D, 4A) and
are therefore not confined by groove walls. Similarly, hippocampal neurites are able to cross single steps as high as 3.16
µm (Fig. 5) and would therefore be able to step out of grooves
even deeper than those used in this study. Extrapolation of the
data in Fig. 5 indicates that grooves would have to be 4.7 µm
high to completely restrain hippocampal neurites. These data
suggest that neurites make an active choice to grow either
parallel or perpendicular to grooves. This idea is also supported
by the observation that hippocampal neurites sometimes grew
parallel to grooves before turning to grow across them (Fig.
4B).
was determined as shown in Fig. 1B. Control (flat) data were collected
from cells on the ungrooved regions of each etched slide.
Angles were considered to be ‘parallel’ to the groove direction if θ
was between 0° and 30° and angles were ‘perpendicular’ if θ was
between 60° and 90°. The remainder of the neurites were classified
as ‘intermediate’ and for the sake of clarity are not included in this
report. The expected frequency in each category for a population of
randomly oriented neurites is 33%. This was confirmed by measurements of the neurites grown on flat quartz surfaces. The percentage
of neurites in each category was therefore compared to the control
percentage for flat substrata using a d-test (Bailey, 1981) or one way
ANOVA followed by a Dunnet multiple comparison test.
RESULTS
The direction of CNS neurite orientation depends on
cell type and groove dimensions
Dissociated neurons from embryonic Xenopus spinal cord and
embryonic rat hippocampus revealed strikingly different
growth patterns on grooved surfaces (Figs 2, 3). Xenopus spinal
neurites exhibited classical contact guidance by growing
parallel to grooves with depths ranging from 14 nm to 1,100
nm and widths of 1, 2 or 4 µm (Fig. 2A). By contrast, hippocampal neurites showed a more complex response. In
general they grew perpendicular to shallow, narrow grooves
and parallel to deep, wide ones (Fig. 2B). At certain groove
depths (130 nm, 520 nm, and 1,100 nm) hippocampal neurites
changed their direction of growth from perpendicular on
grooves 1 µm and 2 µm wide to parallel on grooves 4 µm wide
even though the groove depth was unchanged. The greatest perpendicular response for hippocampal neurites was on grooves
130 nm deep and 1 µm wide and the greatest parallel response
A
Xenopus
spinal neurites
parallel
100
% neurites
80
1 µm
*
B
perpendicular
**
parallel
**
**
**
100
*
*
20
40
**
**
**
**
**
**
*
**
**
*
**
0
14
36
2 µm
**
100
% neurites
**
20
**
**
0
**
130
140
520
**
1100
**
14
**
**
100
36
60
40
**
**
**
**
520
**
**
1100
** **
*
40
**
140
2 µm
60
**
130
80
0
*
20
0
14
36
*
4 µm
**
100
80
% neurites
1 µm
60
40
20
perpendicular
80
60
80
rat hippocampal neurites
130
140
**
**
520
**
1100
**
14
100
4 µm
80
60
36
130
140
520
**
**
**
1100
60
40
40
**
20
**
**
**
**
0
*
20
*
0
14
36
130
140
groove depth (nm)
520
1100
14
36
130
140
groove depth (nm)
520
1100
Fig. 2. Orientation responses of (A)
Xenopus spinal cord neurites and (B) rat
hippocampal neurites on grooved substrata.
The direction of neurite growth was
measured as shown in Fig. 1B and is
presented as the mean percentage of total
neurites parallel (white bars) or
perpendicular (black bars) to the groove
direction ± sd. The groove width is
indicated in the upper left of each graph.
Data were pooled from at least two
experiments on each substrate. The
number of Xenopus neurites on grooves
ranged from 93 (on 130 nm deep, 4 µm
wide grooves) to 188 (on 140 nm deep, 1
µm wide grooves). The number of
hippocampal neurites ranged from 100 (on
14 nm and 520 nm grooves) to 400 (on 130
nm deep, 1 µm wide grooves). Asterisks
represent values significantly different
from the measured control frequencies of
34±3% parallel and 33±3% perpendicular
for Xenopus (n=300) and 34±6% parallel
and 34±5% perpendicular for hippocampal
neurites (n=750) on flat quartz (*P<0.05,
**P<0.01 by one way ANOVA). These
control frequencies were not different from
33% (d-test, Bailey, 1981), which is the
expected freqency of parallel or
perpendicular neurites in a randomly
oriented population.
2908 A. M. Rajnicek, S. Britland and C. D. McCaig
Fig. 3. Phase contrast micrographs of E16 rat hippocampal neurons
(A and B) and Xenopus spinal cord neurons (C and D). Rat and
Xenopus neurons had been growing for 24 and 4 hours, respectively.
Cells are growing on flat quartz (A and C) or on grooves 1 µm wide
and 130 nm deep (B) or 320 nm deep (D). In phase contrast
micrographs of grooved surfaces the ridges appear as phase dark
stripes and the grooves appear as phase bright stripes. Bars: (A,B) 50
µm; (C,D) 100 µm.
Grooved substrata determine the sites at which
neurites emerge from somas
Neurons were grown on grooved substrates to test whether
topographical features of the substratum determined aspects of
neuronal morphology even more subtle than the direction of
overall neurite extension. In particular we examined the
direction of neurite initiation, direction of turning and the rate
of neurite elongation. Grooved substrata determined the site on
the soma that gave rise to neurites. Neurites were uniformly
distributed on flat quartz or on grooves 14 nm deep but most
Xenopus neurites emerged from regions of the somas parallel
to grooves greater than 36 nm deep (Figs 6A, 7C). Hippocampal neurites showed a more varied response (Fig. 6B). One day
after plating hippocampal neurons generally bear one long
process that becomes the axon and several shorter, minor
processes that subsequently become dendrites (Dotti et al.,
1988). By these criteria most presumptive axons emerged from
perpendicular regions and presumptive dendrites emerged from
parallel regions of hippocampal somas on grooves. Presumptive dendrites maintained parallel growth as they extended (Fig.
3B). This implies that the physical contour of the environment
influences the fine structural, and therefore functional, polarity
of differentiating hippocampal neurons.
Fig. 4. Neurites turn in relation to the groove direction. (A) Phase
contrast micrograph of Xenopus neurites on grooves 14 nm deep and
1 µm wide. Cells were plated 8 hours previously. At least three
neurites (arrows) turned through large angles to grow parallel to the
grooves. In one case (double arrows) two neurites appear to have
fasciculated yet they still turn to be parallel. Note that neurite
initiation sites are not biased by the 14 nm grooves. Bar, 50 µm.
(B) Differential interference micrograph of a hippocampal neuron
whose cell body rests on flat quartz (left side of photo). As the
neurite encounters the grooved region of the slide it turns abruptly
(arrow) to become perpendicular to the 140 nm deep, 1 µm wide
grooves. Bar, 50 µm.
Neurites turn to grow in the preferred orientation
Neurites of both cell types often adjust their trajectories by
turning toward their preferred direction of orientation. For
example, Xenopus neurites on substrates with grooves 14 nm
deep and 1 µm wide are symetrically distributed around the
soma (Figs 4A, 6A) yet they demonstrate parallel contact
guidance on these grooves (Fig. 2A). This suggests that
Xenopus neurites change direction as they grow in order to
align parallel to grooves. Observations of single neurons
support this notion (Figs 4A, 8).
Although most presumptive axons on hippocampal neurons
emerged from regions of cell bodies perpendicular to grooves
(Fig. 6B) the degree of turning relative to the groove direction
was assessed to determine whether they would turn to align
perpendicular to grooves. These observations were made on
grooves 130 nm deep and 1 µm across, which is the substrate
that elicited maximum (85±6%) perpendicular orientation (Fig.
2B). Turning was defined as a deviation of 15° or more by the
distal (growth cone) end of a neurite relative to the proximal
end (nearest the soma) of the same neurite. On flat quartz
neurites turned uniformly in all directions: 32% toward
grooves, 33% away from grooves and 43% do not turn (n=200)
but on grooved surfaces more than twice as many turned away
from the direction of the grooves (46%) than turned toward
CNS neurite orientation on grooves 2909
tively. The etched regions were surrounded by flat regions used
for control measurements. Collecting data from a single slide
excluded the potential effects of variation in substratum surface
chemistry and day to day variation in growth. On grooves 1 or
2 µm wide neurites aligned mostly perpendicular to grooves
(Fig. 2B) and were longer than those aligned parallel on the
same substrate (Fig. 9). On grooves 4 µm wide, however,
neurite orientation was random (Fig. 2B) and the length perpendicular was not different from that parallel or that on flat
quartz (Fig. 9). The mean length of neurites parallel to grooves
(regardless of width) were the same as those on flat quartz.
Neurite length (hence, net growth rate) is therefore increased
selectively in neurites growing perpendicular to grooves.
Neurites grow faster in the preferred orientation on
grooves
In general neurites grew faster in their preferred orientation
than in less favored directions or on flat quartz. Xenopus
neurites growing parallel to 130 nm deep, 1 µm wide grooves
(26±1.6 µm/hour, mean ± s.e.m., n=15) grew significantly
faster (P<0.0001, 2-tailed Student’s t-test) than those
oriented randomly on flat quartz (14±0.6 µm/hour, n=74).
Since all neurites grew parallel to grooves in these time lapse
experiments no comparison could be made with growth rates
of neurites on grooved substrata extending in less preferred
orientations.
Hippocampal neurites showed a similar tendency to grow
faster in the preferred orientation. For example, on the substratum that elicited maximum perpendicular orientation (130
nm deep, 1 µm wide grooves) hippocampal neurites perpendicular to grooves were longer than those parallel to grooves
and those on flat quartz (Fig. 9). Neurites on flat quartz had
uniform lengths regardless of orientation (Fig. 9). Since length
measurements were made 24 hours after plating in each case
this suggests that the net rate of neurite growth was increased
for hippocampal neurites growing across grooves. This conclusion was confirmed by measuring hippocampal neurite
lengths on a slide containing three etched regions with grooves
140 nm deep but widths of 1, 2 or 4 µm in each region, respec-
A
100
% initiation sites
them (16%) and 38% did not turn (n=100). The mean angle
turned away from grooves (35±3°) was greater (P=0.0074, 2tailed t-test) than that turned toward grooves (25±2°). Hippocampal neurites were therefore able to correct their course
away from the groove/ridge direction to grow across, rather
than along grooves. In some cases neurites on flat regions of
the slides happened to grow onto the grooved region, changing
course abruptly to grow perpendicular to grooves (Fig. 4B).
This observation further supports the notion that perpendicular orientation on these surfaces is an active choice.
Embryonic age and species affect perpendicular
alignment of hippocampal neurons
Cells dissociated from rat hippocampi of different embryonic
ages were grown on 130 nm deep grooves to explore the
temporal developmental significance of contact guidance. Perpendicular orientation on 1 or 2 µm wide grooves was greater
for neurons isolated from E16 hippocampi than for those from
E19 hippocampi (Fig. 10). E16 neurites aligned perpendicular
at 1 and 2 µm widths but E19 neurites shifted from perpendicular orientation at 1 µm repeats to parallel orientation at 2
µm repeats (Fig. 10).
Although hippocampal neurons isolated from E16 mice or
rats show predominantly perpendicular orientation on 130 nm
deep, 1 µm wide grooves those isolated from mice are less
Xenopus neurons
80
***
***
***
parallel
perpendicular
60
40
*
20
**
***
0
flat
14
36
140
groove depth (nm)
320
***
520
B
B
% initiation sites
Fig. 5. Hippocampal growth cones crossed ridges as high as 3.16
µm. There is a direct correlation between ridge height and the
frequency of neurites that cross over them (R2 = 0.9898).
Extrapolation of the line indicates that ridges would have to be at
least 4.7 µm high to prevent all neurites from crossing. Asterisks
indicate values significantly different from 100%. *0.05≥P≥0.02;
**P<0.001 (d-test; Bailey, 1981). The number of neurites measured
ranged from 36 to 50 at each height. Inset shows a hippocampal
neuron a substratum with ridges (phase dark lines) 0.27 µm high.
Note that both growth cones have crossed the ridges.
60
hippocampal neurons
***
*
parallel
perpendicular
40
**
20
**
0
presumptive
axons
presumptive
dendrites
flat
presumptive
axons
presumptive
dendrites
grooved
Fig. 6. Effect of substratum contour on the site at which neurites
sprout on cell bodies. All grooves were 1 µm wide. The dashed lines
indicate the expected control frequency for randomly oriented
neurites on flat quartz. Asterisks indicate values that are significantly
different than those on flat surfaces by a d test (Bailey, 1981)
*0.05>P>0.02; **0.02>P>0.01; ***P<0.001. (A) For Xenopus
neurons data were collected 2.5 hours after plating and n was 250 for
flat and 50 for each grooved substrate. (B) For hippocampal neurons
initiation sites were measured 24 hours after plating on 130 nm deep
grooves. Data were pooled from 3 experiments. The longest neurite
on each cell was the presumptive axon and the minor processes were
the presumptive dendrites (Dotti et al., 1988). On flat quartz n=50
presumptive axons and 104 presumptive dendrites; on grooves n=135
presumptive axons and 333 presumptive dendrites.
2910 A. M. Rajnicek, S. Britland and C. D. McCaig
Fig. 8. Xenopus neurons that landed fortuitously at the boundary
between grooved regions (right) and flat regions (left) of the slides
during plating. Grooves are 1 µm wide and 320 nm deep (A) or 520
nm deep (B). Note that for each cell neurites on grooves are
straighter and longer than those on the flat region. Bar, 50 µm.
Fig. 7. Time lapse photos of Xenopus neurons on grooves 140 nm
deep and 1 µm wide. (A) The first photo (T = 0 hours) was taken 4
hours after plating. (B) 2 hours later. (C) 4 hours later. Arrows
indicate branch points that resulted from growth cone bifurcation.
Asterisks in (C) indicate newly differentiating cells. Note that
neurites emerge parallel to the groove/ridge axis. Bar, 50 µm.
likely to exhibit perpendicular alignment. After 24 hours in
culture 81±2% of E16 rat neurites (Fig. 10) but only 50±6%
(mean ± s.d., n=100) of E16 mouse neurites aligned perpendicular to grooves (P=0.003, 2 tailed Student’s t-test). This
reduction could reflect species differences or differences in the
relative rates of mouse and rat hippocampal development
because hippocampal development in the mouse begins earlier
(E10) than in the rat (E15) and the development of Ammon’s
horn is more protracted in the mouse than the rat (Reznikov,
1991). The frequency of perpendicular alignment of E16
mouse neurites was therefore compared to E19 rat neurites,
which may be more similar developmentally. There was no difference (Student’s t-test) between the frequency of perpendicular alignment for E16 mouse neurons (50±6%, n=100)
compared to E19 rat neurons (61±7%, n=150).
DISCUSSION
The nature of various attractive, repulsive, permissive and
inhibitory characteristics of neuronal pathways has received
much attention in the context of growth cone guidance
(reviewed by Keynes and Cook, 1995a,b; Tessier-Lavigne and
Goodman, 1996). It is likely that interplay between multiple
factors is required for axon guidance and correct target recog-
nition. Guidance by substratum contours has been largely overlooked and is usually treated as incidental but evidence that
spaces precede outgrowth of some CNS neurons during development and regeneration (e.g. Singer et al., 1979), that CNS
neurites are guided by radial glial cells (e.g. Norris and Kalil,
1991) and that substratum topography affects growth cone
shape (Harris et al., 1985; Nordlander et al., 1991) suggest collectively that guidance by topographical features warrants
investigation in its own right.
Topographical guidance may be important during hippocampus development because alveolar channels in the
embryonic rat hippocampus are presumed to guide pyramidal
axons (Altman and Bayer, 1990b) and non-pyramidal neurons
in the hippocampus may provide contact guidance for later outgrowing septohippocampal fibers (Supèr and Soriano, 1994).
Additionally, Cajal described hippocampal mossy fibers as
being grooved into the irregularities of the surface of regio
inferior neurons suggesting a relationship between topography
and neurite paths (Blackstad and Kjaerheim, 1961).
We used an in vitro system that eliminated simultaneous presentation of potentially competitive guidance cues to explore
the contribution of the physical shape of the substratum in
neuronal morphogenesis. Our data indicate that substratum
contours provide important morphogenetic information to
developing CNS neurons in an age-dependent way by influencing the site of neurite initiation on somas, the direction of
neurite growth (determined by groove dimensions), the presumptive axonal or dendritic identity of neuronal processes and
the rate of neurite elongation. The increase in net perpendicular growth rate on narrow grooves is selective because neurites
that grow parallel to the same grooves are significantly shorter
than perpendicular ones but no different than those on flat
quartz. The net increase could be due to enhanced trophic
support for neurites on grooves because of the increased
CNS neurite orientation on grooves 2911
100
parallel
perpendicular ***
100
**
80
E16 ***
***
80
*
% neurites
length+sem (µm)
120
60
40
0
perpendicular
***
60
*
**
40
20
20
parallel
E19
***
***
0
flat
130×1
140×1
140×2
140×4
groove depth (nm)×width (µm)
Fig. 9. Rat hippocampal neurites are longer when aligned
perpendicular to grooves than when they are parallel or on flat
substrates. All data were collected 24 h after plating and are
expressed as mean neurite length ± s.d. Astersisks indicate
significant differences in the length of neurites growing
perpendicular compared to those growing parallel on the same
substrate; *P=0.0261, **P=0.0012, ***P<0.0001 (2-tailed Student’s
t-test). Number of neurites measured on each substrate (parallel and
perpendicular, respectively): on flat quartz n=32 and 32, on 130 nm ×
1µm grooves n=18 and 91, on 140 nm × 1 µm grooves n=16 and 68,
on 140 nm × 1 µm grooves n=27 and 49, and on 140 nm × 4 µm
grooves n=30 and 40. At a width of 1 µm there is no difference in the
length of neurites perpendicular to grooves 130 nm deep compared to
those perpendicular on 140 nm deep grooves. Neurite length on
grooves 140 nm deep and 4 µm wide is not different from that on flat
quartz. Neurites oriented parallel to grooves 130 nm deep and 1 µm
wide were shorter (P=0.0327) than parallel neurites on flat quartz.
membrane surface area exposed to nutrient-rich culture
medium. This does not appear to be the case however because
on grooved surfaces that induce random orientation neurite
length is the same as that on flat quartz. This suggests that perpendicular neurite growth rate is stimulated only on grooved
surfaces that stimulate perpendicular orientation.
Alignment on grooved surfaces appears to be an
active process
Neurites changed their direction of growth to reflect their
preferred orientation on grooves. Xenopus neurites monitored
over time adjusted their trajectories, turning parallel to grooves.
Whilst only 16% of hippocampal neurites turned parallel to
grooves that induced perpendicular alignment, 46% turned
away from grooves (and through larger angles), yielding perpendicular orientation. Control experiments ruled out the possibility that perpendicular orientation was related to polylysine
treatment of the growth surface. Xenopus and hippocampal
neurites grew parallel to grooves too small to restrain them
physically, suggesting that parallel orientation is not merely due
to neurites being trapped within grooves. This idea is supported
by the observation that hippocampal neurites climbed over steps
at least 3 times higher than those that induced parallel orientation. These observations suggest collectively that alignment on
grooved substrata is an active rather than passive event.
Contact guidance by neurons in situ
Xenopus spinal neurons and rat hippocampal neurons were used
in the present investigation because evidence from anatomical
studies of Xenopus spinal cord and mammalian brain, including
the hippocampus suggest that surface contours affect the pattern
130×1
130×2
130×1
130×2
groove depth (nm)×width (µm)
Fig. 10. Age dependent differences in orientation on grooves.
Hippocampal neurons were isolated from E16 (left side of figure) or
E19 (right side of figure) rat embryos. Perpendicular contact
guidance was enhanced for E16 neurites compared to E19 (P=0.0089
and P=0.0035 for E16 versus E19 neurons at 1 and 2 µm widths,
respectively by a 2-tailed Student’s t-test). On 130 nm × 1 µm
substrates the number of neurites was 150 at E16 and 150 at E19. On
130 nm × 2 µm the number of neurites was 50 at E16 and 100 at
E19. Asterisks indicate differences between the percentage parallel
or perpendicular compared to a control value of 34% for flat quartz
(dashed line); *0.01<P<0.02, **0.0001<P<0.001, ***P<0.0001 by a
2-tailed Student’s t-test.
of neurite growth. For example, in the Xenopus spinal cord
sensory ganglion neurons grow along tracts of preexisting,
longitudinally aligned Rohon-Beard neurons (Nordlander et
al., 1991) and aligned spaces between neighboring neuroepithelial cells form channels which the earliest axonal outgrowths
subsequently invade (Nordlander and Singer, 1982a,b).
Published electron micrographs (e.g. Fig. 2 of Nordlander and
Singer, 1982a) indicate that the channels range from approximately 0.6 µm to 3 µm across. Even the smallest spaces are
therefore at least five times larger than the minimum groove
depth that induced parallel orientation of Xenopus spinal cord
neurites (our Fig. 2A). The first spaces to appear in the Xenopus
spinal cord form adjacent to the differentiating Rohon-Beard
neurons (Nordlander and Singer, 1982a). Cultures of dissociated neural tubes yield a heterogeneous population of
sensory neurons (e.g. Rohon-Beard neurons), motor neurons
and interneurons (Tabti and Poo, 1991). Since greater than 80%
of Xenopus neurites grew parallel to grooves 130 nm deep and
deeper in our experiments (Fig. 4A) it is probable that the
majority of embryonic spinal cord neurons, including RohonBeard neurons, would align with parallel topographical features
in situ. Indeed, growth cones of Xenopus sensory ganglion
neurons appear to be directed by the geometry of their surroundings so that axons, their growth cones and filopodia align
parallel to the dorsolateral fasciculus (Nordlander et al., 1991).
The highly stereotyped crisscross pattern of neuronal
processes within the mammalian hippocampus was noted
during the early histological studies by Santiago Ramón y Cajal
(1893) but the mechanism that generates such striking geometry
has not been established clearly. The adult mammalian hippocampus is one of the most widely studied brain structures
(Reznikov, 1991) largely because of its well defined neuronal
circuitry, yet little effort has been made to determine how appropriate connections are established during development. Most in
situ studies of embryonic hippocampal development have concentrated on neurogenesis (‘birth dates’ and migration patterns
2912 A. M. Rajnicek, S. Britland and C. D. McCaig
of undifferentiated neurons) rather than on the process of axon
outgrowth per se (e.g. Bayer, 1980; Altman and Bayer, 1990a).
It would be instructive to identify factors that determine fiber
outgrowth patterns in the hippocampus because they may also
be relevant to other brain regions. This seems likely because
neurites from a variety of mammalian CNS regions exhibit perpendicular contact guidance on parallel arrays of neurites
(Hekmat et al., 1989; Nagata and Nakatsuji, 1991) and artificial
microstructures in vitro (Nagata et al., 1993). Although Nagata
et al. (1993) state that mouse E17-18 hippocampal neuroblasts
showed ‘relatively higher frequency of parallel orientation’ on
grooved quartz it is difficult to compare our results with theirs
directly because no data were provided for hippocampal
neurons. Our study extends that of Nagata et al. (1993) to
indicate that grooved substrata determine the rate of neurite
extension and the direction of presumptive axon or dendrite
initiation as well as the direction of neurite growth.
Relevance of contact guidance to development and
regeneration
Our data that neurons from hippocampi of different embryonic
ages respond differently to identical topographical guidance
cues support the notion that contact guidance acts during hippocampus development by suggesting that the ability to
respond to topographical cues is regulated temporally. Our data
do not indicate whether the temporal differences reflect
changes in the types of neurons present in the cultures at
different embryonic ages or changes in the responsiveness of
individual neuronal types. Hippocampal development in the rat
begins at E15 and development of Ammon’s horn is most rapid
between E17-19 (Reznikov, 1991). It is likely therefore that
E16 cultures contain a smaller proportion of pyramidal
neurons, which may represent as much 80-85% (probably from
fields CA1, or CA2-CA4) of the population in E19 cultures
(Banker and Cowan, 1979). This variation in cellular composition may account for the age-related decrease in perpendicular orientation of E19 compared to E16 cultures. However,
neurons of the same age, from the same cell suspension
respond orthogonally to substrata differing only in groove
width. This indicates that subtle variations in embryonic
geometry have profound effects on neuronal morphology and
suggests that topographic cues encode more complex guidance
information than they are generally attributed with.
Anatomically aligned spaces exist in the developing and in
the regenerating spinal cord of newts and lizards (Singer et al.,
1979) and growth cones often grow along other neurites during
development and regeneration (Nordlander et al., 1991).
Indeed, some procedures aimed at enhancing CNS regeneration have exploited this observation by using carbon filaments
as substrates for contact guidance of regenerating nerve fibers
(Khan et al., 1991). Our data indicate that the dimensions of
substrata used for such treatments are crucial if they are to
produce effective guidance of CNS neurites across lesion sites.
Possible interactions with other guidance cues
We do not propose that guidance of neurites by physical substratum cues alone explains axonal guidance to targets but it is
likely that contact guidance acts in concert with other neuronal
stimulatory and inhibitory growth cone guidance factors. For
example, steady, DC electric fields associated with the amphibian
neural tube (Hotary and Robinson, 1991; Shi and Borgens, 1994)
and the mammalian primitive streak been implicated in neuronal
morphogenesis (Winkel and Nuccitelli, 1989; Hotary and
Robinson, 1992; Shi and Borgens 1994). Interestingly, Xenopus
spinal neurites in weak, DC electric fields respond parallel to the
electric field lines (see McCaig et al., 1994, for review) and rat
hippocampal neurites respond perpendicular to them (Rajnicek et
al., 1992), thus mimicing their respective contact guidance preferences in the present study. On shallow grooved substrata with
an overlying orthogonal adhesive track chick dorsal root ganglion
neurites aligned preferentially on the adhesive tracks but on
deeper grooves they ignored the adhesive paths and aligned
parallel to grooves (Britland et al., 1996b). Similar competition
experiments have not been done using mammalian CNS neurons.
Future experiments will therefore examine the heirarchy of CNS
guidance cues by establishing competing gradients of electrical
cues, substratum gradients of tropic molecules and gradients of
soluble chemoattractants.
In summary, CNS neurites are very sensitive to topographical
contact guidance cues in the absence of simultaneous chemical
or electrical gradients. Xenopus neurites grew parallel to grooves
but hippocampal neurites regulated their direction of neurite
growth depending on groove dimensions and developmental age.
Grooved substrata influenced the site at which neurites emerged
from cell bodies, the presumptive axonal or dendritic identity of
neurites, the direction of neurite growth and the rate of neurite
elongation. Taken together, these data suggest that perpendicular alignment has at least three contributing factors: (1) Presumptive axons emerge perpendicular to the groove/ridge axis.
(2) Neurites turn more frequently and through larger angles to
grow across, rather than along grooves. (3) Neurites grow more
quickly across grooves than parallel to them or on flat substrata.
The companion paper (Rajnicek and McCaig, 1997) addresses
the question of how growth cones sense small surface contours
and the signal transduction events that lead to directional growth.
This work was supported by the Wellcome Trust (A.M.R. and
C.D.McC.) and EPSRC (S.B.). Polystyrene replica substrates were a
generous gift from Harvey Hoch at Cornell University (USA). We
acknowledge the assistance of Chris Wilkinson, Department of Electronics and Electrical Engineering, Glasgow University and especially
the technical assistance of Bill Monaghan at Glasgow University for
advice and assistance with the fabrication of grooved substrata.
REFERENCES
Altman, J. and Bayer, S. A. (1990a). Mosaic organization of the hippocampal
neuroepithelium and the multiple germinal sources of dentate granule cells.
J. Comp. Neurol. 301, 325-342.
Altman, J. and Bayer, S. A. (1990b). Prolonged sojourn of developing
pyramidal cells in the intermediate zone of the hippocampus and their settling
in the stratum pyramidale. J. Comp. Neurol. 301, 343-346.
Bailey, N. T. J. (1981). In Statistical Methods in Biology, 2nd edn, pp. 38-39.
Hodder and Stoughton, London.
Banker, G. A. and Cowan, M. W. (1979). Further observations on hippocampal
neurons in dispersed cell culture. J. Comp. Neurol. 187, 469-494.
Bayer, S. A. (1980). Development of the hippocampal region in the rat: I.
Neurogenesis examined with 3H-thymidine autoradiography. J. Comp.
Neurol. 190, 87-114.
Blackstad, T. W. and Kjaerheim, Å. (1961). Special axo-dendritic synapses in
the hippocampal cortex: electron and light microscopic studies on the layer of
mossy fibers. J. Comp. Neurol. 117, 133-159.
Boocock, C. A. (1989). Unidirectional displacement of cells in fibrillar
matrices. Development 107, 881-890.
Britland, S., Morgan, H., Wojiak-Stodart, B., Riehle, M., Curtis, A. and
CNS neurite orientation on grooves 2913
Wilkinson, C. (1996a). Synergistic and hierarchical adhesive and
topographic guidance of BHK cells. Exp. Cell Res. 228, 313-325.
Britland, S., Perridge, C., Denyer, M., Morgan, H., Curtis, A. and
Wilkinson, C. (1996b). Morphogenetic guidance cues can interact
synergistically and hierarchically in steering nerve cell growth. Exp. Biol.
Online 1:2 (http://www. science. springer. de).
Clark, P., Connolly, P., Curtis, A. S. G., Dow, J. A. T. and Wilkinson, C. D.
(1990). Topographical control of cell behaviour: II. multiple grooved
substrata. Development 108, 635-644.
Davenport, R. W. and McCaig. C. D. (1993). Hippocampal growth cone
responses to focally applied electric fields. J. Neurobiol. 24, 89-100.
Dotti, C. G., Sullivan, C. A. and Banker, G. A. (1988). The establishment of
polarity by hippocampal neurons in culture. J. Neurosci. 8, 1454-1468.
Ebendal, T. (1976). The relative roles of contact inhibition and contact
guidance in orientation of axons extending on aligned collagen fibrils in vitro.
Exp. Cell Res. 98, 159-169.
Elsdale, T. and Bard, J. (1972). Collagen substrata for studies on cell behavior.
J. Cell Biol. 54, 626-637.
Harris, W. A., Holt, C. E., Smith, T. A. and Gallenson, M. (1985). Growth
cones of developing retinal cells in vivo, on culture surfaces and in collagen
matrices. J. Neurosci. Res. 13, 101-122.
Harrison, R. G. (1914). The reaction of embryonic cells to solid structures. J.
Exp. Zool. 17, 521-544.
Hatten, M. E. (1990). Riding the glial monorail: a common mechanism for
glial-guided neuronal migration in different regions of the developing
mammalian brain. Trends Neurosci. 13, 179-181.
Hekmat, A., Künemund, V., Fischer, G., and Schachner, M. (1989). Small
inhibitory cerebellar interneurons grow in a perpendicular orientation to
granule cell neurites in culture. Neuron 2, 1113-1122.
Hinkle, L., McCaig, C. D. and Robinson, K. R. (1981). The direction of
growth of differentiating neurones and myoblasts from frog embryos in an
applied electric field. J. Physiol. 314, 121-135.
Hirono, T., Torimitsu, K., Kawana, A. and Fukuda, J. (1988). Recognition
of artificial microstructures by sensory nerve fibers in culture. Brain Res.
446, 189-194.
Hotary, K. B. and Robinson, K. R. (1991). The neural tube of the Xenopus
embryo maintains a potential difference across itself. Dev. Brain Res. 59, 65-73.
Hotary, K. B. and Robinson, K. R. (1992). Evidence of a role for endogenous
electrical fields in chick embryo development. Development 114, 985-996.
Khan, T., Dauzvardis, M. and Sayers, S. (1991). Carbon filament implants
promote axonal growh across the transected rat spinal cord. Brain Res. 541,
139-145.
Keynes, R. J. and Cook, G. M. W. (1995a). Repulsive and inhibitory signals.
Curr. Opin. Neurobiol. 5, 75-82.
Keynes, R. J. and Cook, G. M. W. (1995b). Axon guidance molecules. Cell
83, 151-169.
Krayanek, S. and Goldberg, S. (1981). Oriented extracellular channels and
axonal guidance in the embryonic chick retina. Dev. Biol. 84, 41-50.
Mattson, M. P. and Kater, S. B. (1988). Isolated hippocampal neurons in
cryopreserved long-term cultures: development of neuroarchitechture and
sensitivity to NMDA. Int. J. Dev. Neurosci. 6, 439-452.
McCaig, C. D., Allan, D. W., Erskine, L. E., Rajnicek, A. M. and Stewart, R.
(1994). Growing nerves in an electric field. Neuroprotocols 4, 143-141.
Nagata, I. and Nakatsuji, N. (1991). Rodent CNS neuroblasts exhibit both
perpendicular and parallel contact guidance on the aligned parallel neurite
bundle. Development 112, 581-590.
Nagata, I., Kawana, A. and Nakatsuji, N. (1993). Perpendicular contact
guidance of CNS neuroblasts on artificial microstructures. Development 117,
401-408.
Newgreen, D. F. (1989). Physical influences on neural crest migration in avian
embryos: contact guidance and spatial restriction. Dev. Biol. 131, 136-148.
Nordlander, R. H. and Singer, M. (1982a). Spaces precede axons in Xenopus
embryonic spinal cord. Exp. Neurol. 75, 221-228.
Nordlander, R. H. and Singer, M. (1982b). Morphology and position of growth
cones in the developing Xenopus spinal cord. Dev. Brain Res. 4, 181-193.
Nordlander, R. H., Gazzerro, J. W. and Cook, H. (1991). Growth cones and
axon trajectories of a sensory path in the amphibian spinal cord. J. Comp.
Neurol. 307, 539-548.
Norris, C. R. and Kalil, K. (1991). Guidance of callosal axons by radial glia in
the developing cerebral cortex. J. Neurosci. 11, 3481-3492.
Oakley, C. and Brunette, D. M. (1993). The sequence of alignment of
microtubules, focal contacts and actin filaments in fibroblasts spreading on
smooth and titanium substrata. J. Cell Sci. 106, 343-354.
Rajnicek, A. M., Gow, N. A. R. and McCaig, C. D. (1992). Electric fieldinduced orientation of rat hippocampal neurones in vitro. Exp. Physiol. 77,
229-232.
Rajnicek, A. M. and McCaig, C. D. (1997). Guidance of CNS growth cones by
substratum grooves and ridges: effects of inhibitors of the cytoskeleton, calcium
channels and signal transduction pathways. J. Cell Sci. 110, 2915-2924.
Ramón y Cajal, S. (1893). The Structure of Ammon’s horn (English translation
by L. M. Kraft, 1968). 78 pp. Charles C. Thomas, Springfield.
Reznikov, K. (1991). Hippocampal formation in the mouse and rat – structural
and organization and development: a review. In Cell Proliferation and
Cytogenesis in the Mouse Hippocampus (ed. F. Beck, W. Hild, W. Kriz, J. E.
Pauly, Y. Sano and T. H. Schiebler), pp. 1-81. Springer-Verlag,
Roberts, A. and Taylor, J. S. H. (1983). A study of the growth cones of
developing embryonic sensory neurites. J. Embryol. Exp. Morphol. 75, 3147.
Schneider, B. F. and Silver, J. (1990). Failure of the subcallosal sling to
develop after embryonic X-irradiation is correlated with absence of the
cavum septi. J. Comp. Neurol. 299, 462-469.
Shi, R. and Borgens, R. B. (1994). Embryonic neuroepithelial sodium
transport, the resulting physiological potential, and cranial development.
Dev. Biol. 165, 105-116.
Silver, J. and Sidman, R. L. (1980). A mechanism for the guidance and
topographic patterning of retinal ganglion cell axons. J. Comp. Neurol. 189,
101-111.
Singer, M., Nordlander, R. H. and Egar, M. (1979). Axonal guidance during
embryogenesis and regeneration in the spinal cord of the newt: the blueprint
hypothesis of neuronal pathway patterning. J. Comp. Neurol. 185, 1-22.
Supèr, H. and Soriano, E. (1994). The organization of the embryonic and early
postnatal murine hippocampus. II. Development of entorhinal, commissural
and septal connections studied with the lipophilic tracer DiI. J. Comp.
Neurol. 344, 101-120.
Tabti, N. and Poo, M.-m. (1991). Culturing spinal neurons and muscle cells
from Xenopus embryos. In Culturing Nerve Cells (ed. G. Banker and K.
Goslin), pp. 137-153. The MIT Press, Cambridge, Mass.
Tessier-Lavigne, M. and Goodman, C. (1996). The molecular biology of axon
guidance. Science 274, 1123-1133.
Winkel, G. K. and Nuccitelli, R. (1989). Large ionic currents leave the
primitive streak of the 7.5-day mouse embryo. Biol. Bull. 176(S), 110-117.
(Accepted 30 September 1997)