Extracellular matrix dynamics during vertebrate axis formation

www.elsevier.com/locate/ydbio

Developmental Biology 268 (2004) 111–122

Extracellular matrix dynamics during vertebrate axis formation$

Andras Czirok,a,b Brenda J. Rongish,a and Charles D. Littlea,*

aDepartment of Anatomy and Cell Biology, University of Kansas Medical Center, Kansas City, KS 66160, USAbDepartment of Biological Physics, Eotvos University, Pazmany Stny 1A, Budapest 1117, Hungary

Received for publication 30 April 2003, revised form 29 September 2003, accepted 30 September 2003

The authors dedicate this article to the memory of John Phillip Trinkaus, Ph.D. (1918–2003). Trink helped us, and thousands of other developmental

biologists, understand the ‘‘forces that shape the embryo.’’ While a postdoc, one of us (C.D.L.) had the great fortune to have a desk next to Trink during his

visiting professorship in Jon Singer’s Lab (UCSD, 1981)—it was a glorious experience.

Abstract

The first evidence for the dynamics of in vivo extracellular matrix (ECM) pattern formation during embryogenesis is presented below.

Fibrillin 2 filaments were tracked for 12 h throughout the avian intraembryonic mesoderm using automated light microscopy and algorithms

of our design. The data show that these ECM filaments have a reproducible morphogenic destiny that is characterized by directed transport.

Fibrillin 2 particles initially deposited in the segmental plate mesoderm are translocated along an unexpected trajectory where they eventually

polymerize into an intricate scaffold of cables parallel to the anterior–posterior axis. The cables coalesce near the midline before the

appearance of the next-formed somite. Moreover, the ECM filaments define global tissue movements with high precision because the

filaments act as passive motion tracers. Quantification of individual and collective filament ‘‘behaviors’’ establish fate maps, trajectories, and

velocities. These data reveal a caudally propagating traveling wave pattern in the morphogenetic movements of early axis formation. We

conjecture that within vertebrate embryos, long-range mechanical tension fields are coupled to both large-scale patterning and local

organization of the ECM. Thus, physical forces or stress fields are essential requirements for executing an emergent developmental pattern—

in this case, paraxial fibrillin cable assembly.

D 2004 Elsevier Inc. All rights reserved.

Keywords: Gastrulation; Fate map; Extracellular matrix; Time-lapse; Dynamics; Tissue deformation

Introduction

Gastrulation and vertebral axis formation require exten-

sive changes in the shape of the early vertebrate embryo. In

avians, after developmental stage 4 (Hamburger and Ham-

ilton, 1951), the primitive streak gradually regresses and

Hensen node moves caudally. A concomitant process,

somitogenesis, leads to the parsing of discrete segments of

mesodermal tissue into small packets at regular intervals

along the anterior–posterior axis, thereby establishing the

segmented vertebrate body plan. Neurulation, which in-

cludes formation and extension of the notochord, is another

0012-1606/$ - see front matter D 2004 Elsevier Inc. All rights reserved.

doi:10.1016/j.ydbio.2003.09.040

$ Supplementary data associated with this article can be found, in the

online version, at doi: 10.1016/j.ydbio.2003.09.040.

* Corresponding author. Department of Anatomy and Cell Biology,

University of Kansas Medical Center, 3901 Rainbow Boulevard, Kansas

City, KS, 66160. Fax: +913-588-2710.

E-mail address: [email protected] (C.D. Little).

contemporaneous event in vertebrate development. The

notochord is laid down as a ‘‘rod’’ of mesoderm by

Hensen’s node, and elongates as the primitive streak

regresses. The neural tube develops dorsal to the notochord;

subsequently, the notochord and neural tube derivatives

become encased in the vertebrae.

Large numbers of cells move to new locations during

these changes in embryonic anatomy. Recent cell fate

mapping studies of gastrulation/neurulation–stage avian

embryos established that mesodermal precursor cells first

dislocate medially towards the primitive streak. Their in-

gression at the axis is followed by various degrees of lateral

displacement, targeting the notochord, somites, or the seg-

mental mesoderm depending on the position and time of

ingression (Bortier and Vakaet, 1992; Catala et al., 1996;

Psychoyos and Stern, 1996; Sawada and Aoyama, 1999;

Schoenwolf et al., 1992; Yang et al., 2002). While the

cellular fate map of gastrulation is well established, the

mechanisms underlying the morphogenetic movements are

A. Czirok et al. / Developmental Biology 268 (2004) 111–122112

still far from clear (for recent reviews, see Wallingford et al.

(2002); Keller et al., 2003).

It has been long understood that the extracellular matrix

(ECM), which is known to provide signals to cells and

affect cell behavior, plays an important role in morphogen-

esis. However, the ECM is not simply a static scaffold: cell

motion involves cell traction, a mechanical interaction

between cells and the ECM, which, in turn, results in a

local realignment of the ECM (Oliver et al., 1995; Stoplak

and Harris, 1982; Tranquillo, 1999). ECM fibers can also be

reoriented by long-range (nonlocal) mechanical forces with-

in the tissue (Olsen et al., 1999). Nevertheless, to the best of

our knowledge, in vivo ‘‘fate maps’’ of ECM component

movements have not been determined—nor has it been

established whether ECM molecular complexes, like cells,

‘‘behave’’ in a predictable, developmental stage- and posi-

tion-dependent fashion.

Proteins of the fibrillin family are promising candidates

for ECM tracking studies because they are ubiquitous and

easily recognizable, discrete filamentous structures when

made visible by immunolabeling methods (Burke et al.,

2000; Gallagher et al., 1993; Rongish et al., 1998; Visconti

et al., 2003; Wunsch et al., 1994). Fibrillin 1 and 2 are

constituents of connective tissue microfibrils (Sakai et al.,

1986; Zhang et al., 1994) and are both observed in the avian

embryo before gastrulation in the form of slender filaments

marking the future anterior–posterior axis (Gallagher et al.,

1993; Wunsch et al., 1994). During gastrulation, the fibril-

lins continue to display a remarkable pattern widely distrib-

uted in association with the mesoderm (Burke et al., 2000;

Rongish et al., 1998; Visconti et al., 2003). The importance

of these proteins to the structural and functional properties

of connective tissues has been demonstrated by the finding

that mutations in the human FBN1 and FBN2 genes result in

heritable disorders, with patients manifesting ocular, skele-

tal, and cardiovascular abnormalities (Dietz et al., 1994). In

accord, fibrillin-deficient mouse models demonstrated that

the absence of normal levels of either fibrillin 1 or 2 protein

is incompatible with the proper function of mature elastic

fibers (Artega-Solis et al., 2001; Pereira et al., 1997; Pereira

et al., 1999). Deficiency in one type of fibrillin was found to

be compatible with gastrulation of mouse embryos; never-

theless, exogenous expression of a truncated fibrillin in the

presumptive dorsal mesoderm of gastrulation stage frog

embryos disturbed endogenous fibrillin localization and

blocked gastrulation (Skoglund, 1996).

Motivated by the excellent tracking properties and the

developmental importance of fibrillin fibers, we investigated

the reorganization and assembly of this ECM component

during gastrulation/neurulation. We demonstrate the coales-

cence of an extended, loose meshwork into larger cables.

Our high-resolution map of filament displacements shows

that morphogenesis-associated ECM movements essentially

repeat the same basic pattern at progressively more caudal

positions. These quantitative data open the door to future

studies focusing on the mechanical forces that drive mor-

phogenesis. We conclude with the hypothesis that large-

scale ECM reorganization is primarily driven by changing

mechanical tension fields.

Materials and methods

Microinjection

The location of fibrillin 2-containing ECMwas monitored

with Cy3 (Amersham Pharmacia Biotech Inc., Piscataway,

NJ) fluorochrome-conjugated monoclonal JB3 IgG anti-

bodies (Developmental Studies Hybridoma Bank, University

of Iowa, Iowa City, IA). Delivery of these nonperturbing

(Rongish et al., 1998) marker antibodies was accomplished

by injecting 25-nl volumes, one to four times per embryo,

at 1 ng/nl concentrations (Pico-Injector, Harvard Apparatus

Inc, Holliston, MA, mounted on a micromanipulator,

Narishige Scientific Instrument Lab, Tokyo, Japan).

Embryo culture

As detailed in Little and Drake (2000), quail embryos

(Coturnix coturnix japonica, John Smith Farms, Bucyrus,

KS) at Hamburger and Hamilton (HH) stages 4–5 were

encircled with paper rings that adhere to the embryo’s

vitelline membrane and then excised from the egg. After

removing the adherent yolk by washing with sterile embryo

phosphate-buffered saline (EPBS), the embryos are placed on

a solid agarose bed (5% agarose in EPBS) for microinjection.

After introduction of the antibodies, the embryos are

placed in a custom designed, microscope-attached incuba-

tion chamber (Czirok et al., 2002; Rupp et al., 2003). The

wells of a suitably modified Costar polystyrene 6 well plate

(Corning Inc., Corning, NY) are filled with 3 ml of CO2-

independent culture medium (Leibovitz-L15 medium sup-

plemented with 2 mM L-glutamine, 10% chicken serum, and

1% penicillin–streptomycin, GibcoBrl, Grand Island, NY)

in which a Millicell (Millipore Co., Bedford, MA) tissue

insert is placed. The plastic filter material of the insert was

replaced with a sparse array of polypropylene surgical

suture fibers (7–0 Surgipro, USSC, Norwalk, CT). Embryos

are cultured under sterile conditions, at the medium–air

interface, ventral side-up, with their vitelline membrane in

contact with the suture mesh. Their normal development

was established by the lack of morphological defects as well

as by the timely addition of somites and elongation of the

notochord (Rupp et al., 2003).

Microscopy and image processing

As described in detail in Czirok et al. (2002), an

automated Leica DMR upright microscope is used in DIC

and epifluorescence modes with a long working distance

10 � (0.25 NA) objective. Fields of 875 � 688 Am are

imaged with a cooled Quantix CCD camera (Photometrics,

Table 1

Summary of the experimental data

Embryo # of somites at Observation # of

A. Czirok et al. / Developmental Biology 268 (2004) 111–122 113

Tucson, AZ) at a spatial resolution of 658 � 517 pixels, 12

bit light intensity resolution, and typically 100 ms exposure

times. In each period of the imaging cycle, 2–4 embryos are

sequentially positioned under the objective. For each em-

bryo, a rectangular area of 1 � 4 or 2 � 4 slightly over-

lapping microscopic fields are scanned, and each of these

fields is recorded at multiple (10–12) focal planes, separated

by 20 Am, first in DIC and immediately thereafter in

epifluorescence mode. The acquisition of the ‘‘z-stack’’ pairs

is accomplished within a short period of time (typically a

minute) ensuring the correct spatial registration of the

corresponding DIC—epifluorescence image pairs. The prac-

tical result of this technology is that every position of the

embryonic plate lies in one of the focal planes; thus, no

feature can be lost.

Embryos change their position considerably during de-

velopment, and this motion must be compensated to keep

the specimen centered in the field of view. This is accom-

plished in two stages: First, during recording, the stage

positions are periodically (about every 2 h) revised either

manually or using software-generated feedbacks. During the

processing of the stored images, a second, automatic frame-

by-frame realignment is performed to eliminate the drift and

sudden shifts of the object. These tasks, the merger of the

neighboring microscopic fields into a composite image, and

the local contrast enhancement of the immunofluorescence

data were achieved using our algorithms (Czirok et al.,

2002), which are available to download at http://www.

kumc.edu/mic-utils.

Laser scanning confocal microscopy was performed as

described in Rongish et al. (1998).

Tracking of image details

A modified version of a program used in previous cell

tracking studies (Hegedus et al., 2000) allowed us to browse

through the stored images (time-lapse ‘‘z-stacks’’) and fol-

low various details, such as fluorescent filament segments,

5–10 Am in size, through consecutive images. Although this

tracking procedure is performed in three dimensions (using

each optical section), due to the large depth of field of the

microscope objective, the resolution along the vertical (z)

direction is rather limited. Therefore, here we confine our

studies to the two-dimensional (x–y) projections of the

obtained position data. In the following, two-dimensional

vector quantities are denoted by boldface letters. The data

analyses are detailed in the Supplementary Material.
Injection Start End Duration Frames Fields
features

1. PS 3+ 8 6 h 55V 26 4 � 1 1324

2. PS 2+ 10 10 h 30V 42 4 � 2 2665

3. PS 2 9 11 h 10V 108 4 � 2 3074

4. 5 6+ 12 8 h 40V 30 4 � 2 864

5. 4 5 9 6 h 55V 26 4 � 1 877

6. PS 2 7+ 9 h 30V 33 4 � 2 –

7. PS 4 9+ 9 h 30V 33 4 � 2 –

8. PS 3+ 8+ 9 h 30V 33 4 � 2 –

9. PS 1 8 11 h 10V 108 4 � 2 –

Results

Fibrillin immunolabeling

Extracellular fibrillin 2 was labeled in quail embryos by

the microinjection of fluorochrome-conjugated JB3 mono-

clonal, bivalent, IgG antibodies. After the antibody micro-

injection at HH stages 4–5, embryos were recorded until

HH stage 10 with scanning time-lapse microscopy (Table 1).

Within 1 h, the JB3 antibodies reach a substantial portion

(up to 50–70%) of the area pellucida (Fig. 1). The fluores-

cence intensity profile, that is, the extent of areas with

various fluorescence intensities, becomes stable after the

initial diffusion period (data not shown). This stability

indicates very low rates for photo bleaching or other

antibody degradation, since these processes would be

expected to eliminate brightly stained foci.

The resulting labeling is specific, as demonstrated by

experiments with microinjected control (non-ECM-binding)

antibodies, including the avian endothelial cell marker QH1

(Czirok et al., 2002) and mouse mAb G that recognizes

avian h1 integrins (Drake et al., 1991). Further, the labeling

pattern observed in JB3-microinjected embryos was com-

pared to that in postexperimental whole-mount immunos-

taining of fixed embryos (Fig. 2D) and the pattern appears

identical (Rongish et al., 1998). In contrast, this pattern is

strikingly different from that observed following microin-

jection of antibodies to other ECM proteins, including

tropoelastin and fibulin 1 (Visconti et al., 2003), as well

as fibronectin, collagen IIA, and latent TGF-h binding

protein 1 (data not shown). Occasionally some highly

fluorescent precipitates remain near microinjection sites,

which exclude these areas from further analysis (star in

Fig. 1).

The perturbative effect of microinjected JB3 antibodies

was assayed by comparing whole-mounted JB3-injected

embryos to non-injected controls after 10 h of development.

No observable differences were found either in the post-

fixation JB3 staining pattern or in the anatomy of the

embryos. The timing of major morphogenic processes were

also the same in JB3-injected embryos and non-injected

controls (n > 20). In particular, the somites form on

schedule, every 90 min, the heart loops and commences

beating normally (data not shown).

Changes in fibrillin organization

During the recorded 8- to 12-h-long period, the fibrillin 2

immunolabeling pattern changes remarkably. As Fig. 1

http:\\www.kumc.edu\mic-utils

Fig. 1. Relevant developmental stages of a quail embryo. The images in panels (a, aV), (b, bV), and (c, cV) display corresponding epifluorescence-DIC image pairs,

taken at 3-h intervals, following the microinjection of Cy3-JB3 anti-fibrillin 2 antibodies. The DIC images show the characteristic changes in embryonic

anatomy, most notably the elongation of the notochord (nc), formation of the anterior intestinal portal (aip), neural folds (nf), and somites. The progressively

longer red arrows demonstrate the elongation of the embryonic axis and addition of somites during the 6-h elapsed time. Positions of the first and the newest

somites are marked on one side of the embryo with an asterisk and a caret, respectively. The green immunofluorescence patterns reveal substantial

rearrangements of the fibrillin 2-containing ECM. Most notable are parallel cables just lateral to the notochord (brackets, panels a, b, c). These fibrillin cables

enclose the somites and also extend into the cranial aspect of the segmental plate mesoderm (sp, panel cV); this is readily observed at the interval between the

caret and the right bracket in panels b and c. Also, an array of curved cables connect the aip to the somitic field (parentheses, panels b, c). The meshwork of

fibrillin 2 that ensheaths the notochord progressively diminishes at the level of the last two somites (white arrows). The overexposed area (black star, panels a,

b, c) is precipitated material at the microinjection site.


demonstrates, a fibrillin 2-containing ECM meshwork en-

cases the notochord. At HH stage 7, a bundle of fibrillin

fibers connect the AIP to the somitic regions, reaching the

first few somite pairs. Most notable are fibrillin 2-containing

fibers that assemble parallel to the embryonic axis. The

cranial portion of these bundles encloses the somites, while

caudally they extend more than 200 Am into the segmental

plate mesoderm and approach Hensen node. Caudal to the

node, fibrillin 2 exhibits a different pattern, which consists of

punctate, disconnected fluorescent foci. Fig. 2 shows the

somitic array and the segmental plate mesoderm in greater

detail. These images reveal an oriented, dense array of

paraxial filaments, remarkably different from the loose

meshwork-like organization characteristic of the more lateral

regions. (See supplementary data, movie 1.)

Filaments retain their identity

Figure 3 demonstrates the progressive sequence of in-

cremental changes that constitutes the tissue-scale ECM

reorganization. Filament shapes and positions remain fairly

similar between consecutive image pairs; these observations

form the basis of our tracking procedure. As multiple focal

planes were recorded with our multi-field scanning time-

lapse (4D) microscopy system, no intraembryonic object is

lost or rendered out of focus. Thus, one can unambiguously

follow filaments, frame-by-frame, over the recorded time.

Note that although only minor alterations are seen in the

ECM meshwork between short (30 min) time intervals in

Fig. 3,while tracking filaments over a several hours reveals

substantial changes in ECM organization (see circled struc-

tures in Fig. 2).

Filament trajectories reveal substantial relocation

Five embryos were selected for a systematic tracking

procedure in which filaments were identified and followed

frame-by-frame, resulting in a database of spatial and tem-

poral coordinates (Table 1). No disassembly was observed

within the sample of more than 200 traced filaments.

Changes in the obtained coordinates can reflect both genuine

alterations in embryonic anatomy, as well as anatomically

irrelevant movements of the whole specimen relative to the

microscope objective. To eliminate the latter, intersomitic

Fig. 2. Fibrillin 2 reorganization in the axial areas. The individual images in panels (a, aV), (b, bV), and (c, cV) depict corresponding areas outlined by the yellow

boxes in Fig. 1. Selected filaments, marked with the colored circles, visibly alter their relationship with nearby filaments and the axis. Despite this motion, the

filaments retain connectedness and shape to some degree. Note that a ‘‘new’’ filament (orange circle), absent in panel aV, appears in the next two images—and is

shown to move medially. The blue boxes denote the formed somites. Panel d shows the somitic array at a higher magnification. This double-labeled image was

collected at the termination of a time-lapse recording, when a second, whole-mounted JB3 immunostaining was performed. The red channel contains the

labeling pattern generated by the microinjected Cy3-JB3, while the second, whole-mounted Cy2-JB3 staining yields the green channel. The yellow color

indicates staining by both labeling methods. Panel e shows a high-resolution laser scanning confocal microscope (LSCM) image of a comparable area in a

specimen prepared for ideal immuofluorescence quality. The LSCM image is a projection of 10 focal planes into a single field. Comparing our double-labeled

whole-mount wide-field epifluorescence image with the LSCM micrograph reveals a surprising amount of similar detail, including fluorescent filaments on the

order of 1 Am in size.


cleft positions were extracted from DIC images and used as

anatomical reference points. For each filament, a mathemat-

ical procedure, described in the Supplementary Material,

resulted in its trajectory, i.e., its position on each frame

relative to the somitic array. Thus, changes in these positions

indicate anatomically relevant filament displacements.

Fibrillin 2 trajectories, obtained from multiple embryos,

reveal that the ECM is substantially relocated during devel-

opment. Filament motion seems to be ordered and its

tendencies highly reproducible: trajectories of adjacent

filaments are similar. The direction of motion is medial in

the lateral plate mesoderm, while the direction is caudal in

the segmental plate mesoderm near (and posterior to)

Hensen node (Fig. 4). Craniolaterally to the first somite

pair, filament trajectories reflect that ECM is pulled toward

the forming AIP. However, the tendency of ECM motion is

time-dependent: early trajectory segments can be perpen-

dicular to later segments of the same and nearby filaments

(boxed area in Fig. 4).

Velocities characterize anatomical changes

Trajectory plots are tools commonly used to visualize

rearrangements in a given dynamic sample. Nevertheless,

the actual trajectory shapes strongly depend on the partic-

ular timing and duration of the analyzed recording, as well

as on the locations of available tracer particles. Local

tendencies of motion during a short period are much better

suited for quantitative, statistical comparison of different

embryos, or various developmental stages of the same

embryo. In this study, we subdivided the observational

period into 1-h-long segments and determined filament

displacements during these time intervals. The obtained

displacements give the distance between filament positions

at the beginning and at the end of the time segment. The

displacement during a short period is proportional to the

velocity; thus, we use velocities instead of trajectories as

the fundamental descriptor of embryonic rearrangements.

As we noted in Fig. 4, displacements of nearby filaments

Fig. 3. A time-lapse sequence of ECM remodeling at high resolution. The four panels show the movements and rearrangements of two selected fibrillin 2

filaments (circle and square) during a 90-min interval. Despite changes in overall ECM organization, filaments can be identified with high fidelity in the

consecutive frames.


are similar. This local average filament velocity (both

magnitude and direction) within embryo a at time t around

a certain location (x, y) is denoted as the Va(x, y, t) velocity

field (or snapshot).

Traveling wave-like velocity pattern

The velocity field V can be described as a traveling-wave

Vaðx; y; tÞcUðx; y� YaðtÞÞ ð1Þ

where Ya(t) is the position of the somitogenesis front along

the cranial–caudal axis and U is a time-independent, axially

symmetric (Ux(x,y) = �Ux(�x,y), Uy(x,y) = Uy(�x,y)) ve-

locity pattern. Eq. (1) summarizes the assumptions that

filament motion is similar on both the left and right sides

of the embryo, and the same pattern is repeated at gradually

more caudal locations as the primitive streak regresses. In

other words, if an observer’s motion is coupled to Hensen’s

node, then that observer would witness a stationary velocity

pattern: Similar velocities would be obtained at the same

locations (relative to the node), regardless of time or devel-

opmental stage. This time-independent behavior was statis-

tically verified by calculating C(s), the temporal correlation

function of the velocity field. As described in the Supple-

mentary Material, C(s) measures the average similarity of

velocity snapshot pairs, in which the elements are separated

by a time interval of length s. We found that the velocity field

V becomes uncorrelated (C(s) = 0) within 4 h (P > 0.2),

while the same velocity data, when viewed from a reference

system co-moving with the node, remained significantly (P

< 0.01) correlated (C(s) > 0) during the entire observational

period.

To extract the velocity pattern U, velocity snapshots

were shifted so that the locations of Hensen’s node and the

somitogenesis front were held stationary and subsequently

averaged over time as Uaðx; yÞ ¼ hVaðx; yþ YaðtÞ; tÞit:These patterns, characterizing the fibrillin 2 velocity pattern

in individual embryos, are similar (Fig. 5A) and form the

basis of our statistical sample-to-sample comparison. The

velocity pattern, which is also averaged over the individual

embryos (Fig. 5B), reveals tendencies more clearly than

trajectories do. The prevailing motion is caudally oriented

around the node. This directionality changes to a medial

orientation at lateral positions, while at the extra-embry-

onic boundary displacements are directed cranially—result-

ing in a vortex-like velocity field. The medial component

of the velocity field thus substantially contributes to the

accumulation of fibrillin-containing ECM between the

node and the somitogenesis front. Within the same area,

the gradually increasing cranial velocity component reflects

an anterior–posterior axial strain of the tissue. The statisti-

cal errors of the data presented in Fig. 5B can be directly

estimated from the scatter of velocities characterizing indi-

vidual embryos (Fig. 5A). While the sample size and thus

the error depends on the location, the estimated average

statistical error is F20 Am/h for each velocity component.

Discussion

In vivo ECM assembly

The biological function of the ECM—its assembly,

turnover, and remodeling—has been the focus of numerous

studies (Dzamba et al., 2001; Hay, 1982; Ohashi et al.,

1999; Schwarzbauer and Sechler, 1999). Recently a three-

dimensional structure of the fibrillins was proposed to

explain its elastic properties in molecular terms (Baldock

et al., 2001). However, very little is known about the

dynamics of these molecules, particularly regarding the

mechanisms by which the ECM undergoes large-scale

(tissue level) remodeling. Yet, the distribution of the

ECM can be as crucial as its molecular composition.

ECM organization can determine tissue mechanical prop-

erties (Barocas and Tranquillo, 1997; Olsen et al., 1999),

as well as serve as a guidance system for cells. For

example, some cell types, such as fibroblasts (Dickinson

et al., 1994; Stoplak and Harris, 1982), endothelial cells

(Vernon et al., 1995), and neurons (Dubey et al., 2001), are

known to preferentially follow oriented fibers. In addition

to establishing a detailed time-dependent record of fibrillin

localization, we also determined the dynamics of ECM

redistribution. To achieve this, we traced filaments, that is,


certain identifiable segments within the immunolabeled

network. Although the ultrastructure of the early ECM is

not yet determined at early developmental stages, it is

highly likely that filaments are composed of multiple ECM

molecules (Visconti et al., 2003). Because filaments retain

their shape between consecutive frames, position changes

were interpreted as physical displacements of intact objects

rather than as a process involving filament disassembly

followed by reassembly and antibody reattachment. If

disassembly–reassembly were occurring, that would be

manifested as the visible, frequent disappearance of entire

filaments; however, this was never observed. Moreover,

the finding that displacements of adjacent filaments are

similar suggests that filaments are embedded in a visco-

elastic continuum (most likely including attached cells),

rather than existing as free-moving or quickly disconnect-

ing and reconnecting polymers.

Fibrillin 2 in somitogenesis

Our data show that the fibrillin-containing ECM is

substantially reorganized during development. Among the

observed processes, the formation of fibrillin bundles in

the somitic and pre-somitic mesoderm is of special interest.

These fibers delineate existing and future somites and may

contribute to the integrity of the somitic array by providing

a suitable anchoring mechanism. Since somites are sur-

rounded by an acellular ECM, such a scaffold is likely to

be required before the physical separation of the cells from

the pre-somatic mesoderm and at the intersomitic clefts.

The spatial and temporal pattern of fibrillin 2 localization

is thus fully compatible with a presumptive scaffolding

role.

The condensation of fibrillin-containing bundles in the

medial segmental plate mesoderm is also an unsuspected

process of pattern formation: The demarcation between the

axial cables and the less organized lateral meshwork is

fuzzy and does not coincide with preexisting anatomical

structures. Moreover, a substantial portion of fiber compo-

nents are not secreted locally, but collected from a large

area (Fig. 2). This is in contrast, for example, with the

ECM ensheathment of the notochord, where the creation of

the notochord boundary clearly precedes the gradual ac-

Fig. 4. Trajectories of fibrillin filaments and reference points. Eighty

filaments were selected randomly in the embryo shown in Fig. 1, and their

respective trajectories plotted with colored lines. Each color represents

positions within 2-h-long time intervals: red denotes the original position of

each filament (at frame 1), followed by orange, green, cyan, and blue

positions, respectively. Concurrent trajectories of the spatial reference

system, the intersomitic clefts, are plotted with cyan lines (e.g., white

circle). The DIC image, upon which the trajectory lines are superimposed,

shows the embryo at an intermediate time point, taken 4 h after the start of

the observation. The motion of the filaments within the lateral plate

mesoderm (bracket) is strongly medialward. In contrast, filament tra-

jectories near Hensen’s node show a caudalward directional bias (asterisk).

Between these two regions, the filament trajectories gradually change from

medialward to caudalward. This is in distinct contrast to the displacements

of the intersomitic clefts or reference points that are typically very small—

an indication that the intersomitic clefts are well suited to serve as an

anatomical reference system. The yellow box encloses an area in which

early (red) fibrillin trajectory segments are perpendicular to later (blue)

trajectory segments. Thus, the prevalent directionality of a fibrillin 2

filament can be highly time-dependent at one region (yellow box), while

very little change in directionality is exhibited at positions caudal to

Hensen’s node (asterisk). These data demonstrate that fibrillin filaments are

not moving in lockstep; rather, the trajectories appear as a part of a swirl-

like pattern (see Fig. 5).

Fig. 5. Experimentally derived velocity patterns of fibrillin 2. Locally averaged velocities of Cy3-labeled fibrillin filaments are plotted at grid-points of a square

mesh. For each grid-point, velocities are represented by scale bars giving the local average displacement in 1 h. The vertical and horizontal yellow lines show

the positions of the embryonic axis and the somitogenesis front, respectively. Small and large red circles indicate the somitic array and Hensen’s node. The

velocity patterns characterizing five individual embryos are superimposed in panel a; thus, at each grid-point, the different scale bars represent velocity data

obtained in different embryos. The green box delineates an area, where most grid-points contain data from at least two embryos. For each grid-point, the

average of the various velocities was calculated and plotted in panel b. The blue arrow depicts the prevailing filament motion lateral to the node.


cumulation of the fibrillin-containing meshwork (see Figs.

1 and 2; supplementary data movies 1 and 2). Thus, in the

latter case, tissue boundaries can direct a local ECM

accumulation. In accord with this assumption, we found

no evidence of fibrillin 2-containing filaments being trans-

ported to the notochord from distant sites.

ntal Biology 268 (2004) 111–122 119

Cranial–caudal developmental gradient is reflected in the

traveling velocity pattern

The similarities between a caudal region of an older

embryo and a corresponding cranial part of a younger

embryo underlies the hypothesis that the velocity field

reiterates a basic pattern at progressively more caudal loca-

tions—following the regression of the primitive streak. Since

the distance between the newest somite and Hensen node is

stable during these developmental stages, we chose to follow

the somitogenesis front as a positional indicator that moves

in concert with this cranial–caudal developmental gradient.

Our observations are confined to the caudal region of the

embryo, between HH stages 5 and 10. Later, or concomitant

but more cranial events of ECM remodeling are further

complicated by additional morphogenetic processes. An

important example of these is vasculogenesis, since fibrillin

fibers incorporate into, and appear to delineate the devel-

A. Czirok et al. / Developme

Fig. 6. The conjectured interdependence of ECM and mesodermal cell motion. In a

relative to the surrounding ECM (including fibrillin 2), while the ECM was also

summarized in Eq. (1). The simulation was two-dimensional; only anterior–po

populations were followed—three groups on each side of the axis (panel a). The d

(yellow arrow) of the involuting mesodermal cells at a position along the anterior

assumed to move laterally with a constant speed of 25 Am/h (yellow arrow). The em

these assumptions and empirical data, the computer simulation yielded a patterned

as black dots. Thus, the combination of lateral cell locomotion and a caudalward p

fates described in Fig. 12 of Schoenwolf et al. (1992) and Fig. 7 of Psychoyos an

node, respectively, while the red lines indicate the notochordal axis.

oping vascular network (not shown). Moreover, as Hensen’s

node approaches the caudal boundary of the area opaca, its

linear motion slowly diminishes (Spratt, 1946, 1947).

Therefore, we do not expect our scheme of an evenly

shifting velocity pattern to hold for later stages (above HH

13). While the scope of the present investigation is limited

to stages and locations in which early axis patterning is

occurring, our method of visualization, tracking, and quan-

tifying the motion of ECM filaments should be useful in

studies of other, perhaps more complex, morphogenic

processes (as evidenced by the data obtained from regions

of AIP regression and bilateral heart tube fusion).

Forces that shape the embryo

Several investigators have drawn attention to the impor-

tance of physical forces in the shaping of an embryo (Oster et

al., 1983; Trinkaus, 1984). This conceptual framework

computer simulation, cells were assumed to engage in autonomous motility

assumed to move—according to the experimentally obtained velocity field

sterior and lateral–medial movements were considered. Six distinct cell

rawing in panel b, modified from Hay (1982), indicates the postulated path

–posterior axis, denoted by the blue line in panel a. In particular, cells were

pirically determined ECM motion is indicated by red arrows. As a result of

array of cells dispersed along the anterior–posterior axis, depicted in panel

assive motion can result in a pattern that bears a striking resemblance to cell

d Stern (1996). The red circles in panel a denote the somites and Hensen’s


asserts that (1) as tissues are physical objects, their deforma-

tion inevitably involves mechanical forces. (2) The relevant

forces are exerted internally within the embryo by cells.

Morphogenetic movements are thus determined by the inter-

play between the spatio-temporal pattern of exerted forces

and the mechanical properties of the responding tissues.

Recently, viscoelastic parameters of embryonic cell aggre-

gates were established (Forgacs et al., 1998) and gastrulation

movements of sea urchin embryos have been analyzed using

viscoelastic models (Davidson et al., 1995, 1999; Drasdo and

Forgacs, 2000). These latter studies also shed light on the

crucial role of the ECM in the mechanics of gastrulation.

Mechanical tension fields within the avian embryo are also

thought to play a significant role during avian gastrulation

and axial morphogenesis. It has been postulated that at very

early stages, the tension generated by epiboly, the expansion

of the embryonic plate (epiblast) on the vitelline membrane,

is partially responsible for the elongation of the primitive

streak (reviewed in Stern and Canning, 1988; Trinkaus,

1984). Although no stress or strain data are available yet

for avian embryo development, such experiments were

carried out in Xenopus embryos (Moore et al., 1995).

Unlike cells, ECM components cannot move themselves;

thus, they are ideal ‘‘passive tracers’’ of tissue movements.

Although cells are known to reorganize matrix molecules on

their surface, displacements exceeding 100 Am clearly

cannot be generated in that manner. Thus, we propose that

(1) the observed displacements reflect—at least partially—

large-scale tissue movements driven by mechanical forces,

(2) an existing ECM structure can be substantially reorgan-

ized by a mechanical tension-driven meshwork-to-cable

transition, and (3) formation of these structures is strongly

dependent not only on local cellular activity, but also on the

global developmental movements of the embryo. Based on

the nonspecificity of mechanically transduced effects, we

expect that fibrillin 2 is not unique and that other ECM

constituents exhibit similar behaviors. Further experiments

are underway to test these conjectures directly.

Although our results conclude with the determination of

a velocity field that characterizes the nodal region of an

avian embryo, we expect that these data will eventually be

used to estimate the distribution and origin of the forces

acting on the embryonic plate. A similar approach was

recently applied to determine the traction forces of cells

embedded in collagen gels (Vanni et al., 2001). In our case,

however, the viscoelastic properties of the tissue are more

complex than those of a collagen gel undergoing small

deformations, and the full determination of the stress field

requires knowledge about all three components of the

velocities (Helmke and Davies, 2002).

Gastrulating cells are conjectured to move relative to the

surrounding ECM

Embryonic cells can be rearranged by at least two

different means. Cells can actively migrate relative to their

local environment, or large-scale tissue movements can

passively transport them. For example, heart looping clearly

involves tissue deformations that move cells—relative to the

somites, but also relative to each other (Voronov and Taber,

2002). These active and passive modes of relocation are not

mutually exclusive; in fact, both are expected to shape the

embryo.

Our data inevitably raise the question about the possible

relation between the studied ECM motion and the reported

features of mesodermal cell migration (Psychoyos and

Stern, 1996; Schoenwolf et al., 1992). Based on the idea

of superimposing the ECM and cell movements, we

employed a computer simulation to model the lateral

spreading of mesodermal cells (Fig. 6). In this hypothetical

model, (a) the traced cell population is ingressing contin-

uously along the primitive streak, (b) the ingression is

followed by a steady, laterally directed active movement

relative to the surrounding ECM, (c) the ECM moves

according to the experimentally obtained pattern, summa-

rized in Eq. (1). In Fig. 6, a subset of the cells is visualized

as being both carried along with and moving relative to the

ECM environment. Although we assumed a uniform

locomotory activity, the combination of the lateral migra-

tion of the cells together with the experimentally deter-

mined nonuniform motion of the ECM results in a pattern

recapitulating the experimental findings of Schoenwolf et

al. (1992) and Psychoyos and Stern (1996). The validity of

this picture needs to be established by comparing time-

resolved experimental data on concurrent cell and ECM

movements.

Systems approach

It has been evident for decades that morphogenesis

involves an interplay of physical forces, cell behavior, and

genetic regulation. Unfortunately, models encompassing

pertinent developmental complexities are difficult to develop

due to the paucity of nongenetic quantitative experimental

data. Recent advances in optical microscopy, image process-

ing, and computing power have extended the scope of time-

lapse investigations into the quantitative realm. One resulting

advantage is the ability to characterize developmental pro-

cesses with higher precision than earlier fate mapping

methods. These precise quantitative data can be used to

construct and test numerical models. Such models, used as

part of a ‘‘systems’’ or integrative approach (Hove et al.,

2003; Noble, 2002), will permit a new level of understanding

of cell, tissue, and ECM dynamics during morphogenesis.

Acknowledgments

We are grateful to Adrienn Gabor, Erica Perryn, and Paul

Rupp for their valuable help in this project. We thank Tom

Skalak (UVa) for critical comments on the manuscript. This

work was supported by awards of NSF (NATO-NSF DGE-


0075179, to A.C.), the Institutional Biomedical Research

Training Program of KUMC (to A.C.), the Hungarian

Research Fund (OTKA T034995, to A.C.), the Heartland

Affiliate of the American Heart Association (0160369Z, to

B.J.R.), the Mathers Charitable Foundation (to C.D.L.), and

NIH (5P01 HL58213-08, to C.D.L.; 1R01 HL73700 to

B.J.R.). The JB3 antibody was obtained from the devel-

opmental studies Hybridoma Bank at the Univ. of Iowa.

References

Artega-Solis, E., Gayraud, B., Lee, S., Shum, L., Sakai, L., Ramirez, F.,

2001. Regulation of limb patterning by extracellular microfibrils. J. Cell

Biol. 154, 275–281.

Baldock, C., Koster, A., Ziese, U., Rock, M., Sherratt, M., Kadler, K.,

Shuttleworth, C., Kielty, C., 2001. The supramolecular organization

of fibrillin-rich microfibrils. J. Cell Biol. 152 (5), 1045–1056.

Barocas, V., Tranquillo, R., 1997. An anisotropic biphasic theory of tissue-

equivalent mechanics: the interplay among cell traction, fibrillar net-

work deformation, fibril alignment, and cell contact guidance. J. Bio-

mech. Eng. 119 (2), 137–145.

Bortier, H., Vakaet, L., 1992. Fate mapping the neural plate and the intra-

embryonic mesoblast in the upper layer of the chicken blastoderm with

xenografting and time-lapse videography. Development, 93–97

(Suppl.).

Burke, R., Wang, D., Mark, S., Martens, G., 2000. Distribution of fibrillin I

in extracellular matrix and epithelia during early development of avian

embryos. Anat. Embryol. (Berlin) 201, 317–326.

Catala, M., Teillet, M.-A., De-Robertis, E., Le-Douarin, N., 1996. A spinal

cord fate map in the avian embryo: while regressing, Hensen’s node

lays down the notochord and floor plate thus joining the spinal cord

lateral walls. Development 122, 2599–2610.

Czirok, A., Rupp, P., Rongish, B., Little, C., 2002. Multi-field 3D

scanning light microscopy of early embryogenesis. J. Microsc. 206,

209–217.

Davidson, L., Koehl, M., Keller, R., Oster, G., 1995. How do sea urchins

invaginate? Using biomechanics to distinguish between mechanisms of

primary invagination. Development 121, 2005–2018.

Davidson, L., Oster, G., Keller, R., Koehl, M., 1999. Measurements of

mechanical properties of the blastula wall reveal which hypothe-

sized mechanisms of primary invagination are physically plausible

in the sea urchin Strongylocentrotus purpuratus. Dev. Biol. 209,

221–238.

Dickinson, R., Guido, S., Tranquillo, R., 1994. Biased cell migration of

fibroblasts exhibiting contact guidance in oriented collagen gels. Ann.

Biomed. Eng. 22 (4), 342–356.

Dietz, H., Ramirez, F., Sakai, L., 1994. Marfan syndrome and other micro-

fibrillar diseases. Adv. Hum. Genet. 22, 153–186.

Drake, C., Davis, L., Hungerford, J., Little, C., 1991. Perturbation of beta1

integrin-mediated adhesions results in altered somite cell shape and

behavior. Dev. Biol. 149, 327–338.

Drasdo, D., Forgacs, G., 2000. Modeling the interplay of generic and

genetic mechanisms in cleavage, blastulation, and gastrulation. Dev.

Dyn. 219 (2), 182–191.

Dubey, N., Letourneau, P., Tranquillo, R., 2001. Neuronal contact guidance

in magnetically aligned fibrin gels: effect of variation in gel mechano-

structural properties. Biomaterials 22 (10), 1065–1075.

Dzamba, B., Keene, D., Isogai, Z., Charbonneau, N., Karaman-Jurukovska,

N., Simon, M., Sakai, L., 2001. Assembly of epithelial cell fibrillins.

J. Invest. Dermatol. 117, 1612–1620.

Forgacs, G., Foty, R., Shafrir, Y., Steinberg, M., 1998. Viscoelastic

properties of living tissues: a quantitative study. Biophys. J. 74,

2227–2234.

Gallagher, B., Sakai, L., Little, C., 1993. Fibrillin delineates the primary

axis of the early avian embryo. Dev. Dyn. 196, 70–78.

Hamburger, V., Hamilton, H., 1951. A series of normal stages in the devel-

opment of the chick embryo. J. Morphol. 88, 49–92.

Hay, E. (Ed.), 1981. Cell Biology of Extracellular Matrix. Plenum, New

York, pp. 1–417.

Hegedus, B., Czirok, A., Fazekas, I., Babel, T., Madarasz, E., Vicsek, T.,

2000. Locomotion and proliferation of glioblastoma cells in vitro: stat-

istical evaluation of videomicroscopic observations. J. Neurosurg. 92,

428–434.

Helmke, B., Davies, P., 2002. The cytoskeleton under external fluid me-

chanical forces: hemodynamic forces acting on the endothelium. Ann.

Biomed. Eng. 30 (3), 284–296.

Hove, J., Koster, R., Forouhar, A., Acevedo-Bolton, G., Fraser, S., Gharib,

M., 2003. Intracardiac fluid forces are an essential epigenetic factor for

embryonic cardiogenesis. Nature 421, 172–177.

Keller, R., Davidson, L.A., Shook, D.R., 2003. How we are shaped: the

biomechanics of gastrulation. Differentiation 71, 171–205.

Little, C., Drake, C., 2000. Whole-mount immunolabeling of embryos by

microinjection. Methods Mol. Biol. 135, 183–189.

Moore, S., Keller, R., Koehl, M., 1995. The dorsal involuting marginal

zone stiffens anisotropically during its convergent extension in the gas-

trula of xenopus laevis. Development 121, 3131–3140.

Noble, D., 2002. Modeling the heart—from genes to cells to the whole

organ. Science 295, 1678–1682.

Ohashi, T., Kiehart, D., Erickson, H., 1999. Dynamics and elasticity of

the fibronectin matrix in living cell culture visualized by fibronectin-

green fluorescent protein. Proc. Natl. Acad. Sci. U. S. A. 96 (5),

2153–2158.

Oliver, T., Dembo, M., Jacobson, K., 1995. Traction forces in locomoting

cells. Cell Motil. Cytoskeleton 31 (3), 225–240.

Olsen, L., Maini, P., Sherratt, J., Dallon, J., 1999. Mathematical model-

ling of anisotropy in fibrous connective tissue. Math. Biosci. 158,

145–170.

Oster, G., Murray, J., Harris, A., 1983. Mechanical aspects of mesenchymal

morphogenesis. J. Embryol. Exp. Morphol. 78, 83–125.

Pereira, L., Andrikopoulos, K., Tian, J., Lee, S., Keene, D., Ono, R.,

Reinhardt, D., Sakai, L., Biery, N., Bunton, T., Dietz, H., Ramirez,

F., 1997. Targeting of the gene encoding fibrillin-1 recapitulates the

vascular aspect of Marfan syndrome. Nat. Genet. 17, 218–222.

Pereira, L., Lee, S., Gayraud, B., Andrikopoulos, K., Shapiro, S., Bunton,

T., Biery, N., Dietz, H., Sakai, L., Ramirez, F., 1999. Pathogenic se-

quence for aneurysm revealed in mice underexpressing fibrillin-1. Proc.

Natl. Acad. Sci. 96, 3819–3823.

Psychoyos, D., Stern, C., 1996. Fates and migratory routes of prim-

itive streak cells in the chick embryo. Development 122 (5),

1523–1534.

Rongish, B., Drake, C., Argraves, W., Little, C., 1998. Identification of the

developmental marker, JB3-antigen, as fibrillin-2 and its de novo organ-

ization into embryonic microfibrous arrays. Dev. Dyn. 212, 461–471.

Rupp, P., Rongish, B., Czirok, A., Little, C., 2003. Culturing of avian

embryos for time-lapse imaging. Biotechniques 34, 274–278.

Sakai, L., Keene, D., Engvall, E., 1986. Fibrillin, a new 350-kD glycopro-

tein, is a component of extracellular microfibrils. J. Cell Biol. 103,

2499–2509.

Sawada, K., Aoyama, H., 1999. Fate maps of the primitive streak in

chick and quail embryo: ingression timing of progenitor cells of

each rostro-caudal axial level of somites. Int. J. Dev. Biol. 43 (8),

809–815.

Schoenwolf, G., Garcia-Martinez, V., Dias, M., 1992. Mesoderm move-

ment and fate during avian gastrulation and neurulation. Dev. Dyn. 193,

235–248.

Schwarzbauer, J., Sechler, J., 1999. Fibronectin fibrillogenesis: a para-

digm for extracellular matrix assembly. Curr. Opin. Cell Biol. 11 (5),

622–627.

Skoglund, P., 1996. The role of Xenopus fibrillin in the early embryo: clues

from a dominant negative approach. PhD thesis, UCSD, San Diego.


Spratt, N., 1946. Formation of the primitive streak in the explanted chick

blastoderm marked with carbon particles. J. Exp. Zool. 128, 121–163.

Spratt, N., 1947. Regression and shortening of the primitive streak in the

explanted chick blastoderm. J. Exp. Zool. 104, 69–100.

Stern, C., Canning, D., 1988. Gastrulation in birds: a model system for the

study of animal morphogenesis. Experientia 44 (8), 651–657.

Stoplak, D., Harris, A., 1982. Connective tissue morphogenesis by fibro-

blast traction. Dev. Biol. 90, 383–398.

Tranquillo, R., 1999. Self-organization of tissue-equivalents: the nature and

role of contact guidance. Biochem. Soc. Symp. 65, 27–42.

Trinkaus, J., 1984. Cells into Organs. The Forces That Shape the Embryo,

second ed. Prentice-Hall Inc, Englewood Cliffs, NJ.

Vanni, G.S., Lagerholm, B.C., Otey, C., Lanni, F., 2001. Fibroblast con-

tractility and mechanics in a model extracellular matrix. Mol. Biol. Cell

12, 171a.

Vernon, R., Lara, S., Drake, C., Iruela-Arispe, M., Angello, J., Little, C.,

Wight, T., Sage, E., 1995. Organized type I collagen influences endo-

thelial patterns during ‘‘spontaneous angiogenesis in vitro’’: planar cul-

tures as models of vascular development. In Vitro Cell. Dev. Biol.:

Anim. 31 (3), 120–131.

Visconti, R., Barth, J., Keeley, F., Little, C., 2003. Elastin and

related fibrillar proteins in early development. Matrix Biol. 22,

109–121.

Voronov, D., Taber, L., 2002. Cardiac looping in experimental conditions:

effects of extraembryonic forces. Dev. Dyn. 224, 413–421.

Wallingford, J., Fraser, S., Harland, R., 2002. Convergent extension: the

molecular control of polarized cell movement during embryonic devel-

opment. Dev. Cell 2 (6), 695–706.

Wunsch, A., Little, C., Markwald, R., 1994. Cardiac endothelial hetero-

geneity defines valvular development as demonstrated by the diverse

expression of JB3, an antigen of the endocardial cushion tissue. Dev.

Biol. 165, 585–601.

Yang, X., Dormann, D., Munsterberg, A., Weijer, C., 2002. Cell move-

ment patterns during gastrulation in the chick are controlled by positive

and negative chemotaxis mediated by FGF4 and FGF8. Dev. Cell 3,

425–437.

Zhang, H., Apfelroth, S., Hu, W., Davis, E., Sanguineti, C., Bonadio, J.,

Mecham, R., Ramirez, F., 1994. Structure and expression of fibrillin-2,

a novel microfibrillar component preferentially located in elastic

matrices. J. Cell Biol. 124 (5), 855–863.

Documents

Extracellular matrix dynamics during vertebrate axis formation